Flacs v9 Manual

FLACS v9.
0 Users Manual
Copyright GexCon AS
Thursday March 12 2009
Contents
1 Introduction 1
1.1 About this publication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 About this manual . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Feedback from users . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Getting started 7
2.1 Prerequisites for users . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Hardware and software requirements . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Software installation and setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Running FLACS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5 Help and support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.6 Introductory example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3 CASD 31
3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2 File menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.3 Geometry menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.4 Object window in CASD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.5 Grid menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.6 Porosities menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.7 Scenario menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.8 Block menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.9 View menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.10 Options menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3.11 Macro menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.12 Help menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.13 Potential bugs or problems with CASD . . . . . . . . . . . . . . . . . . . . . . . . . 110
ii CONTENTS
4 Flacs simulator 113
4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.2 The Run Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.3 Running several simulations in series . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.4 Output variables in FLACS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
4.5 Files in FLACS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
4.6 Input les to FLACS simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
4.7 Output les from FLACS simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 149
4.8 Potential bugs or problems with Flacs . . . . . . . . . . . . . . . . . . . . . . . . . . 153
4.9 Warning and error messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
5 Flowvis 155
5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
5.2 Creating a new presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
5.3 File menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
5.4 Edit menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
5.5 Page menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
5.6 Plot menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
5.7 Verify menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
5.8 Options menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
5.9 Help . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
5.10 Flowvis examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
6 Utility programs in FLACS 187
6.1 Geometry, grid and porosities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
6.2 Release source modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
6.3 Modifying simulation les . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
6.4 Post-processing of simulation data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
7 Best practice examples 205
7.1 Combined dispersion and explosion simulations with FLACS . . . . . . . . . . . . 206
7.2 Simulation Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
7.3 Equivalent Stoichiometric Gas Cloud . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
7.4 Dispersion simulation with wind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
7.5 Hydrogen explosions and DDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
8 Technical Reference 221
8.1 Denitions and gas thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
FLACS v9.0 Users Manual
CONTENTS iii
8.2 Stoichiometric reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
8.3 Governing equations for uid ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
8.4 Wall functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
8.5 Wind boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
8.6 Combustion modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
8.7 Modelling of jet sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
8.8 Numerical Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
8.9 Linux Quick Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
9 Nomenclature 237
9.1 Roman letters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
9.2 Greek letters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
9.3 Subscripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
9.4 Dimensionless groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
9.5 Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
9.6 FLACS variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
10 References 243
Chapter 1
Introduction
1.1 About this publication
Copyright 2009 GexCon AS
All rights reserved
Updated: January 26 2009
Typeset in Doxygen
Printed in Norway
Intellectual property notice
No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in
any form or by any means, electronic, mechanical, photocopying, or otherwise, without written
permission from GexCon AS.
GexCon AS hereby grants permission to use, copy, and print this publication to organizations or
individuals holding a valid licence for one or several of the software packages described herein.
For further information about GexCon AS, please visit the web site: http://www.gexcon.com
Exclusion of liability
GexCon AS has distributed this publication in the hope that it will be useful, but without any
warranty, without even the implied warranty of merchantability or tness for a particular pur-
pose.
Although great care has been taken in the production of this publication to ensure accuracy,
GexCon AS cannot under any circumstances accept responsibility for errors, omissions, or advice
given herein.
2 Introduction
Registered trademarks
FLACS, DESC, CASD, and Flowvis are registered trademarks of GexCon AS.
Linux is a registered trademark of Linus Torvalds.
Windows is a registered trademark of Microsoft Corporation.
Other product names mentioned herein are used for identication purposes only and may be
trademarks of their respective companies.
1.2 Preface
Computational uid dynamics (CFD) is a branch of uid mechanics that uses numerical methods
and algorithms to solve and analyze problems that involve uid ow, with or without chemical
reactions. Current use of CFD covers a broad range of applications, from fundamental theoret-
ical studies involving models primarily derived from rst principles, to practical engineering
calculations utilizing phenomenological or empirical correlations.
Many of the hazards encountered in the society, and especially in the process industries, involve
accident scenarios where uid ow in complex, large-scale, three-dimensional (3D) geometries
play a key role. FLACS is a specialized CFD toolbox developed especially to address process
safety applications such as:
Dispersion of ammable or toxic gas
Gas and dust explosions
Propagation of blast and shock waves
Pool and jet res
The development of FLACS started in 1980 at the Department of Science and Technology at Chris-
tian Michelsen Institute (CMI). CMI established GexCon (Global Explosion Consultants) as a con-
sultancy activity under the Process Safety Group in 1987. In 1992, the Science and Technology
department at CMI became Christian Michelsen Research (CMR), and CMR established GexCon
as a private limited company in 1998. GexCon AS is a wholly owned subsidiary of CMR, and
holds the full proprietary rights to the CFD code FLACS.
The purpose of this manual is primarily to assist FLACS users in their practical work with the
software. In addition, the manual aims at documenting both the physical and chemical models,
and the numerical schemes and solvers, implemented in the CFD code. Ample references to
published literature describe the capabilities and inherent limitations of the software.
1.3 Acknowledgements
The development of the FLACS software would not have been possible without the generous
contributions received from supporting companies and government institutions throughout the
years. The activity started at Christian Michelsen Institute (CMI) in 1980 with the Gas Explosion
Programmes (GEPs), and FLACS-86 was the rst version distributed to the supporting compa-
nies.
1.3 Acknowledgements 3
Figure 1.1: The M24 compressor module represented in FLACS-86
The development of FLACS continued with the Gas Safety Programs (GSPs) and related projects
up to around 2000:
BP, Elf, Esso (Exxon), Mobil, Norsk Hydro, and Statoil supported the development of
FLACS-86 during the First GEP (1980-1986).
BP, Mobil, and Statoil supported the development of FLACS-89 during the Second GEP
(1986-1989).
BP, Elf, Esso, Mobil, Norsk Hydro, Statoil, Conoco, Philips Petroleum, Gaz de France, NV
Nederlandse Gasunie, Bundes Ministeriumfr Forschung und Technologie (BMFT), Health
and Safety Executive (HSE), and the Norwegian Petroleum Directorate supported the de-
velopment of FLACS-93 during the First GSP (1990-1992).
BP, Elf, Esso, Mobil, Statoil, Philips Petroleum, Gaz de France, HSE, and the Norwegian
Petroleum Directorate supported the development of FLACS-94, FLACS-95, and FLACS-96
during the Second GSP (1993-1996).
BP, Elf, Exxon, Mobil, Norsk Hydro, Statoil, Philips Petroleum, Gaz de France, HSE,
Agip, MEPTEC, and the Norwegian Petroleum Directorate supported the development of
FLACS-97, FLACS-98, and FLACS-99 during the Third GSP (1997-1999).
BP, TotalElfFina (TEF), Norsk Hydro, Statoil, Gaz de France, Philips Petroleum, Mobil and
supported the LICOREFLA project (2000-2001).
Since 2000, various Joint Industry Projects (JIPs), funding from the European Commission (EU)
and the Norwegian Research Council (NFR), and support and maintenance fees (S&M) from an
increasing number of commercial costumers have supported the development of the more recent
FLACS releases, including several specialized versions of FLACS, such as DESC (Dust Explosion
Simulation Code), FLACS-Dispersion, and FLACS-Hydrogen:
FLACS-Dispersion and FLACS-Hydrogen became available in 2001.
FLACS v8.0 came in 2003, including a test release of FLACS-Explo.
FLACS v8.1 came in 2005.
4 Introduction
DESC 1.0 came in 2006.
FLACS v9.0 came in 2008, including a test release of FLACS-Fire.
GexCon also develops several in-house R&D tools, including FLACS-Explo, FLACS-
Aerosol, and FLACS-Energy.
GexCon is grateful to all companies, government institutions, and individuals that have partici-
pated in the development of FLACS. We intend to honour these contributions by continuing to
develop the software, and thereby contribute to improved safety in the process industries.
1.4 About this manual
This Users Manual describes a family of computational uid dynamics (CFD) software products
from GexCon AS, generally referred to as FLACS:
The preprocessor CASD
The CFD simulator Flacs
The postprocessor Flowvis
Utility programs in FLACS such as:
geo2acs, gm, and Porcalc
jet and ash
rdle, cole, and comerge
r1le, r3le, and a1le
These programs constitute a specialized CFD tool, FLACS, or standard FLACS, designed to
study releases of ammable gas and gas explosions in complex congested geometries, both on-
shore and offshore. This manual also describes specialized versions of FLACS:
FLACS-Hydrogen
FLACS-Dispersion
FLACS-Aerosol
FLACS-Energy
FLACS-Explo
FLACS-Fire
DESC
A full version of Standard FLACS exhibits the full functionality of FLACS-Hydrogen and FLACS-
Dispersion, whereas DESC and FLACS-Fire are separate software products. FLACS-Energy,
FLACS-Explo, and FLACS-Aerosol are still in-house R&D tools. The acronym FLACS (FLame
ACceleration Simulator) refers to the complete package of software products, whereas the term
Flacs refers specically to the numerical solver in the CFD code.
The latest release of FLACS is version 9.0 (FLACS v9.0). This version represents a major upgrade
to the graphical user interfaces (GUIs), and is the rst version that runs under both the Linux and
Windows operating systems.
Getting started presents a detailed example for new users of FLACS, and Best practice examples
contains further examples that highlight various applications of FLACS, including some of the
specialized versions.
Technical reference contains technical reference material.
1.4 About this manual 5
1.4.1 Printing conventions in this manual
The symbol > followed by text in typewriter font indicates command line input, e.g.:
> command -options arguments (general syntax for commands)
> find -name flacs (command line input in Linux)
The symbol followed by text in typewriter font eld input commands, e.g.:
exit yes yes
The symbolindicates a path through nested menu items or dialog box options, e.g.:
FileSave
ScenarioIgnitionTime of ignition
Certain features of the software may only be accessible through text le input, and the
content of a text le is also printed in typewriter font:
THE FIRST LINE OF THE FILE ...
THE SECOND LINE OF THE FILE ...
... ...
The format for describing keyboard and mouse input follows the pattern:
CTRL+C
CTRL+MOUSE+LEFT
The use of bold or italic font emphasizes specic words or phrases in the text.
The Nomenclature chapter lists the symbols and abbreviations adopted in this manual.
1.4.2 Special messages
Warning:
Look out for the potential pitfalls pointed out by this heading!
Attention:
Be aware of practical information pointed out by this heading.
Remarks:
Take notice of the points summarized under this heading.
See also:
Follow up the additional sources of information suggested by this heading if required.
1.4.3 Job numbers
The typical application of the FLACS software is to quantify potential consequences of industrial
accident scenarios involving compressible uid ow, with or without chemical reactions. Proper
characterization of a particular problem may involve several simulations, and it is usually conve-
nient to organize the les from related scenarios in a dedicated directory. The individual FLACS
simulations are assigned job numbers, or simulation numbers, or simply jobs. A user may for
instance type:
> run9 flacs 010100
on the command line in Linux to start a FLACS simulation for job number 010100.
The job numbers are constructed from a six-digit string ijklmn, where traditionally:
6 Introduction
ij is the project number.
kl is the geometry number.
mn is the sequence number.
The default job number used in many of the examples in this manual is 010100, i.e. project 01,
geometry 01, simulation 00. However, each of the six digits in the job number may in principle
take on any integer value from zero to nine, and the references to project, geometry, and sequence
numbers only apply when the job numbers are derived from the le database in CASD.
Any updated version of this manual may be found on the FLUG web site.
1.5 Feedback from users
Feedback on the content in this manual is most welcome, and FLACS users may submit their
comments or suggestions by e-mail to: flacs@gexcon.com
When submitting comments or suggestion to the content of the manual, or when pointing out
misprints in the text, please indicate the relevant page numbers or sections, and the correspond-
ing version of the manual (date issued).
Chapter 2
Getting started
8 Getting started
This chapter describes the basics of setting up the FLACS software for new users, including rec-
ommendations concerning the user threshold, typical hardware requirements, and procedures
for installing FLACS on both Linux and Windows.
2.1 Prerequisites for users
Efcient use of FLACS does not require detailed knowledge about computational uid dynam-
ics (CFD). However, users should possess some experience in the application of computers for
routine tasks, such as text editing. Proper interpretation of simulation results requires adequate
knowledge within the eld of uid dynamics. A suitable starting point for the novice in the eld
of gas explosions is the Gas Explosion Handbook (Bjerketvedt et al., 1993) from Christian Michelsen
Research (CMR), and newusers of FLACS should attend a three-day introductory course arranged
by GexCon AS (http://www.gexcon.com).
2.2 Hardware and software requirements
FLACS v9 is available on Linux and on Microsoft Windows. The hardware requirements for
running the FLACS software depend to some extent on the size of the problem in question, i.e.
the number of grid cells required to resolve the computational domain properly. Most modern
computers, be it desktops and laptops, will perform well for small or medium sized problems.
A powerful screen card may be required to handle large geometries in CASD, extra memory
(RAM) is necessary for simulating large problems, and storage of large amounts of simulation
data dictates the requirements for disk space.
Hardware requirements:
Processor: Intel or AMD ix86 32 bit, Intel EM64T or AMD64. Intel IA64 is not supported.
Internal memory; 2GB or more recommended.
Free harddrive capacity: 350MB for software installation and typically 100GB simulation
space.
Graphics card using NVIDIA chip set. Graphics cards using for instance ATI or Intel
chipsets are in general not supported.
DVD-RW drive recommended.
High resolution colour screen (minimum 19", 1600x1200, 24 bit color depth).
FLACS v9 has been tested on the following platforms.
Linux:
OpenSuse 10.0, 10.2, 10.3, 11.0
CentOS 4.6, 5.1
Ubuntu 7.10
Fedora 8
Microsoft Windows:
XP (32 bit)
Vista (32 bit)
2.3 Software installation and setup 9
Red Hat Enterprise Linux 4.6, 5.1 is expected to be FLACS v9 compatible since it is compatible
with CentOS 4.6, 5.1.
For updated hardware and software requirements, please refer to GexCons website,
http://www.gexcon.com .
2.3 Software installation and setup
A license server is necessary for running FLACS. This section presents FLACS installation, the
FLACS Licence Server and the FLACS Conguration Wizard that guides users through the basic
steps of setting up a FLACS Licence Server. All FLACS installations on a network acquire their
individual licenses from a central licence server, and only one FLACS License Server should
therefore be running on a given network.
FLACS is distributed in a single setup le.
2.3.1 On Linux
On Linux FLACS can be installed system wide, in which case FLACS will be available to all users,
or in a users home directory, in which case it will be available to this user only.
2.3.1.1 Installing in users home directory
If only one person will be using FLACS, the software can be installed in this users home directory.
FLACS will by default be installed under /home/my_user/GexCon.
Save the installation package to a convenient location.
Make sure the le is executable:
> chmod u+x /home/my_user/flacs-v9.0-installer
Run the installation program:
> /home/my_user/flacs-v9.0-installer
Please follow the instructions given. It is recommended to keep the default parameters.
FLACS requires a license to run. The license is provided by a license server, which is installed on
only one machine on the local network. During the installation the user can choose to install:
1. Both FLACS software and FLACS license manager
2. FLACS license manager only
For a user home directory installation option 1 should be selected.
The FLACS license manager must be set up before using FLACS. Please refer to the section about
FLACS congure wizard.
2.3.1.2 Installing system wide as super user
To install FLACS system wide, access to the system super user ("root") is required.
/path/to/installation is the path to the location of the FLACS installation package.
10 Getting started
Change user to super user ("root"):
> su <give password>
Make sure the le is executable:
> chmod u+x /path/to/installation/flacs-v9.0-installer
Run the installation program:
> /path/to/installation/flacs-v9.0-installer
Please follow the instructions given. It is recommended to keep the default parameters.
Option 2 can be used to install a FLACS license manager on a system not running FLACS. Alter-
natively one FLACS workstation in the network can be set up to serve licenses to all other FLACS
installations in the network.
The FLACS license manager must be set up before using FLACS. Please refer to the section about
FLACS congure wizard.
2.3.2 On Windows
To install FLACS on Windows please double-click the installation package "acs-v9.0-
installer.exe". This will start the installation wizard. Please follow the instructions given. It is
recommended to keep the default parameters.
Option 2 can be used to install a FLACS license manager on a system not running FLACS. Alter-
natively one FLACS workstation in the network can be set up to serve licenses to all other FLACS
installations in the network.
The FLACS license manager must be set up before using FLACS. Please refer to section Setting
up the FLACS license server.
2.3.3 Setting up the FLACS license server
FLACS version 9.0 has a completely newlicense server/manager system, which operates through
a network protocol. This means that the license manager can be installed anywhere on the net-
work, as long as it is available to the FLACS clients through the local network. The license man-
ager can be installed locally on the machine where the FLACS simulation software is installed, or
separately from the simulation software. Only one license manager should be running on your
network, and this is where the FLACS license is installed. All other FLACS installations should
be set up using this license manager.
After the installation is nished, the FLACS license conguration utility should start automat-
ically. In the event that this does not happen please start the conguration utility as follows,
depending on your installation.
Linux:
> /usr/local/GexCon/FLACS_v9.0/bin/run configureWizard
Windows:
> C:\Program Files\GexCon\FLACS_v9.0\bin\configureWizard.exe
Alternatively it can be started from the FLACS Runmanager HelpStart Conguration Wizard.
If FLACS is installed system wide (installed as root), on Linux, the license manager must be
running as user root.
The conguration utility will guide you through the setup of the license manager. The cong-
uration utility is also used to congure a FLACS installation that gets its license from a license
manager on a separate machine.
2.3.3.1 Setting up the license server on client only FLACS installation
If a FLACS license server is installed and running somehwere on the local network, the FLACS
installation must be congured to connect to the license server.
Figure 2.1: Setting up the license server on client only FLACS installation
2.3.3.2 Setting up the license server on a combined license server and client FLACS installa-
tion
If there is no FLACS license server available on the local network, a license server must be in-
stalled. To install a license server together with the FLACS simulation software, on the same
12 Getting started
machine, please use the following procedure. Alternatively a FLACS license server can be in-
stalled on a separate machine, with or without FLACS software. Please refer to section Stan-
dalone FLACS license manager installation.
Figure 2.2: Setting up the license server on a combined license server and client FLACS installa-
tion (steps 1 and 2)
Figure 2.3: Setting up the license server on a combined license server and client FLACS installa-
tion (steps 3 and 4)
2.3.3.3 Standalone FLACS license manager installation
It is possible to install the FLACS license manager only. This is useful if you would like to have
the license manager on a separate machine. To do this select the appropriate option during in-
stallation (see Software installation and setup).
To congure a standalone FLACS license manager prompt the license manager for an activation
key, by running the following command in a terminal window.
Linux:
> /usr/local/GexCon/FLACS_LicenseManager/bin/FLMserver --get-ActivationKey
Windows:
> C:\Program Files\GexCon\FLACS_LicenseManager\bin\FLMserver.exe --get-ActivationKey
Send the activation key together with the IP address and license manager communication port
number to <flacs@gexcon.com>.
The communication port defaults to 25001. Please make sure that this port is available, and open
on your system. If you are not sure about this please contact your system administrator.
GexCon will, based on the activation key, create a license text le. This le must be saved to:
Linux:
> /usr/local/GexCon/FLACS_LicenseManager/license/license-server.flm
Windows:
> C:\Program Files\GexCon\FLACS_LicenseManager\license\license-server.flm
Note that when using a standalone FLACS license manager, the license manager must be started
manually each time the computer is restarted. This can be done using a startup script (not pro-
vided).
2.3.3.4 Starting FLACS license manager as a service on Windows
The FLACS license manager can be started as a service on Windows using the following proce-
dure.
1. Verify that the FLACS License Manager is working properly as a desktop application
(a) FLACS software and license key must be installed (see procedure above)
(b) Test run FLMserver with the graphical user interface and then quit: > "C:\Program
Files\GexCon\FLACS_LicenseManager\bin\FLMserver.exe"
(c) Make sure to quit FLMserver, the service will not function if there is a desktop FLM-
server running.
2. Download and install the Windows Resource Kit (rktools.exe)
(a) See the following links about Windows Services and related tools:
http://search.microsoft.com/results.aspx?mkt=en-US&setlang=en-US&q=rktools.exe
http://support.microsoft.com/kb/137890
3. Install the FLACS License Manager service "FLMserver" using INSTSRV:
(a) > instsrv FLMserver "C:\Program Files\Windows Resource
Kits\Tools\srvany.exe"
(b) The service can be removed with > instsrv FLMserver REMOVE
(c) The path to srvany.exe might be different on your Windows installation
14 Getting started
4. Run REGEDIT to set up the details of the service
(a) It is strongly advised to backup your current registry before editing
(b) > regedit
(c) Locate and select the FLMserver key:
"HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\FLMserver"
(d) Add one new value for FLMserver: Description
EditNewString Value : "Description"
Description = "FLACS License Manager service."
(e) Add one new key for FLMserver: Parameters
EditNewKey : "Parameters"
Add two new values for FLMserver\Parameters: Application and AppParam-
eters
*
EditNewString Value : "Application"
*
EditNewString Value : "AppParameters"
*
Application = "C:\Program Files\GexCon\FLACS_-
LicenseManager\bin\FLMserver.exe"
*
AppParameters = without-gui
*
IMPORTANT NOTE: options start with double dashes: without-gui
The service will start automatically on reboot, it can also be started/stopped man-
ually:
Control PanelAdministrative ToolsServices
2.3.4 Setting up the FLACS environment
After installation FLACS programs can be accessed from the system menu, in the following loca-
tions:
Linux (KDE): StartApplicationsEdutainmentConstruction
Linux (Gnome): ApplicationsOther
Windows: StartAll ProgramsGexConFLACS_v9.0
Some systems may require the user to log out and restart before FLACS will appear in the system
menu.
Desktops that do not follow the freedesktop.org standards will not install an icon in the Appli-
cations menu. This will happen on older distributions. In these cases, the user may be able to
install icons and associations manually. Refer to your GNU/Linux distribution vendor for details
on how to customize your desktop.
2.3.4.1 FLACS User setup on Linux
For easy access to FLACS from the command line add the following text to you startup le.
If you use the csh/tcsh shell, edit or create the .cshrc le:
alias run9 /usr/local/GexCon/FLACS_v9.0/bin/run
If you use the bash shell, edit or create the .bashrc le:
alias run9=/usr/local/GexCon/FLACS_v9.0/bin/run
FLACS programs can the be started by typing eg. run9 flowvis.
2.4 Running FLACS 15
2.3.5 Uninstalling FLACS
Linux: Run the program "/usr/local/GexCon/uninstall-GexCon.sh".
Windows: FLACS can be uninstalled using Control Panel/Add or Remove Programs.
2.4 Running FLACS
A typical simulation session with the CFD code FLACS involves several steps. Assuming FLACS
is properly installed on the computer, including valid lisence les for the software, users can
initiate a FLACS session by clicking the FLACS icon on the desctop:
Figure 2.4: The FLACS icon
This should open the Run Manager window:
Figure 2.5: The FLACS Runmanager
Some of the main tasks of the Run Manager are:
Starting the Licence Manager
Starting the preprocessor CASD
Running CFD simulations
16 Getting started
Starting the postprocessors Flowvis
The preprocessor should start when clicking the CASD icon in the Run Manager:
Figure 2.6: The CASD icon
The CASD window looks like this:
Figure 2.7: FLACS preprocessor CASD
Work in CASD often involves opening the Database window from the Geometry menu:
GeometryDatabase
The Database window looks like this:
2.4 Running FLACS 17
Figure 2.8: Geometry database window
Typical tasks performed from the Database window include:
Creating a new database and new geometries
Opening existing databases and geometries
Creating new materials (i.e. colours), or modifying existing materials
Creating new objects, or modifying existing objects
The New Object button, available in the Objects tab in the Database window, opens the Object
window:
18 Getting started
Figure 2.9: CASD object window
The main purpose of the Object window is to construct a new object, or to modify an existing
object. Users can build complex objects by adding or subtracting several insides (i.e. boxes or
cylinders). Any geometry can consist of one or several objects, or assemblies of several objects.
An alternative way of working with geometries involves geometry import using the geo2acs
utility . However, this requires that a representation of the geometry already exists on a compat-
ible CAD format (typically Microstation or PDMS).
Apart from geometry building, the menus in CASD also perform the following tasks:
Denition of the computational domain and the computational grid
Porosity calculations with the utility program Porcalc, as well as porosity verication
Scenario setup, including:
Denition of monitor point locations, and selection of output variables
Specication of boundary conditions
Specication of vent panels and leaks
Specication of fuel type
Specication of ignition position and time of ignition
After dening the scenario, the next step is to run the actual FLACS simulations:
Simulations can be started and monitored with the run manager
The same operations can be controlled from the command line in Linux
> run9 flacs 010100
Note that the Run Manager also monitors the simulations while they are running.
2.5 Help and support 19
The nal step in a FLACS session is typically the presentation and verication of simulation re-
sults with the postprocessor Flowvis, as well as data extraction and reporting. The postprocessor
should start when clicking the Flowvis icon in the Run Manager:
Figure 2.10: The Flowvis icon
The Flowvis window looks like this:
Figure 2.11: FLACS postprocessor Flowvis
Some of the most frequently used features in Flowvis include:
Verifying porosities in a geometry
Creating scalar-time plots, 2D-plots, 3D-plots, ...
Creating animations
Data reporting may also include the extraction of numerical simulation results with the utility
programs r1-le and r3-le. These programs run only from command line input in the current
version of FLACS.
2.5 Help and support
FLACS users can get technical support by contacting GexCon software department:
20 Getting started
Email: flacs@gexcon.com
Phone: +47 55574330
Commercial customers are entitled to support and maintenance:
Support: Up to 70 hours of email or phone support per year
Maintenance: New releases of FLACS as they are made available
In addition to the above the user has access to the FLACS User Group web site, which contains
information about FLACS, including a FAQ (Frequently Asked Questions) and a self support
portal where the user can search for answers (as of November 2008 GexCon is working on im-
plementing the self support portal, but a release date is not yet decided)
The support and maintenance requires the user to have a payed and valid support and mainte-
nance contract.
2.6 Introductory example
This chapter contains an introductory example. It gives a rst impression of how to set up and
run a simple FLACS explosion simulation. For additional examples see sections Best practice
examples and Flowvis examples.
2.6.1 Things to keep in mind before you begin
FLACS is a CFD (Computational Fluid Dynamics) Explosion Simulator tool. The input to a CFD
calculation is:
A geometry, either created manually for the specic purpose, or imported from a CAD
system
A grid which divides the simulation domain into cells. In one cell a variable (eg. pressure)
does not vary in space. FLACS use a regular, Cartesian grid, which means box grid cells.
Various scenario parameter, such as boundary conditions, monitoring point locations, gas
cloud size, position and composition, and ignition location.
All of the above is normally handled in the FLACS pre-processor CASD. The geometry is saved to
a le structure, called a le database. The le database le structure starts in a top level directory
given a name with sufx ".db". The le database should not contain user les, or les other
than those created by the le database interface in CASD.
In addition to the le database a number of other les are created before and during the simula-
tion. All les contains the job number, a 6 digit number. The following les are created as input to
the simulation (010101 is the job number).
cg010101.dat3 The grid le
cs010101.dat3 The scenario le
co010101.dat3 The geometry le. This le contains a snapshot of the geometry contained in the
le database.
cp010101.dat3 The porosity le, which is created by Porcalc. Please see section and Porcalc for
details.
2.6 Introductory example 21
During the simulation a set of result les will be created:
r1010101.dat3 Scalar-time output from monitor points
r3010101.dat3 Field output at selected times. Needed to create 2D and 3D plots
rt010101.dat3 Simulation log le
FLACS can also create and use other les. Please see section Files in FLACS for details.
Due to the number of les created by each simulation it is important to create a good le struc-
ture of directories to keep track of the les. See section Files in FLACS for details and further
recommendations.
2.6.2 Initialising the work directory
As FLACS creates a relatively large number of les it is important to have a good system for book
keeping. It is recommended to start out with an empty directory.
2.6.2.1 On Linux
Make a distinct directory (DIRECTORY_NAME) in which you perform the exercise:
> mkdir DIRECTORY_NAME
Move into this directory:
> cd DIRECTORY_NAME
Copy geometry les (notice the space before the ".").
> cp /usr/local/GexCon/FLACS_v9.0/doc/examples/ex2/
*
00001
*
.
Start up the FLACS runmanager:
> run9 runmanager
2.6.2.2 On Windows
1. Make a distinct directory in which you perform the exercise: Open the le browser ("My
Documents") and choose FileNewFolder.
2. Copy les from C:\Program Files\GexCon\FLACS_v9.0\doc\examples\ex2\00001
(00001 means all les containing the text "00001").
3. Start the FLACS runmanager by clicking the desktop icon, or go to Start MenuAll
ProgramsGexConFLACS_v9.0FLACS Runmanager.
2.6.3 Initialising and starting the preprocessor CASD
Use Run Manager Tools CASD (or click the FLACS pre-processor icon)
22 Getting started
2.6.3.1 Open and view the geometry in CASD (Move cursor to the CASD window)
1. choose OPEN in the FILE menu OR file open <CR> OR ALT-f o (<CR> means carrige
return, ie. the enter key)
CASD Ask for opening an existing job le
2. choose 100001.caj <OK>
CASD: Open joble 100001, using MOUSE+LEFT
3. if any error message appears click <OK>
CASD: Ignore error message => error message
CASD: Play with visualisation options, y through geometry etc.
Figure 2.12: The geometry used in example 1
2.6.3.2 Make a grid for the simulation
Make a grid (mesh) for the simulation, calculate porosities (module dim.: 25.6m x 8m x 8m, origin
in corner below the control room).
1. Choose SIMULATION_VOLUME from GRID menu
CASD: To enter the extension of the simulation domain
2. Enter -16 <TAB> -8 <TAB> 0 <TAB> 40 <TAB> 16 <TAB> 16 <OK>
CASD: Volume is dened (16m out from vent, 8m to the sides; observe - sign)
3. In GRID menu, choose DIRECTION X
4. In GRID menu, choose REGION and enter 56 <OK>
CASD: 56 grid cells chosen (1.0m grid size).
5. Repeat steps for Y direction and use REGION 24
CASD: 24 cells in Y-direction
6. Repeat steps for Z direction and use REGION 16
CASD: 16 cells in Z-direction
7. In GRID menu, click INFORMATION, and <OK> to close window
CASD: Check that grid dimension is 1.0m as intended
8. Choose SAVE from the FILE menu
CASD: Save geometry and grid les
9. Choose CALCULATE from POROSITIES menu
CASD: Map geometry information onto the grid, porcalc
10. Choose DISPLAY OFF in the GRID menus
CASD: Dont draw the grid anymore
Figure 2.13: Embedding the grid
24 Getting started
Figure 2.14: Porsity calculations using Porcalc
2.6.3.3 Dene explosion scenario
1. Choose MONITOR_POINTS in SCENARIO menu OR scen mon <CR>
CASD: Dene where to measure variables
2. Click <ADD>, <EDIT> and 0.8 <TAB> 4.7 <TAB> 7.9 <OK>
CASD: Add and dene location of monitor point 1
3. Repeat this for point 2 (12.3, 4, 0.1) and point 3 (24, 7.9, 7.9)
CASD: To edit a non-highlighted monitor, click on its number
4. Click <OK>
CASD: Close MONITOR_POINT window
5. Choose SINGLE_FIELD_SCALAR from SCENARIO menu
CASD: Dene which variables to report at monitors
6. Click on <P>, drag mouse pushing MOUSE+LEFT across all monitors, <OK>
CASD: Log pressure at all three transducers
7. Repeat for <PIMP> and <DRAG>
CASD: Log pressure impulse and dynamic pressure, too
8. Click <OK> and choose SINGLE_FIELD_3D from SCENARIO menu
CASD: Dene variables for contour plots
9. Click on <P>, CTRL-<PROD>, CTRL-<VVEC>, <OK>
CASD: Pressure, ame and velocity vectors. CTRL needed to select more than one
(NB! deselect when using the scroll bar)
10. Choose SIMULATION in SCENARIO menu OR scen sim <CR>
CASD: Choose output and simulation parameters
11. Click on <NPLOT>, enter 50 <OK>, <OK>
CASD: Increase number of contour plots, return to main menu
12. Click on GAS_COMP... in SCENARIO menu OR scen gas_c <CR>
CASD: Dene gas cloud loc., size, comp. and concentration
13. Click on <POS...>, 0 <TAB> 0 <TAB> 0 <OK>
CASD: Position of bounding box describing gas cloud
14. Click on <DIM...>, 25.6 <TAB> 8 <TAB> 8 <OK>
CASD: Dimension of gas cloud equals module dimensions
15. Click on <VOL...>, <METHANE> 91.7 <OK> <ETHANE> 7 <OK> <PROPANE> 1.3
<OK> <OK>
CASD: Gas composition is dened
16. Click on <EQUI...> 1.05 <TAB> 0 <OK> <OK>
CASD: Slightly rich gas mixture is chosen ER=1.05
17. Click on IGNITION in SCENARIO menu <POS...> 12.5 <TAB> 4.1 <TAB> 4.25 <OK>
<OK> OR scen ign pos 12.5 4.1 4.25 OK <CR>
CASD: Dene location of ignition (12.5, 4.1, 4.25)
18. Choose SAVE from the FILE menu
CASD: Save all les, ready to run acs
19. Minimize CASD
CASD: Leave CASD for now, can be activated easily
Figure 2.15: Adding monitoring points
26 Getting started
Figure 2.16: Choosing variables for 3D output
Figure 2.17: Adding a gas cloud and choosing the gas composition
2.6.4 Start FLACS simulation
Select the job in Run Manager and click simulate (if job not visible, use add directory or if di-
rectory is already added, right click and rescan), check how the simulation starts up (click log
le)
Figure 2.18: Running a simulation in the FLACS Runmanager
2.6.5 Study results in post prosessor Flowvis
Use Run Manager Tools Flowvis (or click the FLACS post-processor icon)
1. choose ADD from Page menu (or CTRL+a)
FLOWVIS: Prepare rst page
2. click MOUSE+RIGHT, choose PLOT_TYPE and SCALAR_TIME plot
FLOWVIS: Plotting of time histories of variables
3. choose 100001 and P with MOUSE+LEFT, select all 3 monitors (drag mouse) <OK>
FLOWVIS: Plot pressure time history at all monitors
4. <RESCAN>
FLOWVIS: if simulation is running rescan will update plot
5. Choose MODIFY in the Page menu (or CTRL+m), enter <TAB> 1 <TAB> 2 <OK>
FLOWVIS: divide page into 2 plots
6. Click at lower frame, then MOUSE+RIGHT, PLOT_TYPE, ANNOTATION_ST (or CTRL+0)
FLOWVIS: show numerical values from pressure plots
28 Getting started
7. ADD page and do the same for the DRAG and PIMP variables
8. Choose ADD in Page menu (or CTRL+a), click MOUSE+RIGHT, PLOT_TYPE, 2D... (or
CTRL+2)
FLOWVIS: prepare 2D contour plot
9. Choose 100001, P, click <OK>
FLOWVIS: contour plot of pressure
10. click MOUSE+RIGHT, choose PLOT_DOMAIN, change k-index to 5 <OK>
FLOWVIS: choose XY-cut plane through ignition
11. Click MOUSE+RIGHT, choose VARIABLE_APPEARANCE change Value Range Setting to
Fixed
FLOWVIS: choose a user-dened xed scale for all time steps
12. Choose Min. Value as 0.05 and Max. Value as 2.0
FLOWVIS: dene the scale
Figure 2.19: Showing pressure-time curves with annotation in Flowvis
Figure 2.20: 2D cutplane plot showing over-pressures
Figure 2.21: Setting plot domain for a volume plot
Time steps can now be changed moving the bottom scroll bar to the right, page can be varied
using the right scroll bar.
1. Repeat this method for PROD and VVEC variables (these can be plotted on the same plot)
30 Getting started
FLOWVIS: visualize ame and velocity vectors
Try to show PRESSURE and PROD on the same page using PAGE MODIFY (use a xed scale
for PROD from 0.15 to 0.2 and change Min. Color Index to 9 and Max to 10) Now that you are
familiar with Flowvis, try the volume plot menu to study the development of ame (PROD) and
pressure Use PLOT DOMAIN to narrow the view window and see below the ceiling
2.6.6 Study the effect of ignition location
Enter CASD, open the 100001.caj job-le, save this as a new job number e.g. 100002.caj Change
ignition location in order to study how pressures may vary with different ignition locations End
ignition (0.5, 4.1, 4.25), (job number 100002) Your own assumed worst-case location (job number
100003)
Report highest pressure achieved on monitor point
Make animation of either 2D or volume plots using the export menu (with all timesteps)
Chapter 3
CASD
32 CASD
The preprocessor CASD for the CFD simulator FLACS is used to prepare the input data, or job
data , that denes a FLACS simulation: geometry model, computational grid, porosites, and
scenario description. CASD is an acronym for Computer Aided Scenario Design.
CASD 4 released in 1994, use X11 graphics, but a new version is available based on QT
CASD 5 released in 2001, use Open Inventor graphics
CASD 6 released in 2008, use QT and Coin 3D graphics
This manual describes CASD 6, but the general functionality of CASD 6 is in principle the same
for CASD 4 and CASD 5. CASD 6 is fully backward compatible with CASD 4 and CASD 5.
3.1 Overview
This section provides a general overview of the functionality in CASD.
3.1.1 Starting CASD
Users start CASD by clicking the CASD icon in the run manager window:
Figure 3.1: The CASD desktop icon
or alternatively by executing the command:
> run9 casd6
on the command line in Linux.
3.1.2 CASD command line options
The following options can be given when starting CASD on the command line:
Option Description
-macro macro le name Read input from specied macro le
-numMat maximum number of materials Default is 50
-numObj maximum number of objects Default is 10000
-numAsis maxmimum number of
assemblies/instances
Default is 3500
-stackAsis maxmimu number of nested assembly
levels
Default is 8
-noLock Turns of locking on the database les. Must
not be used if more than one user accesses the
database simultaneously. This option speeds
up the database operations signicantly.
3.1 Overview 33
-display and others Linux: options accepted by X
Table 3.1: CASD command line options
Example:
Linux:
run9 casd -numObj 20000 -numAsis 20000 -noLock
Windows:
casd -numObj 20000 -numAsis 20000 -noLock
Alternatively the options can be set permanently in the FLACS Runmanager,
OptionsPreferences. This will only apply if CASD is started from the Runmanager.
3.1.3 The main window in CASD
Starting CASD 6 opens the main window.
Figure 3.2: The main window in CASD
The main window is divided into the following parts:
The menu bar
34 CASD
The icon bar
The command input eld
The geometry window(s)
The status eld
These parts are described in the following subsections.
3.1.4 The menu bar
The menu bar contains the following menus:
File
Geometry
Grid
Porosities
Scenario
Block
View
Options
Macro
Help
The options on the various menus are described in separate sections in this chapter.
3.1.5 The icon bar
The icon bar contains the following toolbars:
Main toolbar, provides shortcuts to several of the commands on the meny bar:
New, Open, Save, Save as, Import, and Result on the File menu.
Database icon on the Geometry menu.
Calculate and Verify porosities on the Porosity menu.
Graphics toolbar, controls various features of the geometry window(s).
View splitting.
Rectangle zoom.
Spinning (toggle on/off).
Highlighting option, from lled only (0) to various degrees of contour highlighting
(1-5).
Drawing toolbar, opens the plan drawing dialog box:
Specifying le names for texture (e.g. drawings).
File formats: PNG, JPEG, GIF, TIFF
3.1 Overview 35
3.1.6 The command input eld
The command input eld represents an alternative interface between the user and CASD, in
addition to the regular menus on the menu bar. The control input eld contains a scrollable
command history list, and a current command context indicator (left side). The user controls the
command history list from the keyboard:
UP: retrieves the previous line from the command history list
DOWN: retrieves the next line from the command history list
RETURN: processes the content of the command input eld
Hence, the user can choose whether to use a menu options on the menu bar, e.g: FileExitYes
(to exit and save) or to execute, after typing or retrieving, the following command in the com-
mand input eld:
file exit yes yes Command line input will in many situations be the most efcient way
to work with CASD, and other sections in this chapter present additional examples on how to
use this feature.
Examples: Using the command input eld in CASD
Select a box primitive in an object. The following command moves the box to (2, 2, 2), and
would cause the properties dialog to be shown
edit properties 2 2
This is because the position is not completely specied. The user does not have to
specify all parameters, but must include all values for the parameter specied.
If the user wants to edit one of the last parameters in the dialog, it is not necessary to specify
all the parameters in front. The parameter name can be used to indicate which parameter
to edit
edit properties size 2 2 2 vol_por 0.5
The user can also supply the answer to a question in the input eld. To delete an assem-
bly/instance, CASD will ask to conrm the operation. To avoid the question dialog, type
the following command
geometry delete yes
or shorter: ge de y
To direct the output from a list to a le, append the le name after the list command. For
instance, to list geometries in the database, enter the following command, which will create
the text le outfile.txt
geometry list outfile.txt
3.1.7 The graphical area
The graphical area in the main window displays the geometry and the computational grid. In
addition to the options on the View menu, there are several ways of manipulating the view:
Rotation: MOUSE+LEFT
Panning: CTRL+MOUSE+LEFT
Zoom: MOUSE+SCROLL
Rectangle zoom: MOUSE+RIGHT+SELECT
36 CASD
Splitting and closing views: MOUSE+RIGHT+SELECT
The use of these features are quite intuitive, and they will not be described in more detail in this
manual.
3.1.8 The message area
The message area in the main window contains information concerning the active database,
project, geometry, grid, and units.
3.1.9 Files in CASD
CASD stores job data on a set of les. For the arbitrary job number 010100, the most important
les are:
Header le, 010100.caj: ASCII le created by CASD; denes the co, cg, and cm les used by
CASD.
Geometry le, co010100.dat3: binary le created by CASD; contains a list of primitives from
a CASD database that dene the geometry; used by Porcalc and Flowvis.
Grid le, cg010100.dat3: binary le created by CASD; denes the computational mesh; used
by CASD, Flacs, and Flowvis.
Porosity le, cp010100.dat3: binary le created by Porcalc (typically from the Grid menu in
CASD); denes the porosities for each grid cell; used by Flacs and Flowvis.
Polygon le, cm010100.dat3: binary le created by CASD; denes the polygon model; used
by Flowvis (if the le exists).
Scenario le, cs010100.dat3: ASCII le created by CASD; denes the general scenario (mon-
itor points, output variables, fuel region, pressure relief panels, ignition position, etc.); used
by CASD, Flacs, and Flowvis.
The grid-le is also called the obstruction le, or co-le, and is not a direct input to the simulation; it
is however used by Porcalc when generating the porosity le. The File menu in the main window
contains commands for creating, opening, and saving the various job les. See section Files in
FLACS for further information.
3.1.10 Working with geometries in CASD
To implement the geometry model in CASD can often be the most time consuming part of a
project. For modern process facilities it may be possible to import a geometry from an existing
CAD model, but for many installations the geometry must be constructed manually from draw-
ings, photographs, etc.
A large projects, such as a full probabililistic analysis, can involve hundreds of CFD simulations,
and each simulation will typically produce 10-15 different les. Hence, it is very important to
organize the les in a well-structured manner.
The building blocks in a CASD geometry are instances of objects. The structure within an object
is a so-called Constructive Solid Geometry (CSG) model, where simple solid primitives (boxes
and cylinders) are combined by Boolean operators (unions and left differences).
Objects in CASD can be either global or local. Several geometries can contain instances of the
same global object, whereas a local object can only be included in the geometry where it was
created. It is generally recommended to use global objects, and avoid the use of local objects.
3.1 Overview 37
The list of information required to implement a typical process facility, such as an offshore oil
platform or an onshore process plant, is quite extensive:
Plot plan
Sectional drawings
Piping plan
HVAC layout
Cable trays layout
Framing plans
Cladding
Deck plan
Most FLACS users nd it convenient to dene standardized axis directions, and the following
convention is used by GexCon for typical process facilities:
East-West along the x-axis, with positive x towards the east.
North-South along the y-axis, with positive y towards the north.
Up-Down along the z-axis, with positive z pointing upwards.
This results in a conventional right handed coordinate system, where the lower south-western
corner of the facility coincides with the origin (0,0,0).
Each object in a CASD database is assigned a material property, and each material is assigned a
colour hue from the 0-360 colour circle. Many FLACS users nd it convenient to assign certain
hues to various structural elements, and the following convention is used by GexCon for typical
process facilities.
Hue Colour Description
0 Red solid walls and decks
30 Orange pressure relief and and
louvred panels
60 Yellow grated decks
120 Green anticipated congestion
180 Cyan equipment
200 Light blue structure
220 Medium Blue secondary structure
250 Dark Blue piping
300 Pink equipment
Table 3.2: Colour convention used by GexCon
A standardized colour scheme makes it more straightforward to review geometries from old
projects.
3.1.11 About congestion, connement, and vents
In order to have a good representation of the effect of obstacles it is important that they are well
represented geometrically by the chosen grid. In most practical situations it will not be possible
to represent the smaller obstacles on the grid, these should still be included since they may be
treated by proper sub-grid models. Larger obstacles like the oor (or the ground), the ceiling, the
walls and larger equipment will be resolved on-grid. This means that they will be adjusted to
match the grid lines.
38 CASD
The most challenging geometry to represent properly is repeated obstacles of the same size and
spacing as the chosen grid resolution, in such cases the user should consider to change the grid to
achieve a better representation. If this type of geometry is dominant it is of vital importance for
the accuracy of the result that the representation is good enough. In cases where such a geometry
is not dominant one may pay less attention to how it is represented. For normal offshore modules
there will be a range of subgrid sized obstacles which are more or less randomly distributed in
space.
In many experimental setups one will nd repeated obstacles of the same size. The basic research
on gas explosions past many years now has focused on the effect of obstacle arrays, perhaps to
a greater extent than on the effect of more realistic geometries. Both categories are important in
order to be able to validate tools like FLACS.
It is important to represent the vent openings of a semi-conned geometry properly. If obstacles
close to the outer boundaries are adjusted to match the grid, the effective vent area may be af-
fected. In order to verify that the representation of the vent openings is as good as possible the
user should check the porosity elds (using CASD or Flowvis).
3.2 File menu
3.2.1 New
Shortcut CTRL+N
Starts a new simulation job.
The New command in the File menu creates a new empty job. If there were unsaved changes to
the current job, a dialog box is displayed, asking about saving the changes.
3.2.2 Open
Shortcut: CTRL+O
This command opens an existing set of simulation les. The default selection is dened in a .caj
le.
The Open command in the File menu opens an existing job.
If you enter the le name in the command input eld, the path must be encapsulated in apostro-
phes, for instance:
open "../../Test/000000.caj"
If you select the command from the menu bar, or if no name is specied in the command input
eld, the Open dialog box is displayed, allowing you to specify a path and le name to open.
By default, the le lter is initiated for selecting CASD job header les (type .caj). But you
may also select a geometry le (type co.dat3). CASD will then open all les with the same job
number.
If a geometry is open (in the database), the lter string will be constructed from the project and
geometry numbers. It is not possible to open a job that is not compatible with the open project
and geometry numbers.
If there were unsaved changes to the current job, a dialog box is displayed, asking about saving
the changes.
The geometry le is not read when a geometry is open in the database. If no geometry is open
3.3 Geometry menu 39
in the database, CASD will display the contents of the geometry le in the graphic area after
successful open. The contents of the geometry le can be edited using the Edit File command in
the Geometry menu, see section Geometry menu.
3.2.3 Save
Shortcut: CTRL+S
Saves the current simulation job (i.e. the various les that dene the job).
The Save command in the File menu saves the current job.
3.2.4 Save as
Shortcut: CTRL+SHIFT+S
The Save As command saves the current job under a new (user-dened) name (job number).
3.2.5 Import
Imports certain specications from another simulation job (e.g. grid le, scenario le, etc.).
3.2.6 Exit
Shortcut: CTRL+Q
Exits the CASD software.
3.3 Geometry menu
CASD stores the geometry in a database, and on the geometry le (co-le). The commands in
the Geometry menu in the main window, except the Edit File command, are available when
connected to a database. The Save and Save As commands in the File menu writes the geometry
to the geometry le.
The building blocks in a CASD geometry are instances of objects. Objects can be global or local.
Several geometries can contain instances of the same global object, while a local object only can
be included in the geometry where it was created.
Instances can be grouped under assemblies. Several levels of assemblies can be created. Each
instance and assembly has a transformation matrix. The position, scale, and orientation of an
instance is the result of the matrices on all levels above the instance, in addition to the matrix for
the instance itself.
Each geometry is a member of a project. The project is the top level in the CASD data structures.
A project can own a number of geometries.
Instances and assemblies can be made invisible and visible using the following commands:
CTRL+I Make the selected assembly/instance invisible
CTRL+SHIFT+I Make the selected assembly/instance visible.
40 CASD
Use the Position command in the Geometry menu to change the position of the selected assembly
or instance.
3.3.1 Geometry Database
The rst option on the Geometry menu in CASD opens the Database dialog box.
Figure 3.3: The geometry database window in CASD
In the Database dialog box the user can:
Create a new database, project, geometry, or object.
Connect to or save an existing database.
Open or save existing, projects, geometries, or objects.
Insert instances in a geometry.
Dene new materials or edit existing materials.
3.3.1.1 Geometry tab
On the Geometry tab the user can create, open and manipulate projects and geometries. Projects
can be renamed and deleted, geometries can be renamed, copied and deleted.
3.3.1.2 Objects tab
The New Object button in the Database dialog box opens the Object window.
3.3.1.3 Materials tab
Each object in a CASD database is assigned a material property, and each material is assigned
a colour hue from the 0-360 colour circle. To dene a new material click the New Material
button. The new material is dened by a name and a hue, a value between 0 and 360.
3.3.2 Creating a CASD database
To create a database choose GeometryDatabase or type geometry database. The Geome-
try Database window is shown. Click the Connect button. A le selection dialog box is displayed.
Move to the directory where the database should be created, and write the name of the database,
e.g. my_database.db. Alternatively the database can be created using the command input:
database create my_database.db, which will create a database in the current directory.
If the Geometry Database window is not open, choose GeometryDatabase. Use the New Project
button to create a new project, or the Open Project button to open an existing project.
When a project is opened, a new geometry can be created clicking the New Geometry button, or
open an existing geometry clicking the Open Geometry button.
When an existing geometry is opened, the assembly/instance structure and all objects and
materials used are loaded into the CASD program. If the geometry contains many assem-
blies/instances, you may get an error message indicating that there were not room enough in
the CASD data structures. See section CASD command line options for information on how you
can use command line options to allocate more memory for these structures.
3.3.3 Connecting to a database
To create a new database, see section Creating a CASD database.
To connect to an existing database choose GeometryDatabase or type geometry
database. The Geometry Database window is shown. Click the Connect button. A le se-
lection dialog box is displayed. Select the CASD_DB le on the database directory you want to
connect to.
If you enter the le name in the command input eld, the path must be encapsulated in apostro-
phes, for instance:
database connect "MyCasdDB/CASD_DB"
3.3.4 Creating a new or opening an existing object
You can create a new object clicking the New Object button on the Objects tab in the Geometry
Database window, or open an existing object using the Open button.
When you have completed the New or Open Object command, the object window is displayed.
3.3.5 Selecting a node and a subtree
At any time, a part of the binary tree is selected. It may be a single node, or a subtree containing
several nodes. If a subtree is selected, the top node is referred to as the selected node. In the
postx string, the top node is the rightmost node in the subtree.
The selected subtree is highlighted in the graphic window, and underlined in the message area.
There are two different methods for selecting a subtree.
1. Click MOUSE+LEFT while pointing at a primitive. If several primitives are hit, they are
placed on a stack (list). Only one primitive is selected at a time. Press CTRL+TAB command
to parse this stack.
2. Use the following commands:
42 CASD
CTRL+L Select the previous instance
CTRL+R Select the next instance
3.3.6 Maintaining a CASD database
The dbfutil program is available for creating and maintaining CASD le databases.
Linux:
run9 dbfutil database command [option]
Windows:
dbfutil database command [option]
The usage of this program is described in table Using the the dbfutil program. Make sure that no
other users are connected to the database when you execute this program.
Command Description
create Create database
destroy Destroy database
force Destroy database, override any errors
dellock Delete all locks. Use this command if les in
the database are still locked after a crash in
CASD
restoredep Restore dependencies. For each object in the
database, there is a le containing a list of all
geometries that contain instances of the
object. (Executing the Information command
in the File menu in the Object dialog lists the
contents of this le.) This le is used for
determining if the object can be deleted
when you execute the Delete Object
command in the Database menu. CASD
updates these les when required. But if a
problem should occur for some reason, the
restoredep command might help. It updates
the le mentioned above for all objects in the
database.
restorehead Restore header les. This command resets
the process log le for the database. This le
contains a list of (CASD) processes currently
connected to the database.
list List the content of all table les, e.g. list O
lists all objects:
P List the content of all project table les.
O List the content of all object table les.
M List the content of all material table les.
G List the content of all geometry table les.
L List the content of all local object table
les.
U List the content of all objects-used table
les.
A List the content of all asis table les.
Table 3.3: Using the dbfutil program
We strongly recommend that you make backups of your databases on a regular basis.
3.3.7 Local objects
Local objects consist simply of one box or one cylinder. Use local objects to dene entities like
walls, oors etc. Dene global objects for more complicated things.
The name of a local object must start with an underscore character (_).
The Local Object command in the Geometry menu creates a local object, and one instance of it.
You can of course create several instances of the local object using the Instance command.
The Local Object command has two sub choices, Box and Cylinder. Select the appropriate primi-
tive type.
CASD will rst ask for the material name. Enter the name of an existing material. The material
decides the colour of the object. If you havent dened any materials, use the New Material
command in the Geometry Database window to create one.
CASD will then ask for the sizes and porosities for the primitive. CASD creates an instance of the
object in (0, 0, 0). Use the Position or Translate command to move it to the correct position.
You can use the Properties command to edit material, sizes and porosities for a local object. The
Rename command changes the name of the object.
3.3.8 Global objects
A global object is edited in a separate object window. All the commands described in this chapter
refers to the menus in the object window.
Global objects can have instances in several geometries. The structure within a global object is a
constructive solid geometry (CSG) model where simple solid primitives are combined by means
of Boolean set operations. The primitives and operations are nodes in a binary tree where the
leaves are primitives and the internal nodes are operations.
Boxes, cylinders, ellipsoids, general truncated cones (GTC) and complex polyhedrons (CP8) are
44 CASD
the primitive types supported. The box primitive includes planes as a special case. Available
operation types are union and difference.
Warning:
Only boxes and cylinders should be used in by default, but ellipsoids, general truncated
cones and complex polyhedrons can be used in special cases. These latter primitive types
have the following important limitations:
No subgrid models, thus not contribution to turbulence and drag force
Porosity calculation takes a long time for these primitive types. There should be no
more than 100-200 of these primitives in any given geometry
Figure 3.4: Supported primitive types
A root is a subtree that is not part of another subtree. The object typically contains several roots
during editing. But it must contain only one root when it is saved.
The postx string represents a way of visualising the binary tree dening the object.
The postx string for the open object is displayed in the message area in the object window. The
selected subtree is highlighted.
A material is assigned to each object. The material decides the colour of the object.
Figure 3.5: The binary tree for an objects, and the corresponding postx string
3.3.9 Assembly
Opens a dialog box where the user can specify an assembly of several instances.
3.3.9.1 Adding an assembly
Assemblies represents a way to group the instances in complicated geometries. The Assembly
command in the Geometry menu adds an assembly to the geometry. CASD will ask for the
assembly name. You must enter a name that doesnt exist on the same level, see below. The
assembly is placed in (0, 0, 0). You can transform an assembly in the same way as an instance.
All geometries contains at least one assembly, called the top assembly. That assembly can not be
deleted.
When you create an assembly, it is placed in the geometry structure depending on what was
selected on forehand. If an instance was selected, the new assembly is placed after that instance
under the same assembly. If an assembly was selected, the new assembly is placed under that
assembly.
You can later rename the assembly using the Rename command.
3.3.9.2 Selecting an assembly or instance
The selected instance, or all the instances in the selected assembly, are highlighted in the graphic
window. The name of the selection is written in the message area. The name is concatenated
from the geometry name, the names of all assemblies above the selected assembly/instance, and
the name of the selected assembly/instance. Each level is separated by a period (.). An example
is shown below.
Current Geometry Selection: M24.A1.COOLER-2
46 CASD
Here, M24 is the geometry name, A1 is an assembly. The last part of the string is the lowest
level. In this example, it is an instance, identied by the object name, COOLER, and the instance
number.
There are three different methods for selecting an assembly or instance. The rst method is to
select from the graphic window. To do this, click MOUSE+LEFT while pointing at the instance. If
several instances are hit, they are placed on a stack. Only one instance is selected at a time. In
CASD4 use the CTRL+TAB command to parse this stack.
The second method is to use the following commands:
Select the parent assembly: Press CTRL+U.
Select the child assembly/instance: Press CTRL+D.
Select the assembly/instance name: Press CTRL+F. You are asked to enter the concatenated
name to select.
Select the previous assembly/instance on the same level: Press CTRL+L.
Select the next assembly/instance on the same level: Press CTRL+R.
The third method is to use the List command in the Geometry menu to pop up a list of the
contents of the open geometry. You can use the mouse to select from the list.
3.3.10 Instance
Creates an instance in the current geometry and/or assembly.
3.3.10.1 Adding an instance
To add an instance of an object, use the Instance command in the Geometry menu. CASD will
ask for the object name. You must enter the name of an existing object. The instance is placed in
(0, 0, 0). Use the Position or Translate command to move it to the correct position.
Alternatively the Instance button on the Objects tab in Geometry Datbase dialog can be used.
When a new instance is created, it is placed in the geometry structure depending on what was
selected on forehand. If an instance was selected, the new instance is placed after that instance
under the same assembly. If an assembly was selected, the new instance is placed under that
assembly.
3.3.11 Local object
Creates a local object in the current geometry.
3.3.12 Delete
Deletes either the currently selected instance, local object, or the current assembly (must be
empty).
3.3.13 List
Lists all assemblies and instances in the current geometry, including modied positions.
3.3.14 Duplicate
Duplicates the selected instances in the current geometry.
3.3.15 Position
Denes the position of an instance.
3.3.16 Translate
Translates the current instance.
3.3.17 Rotate
The Rotate command rotates the selected assembly or instance. Note that CASD only accepts axis
parallel geometry. That means that the rotation angle must be a multiples of 90 degrees.
3.3.18 Scale
Scales the current instance by a certain factor in each spatial direction
3.3.19 Matrix
Species the transformation matrix of the current instance. This command is normally not used
directly, but is available for macro reading and writing.
3.3.20 Making an assembly or instance visible or invisible
Shortcut: CTRL+I CTRL+SHIFT+I
This command lets the user make the current instance invisible/visible.
3.3.21 Select
Selects an instance in the current geometry through the following short cut options. See section
Selecting an assembly or instance.
3.3.22 Substitute
Substitutes all instances of one object in the current geometry with instances of another object.
The user species the name of the existing object and new objects.
3.3.23 Properties
Opens a dialog box where the user can observe and edit the properties of a local object.
48 CASD
3.3.24 Rename
Opens a dialog box where the user can rename assemblies or local objects.
3.3.25 Object
Opens the currently selected object.
3.3.26 Edit le
The Edit File command in the Geometry menu makes it possible to edit the geometry le (co le)
for the open job. This command is only available when no geometry is open in the database.
The geometry is saved on the geometry le as one single object, when selecting Save in CASD.
Upon the Edit File command, an object window is therefore shown for editing this object, if
the geometry database is not available, or the user wants to make small modications to the
geometry outside of the database.
Since the object structure lacks the assembly/instance mechanism, editing the geometry le di-
rectly without using the database is recommended only for geometries with a relatively small
number of primitives. For geometries with many primitives, the postx string is long and dif-
cult to manage.
Editing the geometry le for FLACS simulations may be advantageous when the user want to
test the impact of small changes in the geometry on the simulation results. Note that there is no
way to update the database from the geometry le.
3.4 Object window in CASD
The object window opens from the New Object button in the database dialog box.
The object window opens from the database window.
3.4 Object window in CASD 49
Figure 3.6: The object window in CASD
The message area in the object window shows the postx string.
3.4.1 File menu in the Object window
The options on the le menu in the object window are explained below.
3.4.2 Save
If the user is editing an object in the database, the Save command in the File menu saves the object
on the database.
If the user is editing the geometry le, the changes are stored internally in the geometry database,
and will be written to the le upon the Save and Save As commands in the File menu in the main
window. Exiting from the object window without saving, the changes are lost.
The object is stored only if it is consistent, that is if it has only one root. If the object is not
consistent, an error message is displayed, and a Union or Left Difference should be added.
3.4.3 Information
The Information command in the File menu displays a list of all geometries containing instances
of the open object.
50 CASD
3.4.4 Exit
Upon the Exit command, CASD asks about saving the object, and then whether to exit from the
object. If the answer is yes to the last question, the object window is closed.
3.4.5 Edit menu
The options on the edit menu in the object window are explained below.
3.4.5.1 Operations
The Operation command in the Edit menu changes the operation type if the selected node is an
operation.
3.4.5.2 Properties
The Properties command in the Edit menu changes the primitive properties if the selected node is
a primitive.
If you have selected a subtree containing only one type of primitives, the Properties command can
be used for changing one or more parameters for all these primitives.
3.4.5.3 Translate
Use the Translate command to translate the selected assembly or instance a specied distance in
each axis direction.
Use the Translate command in the Edit menu to translate the selected subtree a specied distance
in each axis direction.
3.4.5.4 From To
Use the From To command to translate the subtree so that one specied position, the base point,
is moved to another, the target point. A dialog box for specifying the two positions is displayed.
A circle is displayed in the graphic window, indication the position being edited. CASD keeps a
list of positions used in the object. By pressing CTRL+L or CTRL+R, you can parse this list. The
coordinates in the dialog box is updated.
3.4.5.5 Rotate
The Rotate command rotates the selected subtree. You must specify a base point for the rotation,
and the rotation angle. As for the From To command, you can parse the position list using the
CTRL+L or CTRL+R commands. Note that CASD only accepts axis parallel geometry. That means
that the rotation angle must be a multiple of 90 degrees.
3.4.5.6 Scale
The Scale command is only legal when an instance of a local object consisting of a box is selected.
The Scale command scales the selected subtree. You must specify a base point for the scaling,
and the scaling factor. You can parse the position list using the CTRL+L or CTRL+R commands
3.4.5.7 Delete
The Delete command in the Edit menu deletes the last node in the postx string, the selected
subtree or the current root. Note that if the postx string for the object is consistent, it consists of
only one root. Therefore deleting the current root deletes the entire object.
3.4.5.8 Mark
The Mark command is used in connection with the Substitute command. Select Mark command
to mark the subtree to be substitued with the subtree selected when the Substitute command is
selected.
3.4.5.9 Substitute
The Substitute command in the Geometry menu substitutes all instances of one object with in-
stances of another object. You are asked to specify the two object names.
The Substitute command in the Edit menu substitutes the selected subtree with another subtree.
Use the Mark command to select the rst subtree.
The substitute command implies the following steps. (Let subtree 1 denote the rst subtree and
subtree 2 the second subtree.)
1. Make a copy of subtree 2 and give it a new identity, say subtree 3.
2. Delete subtree 1 from the postx string.
3. Insert subtree 3 in the postx string in the position where subtree 1 was situated.
3.4.5.10 Duplicate
The Duplicate command in the Geometry menu duplicates the selected instance. You are asked
to enter the number of copies, and the distance between each copy in the three axis directions.
Click on Ok, and a dialog box pops up for each copy, allowing you to edit the position.
The Duplicate command in the Edit menu duplicates the selected sub tree. You are asked to enter
the number of copies, and the distance between each copy in the three axis directions. Union
operations are added automatically, so that the resulting sub tree includes the original one.
Creating pipe bundles Start with creating one cylinder with the appropriate diameter, length
and direction. Use the Duplicate command in the Edit menu to duplicate the cylinder in one
direction. Use the same command once more to duplicate the resulting row of cylinders in the
other direction.
If you need to change some parameters for all the cylinders, select the entire pipe bundle sub tree
and use the Properties command. If you want to change the distances between the cylinders, this
can be done by scaling the entire sub tree. Afterwards you can use the Properties command to
reset the cylinder diameters and lengths.
52 CASD
3.4.5.11 Material
The Material command in the Edit menu edits the material name for the object. You must enter
an existing material name.
3.4.5.12 Matrix
The Matrix command was introduced to make it simple to create and run macros for creating
geometries.
Warning:
This command should normally not be used in interactive mode.
3.4.6 Add menu in the Object window
The options on the add menu in the object window are explained below.
3.4.6.1 Box
The Box command in the Add menu adds a box at the end of the postx string. A dialog box for
dening the box parameters is displayed.
3.4.6.2 Cylinder
The Cylinder command adds a cylinder at the end of the postx string. A dialog box for dening
the box parameters is displayed.
3.4.6.3 Ellipsoid
The Ellipsoid command adds an ellipsoid at the end of the postx string. A dialog box for dening
the ellipsoid parameters is displayed. Note warning about the use of ellipsoid.
3.4.6.4 CP8
The CP8 command adds a complex polyhedron at the end of the postx string. A dialog box
for dening the complex polyhedron parameters is displayed. Note warning about the use of
complex polyhedron.
Figure 3.7: Denition of a complex polyhedron
3.4.6.5 GTC
The GTC command adds a general truncated cone at the end of the postx string. A dialog box
for dening the general truncated cone parameters is displayed. Note warning about the use of
general truncated cone.
Figure 3.8: Denition of a general truncated cone
3.4.6.6 Union
The Union command adds an union operation at the end of the postx string. This command is
only legal if the object contains at least two roots which can be connected by the operation.
3.4.6.7 Left Difference
The Left Difference command adds a difference operation at the end of the postx string. This
command is only legal if the object contains at least two roots which can be connected by the
operation. If using CASD4, use the Shade command in the View menu, to see the result of the
operation. Note that the right hand side operator of a difference operation must be a primitive.
3.4.6.8 Copy
The Copy command adds a copy of the selected sub tree at the end of the postx string.
54 CASD
3.4.6.9 Object
The Object command adds a copy of a specied object at the end of the postx string.
3.4.7 Select menu in the Object window
The options on select le menu in the object window are explained below.
3.4.7.1 Previous
Shortcut: CTRL+L
Selects the previous primitive or subtree
3.4.7.2 Next
Shortcut: CTRL+R
Selects the next primitive or subtree
3.4.7.3 Stack
Shortcut: CTRL+TAB
This command will parse (cycle through) the list of selected primitives or subtrees if more than
one is selected.
3.4.8 View menu in the Object window
The options on the view menu in the object window are explained below.
3.4.8.1 Print
The Print menu allows exporting a screenshot of the CASD window into different formats:
Postscript
RGB
IV
3.4.8.2 Examiner Viewer and Fly viewer
The default and most widely used viewer is the Examiner viewer. The Fly viewer can be used to
y through the geometry.
3.4.8.3 The XY, XZ and YZ views
The option XY View and XZ View display a projection of the geometry in the XY and XZ planes
respectively. The options YZ East View and YZ West View display a projection of the geometry
in the YZ plane along the positive and negative Y-axis respectively.
3.4.9 3D View
The 3D View option displays a default 3D view of the geometry.
3.4.9.1 Axis
The Axis option turns axis display on and off.
3.4.9.2 Maximize
The option Maximize maximizes the visible window to display the entire geometry and grid.
3.4.9.3 Grid Display
Three different options are available in the Grid Display menu:
Off: The grid is not displayed. Only the geometry would be displayed.
Working Direction: The grid would be displayed in the working direction only.
All Directions: The grid would be displayed in the three directions.
3.4.9.4 Annotation
The options in this menu are currently not used.
3.4.9.5 Draw Style
Different options are available in this menu:
Off: The geometry will not be displayed.
Wireframe: Only the edges of the objects that compose the geometry would be displayed.
Filled: Surfaces of the objects that compose the geometry would be displayed.
Scenario Wireframe: Only the edges of scenario objects (for example, a fuel region) would
be displayed.
Scenario Filled: Surfaces of scenario objects would be displayed.
3.4.9.6 LOD and Properties
The LOD (Level Of Details) and properties menus control the details of the geometry displayed.
3.4.10 Macro menu in the Object window
The options on the macro menu in the object window are explained below. for more infromation
about CASD macros see section Macro menu.
3.4.10.1 Run
Read a macro le dening a single object.
56 CASD
3.4.10.2 Record
Writes all commands given to a user specied macro le.
3.4.10.3 Write Object
Writes a macro le containing the current object.
3.5 Grid menu
The simulation volume is divided into a set of control volumes by three sets of grid planes, one
in each axis direction.
There is always a current grid working direction, and a selected region of grid planes in this
direction. The current working direction is shown in the message area. The lines indicating the
selected region is highlighted.
3.5.1 Simulation volume
The Simulation Volume command lets you change the simulation volume extent in all three di-
rections. If you increase the volume, the original grid planes are kept, but one additional plane is
added in each direction. If you decrease the volume, planes outside the new volume are deleted,
and new planes are created on the volume borders.
3.5.2 Direction
The Direction command changes the working direction. Legal input is x, y or z. The Grid menu
commands Region, Add, Position, Move, Delete, Smooth, Stretch and List affects the grid planes
in the working direction.
3.5.3 Region
The Region command substitutes the selected grid planes by a new set of grid planes. CASD asks
you to enter the new number of control volumes in the region.
3.5.4 Add
The Add command adds a new grid plane in the working direction. You are asked to enter the
coordinate value for the new plane.
3.5.5 Position
The Position command lets you edit the position for the selected grid plane.
3.5 Grid menu 57
3.5.6 Move
The Move command moves the selected grid planes a specied distance.
3.5.7 Delete
The Delete command deletes the selected grid planes.
3.5.8 Smooth
The Smooth command substitutes the selected grid planes by a new set of grid planes.
For the Smooth command, the sizes of the control volumes at each end of the region is kept
unchanged. The sizes of the control volumes between them varies gradually.
This function is typically used when rening the grid around a leak.
3.5.9 Stretch
The Stretch commands substitutes the selected grid planes by a new set of grid planes. This is
particularly useful when stretching the grid towards the outer boundaries.
The Stretch command has two sub-choices:
Negative Direction (typically used at the boundaries at the negative end of the axis)
Positive Direction (typically used at the boundaries at the positive end of the axis)
You must enter the size of the control volume at one end of the region, default is the current size.
Then you must enter a factor by which the sizes of the control volumes in the specied direction
increases/decreases.
Attention:
Note that stretching of the grid should be avoided in areas of the simulation domain where
the main combustion is happening. The ame model in FLACS has been validated for cubical
control volumes, thus the user should not stretch the grid in areas where accurate results are
required. It is however good practice to stretch the grid towards the boundaries, to concerve
simulation time and computer memory.
3.5.10 Information
The Information command displays status information about the dened grid, while the List
command lists the grid coordinates in the working direction.
3.5.11 List
The Information command displays status information about the dened grid, while the List
command lists the grid coordinates in the working direction.
58 CASD
3.5.12 Display
The Display command turns grid display off, displays the grid in the working direction only, or
displays the grid in all three directions.
3.5.13 Select
The selected region of grid planes is limited by two planes, the lower and upper limit. If only
one plane is selected, the upper and lower limit is the same grid plane. Grid planes are selected
using the following commands:
Lower boundary
Select the next grid plane: CTRL+RIGHT
Select the previous grid plane: CTRL+LEFT
Upper boundary
Select the next grid plane: CTRL+UP
Select the previous grid plane: CTRL+DOWN
3.5.14 Grid-related operations
3.5.15 Importing the grid from another job
Use the Import command in the File menu to import the grid from another job.
If you enter the grid le name in the command input eld, the path must be encapsulated in
apostrophes, as described in section . If you select the command from the menu bar, or if no
name is specied in the command input eld, the Import dialog box is displayed, allowing you
to specify the path and le name for a grid le. You will be asked to verify that the current grid
is overwritten by the grid from the specied le.
3.5.16 Saving the grid
The Save and Save As commands in the File menu saves the grid, together with the rest of the
job data. If the grid is changed, you will need to recalculate the porosities.
3.5.17 Grid-related utilities
FLACS is deleivered with a command line tool for creating an manipulating the grid. This tool
can be used to quickly edit or get information about the grid. Please see section gm for further
information.
3.5.18 Grid guidelines
The grid resolution should be chosen to obtain a simulation result within an acceptable time
frame. In most cases a reasonably good result can be obtained on a coarse grid within less than
one hour (in some cases 5 minutes), and high quality results can normally be generated in a few
hours (or at least over the night).
3.5 Grid menu 59
Never start a project with a calculation on a grid that will be running for days. If such long sim-
ulation times are necessary, always start simulating on a much coarser grid [even if this violates
guidelines] to check that the scenario and setup are OK.
The user should keep the position of the grid lines in mind while dening the geometry. The
geometry details such as walls and decks should be adjusted to the closest grid line when in-
putted. Thereby the user keeps track of the positioning instead of having the geometry moved in
an unwanted direction by the porosity calculation program.
In the grid embedding process, it is highly recommended to use GridInformation in Casd to
check different aspects of the grid. Grid sensitivity tests are also recommended.
3.5.18.1 Gas explosion simulations
Attention:
The user should always apply cubical grid cells in the combustion region. Deviations from
this will give different ame propagation and pressures, and the validation work done is
no longer valid. Deviations of the order 10% in aspect ratio is OK, deviations by a factor
of 2 in aspect ratio is not OK. If one chooses not to follow this guideline, the results can be
somewhat improved by setting a xed control volume size for the time stepping routine (see
section The SETUP namelist, example TIME_STEPPING=" STRICT:L_FIX=1.0" ).
Channels and conned vessels and rooms (lled with gas from wall to wall) must always be
resolved by a minimum of 5-6 grid cells in smallest direction if ame acceleration shall be modelled.
This also applies for pipes where ame acceleration along the pipe is of interest.
A pipe connection from one vessel to the next may have less grid cells across the diameter (but
preferably more than 1 CV) if only ame transport by pressure difference and not ame accelera-
tion along the pipe shall be modelled. Increase the inner diameter of angles and bends somewhat
when modelling pipes with cylinder minus primitives. Remember that one full grid cell is re-
quired inside the solid walls around " minus primitive holes" to ensure that the walls will not be
leaking.
Unconned gas clouds as well as partially lled clouds should have a minimum of 13 grid cells
across the cloud if both sides are unconned, and a minimum of 10 grid cells in directions where
cloud meets connement on one side (example vertical direction for dense gas cloud in chemical
plant).
It is not recommended to use non-cubical grids for explosion simulations. As they are often used
for dispersion simulations, the dispersion simulation results should be dumped, thereafter the
rdle utility program should be used to transfer the results from the dispersion grid to a grid
better suited for explosions, see example below:
> run9 rdfile rd111111.n001 rd222222.n001
Here 111111 is the dispersion calculation job number and 222222 is copy of the job, in which the
grid has been modied to follow explosion grid embedding guidelines. The grid of job 222222
must be completely inside the grid of 111111.
The grid can be stretched outside the combustion region in directions where pressure recordings
are not of interest. In directions where pressure wave propagation is of interest, one should not
stretch the grid because this will reduce the sharpness and quality of the pressures.
A proper distance to external boundaries is important. At least 5-10 grid cells from vent opening
to external boundary should be used in situations where the external explosion is not important
(small vent area or strong turbulence inside vessel).
60 CASD
In low-congested vessels with signicant vent opening, external explosions and reversing of ow
may give a strong feedback into the vessel in connection with venting. To pick up this properly,
the distance to the external boundaries should be signicant (maybe 3-4 times the length of the
vessel).
EULER boundary will reect negative pressures, which can destroy results when simulating far
eld pressure propagation. In this case PLANE_WAVE may be recommended (but then the bound-
aries must be far away so no products from combustion reaches the boundaries). For unconned
situations, try to have the same distance to boundaries in all directions (use stretching in direc-
tions with less interest in results).
3.5.18.2 Blast wave propagation in the far eld
Maximum control volume size should be 10% of gas cloud diameter:
max CV = 0.1 (gascloudvolume)
1/3
(3.1)
The grid cells must be approximately cubical in explosion simulations and the cell size should be
maintained in directions of interest for blast propagation simulations. For vessel burst the same
guidelines applies. If the pressure is much higher than 10 barg, somewhat larger grid cells than
this criterion can be acceptable. Remember to use PLANE_WAVE boundary condition and proper
distance to the boundaries.
3.5.18.3 Dispersion simulations near eld
Calculations of ammable gas requires grid renement near the leak. The area of the expanded
jet (at ambient pressure) must be resolved by one grid cell (A
CV
< 2 A
jet
) except for low-
momentum releases of highly buoyant gases such as hydrogen where the guideline (A
CV
<
1.25 A
jet
) should be followed. In most cases, a grid renement near the leak helps in keeping
moderate calculation times while getting acceptable results.
Grid renement guidelines for efcient simulation of high velocity jets recommend only rene-
ment across jet direction (not along). CFLC should be increased by renement factor (CFLV
should not be changed). Smoothing from ne grid cells near jet to normal grid cells further
away from jet is recommended.
If the jet is not along the axes, is impinging or has a low momentum with positive/negative
buoyancy, then extending the rened region of the grid in one or more directions may be re-
quired. Renement along the jet may then also be required.
If only far eld concentrations are of interest, the renement near the leak may not be needed.
Quicker calculations and less stability problems will be seen without the renement.
For dense gas calculations, it may be a good idea to use a ner resolution in the vertical direction
near the ground than in the other two directions. Increase CFLC by this renement factor.
Outside the main area of interest, further stretching to the boundaries is recommended to min-
imize the inuence of the boundaries. If in doubt whether the distance to the boundaries in-
uences your results, increase the distance further to check the sensitivity. The general recom-
mended procedure for setting up the grid is:
1. Cover the computational domain with a uniform grid
2. Rene the grid in the near region of a jet (perpendicular to the jet axis)
3. Stretch the grid outside the main region towards the boundaries
3.6 Porosities menu 61
3.6 Porosities menu
The Porosities menu supplies commands for calculating and verifying porosities.
Calculate This command starts the porosity calculation program, Porcalc
Verify This command starts Flowvis for porosity verication
3.6.1 Calculate
The Calculate command starts the porosity calculation program, Porcalc.
The version of Porcalc can be selected on the Preferences dialog in CASD. The default version
of Porcalc uses a resolution of 64, i.e. the control volume is divided into 64 parts when calculat-
ing the porosities. It can be useful (and necessary) to use the porcalc_16 version if the porosity
calculations are very slow, e.g. if the geometry contains slanted primitives.
If you are working with a multiblock simulation, only the porosities for the selected block are
calculated. Upon the Save and Save As command in the File menu, CASD will warn you about
blocks where the grid and/or geometry is changed since the last time you calculated the porosi-
ties.
Porcalc can also be started from the command line or the Runmanager.
3.6.2 Verify
The Verify command lets you view the calculated porosities. Flowvis is started for porosity ver-
ication. A 2D Cut Plane plot for the appropriate job is automatically created with volume and
area porosities shown. The Plot Domain dialog box pops up. This dialog box lets you select other
planes as wished.
By clicking inside a control volume, you can verify the porosity values for that volume.
If you are working with a multiblock simulation, only the porosities for the selected block are
veried.
3.7 Scenario menu
The purpose of this chapter is to outline how to edit the scenario sections. A short description of
each section and the impact on the FLACS simulations will be given. The scenario-le (cs-le) is
an ASCII le and it is easy to edit manually as well as using CASD.
The items in the Scenario menu are read from a scenario denition le. There is one default sce-
nario denition le, but several other choices can be activated by changing the scenario template.
This can be done using the Preferences command in Options menu. For instance, the default+1
template activates several advanced options, especially in the Simulation and Output Control sec-
tion. The current job should then be saved, closed, and reopened in order for the new options to
be available.
When a section has been selected, the items in the section are displayed as a list. The method for
editing a scenario section depends on the type of section.
Sections such as INITIAL_CONDITIONS and IGNITION contain a list of parameters, each with
one or more values. A parameter is selected for editing by clicking on it, or by typing the param-
eter name in the command input eld in the main window.
62 CASD
Sections such as SINGLE_FIELD_SCALAR_TIME_OUTPUT and SINGLE_FIELD_3D_OUTPUT
contain a list of items, which can be selected. For some sections, each item has a subsection.
For SINGLE_FIELD_SCALAR_TIME_OUTPUT the user must select from a list of monitor points
for each variable selected. An item is selected by clicking on it, or by typing the item name in the
command input eld in the main window. To select several items using the mouse apply CTRL
or SHIFT keys. If a selected option shall be deselected without selecting an alternative option,
deselect by clicking while pressing the CTRL key. Typing an item name in the command input
eld in the main window selects/deselects the item if previously not selected/selected.
Importing the scenario from another job
Use the Import command in the File menu to import the scenario from another job. The user will
be asked to verify that the current scenario is overwritten by the scenario from the specied le.
Saving the scenario
The Save and Save As commands in the File menu saves the scenario (together with the rest of
the job data).
3.7.1 Monitor points
Monitor points are user dened locations in the simulation domain where one or more variables
are to be monitored during the simulation. The maximum number of monitor points allowed is
currently 1000. Positions for monitor points are given in the unit selected in Options Preferences
(normally meter)
When the user has dened all the desired monitor points, he/she may specify a list of variables
to be monitored and the relevant monitor numbers for each variable (see the next subsection).
FLACS identies the 8 surrounding control volume centres and writes an interpolated value of
the specied variables to the scalar-time output le (nodes on other side of wall or zero porosity
will not be used when interpolating).
Attention:
The user should avoid putting monitor points exactly on a grid line or within fractions of a
grid cell size from a wall. Usually it is best to enter monitor points according to the grid, not
according to the geometry. E.g. if monitor points are placed on each side close to a solid wall
they may not necessarily be in two different control volumes (as was the intention). During
the porosity calculation the wall will be adjusted to the nearest control volume face (grid
line) and might therefore move to the wrong side of a monitor point! It may also be a good
idea to ensure that none of the monitor points are inside fully blocked control volumes.
3.7 Scenario menu 63
Figure 3.9: Specication of Monitor Points
64 CASD
Figure 3.10: How to position the monitor points
3.7.2 Single eld scalar time output
As indicated in the previous sub-section, the user must dene all the monitor points and panels
before he/she can specify the list of output variables. For each output variable, the user may enter
one or more numbers indicating the monitor point number(s), or panel number if it is a panel
averaged output variable, for which you want this variable to be measured. An example from
a scenario-le section is shown below [NP identies variable P (pressure) whereas NPP identies
variable PP (panel pressure) which gives average pressure load across a surface described by
panel]:
SINGLE_FIELD_SCALAR_TIME_OUTPUT
NP 1 2 3 4 5
NPP 1 2 3
This shows that pressure is reported for monitor points 1-5 and panel pressure is reported for
panels 1-3 (The denition of measurement panels is described in the section on Pressure Relief
Panels).
To select more than 1 monitor point use CTRL+MOUSE+LEFT or CTRL+SHIFT+MOUSE+LEFT, or
simply drag the mouse over the list of monitor points while pressing LEFT. If a variable and mon-
itor points are selected by a mistake, these can be deselected by using the CTRL+MOUSE+LEFT to
deselect the last monitor point of a variable.
The most commonly used monitor point variables for explosion simulations are pressure (P),
dynamic pressure (DRAG), panel pressure (PP), pressure impulse (PIMP) and sometimes ow
velocity (UVW). For dispersion calculations volume gas concentration (FMOLE) and ow velocity
(UVW) are among the most commonly used variables.
The monitor point results for job 010100 are written to the r1010100.dat3 le which can be read
by Flowvis. If ASCII data is required, the r1le-utility program can be used (see the section on
FLACS utilities).
The rst section on the scenario le denes the names and identiers for all the variables which
may be selected for output from FLACS. In order to select alternative units for certain variables
(e.g. psi or kPa for pressure or K for temperature) the scenario-le should be manually edited.
The variable pressure is described in the top of the scenario-le as follows:
NP "P " 1 "(barg) " N
"Pressure"
An output in a different unit (psig) can simply be obtained by editing this as follows:
NP "P " 1 "(psig) " N
"Pressure"
Similar changes can be made for other variables and other units. Please note that the units of
time must always be seconds (however, it is possible to change them to ms in Flowvis).
A complete list of all variables available can be found in section Output variables in FLACS.
3.7.3 Single eld 3D output
This is an output facility in FLACS which enables the user to generate plots of the spatial distri-
bution of the variables (e.g. cut plane plots and volume plots) at different moments in time. The
user needs to specify the list of desired variables for SINGLE_FIELD_3D_OUTPUT, an example
from scenario-le is shown below (here P, PROD and VVEC are selected in CASD):
SINGLE_FIELD_3D_OUTPUT
NP
NPROD
NVVEC
NU
NV
NW
To select more than one variable press the CTRL-key selecting variable 2 and 3 etc.
Please observe that when velocity vectors are selected for output (VVEC), directional velocities
U, V and W will automatically be selected. These should not be deselected while VVEC remains
selected (this is possible e.g. by manual editing of scenario-le), if this is done very strange
results will be seen in Flowvis as result le is not consistent with scenario le used by Flowvis to
interpret results.
The user should be aware that this type of output may give very large les (r3-les). If the user
wants to save disk space, the number of output variables and the number of time instants for
output must be limited. The r3-les are binary, if ASCII data is wanted the utility program r3le
can be applied.
The most commonly used variables for explosion modelling are pressure (P), ame (PROD), some-
times gas volume concentration (FMOLE), dynamic pressure (DRAG), maximum pressure (PMAX)
66 CASD
and velocity vectors (VVEC). For dispersion FMOLE and VVEC will be the most common variables
to report.
In certain situations the variable PMAX may not be written to even if specied (zero results ev-
erywhere). This may happen if the simulation job requires more RAM than allowed (e.g. due to
self-dened limits in Linux)
To specify output times DTPLOT and NPLOT will normally be used (see section Simulation and
output control), and sometimes also cc-les (see section Runtime simulation control le). To
create animations it is normally recommended to have plots at 100-200 different moments in
time. When creating results les to be used for animations a combination of DTPLOT and NPLOT
is usually recommended for explosions, for dispersion only DTPLOT should be used.
Units of output variables can be changed using the technique described in the previous section.
A complete list of all variables available can be found in section Output variables in FLACS.
3.7.4 Simulation and output control
This section describes parameters for general simulation and output control. The default
scenario-le setup which is suitable for gas explosion simulations is listed below:
TMAX -1
LAST -1
CFLC 5
CFLV 0.5
SCALE 1
MODD 1
NPLOT -1
DTPLOT -1
GRID "CARTESIAN"
WALLF 1
HEAT_SWITCH 0
In addition the following entries are available in the default+1 template:
TSTART -1
TMIN -1
LOAD -1
STEP ""
KEYS ""
For dispersion simulations, higher value of CFL numbers (20 and 2) are recommended. NPLOT
should be -1 (it has no meaning) and a nite value of DTPLOT should be given.
A detailed description of each parameter is given below.
3.7.4.1 TMAX
This is the maximum time interval (seconds) that the simulation will last. For explosion simu-
lations, default value in CASD can typically be used. The default value set by CASD is -1, this
means there is no maximum time specied, and automatic stop criteria will be applied. Auto-
matic stop criteria will usually work well for explosion calculations. The simulation stops 20 %
after either
90 % of fuel is burnt or pushed out of the domain
50 % of fuel is burnt or pushed out of the domain (and average pressure becomes negative)
Automatic stop criteria can not be used for dispersion, and will also be less useful for some
special situations. If pressure If far-eld blast pressures are of interest, the automatic stop criteria
should not be used as it may stop the simulation before blast waves have hit their target.
In a gas dispersion simulation TMAX will typically be from a few seconds to a few minutes. Sim-
ulation results are not affected by variation of this parameter.
3.7.4.2 LAST
This is the maximum number of iterations allowed for the simulation. The default value set by
CASD is -1 that means that there is no limitation to the number of iterations. This value may be
changed if additional control of when to stop a simulation is required but this is generally not
used.
3.7.4.3 CFLC
This is a Courant-Friedrich-Levy number based on sound velocity. The value of CFLC connects
simulation time step length to control volume dimension through signal propagation velocity (in
this case the velocity of sound), in the following way:
Each time step length is chosen so that sound waves may propagate only a limited distance,
which is the average control volume length multiplied by the value of CFLC. The default value
set by CASDis 5.0. The time step limit imposed by this criterion is normally dominant in the early
phase of an explosion, when ow velocities and combustion rate are still low (see also CFLV).
Simulation results may change with this parameter. Therefore, it is not recommended to change
this value for explosion simulations as the validation work is nullied. If convergence problems
occur (a rare occurrence), CFLC may be reduced by a factor of 2. However, other problems should
rst be ruled out. Extreme changes of the CFL numbers (i.e. by an order of magnitude) are never
recommended for normal simulations.
Note that for multi-block simulations a maximum CFLC=0.5 should be used for the BLAST
blocks. It is recommended to use CFLC=0.2 in the BLAST blocks in order to ensure numerical
stability and good representation of the blast wave.
For dispersion simulations, a default value of 20 is normally recommended. This can be increased
by the grid renement factor (if applicable) i.e. if the grid is rened near the leak by a factor of
5, a CFLC number of 205 = 100 may be used (a lower value should be used in case of stability
problems).
For far-eld blast simulations, this should be combined with STEP="KEEP_LOW" in order to
keep the time step short even after the explosion is outside the "core" area (more information is
given below).
3.7.4.4 CFLV
This is a Courant-Friedrich-Levy number based on uid ow velocity. The value of CFLV con-
nects simulation time step length to control volume dimension through signal propagation ve-
locity (in this case the uid ow velocity), in the following way:
Each time step length is chosen so that the uid may propagate only a limited distance, which
is the average control volume size multiplied by the Courant number. The default value set by
CASD is 0.5. The time step limit imposed by this criterion is normally dominant in the later phase
of an explosion, when ow velocities and combustion rate are high (see also CFLC).
68 CASD
Simulation results may change with this parameter. Therefore, it is not recommended to change
this value for explosion simulations as the validation work is nullied. If convergence prob-
lems occur (rare), CFLV may be reduced by a factor of 2. However, other problems should rst
be ruled out. Extreme changes of the CFL numbers (i.e. by an order of magnitude) are never
recommended for normal simulations.
For dispersion simulations, a default value of 2 is normally recommended (a lower value should
be used in case of stability problems).
3.7.4.5 SCALE
This parameter is used to scale all linear dimensions in a scenario. This means that a 10 m long
explosion vessel is calculated as being 20 m long if SCALE is set to 2.0. Positions and sizes of
equipment, gas cloud, ignition region, monitor points, panels etc. are scaled accordingly. The
CASD default value is 1.0. This parameter will inuence simulation results, typically explosion
pressures increase with increasing scale.
This is practically never used for realistic geometries.
3.7.4.6 MODD
This is a parameter that may be used to determine how often data for scalar-time plots are written
to the results le during a simulation: data are namely stored every MODD timesteps. CASD
default is set to 1. This variable does not inuence simulation results, only the amount of data
stored.
This is normally not used in explosion simulations, but a value of MODD=10 (or higher) may be
used for long dispersion simulations.
3.7.4.7 NPLOT
This is a parameter that may be used to determine how often data for eld plots are written to
le during a simulation: data are namely stored at given fuel levels where NPLOT
is the number of fuel levels equally spaced between zero and a maximum. Fuel level is dened
as the current total mass of fuel divided by the initial total mass of fuel. This output mechanism
is not active in the case of a gas dispersion simulation (leaks are specied). This variable does not
inuence simulation results, only the amount of data stored.
3.7.4.8 DTPLOT
This is the time interval (in seconds) for eld output. This is useful in gas dispersion simulations
and also in gas explosion simulations when frequent output is required. Note that the eld output
le will become very large if DTPLOT is set small. This variable does not inuence simulation
results, only the amount of data which is stored.
3.7.4.9 WALLF
This is a control switch that species the use of wall-functions in FLACS. Wall-functions are used
to resolve the effect of momentum and thermal boundary layers on the momentum and energy
equations in near wall regions. The following choices are available:
0 = OFF
1 = ON
The CASD default value of WALLF is set to 1. This parameter will inuence the simulation results.
When WALLF equals 1 wall, functions are employed based on theory explained in [Sand and
Bakke, 1989]. A slightly modied version of this wall-functions procedure is employed when
WALLF equals 2. However, no validation work for WALLF=2 is available.
The default value of WALLF=1 should always be used.
3.7.4.10 HEAT_SWITCH
This parameter is meant to control the activation of thermal attributes on objects in Flacs.
Default choice is zero (0) as large scale explosions is not much inuenced by heat loss. Choos-
ing one (1) will let walls and objects have background temperature, and if gas temperature
changes, some heat transfer into or out of gas will take place. This is useful for small-scale
conned explosions and dispersion with important heat effects. This can be combined with
KEYS="RADIATE=04" in order to activate radiation heat losses (see below).
If heat switch is activated all solid surfaces will now be initialized with ambient temperature (in
previous versions of FLACS a cs [jobno].HEAT le had to be written). Further heat objects can be
specied at different temperatures, see manual.
Two models for radiation heat loss can be activated. One simplied model can be activated using
a "cs \e [jobno].RAD" le, see description in previous FLACS manual. Alternatively a 6-ux
model can be activated with the KEYS-string in the scenario-le or setup-le:
KEYS = " RADIATE=04"
This model will calculate gas heat loss (and absorption/scatter) from radiation as well as radi-
ation from hot objects around. Walls absorb 100% of the incoming radiation and emit radiation
based on its own temperature. Symmetry planes will reect 100%.
3.7.4.11 TSTART
Remarks:
Only available in the default+1 template.
This variable makes it possible to specify a start time for simulation (-1 means not applied =>
default is zero or time of dump-le). If dump-le exists, one can still adjust TSTART, but the
previous history of the simulation can then not be kept using the KEEP_OUTPUT in a setup-le.
3.7.4.12 TMIN
Remarks:
This variable makes it possible to dene a minimum time for simulation. Automatic stop criteria
will not activate before TMIN has been reached (-1 means not applied). This can be useful in blast
simulations.
70 CASD
3.7.4.13 LOAD
Remarks:
This makes it possible to load a dump le directly from CASD (instead of using the cc-le). The
number of dump le should be specied here.
3.7.4.14 STEP
Remarks:
It is possibility to give time stepping input (ref. Manual, options for TIME_STEPPING). The
options include KEEP_LOW that is recommended for calculations of far-eld blast propagation,
and effectively prevents time step from growing when explosion calms down. Another option is
STRICT:L_FIX=1.00
that instructs the simulation to use 1m grid size as basis for timestep (and ignore local grid re-
nement). This can be used instead of increasing the CFLC number as a result of grid renement.
3.7.4.15 KEYS
Remarks:
This provides the option for entering setup-le options directly in CASD, e.g.
RADIATE=04
for enabling radiation heat loss.
3.7.5 Boundary conditions
In the Boundary condition menu, the user must specic boundary conditions for the outer bound-
aries of the simulation domain. The lower boundaries in X- Y- and Z-direction are denoted by
XLO, YLO and ZLO respectively, and the upper boundaries likewise by XHI, YHI, ZHI. Recom-
mended boundary conditions are as follows:
EULER: Euler equations
NOZZLE: Nozzle formulation
PLANE_WAVE: Plane wave condition
WIND: Wind inow or outow
SYMMETRY boundary condition is generally not used for realistic geometries. In addition,
it is also possible to choose EQCHAR, and BERNOULLI, but these boundary conditions are not
recommended.
Remarks:
For most explosion simulations, EULER can and should be used.
For wind and dispersion simulations, NOZZLE boundary condition (similar to EULER)
is more robust.
PLANE WAVE boundary condition is recommended for explosion in low connement
and for far eld blast propagation. Boundary must be extended far outside the explo-
sion (Flames should not reach boundary).
Different boundaries do not need to have the same condition.
Boundary conditions try to model what happens beyond the boundary. Except for solid
walls, this is not straightforward. Sometimes the boundary condition will disturb or
even destroy a simulation. Then the user should:
1. Ensure that the chosen boundary conditions are those that t best to the problem.
2. Consider to increase the Simulation volume and move the boundaries to regions
where less steep gradients will cross the boundaries.
The details of various boundary conditions are given below:
3.7.5.1 Euler
The inviscid ow equations (Euler equations) are discretized for a boundary element. This means
that the momentum and continuity equations are solved on the boundary in the case of outow.
The ambient pressure is used as the pressure outside the boundary. A nozzle formulation is used
in the case of inow or sonic outow.
Warning:
EULER boundary condition may give too low explosion pressures in unconned situations.
In such cases, the Simulation volume should be extended and the Plane wave boundary
condition should be applied.
3.7.5.2 Nozzle
A nozzle formulation is used for both sub-sonic inow and outow and sonic outow. This
condition is suitable for porous areas with small sharp edged holes or grids (e.g. louvres and
gratings). A discharge coefcient is calculated from the area porosity and a drag coefcient.
NOZZLE condition has shown to give a bit higher explosion pressures than EULER, but it is more
robust.
Warning:
NOZZLE boundary condition may give too low explosion pressures in unconned situations.
In such cases, the Simulation volume be extended and the Plane wave boundary condition
should be applied.
3.7.5.3 Plane wave
This boundary condition was designed to reduce the reection of the pressure waves at open
boundaries which occurs when using EULER or NOZZLE. The pressure wave reection is caused
by setting a xed pressure at the boundary. PLANE_WAVE boundary condition extrapolates the
pressure in such a way that reections are almost eliminated for outgoing waves.
72 CASD
Attention:
The pressure might stabilize at a slightly elevated level after an explosion. For low conne-
ment scenarios it is recommended to use Plane wave boundary condition and to extend the
domain such that the total volume is about 100 times larger than the volume of the initial gas
cloud.
Warning:
In semi-conned situations where the boundaries are close to the vents, PLANE_WAVE should
not be applied.
3.7.5.4 Wind
WIND boundary condition models an external wind eld. Velocity and turbulence proles are
specied at the wind boundaries, either by setting some turbulence parameters manually or by
choosing one of the atmospheric stability classes, see Pasquill class . WIND boundary conditions
are particularly applicable to dispersion scenarios. It is possible to apply WIND on both inow
and outow boundaries and on boundaries where the ow is parallel to the boundary.
Warning:
In cases where a generated internal ow has a strong impact on the boundary ow, e.g. gas
explosions, WIND should not be used.
Figure 3.11: Specication of Wind boundary condition
Wind speed WIND_SPEED, U
0
, is the velocity on the boundary at a given Reference height. It is
possible to set WIND_SPEED to positive, zero or negative values, but GexCon recommends to set
a postive value and use the Wind direction parameter to specify the direction of the wind. In case
of no wind, the user should consider to use another boundary conditions. A uniform velocity
prole is obtained by setting the Reference height equal to zero. Then the total volumetric ux
over the boundary is as follows:
V = U
0
n
A
n
2
n
(3.2)
Wind direction WIND_DIRECTION is a vector and each component may be given a positive,
zero or negative value. The sign of this parameter determines the ow direction. A positive value
means wind ow in positive direction, that is inow over the lower boundaries and outow at
the upper boundaries. Wind at an angle different from axis directions may be specied using the
WIND_DIRECTION vector.
Relative turbulence intensity RELATIVE_TURBULENCE_INTENSITY, I
T
, is the ratio between
the isotropic uctuating velocity, u
, and the mean ow velocity U

0
:
I
T
=
u
U
0
(3.3)
I
T
will typically have a value in the range 0.0 to 0.1. This parameter is used to calculate the value
for turbulent kinetic energy, k = 3/2u
2
, at the boundary.
Attention:
It is not necessary to set RELATIVE_TURBULENCE_INTENSITY for inow boundaries
when a Pasquill class is specied. When a Pasquill class is set, FLACS will automatically
create proles for velocities and turbulence parameters at the boundary.
Turbulence length scale TURBULENCE_LENGTH_SCALE,
LT
is a typical length scale on the
boundary. It is used to calculate the rate of dissipation of turbulent kinetic energy, at the bound-
ary:
=
C
k
3/2
LT
(3.4)
For internal ows, the length scale should be about half of the hydraulic diameter. It is not
necessary to give a turbulent length scale when a Pasquill class is set.
Wind buildup time WIND_BUILDUP_TIME is the time velocities on the boundaries used to rise
from zero to WIND_SPEED. A value for WIND_BUILDUP_TIME larger than zero gives a smooth
start of the simulation. GexCon recommends to use WIND_BUILDUP_TIME and to start eventual
leaks after the wind eld has reached steady state.
3.7.5.5 Fluctuating wind
One can also specify uctuating wind eld from the boundaries using a setup-le. Two different
frequencies in horizontal directions and one in vertical direction are applied. This can be dened
with:
VERSION 1.1
$WINDGUST
USE=.TRUE.
$END
74 CASD
To change the frequencies of the uctuations, more options must be specied:
VERSION 1.1
$WINDGUST
USE = .TRUE.
AMP = 2.40, 1.90, 1.84
, 2.40, 1.90, 0.00
TAU = 15.0, 10.0, 10.0
, 70.0, 50.0, 0.0
$END
Two different uctuating periods are assumed along (15s and 70s) and across (10s and 50s) the
wind direction. In the vertical direction one period of 10s is used. Fluctuations are done as har-
monic periods with average velocity uctuation equal to 2.4 (along), 1.9 (across) and 1.3 (vertical
direction, here constant is multiplied by square root of 2 since only one period is used) times the
friction velocity, u.
3.7.5.6 Using TRACER mode when simulating dispersion
In acs2.2.6 a new simulation mode called tracer has been implemented. In this mode Flacs will
only solve a passive transport equation. Below is an example of how to enable the tracer mode:
VERSION 1.1
$SETUP
KEYS="TRACE:T=100,DT_MUL=Y:5"
$END
In the new scenario templates there is an option to dene KEYS within the scenario le (no need
to use a separate setup-le). The above setup-le (or the KEYS-line dened in the scenario le)
will simulate normally until time=100s, thereafter only the fuel transport equation will be solved
(ow eld will be kept constant). When the ow eld equations are switched off at 100s, the time
step is at the same time increased by a factor of 5 if using the DT_MUL=Y:5 string.
This option can be useful when a dispersion of neutral gas (or small quantities of gas) shall be
simulated in an established wind eld. This option should be used with care!
3.7.5.7 Effect of temperature gradients
To simulate the effect of an inversion layer, it is possible to dene a cold or hot temperature region
(layer) in FLACS. One can for instance dene a cold valley by using the setup-le as described in
previous sections. Boundary conditions can not take a temperature prole.
3.7.5.8 General considerations for boundary conditions
It is generally advantageous to place the outer boundaries of the simulation domain far away
from the geometrical extent, but limitations of memory and computing speed may restrict the
practical size of the problem, and in most cases one is forced to compromise between quality and
cost.
Solid wall boundary The solid wall boundary condition is straightforward to model. The ve-
locity vectors are zero at solid walls, both in the tangential and the perpendicular directions.
Hence, a zero gradient perpendicular to the boundary, or a xed value, works well for the scalar
variables. Furthermore, wall-functions may improve the modelling of the ow in near-wall re-
gions, both at the outer boundaries and in the interior space.
External inuences In cases where there are obstacles outside the vent openings of a semi-
conned volume they should be included in the total simulation volume, because they may have
an effect on the explosion. One effect may be that the total venting is reduced due to the external
obstacles, especially if they are large and are placed close to the vents. Since the vent ow is
changed, also the internal ow past obstacles is modied and the explosion becomes different
(higher or lower pressure results). An effect which also may be important is the appearance of
an external explosion which will start when the ame reaches any unburnt gas which may have
escaped through the vent openings. The pressure waves from the external explosion will prop-
agate back into the semi-conned volume and give rise to higher pressures there. The strength
of the external explosion will depend on the local turbulence in the external space, this again
depends on the properties of the vent openings and on any obstacles which may be positioned
in the external space.
3.7.6 Initial conditions
Initial conditions set values for turbulence elds, temperature and pressure at the beginning of
the simulation. Information about the gravity condtions, parameters for the atmospheric bound-
ary layer and the composition of the air is also set here.
The default values are as follows:
UP-DIRECTION 0 0 1
GRAVITY_CONSTANT 9.8
CHARACTERISTIC_VELOCITY 0.0
RELATIVE_TURBULENCE_INTENSITY 0.0
TURBULENCE_LENGTH_SCALE 0.0
TEMPERATURE 20.0
AMBIENT PRESSURE 100000
AIR "NORMAL"
GROUND_HEIGHT 0
GROUND ROUGHNESS 0
REFERENCE_HEIGHT 0
LATITUDE 0
SURFACE_HEAT_P1 0 0 0
SURFACE_HEAT_P2 0 0 0
MEAN_SURFACE_HEAT_FLUX 0
PASQUILL_CLASS "NONE"
GROUND_ROUGHNESS_CONDITON "RURAL"
Attention:
GexCon recommends to use the default values for explosion scenarios.
3.7.6.1 Up-direction
This is a three-component vector dening the upward direction, i.e. opposite of the acceleration
due to gravity. The three components denote the spatial components x, y, and z, respectively. It is
possible to dene any direction, not only directions aligned with the main axes. The vector does
not need to be of unit length.
3.7.6.2 Gravity constant
GRAVITY_CONSTANT is the magnitude of the gravitational acceleration, normally 9.8m/s
2
. Set-
ting a zero value for this parameter means that there are no gravitational inuences on the ow.
Note that panels with inertia are also inuenced by this parameter if the direction of panel release
is in the direction of the Up-direction, see also Pressure relief panels .
76 CASD
3.7.6.3 Characteristic velocity
CHARACTERISTIC_VELOCITY, U
0
, is used to nd values for initial turbulence elds and it
should take a positive or a zero value.
3.7.6.4 Relative turbulence intensity
RELATIVE_TURBULENCE_INTENSITY, I
T
, is the ratio between the isotropic uctuating veloc-
ity, u
, and the mean ow velocity U

0
:
I
T
=
u
U
0
(3.5)
I
T
will typically have a value in the range 0.0 to 0.1. This parameter is used together with Charac-
teristic velocity to calculate the value for the turbulent kinetic energy, k = 3/2u
2
at the beginning
of the simulation.
3.7.6.5 Turbulence length scale
TURBULENCE_LENGTH_SCALE,
LT
, is the length scale of the initial turbulence. A reasonable
choice for TURBULENCE_LENGTH_SCALE is a typical value for the grid size
g
.
LT
is used to
calculate an initial value for dissipation of turbulent kinetic energy, :
=
C
k
3/2
LT
(3.6)
3.7.6.6 Temperature
The initial temperature, T
0
in C
. The default value is T

0
= 20.0 C
3.7.6.7 Ambient pressure

AMBIENT PRESSURE, P
0
is the initial pressure in the simulation and the pressure outside the
simulation volume. The default ambient pressure is P
0
= 100000 Pa = 1 bar.
3.7.6.8 Air
AIR is used to dene the composition of the air. "NORMAL" denotes a standard composition of
20.95% oxygen and 71.05% nitrogen in mole fractions. The composition of air can be changed by
setting an other oxygen content, either as a mole fraction:
AIR "25%MOLE"
or mass fraction:
AIR "10%MASS"
Changing the air composition will inuence the lower ammability limit and upper ammability
limit.
3.7.6.9 Wind boundary layer parameters
In the Wind boundary condition, the reference wind speed, U
0
and the direction of the wind
are specied. In additions, information about the ground conditions is needed to setup inlet
proles for velocity and turbulence parameters. It is possbile to specify an atmospheric boundary
layer stability class, Pasquill class. Then turbulence parameter proles are generated at the wind
boundaries. If no Pasquil class is given, uniform values for k and are obtained on the boundary
according to the expressions in Relative turbulence intensity and Turbulence length scale.
The velocity prole at the wind boundaries is given by the following expression:
u(z) =
_
u
_
ln
_
z+z
0
z
0
__
if z
0
> 0
U
0
if z
0
= 0
(3.7)
where z is the height abouve the ground, z
0
is the atmospheric roughness length and u
is a
friction velocity. For the neutral and none Pasquill class, u
is dened by:
u
=
U
0
ln
_
z
re f
z
0
_ (3.8)
Ground Height GROUND_HEIGHT is height above the ground where the boundary layer ac-
tually starts, for instance due to vegetation. Usually GROUND_HEIGHT=0.
Ground roughness GROUND_ROUGHNESS, z
0
, refers to the aerodynamic roughness length .
Typical values for z
0
are given in table Typical values for aerodynamic roughness length. z
0
should in not be larger than the control volume height close to the surface. aerodynamic rough-
ness length should not be mixed with pipe roughnes etc., but a rule of dump is to relate z
0
to the
average height
g
of the surface irregularities by z
0
=
g
/30.
Terrain description z
0
(m)
Open water, fetch at least 5 km 0.0002
Mud ats, snow; no vegation, no obstacles 0.005
Open at terrain; grass, few isolated
obstacles
0.03
Low crops; occasional large obstacles 0.10
High crops, scattered obstacles 0.25
Parkland, bushes, numerous obstacles 0.5
Regular large obstacle coverage (suburb,
forest)
1.0
Table 3.4: Typical values for aerodynamic roughness length
Reference height REFERENCE_HEIGHT, z
ref
is the height relative to the ground where the ve-
locity equals the wind speed.
Latitude Due to the rotation of the earth, the height of the atmospheric boundary layer is much
larger at equator than at the poles. LATITUDE will only have an effect if a Pasquill class is chosen
and the simulation volume is very large (> 200 m) in the z direction.
78 CASD
Surface heat P1 If the temperature and velocity are known at two different altitudes, for in-
stance from experimental data, it is possible to estimate the surface heat ux.
SURFACE HEAT P1 is a vector Z1 U1 T1, where Z1 is the altitude, U1 is the velocity at Z1 and
T1 is the temperature at Z1
Surface heat P2 SURFACE HEAT P2 is a vector Z2 U2 T2, where Z2 is the altitude, U2 is the
velocity at Z2 and T2 is the temperature at Z2. Z2 must larger than Z1 and U2 must larger than
U1.
Mean surface heat ux MEAN_SURFACE_HEAT_FLUX is the heat ux in W/m
2
from the
ground to the ow. This is a parameter in the boundary layer proles for the unstable Pasquill
class A, B, and C. MEAN_SURFACE_HEAT_FLUX will not apply as heat contribution from the
ground to the ow in the simulations.
Pasquill class Pasquill atmospheric stability classes is a method of categorizing the amount of
atmospheric turbulence present. Pasquill (1961) made six stability classes where:
A is very unstable
B is unstable
C is slightly unstable
D is neutral
E is slightly stable
F is stable
An overview when to apply the different Pasquill classes is given in table: Pasquill stability
classes.
Wind speed Day, strong sun Day, moderate
sun
Night, clouds
>50%
Night, clouds
<50%
< 2m/s A B E F
2 - 3 m/s A-B B-C E F
3 - 5 m/s B B-C D E
5 - 6 m/s C C-D D D
> 6 m/s C D D D
Table 3.5: The Pasquill stability classes
Ground roughness conditions At present the only option is:
GROUND_ROUGHNESS_CONDITON "RURAL"
3.7.7 Pressure relief panels
Pressure relief panels are commonly used in the process industry as a mitigating device in the
case of an explosion. When the pressure forces on the panel exceed a certain limit, the panel yields
and the pressure is relieved. In FLACS, special planar elements inside the simulation volume are
used to model the effect of explosion relief panels and yielding walls. They are specied as boxes
where one, and only one, of the three side lengths must be zero. In addition several parameters
such as the yield pressures, area porosities and drag factor may be specied.
A panel may be active or passive. An active panel will initially modify the porosity in the region
it covers, and again when the pressure difference over the panel exceeds the given limit. If the
initial porosity is set to 0 and the nal porosity is set to 1 the panel will start as being closed and
end up as being open. This is how the behaviour of a real yielding panel is imitated. A passive
panel will never modify the porosity, but it may be used to monitor the same data as an active
panel. There are 5 different panel types (active panels):
Panel type Description
UNSPECIFIED Panel with linear displacement, full
parameter set
POPOUT Panel with a linear displacement, limited
parameter set
HINGED Panel with a rotational displacement, limited
parameter set
PLASTIC Simulates the presence of plastic sheets
(commonly used in experiments)
OVERLAY Panel properties are blended with existing
equipment
Table 3.6: Panel types in FLACS
Note that the PLASTIC and OVERLAY panels can not be chosen directly in CASD. It is possible
to edit the scenario le manually to use these panel types, by initially creating an UNSPECIFIED
panel and change it to the desired type. It is advised that only experts attempt to do this.
The edges of a panel will be automatically adjusted to the closest grid line. It is advised that the
user specify panels whose dimensions match the grid, either by adjusting the grid or modifying
the panel.
One topic which needs special attention is the presence of solid structures close to the panel area.
For example structural beams may constitute quite large blockages which must be accounted for,
especially since they also occur at the vent openings of typical offshore modules. In such cases
the panel area must be dened using several panels with solid beams in between. Smaller beams,
often an integral part of the support frame of the pressure relief panels, may be accounted for by
adjusting the nal panel porosity for the panel itself.
In addition to different types of pressure relief panels, an INACTIVE panel used to monitor vari-
ables is incorporated in the FLACS code. Use an INACTIVE panel to measure for instance average
pressure across surfaces such as decks and walls. The panel will be available under Single eld
scalar time output in the Scenario menu.
Note that if several panels are specied, they should not overlap each other anywhere (in the
model it is assumed that the panels are non-overlapping).
The full list of parameters from the scenario le is shown below. The type of the panel determines
which of these parameters are relevant.
INSERT 1
POSITION 0.0, 0.0, 0.0 (m)
SIZE 1.0, 1.0, 0.0 (m)
MATERIAL BLUE
PANEL_TYPE UNSPECIFIED
OPENING_PRESSURE_DIFFERENCES 0.0, 0.0 (bar)
INITIAL_AND_FINAL_POROSITY 0.0, 1.0 (-)
80 CASD
WEIGHT 0.0 (kg/m2)
DRAG_COEFFICIENT 1.0 (-)
MAXIMUM_TRAVEL_DISTANCE 0.0 (m)
SUB_SIZES 1.0, 1.0 (m)
The parameters for specifying panel properties are described in the following sections.
3.7.7.1 Insert
Integer identifying the panel.
3.7.7.2 Position
Cartesian coordinates [m] of the corner of the panel (the corner with lowest value of the coordi-
nate in each axis direction).
3.7.7.3 Size
The panel is assumed to be a plane of rectangular shape. The dimension [m] in each of the axis
directions is given. One dimension should be zero, this shows how the panel is oriented. If e.g.
the dimension in x-direction is zero, the normal of the panel points in x-direction. The other two
dimensions should be positive. The panel edges will be adjusted
to match the grid lines perfectly in the FLACS code, so it is advised that you only specify panels
which coincide with the grid lines, in order to avoid any confusion concerning the geometrical
representation of the panel.
3.7.7.4 Material
Colour used when visualizing the panel as part of the geometry considered.
3.7.7.5 Panel type
There are 6 different panel types described here. A panel is either active or passive. They are
listed and described in more detail below.
Inactive is a passive panel that does not affect the geometry or the ow eld. such panels are
used to monitor certain variables
Unspecied is a panel with linear displacement when the panel yields
Popout is a panel with a linear displacement when the panel yields
Hinged is a panel with a rotational displacement when the panel yields
Plastic is a panel that imitates a plastic sheet (commonly used in experiments)
Overlay is a panel superimposed on other equipment
Inactive panel A passive panel, corresponding to the case when PANEL_TYPE equals
INACTIVE, does not affect the numerical simulation in any respect (neither the geometrical de-
scription including the porosities, nor the ow eld), it is only included in order to monitor
variables related to the area which the passive panel occupies (e.g. area-averaged pressure over
all the control-volume faces of the panel). Only the parameters INSERT, POSITION, SIZE, and
MATERIAL are relevant when the PANEL_TYPE is INACTIVE.
Unspecied, Popout and Hinged panel An active panel (all the panel types except INACTIVE)
will in general affect both the porosities representing the geometry and the ow eld. For the
active panels both an initial and a nal area porosity value is given (open area divided by total
area). The initial porosity corresponds to the state of the panel before it has started to yield due
to external pressure forces. The nal porosity corresponds to the state of the panel after it has
yielded completely due to pressure forces.
Note that each control-volume face covered by the panel is treated separately. Thus it is possible,
depending on the scenario, that one part of the panel is open, while at the same time other parts
are closed.
The way of using the initial and nal porosities in the FLACS code depends on the type of panel
considered. In the case of the panel types UNSPECIFIED, POPOUT, and HINGED; for each control-
volume face which is covered by the panel, the porosity representing the panel is also the porosity
of the control-volume face, regardless of any other equipment or structure included in the geom-
etry.
Overlay and Plastic panel For a panel of type OVERLAY the value of the area porosity deter-
mined by the panel is multiplied by the value of the area porosity determined by other geometri-
cal objects (in the absence of any panel) giving the effective area porosity, for each control-volume
face covered by the panel. Let us consider an example where the nal porosity of the panel is 1
(fully open). After the panel has yielded completely for all the control-volume faces covered by
the panel, the area porosities have the same values that they would have if no panel was included
in the geometry. So in this case other geometrical objects than the panel itself can not "blow out"
together with the yielding panel (cf. the comments above).
For a panel of type PLASTIC, and in the case of the initial area-porosity; for each control-volume
face which is covered by the panel, the porosity representing the panel is also the porosity of the
control-volume face, regardless of any other equipment or structure included in the geometry
(as in the case of the panel types UNSPECIFIED, POPOUT, and HINGED). Usually the initial area-
porosity of a PLASTIC panel is zero (completely blocked). If the PLASTIC panel has yielded
completely, the area porosities have the same values that they would have if no panel was in-
cluded in the geometry (as in the case of an OVERLAY panel with nal area-porosity 1). Note
that for a PLASTIC panel the nal area-porosity is always assumed to be 1, regardless of the
user-given value.
Panels of the types PLASTIC or OVERLAY are assumed to be light panels, and inertial forces
are neglected by the FLACS code during the dynamical process when the panel yields due to
pressure forces. In the case of panels of types UNSPECIFIED, POPOUT, or HINGED, inertial forces
are in general taken into account.
Attention:
The PLASTIC and OVERLAY panels cannot be chosen in the present version of CASD. It is
possible to edit the scenario le manually to use these panel types, by initially creating an
UNSPECIFIED panel and change it to the desired type. It is advised that only experts attempt
to do this.
82 CASD
The +IMP panel types The standard panel types will yield at a give negative or positive pres-
sure. In some situations it can of necessary for panels to also take into account the pressure
impulse. To activate this the user can add the sufx +IMP to the end of the panel type string, e.g
PANEL_TYPE POPOUT+IMP
OPENING_PRESSURE_DIFFERENCES 0.05 0.005 (bar bar
*
s)
When using this the input to OPENING_PRESSURE_DIFFERENCES changes to
Parameter 1 is yield pressure for both negative and positive direction [bar]
Parameter 2 is yield pressure impulse both negative and positive direction [bars]
The panel will yield if both the pressure and pressure impulse yield values are exceeded.
This is valid for panel types HINGED and POPOUT.
Attention:
This functionality is not available through CASD. The user must manually edit the cs-le.
3.7.7.6 Opening pressure differences
When the net pressure over the panel (pressure on the negative side relative the coordinate axis
minus pressure on the positive side, i.e. the net pressure from the uid acting on the panel)
exceeds limits given by the user, the panel starts to yield. Two opening pressure-differences, in
units of [bar], are given by the user, the rst corresponds to the case when the net pressure acts in
negative direction, the second when the net pressure acts in positive direction. If for example the
OPENING_PRESSURE_DIFFERENCES are given by -0.1 and 0.2, the panel starts to yield when the
net pressure is less than -0.1bar (negative direction) or greater than 0.2 bar (positive direction).
Note that the FLACS code makes the same interpretation of the input data, regardless of the
sign of the opening pressure-differences given by the user. Hence, if the OPENING_PRESSURE_-
DIFFERENCES were given instead by 0.1 and 0.2, or by 0.1 and -0.2, the effect on the numerical
simulation would be exactly the same. The opening pressure-differences must be given for the
active panels (all the panel types accept INACTIVE).
3.7.7.7 Initial and nal porosity
The initial area porosity [-] of the panel (open area divided by total area) corresponds to the
state of the panel before it has started to yield due to external pressure forces. Similarly the nal
porosity corresponds to the state of the panel when it has yielded completely. In the case of
a PLASTIC panel, the nal area-porosity should be set to 1 by the user in order to correspond
to how the FLACS code works (the nal area-porosity is always assumed by FLACS to be 1,
regardless of the user-given value). Note that the way of using the initial and nal porosities in
the FLACS code depends on the type of panel considered. See the comments above. The initial
and nal porosity must be given for the active panels (all the panel types accept INACTIVE). The
value of the porosity range from 0 (totally blocked) to 1 (fully open). It is expected that the initial
porosity is less than the nal porosity (the net pressure opens up the panel). If this is not the case,
the FLACS code gives a warning.
3.7.7.8 Weight
The parameter WEIGHT is specied for the panel types UNSPECIFIED, POPOUT, and HINGED (for
the other panel types it is not relevant to specify this parameter). This parameter is the mass per
unit area [kg/m
2
] of the part of the panel that is moved when the net pressure acting on the panel
exceeds the opening pressure-difference given by the user. The value of WEIGHT is either zero or
positive. If the value of WEIGHT is zero, it is assumed in the FLACS code that the panel is so light
that inertial forces can be neglected during the dynamical process when the panel yields due to
pressure forces. If the value of WEIGHT is positive, the dynamics when the panel yields and part
of the panel is accelerated and move away from its initial position, is modelled in the FLACS
code. Note that when you specify the panel type POPOUT, the inertial forces must be there when
the panel yields, that is the WEIGHT must be a positive number. Note that it is only the part of
the panel that is accelerated and move away from its initial position, that should contribute to
the mass per unit area. The mass of a rigid frame that does not yield, should not be included.
3.7.7.9 Drag coefcient
A dimensionless drag-coefcient [-] is given by the user for the panel type UNSPECIFIED (for
the other panel types it is not relevant to specify this parameter). The drag coefcient is used
when the drag force from the panel on the uid is modelled (both before, during, and after the
panel has yielded). The value of the DRAG_COEFFICIENT is zero or positive. A typical value is
2.0 (the value 2.0 is set by the FLACS code as a xed preset value for the panel types POPOUT and
HINGED).
3.7.7.10 Maximum travel distance
The maximum travel-distance [m] is given by the user for the panel type UNSPECIFIED (for
the other panel types it is not relevant to specify this parameter). The use of the maximum
travel-distance is based on a rough approximation. The maximum travel- distance is the small-
est distance from the initial position of the panel to the position where the yielded panel no
longer affects signicantly the effective area-porosity at the initial position of the panel. The
maximum travel-distance is used to model the effective area-porosity at the initial position of the
panel during the dynamical process when the panel yields. A typical value of the MAXIMUM_-
TRAVEL_DISTANCE is in the order of 1m (for the panel types POPOUT and HINGED there is an
in-built model in the FLACS code for the effective area-porosity that does not need a value of the
MAXIMUM_TRAVEL_DISTANCE as input from the user).
3.7.7.11 Sub sizes
The parameter SUB_SIZES is given by the user for the panel types POPOUT or HINGED (for the
other panel types it is not relevant to specify this parameter). These panel types are assumed to
consist of sub-panels mounted on a frame. The width [m] and the height [m] of the sub-panels
are given by the user (it is assumed in the numerical model that all the sub-panels are of uniform
size). Both the width and the height should be positive. In the case of the panel type HINGED it
is assumed that each sub-panel turns on a hinge when it yields. The width of the sub-panel is
dened as the dimension in the direction normal to the axis of the hinge. It is important (since
it affects the model in FLACS) to specify the width of a HINGED panel rst and then the height,
in the parameter SUB_SIZES (for a POPOUT panel it is of no importance which dimension that is
dened as the width).
3.7.8 Gas composition and volume
This section allows you to dene a box shaped cloud region and the gas concentration and com-
position. The menu is shown below:
84 CASD
POSITION_OF_FUEL_REGION 0.0, 0.0, 0.0 (m)
DIMENSION_OF_FUEL_REGION 0.0, 0.0, 0.0 (m)
TOXIC_SPECIFICATION ""
VOLUME_FRACTIONS <MENU>
EQUIVALENCE_RATIOS_(ER0_ER9) 1.0, 0.0
The position of the cloud is the location of the minimum point (x
min
, y
min
, z
min
) of the box, and the
sizes (x
siz
, y
siz
, z
siz
) give the side lengths of the box (only positive values allowed for the sizes).
The gas composition is specied by entering volume fractions of the listed gas components, and
the concentration is given by the equivalence ratios (ER0 inside the gas cloud and ER9 outside)
METHANE 1.0
ACETYLENE 0.0
ETHYLENE 0.0
ETHANE 0.0
PROPYLENE 0.0
PROPANE 0.0
BUTANE 0.0
HYDROGEN 0.0
CO 0.0
CO2 0.0 (inert)
H2S 0.0
TOXIC 0.0
The sum of the volume fractions does not need to be 1, FLACS will interpret the values as volu-
metric parts (normalization is done in FLACS by dividing each value by the sum of all values).
For a mixture of two or more gas components, the combustion properties are calculated as a sort
of average of all the selected components, taking into account the mole (volume) fraction and the
relative consumption of O2 (parameter a in the table above). This has shown to give the proper
blending of the component properties into mixture properties.
The laminar burning velocity of a gas mixture in FLACS depends on the concentration of the fuel
relative to the concentration of oxygen as well as on the type of fuel. A widely used measure for
the relative fuel-oxygen concentration is the Equivalence ratio (ER) which is dened as follows:
ER = =
_
m
f uel
/m
oxygen
_
actual
_
m
f uel
/m
oxygen
_
stoichiometric
=
_
V
f uel
/V
oxygen
_
actual
_
V
f uel
/V
oxygen
_
stoichiometric
(3.9)
Now, the dependency of the laminar burning velocity (Slam) is illustrated for methane in the
gure below. The factor, ERfac, which is a function of ER and ranges from 0 to 1 is multiplied
with the tabular value of the laminar burning velocity for the gas component, thus a relation
between Slam and ER is established.
Each gas component has its own ERfac curve in FLACS. The next gure shows the laminar burn-
ing velocity curve as function of ER for all the gas components separately. Note that hydrogen,
acetylene and propylene have the three highest laminar burning velocities, and notably also the
widest range of ammability.
For a gas mixture the resulting curve is based on the blended properties of the components as
have been described initially in this section.
3.7.9 New gas data
The gas-data for acs2.2.7 are found in " acs/FLACS/bin/les/gasdata2.2.7/" , these have
been adjusted for better conformance with published values (especially for the LFL and UFL
values). In addition, the model for the effect of inert gases on the ammability limits and burning
velocity has been revised.
The effect of inert gas on the ammability limits can be better understood by looking at the plot
below:
Notice that He or N2 cannot be modelled as " fuels" in FLACS yet. H2O can now be modelled.
The two plots below show the effect of different inert gases on the laminar burning velocity of
methane at varying equivalence ratio (ER):
3.7.10 Leaks
Gas leaks occur quite frequently in many offshore and onshore installations and elsewhere, either
during production, processing or transport of the gas. Hydrocarbons and many other types of
highly ammable gases are present all over the industry and may be a source for potentially
dangerous gas explosions. In addition, spills of ammable liquids may cause accumulation of
explosive gas clouds under certain conditions. Experiments have clearly shown that even small
amounts of gas mixed with air at the right concentration can result in very strong explosions.
By using the dispersion and explosion simulation capabilities of FLACS it is possible to analyse
quite realistic scenarios, where the effect of a single gas leakage developing into an explosive
gas cloud and nally resulting in several possible gas explosion scenarios may be studied. One
primary interest may be to vary the leakage itself, changing the ow rate, location, and duration.
Other sensitivities may also be studied such as modifying the ventilation and wind conditions
for a given geometrical layout. Finally, the variation of ignition time and location may be made
during the explosion studies.
The release model for leaks in FLACS ensures that the desired mass ow or velocity of gas is set
at the control volume where the leak is located. Also other properties such as the temperature
and relative turbulence intensity and length scale must be specied. The gure below shows how
a jet is represented on the grid:
Figure 3.12: Leak denition
86 CASD
The area porosity of the open control volume face is adjusted to ensure that the ow rate is
correct. If the grid resolution is coarse compared to the size of jet orice, a good representation of
the detailed ow in the near-orice region is not possible. This may lead to excessive smearing
of the gradients of the ow. In such cases (which may be the normal practical case) one must
realize that the simulated gas concentration is locally underestimated, although the total amount
is correct. The grid resolution may be especially important for jets impinging on nearby walls
and jets with a strong cross-wind inuence.
It is possible to specify one or more leakage points in a FLACS simulation. The leakage is in-
tended to model the ow of gas from a reservoir into the simulation volume, the available pa-
rameters controlling the leakage are shown next:
LEAKS
INSERT 1
TYPE "JET"
POSITION 1.0, 1.0, 1.0 (m)
OPEN_SIDES "+X"
START_TIME 1.0 (s)
DURATION 10.0 (s)
OUTLET
VESSEL
Attention:
Remember to set the ER0 parameter under GAS_COMPOSITION_AND_VOLUME to a large
nunber if the leak is a pure fuel leak.
EXCESS LEAK RATE
3.7.10.1 Insert
This command is used to insert a new leakage, the parameters for this leakage will be explained
in the following next pages. The maximum number of leakages is 50.
3.7.10.2 Type
The type of leakage must be specied as one of the following:
DIFFUSE Low momentum point leak (deprecated)
JET Low and high momentum point leak (preferred)
AIR Similar to "JET", but leak mixture is air (ER9)
SUCTION Negative point source, i.e. removal of gas/air mixture
FAN Fixed momentum in CV
The pool leak is not specied in this menu but is rather dened using the cs.POOL and the
cl.POOL les (see section Pool setup le).
Diffuse The DIFFUSE leak is a no momentum (same velocity as surrounding ow) release of
gas into the grid cell chosen. One must give a direction, but this direction will not be applied.
DIFFUSE should normally not be used.
Jet Low and high momentum point leak.
Air Low and high momentum point leak of air (ER9).
Suction Negative point source, i.e. removal of gas/air mixture.
Air If a FAN with direction +X is specied, this will be translated into the string "!J+X=X:fan"
in the leak le. Fans should be along the axis directions.
3.7.10.3 Leak build-up
Leaks previously were built up in a time equal to 1% of the leak duration, and thereafter immedi-
ately shut down when killed. Now the maximum build-up time has been set to 1s, the shut-down
has also been made smoother, with a gradual shut-down over 1s. These values can be changed
manually in cl-les.
3.7.10.4 Leak concentration between ER0 and ER9
Concentration between ER0 and ER9 can be specied for a leak. The control string must be
changed manually.
Example: To get a concentration of 60% ER0 and 40% ER9 one must change leak control string in
cl-le.
From:
"!J+X"
to:
"J+X:mix=0.6"
3.7.10.5 Q8 and Q9 output
The rt.FUEL and rt.MON les have now two columns, called Q8 and Q9. These les are gen-
erated from GAS_MONITOR_REGION" in cs-le or cs-MON les. Q8 and Q9 is used to create
equivalent stoichiometric gas clouds from real gas clouds.
Q8 This column reports expansion weighted volume, i.e. closed volume equivalent cloud at
concentration for maximum expansion (normally near stoichiometry).
Q9 This column gives an improved version of Q5 for gases where maximum ame speed and
maximum expansion deviates. Q9 will for other gases be marginally higher than Q5.
For enclosed situations, and situations where combustion is much quicker than venting
(including quasi-detonation and ames faster than speed of sound ahead) Q8 will be a
recommended equivalent cloud size.
For well-vented situations we generally recommend to use Q9 (to replace Q5) as equivalent
stoichiometric cloud.
3.7.10.6 Position
This menu is used to specify the leak position (in metres).
88 CASD
3.7.10.7 Open sides
The open sides indicates which side(s) of the control volume affected by the leak should be made
to open up as the leak starts (direction of the leak). OPEN_SIDES is a text string containing char-
acters taken from "+-XYZ". A leakage directed in the positive x-direction should have OPEN_-
SIDES = "+X". Several sides may be open for a single leak, e.g. OPEN_SIDES = "+X-X" or
OPEN_SIDES = "+XY-ZX". This will give two or more identical leaks
3.7.10.8 Start time
This is the time (in seconds) when the leakage should start. Normally one would allow a certain
time for wind build-up prior to starting any leaks.
3.7.10.9 Duration
This is the duration (in seconds) of the leakage. The nish time for the leakage is the start time
plus the duration. Hence, the user should set TMAX accordingly. The user should also note that
the build-up time of the leak is 1 % of leak duration (maximum 1 second).
3.7.10.10 The Outlet menu
Users specify the leak conditions in the outlet menu, e.g.
AREA 0.1 (m2)
MASS_FLOW 0.0 (kg/s)
VELOCITY 10.0 (m/s)
RELATIVE_TURBULENCE_INTENSITY 0.15
TURBULENCE_LENGTH_SCALE 0.01 (m)
TEMPERATURE 6.0 (C)
DIRECTION_COSINES 0.0, 0.0, 0.0
3.7.10.11 Outlet: area
The effective cross-section area A of the leak outlet must be specied for any type of leak. For a
specied mass ow m (or volume ow

V and density ) the velocity u at the outlet should not be
above the speed of sound. The following relations apply for the conditions at the outlet:
m =

V = uA (3.10)
Assuming ideal gas properties:
pV = nT =
m
M
T = mRT (3.11)
where is the ideal gas constant (8314.3 [J/kmole K]), T [K] the and the specic gas constant R
is equal to the ideal gas constant divided by the molecular weight M, the gas density at the outlet
is:
=
m
V
=
p
RT
(3.12)
m = uA Mass ow at the leak outlet (kg/m3/s)
= p/(RT) Gas density at the outlet (kg/m3)
p = p
amb
Pressure at the outlet is the ambient pressure (Pa)
R = /W Specic gas constant (J/kg/K)
W Mole weight for the given gas composition (kg/kmole)
T Temperature at the outlet (K) (0.0 C = 273.15 K
u Velocity at the outlet (m/s)
The relations shown above may be used to ensure that the outlet conditions are set in a consistent
way.
3.7.10.12 Outlet: mass ow
The user can specify the mass ow rate at the leak outlet here. Note that mass ow takes priority
if both mass ow and velocity are specied (see below). Make sure that the generated velocity is
not larger than the speed of sound (approximately 340 m/s for air). Setting the outlet area large
enough for a given mass ow rate causes the outlet velocity to be as low as desired. FLACS will
report the generated velocity on the log le (rt-le). The mass ow rate must have a positive
value (zero if not specied).
3.7.10.13 Outlet: velocity
Instead of specifying a mass ow rate at the outlet (see above) the user may specify the velocity.
FLACS will report the generated mass owrate on the log le (rt-le). If you specify both MASS_-
FLOW and VELOCITY then the mass ow rate takes precedence. The velocity must have a positive
value (zero if not specied). The specied velocity should be lower than the speed of sound.
3.7.10.14 Outlet: relative turbulence intensity
This parameter must always be specied for a leakage. Normally a value in the range 0.01 to
0.10 will be appropriate. If the relative turbulence intensity cannot be obtained from any source
(literature or experiments) for the given leakage, the following rough classication may be used:
0.01-0.03 Low turbulence intensity
0.03-0.06 Medium turbulence intensity
0.06-0.10 High turbulence intensity
It is generally believed that the turbulence generated in the jet zone is less important than the
turbulence generated by the induced downstream ow eld, in which case the k model in
FLACS represents the actual modelling. It should be noted here that on the coarse grids nor-
mally used for dispersion calculations in large geometries the generated turbulence due to uid
stresses may be underestimated, and since numerical diffusion may be high on such coarse grids
one might expect that the turbulent (or effective) mixing process is not represented with high
accuracy.
3.7.10.15 Outlet: turbulent length scale
This parameter must always be specied for a leakage. It may be difcult to obtain the value for
the turbulence length scale. It is generally related to the size of the nozzle or the feeding pipe
90 CASD
for the leakage. A rough estimate is that the turbulence length scale is in the range of 10% to
20% of the nozzle diameter. When the turbulence length scale is small the dissipation rate of the
turbulence kinetic energy is large, thus the turbulence will dissipate quickly for small diameter
nozzles. In such cases, the turbulence generated by the downstream ow eld is the important
factor, and the k model in FLACS takes care of that.
3.7.10.16 Outlet: temperature
This parameter must always be specied for a leakage.
3.7.10.17 Outlet: direction cosines
This is a vector which may be used to dene the direction of an oblique jet.
3.7.10.18 The Vessel menu
This menu was developed to calculate the outlet conditions for a leak from a pressurized vessel.
Currently, it is not recommended to use the vessel menu. The Jet utility program should be used
instead.
3.7.11 Ignition
In case of a gas explosion simulation the user must specify the location and size of the ignition
source and also a time for the ignition to occur. The available parameters here are as follows:
POSITION_OF_IGNITION_REGION 0.0, 0.0, 0.0 (m)
DIMENSION_OF_IGNITION_REGION 0.0, 0.0, 0.0 (m)
TIME_OF_IGNITION 0.0 (s)
RADMAX 99999.0 (m)
The ignition region can be a point, a line, a plane or a volume. Normally the user would choose
an ignition point (zero dimension), with ignition effected at time zero. The RADMAX parameter
will be ignored by all FLACS versions after 1998, but it is only present for backward compatibility.
If the ignition point is inside a partially blocked control volume, the ame might quench. It is
therefore not recommended to ignite inside a partially blocked control volume. But if the user still
chooses to do so, and if there are any problems to obtain a proper ignition and ame propagation,
the user might try to increase the DIMENSION_OF_IGNITION_REGION side lengths up to about
0.05 to 0.10 m.
Ignition in FLACS is usually set to occur in just one control volume. It is possible to dene a larger
region for the ignition, but this is not advised. The time of the ignition may also be specied, this
is useful for igniting gas clouds which are generated during gas dispersion simulations. In a
normal gas explosion simulation the time of ignition should be set to zero.
Attention:
The user should avoid specifying the ignition point exactly on a grid line, and should rather
specify it at the centre of a control volume. Usually, it is best to specify the ignition point
according to the grid, and not according to the geometry. For instance, if the ignition point is
placed on one side close to a solid wall,it may not necessarily ignite at that side of the wall.
During the porosity calculation, the wall may be adjusted to the nearest control volume face
(grid line) and might therefore move to the wrong side of an ignition point! It may also be a
good idea to ensure that the ignition point is not inside a fully blocked control volume.
Figure 3.13: How to position the ignition
3.7.12 Water spray
Water deluge systems can be an effective way to mitigate the consequences of gas explosions
in several situations. The mitigating effect has been seen in many experiments. However, the
phenomena related to the interaction between water sprays and an accelerated ow eld and the
ame are quite complex. The explosion accelerates the ow as well as the water droplets. The
acceleration depends on the local ow conditions, and the droplet size. When hydrodynamic
forces become large enough to overcome surface tension, the droplets break up (again dependent
on the droplet size). When the droplets are small enough (either because the droplets produced
by the nozzle are small, or because large droplets break up into small droplets when the ow is
accelerated), they tend to reduce the burning rate due to cooling of the ame, and dilution of the
gas mixture by evaporation. However, large droplets in the ame region (before break-up of the
large droplets) tend to increase the turbulence level and thereby increase the burning rate. So
there are competing effects when using a water spray system. In some cases water spray may
even increase the maximum overpressure of an explosion.
The problem when modelling water sprays is how to represent all this on a coarse simulation
grid. It was necessary to do simplications, and also to use experiments performed for tuning of
the models. Two distinct effects were identied.
1. The acceleration of the ames due to turbulence from the sprays and the presence of the
droplets.
92 CASD
2. The reduction of burning rate experienced in certain situations because of the water sprays.
For each of the nozzles an acceleration factor, denoted F1, is determined. F1 is used to increase the
burning rate if any watersprays are present. A quenching factor, denoted F2, is also determined.
F2 is used to reduce the burning rate if the conditions for droplet break-up are present.
Remarks:
Due to model simplications, there is no need for a very accurate positioning of the sprays.
In regions where sprays of the same type will overlap, the user should dene one water
spray region for the whole system. If more than one waterspray region is to be dened, the
user should make sure that they do not overlap, as FLACS will then stop. In FLACS, the
droplets are assigned a velocity but the transportation of droplets is not modelled. Regions
where it is obvious that a lot of droplets will be transported ahead of the ames (for instance
directly outside vent openings), should be included in the spray region (i.e. larger water
spray regions should be dened).
The models are validated in stoichiometric gas concentrations in a 180 m
3
vented box, with vari-
ous obstruction levels, in a 50 m
3
model of an offshore module, and also in full- scale experiments.
It is reasonable to believe that the models represent mechanisms in connection to water mitiga-
tion well.
The water spray model implemented in the FLACS code is relatively simple. One or more non-
overlapping water spray regions are dened. In each region there is assumed to be droplets of
a given diameter (before break up), and a given water volume-fraction. If the relative velocity
between the droplets and the gas ow exceeds a so-called critical break-up velocity (depending
on the diameter of the droplet), it is assumed that the droplets break up.
Two non-dimensional factors are employed in the numerical model. When there is water in the
reactive mixture (i.e. inside a water spray region), this is assumed to enhance the burning rate.
Before break-up of the droplets, the burning enhancement-factor denoted F1 [-] (positive number)
is multiplied by the laminar burning velocity and added to the ordinary burning velocity (i.e. the
burning velocity employed in the FLACS code without any water spray, this burning velocity is
in general in the turbulent regime) to give the effective burning velocity with water spray. When
the droplet break-up criterion is fullled, the burning velocity (without break-up) is multiplied
by the burning reduction-factor F2 [-] (positive number less than 1), to give the effective burning
velocity
S
water
= (S
turb
+ F1 S
lam
)F2 (3.13)
The water spray model is validated for explosions with stoichiometric fuel-air mixtures. For non-
stoichiometric fuel-air mixtures it is assumed that the pressure reduction caused by water spray
might be slightly under predicted.
Below is shown a list of parameters which may be visible on the scenario le:
INSERT 1
POSITION 0.0, 0.0, 0.0 [m]
SIZE 1.0, 1.0, 1.0 [m]
VOLUME_FRACTION 0.2 [per thousand]
MEAN_DROPLET_DIAMETER 300.0 [mm]
NOZZLE_TYPE "FACTORS: 10.0 0.3" [-]
3.7.13 Insert
Integer identifying the water spray region considered. A maximum number of 25 regions may
be dened in the current version of the FLACS code.
3.7.14 Position
Cartesian coordinates [m] of the corner of the box-shaped water spray region (the corner with
lowest value of the coordinate in each axis direction). Make sure not to dene any overlapping
regions. The FLACS code will assign the nearest grid line to your given position as the actual
position of the water spray region. In other words, FLACS will make the water spray region stick
to the grid lines.
3.7.15 Size
The dimension [m] in each of the axis directions is given for the box-shaped water spray region.
All three dimensions should be positive. Make sure not to dene any overlapping regions, the
FLACS code will assign the nearest grid line to your given (position + size) as the actual (position
+ size) of the water spray region. In other words, FLACS will make the water spray region stick
to the grid lines. The FLACS code will issue an error message and stop if you have dened any
overlapping water spray regions. You should check the position and size parameters if this error
occurs.
3.7.16 Volume fraction
Volume fraction [per thousand] of water, dened by volume of liquid water in litre divided by
total volume in cubic meter (one cubic meter equals 1000 litre), inside the water spray region.
Thus, if the volume fraction is 0 there is no liquid water, and if the volume fraction is 1000 there
is only liquid water. If this parameter is less than 0.01 (corresponds to a mass fraction less than
1%), the FLACS code will give you a warning that the water spray is assumed not effective in
the numerical model. As long as the volume fraction of water is larger than this small minimum
value, the water spray model in FLACS is effective (in the current version of FLACS the value
of the parameter VOLUME_FRACTION does not affect how the water spray model works, so any
value larger than the minimum value will give the same numerical results in the simulation).
3.7.17 Mean droplet diameter
The mean diameter [mm] of the water droplets before break-up due to acceleration of the gas
ow, is given by the user. In the water spray model it is assumed that all droplets have the same
size, and that the droplets are uniformly distributed in space inside the water spray region. This
is an approximation. In most real situations there will be a droplet size distribution, which in
most cases is non-uniform in space, i.e. the distribution may change from one region to another
region.
In the water spray model, the mean droplet diameter is dened to be the so-called Sauter diam-
eter. This diameter is dened by the volume-based mean diameter cubed divided by the area-
based mean diameter squared. The Sauter diameter depends on the operating water-pressure
forcing water out the nozzle. The empirical relation
D = P
0.333
(3.14)
94 CASD
( D in [mm] and P in [barg]) seems to give good estimates for the Sauter diameter over a range of
values for the water pressure, and for various nozzles. However, for some special nozzle types
giving very large (e.g. nozzle type LDN) or very small (e.g. nozzle type P120) droplet diameters,
this empirical relation may be less accurate. One may then try to estimate otherwise the Sauter
diameter for the water-pressure considered. You may use either your estimate or data given by
the vendor of the nozzle.
3.7.18 Nozzle type
The parameter NOZZLE_TYPE should be set equal to the text string "FACTORS: F1 F2" (note
that the double quote character " is included twice in the text string), where F1 and F2 are the
numerical values of the two factors. The two non-dimensional factors F1 and F2 are modelled
by
F1 = 14U
z
water
(3.15)
and
F2 =
0.03
D
water
(3.16)
where U
z
[m/s] is the average droplet velocity vertically downward (absolute value),
water
[per
thousand] is the water volume-fraction, and D [mm] is the Sauter diameter. How to obtain the
Sauter diameter (mean droplet diameter) is described above. If the nozzle spreads the water
horizontally, the droplets will soon fall down with a constant velocity (gravity forces are balanced
by drag forces). This constant velocity depends on the droplet diameter. It can be estimated from
the empirical relation
U
z
= 2.5D
0.94
(3.17)
where the units of U
z
is [m/s] and the units of D is [mm]. For some nozzle types the droplets
leave the nozzle with a signicant velocity component vertically downward. In this case the
average downward velocity should be estimated otherwise (giving a larger value than from the
expression above). The water volume-fraction is estimated by
water
=
n(Q/60)
X
length
Y
length
Z
length
(3.18)
where X
length
[m] is the length in x-direction of the assumed rectangular waterspray region,
Y
length
[m] is the length in y-direction (it is assumed that the xy-plane is the horizontal plane), n
is the number of nozzles (it is assumed that all of the nozzles within the same waterspray region
is of the same type and with the same water ow-rate), and Q is the water ow-rate [litre/min]
for a single nozzle (thus Q/60 is the water ow-rate in units of [litre/s]). The water ow-rate
depends on the operating water-pressure. It is assumed that the ow-rate is related to the water
pressure P [barg] by
Q = k
P (3.19)
where the so-called k-value of the nozzle depends on the type of nozzle considered.
3.7.19 Louvre panels
Warning:
The louvre panel model has not been thoroughly validated, and some limitations have been
identied. Especially the thickness of the louvre slats (section Area porosity) does not seem
to have a large enough effect on the results. As the louvre panel model is relatively compli-
cated to set up, it is recommended to use a porous plate in most situations.
Louvre panels are common in offshore installations and also in land-based process industry. The
louvre panels affect the ow eld both due to drag forces, and deection of the ow downstream
of the panel, determined by the geometry of the louvre slats. A new subgrid model for ow
through louvres is described in detail in [Salvesen, October 1996]. Some validation exercises
using the new subgrid model are documented in [Salvesen, November 1996].
A louvre panel is assumed to consist of slats mounted on a frame. The louvre slats may have an
arbitrary form depending on the type of louvre panel considered, but it is assumed that the slats
of a louvre panel are more or less equal in shape, uniformly oriented, and uniformly distributed.
It is assumed that the velocity vector downstream of the louvre panel is forced to be within a
xed plane (a mathematical plane of zero thickness) determined by the louvre slats, when the
ow exits the louvre, independently of the upstream velocity.
To be specic, let us consider an example: A louvre panel is oriented with normal in x- direc-
tion. The louvre slats dene a plane with unit normal vector in the xy-plane, (sin(q), cos(q), 0),
where q is an angle with absolute value less than 90. The angle q is dened as the angle between
the slats of the louvre (assuming that the slats are more or less plane, or if they are curved that
they dene a tangent at the exit on the downstream side) and the normal of the louvre (in our
example the x-axis). The assumption is that the projection of the downstream velocity vector for
ow in positive x-direction, on to the xy-plane, is directed a xed angle q relative the positive
x-axis (corresponding to the tangent vector (cos(q), sin(q), 0)), independently of the upstream
velocity. How good this assumption is in practice depends on several factors: the form (plane
or curved) and the thickness of the slats, and how wide the slats are compared to the distance
between two neighbouring slats. The component of the velocity vector in z-direction, is assumed
to be unaffected by the presence of the louvre panel.
In the example above, a louvre panel is oriented in x-direction. It may also be oriented in y- or
z-direction. In general the louvre slats dene two distinct planes, one plane for ow exiting in
positive direction and one plane for ow exiting in negative direction. Often these two planes are
identical. This is the case e.g. if the slats are of rectangular shape. But if the slats are e.g. V-formed,
the two planes are distinct. The two planes may be arbitrary oriented as long as two conditions
are satised: Firstly, none of the planes should coincide with the louvre plane itself (this would
imply that the louvre panel is completely blocked). In the numerical code it is checked that the
angle between the unit normal of the louvre plane and the unit normal of the plane determined
by the slats (for ow in positive or negative direction), is not too small, that is less than one
degree. Secondly, if the plane determined by the slats for ow in positive direction is distinct
from the plane for ow in negative direction, the vector cross product of the normal vectors of
these two planes should lie in the louvre plane (in other words, these two normal vectors and the
normal of the louvre plane all lie in the same plane).
Note that the subgrid model affects how the momentum ux is calculated for the face of the
staggered control volume adjacent to the louvre plane on the downstream side. The velocity of a
control volume is determined by a balance of momentum uxes over all the faces of the control
volume. The subgrid model gives the momentum ux at the CV face adjacent to the louvre
panel, the momentum uxes at the other CV faces are determined by the ow eld otherwise.
A consequence of this is that the rst velocity vector downstream of the louvre plane need not
be directed exactly according to the plane determined by the louvre slats, since the ow eld
96 CASD
otherwise also affects the velocity vector.
In the subgrid model it is assumed that the total drag force can be represented as the sum of
three terms; drag due to acceleration of the magnitude of the velocity of the ow, drag due to
bending of the ow, and drag due to friction. In the description of these three terms below, it is
assumed that the ow is in positive direction (either x-, y- or z-direction). The subscript e (east)
corresponds to downstream values, the subscript w (west) corresponds to upstream values. A
similar description is valid for ow in negative direction (not given here). Drag (absolute value)
due to acceleration of the ow is represented by:
F
D,acc
= C
acc
max
_
0.05
e
[U
e
]
2
0.5
w
[U
w
]
2
_
(3.20)
in the case that the plane determined by the louvre slats for ow in positive direction is the same
as the one for ow in negative direction. Here is the density, and U is the velocity vector. The
drag coefcient C
acc
is typically set to 1. Note that no contribution from drag due to acceleration
is included if the norm of the velocity vector upstream is larger than the one downstream (this
would correspond to a pressure increase instead of a pressure drop over the louvre).
In the case that the plane determined by the louvre slats for ow in positive direction is distinct
from the one for ow in negative direction (e.g. if the slats are V-formed), it is assumed that the
drag is represented by:
F
D,acc
= C
acc
_
max
_
0
0.5
_
int
[U
int
]
2
w
[U
w
]
2
_
_
+ max
_
0
0.5
_
e
[U
e
]
2
int
[U
int
]
2
_
__
(3.21)
Here U
int
is the so-called intermediate velocity vector in the plane determined by the louvre slats
for ow in negative direction. It is assumed that the velocity vector rst is forced to be within this
plane, and then is forced to be within the plane determined by the louvre slats for ow exiting in
the positive direction.
Drag due to bending of the ow is modelled as:
F
D,bend
= 0.5
w
[U
w
]
2
C
bend
(1 + 2) (3.22)
where 1 is the angle between the upstream and the intermediate velocity vector, 2 is the angle
between the downstream and the intermediate velocity vector, 0 i < , i = 1, 2. Note that
the intermediate velocity vector and the downstream velocity vector are identical when the two
planes determined by the louvre slats (for ow exiting in positive and in negative direction)
are identical (and then 2 = 0). The coefcient C
bend
(units of 1/radian) could be estimated from
experiments or ne grid simulations. In the validation simulations a value of C
bend
= 0.11/(/4)
is used. This corresponds to a result found in the literature; a resistance coefcient for ow in
pipes of 0.11 for a bending of /4radian. Confer [Salvesen, November 1996] for more details.
If the ow is neither bended nor accelerated, there is still a drag due to skin friction assumed to
be represented by
F
D, f ric
= 0.5
w
[U
w
]
2
C
f ric
(3.23)
where C
f ric
is a drag coefcient.
The coefcients C
acc
, C
bend
, and C
f ric
are expected to depend on the specic type of louvre con-
sidered, and on both the Reynolds number and the Mach number of the ow. The direction of
the upstream velocity vector may also be of importance. To investigate how the coefcients de-
pend on the various parameters mentioned above, seems to be a challenging task. Performing
ne-grid simulations or experiments are ways of approaching the problem. In the present model
it is assumed that the coefcients depend on only the geometry of the louvre (neglecting the
dependence on the other parameters mentioned above).
If experimental values of the coefcients C
acc
, C
bend
, and C
f ric
are known for the specic type of
louvre panel considered, these values should be used in the numerical calculations. In many
cases only the so-called pressure-loss coefcient (sometimes in the literature it is also called
pressure-drop or resistance coefcient) is known from experiment. This coefcient is dened
by
p
w
p
e
= 0.5
w
[U
w
]
2
C
pressure
(3.24)
for upstream ow in positive direction with velocity vector pointing normally on the louvre
plane. In the case of ow with low Mach number (incompressible or nearly incompressible uid),
the static pressure drop over the louvre panel is essentially balanced by the drag force (i.e. the
component of the force from the louvre on the uid along the normal of the louvre plane) per unit
area (see [Salvesen, October 1996] for details). Thus for incompressible ow, when U
w
= U
e
, with
upstream velocity vector pointing normally on the louvre plane, the pressure-loss coefcient can
be related to the coefcients C
acc
, C
bend
, and C
f ric
by
C
pressure
= C
acc
(1/(cos
2
) 1) + C
bend
1 + C
f ric
(3.25)
where the relation |U
e
| cos q = U
e
is utilized (confer the example above). Here it is assumed that
the plane determined by the louvre slats for ow in positive direction is the same as the plane
for ow in negative direction. One approach is to set C
acc
equal 1 (this value is supported by
theoretical considerations, cf. [Salvesen, October 1996] ), set C
bend
equal zero or a guessed value
based on experimental results found in the literature (e.g. set C
bend
= 0.11/(/4) as described
above), and set C
f ric
so that the relation above is satised for given values of C
pressure
, C
acc
, and
C
bend
. Another approach is to set both C
acc
and C
bend
equal zero, and set C
f ric
equal C
pressure
. Both
these approaches are expected to give good results when performing simulations. If not even the
pressure-loss coefcient C
pressure
is known, one may use a guessed value for this coefcient
Values of the pressure-loss coefcient are reported in the range from 9.8 to 13.9 for commercially
available louvre panels for industrial use. So a value in the range, say between 10 and 11, is in
many cases expected to be a reasonable estimate for the pressure-loss coefcient.
An example of a setup for a louvre panel is shown below:
NAME "NoName"
POSITION 0.0, 0.0, 0.0 (m)
SIZE 0.0, 1.0, 1.0 (m)
MATERIAL BLUE
NORMAL_VECTOR_SLATS_POSITIVE 1.0, 0.0, 1.0 (-)
NORMAL_VECTOR_SLATS_NEGATIVE 1.0, 0.0, 1.0 (-)
DRAG_COEFFICIENT_ACCELERATION 1.0 (-)
DRAG_COEFFICIENT_BENDING 0.14 (1/radian)
DRAG_COEFFICIENT_FRICTION 9.0 (-)
AREA_POROSITY 0.7 (-)
3.7.19.1 NAME
Character string identifying the louvre panel considered (not used by FLACS).
98 CASD
3.7.19.2 POSITION
Cartesian coordinates of the corner of the louvre panel (the corner with lowest value of the coor-
dinate in each axis direction).
3.7.19.3 SIZE
The louvre panel is assumed to be of rectangular shape. The dimension in each of the axis direc-
tions is given. One dimension should be zero, this shows how the louvre panel is oriented, and
denes the louvre plane. If e.g. the dimension in x-direction is zero, the normal of the louvre
plane points in x-direction. The other two dimensions should be positive.
3.7.19.4 MATERIAL
Colour used when visualizing the louvre panel as part of the geometry considered.
3.7.19.5 Normal vector slats positive
Cartesian coordinates of normal vector of plane determined by the louvre slats (ribs) for ow
in positive direction. This vector need not be specied as a unit vector. But the normal vector
should not be parallel to the normal vector of the louvre plane. This would mean that the louvre
panel is completely blocked by the louvre slats (if e.g. the louvre panel is oriented in x-direction,
to specify the normal vector determined by the louvre slats as (1.0, 0.0, 0.0) would be an illegal
choice). Note that the normal vector need not lie in the xy-, xz-, or yz-plane. To specify the vector
as for example (1.0, 1.0, 1.0) would be a valid choice.
If for example the louvre panel is oriented in x-direction, and the normal vector determined by
the louvre slats is given by (sin(q), 0, cos(q)) (here q is an angle with absolute value less than
90), the interpretation is the following: The angle q is the angle between the slats of the louvre
(assuming that the slats are more or less plane, e.g. of rectangular shape, or if they are curved that
they dene a tangent at the exit on the downstream side) and the normal of the louvre plane (in
this example the positive x-axis). The tangent vector in the xz-plane of the plane determined by
the louvre slats is given by (cos(q), 0, sin(q)). A typical value of the angle q would be -45. This
would correspond to a case where the louvre panel is a shield for rain (assuming that vertically
upwards is in positive z-direction).
3.7.19.6 Normal vector slats negative
Cartesian coordinates of normal vector of plane determined by the louvre slats for ow in neg-
ative direction. Similar comments apply here as those given above for NORMAL_VECTOR_-
SLATS_POSITIVE.
In many cases this vector equals the normal vector for ow in positive direction. This is the case
if e.g. the louvre slats have a rectangular shape. But if the louvre slats are e.g. V-formed the plane
determined by the louvre slats for ow in negative direction is different from the plane for ow
in positive direction.
Note that the normal vector determined by the louvre slats for ow in negative direction should
lie in the plane dened by the normal vector determined by the louvre slats for ow in positive
direction and the normal vector of the louvre plane. If for example the louvre panel is oriented in
x-direction and the normal vector of the plane determined by the louvre slats for ow in positive
direction lies in the xz-plane, then the normal vector for ow in negative direction should also lie
in the xz-plane.
3.7.19.7 Drag coefcient acceleration
Drag coefcient C
acc
of drag due to acceleration of the magnitude of the velocity of the ow. This
coefcient should be zero or positive. A typical value is 1 (this value is supported by theoretical
considerations). Further guidance is given above.
3.7.19.8 Drag coefcient bending
Drag coefcient C
bend
of drag due to bending of the ow. This coefcient should be zero or
positive. The units of this coefcient is (1/radian). Further guidance is given above.
3.7.19.9 Drag coefcient friction
Drag coefcient C
f ric
of drag due to friction. This coefcient should be zero or positive. Further
guidance is given above.
3.7.19.10 Area porosity
An area porosity of the louvre panel is specied. This is not a projected area porosity on to the
louvre plane (in many cases it is not possible to see through the louvre watching normally on it,
i.e. the projected area porosity is zero), but is viewed as the ratio of open area to total area in
a representative cut plane parallel to the louvre plane. If e.g. the louvre consists of rectangular
slats of thickness D uniformly inclined relative to the louvre plane and uniformly spaced with
centre/centre distance 4D, the area porosity is (4D D)/4D = 0.75 (this value is independent
of the angle of inclination). If the frame of the louvre blocks the panel considerably, this may be
taken into account. If the blockage of the frame corresponds to an area porosity
f rame
, the total
effective area porosity in our example will be 0.75
f rame
Note that in the subgrid model for louvre panels, the value of the area porosity affects the drag
force only indirectly. The drag force is determined by the drag coefcients, the conguration of
the louvre slats (determining the deection of the ow downstream of the louvre), and the ow
eld. The value of the area porosity affects how the uxes (of mass, momentum, etc.,) are set
when solving the conservation equations at the louvre plane, and this will affect the ow eld.
If the prole of the louvre slats has a complicated form, it may not be obvious how to estimate
the effective area porosity. If the area porosity varies in different cut-planes parallel to the louvre
plane, it is expected that a cut-plane with a minimum area porosity is the most representative.
The user may perform several simulations varying the value of the area porosity, if one wants to
investigate the sensitivity of the value of the area porosity on the numerical results. The limited
testing in the validation simulations reported in [Salvesen, November 1996], seems to indicate
that the maximum overpressure of an explosion simulation is in general not very sensitive to
changes in the value of the area porosity. In one scenario the maximum overpressure dropped
about 12% when the value of the area porosity was increased from 0.6685 to 1.0.
100 CASD
3.7.20 Grating
Warning:
The grating model should only be used in dispersion and ventilation simulations. The cur-
rent model is designed for ow through the grating, but the important effect on ame accel-
eration along the grating is not handled correctly. It is recommended to use a porous plate in
explosion simulations.
The static pressure-loss coefcient C
spl
is dened by
p = 0.5[U]
2
C
spl
(3.26)
where p is the pressure loss across the grating, is the density, and U is the upstream velocity
component normal to the grating. The component normal to the grating of the force per unit area
from the grating on the uid owing through it, i.e. the drag force per unit area, is assumed to be
equal to the static pressure drop over the grating. The static pressure-loss coefcient is modelled
by:
C
spl
= f
1
(R
e
) f
2
() f
3
(M) (3.27)
where the factor depending on the geometry, characterized by the area porosity , is given by:
f
2
() =
(1
2
)
2
(3.28)
and the factor depending on the upstream Mach number M, and the upstream choking Mach
number M
(upstream Mach no. corresponding to choking at the grating) is given by:

f
3
(M) =
_
M
M
_
1/7
(3.29)
The choking Mach number is related to the area porosity, M
= M
(). This relation is modelled

by:
M
() = 0.675 + 0.325
4
(3.30)
being a curve t to experimental data.
The factor depending on the Reynolds number, R
e
, is modelled as a curve t to experimental
results. The characteristic length used in the Reynolds no. is the diameter of the wire (or of
the rod if the grating is constructed by rods rather than wires). This curve t and experimental
data are shown in the gure below. Further details about modelling drag forces for ow through
grating, are found in [Salvesen & Storvik, 1994].
Figure 3.14: Factor f1 related to Reynolds number
Factor f
1
(R
e
) in pressure-loss coefcient as function of Reynolds number (logarithmic scale).
Both experimental data and a curve-t to these data are shown.
An example of a setup for a louvre panel is shown below:
NAME "NoName"
POSITION 0.0, 0.0, 0.0 (m)
SIZE 0.0, 1.0, 1.0 (m)
MATERIAL RED
AREA_POROSITY 0.7 (-)
CHARACTERISTIC_LENGTH 0.01 (m)
3.7.20.1 Name
Character string identifying the grating considered.
3.7.20.2 Position
Cartesian coordinates of the corner of the grating (the corner with lowest value of the coordinate
in each axis direction).
102 CASD
3.7.20.3 Size
The grating is assumed to be of rectangular shape. The dimension in each of the axis directions
is given. One dimension should be zero, this shows how the grating is oriented. If e.g. the
dimension in x-direction is zero, the normal of the grating points in x-direction. The other two
dimensions should be positive.
3.7.20.4 Material
Colour used when visualizing the grating as part of the geometry considered.
3.7.20.5 Area porosity
Ratio of projected open area of the grating divided by total area. Value between 0 and 1.
3.7.20.6 Characteristic length
Characteristic length used in the Reynolds number, dened as the wire diameter (or similar di-
mension if it is not a wire, but a rod). A typical value of the characteristic length is 0.01m.
3.7.21 Gas monitor region
The GAS_MONITOR_REGION function is used for monitoring amount of fuel inside a module
or another user dened volume during a dispersion simulation. FLACS will write a text le
called rt010100.FUEL which contains a number of columns showing the amount of fuel inside
the dened volume. The columns Q8, Q9, [LFL:UFL], Q6/Q7 are used for risk assessments in
connection to dispersion studies and estimate of explosion severity and ignition probabilities.
If more than one region is needed the user must create a cs010100.MON le. Please see section
Monitor le for details.
3.7.22 Species
Gases not dened in FLACS can be dened manually. When dening a gas, a long list of vari-
ables must be dened. GexCon plan to provide templates for some of the typically used gases
later (e.g. ammonia and chlorine), however, in the meantime the user must dene these gases. To
quickly increase the molecular weight of a gas, for a simulation in which pressure and tempera-
ture does not change too much, simply copy the properties of another dened gas when dening
the USERSPEC_1 gas.
These properties can be found by running a utility program called listspecie.
Linux:
> run9 listspecie1.0 BUTANE
Windows:
>listspecie1.0 BUTANE
3.8 Block menu 103
To dene chlorine gas for a dispersion calculation (ignoring any kind of aerosol and condensa-
tion) simply modify the WFUEL from 0.581e2 to 0.709e2 when specifying the USERSPEC_1 fuel
in SPECIES menu of CASD. If proper compression and change of temperature is needed, mod-
ication of the HFUEL, AENT, BENT and DENT values is also required (this is more complicated).
Several of the parameters have to do with liquid properties of the fuel, this functionality is not
yet available and these are thus irrelevant.
3.8 Block menu
Analyzing the effects of far-eld pressure waves can be of interest in many applications. The
multi-blocks option in FLACS allows performing far-eld pressure waves, i.e. blast waves, stud-
ies. A description of the settings and running commands for the multi-blocks simulation in
FLACS is given here.
3.8.1 Dening a multiblock grid
Assuming a given geometry in the pre-processor CASD, a multiblock grid can be dened. By
default, in the sub-menu Select of the menu Block in the CASD tool bar, a block called super and
a block called FLACS are dened. The block called super needs to be selected to further dene
the multi-block grid. The block named super will be referred to as SuperGrid in the following.
The extents of the SuperGrid domain (menu Grid and Simulation_Volume) denes the total size of
the simulation domain and the number of control volumes in the SuperGrid gives the number of
blocks.
The block FLACS can be selected in the sub-menu Select of the menu Block. The position of this
block into the SuperGrid can be dened in the sub-menu Properties of the menu Block by specifying
the appropriate indices i,j,k of a SuperGrid cell. A grid relative to the block FLACS selected can be
dened in the usual way going in the sub-menu Region of the menu Grid. The following picture
shows a grid with four different blocks:
104 CASD
Figure 3.15: A multi-block grid
Two different types of blocks exist:
FLACS
BLAST
Setup parameters for the simulation (ignition position, initial conditions,...) can only be dened
for a FLACS block. The ignition region must be contained in a single FLACS block and must not
be divided into several blocks.
The BLAST blocks are used to study the development of the pressure front only. The computa-
tions in a BLAST block are faster and require less memory than in a FLACS block. Combustion
processes should not occur in a BLAST block. Therefore, extra FLACS blocks might need to be
dened to contain the whole combustion area.
A block can be added through the sub-menu Add of the menu Block. The position in the SuperGrid
need to be dened. BLAST blocks are added by default. The type of block for the selected block,
can be changed in the sub-menu Properties of the menu Block. Finally, selected blocks can also be
deleted through the sub-menu Delete of the menu Block.
3.8.2 Running a multiblock simulation
Saving the multiblock setup generates a SuperJob number and one job number for each of the
blocks that have been dened. Once the multiblock setup has been saved, CASD can be exited
and the porosities computed for each of the job numbers:
Linux:
> run9 porcalc 50000i
3.9 View menu 105
Windows:
> porcalc.exe 50000i
where 50000i is the job number relative to the block i.
Assuming 500000 is the SuperJob number corresponding to the considered multiblock simulation,
the following command starts the simulation
Linux:
> run9 flacs setup-500000
Windows:
> flacs setup-500000
The le setup-500000 contains a list of job numbers and corresponding block type relative to the
SuperJob number 500000:
VERSION 1.1
$JOBSPEC
BLOCKS =
"500001" "FLACS"
"500002" "BLAST"
"500003" "BLAST"
"500004" "BLAST"
"500005" "BLAST"
$END
3.8.3 General guidelines for multi-block simulations
The grid resolution in the explosion block should be reasonably ne, but it might be necessary to
use a coarser grid than the one generally used in explosion simulations. The block faces must t
perfectly to their neighbors
The geometry should be as simple as possible at the block boundaries, if not the porosity pattern
might become different on the two sides of the boundary.
In BLAST blocks the porosity is either 0 or 1. For stability, the CFLC number should be 0.5 in the
BLAST blocks as well as in FLACS blocks due to the explicit coupling between the blocks.
In BLAST blocks the PLANE_WAVE boundary condition should be used (if not SYMMETRY). This
PLANE_WAVE condition should also be used in the FLACS blocks to minimize the inuence of
the boundaries.
Both monitor point output and eld output (r1- and r3-les) can be generated from a multiblock
simulation. For 2D eld plots and line plots several blocks can be shown in one plot in Flowvis.
However, for volume plots only one block can be shown for each plot.
With BLAST blocks only a selection of the variables of the FLACS blocks exists. Variables like
PROD and FUEL should not be specied for output in BLAST blocks.
3.9 View menu
The View menu in CASD contains commands for manipulating the view.
106 CASD
3.9.1 Print
The Print menu allows exporting a screenshot of the CASD window into different formats:
Postscript
RGB
IV
3.9.2 Examiner viewer and Fly viewer
The default and most widely used viewer is the Examiner viewer. The Fly viewer can be used to
y through the geometry.
3.9.3 XY View, XZ View, and the YZ Views
The option XY View and XZ View display a projection of the geometry in the XY and XZ planes
respectively. The options YZ East View and YZ West View display a projection of the geometry
in the YZ plane along the positive and negative Y-axis respectively.
3.9.4 3D View
The 3D View option displays a default 3D view of the geometry.
3.9.5 Axis
The Axis option turns axis display on and off.
3.9.6 Maximise
The option Maximize maximizes the visible window to display the entire geometry and grid.
3.9.7 Grid display
Three different options are available in the Grid Display menu:
Off: The grid is not displayed. Only the geometry would be displayed.
Working Direction: The grid would be displayed in the working direction only.
All Directions: The grid would be displayed in the three directions.
3.9.8 Annotation
The options in this menu are currently not used.
3.10 Options menu 107
3.9.9 Draw style
Different options are available in this menu:
Off: The geometry will not be displayed.
Wireframe: Only the edges of the objects that compose the geometry would be displayed.
Filled: Surfaces of the objects that compose the geometry would be displayed.
Scenario Wireframe: Only the edges of scenario objects (for example, a fuel region) would
be displayed.
Scenario Filled: Surfaces of scenario objects would be displayed.
3.9.10 LOD and Properties
The LOD (Level Of Details) and properties menus control the details of the geometry displayed.
3.10 Options menu
The user may select certain options under the Options menu in CASD.
3.10.1 Units
The user may choose between the following units for the spatial dimensions: millimetres (mm),
centimetres (cm), decimetres (dm), meters (m), and inches (in); the default option is meters.
3.10.2 Preferences
The users may set preferences for:
General features: scenario template and version of Porcalc
Colours: background colours for examiner viewer and y viewer
Performance: redraw options
3.10.2.1 General
The user may chose between the available scenario templates in a drop down menu. It is essential
that the selected scenario template match the FLACS version to be used in the simulations.
Warning:
New scenario templates will not necessarily work with old scenario les.
By ticking off the option Write polygons (cm le) when saving, the polygons le read by Flowvis
will not be written. In case of a very complex geometry, ticking off this option allows Flowvis
loading the result les faster than if the cm-le were existing. The option Write polygons (cm
le) when saving should therefore be ticked off if the geometry is very complex. See also section
Polygon le.
108 CASD
3.10.2.2 Colours and Performance
Through these two menus the CASD window can be customized.
3.11 Macro menu
The Macro menu contains commands for running and recording macros.
Run: This command processes all the commands on a specied le before turning control
over to the user again. If an error occurs, the processing is interrupted.
Record: This is a toggle button for turning command recording on/off. When turning
recording on, the command requires a macro le name. All subsequent commands are
recorded on the specied le, until the recording is turned off.
Write Geometry: This command writes the macro les needed to dene the open geometry,
including global objects and materials. CASD asks for the path to a directory where the
les are placed. See section and for more information.
3.11.1 Run
To create a geometry from a set of macro les, use the Run command in the Macro menu. Alter-
natively use the command input to read the macro le geometry_name.mcr: macro run
geometry_name
If the project or geometry already exists in the database, an error message is displayed and CASD
exits from the macro.
Macro Description
geometry_name.mcr Creates a new geometry
geometry_name_materials.mcr Creates all materials used in the geometry
material_name.mcr Creates material (one le for each material)
geometry_name_objects.mcr Creates all objects used in the geometry
object_name.mcr Creates object (one le for each object)
geometry_name_instances.mcr Creates all assemblies/instance
Table 3.7: Macro les
If some objects or materials on the macro les already exist in the database, an error message is
displayed, and the object/material is not overwritten.
3.11.2 Record
The option Record is used to save a macro in a 000000.caj.mcr le, where 000000 is a given job
number. The macro le is written simultaneously as the user use CASD, thus this function acts
like a log of the performed actions.
This le can be read as executing a macro with the option Run.
3.12 Help menu 109
3.11.3 Write geometry
The Write Object command in the Macro menu writes a macro le that denes the open object.
CASD asks for the path to a directory where the macro le is to be placed.
Note that this macro le must be started in the CASD main window.
The Write Geometry command in the Macro menu causes CASD to write a complete set of macro
les for the open geometry. The les include macro les which create the project, geometry, all
materials needed, all objects needed in addition to the assemblies/instances. The les are listed
in table Macro les created by the Write Macro command.
Figure 3.16: The macro le hierarchy
Note that the macro le format is not intended as a backup format. Future versions of CASD may
not be backwards compatible with the menu structure and commands in the current version.
The Copy command in the Database menu can be used to make a copy of a geometry within the
same project. The macro les created by the Write Geometry command in the Macro menu can be
used for copying the geometry from one database to another. They can also be used for copying
one geometry into another (existing) geometry.
To copy one geometry (geo1) into another (geo2) in the same database, open geo1 and execute
the Write Geometry command in the Macro menu. Exit from geo1 and open geo2. Create a new
assembly and execute the macro:
*
macro run geometry_name_instances
Problems occur if geo2 contains local object(s) with the same name(s) as in geo1.
3.12 Help menu
The purpose of the Help menu is to provide the user with relevant information concerning the
general use of FLACS and the active FLACS licence.
110 CASD
3.12.1 Online help
This menu opens this Users manual in the default HTML browser on the computer (see the
Options menu for information on how to change the default HTML browser).
3.12.2 Quick reference
This menu opens a window summarizing various controls
Figure 3.17: Quick References
3.12.3 Licence terms
This menu opens a window that contains the FLACS licence terms.
3.12.4 About CASD
This menu displays the FLACS splash screen with information about the version of FLACS and
the version of CASD.
3.13 Potential bugs or problems with CASD
This chapter contains a list of potential bugs or problems with CASD, and some possible
workarounds.
3.13 Potential bugs or problems with CASD 111
3.13.1 Problem with macro les and local objects
Using the WRITE GEOMETRY option in CASD does not correctly write local objects. The trans-
formation matrix is missing, resulting in a faulty geometry when reading macro les back into
CASD.
3.13.2 CAD import
Geometry can be imported from CAD programs. Please refer to section geo2acs.
3.13.3 Heavy hydrocarbons C5H12 and upwards
Several vapours from hydrocarbons that are heavier than butane can be modelled, but combus-
tion properties have been copied from butane. Enthalpies for Dodecane have been copied from
Decane, and this may lead to strange results if burning Dodecane.
3.13.4 Reading old dump le
In previous FLACS versions the freezing point of water (0 C) was dened to be 273 K, whereas
in the newest version this has been changed to 273.15 K. For wall temperatures it has always been
assumed that 0 C translates to 273.15 K. The gas constant, , has similarly been changed from
8314 J/(kmol K) to 8314.472J/(kmol K).
If an old dump-le is read in the new version of FLACS, small inconsistencies/disturbances may
be seen in the temperature eld of this reason.
112 CASD
Chapter 4
Flacs simulator
114 Flacs simulator
This chapter describes various aspects the CFD simulator Flacs:
how to start and stop a simulation
how to monitor the progress of a running simulation
a list of input les
a list of output les
a list of output variables
4.1 Overview
See Getting started for a detailed description of how to install FLACS and the basic steps to get
started using it.
On Linux it is recommended that the user denes an alias for running FLACS v9.0 programs:
> alias run9 /usr/local/GexCon/FLACS_v9.0/bin/run
On Windows a desktop icon for the Run Manager is created during the installation.
4.2 The Run Manager
Linux:
> run9 runmanager
Windows:
> "C:Program Files\GexCon\FLACS_v9.0\bin\runmanager"
4.2 The Run Manager 115
Figure 4.1: The FLACS Runmanager
4.2.1 Starting simulations
See Running FLACS for the basic introduction on how to get started using the FLACS package.
The FLACS simulator (Flacs) can be started from the command line or from the Runmanager.
The command to start a FLACS simulation from the command line:
> run9 runflacs 010101
4.2.2 Monitoring simulations
On Linux the user can use the tail command to monitor the progress of a FLACS simulation.
> tail -f tt010101
116 Flacs simulator
4.2.3 Check list for running simulations
The sequence of tasks involved in a general FLACS simulation includes:
1. Stating your problem
2. Dening sensitivities, or parameter variation
3. Dening and verify the geometry
4. Dening and verify the grid
5. Calculating and verify the porosities
6. Dening and verify the scenario
7. Running the simulations
8. Checking the simulation log les for errors
9. Presenting the results
10. Storing all data for later use
Remarks:
It is important to check the correctness of all input parameters.
Below is a recommended check list for basic QA of the simulation set-up:
1. Avoid large Courant numbers (CFLV and CFLC)
2. Locate ignition in an unblocked control volume
3. Locate monitors in unblocked control volumes
4. Dene realistic discharge parameters for leaks
5. Verify vent areas
6. Verify gas composition
7. Avoid strong transient wind build-up
8. Check disk space and access rights
9. The required les:
Grid le
Obstruction le
Porosity le
Scenario le
4.3 Running several simulations in series
An efcient way to handle many simulations is to use run scripts (text le with commands).
Running many simulations in parallel may exhaust the computer memory and actually increase
the total computation time for the simulations. On Linux if the user wants to stop a running
process the kill command can be used, the ps command will report the process id of all running
processes. On Windows the Task Manager can be used.
Remarks:
See section Linux Quick Reference for useful Linux commands and examples on how to run
and monitor FLACS commands effectively on a Linux system.
4.3 Running several simulations in series 117
4.3.1 Create and use run scripts
Flacs simulations are easy to run in batch, in parallel or series, using the Run mananger, however
if the user wants more control a run script can be created using a text editor, the script le my_-
runfile could look like this, on Linux:
#!/bin/csh -f
# Set up an alias for running the FLACS simulator:
alias my_runflacs /usr/local/GexCon/FLACS_v9.0/bin/run_runflacs
# Run the simulations in series:
my_runflacs 010101
my_runflacs 010102
my_runflacs 010103
Make the script le my_runfile executable:
> chmod u+x my_runfile
Run the script in the background, messages are sent to the le my_listfile:
> ./my_runfile >& my_listfile &
Similarly a bat script, my_runfile.bat can be created on Windows:
@echo off
setlocal
rem Set up an alias for running the FLACS simulator:
set my_runflacs="c:\program files\gexcon\flacs_v9.0\bin\runflacs.exe"
rem Run the simulations in series:
%my_runflacs% 010100
Run the script, messages are sent to the le my_listfile:
> my_runfile.bat > my_listfile
4.3.2 Optimizing computer loads
In general Flacs simulations will use 100% of a single CPU core if available, thus optimal use of a
computer for runninig simulations is to start as many simulations in parallel equal to the number
of CPU cores available. It is however necessary to keep the total memory consumption within
physical computer memory. If this is not done the computer will start to use virtual disk memory,
which is signicanly slower than physical memory. This will result in longer simulation time.
As a rule of thumb a simulation computer should have 2GB of memory per CPU core.
4.3.3 Stopping simulations
Flacs simulations can be stopped prematurely either by using the Task Manager (Windows) or
using a combination of the command line utilities ps and kill on Linux.
To nd the Flacs simulation process ID (PID) run the following command on Linux:
118 Flacs simulator
> ps -edlaf|grep flacs
This will list all FLACS related programs running on the computer. The PID is the number found
in the 4th column. Also note the user name of the process, which is found in the 3rd column. The
process can then be stopped using the following command:
> kill 1234
where 1234 is the PID.
Alternatively the simulation can be stopped by using the Runtime simulation control le (cc le).
Adding the following line to the cc le will stop the simulation at the time 123.4 sec:
TSTOP 123.4
4.4 Output variables in FLACS
This chapter describes the output variables that can be selected for output in selected monitor
points or over selected monitor panels (under Single Field Scalar Time Output on the Scenario
menu in CASD), or throughout the entire calculation domain (under Single Field 3D Output on
the Scenario menu in CASD).
The normal output variables in FLACS are:
Name Dim Units Description
H 1 (J/kg) Enthalpy
FUEL 1 (-) Fuel mass fraction
FMIX 1 (-) Mixture fraction
FVAR 1 (-) Mixture variance
K 1 (m2/s2) Turbulent kinetic
energy
EPK 1 (1/s) Turbulence ratio
EPS 1 (m2/s3) Dissipation rate of
turbulent kinetic
energy
GAMMA 1 (-) Isentropic gas
constant
LT 1 (m) Turbulent length
scale
MU 1 (kg/(ms)) Effective dynamic
viscosity
OX 1 (-) Oxygen mass
fraction
P 1 (barg) Pressure
PMAX 1 (barg) Maximum pressure
DPDT 1 (bar/s) Rate of pressure rise
PIMP 1 (Pas) Pressure impulse
PIMPMAX 1 (Pas) Maximum pressure
impulse
PROD 1 (-) Combustion product
mass fraction
RFU 1 (kg/(m3s)) Combustion Rate
4.4 Output variables in FLACS 119
RET 1 (-) Turbulent Reynolds
number
FMOLE 1 (m3/m3) Fuel mole fraction
FDOSE 1 (m3/m3s) Fuel mole fraction
DOSE
RHO 1 (kg/m3) Density
T 1 (K) Temperature
TURB 1 (m/s) Turbulence velocity
TURBI 1 (-) Relative turbulence
intensity
VVEC 3 (m/s) Velocity vector
U 0 (m/s) Velocity component
x-direction
V 0 (m/s) Velocity component
y-direction
W 0 (m/s) Velocity component
z-direction
UVW 1 (m/s) Velocity value
UDRAG 1 (Pa) Drag component
x-direction
VDRAG 1 (Pa) Drag component
y-direction
WDRAG 1 (Pa) Drag component
z-direction
DRAG 1 (Pa) Drag value
DRAGMAX 1 (Pa) Maximum drag
value
UDIMP 1 (Pas) Drag-impulse
component
x-direction
VDIMP 1 (Pas) Drag-impulse
component
y-direction
WDIMP 1 (Pas) Drag-impulse
component
z-direction
DIMP 1 (Pas) Drag-impulse value
DIMPMAX 1 (Pas) Maximum
drag-impulse value
UFLUX 1 (kg/(m2s)) Flux component
x-direction
VFLUX 1 (kg/(m2s)) Flux component
y-direction
WFLUX 1 (kg/(m2s)) Flux component
z-direction
FLUX 1 (kg/(m2s)) Flux value
UMACH 1 (-) Mach number
component
x-direction
VMACH 1 (-) Mach number
component
y-direction
120 Flacs simulator
WMACH 1 (-) Mach number
component
z-direction
MACH 1 (-) Mach number value
CS 1 (m/s) Sound velocity
TAUWX 1 (-) Wall shear force
tauwx
TAUWY 1 (-) Wall shear force
tauwy
TAUWZ 1 (-) Wall shear force
tauwz
NUSSN 1 (-) Nusselt number
RESID 1 (-) Mass residual in
continuity equation
ER 1 (-) Equivalence ratio
ERLFL 1 (-) Equivalence ratio,
LFL
ERNFL 1 (-) Equivalence ratio,
normalized
ammable range
EQ 1 (-) Equivalence ratio,
nite bounded
EQLFL 1 (-) Equivalence ratio,
LFL
EQNFL 1 (-) Equivalence ratio,
normalized
ammable range
TMOLE 1 (m3/m3) Toxic mole fraction
TCONS 1 (mg/m3) Toxic concentration
TDOSE 1 (mg/m3minute) Toxic dose
PROBIT 1 (-) Toxic probit
PDEATH 1 (-) Probability of death
as function of toxic
probit
Table 4.1: Output variables in FLACS
The panel output variables in FLACS are:
Name Dim Units Description
PPOR 1 (-) Panel average area
porosity
PP 1 (Pa) Panel average
pressure
PPIMP 1 (Pas) Panel average
pressure impulse
PDRAG 1 (Pa) Panel average drag
PDIMP 1 (Pas) Panel average drag
impulse
Table 4.2: Panel output variables in FLACS
The following sections gives a description of the most commonly used output variables.
4.4.1 Mass fraction of fuel: FUEL
This is the mass fraction of fuel in the mixture of fuel, air, and combustion products. The fuel
may be a mixture of several components, such as hydrocarbons and hydrogen. Plots of FUEL are
useful for displaying the fuel cloud.
4.4.2 Pressure: P
This is the static overpressure (bar g). In a owing uid, the total pressure is the sum of the
static pressure and the dynamic pressure. The static pressure is isotropic, whereas the dynamic
pressure, caused by the relative motion of the uid, is anisotropic. A pressure transducer placed
in a ow eld will in general measure the static pressure and a certain portion of the dynamic
pressure, depending on the orientation of the face of the pressure transducer relative to the ow
direction. Head on measurements give the total pressure, whereas side on measurements give
the static pressure.
4.4.3 Pressure impulse: PIMP
PIMP is the time integral of the pressure:
I
p
=
_
t2
t1
pdt (4.1)
The pressure impulse is simply the area below the pressure-time curve, and since it is the prod-
uct of pressure and time it holds information about both the amplitude and the duration of the
pressure-time curve.
4.4.4 Mass fraction of combustion products: PROD
This is the ratio of mass (kg) of combustion products per unit mass (1 kg) of the total mixture
of fuel, air and combustion products for each control volume. The combustion products consist
of carbon-dioxide and water vapour. Plots of PROD are useful for displaying the ame (or more
correctly the burnt volume).
See sections Denitions and gas thermodynamics, Stoichiometric reaction and Gas composition
and volume for more information on the reactions that convert fuel onto combustion products.
4.4.5 Gas density: RHO
The gas density is:
p = RT (4.2)
This is the uid mass (kg) per unit volume (1 m3). The equation of state gives the relation between
pressure density and temperature.
122 Flacs simulator
4.4.6 Gas temperature: T
This is the absolute temperature (K) of the uid. See RHO above for a description of the relation
between pressure density and temperature. The temperature may be increased by compression
which converts mechanical energy into thermal energy, and by combustion which converts chem-
ical energy into thermal energy.
4.4.7 Velocity vector: VVEC
This is the entity which gives the three velocity components of the time averaged uid ow. The
energy contained in the temporal uctuations of the ow which are not captured using a given
spatial and temporal resolution is handled by a turbulence model. VVEC consists of the three
components U, V, and W. If you are editing the cs-le manually, always remember to include the
components if you have specied VVEC for output (CASD includes them automatically).
4.4.8 Drag value: DRAG
The drag value (in FLACS dened as drag force per unit area) is proportional to the dynamic
pressure for the uid ow. The expression for the dynamic pressure is:
p
dyn
= u
2
/2 (4.3)
An obstacle submerged in a uid ow will interact with the uid, thereby a drag force results.
The drag force may be measured in experiments and if the Reynolds number is high, the ratio
drag force / dynamic pressure is constant:
Drag coefcient C
D
= (F
D
/A)/(u
2
/2)
Drag force F
D
= C
D
(u
2
/2)A
The drag value is calculated by assuming the drag coefcient C
D
= 1 and the cross-section area
A = 1 , with this denition the drag value is the same as the dynamic pressure.
4.4.9 Drag-impulse value: DIMP
Drag-impulse value is the time integral of the dynamic pressure:
I
Pdyn
=
_
t2
t1
p
dyn
dt (4.4)
The drag-impulse value is equivalent to the pressure impulse, with the difference that the dy-
namic pressure is being integrated instead of the static pressure.
4.4.10 Equivalence ratio: ER
The equivalence ratio, ER, is a measure of the concentration of fuel compared to the stoichiomet-
ric concentration, i.e. ER equals unity at stoichiometric concentration. If (F/O) is the ratio of fuel
to oxygen, the equivalence ratio is dened as follows:
ER = (F/O)/(F/O)
stoichiometric
(4.5)
For zero fuel, ER equals zero and for pure fuel ER goes to innity.
4.4.11 Equivalence ratio %LFL: ERLFL
This is a measure for concentration of fuel compared to the LFL concentration, where LFL is the
lower ammable limit. The LFL value normally varies with gas type and oxygen concentration
(again depending on the amount of inert gases) in the mixture. In FLACS the fuel is always
mixed with air which has a preset oxygen concentration, so only the variation of LFL with gas
type remains. The denition of ERLFL is as follows:
ERLFL = 100 ER/ERLFL% (4.6)
4.4.12 Equivalence ratio, normalized DFL: ERNFL
The ammable range is dened to be from LFL to UFL, where LFL is the lower ammable limit
and UFL is the upper ammable limit. ERNFL is dened as follows:
ERNFL = (ER ERLFL)/(ERUFL ERLFL) (4.7)
ERNFL is zero at LFL and one at UFL.
4.4.13 Equivalence ratio, nite bounded: EQ
This is a measure for concentration of fuel similar to the equivalence ratio (see ER above). Say
that (F/O) is the ratio of fuel to oxygen, then the nite bounded equivalence ratio is dened as
follows:
EQ = (F/O)/[(F/O) + (F/O)
stoichiometric
] (4.8)
At stoichiometric concentration EQ equals 1/2. For zero fuel EQ equals zero and for pure fuel EQ
equals one.
4.4.14 Equivalence ratio, %LFL: EQLFL
This is a measure for concentration of fuel compared to the LFL concentration, where LFL is the
lower ammable limit. The LFL value normally varies with gas type and oxygen concentration
(again depending on the amount of inert gases) in the mixture. In FLACS the fuel is always
mixed with air which has a preset oxygen concentration, so only the variation of LFL with gas
type remains. The denition of EQLFL is as follows:
EQLFL = 100 EQ/EQLFL% (4.9)
124 Flacs simulator
4.4.15 Equivalence ratio, normalized DFL: EQNFL
Equivalence ratio, normalized ammable range. The ammable range is dened to be from LFL
to UFL, where LFL is the lower ammable limit and UFL is the upper ammable limit. EQNFL is
dened as follows:
EQNFL = (EQEQLFL)/(EQUFL EQLFL) (4.10)
EQNFL is zero at LFL and one at UFL.
4.4.16 Panel average pressure: PP
This is the average pressure (Pa) acting on the panel surface in the perpendicular direction. It is
the sum of the directional pressure forces acting on the panel divided by the net surface area of
the panel (also accounting for the area porosity for each control volume). The sign of PP indicates
the direction of the total force, +/- along the positive/negative direction respectively.
The panel average pressure PP is calculated in the following way: For each control volume face
(CV face) covering the area of the panel considered, the net static pressure is calculated (static
pressure on the negative side relative the coordinate axis minus the static pressure on the positive
side). This net static pressure is then integrated over the blocked area of the panel and the integral
is then divided by the total blocked area of the panel, to give the panel average pressure PP. If
e.g. the area porosity of the panel is zero, the panel is totally blocked and the integral is divided
by the total area of the panel. And if e.g. the panel is totally open (area porosity one, no blocked
area), panel average pressure PP vanishes (is zero), since there is no blocked area to integrate
over (integral with respect to net static pressure over blocked area is zero). In general the panel
can be porous (partially blocked and partially open).
4.4.17 Panel average porosity: PPOR
This is the average pressure porosity, it is the amount of open surface on the panel divided by the
total panel area. Output of PPOR may be used to verify when the panel yields.
4.4.18 FMOLE and FDOSE
FMOLE is the mole, or volume, fraction of the gas in the gas/air mixture, and FDOSE is the inte-
grated (accumulated) FMOLE. For 60s dose for monitor points, you can simply export FMOLE to
ASCII-format using r1-le, import to excel, and subtract the FDOSE value of time-60s. The FDOSE
variable can also be selected for 3D output. If the user wants a contour plot of the 60s exposure
one or more plots must be selected for each 60s period. Using the r3le utility program one can
then generate new r3-les with FDOSE(time)-FDOSE(time-60s).
The utility programme r3le can generate the so-called dose (i.e. exposure) output:
FDOSE(t) =
_
t
0
FMOLE(t)dt
dose(t2) = FDOSE(t2) FDOSE(t1)
(dose/time)(t2) = dose(t2)/(t2 t1)
The times (t2 and t1) are taken from the output times with a certain integer interval given by the
option dose= or dose/time=:
dose=2 means that t2-t1 = 2DTPLOT, output is then dose(t)
dose/time=2 is similar, but you get (dose/time)(t) output
The following starting point is assumed:
FLACS result les in the current directory, and
FDOSE output at regular time intervals (e.g. DTPLOT = 60)
To generate the dose (i.e. exposure) output in a separate directory:
1. create the directory work and enter into it
2. run the r3le utility (assuming the job number is 010100)
> mkdir work
> cd work
> run9 r3file1.3 ../r3010100.dat3 format=r3file dose=1 name=NFDOSE force
Now the following les in the work directory can be found:
a3010100.NFDOSE
cgNFDOSE.dat3 -> ../cg010100.dat3
coNFDOSE.dat3 -> ../co010100.dat3
cpNFDOSE.dat3 -> ../cp010100.dat3
csNFDOSE.dat3
r3NFDOSE.dat3 -> a3010100.NFDOSE
The results can be viewed using Flowvis.
Warning:
region= cannot yet be used with the dose output.
4.4.19 Variables for toxic substances
It is possible to model the effect of toxic substances with FLACS. The toxic component is specied
in the section GAS_COMPOSITION_AND_VOLUME on the scenario menu.
Remarks:
The toxic substance variables are only available using the default+1 scenario template.
The parameter TOXIC_SPECIFICATION can be specied in one of the following ways:
Selecting a predined substance
Creating a user specied toxic data le, and specifying the substance name
Specifying the substance properties directly
The following predened substances are available:
126 Flacs simulator
Substance a b n Formula Molar
mass
(g/mol)
Boiling
point (C)
Acrolein -4.1 1 1 C3H4O 56.06 53
Acrylonitrile
-8.6 1 1.3 C3H3N 53.06 77
Allyl
alcohol
-11.7 1 2 C3H6O 58.08 97
Ammonia -15.6 1 2 NH3 17.0306 -33.34
Azinphos-methyl
-4.8 1 2
C10H12N3O3PS2
317.32 200
Bromine -12.4 1 2 Br2 159.808 58.85
Carbon
monoxide
-7.4 1 1 CO 28.010 -192
Chlorine -15.6 1 2 Cl2 70.906 -34.4
Ethylene
oxide
-6.8 1 1 C2H4O 44.05 10.7
Hydrogen
chloride
-37.3 3.69 1 HCl 36.46 -85.1
Hydrogen
cyanide
-9.8 1 2.4 HCN 27.03 26.0
Hydrogen
fluoride
-8.4 1 1.5 HF 20.01 19.54
Hydrogen
sulfide
-11.5 1 1.9 H2S 34.082 -60.28
Methyl
bromide
-7.3 1 1.1 CH3Br 94.94 3.56
Methyl
isocyanate
-1.2 1 0.7 C2H3NO 57.1 39.1
Nitrogen
dioxide
-18.6 1 3.7 NO2 46.01 21.1
Parathion
-6.6 1 2
C10H14NO5PS
291.3 375
Phosgene -10.6 2 1 CCl2O 98.92 8
Phosphamidon
-2.8 1 0.7
C10H19ClNO5P
299.70 162
Phosphine
-6.8 1 2 PH3 34.00 -87.8
Sulphur
dioxide
-19.2 1 2.4 SO2 64.07 -10
Tetraethyl
lead
-9.8 1 2 C8H20Pb 323.44 84
Table 4.3: Predened toxic substances
The user can specify one of these substances as the toxic component in scenario section GAS_-
COMPOSITION_AND_VOLUME:
GAS_COMPOSITION_AND_VOLUME
POSITION_OF_FUEL_REGION 0 0 0
DIMENSION_OF_FUEL_REGION 0 0 0
TOXIC_SPECIFICATION "Ammonia"
VOLUME_FRACTIONS
TOXIC 1
EXIT VOLUME_FRACTIONS
EQUIVALENCE_RATIOS_(ER0_ER9) 1E+30 0
EXIT GAS_COMPOSITION_AND_VOLUME
For a pure toxic release, specify ER0 = 1E+30, although the toxic substance might not be com-
bustible. This is not formally correct, but is a workaround and gives a mass fraction of one.
The alternative is to write a user dened toxic data le (./my_toxic_data.dat):
: substance , a , b , n , formula , MolarMass(g/mol) , BoilingPoint(C)
"Acrolein" -4.1 1. 1. "C3H4O" 56.06 53.
="Acrylaldehyde"
Attention:
The rst sign in the toxic data le must not be space or tab. Comments starts with #, ! or
(space).
The scenario speciaction should then look like this (in this case there is also a fraction of methan
in the gas mixture):
POSITION_OF_FUEL_REGION 0 0 0
DIMENSION_OF_FUEL_REGION 0 0 0
TOXIC_SPECIFICATION "Acrolein, data_file=./my_toxic_data.dat"
VOLUME_FRACTIONS
METHANE 1
TOXIC 9
EXIT VOLUME_FRACTIONS
EQUIVALENCE_RATIOS_(ER0_ER9) 1E+30 0
EXIT GAS_COMPOSITION_AND_VOLUME
The substance properties can be specied directly in the GAS_COMPOSITION_AND_VOLUME sec-
tion:
TOXIC_SPECIFICATION "probit_constants=-4.1,1,1,molar_mass=56.06"
Descrition of the variables used to monitor toxic substances can be found in the section Output
variables in FLACS.
The keywords for TOXIC_SPECIFICATION are:
substance =name of toxic substance (or formula)
probit_constants =a,b,n
molar_mass =M
data_le =name of the datale (default toxic_data.dat)
The TOXIC_SPECIFICATION string is stripped of spaces and converted to lowercase before
parsing. The name of the datale and formula are case sensitive.
Warning:
The implemented models for toxic components are limited to substances with purly gaseous
behaviour. Toxic substances with a boiling point above ambient temperature will typically
spread as a mist and their toxic effect could for instance require direct skin contact. Such
effects are not currently handled by FLACS.
128 Flacs simulator
Attention:
A gas cloud with toxic components should not be specied in the GAS_COMPOSITION_-
AND_VOLUME scenario section. To specify a given mass fraction in a predened gas cloud
use the cloud interface.
4.4.19.1 Toxic dose TDOSE
Toxic dose TDOSE is dened as:
TDOSE(mg/m
3
minute) =
_
t
0
C
n
dt
For constant C:
TDOSE(mg/m
3
minute) = C
n
t
4.4.19.2 Toxic probit function PROBIT
Toxic probit function PROBIT is dened as:
PROBIT() = a + b ln(TDOSE)
4.4.19.3 Probabilty of death as function of toxic probit PDEATH
Probabilty of death as function of toxic probit PDEATH is dened as:
PDEATH() = 0.5
_
1 + er f
_
PROBIT 5
2
__
where:
er f (x) =
2
_
x
0
e
t
2
dt
4.4.20 Modifying names and units for output variables
It is possible to use alternative names for some output variables in FLACS. For example the
old DRAG name may be substituted by PDYN (dynamic pressure), and the units for the pressure
variables may be set in the VARIABLE_DEFINITION section (Pa, hPa, kPa, barg or mbarg):
New Old Units Description
P P (Pa) Static pressure
PIMP PIMP (Pas) Static pressure
impulse
UPDYN UDRAG (Pa) Dynamic pressure
x-component
VPDYN VDRAG (Pa) Dynamic pressure
y-component
WPDYN WDRAG (Pa) Dynamic pressure
z-component
4.5 Files in FLACS 129
PDYN DRAG (Pa) Dynamic pressure
UDIMP UDIMP (Pas) Dynamic pressure
impulse
x-component
VDIMP VDIMP (Pas) Dynamic pressure
impulse
y-component
WDIMP WDIMP (Pas) Dynamic pressure
impulse
z-component
DIMP DIMP (Pas) Dynamic pressure
impulse
PTOT N/A (Pa) Total pressure
PTOT=P+PDYN
PRIMP N/A (Pas) Total pressure
impulse
Table 4.4: Modied variables names.
Remarks:
N/A = not available
4.5 Files in FLACS
Input and output data for FLACS are stored in les. The name of each le consists of two parts
separated by a dot (.). The rst part of the le name contains a two-letter prex followed by the
6-digit job number. The second part of the le name, called the le type, contains a prex of one
or more letters followed by one or more digits (dat3 for most of the les, n001 etc. for the leak
data les).
Summary of the les used by FLACS:
File name Contents of le
cs000000.dat3 Scenario
cg000000.dat3 Grid
cp000000.dat3 Porosities
co000000.dat3 Obstructions
cc000000.dat3 Runtime simulation control
cn000000.dat3 Time dependent CFL-numbers
cl000000.n000 Time dependent leak data
r1000000.dat3 Scalar-time output
r3000000.dat3 Field output
rt000000.dat3 Simulation log
tt000000 Simulation log, terminal printout
rd000000.n000 Simulation dump
rx000000.n000 Simulation save, temporary le created by
FLACS
Table 4.5: Summary of les in FLACS.
Summary of the identication numbers used in the les:
130 Flacs simulator
File name Meaning of le type digits
??000000.dat3 Number of dimensions (1, 2, or 3)
cl000000.n000 Number identifying a leak
rd000000.n000 Number identifying a Dump or Load
rx000000.n000 Number identifying a variable
Table 4.6: Identication numbers
4.6 Input les to FLACS simulations
This section summarizes the various input les to FLACS.
4.6.1 Obstruction le
Denes the geometry, contains a list of primitives (boxes and cylinders) from a CASD database,
generated by CASD, required by Porcalc and Flowvis.
File name template: co000000.dat3
For briefness this le may be called the co-le or obstruction le hereafter. It is a binary le
and will be generated when starting the Porcalc programme from CASD. It contains a list of
geometrical primitives (boxes and cylinders) which are extracted from the CASD database. The
co-le is not accessed by FLACS. The Porcalc programme will require read access for the co-le,
and it is also used by Flowvis when geometry is specied for a plot.
4.6.2 Grid le
Denes the computational mesh, generated by CASD, required by Porcalc.
File name template: cg000000.dat3
For briefness this le may be called the cg-le or grid le hereafter. It is a binary le and will
be created when using the grid denition menu in CASD. The grid le stores the computational
grid, i.e. the discrete representation of the simulation volume. The simplest form of a grid is a
uniform grid, where all control volumes have the same size and shape.
The gure below shows a two-dimensional section of a uniform grid, and illustrates the position
of the internal nodes and the boundary nodes relative to the grid lines.
4.6 Input les to FLACS simulations 131
Figure 4.2: Section of a grid showing internal and boundary nodes
The numbering scheme for the nodes and grid lines may need some attention. In FLACS and
CASD one type of numbering is used, whereas in Flowvis a different type of numbering has
been adapted. The explanation to this is that FLACS and CASD have kept the original numbering
scheme which was developed initially, and at a later time when Flowvis was developed a new
and more intuitive numbering scheme was selected. Below is a table which shows the details of
the two numbering schemes:
Numbering schemes for nodes and grid lines:
Sizes Internal nodes Boundary
nodes
Grid lines Used in
N = NX, NY, NZ 2 to N-1 1 and N 2 to N Flacs and CASD
M = N-1 1 to M 0 and M+1 1 to M+1 Flowvis
Table 4.7: Numbering schemes for nodes and grid lines.
The Flowvis number is one less than the FLACS and CASD number. This should be kept in mind
when setting up plot domains in Flowvis. (see section Flowvis for details).
The total number of nodes and control volumes will be the same for both numbering schemes.
Below is a gure showing a three-dimensional grid on a typical offshore module geometry.
In the gure above the grid is extended outside the module walls and also above the module
roof, this is because it was used for a multi-block simulation.
132 Flacs simulator
In typical semi-conned geometries like an offshore module one may create a grid only covering
the interior including the outer walls of the module, but it is recommended to extend the grid
to improve the boundary conditions. In the cases of very open geometries and in multi-block
simulations one should always extend the grid well outside the main explosion area.
More information can be found in sections Porcalc and Calculate porosities.
4.6.3 Polygon le
Denes the polygon model used by CASD, used by Flowvis if present.
4.6.4 Header le
Denes the co, cg and cm les to be used by CASD.
4.6.5 Porosity le
Denes the porosities for each grid cell, calculated by the program Porcalc from the co and cg-
les, required by Flacs.
File name template: cp000000.dat3
For briefness this le may be called the cp-le or porosity le hereafter. It is a binary le and will
be generated when using the Porcalc programme.
It contains data which are calculated based on the geometry and the grid. FLACS will stop if the
cp-le is not accessible for reading. The cp-le is quite large, the size (in bytes) may be calculated
as follows:
SIZE = 10 4 NX NY NZ
In order to save space the le may be deleted and regenerated when needed using Porcalc, or the
size may be reduced using the unix command compress, remember to uncompress the le before
using Flacs, CASD or Flowvis.
4.6.6 Scenario le
Denes the general scenario (monitor points, output variables, fuel region, vents, ignition posi-
tion, etc.), required by Flacs.
File name template: cs000000.dat3
For briefness this le may be called the cs-le or scenario le hereafter. It is a text le and will be
created when using the scenario denition menu in CASD.
The term scenario was dened in the introduction to this Users Guide, briey as being the set of
parameters which may be used to control the behaviour of a given FLACS simulation.
The rst line of the scenario le identies the le format, for FLACS v9.0 this is set to the following
text string: "VERSION 0.5". This must not be changed manually since it will be used by FLACS
to determine how to read and interpret the le.
Only the most recent format is described here.
The scenario le contains several sections, which are structured as follows:
SECTION_NAME
A section may contain several lines ...
EXIT SECTION_NAME
Sections in the scenario le:
SINGLE_FIELD_VARIABLES
MONITOR_POINTS
PRESSURE_RELIEF_PANELS
SINGLE_FIELD_SCALAR_TIME_OUTPUT
SINGLE_FIELD_3D_OUTPUT
SIMULATION_AND_OUTPUT_CONTROL
BOUNDARY_CONDITIONS
INITIAL_CONDITIONS
VOLUME_FRACTIONS
IGNITION
LEAKS
OUTLET / VESSEL
WATERSPRAYS
LOUVRE_PANELS
GRATING
MONITOR_VOLUMES
4.6.7 Setup le
This is an optional le used to set certain user-dened variables, such as constants in the com-
bustion model. FLACS Run manager detects the le automatically if the le exists in the same
directory as the regular job les. When using the run9 runflacs command the setup le must
be supplied as argument #2.
> run9 runflacs 010101 cs010100.SETUP
The setup le may contain the following so-called namelists: JOBSPEC, SETUP and
PARAMETERS. Note that the $ in the name list must be positioned in column 2 (only on cer-
tain machine types). The namelists may be entered in any order or may alternatively be left out
entirely.
VERSION 1.1
$JOBSPEC
...
$END
$SETUP
...
$END
$PARAMETERS
...
$END
4.6.7.1 The JOBSPEC namelist
The available keywords in JOBSPEC and their default values are summarized below:
VERSION 1.1
$JOBSPEC
BLOCKS = " ", " "
134 Flacs simulator
IGNITION = " "
SYNC_OUTPUT = .TRUE.
KEEP_OUTPUR = .FALSE.
RESET_LOAD = .TRUE.
$END
The meaning of the keywords is as follows:
Keyword Description
BLOCKS List of job numbers and block type for each
block:
there should normally be only one
FLACS block, but it is possible to use
more than one
there may be zero or more BLAST
blocks
a maximum of 10 blocks are allowed
IGNITION Job number for the block where ignition shall
occur:
this must be a FLACS block
SYNC_OUTPUT Synchronize output (r3-le) so that all blocks
write at the same time:
plots at same time can thereby be
shown properly in Flowvis
KEEP_OUTPUT Keep old results on existing r1-le and r3-le
(append new results):
.TRUE. if you want to run a
continuation run
.FALSE. if you restart a new scenario
RESET_LOAD Set equal to .TRUE. if you want to reset
initial condition at LOAD time:
useful for starting an explosion after a
dispersion
4.6.7.2 The SETUP namelist
The available keywords in SETUP and their default values are summarized below:
VERSION 1.1
$SETUP
TIME_STEPPING = "STRICT"
EQUATION_SOLVER = "BI_CGSTAB"
MASS_CONSERVATION = "BEST"
COMBUSTION_MODEL = "BETA3"
TURBULENCE_MODEL = "K-EPS"
DIFFUSION_MODEL = "LINEAR"
GASDATA_MODEL = "DEFAULT"
AIR = "NORMAL"
AMBIENT_PRESSURE = "1.0E5"
$END
These controls will assume default values if not specied by the user.
The meaning of the keywords is as follows:
Keyword Description
TIME_STEPPING Selection of time stepping method:
STRICT
STRICT:V_MEAN=<real_number>
NORMAL
EQUATION_SOLVER Selection of linear equation solver:
BI_CGSTAB
TDMA
MASS_CONSERVATION Selection of mass conservation quality:
GOOD
BETTER
BEST
COMBUSTION_MODEL Selection of combustion model:
BETA3
SIF (deprecated)
BETA2 (deprecated)
BETA1 (deprecated)
TURBULENCE_MODEL Selection of turbulence model:
K-EPS
DIFFUSION_MODEL Selection of turbulence model:
HARMONIC
LINEAR
GASDATA_MODEL Selection of gasdata model, laminar burning
velocity as function of ER for the fuels:
DEFAULT
name of directory containing gasdata
les
BUILT-IN
136 Flacs simulator
AIR Specication of O2 fraction in air (<percent>
is the fraction multiplied by 100):
NORMAL
<percent>VOLUME
<percent>MOLE
<percent>MASS
AMBIENT_PRESSURE Specication of ambient pressure, the default
value is:
1.0E5
Time stepping for ventilation simulations It is possible to choose a time-stepping algorithm
which only includes the convective speed, by specifying TIME_STEPPING as STRICT:V_-
MIN=<real number>. The CFL-number based on convective speed (CFLV) is given as usual
(in the cs-le). CFL-number based on speed of sound (CFLC) is not used (the value of CFLC given
in the cs-le is not employed). Acoustical waves are not sought resolved using this approach.
This criterion is intended as an alternative to the default criterion when the ow is station-
ary/slowly varying or nearly incompressible. It has been tested in wind/ventilation simulations.
It should not be used for an explosion simulation. A speed-up factor of ca. 8 is seen from test
simulations of wind/ventilation using this criterion compared to the default setup in FLACS.
However, the speed-up factor depends on the scenario.
When high-momentum leaks are modelled, the convective speed is relatively large, and the
speed-up effect when using this criterion may be limited in this case. To employ this criterion,
the user must specify a velocity V_MIN [m/s], for example as in STRICT:V_MIN=1.0. This
velocity V_MIN is used by the time-stepping algorithm as a minimum speed when determining
the time-step, its value should be positive and not too small.
If a wind eld is specied, a natural choice would be to set V_MIN equal the value of WIND_-
SPEEED, if no wind eld is specied a value of 1.0 m/s for V_MIN would be a typical value.
Note that the value of V_MIN is in general only used in the initial phase of the simulation (when
normally a ow eld is started from a condition at rest to ensure that the time-step is not too large
even though the velocity of the ow is zero or very small (when the default time-step criterion is
used the value of CFLC limits the time-step even when the convective speed is zero).
Attention:
Note that the intention by using this time-step criterion is to speed up the calculation by
using a coarser resolution in time (longer time-steps), and this may change the simulation
results compared to a ner resolution in time (smaller time-steps) when the ow eld is
transient. A typical choice of CFLV would be 1.0, a larger value of CFLV may lead to unstable
results (depending on the scenario).
4.6.7.3 The PARAMETERS namelist
The available keywords in PARAMETERS and their default values are shown below:
VERSION 1.1
$PARAMETERS
ZERO_APOR = 0.0
ZERO_VPOR = 0.0
ER_LOW = 0.0
ER_HIGH = 0.0
FLUX_CONTROL = 1
MAX_ITERATIONS = 100
ERROR_LIMIT = 1.0E-6
ITERATE = 1
TIMEORDER = 0
$END
Note that the values in the PARAMETERS namelist are numerical, not text strings as in the SETUP
namelist.
The meaning of the keywords in the PARAMETERS namelist is as follows:
Keyword Description
ZERO_APOR Lower limit for area porosities, may be used
to avoid problems associated with small area
porosities:
0.0 not in effect (default)
0.05-0.1 to avoid MASS_RESIDUAL
ZERO_VPOR Lower limit for volume porosities, may be
used to avoid problems associated with
small volume porosities:
0.0 not in effect (default)
0.05-0.1 to avoid MASS_RESIDUAL
ER_LOW Lower bound for ER range in
GAS_MONITOR_REGION:
0.0 not in effect (using ER_LFL as
default)
ER_HIGH Upper bound for ER range in
GAS_MONITOR_REGION:
0.0 not in effect (using ER_UFL as
default)
FLUX_CONTROL Controlling oscillating velocities on stretched
grids:
1 = normal (default)
2 = reduced to avoid oscillations
MAX_ITERATIONS Maximum number of iterations in the linear
solver, this may be used to speed up the
computation:
100 = high effort (default)
50 = medium effort
5 = low effort
138 Flacs simulator
ERROR_LIMIT Error residual limit in the linear solver, this
may be used to speed up the computation:
1E-6 = high accuracy (default)
1E-4 = medium accuracy
1E-2 = low accuracy
ITERATE Residual limit in the linear solver, this may
be used to speed up the computation:
1 = normal (default)
n = iterate n times (n>1)
TIMEORDER Discretization order of time differentials:
0 = 1. order (default), enforces
ITERATE = 1
1 = 1. order, may use ITERATE > 1
2 = 2. order, may use ITERATE > 1
Changing the values in the PARAMETERS namelist will affect the accuracy and stability of the
code. In cases where the simulation gives MASS_RESIDUAL problems it may be benecial to set
the values of ZERO_APOR=0.1 and ZERO_VPOR=0.1 (or similar values in the order of 0.01 - 0.1).
In cases with stretched grids one may see oscillating ow in where the ratio between smallest
and largest side length of the control volume is large, try to set FLUX_CONTROL=2 to avoid the
problem.
A speed-up of 10-20% may be achieved by changing the accuracy and effort level from high to
low (MAX_ITERATIONS=5 and ERROR_LIMIT=1E-2).
Increasing the ITERATE value will increase the calculation time drastically. The memory usage
will also increase when TIMEORDER is increased. In cases where a converged solution is not
achieved otherwise one may try to set TIMEORDER=1 and ITERATE=3. Note that this option is
still in the phase of testing and should be used with caution (it does not seem to help very much
at present state).
4.6.8 Example: using a setup le for vessel burst calculations
A region with given pressure and/or temperature has been available in FLACS for some time,
now this possibility has been further enhanced:
1. The region can also contain ammables (used for special shape clouds)
2. The calculation can be carried out using BLAST block only
The following setup-le will dene a 12m diameter spherical high-pressure region at 300 barg
and 2500C to be calculated using the BLAST solver in FLACS. This is 4-5 times faster than
the FLACS solver, requires much less memory (i.e. larger jobs can be simulated), but does not
have panel and porosity functionality. If such functionality is required, one should instead use a
FLACS block (remove jobspec-section below).
VERSION 1.1
$PARAMETERS
PFAC = 1
HPPOS = 44, 44, 0
HPSIZ = 12, 12, 12
HPEXP = 2.0, 2.0, 2.0
HPTYP = 1, 1, 1
$END
$SETUP
KEYS="PS1=01,P_SET=Y:301,T_SET=Y:2500"
$END
$JOBSPEC
BLOCKS = "900000" "BLAST"
$END
Use short time step (of the order CFLC=0.05) and start FLACS with
> run9 flacs setup-file (the job number is given in the setup-file)
If the user wants to simulate the rupture of a pressurized vessel lled with evaporating liquid
(BLEVE), one should consider the following approach (not validated):
1. Use a vessel of larger dimensions with correct pressure and boiling point temperature.
2. Transfer all liquid into pressurized gas.
To create a cylindrical region instead of a spherical, HPTYP can be changed from 1,1,1 to 1,1,0
(vertical cylinder) or 0,1,1, (x-direction cylinder) etc.
If only a z>0 hemisphere is wanted, one can change the KEY string:
KEYS="PS1=01,P_SET=Y:301,T_SET=Y:2500,HPCON===+1"
Explanation: HPCON=XYZF
X/Y/Z can have the following values (one for each direction XYZ):
- (negative half),
+ (positive half) or
= (both halves)
F can have the following values (FUEL lean or FUEL rich):
0 (lean, i.e. ER9) or
1 (rich, i.e. ER0) concentration of fuel
This is thus an alternative to the cloud interface described below. A hemispherical cloud at
ambient T and P with diameter 20m in origin can be dened like this:
VERSION 1.1
$PARAMETERS
PFAC = 1
HPPOS = -10, -10, 0
HPSIZ = 20, 20, 10
HPEXP = 2.0, 2.0, 2.0
HPTYP = 1, 1, 1
$END
$SETUP
KEYS="HPCON===+1"
$END
(HPTYP = 1,1,0 gives a vertical cylinder etc.)
140 Flacs simulator
4.6.9 Cloud le
Optional le used to dene fuel clouds of arbitrary shape.
File name template: cs000000.CLOUD
The cloud interface module in FLACS can be used to specify rectangular or other than rect-
angular shapes of clouds with uniform or non-uniform concentration of fuel. The cloud
interface module is automatically invoked if FLACS nds the le cs000000.CLOUD in the
working directory, where 000000 is a given 6-digit job number. The hull software estab-
lishes nearest neighbor connectivity for the scattered data points in the CLOUD le (see
http://cm.bell-labs.com/who/clarkson).
A simple example of a CLOUD le dening a 1.5 m x 1.5 m x 0.34 m rectangular cloud of homo-
geneous concentration, is shown below:
VERSION 1.0
POINT
:
9.2500 -0.8500 0.0240 1.0000
10.7500 -0.8500 0.0240 1.0000
10.7500 0.6500 0.0240 1.0000
9.2500 0.6500 0.0240 1.0000
9.2500 -0.8500 0.3640 1.0000
10.7500 -0.8500 0.3640 1.0000
10.7500 0.6500 0.3640 1.0000
9.2500 0.6500 0.3640 1.0000
In this le, the rst line:
VERSION 1.0
is a version identication for future compatibility checking and the value of 1.0 should be entered
to be consistent with future interpretation.
The third line denes the type of data object used:
POINT
The following data objects can be used with the cloud interface:
POINT: the next lines contain points :N(x,y,z,f)
TETRAHEDRON: the next lines contain points :4(x,y,z,f)
HEXAHEDRON: the next lines contain points :8(x,y,z,f)
ARRAY: the next lines contain points :NINJNK(x,y,z,f)
FLUENT-UNS: the next lines contain FLUENT-UNS prole data
(( : the start of FLUENT-UNS prole data
The sequences of coordinates follow the type of data object. The coordinate of the rst point is:
9.2500 -0.8500 0.0240 1.0000
where the denition of each of the four numbers is:
x = x-coordinate
y = y-coordinate
z = z-coordinate
f = fuel mass fraction
Several point sets or more generally several data object sets, can be specied by initiating the
sequence of coordinates by a
:
as on the fourth line of the example.
Additional options are available with the cloud interface, such as lters:
ER_FLAT a b
This option lters the f-eld with a cutoff value (a) and an insert value (b). If the option ER_-
FLAT is not specied the program does:
if (ERf<ER9) ERf=ER9; else if (ERf>ER0) ERf=ER0
and if the option ER_FLAT is given the program does
if (ERf<a) then ERf=ER9; else ERf=b, ER9<b<ER0
The cloud interface module can recognize ASCII data les of certain format, in particular the
FLUENT-UNS prole le format. Geometrical transformations to align the FLUENT coordinate
system with the one used in FLACS are available. These are:
INIT: intialise the transformation matrix
TRANSLATE tx,ty,tz
SCALE tx,ty,tz,scale
ROTATE tx,ty,tz,ax,ay,az,revolution,angle
4.6.10 Layer le
Optional le used to dene dust layers and dust lifting parameters.
File name template: cs000000.LAYER
4.6.11 Events le
Optional le used to specify triggers and events.
File name template: cs000000.EVENTS
4.6.12 Heat le
Optional le used to specify heat transfer by convection.
File name template: cs000000.HEAT
142 Flacs simulator
4.6.13 Radiation le
Optional le used to specify heat loss by radiation.
File name template: cs000000.RAD
4.6.14 Pool setup le
This is an optional le used to specify a liquid spill on the ground. The pool model is enabled by
setting : Run Manager Parameters FLACS version = pool.
File name template: cs000000.POOL
It is possible to dene both a static pool and a spill that moves on the ground. For a moving
spill, the shallow-water equations are solved in two dimensions on the ground. Heat from the
ground, the ow above the pool/spill, and the radiation from the sun determine the evaporation
rate. The composition of the liquid equals that for the gas composition, see Gas composition and
volume. Remember to set
EQUIVALENCE_RATIOS_(ER0_ER9) 1E+30, 0.0
Input variables are given as follows:
VERSION 0.1
$POOL_SETUP
POOL_MODEL = 3
POOL_GROUND = "CONCRETE"
START_POOL = 5.0
MASS_POOL_0 = 0.0
DMDT = 400.0
RAD_I = 0.0
RAD_O = 5.0
XT = 0.0
YT = 0.0
ZT = 5.0
T_SOIL = 283.0
HEAT_SUN = 100.0
ROUGH_L = 0.001
$END
POOL_MODEL tells which pool model that should be used. POOL_MODEL=1 (PM1) is the
static circular pool with macroscopic correlations for the heat and mass transfer. POOL_-
MODEL=3 (PM3) refers to the spill model with a moving spill, where heat and mass trans-
fer is calculated locally in each control volume.
POOL_GROUND describes the ground. Possible grounds are "CONCRETE", "AVERAGE",
"WATER", "SOIL", "PLATE", "INSULATED", and "USER". Only "CONCRETE" and "AV-
ERAGE" (soil) work for PM1. If "USER" is specied, values for ground conductivity
"CONDUC_S" and ground diffusivity "DIFFUS_S" must be added to the setup le. It is
also possible to specify more than one ground property by doing as follows:
1. Give the general ground at the beginning of the text string: "GROUND1"
2. Specify the other ground and a box where this ground properties should be
used: "GROUND1,GROUND2 [x_low,x_high,y_low,y_high,z_low,z_high]". Example:
"SOIL,WATER[:: -1,0.1]". ":" means the entire range and can only be used for x and y
directions.
Conductivities and diffusivities for predened ground are given in Ground properties ta-
ble.
START_POOL is the point in time when pool leakage starts and the pool starts to evaporate
[s].
MASS_POOL_0 is the pool mass at time START_POOL [kg].
DMDT is the mass rate inserted uniformly over the initial pool area [kgs
1
].
RAD_I denes the inner radius of the pool donut. (PM1) or leak donut shaped area. Use
RAD_I=0.0 is default and gives a circular pool (PM1) or leakage area (PM3) [m].
RAD_O is the outer radius of the circular pool (PM1) or leak area (PM3) [m].
XT is the x coordinate of the centre of the pool (PM1) or leakage area (PM3) [m].
YT is the y coordinate of the centre of the pool (PM1) or leakage area (PM3) [m].
ZT denes the altitude from where Flacs searches for the ground in the negative z-direction.
If ZT is inside an object, there are two possible outcomes:
1. If there is no gap below ZT, then ZT = ZMAX.
2. If there is a gap (free space) below ZT, then ZT is the altitude of the surface below the
gap.
[m].
T_SOIL is the ground temperature [K].
HEAT_SUN is radiative heat from the sun [Wm
2
].
ROUGH_L is the surface roughness. Default value is 0.0005 m.
PLATE_L is thickness of the steel plate and it must be given an value when "PLATE" is in
POOL_GROUND [m].
WATER_VEL = vel_x,vel_y is a two-dimensional vector describing the velocity of wa-
ter. Default value is 0.0,0.0 [ms
1
].
T_POOL should be given in the pool or leak temperature differs from the boiling point
temperature [K].
CONDUC_S is ground conductivity when "USER" is dened in POOL_GROUND
[Wm
1
K
1
].
DIFFUS_S is ground thermal diffusivity when "USER" is dened in POOL_GROUND
[m
2
s
1
].
Ground Conductivity (Wm
1
K
1
) Thermal diffusivity (m
2
s
1
)
Concrete 1.1 10
6
Average/Soil 0.9 4.3 10
7
Plate 15 3.9 10
6
Insulated 0.0 10
30
User Set CONDUC_S Set DIFFUS_S
Water Heat transfer coefcient Heat transfer coefcient
Table 4.11: Ground properties
See also:
Example le included with the FLACS installation. Linux:
> cp /usr/local/GexCon/FLACS_v9.0/doc/examples/pool/cs000006.POOL .
Windows:
Copy les from C:\Program Files\GexCon\FLACS_-
v9.0\doc\examples\pool\cs000006.POOL
144 Flacs simulator
4.6.15 Pool leakage le
This is an optional le and it requires that cs000000.POOL exist and that Run Manager Param-
eters FLACS version = pool.
File name template: cl000000.POOL
The cl-le contains:
Line 1: POOL LEAK FILE
Line 2: Number of variables
Line 3: Name of columns. TIME should always be in the rst column.
Line 4 - : Values at different points in time.
Example:
POOL LEAK FILE
2
TIME [s] DMDT [kg/s]
0.0 0.0
1.0 1000.0
100.0 1000.0
101.0 0.0
1200 0.0
Attention:
Time in the cl000000.POOL is relative to START_POOL in cs000000.POOL.
See also:
Example le included with the FLACS installation. Linux:
> cp /usr/local/GexCon/FLACS_v9.0/doc/examples/pool/cl000006.POOL .
Windows:
Copy les from C:\Program Files\GexCon\FLACS_-
v9.0\doc\examples\pool\cl000006.POOL
4.6.16 Leak le
Denes time dependent leaks of either fuel or suppressant the number i = 1, 2, 3, ... identies
the leak; generated by the Jet program.
File name template: cl000000.n001
For briefness this le may be called the cl-le or leak le hereafter. It is a text le and may be
generated using a text editor, it may also be generated using CASD which is the preferred way
to do it.
It is possible to simulate realistic leaks in FLACS. This is done by proper subgrid modelling of
the conditions at the leak outlet and by allowing time varying leak data and multiple leaks. The
position and direction vector for each leak must be specied on the scenario le. For each leak
there must exist one le cl-le containing the time dependent leak data. The leak number is used
as identication number in the le type.
The format of the cl-le is as follows:
LEAK_CONTROL_STRING
7
TIME AREA RATE VEL RTI TLS T
data (1-7)
...
...
The seven parameters that constitute the time dependent leak data are chosen so that it is possible
to specify rate, velocity, turbulence, and temperature for the leak. (at the nozzle throat) The names
in the le format described above have the following meaning:
Parameter Units Description
TIME (s) Time
AREA (m2) Area
RATE (kg/s) Mass ow
VEL (m/s) Velocity
RTI (-) Relative turbulence intensity
TLS (m) Turbulence length scale
T (K) Temperature
Table 4.12: Leak parameters
Details of the calculation of turbulent kinetic energy and its rate of dissipation from the set of
parameters listed above (VEL, RTI and TLS) are given for the WIND condition.
Line number one of the leak le contains the LEAK_CONTROL_STRING which is used to specify
the type of leak and the leak outlet. The following letters are recognized:
Letter Description
J Jet leak
D Diffuse leak
+ Direction is now positive
- Direction is now negative
X Leak through control volume face
perpendicular to the x-axis
Y Leak through control volume face
perpendicular to the y-axis
Z Leak through control volume face
perpendicular to the z-axis
Table 4.13: Leak control string letters
Diffuse leaks are handled by setting xed values for the scalar quantities and introducing a mass
source equal to the specied leak rate at the leak node. The area porosity at the leak outlet is set
so that the specied area is obtained. A leak velocity that satises mass conservation is obtained
automatically through the ow eld solution procedure.
Jet leaks are handled by setting xed values for the scalar quantities and one or more of the
momentum components at the leak node. The area porosity is calculated so that the specied
mass ux and momentum ux is obtained. The mass conservation is not solved for at a jet leak
node during the ow eld calculation.
The default leak type is diffuse (D) and the default direction is positive (+). Take care not to
specify a control volume face more than once. The following examples deal with the LEAK_-
CONTROL_STRING.
146 Flacs simulator
4.6.16.1 Example 1: Diffuse leak in positive x-direction
X
4.6.16.2 Example 2: Diffuse leak in all directions:
D+XYZ-XYZ
4.6.16.3 Example 3: Erroneous specication:
DJ+X+XY
The following error messages would be issued on the log le(s) for example 3:
***
(LKINIT) ALREADY SET DIFFUSE
***
(LKINIT) ALREADY SET +X
Note that the simulation does not stop unless severe errors occur. In example 3 the leak speci-
cation would be taken as:
D+XY
Note that the LEAK_CONTROL_STRING must be enclosed in apostrophes (...) and may contain
blanks. The maximum length is 16 characters.
Line number two contains the number of data values to be found in each of the next lines.
Line number three contains the names of the variables. The sequence of the variables is prede-
ned and may not be changed. The time will always be the rst value on a line containing data
values.
Line number four and onward contain values for each variable. As the simulation proceeds
the le will be scanned for new data values when old values expire. Intermediate values are
calculated by linear interpolation or time integral (as appropriate) between old time and new
time.
A leak is started when the rst time on its leak le is reached and is stopped when the last time
on the le is reached.
In the case of a jet leak the nozzle density, DENS, and pressure, PRES, are calculated as follows:
DENS = RATE/(VEL AREA)
PRES = DENS R T
Where R is the specic ideal gas constant.
If the nozzle pressure is higher than the ambient pressure an analytic jet expansion will be calcu-
lated.
Note that the gas composition for each leak is taken to be the one specied by the equivalence
ratio for rich gas, ER0, and volume fractions given on the scenario le.
The leak position and direction vector is given on the scenario le. Note that the direction vector
does not need to be of unit length. FLACS will calculate the direction cosines by dividing each
component in the direction vector by the length of the vector. The direction cosines are used to
modify the ow velocity and the area for the X-, Y- and Z- direction independently. This may be
used to simulate jet leaks that are not parallel to the main axes. It is noted that this option should
be used with care in cases where the grid resolution is coarse due to the fact that numerical
diffusion tends to smear out the jet.
4.6.16.4 Example 4: Jet at 45.0 degrees:
J+XY
Direction vector given as 1.0 1.0 0.0
4.6.16.5 Example 5: Jet at 0.0 degrees:
J+X
Direction vector given as 1.0 0.0 0.0
4.6.17 Runtime simulation control le
Optional le, can be used to dene output, generate or load dump les, terminate the simulation,
etc. (further details can be found in the FLACS manual).
File name template: cc000000.dat3
For briefness this le may be called the cc-le or control le hereafter. It is a text le and may be
generated using a text editor. It is a facility for giving instructions to FLACS during a simulation.
Keywords and values may be given on the cc-le which will be read and interpreted each time
step. It may contain any number of lines. Each line shall contain a keyword followed by a value
separated by a comma (,) or a space ( ). Float values may be entered with or without a decimal
point, integer values are taken to be the integer part of a oat value. The following keywords are
currently recognized:
Keyword Meaning
NLOAD Set load identication and load data at
simulation start
NDUMP Set dump identication
TDUMP Dump data at given time
IDUMP Dump data at given iteration number
FDUMP Dump data at given fuel level
TOUTF Write results at given time
IOUTF Write results at given iteration number
FOUTF Write results at given fuel level
TSTOP Stop simulation at given time
ISTOP Stop simulation at given iteration number
FSTOP Stop simulation at given fuel level
Table 4.14: Valid cc le keywords
Fuel level is dened as the current total mass of fuel divided by the initial total mass of fuel. For
gas dispersion simulations, where the initial mass of fuel is zero, the fuel level cannot be used.
148 Flacs simulator
The control le is useful when you want to produce output that is not possible to specify on the
scenario le and when you want to monitor a simulation while it is running. The dump/load
option is one example of output and input that is not possible to specify on the scenario le. The
following examples shows how to use the cc le.
4.6.17.1 Example 1
Contents of cc-le (including comments):
"NDUMP" 1 dump file name is "rd000000.n001"
"TDUMP" 0.20 dump when time is 0.2 s
"NDUMP" 2 dump file name is "rd000000.n002"
What happened:
1. The le "rd000000.n001" was written at time 0.20 s.
4.6.17.2 Example 2
Contents of cc-le (including comments):
"NLOAD" 2 load file name is "rd000000.n002"
What happened:
The le "rd000000.n002" was read at simulation startup restoring the exact conditions at the time
when the le was written. Simulation output was appended to the existing result les. An error
message was given if the le did not exist.
4.6.18 Fuel le
Denes that the combustible dust.
4.6.19 Time dependent CFL le
Optional le used to specify time dependent CFL numbers (Please see sections CFLV and CFLV
for details).
File name template: cn000000.dat3
For briefness this le may be called the cn-lecn-le or CFL-numbers leCFL-numbers le here-
after. It is a text le and may be generated using a text editor. It is a facility for giving instructions
to FLACS during a simulation.
High ow velocities and change rates and small control volumes put severe constraints on the
size of the time step in computational uid dynamics. Using large time steps may cause the
solution to become unstable. However, to be able to simulate over a long period with todays
computer resources, it is necessary to maintain as large a time step as possible. During a gas
4.7 Output les from FLACS simulations 149
dispersion simulation the stability requirements vary, characterized by stronger requirements in
periods of highly unsteady ows and weaker requirements in steady ow periods.
To be able to effectively simulate gas dispersion scenarios including high momentum jets a facil-
ity for specifying time dependent Courant numbers is available. FLACS will read time varying
Courant numbers from the cn-le if it exists. Courant numbers specied on the scenario le are
then overridden.
An example showing the format of the cn-le is given below:
HEADLINE
3
TIME CFLC CFLV
data (1-3)
...
...
Line number one is intended to be just for the users information and contains no information to
the system other than being the rst line.
Line number two contains the number of data values to be found in each of the next lines, in this
case the number of data values is 3.
Line number three contains the names of the variables. The sequence of the variables is prede-
ned and may not be changed. The time will always be the rst value on a line containing data
values.
Line number four and onward contain values for each variable. As the simulation proceeds
the le will be scanned for new data values when old values expire. Intermediate values are
calculated by linear interpolation or time integral (as appropriate) between old time and new
time.
The cn-le must be generated manually using a text editor. It is advised that only experts make
use of the cn-le to change the default time setting in FLACS.
4.7 Output les from FLACS simulations
This section summarizes the various output les from FLACS simulations.
4.7.1 Simulation log le
Output log containing the information on the rt-le (and output messages from the operating
system); it can be useful to monitor the writing of this le during simulations:
Linux:
> tail --f tt010100
Remarks:
This le is not created on Windows.
4.7.2 Scalar-time output le
Binary output le containing the value of parameters dened in the cs-le in the specied moni-
tor points, required by Flowvis for scalar-time plots (further details can be found in section Plot
150 Flacs simulator
type).
File name template: r1000000.dat3
For briefness this le may be called the r1-ler1-le or scalar-time output lescalar-time output
le hereafter. The values of selected variables for scalar-time output at selected locations (monitor
points) are written to the r1-le at a given iteration interval.
In the case of a simulation restart the r1-le will by default be deleted (i.e. old results are lost)
when using the 2.X le format (but it is kept when using the 1.2 le format). If you want to keep
the old results at restart you must set KEEP_OUTPUT=.TRUE. using a job le. During a LOAD an
existing r1-le will be scanned up to the time recorded on the specied dump le before any new
results are written. It is usually wise to make a copy of the original r1-le before a load/rerun is
performed.
The r1-les generally require modest amounts of disk space. The number of bytes in one data
record can be calculated from the following expression (the 1.2 and 2.X le formats have approx-
imately the same record size):
RecordSize4 (2 + SUM)
Where SUM is the total number of references made to monitor points and panels.
4.7.3 Field output le
Binary output le containing the value of parameters dened in the cs-le in all grid cells, re-
quired by Flowvis for 2D and 3D plots (further details can be found in section Plot type).
File name template: r3000000.dat3
For briefness this le may be called the r3-le or eld output leeld output le hereafter. The
whole matrix of each selected variable for eld output is written to the r3-le, when triggered by
one or more events (or signals). Simulation start and stop will trigger output as will the event
of passing certain fuel levels and time values. Runtime specied events may also trigger output.
A message is issued on the log le(s) for each event that triggers output. Several events may
happen to trigger output simultaneously, this will generate just one instance of output.
In the case of a simulation restart the r3-le will by default be deleted (i.e. old results are lost)
when using the 2.X le format (but it is kept when using the 1.2 le format). If you want to keep
the old results at restart you must set KEEP_OUTPUT=.TRUE. using a job le. During a LOAD an
existing r3-le will be scanned up to the time recorded on the specied dump le before any new
results are written. It is usually wise to make a copy of the original r3-le before a load/rerun is
performed.
The r3-les generally require large amounts of disk space. The number of bytes in one data record
can be calculated from the following expression (the 1.2 and 2.X le formats have approximately
the same record size):
RecordSize4 (2 + NX NY NZ) NVAR
NVAR number of variables specied
NX,NY,NZ matrix dimensions
4.7.4 Simulation log les
Binary output log of general messages
4.7 Output les from FLACS simulations 151
File name template: rt000000.dat3
For briefness this le may be called the rt-le or log le hereafter. It contains a log of the messages
issued during a simulation. It is wise to check if any error messages have been issued. Error
messages are identied by "" at the beginning of the message, plain messages are introduced
by a space ( ) or a number sign (#).
In the case of a simulation restart the rt-le will be deleted. It is usually wise to make a copy of
the original rt-le before a load/rerun is performed.
The information on the log le is also written to a connected terminal (standard output) during
the simulation.
File name template: rt000000. (new specic log les, see below).
Several log les will be generated when using FLACS, they are all text les and may be printed
as time series using common graphical tools (such as gnuplot on Linux or Excel on Windows).
The contents of each le is explained at the start of the le, below you will only nd a brief
explanation of the new log les:
File name Contents of le
rt000000.CFL CFL-numbers and time step log le
rt000000.EQSOL Equation solver (residual) log le
rt000000.FUEL Amount of fuel log le
rt000000.MASS Total mass log le
rt000000.P Pressure log le
rt000000.dat3 The usual simulation log le
Table 4.15: Flacs log les
Note that some of the les listed above may not appear for a given installation, this is because
they are intended for debugging and testing purposes only. The most relevant le for a regular
FLACS user is the "rt000000.FUEL" le, where the amount of fuel in the simulation volume (total
amount and combustible amount) is logged.
4.7.5 General simulation log le
4.7.6 CFL log le
Output log of CFL numbers.
4.7.7 Fuel log le
Output log of the amount of fuel in the simulation volume.
4.7.8 Mass log le
Output log of the mass in the simulation volume.
4.7.9 Pressure log le
Output log of pressure.
152 Flacs simulator
4.7.10 Simulation dump le
Creates a snapshot of the simulation at a selected time dened in the cc-le the number i (i =
1, 2, 3, ... , 999) identies the dump (further details can be found in the FLACS manual).
For briefness this le may be called the rd-le or dump le hereafter. It contains the information
necessary to restart a simulation from a given time required that the initial input data have not
been changed drastically. There may be more than one rd-le.
Attention:
If you want to append new data to existing result les, do not change the output specication
or the grid or the geometry when using a dump le!
4.7.11 Monitor le
4.7.12 The frozen cloud concept
The frozen cloud principle assumes that for a given dispersion/ventilation scenario the variation
in fuel concentration is approximately proportional to the ratio [leak rate/ventilation rate]. This
concept may be used in the following way:
4.7.13 Example: dispersion and ventilation study
Prepare a FLACS dispersion/ventilation simulation in the usual way, say for run no. 123456
as in the following example. The simulation les for run no. 123456 should include the le
cs123456.MON (in addition to the other needed les, for example cg123456.dat3). The content of
the text le (could be edited with any text editor) cs123456.MON could be something like as in
the following example.
Attention:
Note that you may have comment lines starting with #
VERSION 1.0
# name=<suffix>; output file will be named rt123456.MON.<suffix>
# start=x,y,z; start position of region in space considered, unit [m]
# end=x,y,z; end position of region in space considered, unit [m]
# see=1.0; in general this parameter is set to a fixed value 1.0
# mix=<fraction>; when the output in the file rt123456.MON.<suffix> is calculated,
# the extrapolated fuel concentration in every control volume is calculated
# as the fuel concentration of standard FLACS multiplied with the factor <fraction>
volume(name="123455",start=138,-21,32,end=198,21,48,output="FUEL(see=1.0,mix=0.8)")
volume(name="123456",start=138,-21,32,end=198,21,48,output="FUEL(see=1.0,mix=1)")
volume(name="123457",start=138,-21,32,end=198,21,48,output="FUEL(see=1.0,mix=1.2)")
The le cs123456.MON will be automatically read by FLACS when the simulation for run no.
123456 is started. Three output les (text les) rt123456.MON.123455, rt123456.MON.123456,
and rt123456.MON.123457 will be generated. Output from these les can be used during e.g. a
probabilistic analysis. Using the frozen cloud concept, it could for example be assumed that the
output in rt123456.MON.123455 are approximately equal similar output that could be obtained
by a new full FLACS simulation with scenario equal run no. 123456 except that the leak rate is
reduced by multiplying with the factor 0.8 (ventilation and other parameters are kept xed).
4.8 Potential bugs or problems with Flacs 153
4.7.14 Simulation save le
Temporary le generated by DESC/FLACS in some situations (further details can be found in
the FLACS manual).
For briefness this le may be called the rx-le or save le hereafter. It is a temporary le that
FLACS uses to store intermediate data for output. There may be more than one rx-le. These
les are needed when time integrals of whole matrices are to be calculated. It avoids using large
amounts of core memory but accessing large les is slow compared to memory access. Note that
the simulation save les are only written if you have specied eld output of variables that need
temporary le storage. These les are deleted when the simulation terminates normally. In cases
when the simulation is not allowed to terminate normally these les will exist after programme
stop, they must then be deleted manually. Existing temporary les may confuse the programme
at start-up. Below is presented a list of the variables that require temporary le storage when
specied for eld output:
Name Description
NPIMP Pressure impulse
UDIMP Drag-impulse component x-direction
NVDIMP Drag-impulse component y-direction
NWDIMP Drag-impulse component z-direction
NDIMP Drag-impulse value
NRESID Mass residual in continuity equation
It is advised that the above listed variables are not specied for eld output. The le handling
will increase the execution time for the programme drastically!
4.8 Potential bugs or problems with Flacs
This chapter contains a list of potential bugs or problems associated with the CFD simulator
Flacs, and some possible workarounds.
4.9 Warning and error messages
Flacs will write any error or warning messages to the Simulation log le. The following sec-
tions give a description of the most common warnings and errors, and possible causes and
workarounds.
4.9.1 Mass residual
The message " MASS RESIDUAL =..." indicates that there are problems to achieve a converged
solution. If the mass residual is not too large (should be smaller than order of 1/10) or it does
not last for many iterations it may be ignored. The built in limit for the mass residual is rather
strict to enable an accurate solution over long time, a short lasting violation of the limit does not
necessarily lead to a suspect result.
There are several possible solutions to this problem
Lower the CFLV and CFLC numbers. This will result in longer simulation time. It is rec-
ommended to try to divide both numbers by 2 in case of mass residual
154 Flacs simulator
Change the values of ZERO_APOR and ZERO_VPOR. (details)
In cases with stretched grids set FLUX_CONTROL=2. (details)
4.9.2 Leak excess area
If a leak is dened with a larger area than the control volume face where the leak is located, the
warning " LEAK EXCESS AREA ..." will be issued.
The solution is to:
Increase the grid cell size where the leak is dened
or redene the leak so that the leak area is decrease
Chapter 5
Flowvis
156 Flowvis
Flowvis is the postprocessor for the CFD-code FLACS, and is a program for visualizing results
from computer aided simulations of gas explosions, gas dispersion and multi phase ow.
Flowvis was originally developed for presenting output from the multi-phase simulation pro-
gram GALIFA (GAs LIquid Flow Analyser). This work was nanced by Christan Michelsen Re-
search (formerly Christian Michelsen Institute). The basic concepts were based on experience
gained from developing and using the VIEW program which was the result visualization tool for
FLACS86 and FLACS89.
The development of Flowvis has been nanced through several different projects (see Acknowl-
edgements), starting with the Gas Safety Programme 1990-1996.
This chapter describes the menus in Flowvis.
5.1 Overview
The user starts Flowvis either by clicking the Flowvis icon in the run manager window,:
Figure 5.1: The Flowvis desktop icon
or by executing the following command (in Linux):
> run9 flowvis
A Flowvis save le is called a presentation le, and has the sufx .pres, eg. "2D-plots.pres".
The command for starting Flowvis is usually
> run9 flowvis options
The only option available is -verify_por.
The purpose of Flowvis is to visualize simulation results fromFLACS. The results froma simulation
are called job data in Flowvis and in this guide. The job data, which are stored on a set of les,
include grid, geometry ( obstructions), bulk data ( porosities), scenario and results.
The user visualizes simulation results by creating a presentation. A presentation contains one or
more pages, each containing one or more plots. Presentations can be saved on les. Available plot
types include curve plots, 2 dimensional contour and vector plots, and 3-dimensional volume
plots.
One page is displayed at a time in the graphical area. 220 range colours are available for producing
lled contours and shaded plots. In addition, a set of single colours are used for plotting curves,
text etc. Flowvis supplies functions for colour table manipulation. Different colour tables can be
used for each page within a presentation. In addition, the value to colour or phase to colour mapping
can be set individually for each plot.
Flowvis offers a high degree of exibility with regards to combining different types of data (grid,
geometry, porosities, scalar variable, vector variable etc.) in one plot.
5.1 Overview 157
Cut, copy and paste commands are available for both pages and plots. Geometric primitives can
be blanked to make a better view of data inside the geometry. The contents of the plots can be
rotated and mirrored.
Flowvis can export to several different formats, including image and video formats. Flowvis can
show animations directly on the screen, or write them to les.
Experimental data on ASCII les can be imported into curve plots. This makes it possible to com-
bine experimental and simulated data in the same plots.
If one wants to start Flowvis for porosity verication, a start-up option is available which instructs
Flowvis to automatically create a plot for this purpose.
5.1.1 The Flowvis window
The main window is displayed all the time while Flowvis is running. The window contains a menu
bar, a graphical area, a time slider, a page slider and a button row.
Figure 5.2: Flowvis elements
The user interface consists of a main window, and a set of pop-up dialog boxes. The main window is
displayed all the time while Flowvis is running. It contains a menu bar which lists the available
menus. One page at a time is displayed in the graphical area. In addition, there are sliders for
selecting which timestep and page to display, and several push buttons for eg. redrawing the
selected plot and for re-scanning the input les.
Many of the Flowvis commands have a pop-up dialog box associated with them. The dialog box
pops up when the command is selected. The main window does not accept input while a pop-up
dialog box is open.
Usually, only one dialog box is open at a time. But under certain conditions, for example upon an
error, a message dialog box may pop up in addition to the one already open. The message dialog
box contains a text and a push button labelled Ok. No other Flowvis dialog boxes takes input
158 Flowvis
while this dialog box is open. The user must verify that the information is perceived by clicking
the Ok button. The message dialog box then pops down.
The Ok and Cancel push buttons are present in most Flowvis pop-up dialog boxes. Clicking the
Ok button applies all changes specied in the dialog box. The graphics are redrawn if applicable,
and the dialog box is closed. The Cancel button dismisses all the changes, and the dialog box is
closed. Many dialog boxes contain an Apply button in addition. Clicking this button causes the
selected plot or page to be redrawn with the current changes applied. The dialog box remains
open.
5.1.2 The menu bar
The menu bar lists the available menus.
1. The File menu contains commands for opening, saving, exporting and printing the presen-
tation.
2. The Edit menu contains commands which manipulates both pages and plots.
3. The Page menu contains commands for manipulating the displayed page.
4. The Plot menu contains commands for creating and editing plots.
5. The Verify menu contains a command for verifying porosities, and other control volume
attributes.
6. The Options menu contains toggle buttons for turning on and off the display of page head-
ers, implicit redraw and choosing between white and black background.
7. The Help menu contains on-line help and about Flowvis.
Not all menu items are available all the time while running Flowvis. This depends on the con-
tents of the current presentation and the selected page or plot. Unavailable menu item texts are
dimmed.
5.1.3 The graphical area
One page at a time is displayed in the graphical area. Either a page or a plot is selected at every
stage while running Flowvis, except when the current presentation is empty. If a plot is selected,
this plot is indicated by a frame surrounding the plot. If no frame is displayed, the displayed
page is selected.
A plot is selected by clicking MOUSE+LEFT inside the plot. The displayed page is selected by
double clicking MOUSE+LEFT somewhere in the graphical area.
5.1.4 The Page slider
The Page slider is for selecting which page to display in the graphical area. The scale is only
visible if the current presentation contains more than one page. The pages are numbered from 1.
5.1.5 The Time slider
The Time slider is for selecting which time step to display. The scale is only visible if the displayed
page contains plots with data for more than one time step. The time steps are numbered from 0
(zero).
5.1 Overview 159
5.1.6 The Play / Pause button
The Play / Pause button starts an animation sequence of the current plot(s). Clicking the button
while the animation is ongoing, will pause the animation.
5.1.7 Add Page button
Shortcut: CTRL+A
This command adds a new page after the displayed page. The new page contains one plot.
5.1.8 The Previous Plot push button
Shortcut: CTRL+LEFT
Clicking the Previous Plot push button causes the selection of the previous plot. If the previously
selected plot was the rst plot in the page, the page is selected. If the page was selected, the last
plot in the page is selected.
5.1.9 The Next Plot push button
Shortcut: CTRL+RIGHT
Clicking the Next Plot push button causes the selection of the next plot. If the previously selected
plot was the last plot in the page, the page is selected. If the page was selected, the rst plot in
the page is selected.
5.1.10 The Rescan Data push button
Shortcut: CTRL+R
Clicking the Rescan Data push button causes Flowvis to rescan all data les referenced in the
current presentation. This makes it possible to view the output from a simulation while the
simulation is running.
5.1.11 The Redraw Plot push button
Shortcut: F5
Clicking the Redraw Plot push button causes Flowvis to redraw the selected plot.
5.1.12 The Cancel Redraw Button
Shortcut: ESCAPE
A time consuming plot operation can be interrupted by pressing the cancel redraw button.
160 Flowvis
5.1.13 The pop-up menu
Some commonly used choices from the Edit and Plot menus are collected in a pop-up menu
available anywhere in the main window (press MOUSE+RIGHT).
5.1.14 Keyboard short-cuts
The following shortcut key combinations are available.
CTRL+N Create new presentation
CTRL+SHIFT+N Create new presentation in directory
CTRL+O Open existing presentation
CTRL+S Save presentation
CTRL+SHIFT+S Save presentation with le new name
CTRL+I Import le
CTRL+E Export presentation to image or movie format
CTRL+P Print presentation
CTRL+Q Exit Flowvis
CTRL+X Cut plot
CTRL+C Copy plot
CTRL+V Paste plot
CTRL+A Add a page to the presentation
CTRL+M Modify the page
Page Up, MOUSE+WHEEL Display previous page
Page Down, MOUSE+WHEEL Display next page
LEFT RIGHT, ALT+WHEEL Select previous/next time step
CTRL+1 Create new scalar-time plot
CTRL+2 Create new 2D cut plane plot
CTRL+3 Create new 3D cut plane plot
CTRL+4 Create new volume plot
CTRL+5 Create new scalar line plot
CTRL+6 Create new grid plot
CTRL+7 Create new monitor points plot
CTRL+8 Create new panels plot
CTRL+9 Create new particle traces plot
CTRL+0 Create new scalar time annotation
CTRL+LEFT, CTRL+RIGHT Select previous/next plot
F1 Online help
5.2 Creating a new presentation 161
5.2 Creating a new presentation
This section introduces the structure of a Flowvis presentation. A Flowvis presentation is a col-
lection of plots showing the results from one or several simulations.
5.2.1 The structure of a Flowvis presentation
Flowvis can plot results from simulations located in the directory selected by the user. It is not
possible to include results from more than one directory in a single presentation. To work around
this limitation FLACS les can either be copied to a single directory, or on Linux, softlinks can be
created using the following command:
> ln -s ../simulations2/cs010100.dat3
Note that the command must be repeated for all neccesary le (cs, co, cp, cg, r3, r1 les).
To start a new presentation either choose FileNew or FileNew At. This will create an empty
presentation. The New At command opens a directory selection dialog enabling the user to select
the working directory. The New command will look for simulation results in the current directory,
that is the directory where Flowvis was started.
Remarks:
Only variables saved during a Flacs simulation can be plotted in Flowvis. If a variable is
missing, the simulation scenario must be edited using CASD and simulation must be run
over again.
Pages are added by clicking the Add Page button, or CTRL+A
Subdivide a page to have more than one plot by typing CTRL+M
Add a plot by going MOUSE+RIGHT in the plot area, and select a plot type, or type for
instance CTRL+1 to have a scalar time plot
A dialog where the user can select simulation number and variable to plot is show
Modify the plot by MOUSE+RIGHT, selecting Plot Speciaction, Plot Domain and Variable Ap-
pearance
The presentation can be saved on le using the Save or Save As commands
HINT! It is not necessary to create a new presentation from scratch for each new simulation.
An existing presentation can easily be adapted to new simulations by using the Substitute Job
command.
5.3 File menu
The File menu supplies commands for creating, opening, saving, exporting and printing presen-
tations.
The default le type for presentation les is .pres.
162 Flowvis
5.3.1 New
Shortcut: CTRL+N
This command creates a new empty presentation. If there were unsaved changes to the current
presentation, a dialog box is displayed, asking the user about saving changes.
5.3.2 New at
Shortcut: CTRL+SHIFT+N
This command supplies the possibility to read data from a new directory. A dialog box for spec-
ifying the directory pops up. When a directory has been specied, Flowvis proceeds as for the
New command.
The New At dialog box is a File Selection Box. It lets the user traverse through directories, view the
subdirectories in them and select a directory.
A directory is selected by clicking in the list, or by typing the directory name in the Directory eld.
To open the selected directory, click the Choose button.
5.3.3 Open
Shortcut: CTRL+O
This command opens an existing presentation le. The dialog box allows the user to specify
the path and le name for the presentation le. If there were unsaved changes to the current
presentation, a dialog box is displayed, asking the user about saving changes.
5.3.4 Save
Shortcut: CTRL+S
This command saves the current presentation on a le. If no le name has been specied, the
Save As dialog box is displayed. This command is available only when the current presentation
is not empty.
5.3.5 Save as
Shortcut: CTRL+SHIFT+S
This command saves the current presentation on a new le. The Save Asdialog box is displayed,
allowing the user to specify the path and le name for the presentation le. If an existing le
name is specied, Flowvis displays a dialog box asking if it is OK to overwrite the le. This
command is available only when the current presentation is not empty.
The le selection box lets the user traverse through directories, view the les and subdirectories
in them and select les. A presentation le is selected by clicking in the lists, or by typing the
le name in the File name eld. The selected le is opened by clicking the Open button or double
clicking the selected le name in the Files list.The Cancel button cancels the operation.
5.3 File menu 163
5.3.6 Import
Shortcut: CTRL+I
This command imports data from an ASCII le into a Scalar Line plot, and thus makes it possible
to include experimental data in this plot type. This command is available only when the selected
plot is a Scalar Line plot.
5.3.7 Export
Shortcut: CTRL+E
This command exports the displayed page or all pages to various image and movie formats. This
command is only available when the current presentation is not empty.
The Export dialog box allows choosing between exporting the Displayed Page or All Pages, and
the Displayed Timestep or All Timesteps. The output formats are as follows:
Format Type Description
PNG Portable Network Graphics
GIF CompuServe graphics interchange format
JPEG Joint Photographic Experts Group JFIF
format
BMP Microsoft Windows bitmap image le
HTML Web Presentation
MPEG Motion Picture Experts Group le
interchange format (Movie format)
MPEG4 Microsoft MPEG-4v2 video (Movie format)
DIV3 DivX ;-) (Movie format)
WMV7 Microsoft WMV7 (Movie format)
OGG Xiph.org Foundation video format (Movie
format)
GIFAnimated CompuServe graphics interchange format
(Movie format)
Table 5.1: Flowvis export formats
Some formats might not be available on all systems.
The le name is derived from the presentation name. The page number and timestep number are
added to the name, and the le type depends on the selected device, see above. When exporting
all pages and/or timesteps, the le name contains formats for inserting the actual page and/or
timestep number in each le name (one le per. page/timestep). Note that the le sufx must
not be changed as this is what determines which image format to use for export.
During export the desktop must stay open, ie. you can not switch desktops. You can normally
cover the Flowvis window without creating problems for the export.
Note that the HTML export facility has a few limitations.
5.3.8 Print
Shortcut: CTRL+P
This command prints the displayed page or all pages, or creates a pdf (Portable Document For-
164 Flowvis
mat) le. This command is only available when the current presentation is not empty.
The Printdialog box allows choosing between printing the Displayed Page or All Pages, and the
Displayed Timestep or All Timesteps.
To create a pdf le choose Print to le, and give the le a name with sufx ".pdf".
5.3.9 Exit
Shortcut: CTRL+Q
This command ends the Flowvis application. If there were unsaved changes to the current pre-
sentation, a dialog box is displayed, asking the user about saving changes.
5.4 Edit menu
The Edit menu supplies commands for editing the selected plot or page.
The Cut, Copy and Paste commands applies to the selected plot or page. If a plot is selected, this
plot is indicated by a white frame surrounding the plot. If no frame is displayed, the displayed
page is selected. A plot is selected by clicking MOUSE+LEFT inside the plot, or by clicking the
Select Previous or Select Next Plot push buttons. The displayed page is selected by double clicking
MOUSE+LEFT somewhere in the graphical area.
5.4.1 Cut
Shortcut: CTRL+X
This command copies the selected page or plot to the clipboard, and deletes it from the presen-
tation. The command is not available if the current presentation is empty, or if an empty plot is
selected.
5.4.2 Copy
Shortcut: CTRL+C
This command copies the selected page or plot to the clipboard. The command is not available if
the current presentation is empty, or if an empty plot is selected.
5.4.3 Paste
Shortcut: CTRL+V
If the current presentation is empty, or if the displayed page is selected, this command is available
only if the clipboard contains a page. The command then inserts a copy of the clipboard page
after the selected page. If a plot is selected, the Paste command is available only if the clipboard
contains a plot, and the selected plot is empty. The command then copies the clipboard plot to
the selected plot.
5.5 Page menu 165
5.4.4 Paste plot format
This command is available only if the clipboard contains a plot. The command copies the format,
or parts of the format, of the clipboard plot to the selected plot, or all the plots of the same type
in the selected page or the presentation.
The Paste Plot Format dialog contains two sets of buttons. The rst are three radio buttons for
selecting the target for the paste operation. The format of the clipboard plot can be pasted either
to the selected plot only (if same plot type), to all plots of same the type in the displayed page, or to all
plots of same the type in the entire presentation.
In addition, there are toggle buttons for deciding which parts of the plot format to paste. The
options are found in the table below.
Name Description
Relative Timestep This is the timestep for the plot relative to
the timestep for the page
Rotation Matrix This implies the settings in the Rotate dialog
box
Variable Appearance This implies the settings in the Variable
Appearance dialog box
Utilities This implies the Utility settings in the Plot
Specication dialog box
Plot Domain This implies the settings in the Plot
Domaindialog box
Table 5.2: Paste plot format options
5.4.5 Substitute job
This command is not available for empty presentations. The command substitutes a job number
with another in all plots in the selected page or in the entire presentation.
The Substitute Job dialog box contains two text elds. One is for entering the job number to be
substituted, and the other is for entering the new job number. In addition, the dialog box contains
two radio buttons for selecting whether to apply the substitution only to the displayed page or
to all pages in the presentation.
Note that when substituting a job with one with greater grid resolution, the minimum and max-
imum grid indices in the plot specication is unchanged.
5.5 Page menu
The Page menu supplies commands for adding new pages, and editing the displayed page. The
number of plots in a page is dened by the subdivision in the horizontal and vertical directions.
A page includes a time specication which decides the time units used for the page, in addition to
minimum, maximum and delta timestep used for displaying the page.
5.5.1 Add
Shortcut: CTRL+A
This command adds a new page after the displayed page. The new page contains one plot.
166 Flowvis
5.5.2 Modify
Shortcut: CTRL+M
The Modify command makes it possible to change parameters for the displayed page, such as
subdividing the page into several plots.
The upper frame is for Time Specication. It contains an options menu for selecting Time Units.
Available units are Seconds and Milliseconds.
The rest of the Time Specication area is included only if at least one plot in the displayed page
can be plotted for several timesteps. The next eld is for editing the Minimum Timestep for the
displayed page. Another eld shows the Maximum Timestep for which there are datas for this
page. The latter eld is not editable. To modify the maximum timestep, press the Fixed toggle
button and modify the Fixedeld. In addition, the Delta Timestepeld decides the delta used when
printing all timesteps (using the Print command in the File menu).
The second area in the dialog box is the Subdivision of Page. This area contains two elds for
setting number of plots in the horizontal (x) and vertical (y) direction for the displayed page.
When a new page is created, it contains only one plot. Note that if the number of plots is reduced,
the remaining plots are deleted without any warning.
5.5.3 Single colours
This command modies the single colours used in the displayed page.
It contains a set of sliders, a push button and a colour display in addition to the Ok and Cancel
push buttons. The rst scale is for selecting a single colour index. The corresponding colour is
indicated by an arrow in the single colour display. The single colour display contains a horizontal
row of colour rectangles, one for each single colour. In addition to using the index slider, a colour
index can be selected by clicking the mouse in the colour display.
There are three scales for setting Hue, Light and Saturation for the selected colour. Pressing the In-
terpolate push button causes interpolation of the hue, light and saturation values from the second
to the last colour.
5.5.4 Range colours
This command modies the range colours used in the displayed page.
At the top, there is an area called Range Colours Setup, which contains radio buttons for selecting
the number of different colours and the number of shades (light/saturation values) per colour.
There are a total of 220 range colours available. These can be organized as 220, 18, 12, or 6
different colours (hues) with 1, 10, 14 or 22 shades for each colour. The light and saturation range
is used for shading 3 dimensional plots.
The next area is labelled Hue for Range Colour, and contains two scales, one for selecting a range
colour index, and one for setting the hue for the selected colour.
The range colour display contains two horizontal rows of colour rectangles. The upper row con-
tains one rectangle for each range colour, while the lower row shows the light and saturation
range for the selected colour. The selected colour is indicated by an arrow. Other colour(s) are
selected by clicking MOUSE+LEFT in or below the desired rectangle.
The next area is called Light and Saturation Range for each Colour, and contains scales for setting
the minimum and maximum light and saturation for each range colour.
5.6 Plot menu 167
5.5.5 Header
This command turns page header on/off, and modies the contents of the header.
At the top of the Page header dialog box is a toggle button for turning the dispay of page headers
on/off. This can also be done using the Header On command in the Options menu.
Below the Header On/Off toggle button is an area for deciding the contents of the page. These
settings can be applied to all the pages in the current presentation, or to the displayed page only.
The items that can be included in the header are the Flowvis version string, the current date, the
page number and a user dened string.
5.5.6 Next page
Shortcut: Page Down
Shows the next presentation page.
5.5.7 Previous page
Shortcut: Page Up
Shows the previous presentation page.
5.5.8 Next time step
Shortcut: RIGHT
Shows the next time step.
5.5.9 Previous time step (left arrow)
Shortcut: LEFT
Shows the previous time step.
5.6 Plot menu
The Plot menu contains commands for creating and editing plots.
5.6.1 Plot type
This command is available only if the selected plot is empty. The Data Selection dialog box pops
up after a plot type has been selected (and the selected plot type is not Annotation or Scalar Time
Annotation).
The following plot types are available in Flowvis:
Scalar Line plot. This is a curve plot where the value of one variable is plotted along one grid
curve. Output from several simulations and for several phases can be combined.
168 Flowvis
Scalar Time plot. This is a curve plot where the value of one variable is plotted along the time
axis. Output from several simulations and for several monitor points can be combined in
one plot.
Grid plot. This is a 3 dimensional plot of the grid.
Monitor plot. This is a 3 dimensional plot where the monitor points are displayed together
with the geometry.
2D Cut Plane plot. This is a 2 dimensional plot of a grid plane, or part of a grid plane. The
plot can include output for none, one or two variables. If two variables shall be combined,
one must be a scalar variable and the other must be a vector variable. Scalar variables are
displayed using lled contours, and the vector variables are displayed as arrows. The plot
can include output fromseveral simulations. Grid, geometry and porosities can be included
in the plot.
Monitor Points, Ignition Region and Panels. The colours for the monitor points dene the
point classications as described above. This is also true for the ignition region if it is a
point and not a region. Only the points/planes that intersect with the displayed cut plane
are drawn. In addition, the geometry is now drawn in colours if no data is displayed.
3D Cut Plane plot. This is a 3 dimensional plot of a set of grid planes. The type of output is
identical to that of 2D Cut Plane plots, except that contour curves cannot be drawn, and the
geometry is drawn as a 3 dimensional shaded model. By combining many cut planes and
by limiting the visible value range for the scalar variable, a 3 dimensional view of the value
range is obtained (honeycomb plot).
Volume plot. This is a 3 dimensional plot where the values of a variable are visualized as a
raytraced volume. Output for one scalar variable for one simulation can be visualized in
this plot type. A 3 dimensional shaded model of the geometry can be included.
Particle Traces plot. This is a 3 dimensional plot where a velocity eld is visualized by tracing
a set of particles in the eld.
Annotation plot. This is a text plot which extracts text information from other plots in the
same page. Repetition of the same text in several plots are then avoided.
Scalar Time Annotation plot. This is a text plot which contains a table of minimum and
maximum values for Scalar Time plots in the same page.
A plot is selected by clicking the MOUSE+LEFT inside the plot, or by clicking the Select Previous or
Select Next Plot push buttons.
A plot is selected by clicking MOUSE+LEFT inside the plot. If the selected plot is empty, it can be
dened using the Plot Type command in the Plot menu. The available plot types are described
briey below.
Plot type Time
specica-
tion
No. of
jobs
No. of
variables
Geometry
display
Grid
display
Porosity
display
Scalar
Line
Timestep n 1 None (The
abscissa is
a grid line)
None
Scalar
Time
Time
range
n 1 None None None
Grid None n 0 Volume Volume None
Monitor None 1 0 Volume None None
2D Cut
Plane
Timestep n 0 - 2 2D cut 2D cut 2D cut
3D Cut
Plane
Timestep n 0 - 2 Volume 2D cuts 2D cuts
Volume Timestep 1 0 - 1 Volume None None
5.6 Plot menu 169
Part.
Traces
Timestep 1 0 - 1 Volume None None
Annota-
tion
Timestep 1 0 None None None
Scal.T.
Annot
Time
range
n n None None None
Table 5.3: Summary of plot types in Flowvis
5.6.2 Data selection
This command is available only if the selected plot is not empty and not an Annotation plot. The
command lets the user change the selection of jobs, variables, phases, monitor points and panels
for the selected plot. Legal selections depends on the plot type.
Figure 5.3: Flowvis data selection
The box contains four lists:
Job Numbers Contains the job numbers for all scenario and grid les found in the current di-
rectory
Variables All variables that can be plotted. If a variable is not listed here, it was not selected for
output in CASD
Phases Selecting phases. This is currently not used in FLACS
Monitors Monitor points or monitor panels
Plot type No. of jobs No. of
variables
No. of phases No. of monitor
points
Scalar Line n 1 n 0
Scalar Time n 1 1 n
Grid n 0 0 0
Monitor 1 0 0 n
2D Cut Plane n 0 - 2 n 0
3D Cut Plane n 0 - 2 n 0
Volume 1 0 - 1 (scalar) 1 0
170 Flowvis
Particle Traces 1 0 - 1 (vector) n 0
Table 5.4: Valid data selections depending on plot type
For Grid plots, 2D Cut Plane plots and 3D Cut Plane plots, several jobs which are part of one
multiblock simulation can be combined in one plot.
5.6.3 Plot specication
This command is available for all plot types except Annotation plots. The contents of the Plot
Specication dialog box depends on the plot type. It contains controls for manipulating value
range, the display of axis, geometry etc.
Figure 5.4: Flowvis plot specication
The contents of the Plot Specication dialog box depends on the plot type for the selected plot.
On top of the dialog box for all plot types that are drawn for one timestep at a time, there is a text
eld labelled Relative Timestep. This is the timestep for the selected plot relative to the timestep
for the page.
Instead of Relative Timestep, the Plot Specication dialog box for Scalar Time plots contains an area
for setting the Time Range for the plot.
The Plot Specication dialog box for Monitor plots contains toggle buttons for turning on/off
projections of the monitor points to the xy-, yz- and xz-planes, and a toggle button for turning
geometry extent marking on or off.
The Plot Specication dialog box for all plot types has an area for selecting Utilities. The utilities
that are available for a plot, depends on the plot type. A short description of all the available plot
utilities is given below.
Utility Scalar
Line
plot
Scalar
Time
plot
Grid
plot
Moni-
tor
plot
2D Cut
Plane
plot
3D Cut
Plane
plot
Vol-
ume
plot
Particle
Traces
Axis x x x x x x x x
5.6 Plot menu 171
Legend x x
Scalar
Legend
x x x
Vector
Legend
x x x
Grid x x
Geom-
etry
x x x x x
Area
Porosi-
ties
x x
Vol.
Porosi-
ties
x x
Depth
Shad-
ing
x x x
Trilin-
ear
x
Table 5.5: Plot utilities
Axis Turns on or off the display of the axis.
Legend Turns on or off the display of a legend explaining the meaning of a curve colour or font.
Scalar Legend Turns on or off the display of a legend annotating the value to colour or phase to
colour mapping.
Vector Legend Turns on or off the display of a legend annotating the value to colour and value
to vector length mapping.
Grid Turns on or off the display of the grid.
Geometry Turns on or off the display of the geometry.
Area Porosities Turns on or off the display of the area porosities.
Volume Porosities Turns on or off the display of the volume porosities.
Depth Shading Turns on or off depth shading. If depth shading is turned off, at shading is
applied.
Trilinear Turns on or off trilinear interpolation. This type of interpolation gives a more accurate
reconstruction of the data. (It takes longer time to render the plot.)
5.6.4 Plot domain
This command is available for all plot types except Scalar Time, Monitor and Annotation plots.
The purpose of the Plot Domain dialog box is to dene the part of the simulation volume to
display.
On top of the Plot Domaindialog box is a list of all jobs included in the selected plot. One job is
selected at a time, and the controls in the rest of the dialog box applies to this job.
The Plot Domaindialog box for Scalar Line plots contains radio buttons for selecting the direction
of the grid curve to plot along. In addition, it contains scales for selecting the indices of the curve,
172 Flowvis
and minimum and maximum indices along the curve. For 2 dimensional simulations, only the
minimum and maximum indices can be set.
The Plot Domain dialog box for Grid plots contains scales for selecting the minimum and maxi-
mum grid indices in each direction.
The Plot Domain dialog box for 2D Cut Plane plots contains toggle buttons for selecting plane
orientation in addition to scales for selecting the index of the plane, and minimum and maximum
indices in the plane. For 2 dimensional simulations, only the minimum and maximum indices
can be set. A dialog box for verifying porosities pops up upon clicking MOUSE+LEFT inside the
plot. The porosities in the indicated control volume is displayed in the dialog box.
For a 3D Cut Plane plot, the Plot Domain dialog box contains sliders for selecting the minimum
and maximum grid indices in each direction in the simulation volume. The user can specify
number of planes, step between each plane and start index for the planes to be drawn in each
direction, using the spin boxes.
For a Volume and Particle Traces plots, the Plot Domain dialog box contains scales for selecting
the minimum and maximum grid indices in each direction.
5.6.5 Variable appearance
This command is available for all plot types except Grid, Monitor and Annotation plots, given
that one (or two) variable has been selected in the selected plot. The content of the Variable
Appearance dialog box depends on the plot type. For example, the curve colours and legend texts
can be modied for curve plots.
Each page in a presentation has its own colour table, consisting of a set of single colours, and a
set of range colours. The single colours are used for plotting curves (Scalar Line and Scalar Time
plots), while the range colours are used for lled contours and shaded plots (2D, 3D Cut Plane
and Volume plots). There are a total of 220 range colours available. These can be organized as
220, 18, 12, or 6 different colours (hues) with 1, 10, 14 or 22 shades (light and saturation values)
for each colour. The light and saturation range is used for shading 3 dimensional plots.
5.6.5.1 Scalar Line and Scalar Time plots
The Variable Appearance dialog box for Scalar Line and Scalar Time plots contains four areas. The
uppermost area is the Entry Listcontaining a list of all entries in the plot. The settings in the rest
of the dialog box applies to the selected entry.
Below the entry list is an area for Value Range, controlling the plotted value range for the selected
variable. The area contains radio buttons for selecting Linear or Logarithmic interpolation. Linear
interpolation is default. Logarithmic interpolation is not available for particle plots.
Below the interpolation buttons are radio buttons for selecting Automatic, Fixed, or Variable value
range. Automatic means that the minimum and maximum values used in the plot equal the
minimum and maximum values found on the data le (for all timesteps). Selecting Fixed value
range, and entering the desired minimum and maximum values, makes it possible to manually
control the plotted value range. For Automatic and Fixed value range, the plotted value range is
constant for all timesteps. Selecting Variable, makes Flowvis adjust the plotted value range to the
minimum and maximum values for the displayed timestep. Variable value range is not available
for Scalar Time plots.
The Single Colour Display contains one rectangle for each single colour. The single colour asso-
ciated with the selected entry in the entry list is marked with an arrow. Another colour can be
selected by clicking MOUSE+LEFT in or below the desired rectangle.
5.6 Plot menu 173
Below the colour display are two option menus for selecting Curve Type and Marker Type. The
Index scale can be used for selecting the single colour index for the selected entry. The legend text
for the selected entry can be edited in the Legend Texteld .
5.6.5.2 2D Cut Plane, 3D Cut Plane, Volume and Particle Traces plots
The contents of the Variable Appearance dialog box for these plot types depends of the contents of
the selected plot, but three areas are always present. The rst is the Entry list which contains a
list of all entries in the plot. The settings in the rest of the dialog box applies to the selected entry.
The second area always present is the Value Range settings. This area is described in the previous
section.
The third area always present is the Range Colour Display which contains two horizontal rows
of colour rectangles. The upper row contains one rectangle for each range colour, while the
lower row contains the light and saturation range for the selected colour. If the selected entry
is a variable plotted for only one phase, two arrows mark the colour range that is mapped to
the value range. If the current entry is a variable plotted for several phase, only one arrow is
displayed. It marks the colour used for plotting the current variable/phase combination. Other
colour(s) are selected by clicking MOUSE+LEFT in or below the desired rectangle.
The contents of the rest of the dialog box depends on the entry that has been selected.
The selected entry is a variable plotted for only one phase
If the selected entry is a variable plotted for only one phase, two areas controlling the value to
colour mapping are included below the range colour display. The rst one contains elds dis-
playing the Minimumand Maximum Values in the plot, and Indexscales for setting the correspond-
ing range colour indices.
The next area controls the Colours for Values less than Minimum or greater than Maximum. It contains
two sets of radio buttons. The rst controls how to display values that are less than the minimum
value, and the second controls how to display values that are greater than the maximum value.
The alternatives are to display the value in the same colour as the minimum/maximum values, in
the previous/next colours in the colour table, or not to use any colour at all. The last alternative is
useful when making a honeycomb plot in a 3D Cut Plane plot (by specifying a xed value range
less than the total in the Plot Specication dialog box)
If the plot type is 2D Cut Plane plot, and the selected variable is a scalar variable, an options menu
for selecting Draw mode is included in the dialog box. The available options for drawing sin-
gle phase scalar variables are Filled (lled contours), Contours (iso-curves), or Filled and Contours
(lled contours combined with contour curves). The Contours command controls contours and
contour annotation. The Contours dialog box is described below.
If the selected entry is a variable plotted for several phases, a scale and a text eld are included
below the range colour display. The scale can be used for selecting the range colour index for the
selected entry, while the legend text for the selected entry can be edited in the text eld.
5.6.6 Contours
This command controls contours and contour annotation in 2D Cut Plane plots. The command is
available for 2D Cut Plane plots with scalar variable output for one phase, and is only available
if the current 2D plot is drawing contours.
This dialog box controls contours and contour annotations in a 2D Cut Plane plot. To turn on
contour drawing, select Contours (iso-curves), or Filled and Contours from the Draw options menu
in the Variable Appearance dialog box.
174 Flowvis
At the top of the dialog box is a scale for setting the number of contours. The next area is the
Contour List which contains a list of all contour levels. The settings in the rest of the dialog box
applies to the selected entry.
The Value at Contoureld is for editing the value of the selected contour, while the Linear Distribu-
tion push button is for performing linear distribution of all the contour values. Pressing the Clear
Annotationpush button clears all annotation for all timesteps.
To annotate a contour, press MOUSE+LEFT near the desired contour in the desired position The
contour annotation is only displayed at the time step for which it has been created. If the plot
is changed in a way that makes the annotation wrong, the annotation must be cleared rst. A
pop-up dialog box reminds the user of this.
5.6.7 Vectors
This command is available for 2D and 3D Cut plane plots with vector output. The command
controls the vector scaling, the vector projection mode, the vector arrow type and the vector
legend contents.
At the top of the Vectors dialog box are two radio buttons for selecting automatic or xed vector
scaling. Automatic, which is default, means that the vectors are scaled to t the minimum control
volume sizes. Fixed means that the user species the scaling factor.
Below the scale controls are toggle buttons for controlling the vector legend. The user can select
to include or not include arrows and colour boxes in the legend.
The next control is for selecting projection mode. This is only applicable for 3D plots. The user
may then select to draw the real 3 dimensional vectors, or the projections of the vectors onto the
cut planes.
The last controls decides the arrow types and opening angles for the arrows.
5.6.8 Geometry appearance
This command is available for 3 dimensional plots containing geometry, and not containing grid,
porosities or variable(s). The Geometry Appearance dialog box contains one toggle button for
turning on/off geometry shading. When shading is turned off, the geometry is displayed as
a wireframe model. For plots containing grid, porosities or variable(s), only shaded geometry is
available.
5.6.9 Blank and un-blank primitives
This command is available for 3 dimensional plots containing geometry. The command lets the
user hide selected primitives in the geometry plot.
The dialog boxes for blanking and unblanking are identical. When the dialog box is shown, the
geometry is drawn as a wireframe model. If the command is Blank Primitives, all visible (not
blanked) primitives are drawn. If the command is Unblank Primitives, all blanked primitives are
drawn.
Primitives are selected from the graphical area by clicking MOUSE+LEFT on any point inside
the primitive. To simplify the selection, the dialog box contains push buttons for traversing the
primitive list. Pressing the Closer push button selects the next primitive in the direction towards
the user, while pressing the Farther push button selects the next primitive in the direction away
from the user.
5.6 Plot menu 175
Pressing the Blank or Unblank Primitive push button marks the primitive for blanking/un-
blanking. Pressing the Apply push button redraws the wireframe model with selected primitives
removed.
This Unblank Primitives command is available for 3 dimensional plots containing geometry, and
where one or several primitives have been blanked. The command lets the user select primitives
for un-blanking in the selected plot.
5.6.10 Rotate
This command rotates the contents of the selected plot. The command is not available for empty
plots or annotation plots.
5.6.10.1 2D and line plots
While the mouse pointer is placed inside the Rotate dialog box, the following key functions are
available:
X, x Mirror about x-axis
Y, y Mirror about y-axis
Z, z Rotate about z-axis in steps of 90 degrees.
Pressing the upper case keys gives rotation in counter-clockwise (positive) direction, while lower
case keys gives rotation in the clockwise (negative) direction.
The dialog box contains a push button labelled Default View. Pressing this button reinstalls the
default transformation for the plot.
5.6.10.2 3D plots
While the mouse pointer is placed inside the Rotate dialog box, the following key functions are
available:
X, x Rotate about x-axis
Y, y Rotate about y-axis
Z, z Rotate about z-axis.
Upper case keys means positive rotation angle while lower case means negative.
The Rotate dialog box contains an options menu for selecting a projection. Available projections
are 3D View, XY View, YZ View and XZ View.
While rotating the plot, Flowvis draws a wireframe box marking the simulation volume extent
instead of the plot contents. The Apply push button causes the entire plot to be redrawn.
5.6.11 Particle traces specication
This command is only available for Particle Traces plots. To create a Particle Trace plot the variable
VVEC must be saved during the simulation and plotted in Flowvis.
176 Flowvis
Figure 5.5: A particle trace plot
The Particle Traces Specication dialog box contains four frames. The rst from the top is for
selecting Velocity Field Type, Trace Time, Number of Traces and Minimum Velocity. The next frames
are the Firing Specication, the Firing Volume Specication and the Animation Specication
There are two choices for the Velocity Field Type. Steady State means that the particles are traced
in a constant velocity eld given by the current timestep. Transient means that the particles are
traced in a varying eld from the current timestep to the last.
The innermost loop of the algorithm for drawing the particle traces is a loop through all particles
drawing the particle traces over a time interval given by the Trace Time. Pressing Ok or Apply
for Steady State draws only one trace. The Trace Time decides the length of the particle traces
directly. For Transient, the traces are drawn from the current timestep to the last. The No. of Traces
then decides the number of traces required for drawing the plot. A short trace time increases the
drawing time, but gives more accurate traces, and longer trace lines.
If the velocity of a particle gets lower than the Minimum Velocity, the particle is removed from
the plot and the trace of that particle is ended. In an animation, the particle disappears from the
screen.
The Firing Specication frame contains radio buttons for selecting Random or Fixed Firing Positions.
Random means that the particles are red at random positions inside the volume given by the
Firing Volume Specication. The number of particles red are given by the No of Particles eld.
If Fixed ring positions is selected, the particles are red from a xed grid given by the No of
Particles in X, Y and Z-dir. elds, and the Firing Volume Specication (frame below).
The Animation Specication frame controls particle animation. The No. of Frames Between Firing pa-
rameter gives the interval at which Flowvis res new particles. Flowvis then res a new particle
for each particle that has been killed since the last time. The two next elds decides the minimum
and maximum lifetime for a particle. The Trace Time decides the duration of each frame. A short
Trace Time gives shorter traces.
An animation is started by pressing the Draw Frames push button. If the Record toggle button is
5.6 Plot menu 177
selected, the frames are stored in memory, and can be replayed using the Play push button. The
frames may also be stored on les, by selecting the Record on les toggle button. The les are
stored on the directory where Flowvis was started, and named fl_framedd.png, where dd is
the frame number.
Remarks:
In the current version of Flowvis recording of particle plot images to le is not working.
Figure 5.6: The particle trace properties window
5.6.12 Next plot
Shortcut: CTRL+RIGHT
Choosing Next Plot causes the selection of the next plot. If the previously selected plot was the
last plot in the page, the page is selected. If the page was selected, the rst plot in the page is
selected.
178 Flowvis
5.6.13 Previous plot
Shortcut: CTRL+LEFT
Choosing Previous Plot causes the selection of the previous plot. If the previously selected plot
was the rst plot in the page, the page is selected. If the page was selected, the last plot in the
page is selected.
5.7 Verify menu
The verify menu enables the user to have a detailed look at the area and volume porosities, and
coordinates for a geometry, and pick exact values from lines curves.
5.7.1 Porosities
This command is available only for 2D Cut Plane plots. In addition to porosities, this command
makes it possible to verify coordinates, areas and volumes for control volumes.
This command lets the user select a control volume and verify the porosities, areas and volume
for this control volume. Two dialog boxes pops up. One is the Plot Domain dialog box. It is used
for selecting the cut plane to view. The other dialog box is the Verify Porosities dialog box. The
selected control volume is shaded in the plot.
Since a 2D Cut Plane plot may include several jobs, the Verify Porosities dialog box includes
a eld showing which job the selected control volume belongs to. In addition there are elds
displaying the selected control volume indices, and push buttons for navigating through the
control volumes.
The four next elds show the area and volume porosities for the selected control volume.
At the bottom of the dialog box are two push buttons for verifying Coordinates and Areas and Vol-
umes. Clicking the Coordinates push button makes the Verify Coordinates dialog box pop up. This
dialog box contains 8 x 3 elds showing the X, Y and Z coordinates for the eight corners of the
selected control volume. Clicking the Areas and Volumesbutton makes the Areas and Volumesdialog
box pop up. This dialog box contains 3 x 3 elds showing the X, Y and Zcomponents of the areas
in the three grid directions ( I, J and K). The last eld shows the volume.
Flowvis can be started in a special verify porosity mode, from the command line by using the
following command line option.
Linux:
> run9 flowvis -verify_por 010101
Windows:
> flowvis -verify_por 010101
010101 is the job number.
5.7.2 Value on curve
This command makes it possible to verify the values in Scalar Time and Scalar Line plots.
5.8 Options menu 179
The Values On Curve dialog box lets the user select a point on the abscissa (time for Scalar Time
plots and coordinate value for Scalar Line plots). The corresponding value, or values if several
curves, are shown in the value list in the dialog.
The point on the abscissa can be selected by clicking MOUSE+LEFT on the desired position in the
plot, or by typing the abscissa value in the input eld in the dialog, and then pressing the Select
button.
5.8 Options menu
Options are saved for each user under
Linux: /config/GexCon/Flowvis.conf
Windows: In Windows registry "HKEY_CURRENT_USER\Software\GexCon\Flowvis"
5.8.1 Header on
This option species whether a header is displayed on top of each page. The header contains
Flowvis version number, plotting date and page number.
5.8.2 Background white
By default the page background is white.
5.8.3 Redraw picture on OK
By default, the selected plot is redrawn when the user selects Ok from a popup dialog box which
changes the contents of the plot. This can be avoided by turning off this option. The time needed
for changing a plot through the use of several dialog boxes can thus be reduced.
5.8.4 Interpolate
Flowvis interpolates the results between grid cells, by default when showing 2D, 3D and Volume
plots. In some situations it can be necessary to turn this off, for instance when investigating
possible simulation errors.
5.8.5 Fonts
Font type and size can be changed separately for each of the text areas in a Flowvis plot. The font
setting is global for all plots and pages in the presentation.
Remarks:
Flowvis will not be able to scale all elements in the plots according to the size and type of the
fonts, thus in some cases the fonts will be covered, or cover the plot other graphical elements.
Clicking the Restore Defaults button will normally alleviate this problem.
180 Flowvis
5.8.6 Plots
The Plots options enables some control of the position of plot elements, such as.
Changing plot size
Moving the plot on the page
Moving and resizing the legend
In addition the user can select to have overrun text be highlighted to be warned if some text
elements are not shown on the plot.
5.9 Help
The Help menu gives access to the online help and license terms.
5.9.1 Online help
Shortcut: F1
The online help is shown in the Qt Assistant help browser.
In Qt Assistant the user has full access to a searchable index in addition to a full text search in the
user manual. It is possible to have several tabs open, create bookmarks and print selected pages.
The FLACS Users manual is also available as a single le pdf (Portable Document Format).
5.9.2 Licence terms
Shows the FLACS license terms, including the specic license type for the specic FLACS instal-
lation.
5.9.3 About Flowvis
Shows about Flowvis dialog.
5.10 Flowvis examples
This section gives two examples on Flowvis use. Results from a FLACS simulation in the M24
module are used in both examples. The grid resolution is 40x13x12 m
3
. The simulation les can
be copied from your FLACS installation directory.
Linux:
> cp /usr/local/GexCon/FLACS_v9.0/doc/examples/ex1/
*
110101
*
.
Windows:
Copy les fromC:\Program Files\GexCon\FLACS_v9.0\doc\examples\ex1\110101
(110101 means all les containing the text "110101")
5.10 Flowvis examples 181
The simulation must be run before creating the presentation:
On Linux:
> cd <simulation file directory>
> run9 porcalc 110101
> run9 flacs 110101
On Windows start the FLACS runmanager, click the Add directory button, and select the directory
where the simulation les are located. Select the job number in the list, and click the Simulate
button.
When the simulation has nished continue to the next section.
5.10.1 Creating a presentation with multiple plots
This example illustrates how to create a simple presentation from scratch.
In the examples, there are several tasks which implies the use of several dialog boxes to dene a
plot. This can be time consuming because the plot is redrawn every time an Ok button is clicked.
To avoid this, turn the Redraw Plot On Ok option off (in the Options menu). Use the Redraw Plot
button in the main window when a redraw is wanted.
This example covers 2D and 3D Cut Plane plots, Scalar Time plots and Monitor Point plots.
The presentation shall contain two pages. The rst shall contain 2 dimensional cut plane plots
(the JK cut plane with grid index equal to 20) showing the combustion product, pressure and
velocity at timestep 17. There shall also be a plot showing the position and orientation of the cut
plane compared to the geometry. The porosities in the cut plane shall also be visualized.
The second page shall contain a scalar time plot and a plot showing the positions of the monitor
points. At last, it is outlined how to adapt the presentation to the results from a new simulation.
1. Start Flowvis
2. Change directory to where the simulation results are
(a) Select the New At command from the File menu. The New At dialog box makes it
possible to navigate through the directory structure by double clicking on directory
names. When the desired directory is reached, press the Ok button.
3. Create a page
(a) Select the Add Page command from the Page menu. Select the Modify Page command
and change the subdivision of the page to 2 plots in X-direction and 3 in Y-direction.
4. Create a plot which shows the geometry and cut planes in 3 dimensions
(a) A 3D Cut Plane plot is used for this purpose. Click MOUSE+LEFT while pointing the
mouse pointer somewhere near the top left corner of the page. A frame indicates the
selected plot.
(b) Select the Plot Type command from the Plot menu. Select 3D Cut Plane plot from the
submenu. The Data Selection dialog box pops up. Select the appropriate job number.
When the variable list is displayed, select PROD (combustion product). When the Ok
button is pressed, a plot with default settings is displayed. This means that three cut
planes are displayed, one perpendicular to each axis direction.
(c) Select the Plot Domain command from the Plot menu. The Plot Domain dialog box pops
up. Set the number of IJ and IK planes equal to 0 (keep the number of JK planes equal
to 1). The Start Index for the JK plane is set to 20. Press the Ok button.
182 Flowvis
(d) Select the geometry utility in the Plot Specication dialog box, and press the Ok button.
5. Some primitives need to be blanked to make it possible to see inside the geometry.
(a) Select the Blank Primitive command from the Plot menu. The geometry is redrawn as
a wireframe model.
(b) When the dialog box pops up, click MOUSE+LEFT while pointing the mouse pointer at
the roof of the geometry. The selected primitive is highlighted.
(c) Press the Blank Primitive button to blank the roof. Do the same thing to the XZ wall
closest to the viewer. If it is difcult to hit the wanted primitive when clicking, the
Closer and Farther buttons makes it possible to loop through all primitives intersecting
the mouse pointer position.
(d) The plot is redrawn with shaded geometry when pressing the Ok button.
6. Create an Annotation plot
(a) Select the plot to the right of the previous plot by clicking MOUSE+LEFT while point-
ing the mouse pointer somewhere in that area. Select the Plot Type command, and
Annotation from the submenu.
(b) The Annotation plot is now displayed. It contains the job number, iteration number,
time, minimum and maximum grid indices from the 3D Cut Plane plot. The corre-
sponding text is removed from the 3D Cut Plane plot which is redrawn because the
removed text makes room for larger graphics.
7. Create a 2D Cut Plane plot for visualizing the porosities
(a) Select the plot below the 3D Cut Plane plot. Select the Plot Type command and 2D Cut
Plane from the submenu. Select the appropriate job number in the Data Selection dialog
box, and press the Ok button.
(b) Select the Plot Specication command. Select the Area Porosities and Volume Porosities
utilities and deselect the Axis and Grid utilities. Press the Ok button.
(c) Select the Plot Domain command. Select the JK plane and the appropriate grid index
(20) using the I indexslider. Press the Ok button to see the result.
(d) Note that the Annotation plot changes when this 2D Cut Plane plot is created. In-
stead of the minimum and maximum grids indices, the Annotation plot now contains
information about the plane displayed in the 2D Cut Plane plot.
8. Create 2D Cut Plane plots of the combustion product and the pressure
(a) Select the Copy command fromthe Edit menu to copy the selected plot to the clipboard.
Select the plot to the right, and select the Paste command. A copy of the previous
plot is displayed. Select the Data Selection command from the Plot menu, and select
the combustion product (PROD) from the variable list. Press Ok and select the Plot
Specication command. Deselect the Area Porosities and Volume Porosities utilities and
select the Geometry utility. Press Ok to see the result.
(b) Copy this plot to the clipboard and paste it in the next plot. Select the Data Selection
command and select the P variable instead of PROD.
9. Create a 2D Cut Plane plot of the velocities
(a) Select the last plot in the displayed page and select the Paste command. The contents
of the clipboard is copied to the selected plot. Select the Data Selection command and
select the VVEC variable instead of PROD.
(b) To get longer vectors, select Variable Value Range in the Variable Appearance dialog box
and select Fixed Scale in the Vectors dialog box on the Plot menu. Set the scaling factor
equal to 1.5.
10. Add a new page
(a) Select the Add command from the Page menu. The new page is added after the previ-
ous. Modify the page subdivision to be two plots in Y-direction.
11. Create a Scalar Time plot of the Pressure
(a) Select the top plot in the page. Select the Plot Type command and Scalar Time from
the submenu. Select the appropriate job, the P variable (pressure) and the wanted
monitor points in the Data Selection dialog box. A sequence of adjacent (in the list)
monitor points can be selected by moving the mouse pointer over the items while
holding MOUSE+LEFT down. Several monitor points can also be selected by clicking
while holding the Ctrl key down. Upon Ok, the plot is drawn.
12. Create a Monitor Point plot showing the points used in the previous plot
(a) Select the bottom plot, and select the Plot Type command and Monitor from the sub-
menu. Select the appropriate job, and the same monitor points as in the previous plot.
(b) Press the Monitor Point Listtoggle button in the Plot Specication dialog box.
13. Save the presentation
(a) Select the Save command from the File menu. Since the presentation is new, no le
name is associated with it. The Save As dialog box is therefore displayed. Specify a
name and press the Ok button.
14. Adapt the presentation to the results from a new simulation
(a) Start a new Flowvis instance. Select the Open command from the File menu. Select
the le specied in the previous step. Select the Substitute Job command from the Edit
menu. Specify the previous and the new job numbers. Select the All Pages option and
press the Ok button.
Figure 5.7: The rst page of the Flowvis presentation in example 1 (location of plots differs
slightly from text)
184 Flowvis
Figure 5.8: The second page of the Flowvis presentation in example 1 (location of plots differs
slightly from text)
5.10.2 Creating a honeycomb plot
This example shows how to create a honeycomb plot by specifying a limited visual value range
in a 3D Cut Plane plot.
1. Create a 3D Cut Plane plot
(a) Open the presentation from the previous section.
(b) Copy the 3D Cut Plane plot from the rst page.
(c) Add a new page and paste the plot from the clipboard.
(d) Select the Plot Domain command. Change the number of cut planes in each direction
to be equal to the number of grid indices in that direction. The number of grid indices
can be seen from the scales in the volume specication.
(e) Turn off the grid display in the Plot Specication dialog box.
2. Modify the variable appearance
(a) Select the Variable Appearance command. Select Fixed Value Range Setting, and set the
Fixed minimum value to be equal to 50% of the maximum value for All Timesteps. The
Fixed maximum value is set to be equal to that of All Timesteps. Set the Colour for val-
ues < min. to be None. Press the Ok button. The combustion product is drawn as a
honeycomb volume covering the volume where the combustion product is above the
specied limit.
3. Change the colour table
(a) To get more ame-like colours, select the Range Colours command from the Page menu.
Select the rst index using the index slider. Set the hue for this index to be equal to 0.
Select the last index and set this hue to be 60. Press the Interpolate button. Press Ok to
see the result.
Figure 5.9: The honeycomb plot
186 Flowvis
Chapter 6
Utility programs in FLACS
188 Utility programs in FLACS
This chapter describes the utility programs in FLACS. The user can run the programs from the
command line in the terminal window under Linux, or on the command line in the cmd window
under Windows. It is possible to start the cmd command shell from the FLACS Run Manager.
The utility programs are categorized according to application:
1. Geometry, grid, and porosities: geo2acs, gm, and Porcalc
2. Release source modelling: Jet, Flash
3. Modifying simulation les: rdle, cole, comerge
4. Post-processing simulation data: r1le, r3le, a1le
6.1 Geometry, grid and porosities
6.1.1 geo2acs
geo2acs is a utility designed to import geometry from various 3D geometry formats to FLACS.
The following input formats are supported:
Microstation dgn version 7
AutoReaGas
AutoCAD (when contained in dgn le)
StereoLitography le (STL)
In addition to the dgn format geo2acs can export from the prp format (Propagated steel, Frame-
works format, very similar to dgn format). Often the dgn les are accompanied by a drv le
(Design Review). geo2acs can use the information in the drv le to group single primitives into
larger objects and give the objects the proper names.
The dgn le format can be produced by a number of different software packages. It is the native
format of Microstation, but ie. PDMS can be exported to dgn format using the PDMS module
ExPlant-I.
Microstation can open AutoCAD les and save them in dgn format. The internal structure of the
geometry is not changed, and export from these types of les sometimes result in incomplete or
completely empty geometries.
The output format is CASD macro les.
6.1.1.1 Usage
geo2acs is a command line program using mandatory and optional arguments.
The conversion is a three-step process. During the export process itself the geometry les are
read via an input le.
Running the program with
"--help"
option prints a help message:
Linux:
> run9 geo2flacs --help
6.1 Geometry, grid and porosities 189
Windows:
> geo2flacs --help
The above command gives in the following output:
Usage: geo2flacs [OPTION...] DIRECTORY
Geometry conversion tool for FLACS. Converts Microstation,
AutoReaGas, STL, and other 3D geometry formats, to CASD
macro files.
-a, --noAlign Do not align objects to nearest axis
and allow irregular objects (special
version of CASD required)
-c, --create Create input file
-d, --minDia=DIAMETER Do not export objects with diameter
smaller than DIAMATER (mm)
-e, --export Start export of files
-f, --force Force delete output directory before
exporting
-g, --gexconCol Gexcon colour scheme
-l, --minLength=LENGTH Do not export objects with largest
length smaller than LENGTH (mm)
-n, --group=NUMBER Group NUMBER of primitives into one object
-o, --outdir=DIR Output sub-directory (default = out)
-r, --rotate=ANGLE,X,Y,Z Rotate geometry ANGLE degrees around (X,Y,Z)
-s, --translate=X,Y,Z Translate geometry (X,Y,Z) before export
-t, --terrain[=HEIGHT] Use HEIGHT as bottom value for terrain
export
-v, --verbose Produce verbose output (increase
diagnostic output)
-x, --addLength Add small length to cylinders with 0
length which would otherwise be skipped
-z, --buildingBase=Z0 Create buildings starting at Z0 (only valid for
Shape files)
-?, --help Give this help list
--usage Give a short usage message
-V, --version Print program version
Mandatory or optional arguments to long options are also mandatory
or optional for any corresponding short options.
Object type specification:
P: Piping
E: Equipment
S: Steel
R: Race Way
T: Terrain
B: Bounding Box
Specify option in the "geo2flacs_files.txt" file.
Report bugs to <flacs@gexcon.com>.
6.1.1.2 Step-by-step procedure
1. Begin the export process by creating a new directory and copy all les to be converted to
this directory.
2. Run geo2acs rst time with argument -c: "<tt>run9 geo2acs -c .</tt>". This will create
the input le geo2acs_les.txt in the current directory.
3. Open the input le geo2acs_les.txt in a text editor (eg. emacs, vi, pico, kwrite, etc and add
export arguments as described in section Input le parameters. Save the le.
4. Run geo2acs again with argument -e to do the export: "<tt>run9 geo2acs -e .</tt>".
This will create a sub directory out if it does not exist, and start the export. The exported
les are created in the out directory. If the out directory contains exported les geo2acs
will exit without exporting. A log le geo2acs.log will be created.
5. Start CASD, create a database and read the macro les created by geo2acs. To do this go
Macro Run ... and select the le 0-RunMe.mcr. The geometry will be read into FLACS.
Note that if the geometry has more than 10000 objects (2000 in FLACS versions prior to 9.0)
CASD must be started with arguments -numObj XXXXXX -numAsis XXXXXX, where
XXXXXX is the number of objects. The number of objects created is reported at the end
of the export.
6.1.1.3 Input le parameters
If a dgn or prp le does not have an accompanying drv le geo2acs must have information
about the type of geometry inside the le. This is done in the input le geo2acs_les.txt. When
this le is initially created all dgn les without a drv le will have a trailing "-". Change the "-"
into one of the following letters.
Parameter Geometry type
P Piping
E Equipment
S Steel
R Race Way
T Terrain
B Bounding Box
Table 6.1: Input le parameters
The parameters P, E and S are most commonly used. R is used if the le contains only Race
Way elements. T is for importing grid or triangular terrain geometry. B can be used to export
geometry from les geo2acs otherwise is unable to export from.
6.1.1.4 Command line options
The following command line options are recognized by geo2acs:
Short Option Long Option Description
-a noAlign Do not align objects to
nearest axis and allow
irregular objects (special
version of CASD required)
-c create Create input le
-d minDia=DIAMETER Do not export objects with
diameter smaller than
DIAMATER (mm)
-e export Start export of les
-f force Force delete output directory
before exporting
-g gexconCol Gexcon colour scheme
-l minLength=LENGTH Do not export objects with
largest length smaller than
LENGTH (mm)
6.1 Geometry, grid and porosities 191
-n group=NUMBER Group NUMBER of
primitives into one object
-o outdir=DIR Output sub-directory
(default = out)
-r rotate=ANGLE,X,Y,Z Rotate geometry ANGLE
degrees around (X,Y,Z)
before export
-s translate=X,Y,Z Translate geometry (X,Y,Z)
before export
-t terrain[=HEIGHT] Use HEIGHT as bottom
value for terrain export
-v verbose Produce verbose output
(increase diagnostic output)
-x addLength Add small length to
cylinders with 0 length
which would otherwise be
skipped
-z buildingBase=Z0 Create buildings starting at
Z0 (only valid for Shape
les)
-? help Give this help list
usage Give a short usage message
-V version Print program version
Table 6.2: geo2acs command line options
DIRECTORY is mandatory when using the "-c" or the "-e" argument. The directory can either be
the current directory (use ".") or any other directory (use full path or relative path).
Using the "-f" argument will force overwriting of any existing les in the out directory.
"-t HEIGHT" is used when exporting terrain data. If HEIGHT is negative use the long option, eg:
--terrain=-10.0
If one or more of the coordinates are negative when using the translate option the coordinates
must be specied like this:
-s "\-100,\-100,\-100"
Note the double quotes ("") around the coordinates.
The argument "-g" force geo2acs to export geometry using the Gexcon colour scheme. By default
geo2acs will try to preserve the colours found in the dgn le.
"-a" is a special purpose option that prevents objects from being rotated to the nearest axis. This
requires a special version of CASD and is not validated for general use.
"-v" will output as much information as possible during the export.
6.1.1.5 Geometry verication
As with all conversion from one format to another the process cannot be guaranteed to produce
correct results. It is therefore very important to do a thorough verication of the geometry in
CASD. Because CASD only offers a subset of the functionality in most CAD packages some in-
formation will be lost and some information might be interpreted with errors.
It is important to have special focus on details having signicant inuence on the simulation re-
sults. These includes walls, decks (with and without grating) and major obstacles. The degree of
porosity of the grating can have a signicant inuence on explosions, but no porosity information
is exported.
6.1.1.6 Problems and solutions
Note that geo2acs is under continuous development. This means that the software will have
bugs, although Gexcon has done signicant testing of the program to ensure that the program
will operate well in most situations. Contact GexCon if any bugs are found. GexCon will consider
xing the bug if it is part of the core functionality of the program.
Increased diagnostic output geo2acs will output information during the export. Increased
diagnostic is written if the "-v" option is used. This information can be redirected to a le by
using the ">" sign:
Linux:
> run9 geo2flacs -v -e . > out.txt
Windows:
> geo2flacs -v -e . > out.txt
This will create a le out.txt in the current working directory.
Dgn version supported The current version of geo2acs supports dgn le version 7. Version 8
is not supported and must be converted to version 7 using Microstation.
Note about PDMS ExPlant-I There is bug in ExPlant-I that sometimes exports geometry in
a format that cannot be exported with geo2acs. If an object is represented by only "Shape"
elements it will not be exported. How to x this in ExPlant-I is not known to GexCon, but a
knowledgable ExPlant-I operator will be able to work around the problem.
6.1.2 gm
The gm program allows making grids by scripting. The following command starts the gm pro-
gram:
Linux:
> run9 gm
Windows:
> gm
6.2 Release source modelling 193
The following sequence of commands creates a homogeneous grid of 1 m resolution in all direc-
tions over a simulation domain extending from 0 m to 10 m in all directions.
Executing "gm"
gm> x add 0:10:1
gm> y add 0:10:1
gm> z add 0:10:1
gm> save cg999999.dat3
The grid is saved in a le named cg999999.dat3. Assuming a 999999-job number this grid can be
used in a FLACS simulation. The gm program can read the FLACS cg-le and print the position
of the gridlines on the screen:
Executing "gm"
gm> open cg999999.dat3
gm> x print
gm>
x 11
0.000 1.000 2.000 3.000 4.000 5.000 6.000 7.000 8.000 9.000 10.000
gm> y print
gm>
y 11
0.000 1.000 2.000 3.000 4.000 5.000 6.000 7.000 8.000 9.000 10.000
More options are available with the gm-program and can be listed typing the command help in
the gm-interface:
gm> help
6.1.3 Porcalc
The Porcalc program starts either from the Calculate option under the Porosities menu in
CASD, or from command line input.
Linux:
> run9 porcalc 010100
Windows:
> porcalc 010100
6.2 Release source modelling
6.2.1 Jet
The jet program is used to compute the area and the subsonic velocity after shock of a jet issuing
from a high-pressure reservoir. The jet program is started with the following command:
Linux:
> run9 jet < jet-in > jet-out
Windows:
> jet < jet-in > jet-out
Below is an example of a basic jet-in le:
cl-file ! output format (-,
*
, cl-file, ...).
METHANE=0.8,ETHANE=0.1,PROPANE=0.04,BUTANE=0.06 ! gas type (AIR, METHANE, ...)
100 ! reservoir volume (m3)
50 30 ! reservoir pressure (barg) and temperature (C)
1 30 ! atmospheric pressure (bara) and temperature (C)
0 0 ! heat transfer coefficients (J/s) and (J/sK)
30 ! wall temperature (C)
0.1 0.62 ! nozzle diameter (m) and discharge coefficient (-)
0 ! start time (s)
1 10 ! time step (s) and number of iterations (-)
"+XJ" ! leak control string
1e-5 1e5 ! shutoff pressure (barg) and release mass (kg)
0.1 ! relative turbulence intensity RTI (-)
0.1
*
D ! turbulence length scale TLS (m) + function
Most of the parameters can be dened intuitively, a brief description of the most unconventional
parameters is given below.
The rst line says that the output of the jet program will directly be written with the format of a
cl-le, the leak le used to dene gas releases in FLACS. This is the most useful output format
and it will be described further in the following.
The heat transfer coefcients do not need to be dened, the default zero values can be conserved.
The time step and the number of iterations control the duration and the times at which the prop-
erties of the jet will be written in the cl-le. In this example, the properties will be written in the
cl-le each seconds over a duration of 10 s.
The leak control string will be directly added into the cl-le for the denition of the direction of
the leak.
The output le generated by the jet program used with the previous inputs is:
Starting jet
Enter output format (-,
*
, cl-file, ...):
Enter gas type (AIR, METHANE, ...):
Enter reservoir volume (m3):
Enter reservoir pressure (barg) and temperature (C):
Enter atmospheric pressure (bara) and temperature (C):
Enter heat transfer coefficients (J/s) and (J/s/K):
Enter wall temperature (C):
Enter nozzle diameter (m) and discharge coefficient (-):
Enter start time (s):
Enter time step (s) and number of iterations (-):
Enter leak control string:
Enter shutoff pressure (barg) and release mass (kg):
Enter relative turbulence intensity RTI (-):
Enter turbulence length scale TLS (m) + function:
#
# GENERAL:
# gas type = METHANE=0.8,ETHANE=0.1,PROPANE=0.04,BUTANE=0.06
# mole weight = 21.093 kg/kmol
# heat ratio, kappa = Cp/Cv = 1.291 -
# critical pressure ratio = 0.547 -
#
# RESERVOIR:
# critical pressure = 0.827 barg
# pressure = 50.000 barg
# temperature = 30.000 C
# density = 42.679 kg/m3
6.2 Release source modelling 195
# volume = 100.000 m3
# initial mass = 4267.876 kg
#
# MACH:
# speed of sound, M=0 = 392.713 m/s
# speed of sound, M=1 = 366.957 m/s
# pressure, M=1 = 26.919 barg
# temperature, M=1 = -8.460 C
# density, M=1 = 26.758 kg/m3
# maximum velocity = 1030.249 m/s
#
# NOZZLE:
# effective diameter = 27.953 mm
# effective area = 613.675 mm2
# discharge coefficient = 0.620 -
# sonic massflow = 6.026 kg/s
# jet force = 3863.166 N
#
# ATMOSPHERIC:
# pressure = 1.000 bara
# temperature = 30.000 C
# density = 0.837 kg/m3
+XJ
7
TIME (s) AREA (m2) RATE (kg/s) VEL (m/s) RTI (-) TLS (m) T (K)
0.0000E+00 3.3311E-02 6.0258E-02 2.0741E+00 1.0000E-01 2.0594E-02 2.9086E+02
5.0000E-01 3.3311E-02 6.0258E+00 2.0741E+02 1.0000E-01 2.0594E-02 2.9086E+02
1.0000E+00 3.3280E-02 6.0209E+00 2.0739E+02 1.0000E-01 2.0585E-02 2.9080E+02
2.0000E+00 3.3219E-02 6.0112E+00 2.0734E+02 1.0000E-01 2.0566E-02 2.9068E+02
3.0000E+00 3.3159E-02 6.0015E+00 2.0730E+02 1.0000E-01 2.0547E-02 2.9056E+02
4.0000E+00 3.3098E-02 5.9918E+00 2.0726E+02 1.0000E-01 2.0528E-02 2.9045E+02
5.0000E+00 3.3037E-02 5.9821E+00 2.0722E+02 1.0000E-01 2.0510E-02 2.9033E+02
6.0000E+00 3.2977E-02 5.9724E+00 2.0718E+02 1.0000E-01 2.0491E-02 2.9021E+02
7.0000E+00 3.2917E-02 5.9628E+00 2.0714E+02 1.0000E-01 2.0472E-02 2.9009E+02
8.0000E+00 3.2857E-02 5.9531E+00 2.0710E+02 1.0000E-01 2.0453E-02 2.8997E+02
9.0000E+00 3.2797E-02 5.9435E+00 2.0706E+02 1.0000E-01 2.0435E-02 2.8985E+02
9.5000E+00 3.2797E-02 5.9435E-02 2.0706E+00 1.0000E-01 2.0435E-02 2.8985E+02
The rst 15 lines of this le concern the input parameters dened in the jet-in le. The 34 follow-
ing lines summarize the properties of the jet and physical parameters used in the computations.
Finally, the last 15 lines constitute the cl-le that will be used by FLACS for a dispersion simu-
lation. Therefore, providing the fact that in CASD, for a job number 999999 a leak position has
been dened, this le (with the 49 rst lines removed) can be renamed as cl999999.n001 and di-
rectly used for a dispersion simulation. In the new version of the jet program, specifying the
option -silent in the command line, automatically removes the 49 rst lines of the output le. The
command line is:
Linux:
> run9 jet -silent < jet-in > jet-out
Windows:
> jet -silent < jet-in > jet-out
6.2.2 Flash
The ash program is a utility that computes the physical properties of ashing releases of pres-
surized liqueed gas. The term ashing is usually used to describe vapor formation by pressure
changes. Many materials (such as propane, ammonia or chlorine for example) commonly stored
as pressurized liquids in the industry can ash as released into the atmosphere.
For such pressurized storage conditions the release has, in the free atmosphere, the appearance
of a two-phase jet composed of droplets and vapor. The image below summarizes the thermody-
namic state of the material as the distance with the release location increases.
Figure 6.1: The Flash utility
The end of the near eld region is dened as the position where all the liquid in the jet has
changed phase. At this position, denoted xf , the jet is in a single vapor phase and assuming that
all the required properties needed to dene a gas leak in FLACS are known, the ashing release
can be treated as a so-called jet leak in FLACS.
The following command starts the ash program.
Linux:
> run9 flash
Windows:
> flash
The ash program can handle ashing releases of 9 different species, namely:
acetylene
ammonia
butane
chlorine
ethane
ethylene
methane
propane
propylene
The following inputs are needed:
Area of the exit orice
6.3 Modifying simulation les 197
Temperature of the liqueed gas at the exit orice
Value of the discharge coefcient (by default this value is set to 0.62)
Mass ow rate or pressure at the orice
Temperature of ambient air
Finally, from these ve inputs the ash program gives the following outputs:
Position xf where all the uid is in a single vapor phase
Area of the jet at the position xf
Velocity of the jet at the position xf and mass ow rate
Mass fraction of air and released material at the position xf
Mass fraction of released material that rained-out and formed a pool on the ground.
In the previous list, the rst three outputs can directly be used to dene a leak in FLACS, the
fourth output allowderiving the value for the equivalence ratio of the released material in FLACS
and if signicant, the last output should be part of a pool setup (see pool model).
6.3 Modifying simulation les
6.3.1 rdle
The rdle program is mainly used to adapt a certain dump-le to a new grid resolution. Consider
a job number 999999 with a given simulation volume and a given grid resolution. Consider a job
number 888888 with the same simulation volume than the job number 999999 but with a different
grid resolution. Assume that all the other parameters of the two job numbers are the same and
that a dump le rd999999.n001 exists. The following command creates a dump le rd888888.n001
from the dump le rd999999.n001:
Linux:
> run9 rdfile rd999999.n001 rd888888.n001
Windows:
> rdfile rd999999.n001 rd888888.n001
The command
Linux:
> run9 rdfile
Windows:
> rdfile
lists all the options available for use with the rdle program:
Starting rdfile
FLACS rdfile (version 1.3, 2005-06-24)
Copyright 2005, GexCon AS
usage: rdfile in_file out_file [options]
options (enter in any order after file names):
force force overwrite
version=string set file version
setup=file_name use setup file
convert[=MIX,H,TAU] convert fields
block=number set block number
time=value set time
iter=value set iteration count
6.3.2 cole
The cole program extracts ASCII data from the FLACS geometry le co-le. The program reads
a co-le and writes obstacle size distribution to the screen. The specication of the option list=?
with the cole program leads to the full list of primitives. The cole program can also list the total
length of cylinders/boxes for each diameter size classes providing the option(s) classes_cyl=c1,c2
or/and classes_box=b1,b2,.,.
The following command lists all the options available for use with the cole program.
Linux:
> run9 cofile
Windows:
> cofile
The output of this command is:
Executing "cofile"
FLACS cofile (version 1.0, 2005-05-02)
usage: cofile file_name [options ...]
options (enter in any order after file_name):
region=x,X,y,Y,z,Z region of interest
classes_cyl=c1,c2,,, classes of cylinders
classes_box=b1,b2,,, classes of boxes
inch=0.0254 set inch to meter scale
accu=0.001 set size accuracy
plot=0/1 plot=no/yes (using gnuplot)
list=0/1 list=no/yes
silent=1/2 do not write to screen/file
6.3.3 comerge
The comerge program is used to create new FLACS geometry co-les from existing co-les. Con-
sider a job number 999999 with a given geometry. The following command creates a new co-le
co888888.dat3 for the job number 888888:
Linux:
> run9 comerge region=x_min,x_max,y_min,y_max,z_min,z_max co999999.dat3 co888888.dat3
Windows:
6.3 Modifying simulation les 199
> comerge region=x_min,x_max,y_min,y_max,z_min,z_max co999999.dat3 co888888.dat3
Note: To ensure that region=... is a single argument it must either be without spaces or embedded
between quotation marks.
> ... region=x_min,x_max,y_min,y_max,z_min,z_max
or
> ... "region= x_min , x_max , y_min , y_max , z_min , z_max"
but not
> ... "region = ..."
The geometry of the new job number 888888 is the same than the geometry of the job number
999999 in the region specied in the command line.
Several co-les from existing job numbers can be used to generate a new co-le. The command:
Linux:
> run9 comerge
Windows:
> comerge
lists all the options available for use with the comerge program.
Executing "comerge"
FLACS comerge (version 1.2, 2005-05-10)
usage: comerge [transform] input_file[s] output_file [force]
enter transform before each input_file
init identity transform
translate:tx,ty,tz translate in x,y,z directions
turn:axis,angle,x,y,z turn around axis at point x,y,z
axis = x/y/z, angle = +-90
*
N (+ is CCW)
skip:box skip boxes
skip:cyl skip cylinders
skip:col= skip objects with given colour hue
only:col= only objects with given colour hue
box=cyl convert boxes
beam=cyl convert composite beams (T/I/H/U shaped)
only only keep the converted objects
show show the objects to be converted
skip skip the converted objects
max_W=value set maximum beam width
min_L/W=value set minimum beam length/width ratio
min_T/W=value set minimum beam thickness/width ratio (box beams)
max_T/W=value set maximum beam thickness/width ratio (composite beams)
6.4 Post-processing of simulation data
6.4.1 r1le
The r1le program extracts ASCII data from the FLACS r1-le. Considering a job number 999999
the basic command is:
Linux:
> run9 r1file r1999999.dat3 name=NP force
Windows:
> r1file r1999999.dat3 name=NP force
The previous command creates an ASCII le named a1999999.NP containing the time-history of
the variable NP measured at all the monitor points dened in the job number 999999. The option
force overwrites the le a1999999.NP if it already exists.
The name of the output le can be set using "output=". The following example generates a le
ABC.NP from a le r1999999.dat3:
Linux:
> run9 r1file r1999999.dat3 name=NP force output=ABC
Windows:
> r1file r1999999.dat3 name=NP force output=ABC
Typing the command:
Linux:
> run9 r1file
Windows:
> r1file
lists all the options available for use with the r1le program:
Executing "r1file"
FLACS r1file (version 1.0, March 1998)
Copyright 1998, Christian Michelsen Research AS
usage: r1file file_name [options ...]
name=string variable name
output=string output file name
format=ascii/binary output format
time=start,finish output time range
monitors=+-::,
*
,...? monitor point list:
=+- set, add or remove
:: first:last:step, positive numbers
*
all (same as : or ::)
, separator
? last character, print monitor map
example "monitors=1:20,-3,5,7,+31,33"
example "monitors=2:10:2,13,19?"
6.4 Post-processing of simulation data 201
6.4.2 r3le
The r3le program extracts ASCII data from the FLACS r3-le. It may also process the data in
the r3-le. Considering a job number 999999 the basic command is:
Linux:
> run9 r3file r3999999.dat3 name=NP force
Windows:
> r3file r3999999.dat3 name=NP force
The previous command creates an ASCII le named a3999999.NP containing the values of the
variable NP over the entire simulation domain dened in the job number 999999. The option
force overwrites the le a3999999.NP if it already exists.
In addition to the extraction functionalities, the r3le program can process the data of the NF-
DOSE variable. Assuming a job number 999999 and that the le r3999999.dat3 contains outputs
of the variable NFDOSE at regular time intervals (i.e. the time intervals are given by the DTPLOT
variable in the cs999999.dat3 le) the r3le program can compute an average dose. For example,
considering the following command:
Linux:
> run9 r3file r3999999.dat3 dose=2 name=FDOSE force
Windows:
> r3file1.3 r3999999.dat3 dose=2 name=FDOSE force
This will produce an average dose over dosedtplot seconds which in our case gives a 20 s average
period.
Typing the command
Linux:
> run9 r3file
Windows:
> r3file r3999999.dat3
lists all the options available for use with the r3le program:
Starting r3file
FLACS r3file (version 1.3, 2006-03-15)
usage: r3file file_name [options ...]
name=string variable name
output=string output file name
format=ascii/binary output format
time=value output at time
interpolate[=0/1] time interpolation
grid[=0/1] grid output
load=n### load rd-file.n###
dump=n### dump rd-file.n###
gridfit[=0/1] fit region to grid
dose=integer steps for dose
dose/time=integer steps for dose/time
dtplot=value time between plots
dteps=value time proximity
zero_apor=value apor<value : apor=0
zero_vpor=value vpor<value : vpor=0
small_por=value vpor<value : ?por=0
mix=value mixture mole scale
verbose[=0/1] verbose output
6.4.3 a1le
The a1le program is used to process ASCII les with multicolumn data. Consider a job number
999999 and an ASCII le a1999999.NP containing the pressure measurements at three different
monitor points. The following command writes the time-integrals of the pressure at the three
monitor points:
Linux:
> run9 a1file a1999999.NP force integrate :1 :2 :3
Windows:
> a1file a1999999.NP force integrate :1 :2 :3
The integrals are written in the le a1999999.NP.out. Here is another example with the a1le
program:
Linux:
> run9 a1file a1999999.NP force slope=0.25,0.75
Windows:
> a1file a1999999.NP force slope=0.25,0.75
The previous example writes the slopes on screen (slope between 0.25 and 0.75 of maximum
value). Finally, the last example shows how to process data by combining columns of the data
le:
Linux:
> run9 a1file a1999999.NP force integrate :a=1-2
Windows:
> a1file a1999999.NP force integrate :a=1-2
This last example, writes the time integral of the expression data in column 1 minus data in column
2 into the a1999999.NP.out le.
The following command:
Linux:
6.4 Post-processing of simulation data 203
> run9 a1file
Windows:
> a1file
lists all the options available for use with the a1le program:
Starting a1file
FLACS a1file (version 1.0, 2005-12-31)
usage: a1file file_name [options ...]
clamp=start,finish use clamping
clip=start,finish clip data
scale=scale scale data
:1=2-3 calculate and output
:1 just output
slope=low,high calculate slope
integrate calculate integral
Chapter 7
Best practice examples
206 Best practice examples
This chapter presents examples and best practice guidelines for FLACS. The examples include
simulations that require specialized versions of FLACS, such as FLACS-Fire or DESC.
7.1 Combined dispersion and explosion simulations with
FLACS
There are at least three different ways to perform a combined dispersion and explosion simula-
tion in FLACS:
Run a dispersion simulation where ignition time and position are set before you start
the simulation. If the fuel concentration at the ignition time and position is outside the
ammable region there will be no explosion. With this approach it is not possible to use the
WIND condition because it enforces a xed velocity, which is not applicable in the explo-
sion.
Run a dispersion simulation, look at the results and decide where and when to have the
ignition, rerun the dispersion simulation with ignition time and position set. Since you
have selected a proper ignition time and position there will be an explosion, but you have
spent a lot of extra time to rerun the complete dispersion simulation. With this approach it
is not possible to use the WIND condition because it enforces a xed velocity, which is not
applicable in the explosion.
Run a dispersion simulation, create simulation dump les at selected time instants, look at
the results and restart the simulation from the dump le with time closest to the desired
time of ignition. This gives you the exibility to select several ignition positions without
having to rerun the dispersion simulations. You can monitor the progress of the dispersion
and decide to create dump les also after the simulation has been started (use the cc-le).
With this approach it is possible to use the WIND condition during the dispersion simula-
tion and to switch it off (change to EULER) for the explosion simulation.
7.2 Simulation Example
7.2.1 Initialization
It is recommended to start out with an empty directory for storing the les. Please note that if you
have followed the previous example in the beginning of this manual, you can skip this section.
Linux:
Make a distinct directory (DIRECTORY_NAME) in which you perform the exercise:
> mkdir DIRECTORY_NAME
Move into this directory:
> cd DIRECTORY_NAME
Copy geometry les (notice the space before the ".").
> cp /usr/local/GexCon/FLACS_v9.0/examples/ex3/
*
00001
*
.
Start up the FLACS runmanager (this assumes that you have set up an alias run9 that points to :
7.2 Simulation Example 207
> run9 runmanager
Windows:
1. Make a distinct directory in which you perform the exercise: Open the le browser ("My
Documents") and choose File New Folder.
2. Copy les from C:\Program Files\GexCon\FLACS_v9.0\examples\ex3\00001
(00001 means all les containing the text "00001").
3. Start the FLACS runmanager by clicking the desktop icon, or go to Start Menu AllPro-
grams GexCon FLACS_v9.0 FLACS Runmanager.
7.2.2 Wind and Dispersion Simulations
In the run manager, use Add Directory to nd the folder that contains the geometry les. Use
Run Manager Tools CASD (or click the FLACS pre-processor icon). Open the le 200001.caj
(and ignore any error messages that appear). The geometry is a representation of a full-scale
process module. The dimensions are 28 m 12 m 8 m.
7.2.2.1 Scenario
Monitor Points and Output Variables Dene a regular pattern of 16 monitor points inside
module (X=3, 9, 15, 21, Y=2, 6 and Z=2, 6).
Measure FMOLE and UVW at monitor points (use SINGLE_FIELD_SCALAR_TIME_OUTPUT).
Remember to use your mouse to select all 16 monitor points.
Measure FMOLE, ER and VVEC for 3D-output (use SINGLE_FIELD_3D_OUTPUT). Remember
to hold the CTRL key while selecting multiple variables for output
Figure 7.1: Specication of Monitor Points
Simulation and Output Control Choose NPLOT=0 and DTPLOT = 2.5, CFLC = 100 (increased
due to grid renement, see below. For more details, see Section X.X.X in the manual) CFLV = 2.
Choose total simulation time TMAX=75.
Boundary Conditions Dene Wind inow from XLO and YLO, 2m/s diagonally [use Wind
Direction (1,1,0)], use Wind Build-up time = 0, NOZZLE at other boundaries.
Figure 7.2: Specication of Wind boundary condition
Initial Conditions Choose initial turbulence low for stability (CHARACTERISTIC_-
VELOCITY=0.1, RELATIVE_TURBULENCE_INTENSITY=0.1, TURBULENCE_LENGTH_-
SCALE=0.01). Use Reference Height: 10, Surface Roughness: 0.01 (logarithmic wind prole),
and Pasquill class F. Leave the other parameters unchanged.
Figure 7.3: Specication of Initial Conditions
Gas composition Natural gas (91% Methane, 7% Ethane and 2% Propane) and set ER0 = 1e30
(pure gas release), Zero cloud size (no gas initially)
Leak In the leak menu, specify leak position as (X=6m, Y=5.05m, Z=2.38m). Leak direction +X
(use OPEN_SIDES). Start gas dispersion at T=10 seconds, with a duration of 40 seconds. The 10
second start time is chosen so that the wind eld can reach steady state.
Click on OUTLET to open a new Window. Use a mass release rate of 4 kg/s through a 0.02m2
leak area. Set relative turbulence intensity = 0.2, turbulence length scale = 10% of leak diameter
= 0.014, Temperature = 20 C, leave direction cosines as (0, 0, 0).
Ignition Specify an ignition time of 100 seconds (a random value after the end of the simulation
so that FLACS does not try and ignite the gas cloud). Leave the ignition position unspecied.
Gas Monitor Region Dene gas monitor region to cover the module (X from 0-28m, Y from
0-12m and Z from 0-8m).
7.2.2.2 Grid
Simulation Volume Choose a simulation domain extending from 32 to 60 (X), -28 to 40 (Y) and
0 to 32 (Z).
Grid Resolution Dene a 1.333 m grid resolution in all directions (i.e. 3 grid cells for every 4
m). Stretch the grid from the module to the boundaries (use CTRL and arrow keys, see Section
REF for more details if needed).
Grid Renement The grid needs to be rened across the leak in order to make sure that it does
not get strongly diluted. In this case, the area of the leak is 0.02 m
2
. According to the requirements
presented in the section Grid guidelines, the area of a control volume near the leak should be
larger than 0.02 m2 but not larger than 0.04 m2. Therefore, in order to optimize the simulation
time, it is decided to rene the grid to 0.2m in Y- and Z-direction around leak. It should be kept
in mind that only the grid cell containing the leak and one neighbour cell on each side need
to be rened and then the grid resolution should be smoothly increased to the prevailing grid
resolution (i.e. 1.333 m). This is shown in the sketch below:
Figure 7.4: Schematic of grid renement
The following steps should then be followed:
Y-direction Since the leak position is at 5.05 m, we need to manually create grid lines 3/20.2
= 0.3 m from the leak position in both directions, i.e. 4.75 and 5.35 m. Change the grid direction
to Y and use Grid Add to add these two grid lines. Select these two grid lines using CTRL
and arrow keys (a message in CASD in the yellow box below the geometry can conrm what
grid lines are selected). After selecting these two grid lines, use Grid Region and enter 3 to
create three grid cells of resolution 0.2 m. The next step is to smooth the grid between the two
signicantly different grid resolutions. For this purpose, we select the grid lines between -1.333 m
and 4.95 m. In this way, one ne grid cell is selected and selecting a grid cell 4-5 away provides the
possibility to achieve a reasonably gradual transition (the goal is that the grid resolution should
not be changing by more than 30-40 % from one grid cell to the next). This can be conrmed by
using Grid Information. Another limitation in this case that we need to keep the grid line at
0 m (and 12 m) intact and therefore, we cannot select a grid cell beyond -1.333 m (and 13.333 m).
After these two grid lines have been selected, use Grid Smooth. Repeat on the other side when
the grid lines between 5.15 m and 13.333 m are selected. Check Grid Information to ensure
that the Max. percentage difference factor is acceptable.
Figure 7.5: Grid in Y-direction
Z-direction By a similar logic as above, we need to manually create grid lines at 2.08 and 2.68
m. Change the grid direction to Y and use Grid Add to add these two grid lines. Select these
two grid lines using CTRL and arrow keys and use Grid Region and enter 3 to create three
grid cells of resolution 0.2 m. For smoothing in the +Z direction, select the grid lines between
2.48 and 9.333 m and use Grid Smooth. In the Z direction, it is not possible to use the smooth
command as there is only one large grid cell. In this case, select the grid from 0 m to 2.28 m and
use Grid Stretch Negative. Check Grid Information to ensure that the Max. percentage
difference factor is acceptable.
The last step before starting the simulation is to make a cc-le. In the FLACS runmanager, click
on the dispersion job, click parameters, and edit cc-le and type the following (use capital letters
and extra line shift at the end of cc-le):
NDUMP 1
TDUMP 40
NDUMP 2
TDUMP 55
This gives 2 dumps at 40 and 55 seconds which can be used to restart the calculations.
Figure 7.6: Dening the cc le for dispersion simulation
7.2.2.3 Simulation
Start the simulation by clicking on the job and clicking simulate in the runmanager.
7.2.2.4 Results
The most important result from this simulation is the gas cloud distribution. This can be studied
in Flowvis based on the material presented in sections Flowvis examples and Introductory exam-
ple. 2D pictures of the gas cloud at times 40 s and 55 s are shown below. The concentrations are
plotted in the ammable range for natural gas (i.e. between 5 % and 15 % natural gas).
Figure 7.7: Gas cloud distribution in the ammable range at time 40 seconds
Figure 7.8: Gas cloud distribution in the ammable range at time 55 seconds
7.2.3 Explosion Simulations
In order to carry out explosion simulations (ignition of a realistic gas cloud produced as a result
of the release simulated above), copy les to new job (Use CASD to save as a new job number).
We have dumped the dispersion results at two discrete times: 40 s and 55 s. The gas cloud at 40
s is used to start an explosion job.
7.2.3.1 Scenario and Grid
Use owvis to nd a suitable ignition position e.g. make contour (2D cut plane) of FMOLE or
ER in plane of leak (2.4 m) to nd regions where concentration is close to stoichiometric (that is
expected to lead to Worst-case explosion pressure). In this case, based on the gure above, an
ignition position of (23, 4.5, 2.4) was chosen. Certain changes need to be made to the scenario
and the grid to make the job suitable for explosion simulations. The following steps should be
followed:
1. Change grid to uniform explosion grid e.g. 0.5 m (stretch from module to boundaries)
2. Simulation and Output Control: Change TMAX to -1, NPLOT = 50, CFLC = 5, CFLV = 0.5.
Change DTPLOT back to -1.
3. Change output variables (both scalar time and 3D): Use P and PROD (it is possible to re-
move the other variables)
4. Change ignition position (use step 2 above) and ignition time (40.05)
5. Save and calculate porosities
After that, the user should click on the job in the runmanager (if it is not visible, click on rescan
directory). Click Parameters and dene a cc-le. The cc-le should contain only one line (the
user should remember to include an extra line in the end):
NLOAD = 1
The last step is to generate a new rd le for restarting calculation based on the explosion job. This
transfers all required information from the dispersion grid to the explosion grid.
Linux: In terminal window, type:
> run9 rdfile rdXXXXXX.n001 rdZZZZZZ.n001
where XXXXXX is the dispersion job and ZZZZZZ is the explosion job. The user should make
sure that you are in the correct directory.
Windows: In command window, type:
> rdfile rdXXXXXX.n001 rdZZZZZZ.n001
where XXXXXX is the dispersion job and ZZZZZZ is the explosion job. The user should make
sure that you are in the correct directory.
7.2.4 Simulation
Start the simulation by clicking on the job and clicking simulate in the runmanager.
7.2.5 Results
It is possible to generate results using Flowvis. The scalar time plot of pressure at all monitor
points is shown below. It can be seen that the maximum pressure of 1.3 barg occurs at t = 40.34
second at monitor point 1.
7.3 Equivalent Stoichiometric Gas Cloud 215
Figure 7.9: Overpressures at monitor points as a result of the explosion of the dispersed gas cloud
at 40 seconds
7.3 Equivalent Stoichiometric Gas Cloud
7.3.1 General Principles
For a dispersion study following a leak in a process area, the main evaluation parameter is the
size of the ammable gas cloud. In order to evaluate the hazard of a given gas cloud, we have
developed methods used for natural gas in the oil and gas industry that aim at estimating an
equivalent stoichiometric gas cloud with comparable explosion consequences. These methods
have been developed in order to reduce the number of simulations that need to be carried out
in order to do a thorough risk evaluation. The size of the equivalent stoichiometric cloud at the
time of ignition is calculated as the amount of gas in the ammable range, weighted by the con-
centration dependency of the ame speed and expansion. For a scenario of high connement,
or a scenario where very high ame speeds (faster than speed of sound in cold air) are expected
(either large clouds or very congested situations), only expansion based weighting is used (de-
noted as Q8). For most situations lower ame speeds are expected and the conservatism can be
reduced. Here a weighting of reactivity and expansion is used (denoted as Q9). The Q8 and Q9
equivalent volumes can be dened as:
Q8 =
V E/E
stoich
(7.1)
and
Q9 =
V BV E/(BV E)
stoich
(7.2)
Here, V is the ammable volume, BV is the laminar burning velocity (corrected for ame wrin-
kling/Lewis number effects), E is volume expansion caused by burning at constant pressure in
air, and the summation is over all control volumes. Thus, Q9 cloud is a scaling of the non-
homogeneous gas cloud to a smaller stoichiometric gas cloud that is expected to give similar
explosion loads as the original cloud (provided conservative shape and position of cloud, and
conservative ignition point). This concept is useful for QRA studies with many simulations, and
has been found to work reasonably well for safety studies involving natural gas releases (NOR-
SOK, 2001). Figure 1 presents the pressure as a function of equivalent stoichiometric volume Q9
for several full-scale experiments carried out by HSE as a part of the Phase 3B project. It can be
seen that the pressure values correlate with the equivalent gas cloud sizes to a reasonable degree.
Figure 7.10: Overpressures predicted by FLACS as a function of equivalent cloud size compared
with HSE large scale experiments (Phase 3B project).
This concept has also been applied to hydrogen systems for the FZK workshop experiments and
has been found to give reasonably good predictions. Figure 2 shows the explosion pressure plot-
ted as a function of the size of the equivalent gas cloud, along with reference values calculated
by using a stoichiometric gas cloud of the same volume. This gure shows that this approach
gives a very good indication of expected overpressures (without initial turbulence, but with con-
servative cloud position and shape, and ignition location). Comparison with experimental data
revealed that in general, the simulations were able to predict very representative values of over-
pressures for both geometries and all release scenarios. More details can be found in Middha, et
al., 2007.
7.3 Equivalent Stoichiometric Gas Cloud 217
Figure 7.11: Overpressures predicted by FLACS in the hood (red) and plate (blue) congurations
for workshop experiments carried out by FZK as function of estimated Q9 equivalent cloud
size for ignited leaks. Reference calculations with homogenous stoichiometric gas clouds are
included.
7.3.2 Shape of Equivalent Gas Cloud
As a practical guideline, the user is recommended to choose the shape of the cloud that will give
maximum travel distance from ignition to end of cloud for smaller clouds. For larger clouds,
end ignition scenarios with longer ame travel should also be investigated. The cloud should
be made as a cubical rectangular box with assumed planes of symmetry towards connement.
The aspect ratio for a free cloud should be 1:1:1, for a cloud towards the ceiling 2:2:1, towards
ceiling with one sidewall 2:1:1, etc. For a free jet in a less conned situation, the jet momentum
will usually dominate the mixing of the jet until the hydrogen concentration has become lean
unless the wind is very strong. The cloud should be assumed to be located a small distance
downwind of the jet, if possible conservatively towards obstructions/walls. Alternatively, for
a low momentum release in a conned situation with strong stratication, the cloud should be
assumed to cover full ceiling area (or area between beams). For highly buoyant gases such as
hydrogen, it can always be assumed that a conned and semi-conned cloud is located near the
ceiling or below any other horizontal connement (a possible deviation from this may be large
liquid hydrogen releases in hot and dry surroundings).
7.3.3 Ignition of Equivalent Gas Cloud
For smaller clouds (ame travel less than 1mto open boundary), ignition with maximumdistance
to edges of gas cloud is normally a conservative choice (for a cloud located in a corner, this means
corner ignition, not central ignition). This could be used to represent all scenarios. Alternatively
a distribution of ignition points could be applied. For larger clouds, a homogenous distribution
of ignition positions should be applied. It should be kept in mind that the gas clouds with the
possibility of the longest ame travel are often the most dangerous ones.
For a quantitative explosion risk assessment the explosion simulations should be performed with
various idealized clouds of variable size and typically using stoichiometric concentration. For the
purpose of QRA, the distribution of ignition locations should be chosen to represent reality. If
there are one or more highly likely ignition locations that dominate the ignition frequencies, these
may be used. Otherwise, it should be assumed that a constant ignition source might lead to end
ignition (where concentration reaches LFL), whereas intermittent ignition sources will be more
arbitrarily distributed (with higher likelihood centrally in the cloud where concentrations are
above LFL). For stratied clouds, end ignition will mean ignition in the lower end of the cloud.
Two ignition probabilities should generally be established (probability for spontaneous ignition
should also be considered in case of hydrogen):
PI
const
= ignition probability from constant ignition sources (per m3 that is exposed to
ammable gas for the rst time in the last 1 second)
PI
intermittent
= ignition probability for intermittent ignition sources (per m3 ammable vol-
ume and second)
From the CFD calculations the volume of the gas cloud with concentration between LFL and
UFL will give the volume that may be ignited if exposed to an intermittent ignition source. This
information should thus be related to the intermittent ignition frequencies dened. In the case
of constant ignition source, the Q6 output from FLACS (or similar from another CFD tool) gives
the cloud volume that was exposed to ammable gas concentrations for the rst time last sec-
ond. This information should be combined with the probabilities for ignition by constant ignition
sources. Thus for each time step (or each 1s) for each dispersion calculation, the probability for
ignition from spontaneous, intermittent and constant ignition sources should be established, and
this probability should be added to a gas cloud size class (based on corresponding Q8, Q9 or a
combination).
If gas cloud becomes rich (in a well mixed state, but more gas available than needed to ll room
with stoichiometric concentration) this may be represented either as a stoichiometric cloud or
as a rich cloud with a slightly higher reactivity than observed. This can be done by using the
ammable volume to establish the volume of the cloud and the ammable mass to establish its
concentration.
7.3.4 Further Guidelines
As described above, for scenarios (cloud/congestion combinations) with low or moderate con-
gestion (or cloud sizes much smaller than room dimension, e.g. < 10%) where very high ame
speeds can be achieved, one should use Q8 value instead of Q9. To evaluate this, a number of
cloud sizes can be simulated to identify a critical cloud size, Qcrit, for which ame speeds ex-
ceed e.g. 200 m/s. For clouds with Q8 < Q
crit
, the weighting procedure above can be applied,
for clouds with Q8 > Q
crit
, one should apply the Q8 cloud as representative cloud size. For
vented rooms and other situations with a signicant connement, a weighting between Q8 and
Q9 volumes is suggested.
If the ventilation or a high-momentum leak (jet) creates a signicant turbulence in the region
where ignition is expected, this turbulence should be dened as an initial condition for the CFD
solver. The main parameter that should be dened should be turbulent uctuations (product
of characteristic velocity and turbulence intensity). The turbulent length scale should not be
assigned a too large value (As a guideline, maximum 10% of smallest cloud dimension and max-
imum 50% of grid cell size should be used).
7.4 Dispersion simulation with wind 219
7.4 Dispersion simulation with wind
FLACSDispersion is a CFD-code for predicting the spread of ammable or toxic gases in com-
plex geometries or terrain, including density effects. GexCon released the rst FLACS version
with dispersion capabilities in 1989, and since 2001 it has been possible to purchase dispersion
capabilities. All functionality of FLACS-DISPERSION is also found in the full FLACS version.
7.5 Hydrogen explosions and DDT
FLACS-Hydrogen is a CFD-code for predicting dispersion and explosion scenarios with hydro-
gen gas in complex geometries. Hydrogen has been available as a gas in FLACS since 1989,
and could be purchased as a dedicated tool since 2001. The validity was strongly improved with
FLACS 8.1 in 2005. All functionality of FLACS-Hydrogen can be found in the full FLACS version.
Important changes are carried out for hydrogen combustion properties:
Modied laminar burning velocity curve, LFL has been reduced to 4%
Adjustment to include effect of Lewis number (more than normal wrinkling on lean con-
centrations, less than normal wrinkling on rich concentrations)
Stronger enhancement due to ame wrinkling (3.5 times increase in burning velocity with
distance from ignition point at stoichiometric concentration).
Normal time step guidelines should be applied, i.e. CLFC=5 and CFLV=0.5
These changes are resulting from the recent active work on validation for hydrogen safety, both
within the activity supported by Norsk Hydro, Statoil and IHI, but also through work within
HYSAFE network. New experiments have become available. Much of the work behind is of
condential nature, and it is not clear that we can share the validation data with all FLACS users.
However an extensive validation matrix has been simulated and we are quite satised with the
results simulating:
20m diameter hemispherical hydrogen cloud (large-scale laminar ames)
1.4m small-scale channel (different bafes, gas concentration, ignition location)
3D corner small-scale pipe arrays (different congestion, gas concentration, ignition)
30m tunnel FLAME facility (different venting, bafes, gas concentration)
And more ...
Based on the results we consider the FLACS V8.1 performance simulating hydrogen comparable
to what is generally seen when simulating other gases.
The laminar diffusivity for hydrogen in FLACS may historically be somewhat too high. This
will only be important in situations with absolutely no turbulence, i.e. mainly in closed vessels
with no ventilation and weak temperature gradients. When discovering the possible error, the
constant was not changed in FLACS. Instead we included a possibility for the user to dene the
constant for dynamic laminar viscosity manually through setup-le or KEYS-string in scenario-
le.
The default constant for dynamic laminar viscosity in FLACS is 2.0e-5 (see below), a more correct
value is probably 0.6e-5.
VERSION 1.1
$SETUP
KEYS="AMUL=Y:2.0e-5"
$END
7.5.1 Use normal time step for hydrogen explosion simulations
One important observation done is that you should not change CFLC-number (time step crite-
ria) when simulating deagrations (explosions) involving hydrogen. We previously (last year)
recommended using a shorter time step for hydrogen simulations (CFLC=0.5), and we also have
been recommending this for far eld blast calculations (CFLC=0.5). We have for a long time been
aware that shorter time-step in some situations will give different explosion development. Now
we have seen that results in hydrogen simulations become better keeping the normal time step.
For pressure wave propagation a distance away from the calculation there is a possibility to set
an option in the scenario-le that keeps a short time step when blast wave propagates away from
the explosion (STEP=KEEP_LOW). For multi-block simulations with BLAST solver you may still
need to use a short time step for stability reasons. Note that "STRICT:" is assumed as default
time stepping mode in the new version of Flacs. You may enter "NOSTRICT:" as your choice,
but this is normally not recommended.
For dispersion calculations we still recommend to use long time steps (CFLC=20, CFLV=2). With
local grid renement (for instance near leak), we also recommend to increase CFLC proportional
to the grid renement factor: CFLC=20 x CVnormal / CVne. If stability problems occur (this
may well happen), it is recommended to reduce the CFLC and CFLV numbers by a factor in the
range of 2 - 4.
For explosive and vessel burst calculations it is recommended to use short time steps. The models
for handling explosives are not included in the standard version. It is recommended to set CLFC
and CFLV=0.025 when using BLAST solver, and slightly larger values could be used with FLACS
solver. Vessel burst calculations can be done as a normal FLACS calculation, then CLFC and
CFLV around 0.1 is recommended.
Chapter 8
Technical Reference
222 Technical Reference
This chapter contains an overview of the theoretical foundation for the FLACS software, includ-
ing physical and chemical models.
8.1 Denitions and gas thermodynamics
This section presents denitions of gas and mixture parameters and relations for ideal gases.
8.1.1 Denitions
Number of moles of species:
n
i
=
m
i
M
i
(8.1)
Mole fractions:
X
i

n
i
N
i=1
n
i
(8.2)
Mass fractions:
Y
i

m
i
N
i=1
m
i
(8.3)
Fuel-oxidant ratio:
(F/O)
m
fuel
m
ox
(8.4)
Equivalence ratio:

(F/O)
(F/O)
stoich
(8.5)
Mixtures of fuel and oxidant are characterised by the equivalence ratio as follows:
> 1 : Fuel rich mixture
= 1 : Stoichiometric mixture
< 1 : Fuel lean mixture
The mixture fraction, , describes the degree of mixing between two well-dened states (0 and 1)
and is dened as follows:
=

0
0
(8.6)
where is a general variable.
The progress variable tells how much of the potential fuel that has burnt and is dened as
follows:
=
Y
fuel
Y
0
fuel
+
_
Y
1
fuel
Y
0
fuel
_ (8.7)
8.1.2 Mixing of several gases
Mole fraction:
X
i

Y
i
/M
i
N
i=1
Y
i
/M
i
(8.8)
Mass fraction:
Y
i

X
i
M
i
N
i=1
X
i
M
i
(8.9)
8.2 Stoichiometric reaction 223
8.1.3 Ideal gas relations
Ideal gas law for a mixture:
p = RT (8.10)
Ideal gas law for single specie:
p
i
=
i
R
i
T (8.11)
Daltons law for a perfect gas:
p =
N
i=1
p
i
=
R
u
T
V
N
i=1
n
i
(8.12)
Isentropic ratio:
=
c
p
c
v
(8.13)
Speed of sound:
c
_
RT =
_
(8.14)
Mach number:
Ma
u
c
(8.15)
Pressure-density-temperature:
_
p
p
0
_
=
_

0
_
=
_
Y
Y
0
_
/(1)
=
_
1 +
1
2
Ma
2
_
/(1)
(8.16)
8.2 Stoichiometric reaction
Combustion is oxidation of a fuel accompanied by the production of heat and light. In most
burning processes, air is oxidant. A simple main reaction can be written as:
C
nc
H
nh
O
no
+ aO
2
ncCO
2
+ bH
2
O + Q (8.17)
This reaction is stoichiometric because there is neither fuel nor oxidant left after the reaction is
completed. The stoichiometric amount of oxidant on mole basis can be calculated by:
a = nc +
nh
4

no
2
(8.18)
The combustion products produced in the reaction are water vapour (H
2
O) and carbon dioxide
(CO
2
).
Some relations for mixing fuel with air are listed below. The mole fraction of O
2
in air is set to
20.95%, which corresponds to a mass fraction of 23.2%. This is the normal air composition in
FLACS, see Initial conditions .
Stoichiometric oxidant-fuel ratio on mass basis:
r
ox
= a
M
O
2
M
fuel
(8.19)
Stoichiometric air-fuel ratio on mass basis:
r
air
=
_
1 +
Y
N
2
Y
O
2
_
r
ox
= 4.31r
ox
(8.20)
Mass fraction of fuel given an equivalence ratio :
Y
fuel
=

+ r
air
(8.21)
8.3 Governing equations for uid ow
This section describes the mathematical model for compressible uid ow used in FLACS.
Conservation of mass:
t
(
v
) +

x
j
_
j
u
j
_
=
m
V
(8.22)
Momentum equation:
t
(
v
u
i
) +

x
j
_
j
u
i
u
j
_
=
v
p
x
i
+

x
j
_
ij
_
+ F
o,i
+
v
F
w,i
+
v
(
0
) g
i
, (8.23)
where F
w,i
is ow resistance due to walls and F
o,i
is ow resistance due to sub-grid obstructions:
F
o,i
=
x
i
u
i
|u
i
| (8.24)
ij
is the Stress tensor.
Transport equation for enthalpy:
t
(
v
h) +

x
j
_
j
u
j
h
_
=

x
j
_
eff
h
h
x
j
_
+
v
Dp
Dt
+

Q
V
(8.25)
Transport equation for fuel mass fraction:
t
(
v
Y
fuel
) +

x
j
_
j
u
j
Y
fuel
_
=

x
j
_
eff
fuel
Y
fuel
x
j
_
+ R
fuel
(8.26)
where R
fuel
is the fuel reaction rate, which will be handled in combustion modelling.
Transport equation for the mixture fraction:
t
(
v
) +

x
j
_
j
u
j
_
=

x
j
_
eff
x
j
_
(8.27)
Transport equation for turbulent kinetic energy:
t
(
v
k) +

x
j
_
j
u
j
k
_
=

x
j
_
eff
k
k
x
j
_
+
v
P
k

v
(8.28)
Transport equation for the dissipation rate of turbulent kinetic energy:
t
(
v
) +

x
j
_
j
u
j
_
=

x
j
_
eff
x
j
_
+
v
P
C
2
2
k
(8.29)
8.3 Governing equations for uid ow 225
The stress tensor in the above equations is given by:
ij
=
eff
_
u
i
x
j
+
u
j
x
i
_
2
3
ij
_
k +
eff
u
k
x
k
_
(8.30)
The effective viscosity is dened as follows:
eff
= + C
k
2
, (8.31)
where the second term is known as the turbulent viscosity or eddy viscosity.
Flow shear stresses, G
s
, wall shear stresses, G
w
, buoyancy, G
b
, and sub-grid objects, G
o
contribute
to the production of turbulent kinetic energy P
k
:
P
k
= G
s
+ G
w
+ G
b
+ G
o
(8.32)
The production rate of turbulent kinetic energy due to shear stresses appears from the derivation
of the transport equation and reads:
G
s
=
ij
u
i
x
j
(8.33)
Production due to buoyant forces modelled by a simple gradient model:
G
b
=
1
eff
b
g
i
x
i
(8.34)
The turbulence generation due to sub-grid obstructions is modelled by:
G
o
= C
o
v
|u| u
2
i
f
i
(8.35)
where C
o
is a model constant and f
i
is a parameter depending on sub-grid objects.
The production of dissipation, P
, is modelled as follows:
P
= C
1
k
P
k
_
1 + C
3
R
f
_
(8.36)
where the model for the buoyancy term follows Rodi (1980):
R
f
=
G
b
P
k
|u g|
|u| |g|
(8.37)
In FLACS, the buoyancy terms G
b
and R
f
are zero when products are present.
8.3.1 Turbulence model
Turbulence is modelled by a two-equation model, the k model. It is an eddy viscosity model
that solves two additional transport equations; one for turbulent kinetic energy and one for dis-
sipation of turbulent kinetic energy. Following Boussinesq eddy viscosity assumption, an eddy
viscosity models the Reynolds stress tensor as follows:
i
u
j
=
eff
_
u
i
x
j
+
u
j
x
i
_
2
3
k
ij
(8.38)
A few constants are included in the equations mentioned above. In FLACS, the following set of
constants is used, which agree with the model of Launder and Spalding (1974):
C
C
1
C
2
C
3
0.09 1.44 1.92 0.8
(8.39)
In addition, there is a set of turbulent Prandtl-Schmidt numbers,
. Prandtl-Schmidt numbers
tells about the diffusion of the variable in question compared to the dynamic viscosity. The
turbulent Prandtl-Schmidt numbers are:
h

fuel

b
0.7 0.7 0.7 1.0 1.3 0.9
(8.40)
8.4 Wall functions
Boundary layers are regions in the ow eld close to walls and obstructions where there are
steep gradients and peak values for turbulent kinetic energy and its dissipation rate. Very close
to the wall surface dominates viscous forces over inertial effects. The motivation for using wall-
functions is to model the inuence of the wall at a point a certain distance from the wall.
A dimensionless wall distance is dened by:
y
+
=
C
1/4
k
1/2
y
, (8.41)
where y is the distance from the wall point to the wall. Wall point is dened as the point closest to
the wall where transport equations are solved. The shear stresses caused by the wall are modelled
by:
w,i
=
_
u
i
y
if y
+
< E
+
u
i
C
1/4
k
1/2
E
+
+ln
_
y
+
E
+
_ if y
+
E
+
(8.42)
Then wall friction term in the momentum equation becomes:
F
w,i
=
v
w,i
A
w
V
(8.43)
Production of turbulent kinetic energy in the wall point is modelled by:
G
w
=
_
_
_
0 if y
+
< E
+
2
2
w
|u|C
1/4
k
1/2
E
+
+ln
_
y
+
E
+
_ if y
+
E
+ (8.44)
Dissipation of turbulent kinetic energy is given a value at the wall point by solving the following
integral:
w
=
1
y
cv
_
y
cv
0
dy (8.45)
The integral is estimated by:
w
=
_
_
1
y
cv
_
2k
y
+ (y
cv
y)
_
if y
+
< E
+
1
y
cv
_
2
_
ky
+
yE
+

_
y
+
E
+
1
_ _
k
+1
k
(y
+1
y)
__
+
C
3/4
k
3/2
ln
_
y
+
E
+
_
+ (y
cv
y)
_
if y
+
E
+
(8.46)
8.5 Wind boundary 227
k
+1
denotes the value of k and y
+1
denotes the wall distance in the point beyond the wall point
in the opposite direction of the wall. denotes the mean value of between the cell point and the
control volume boundary in the opposite direction of the wall.
8.5 Wind boundary
Wind boundaries reproduces the properties of the atmospheric boundary layer close to Earths
surface. Monin and Obukhov (1954) developed a theory to explain buoyancy effects on the at-
mospheric boundary layer and dened a charcteristic lenght scale:
L =
a
c
p
T
a
u
3
gH
s
(8.47)
where H
s
is the sensible heat ux from the surface and u
is the friction velocity. The Monin-

Obukov length is a measure for the stability of the atmospheric boundary layer. Table Monin-
Obukhov lengths and stability shows an interprentation of the Monin-Obukhov lenths with re-
spect to the atmospheric stability. (Bosch and Veterings, 1996)
Monin-Obukhov length Stability
Small negative, 100m < L0 Very unstable
Large negative, 10
5
< L < 100 Unstable /tr>
Very large, |L| > 10
5
Neutral
Large positive, 10 < L < 10
5
Stable
Small positive, 0 < L < 10 Very stable
Table 8.1: Monin-Obukhov lengths and stability.
In FLACS, the Monin-Obukhov length is estimated by using Pasquill classes, which is a method
of categorizing the amount of atmospheric turbulence present. The user have to give average
wind velocity, U
0
, a reference height, z
re f
, an atmospheric roughness length, z
0
and the Pasquill
class under Initial conditions. The velocity prole is logarithmic:
U(z) =
_
u
ln
_
z+z
0
z
0
_
if z
0
> 0
U
0
if z
0
= 0
(8.48)
where, u
, the friction velocity. u
is generally given by:

u
=
_
_
U
0
ln
_
z
re f
z
0
_
+
5
L
(z
re f
z
0)
if L > 0
U
0
ln
_
z
re f
z
0
_
2
if L < 0
(8.49)
Attention:
At present two friction velocities are used in FLACS. The expression in the friction velcoc-
ity equation above is used for k and . For velocity is the following friction velocity used:
u
= U
0
/ln(z
re f
/z
0
). This is due to the development history of FLACS, where a logritmic
velocity prole was implemented before the proles for the turbulence parameters.
Table Wind prole parameters below gives an overview of parameters that are used to calculate
values for velocity, k, and at wind boundaries.
1
and
2
in the equation for the friction velocity
are given by:
1
= 2 ln
_
1 +
3
2
_
+
1
2
ln
_
1 +
2
3
_
2 arctan (
3
) +

2
(8.50)
2
= 2 ln
_
1 +
4
2
_
+
1
2
ln
_
1 +
2
4
_
2 arctan (
4
) +

2
(8.51)
3
=
_
1
16z
re f
L
_
1/4
(8.52)
4
=
_
1
16z
0
L
_
1/4
(8.53)
Pasquill class Stability Boundary layer
height, h
L
s
z
s
A Unstable 1500 m 33.162 m 1117 m
B Unstable 1500 m 33.258 m 11.46 m
C Slightly
unstable
1000 m 51.787 m 1.324 m
D Neutral 0.3u
L
f
1.0 m 0 m
E Slightly stable 0.4
_
u
L
f
-48.33 m 0 1.262 m
F Stable 0.4
_
u
L
f
-31.323 m 0 19.36 m
Table 8.2: Wind prole parameters. Values for L
s
and z
s
are taken
from Bosch and Wetering (1996) and orginates from the graphs of
Golder (1972). Values for h are taken from Han et al. (2000).
From the values in the wind prole parameters table, the Monin-Obukhov length can be calcu-
lated as follows (Golder, 1972):
1
L
=
1
L
s
log
z
0
z
s
(8.54)
The set of expressions for the wind boundary proles for turbulent kinetic energy, k, and its
dissipation, proposed by Han et al. (2000) are implemented in FLACS. Different experssions
were proposed for for unstable and stable/neutral boundary layers. Unstable boundary layers
are caused by heat from the ground that increases the temperature of the air close to the surface.
Hence, the density close to the surface is less than the density of the air above, which gives an
unstable situation. The mean surface heat ux, q
s
, is therefore an important parameter when the
turbulence proles at the inlet is estimated for unstable boundary layers. The inlet proles for
ustable boundary layers (A, B, and C) are:
k(z) =
_
0.36w
2
+ 0.85u
2
_
1 3
z
L
_
2/3
if z 0.1h
_
0.36 + 0.9
_
z
h
_
2/3
_
1 0.8
z
h
_
2
_
w
2
if z > 0.1h
(8.55)
and
(z) =
_
_
_
u
3
z
_
1 + 0.5
z
L
2/3
_
3/2
if z 0.1h
w
3
h
_
0.8 0.3
z
h
_
if z > 0.1h
(8.56)
where the heat velocity , w
, is given by:
w
=
_
g q
s
h
T
0
c
p
_
1/3
(8.57)
8.6 Combustion modelling 229
where the air properties, , and c
p
are obtained at ambient temperature T
0
and pressure p
0
.
Proles for neutral and stable boundary layers depend on the friction velocity and the Monin-
Obukhov length as follows:
k(z) =
_
6u
2
if z 0.1h
6u
2
_
1
z
h
_
1.75
if z > 0.1h
(8.58)
and
(z) =
_
u
3
z
_
1.24 + 4.3
z
L
_
if z 0.1h
u
3
z
_
1.24 + 4.3
z
L
_ _
1 0.85
z
h
_
3/2
if z > 0.1h
(8.59)
8.6 Combustion modelling
Ignition of a premixed cloud of fuel and oxidant may escalate to an explosion. Before escalation,
a steady non-turbulent premix of fuel and oxidant will burn with a laminar burning velocity:
S
0
L
= S
0
L
(fuel, ) (8.60)
The laminar burning velocity depends on the fuel and the equivalence ratio . For mixtures with
fuel contents below the Lower Flammability Limit (LFL) or above the Upper Flammability Limit
(UFL), the laminar burning velocity equals zero, i.e. it will not burn. In an explosion, the ame
will accelerate and become turbulent. The turbulent burning velocity is much larger than the
laminar one due to much better mixing of reactants and products. FLACS uses correlations for
both laminar and turbulent burning velocities that origin from experimental work.
In industrial applications, the reaction zone in a premixed ame is thin compared to practical grid
resolutions. It is therefore necessary to model the ame. In FLACS, the ame zone is thickened by
increasing the diffusion with a factor and reducing the reaction rate with a factor 1/. Hence,
the ame model in FLACS is called the -model.
8.6.1 The FLACS ame model
The diffusion coefcient D for fuel comes from the transport equation for fuel:
D =

eff
fuel
(8.61)
Furthermore, it is possible to dene a dimensionless reaction rate W. In the -model, D and W
are adjusted as follows:
W
=
W
= W
l
LT
g
(8.62)
D
= D = D
g
l
LT
(8.63)
From an eigenvalue analysis of the burning velocity (Arntzen, 1998), the following relation be-
tween the diffusion coefcient D and a dimensionless reaction rate W is obtained for
q
= 0.05:
WD = 1.37S
2
u
= W
(8.64)
q
is the quenching limit of the progress variable . D
and W
depend on the grid size and the

burning velocity as follows:
W
= c
1
S
u
g
(8.65)
D
= c
1
S
u
g
D (8.66)
The reaction rate of fuel is modelled by the following expression:
R
fuel
= W
min
_
H
(
q
), , 9 9
_
(8.67)
where
H
is the Heaviside step function.
8.6.2 Burning velocity model
The burning velocity is laminar when the ame is smooth and governed by molecular diffu-
sion. This is typically the case in the very early phase of an explosion (e.g. spark ignition of
a combustible cloud under quiescent conditions). A short period of time after the ignition, the
ame becomes quasi-laminar when instabilities lead to wrinkling of the ame. After a transition
period, the ame reaches the turbulent regime.
The laminar burning-velocity depends on the type of fuel, fuel-air mixture and pressure. For each
fuel, the laminar burning velocities at different equivalence ratios are tabulated. The laminar
burning-velocity of a mixture of fuels is estimated by taking the volume-weighted average. The
pressure dependency on the laminar burning velocity is described as:
S
L
= S
0
L
_
P
P
0
_
P
, (8.68)
where
P
is a fuel dependent parameter. In the quasi-laminar regime, the turbulent burning
velocity is given by:
S
LQ
= S
L
+ 8S
0.284
L
u
0.912
l
0.196
I
(8.69)
The correlation for the turbulent burning velocity S
T
is a simplication of a general expression
presented by Bray (2000) and reads:
S
T
= 15S
0.784
L
u
0.412
l
0.196
I
(8.70)
FLACS selects burning velocity as follows:
S
u
= max
_
S
L
min
_
S
QL
, S
T
__
(8.71)
8.7 Modelling of jet sources
To model the conditions of a pressurised reservoir which is gradually emptied through a nozzle,
a simple procedure may be utilised which calculates the sonic ow rate through the nozzle. By
assuming isenthalpic expansion, it is possible to calculate expansion of a sonic ow analytically.
Further air entrainment can be accounted for using simplifying assumptions. Finally, the calcu-
lated mass ow mixes in a well-stirred reactor with a constant volume and a constant ventilation
rate. Additional values for the turbulence quantities must be calculated in order to use the data
in FLACS.
Calculation of 5 stage analytic dispersion:
8.7 Modelling of jet sources 231
1. Reservoir (stagnation)
2. Nozzle (sonic)
3. Jet (outlet)
4. Air entrainment
5. Well-stirred reactor
Let (u, h, T, p, , A) be a vector describing the necessary leakage parameters, then

a
refers to the ambient condition,

2
refers to the outlet condition,

1
refers to the nozzle condition and

0
refers to the stagnation condition.
Initial reservoir conditions:
Pressure: p
0
is specied.
Temperature: T
0
is specied
Volume: V
0
is specied
Density:
0
=
p
0
RT
0
Total mass: m
0
=
0
V
0
Heat exchange coefcient: h
wall
is =specied.
Reservoir conditions at time t + dt:
Total mass: m
0,t+dt
= m
0
m
1
dt
Temperature: T
0,t+dt
= T
0
(

Q
0
+ m
1
h
1
)dt
Wall heat ux:

Q = h
wall
(T
0
T
wall
)
Density:
0
= m
0
/V
0
Pressure: p
0
=
0
RT
0
Nozzle (sonic) conditions:
Effective nozzle area: A
1
is specied.
Temperature: T
1
= T
0
(2/( + 1))
Pressure: p
1
= p
0
(T
1
/T
0
)
/(1)
Density:
1
=
p
1
RT
1
Sound speed: c
1
=

RT
1
Velocity: u
1
= c
1
Enthalpy: h
1
= c
p
T
1
Mass ow: m
1
=
1
u
1
A
1
Jet (outlet) conditions:
Velocity: u
2
= u
1
p
1
p
2
1
u
1
Enthalpy: h
2
= h
1
+
1
2
(u
2
1
u
2
2
)
Temperature: T
2
= T
1
+
1
2
u
2
1
u
2
2
c
p
Pressure: p
2
= p
a
Density:
2
=
p
2
RT
2
Effective outlet area: A
2
= A
1
1
u
1
2
u
2
Mass ow: m
1
=
1
u
1
A
1
Air entrainment condition:
Pressure: p
3
= p
a
Temperature: T
3
= T
a
Density:
3
=
p
3
RT
3
Velocity: u
3
= u
2
f
2
f
3
Effective area: A
3
= A
2

2
u
2
3
u
3
Mass ow: m
3
=
3
u
3
A
3
Well-stirred reactor condition:
Volume: V
4
is specied.
Ventilation rate:

V
air
is specied.
Incremented fuel mass: m
fuel
= m
fuel
+ m
fuel
d f
Incremented air mass: m
air
= m
air
+ m
air
d f
Incremented mixture mass: m
mix
= m
fuel
+ m
air
Density:
4
=
m
mix
V
4
+(V
fuel
+V
air
)dt
Total mass inside volume: m
4
= V
4
4
Fuel mass: m
fuel,t+dt
= m
4
m
fuel
m
total
Air mass: m
air,t+dt
= m
4
m
air
m
total
8.8 Numerical Schemes
FLACS is a computational uid dynamics (CFD) code solving the compressible conservation
equations on a 3D Cartesian grid using a nite volume method. The conservation equations
for mass, momentum, enthalpy, and mass fraction of species, closed by the ideal gas law, are
included. The conservation equations can be represented in general as:
t
() +

x
j
(u
i
)

x
j
_
x
j
()
_
= S
(8.72)
The in-house development started around 1980, primarily aimed at simulating the dispersion
of ammable gas in process areas, and subsequent explosions of gas-air mixtures. Hjertager
(1985, 1986) describes the basic equations used in the FLACS model, and Hjertager, Bjrkhaug &
Fuhre (1988) present the results of explosion experiments to develop and validate FLACS initially.
During the course of more than 25 years of development and evaluation of the FLACS software,
the numerical methods have been steadily modied and revised.
The inherent capability of FLACS has been performing explosion and dispersion calculations
to help in the improvement of oil and gas platform safety with initial focus on the North Sea.
Signicant experimental validation activity has contributed to the wide acceptance of FLACS as
a reliable tool for prediction of natural gas explosions in real process areas offshore and onshore.
The numerical model uses a second order scheme for resolving diffusive uxes and a second-
order scheme (hybrid scheme with weighting between 2
{nd} order upwind and 2
{nd} order
central difference, with delimiters for some equations) to resolve the convective uxes.
The time stepping scheme used in FLACS is a rst order backward Euler scheme. Second order
schemes in time have been implemented, but are generally not used due to short time steps.
Based on extensive validation, guidelines for time stepping have been established in order to
get accurate results. These are based on CFL-numbers based on speed of sound (CFLC) and
ow velocity (CFLV). For explosion calculations CFLC=5 and CFLV=0.5 must be applied (which
means that the pressure can propagate 5 cells and the ow 0.5 cells in each time step) to achieve
8.9 Linux Quick Reference 233
good results. For dispersion calculations, the guidelines are less strict as the results do not depend
much on the time steps. Normally it is recommended to increase the time steps by a factor of 4
(CFLC=20 and CFLV=2). When grid is rened near leak, it is also recommended to ignore the
rened region (i.e. multiply CFLC-number with the renement factor).
The SIMPLE pressure correction algorithm is applied (Patankar, 1980), and extended to handle
compressible ows with additional source terms for the compression work in the enthalpy equa-
tion. Iterations are repeated until a mass residual of less than 10
4
is obtained.
8.9 Linux Quick Reference
This section summarises some relevant information for users that run FLACS under the Linux
operating system. Further information concerning Linux may be found at e.g. www.linux.org
8.9.1 Distributions
FLACS work on most recent Linux distrobutions. An updated list of distributions, on which
FLACS v9.0 has been tested, is given in Hardware and Software requirements.
8.9.2 Desktop environments
FLACS works independently of the Desktop environment. The most popular environments are:
KDE
Gnome
Most of the developers at GexCon prefer KDE.
8.9.3 Shell
A command shell is command line interface computer program to an operating system (OS). The
most popular shells are:
bash , setup le in home directory: .bashrc
C shell , setup le in home directory: .cshrc
8.9.4 Text editors
To create, read, write, or edit text les, e.g. Flacs input le the user must know how to use a text
editor. Recommended text editors are:
vi / vim
Emacs
gedit
kate
Warning:
With Emacs: Remember to have an empty line at the end of the le. VIM adds an extra line
automatically.
Attention:
Notepad++ is recommended if you like to edit your text le in Windows. Notepad++ does
not change your text le, which can be the case for other editors.
8.9.5 Communication with other computers
ssh (SSH client) is a program for logging into a remote machine and for executing com-
mands on a remote machine.
scp copies les between hosts on a network. It uses ssh for data transfer.
ftp is the user interface to the Internet standard File Transfer Protocol. The program allows
a user to transfer les to and from a remote network site.
8.9.6 Help in Linux
Most commands in Linux have related manual pages. These can be displayed by:
> man <command>
For instance:
> man ls
The help and info commands provide less extensive outputs than man, whereas apropos also
includes the man output for related commands. Most commands do also show help by writing
> <command> --help
For instance
> r1file --help
8.9.7 Useful commands
Recent commands are saved in a history le, located in the users home directory, and it is listed
with the command:
> history
User can dene aliases. The alias commands are usually wanted to be executed every time a
shell is used and are therefore usually added to the shell setup-le (.bashrc or .cshrc). alias can
be used to make new short-cuts or to change output of already existing commands. The alias
syntax differ slightly from shell to shell. See:
> man alias
8.9 Linux Quick Reference 235
The cd command changes the working directory:
> cd DIR
To move to the parent directory:
> cd ..
ls lists contents of directories:
> ls
There are a lot of options to the ls command. See
> man ls
Other frequently used commands:
chgrp : Changes group.
chmod : Changes permissions.
chown : Changes ownership.
cp : Copies les from one place to another, or duplicates one le under a different name.
diff : Compare les line by line.
df : Keeps track of your hard disk space.
du : Lists the le sizes in kilobyte.
exit : Ends the application.
find : Looks for les with particular content.
free : Outputs the amount of free RAM on the system.
grep : Finds words in les.
gunzip : Expands les.
gzip : Compresses les.
kill : Terminates a process.
less : Views le contents.
mkdir : Creates directories.
more : Views le contents.
mv : Moves or renames les.
ps : Lists running processes on the system.
pwd : Prints the path of the working directory.
rm : Deletes les permanently.
rmdir : Deletes empty directories.
tail : Views the last part of a le. It is an useful command for monitoring Flacs running
les.
tar : Assembles les into a package and extract a package.
top : Provides live summary of running processes.
| : Pipe command, used together with other commands.
8.9.8 Permissions
Permissions are best showed by an example with a directory containing only one le (file.txt)
and one directory (DIR). ls -al gives the following results:
drwxr-sr-x 2 idar gexcon 4096 2008-05-17 13:01 DIR/
-rw-r--r-- 1 idar gexcon 102030 2008-05-17 14:01 file.txt
where:
-rw-r-r- indicates the le permissions.
1 indicates that there is one le.
idar indicates that the le belongs to the user idar.
gexcon indicates the group.
102030 is the size of the le in bytes.
2008-05-17 indicates the date the le was created/modied/moved.
14:01 indicates the time the le was created/modied/moved.
file.txt is the le name.
The rst character in le persmission is d if it is a directory and - else. The next nine characters
indicate the permissions, where the rst three are for the user that owns the le, the next three
are for the group, and the last three are for others. There are three possible attributes:
r : Read permission.
w : Write permission.
x : Execute permission.
In the example,
-rw-r--r--
only the user has both read and write permission, i.e. can modify the le. Members of the group
and others can read the le.
Chapter 9
Nomenclature
238 Nomenclature
9.1 Roman letters
A Area m
2
a Moles of O
2
in a stoichiometric reaction -
c Speed of sound ms
1
c
p
Specic heat capacity at constant pressure JK
1
kg
1
c
v
Specic heat capacity at constant volume JK
1
kg
1
C
1
Constant in the k equation; typically C
1
= 1.44 -
C
2
2
= 1.92 -
C
3
3
= 0.8 -
C
D
Drag coefcient -
C
= 0.09 -
d Diameter m
D Diffusion coefcient m
2
s
1
f Sub-grid obstructions turbulence generation factor -
E
+
Constant in wall functions; typically E
+
= 11 -
F Force N
F
D
Drag force N
F
w
Wall friction force N
F/O Fuel-oxidant ratio, see denition -
g, g Gravitational acceleration (scalar, vector) ms
2
h Specic enthalpy Jkg
1
h Heat transfer coefcient WK
1
m
2
h Height of the atmospheric mixing layer m
I
p
Pressure impulse Pa s
I
T
Relative turbulence intensity -
k Turbulent kinetic energy m
2
s
2
L Monin-Obukhov length scale m
l Length m
l
LT
Mixing length in the -model, l
LT
= C
k
3/2
1
m
M, M
k
Molecular weight of a mixture, specie kgmol
1
m Mass kg
m Mass rate kgs
1
N Total number density -
n Number density -
p Absolute pressure Pa
p
0
Ambient pressure Pa
P Gauge pressure, overpressure Pa, bar
Q Heat J
Q Heat rate Js
1
R, R
k
Gas constant of a mixture, specie R = R
u
/M Jkg
1
K
1
R
u
Universal gas constant 8.314
Jmol
1
K
1
R
fuel
Reaction rate for fuel kgm
3
s
1
r Radius m
r
ox
Stoichiometric fuel-oxidant ratio on mass basis -
r
air
Stoichiometric fuel-air ratio on mass basis -
S
L
Laminar burning velocity ms
1
S
QL
Quasi-laminar burning velocity ms
1
S
T
Turbulent burning velocity ms
1
T Absolute temperature K
t Time s
9.3 Subscripts 239
U
0
Reference, characteristic velocity ms
1
u
i
, u Mean velocity (ith component, vector) ms
1
u
Root mean square of velocity ms

1
u
Friction velocity ms
1
V Volume m
3
V Volume rate m
3
s
1
W
Dimensionless reaction rate -

X Mole fraction -
x Length coordinate m
Y Mass fraction -
y Wall distance m
y
+
Dimensionless wall distance in wall functions -
z Distance above the ground m
z
0
Aerodynamical roughness length m
9.2 Greek letters
Volume fraction -
Thermal diffusivity m
2
s
1
Transformation factor in the -model, see The FLACS ame model -
i
Area porosity in the i th direction -
v
Volume porosity -
Isentropic ratio -
p
Pressure exponent for the laminar burning velocity, see correlation. -
Control volume length m
H
Heaviside step function.
H
(a b) = 1 if a b.
H
(a b) = 0 if a < b. -
ij
Kronecker delta function,
ij
= 1 if i = j.
ij
= 0 if i = j. -
Dissipation of turbulent kinetic energy m
2
s
3
g
Surface roughness m
Surface tension Nm
1
Von Karman constant; typically = 0.41. -
Conductivity Wm
1
K
1
Dynamic viscosity Pa s
t
Dynamic turbulent viscosity Pa s
eff
Effective viscosity,
eff
= +
t
Pa s
Kinematic viscosity m
2
s
1
Mixture fraction -
Density kgm
3
Prandtl-Schmidt number, see overview. -
ij
Stress tensor, see equation. Nm
2
Time scale s
e
Integral time scale in turbulent ows s
w
Wall shear stress Nm
2
Equivalence ratio, see denition. -
General variable -
Progress variable, see denition. -
9.3 Subscripts
240 Nomenclature
a Ambient -
cv Control volume -
D Drag -
g Ground -
f Flow -
i Specie index, spatial index -
j Spatial index -
L Laminar -
n Control volume index -
o Sub-grid objects -
p Particle -
stoich Stochiometric -
T Turbulent -
v Volume -
w Wall -
9.4 Dimensionless groups
Bi Biot number, Bi =
hd
p
p
. -
Da Damkhler number, Da =

e
chem
. -
Fr Froude number, Fr =
inertial force
gravity force
=
u
l
2
g
. -
Le Lewis number, Fr =

D
= c
p
D =
Sc
Pr
-
Ma Mach number, Ma =
u
c
-
Nu Nusselt number, Nu =
hl
-
Pr Prandtl number, Pr =

=
cp
-
Re Reynolds number, Re =
inertial force
viscous force
=
lu
-
Sc Schmidt number, Sc =

D
=

D
-
St Stokes number, St =

p
f
-
We Weber number, We =
inertial force
surface tension force
=
u
2
d
p
-
9.5 Abbreviations
AIT Auto Ignition Temperature -
CAD Computer Aided Design -
CASD Computer Aided Scenario Design -
CFD Computational Fluid Dynamics -
CFL Courant-Friedrichs-Levy -
CFU Central Processing Unit -
CMR Christian Michelsen Research -
CP8 Complex Polyhedron -
CV Control Volume -
DDT Deagration-to-Detonation Transition -
DESC Dust Explosion Simulation Code -
DNS Direct Numerical Simulation -
ER Equivalence Ratio -
FLACS FLame ACceleration Simulator -
GTC General Truncated Cone -
HSL Health and Safety Laboratory -
9.6 FLACS variables 241
HVAC Heating, Ventilating, and Air-Conditioning -
LES Large Eddy Simulation -
LFL Lower Flammability Limit -
MIE Minimum Ignition Energy -
PI Ignition Probability -
QRA Quantitative Risk Analyses -
RAM Random Access Memory -
UFL Upper Flammability Limit -
9.6 FLACS variables
CFLC CFL number based on speed of sound -
CFLV CFL number based on ow velocity -
CS Speed of sound, c ms
1
DPDT Rate of pressure rise,
dp
dt
Pas
1
EPK Turbulence ratio,

k
s
1
EPS Dissipaton of turbulent kinetic energy, m
2
s
3
EQ Equivialence ratio, nite bounded, EQ =
(F/O)
(F/O)+(F/O)
stoich
-
ER Equivialence ratio, -
DIMP Drag impulse,
_
t
0
F
D
dt Pa s
DRAG Drag value,
F
D
Pa
FDOSE Dose, integral of mole fraction of fuel,
_
t
o
X
fuel
dt s
FLUX Mass ux,
m
A
kgm
2
s
1
FMIX Mixture fraction, -
FMOLE Mole fraction of fuel, X
fuel
-
FUEL Mass fraction of fuel, Y
fuel
-
FVAR Variance of mixture fraction,

2
-
GAMMA Isentropic gas constant, -
H Enthalpy, h Jkg
1
K Turbulent kinetic energy, k Jkg
1
LT Turbulent length scale, (output), l
LT
m
MACH Mach number, Ma -
MU Effective dynamic viscosity,
eff
Pa s
NUSSN Nusselt number, Nu -
OX Mass fraction of oxygen, Y
O
2
-
P Gauge pressure, overpressure, P bar
PIMP Pressure impulse,
_
t
2
t
1
Pdt bar s
PMAX Maximum over pressure, P
max
bar
PROD Mass fraction of products, Y
prod
bar
RET Turbulent Reynolds number,
ul
LT
t
-
RFU Combustion rate, R
fuel
kgm
3
s
1
RHO Density, kgm
3
RTI Relative turbulence intensity (input), I
T
-
T Temperature, T K
TAUWX Wall shear stress in x direction,
w,1
Nm
2
TAUWY Wall shear stress in y direction,
w,2
Nm
2
TAUWZ Wall shear stress in z direction,
w,3
Nm
2
TLS Turbulence lenght scale (input), l
LT
m
TURB Root mean square of velocity, u
ms
1
242 Nomenclature
TURBI Relative turbulence intensity (output), I
T
-
U Velocity component in x direction, u
1
ms
1
UDIMP Drag impulse in x direction,
_
t
0
F
D,1
dt Pa s
UDRAG Drag value in x direction, |F
D,1
| Pa
UFLUX Mass ux in x direction, u
1
kgm
2
s
1
UVW Absolute value of velocity, |u| ms
1
V Velocity component in y direction, u
2
ms
1
VDIMP Drag impulse in y direction,
_
t
0
F
D,2
dt Pa s
VDRAG Drag value in y direction, |F
D,2
| Pa
VFLUX Mass ux in y direction, u
2
kgm
2
s
1
VVEC Velocity vector, u ms
1
W Velocity component in z direction, u
3
ms
1
WDIMP Drag impulse in z direction,
_
t
0
F
D,3
dt Pa s
WDRAG Drag value in z direction, |F
D,3
| Pa
WFLUX Mass ux in z direction, u
3
kgm
2
s
1
Chapter 10
References
244 References
Abdel-Gayed, R.G. & Bradley, D. (1981). A two-eddy theory of premixed turbulent ame prop-
agation. Philosophical Transactions of the Royal Society of London, Series A, 301: 1-25.
Abdel-Gayed, R.G., Al-Khishali, K.J. & Bradley, D. (1984). Turbulent burning velocities and
ame straining in explosions. Proceedings of the Royal Society of London, Series A, 391: 389-
413.
Abdel-Gayed, R.G., Bradley, D. & Lawes, M. (1987). Turbulent burning velocities: A general
correlation in terms of straining rates. Proceedings of the Royal Society of London, Series A,
414: 389-413.
Abdel-Gayed, R.G., Bradley, D., Hamid, M.N. & Lawes, M. (1984). Lewis number effects on
turbulent burning velocity. Twentieth Symposium (Int.) on Combustion: 505-512.
Abdel-Gayed, R.G., Bradley, D., Lawes, M. & Lung, F.K-K (1986). Premixed turbulent burning
during explosions. Twenty-rst Symposium (Int.) on Combustion: 497-504.
Altman, I.S. (1999). On condensation growth of oxide particles during gas-phase combustion
of metals. Seventeenth International Colloquium on the Dynamics of Explosions and Reactive
Systems (ICDERS), 25-30 July 1999, Heidelberg, Germany.
Amyotte, P.R., Chippett, S. & Pegg, M.J. (1989). Effects of Turbulence on Dust Explosions. Jour-
nal of Loss Prevention in the Process Industries, 14: 293-310.
Arntzen, B.J. & Hansen, O.R. (1997). Improved thermodynamics in FLACS, modelling the effect of in-
ert gases. Report Christian Michelsen Research, CMR-97-F30009, Bergen, Norway.
Arntzen, B.J. (1993). Combustion modelling in FLACS93. Report Christian Michelsen Research,
CMR-93-F25043 (Condential).
Arntzen, B.J. (1997). Modelling of steady and transient ows in FLACS96. Report Christian Michelsen
Research,CMR-97-Fxxxxx (to be issued in 1997?) (Condential).
Arntzen, B.J. (1998). Modelling of turbulence and combustion for simulation of gas explosions in com-
plex geometries. Dr. Ing. Thesis, NTNU, Trondheim, Norway.
Arntzen, B.J., Salvesen, H.C., Nordhaug, H.F., Storvik, I.E. & Hansen, O.R. (2003). CFD mod-
elling of oil mist and dust explosion experiments. Proceedings Fourth International Seminar
on Fire and Explosion Hazards, 8-12 September 2003, Londonderry, Northern Ireland, UK:
601-608.
ATEX 1999/92/EC (1999). Directive 1999/92/EC (ATEX 118a) of the European Parliament and the
Council, on minimum requirements for improving the safety and health protection of workers po-
tentially at risk from explosive atmospheres.
ATEX 94/9/EC (1994). Directive 94/9/EC (ATEX 100a) of the European Parliament and the Council, on
the approximation of the laws of the member states concerning equipment and protective systems
intended for use in potentially explosive atmospheres.
Babrauskas, V. (2003). Ignition Handbook. Fire Science Publishers, Issaquah, USA.
Bardon, M.F. & Fletcher, D.E. (1983). Dust explosions. Scientig Progress, 68: 459-473.
Bartknecht, W. (1971). Brenngas- und Staubexplosionen. Forschungsbericht F45, Bundesinstitut
fr Arbeitsschutz, Koblenz, Germany (in German).
Bartknecht, W. (1978). Explosionen: Ablauf und Schutzmanahmen. Springer-Verlag, Berlin (in
German, English translation: Bartknecht, 1981).
Bartknecht, W. (1981). Explosions: course prevention protection. Springer-Verlag, Berlin.
245
Bartknecht, W. (1993). Explosionsschutz: Grundlagen und Anwendung. Springer-Verlag,
Berlin (in German).
Barton, J. (2002). Dust explosions: Prevention and protection. IChemE, Rugby.
Batchelor, G.K. & Townsend, A.A. (1947). Decay of vorticity in the isotropic turbulence. Pro-
ceedings of the Royal Society of London, Series A, 190: 534-550.
Batchelor, G.K. & Townsend, A.A. (1948a). Decay of isotropic turbulence in the initial period.
Proceedings of the Royal Society of London, Series A, 193: 539-558.
Batchelor, G.K. & Townsend, A.A. (1948b). Decay of isotropic turbulence in the nal period.
Proceedings of the Royal Society of London, Series A, 194: 527-543.
Beck, H., Glienke, C. & Mhlmann, C. (1997). BIA-Report 13/97: Combustion and explosion char-
acteristics of dusts. HVBG, Saint Augustin, Germany.
Ber, J.M., Chomiak, J. & Smooth, L.D. (1984). Fluid dynamics of coal combustion: a review.
Progress in Energy and Combustion Science, 10: 177-208.
Bjerketvedt, D., Bakke, J.R. & van Wingerden, K. (1993). Gas explosion handbook, Report Chris-
tian Michelsen Research, CMR-93-A25034, Bergen, Norway.
Bosch, C.J.H. van den, Weterings, R.A.P.M. (Editors) (1996) Methods for the calculations of
physical effects. Due to releases of hazardous materials, Report TNO CPR14E.
Bradley, B., Lau, A.K.C. & Lawes, M. (1992). Flame stretch rate as a determinant of turbulent
burning velocity. Philosophical Transactions of the Royal Society of London, Series A, 338: 359-
387
Bradley, D. (1992). How fast can we burn? Twenty-fourth Symposium (Int.) on Combustion: 247-
262.
Bradley, D. (2002). Problems of Predicting Turbulent Burning Rates. Combustion Theory and
Modelling, 6: 361-382.
Bradley, D., Chen, Z. & Swithenbank, J.R. (1988). Burning rates in turbulent ne dust-air ex-
plosions. Twenty-second Symposium (Int.) on Combustion: 1767-1775.
Bradshaw, P. (1994). Turbulence: the chief outstanding difculty of our subject. Experiments in
Fluids, 16: 203-216.
Bray, K.N.C. (1990). Studies of the turbulent burning velocity. Proceedings of the Royal Society of
London, Series A, 431: 315-335.
Butler, T.D., Cloutman, L.D., Dukowicz, J.K., Ramshaw J.D. (1981). Multidimensional numer-
ical simulation of reactive ow in internal combustion engine. Progress in Energy and Com-
bustion Science, 7: 293-315.
Cashdollar, K.L. (2000). Overview of dust explosibility characteristics. Journal of Loss Prevention
in the Process Industries, 13: 183-199.
Cashdollar, K.L., Zlochower, I.A., Green, G.M., Thomas, R.A. & Hertzberg, M. (2000).
Flammability of methane, propane, and hydrogen gases. Journal of Loss Prevention in
the Process Industries, 13: 327-340.
Cassel, H.M. (1964). Some fundamental aspects of dust ames. Report of Investigation 6551, U.S.
Department of the Interior, Bureau of Mines, Washington.
246 References
Catlin, C.A. & Lindstedt, R.P. (1991). Premixed turbulent burning velocities derived from mix-
ing controlled reaction models with cold front quenching. Combustion and Flame, 85: 427-
439.
Cesana, C. & Siwek, R. (2001). Operating instructions 20-l-apparatus, 6.0. Adolf Khner AG.
Chen, C.H. & Jaw, S.Y. (1997). Fundamentals of turbulence modeling. In Combustion: An Interna-
tional Series. Taylor & Francis, Washington.
Clift, R., Grace, J.R. & Weber, M.E. (1978). Bubbles, drops and particles. Academic Press, New
York.
CMR (1996). FLACS 96 Users GUIDE. Report Christian Michelsen Research, CMR-96-F20073
,Christian Michelsen Research AS, Bergen, Norway.
Coulsen, J.M. & Richardson, J. F. (1977). Chemical engineering. Pergamon Press, Oxford.
Crowe, C., Sommerfeld, M & Tsuji, Y. (1998). Multiphase ows with droplets and particles. CRC
Press, Boca Raton.
Dahoe, A.E. & de Goey, L.P.H. (2003). On the determination of laminar burning velocity from
closed vessel gas explosions. Journal of Loss Prevention in the Process Industries, 16: 457-478.
Dahoe, A.E. (2000). Dust explosions: a study of ame propagation. PhD-thesis, Delft University of
Technology, Delft, Holland.
Dahoe, A.E., Cant, R.S. & Scarlett, B. (2001). On the decay of turbulence in the 20-litre explosion
sphere. Flow, Turbulence and Combustion, 67: 159-184.
Dahoe, A.E., Cant, R.S., Pegg, R.S. & Scarlett, B. (2001). On the transient ow in the 20-litre ex-
plosion sphere. Journal of Loss Prevention in the Process Industries, 14: 475-487.
Dahoe, A.E., Hanjalic, K. & Scarlett, B. (2002). Determination of the laminar burning velocity
and the Markstein length of powder-air ames. Powder Technology, 122: 222-238.
Dahoe, A.E., van der Nat, K., Braithwaite, M. & Scarlett, B. (2001). On the sensitivity of the
maximum explosion pressure of a dust deagration to turbulence. KONA, 19: 178-195.
Dahoe, A.E., Zavenbergen, J.F., Lemkowitz, S.M. & Scarlett, B. (1996). Dust explosions in
spherical vessels: the role of ame thickness in the validity of the cube-root law. Journal
of Loss Prevention in the Process Industries, 9: 33-44.
Eckhoff, R.K. (1984). Relevance of using (d p/d t)max data from laboratory-scale tests for pre-
dicting explosion rates in practical industrial situations. VDI-Berichte, 494: 207-217.
Eckhoff, R.K. (2003). Dust explosions in the process industries. Third edition, Gulf Professional
Publishing, Amsterdam.
Elghobashi, S. & Truesdell, G.C. (1993). On the two-way interaction between homogeneous
turbulence and dispersed solid particles. I: Turbulence modication. Physics of Fluids A
(Fluid Dynamics), 5: 1790-1801.
Elghobashi, S. (1991). Particle-laden turbulent ows: direct simulation and closure models. Ap-
plied Scientic Research, 48 :301-314.
Elghobashi, S. (1994). On predicting particle-laden turbulent ows. Applied Scientic Research,
52: 309-329.
EN 1127-1 (1997). Explosive atmospheres Explosion prevention and protection, Part 1: Basic
concepts and methodology. European Standard, CEN, Brussels, _XXX_Month, 1997.
247
EN 14491 (2006). Dust explosion venting protective systems. European Standard, CEN, Brussels,
March 2006.
EN 50281-3 (2000). Equipment for use in the presence of combustible dust, Part 3: Classication of
areas where combustible dusts are or may be present. European Standard, CEN, Brussels, _-
XXX_Month, 1997.
Ertesvg, I.S. (2000). Turbulent stryming og forbrenning: Fr turbulensteori til ingenirverkty. ISBN
82-519-1568-6. TAPIR Forlag, Trondheim (in Norwegian).
Gibbs, G.J. & Calcote, H.F. (1959). Effect of molecular structure on burning velocity. Journal of
Chemical and Engineering Data, 4: 226-237.
Glassman, I. (1996). Combustion. Third Edition. Academic Press, San Diego.
Golder, D. (1972). Relations among stability parameters in the surface layer. Boundary-Layer Me-
teorology 3: 47-58.
Gore, R.A. & Crowe, C.T. (1989). Effect of Particle Size on Modulating Turbulent Intensity. In-
ternational Journal of Multiphase Flow, 15: 279-285.
Goroshin, S. & Lee, J. (1999). Laminar dust ames: A program of microgravity and ground
based studies at McGill. Fifth International Microgravity Combustion Workshop, 18-20 May
1999, Cleveland, Ohio.
Glder, O.L. (1990). Turbulent premixed ames for different combustion regimes. Twenty-third
Symposium (Int.) on Combustion: 743-750.
Han, J., Arya, S. P., S. Shen, Lin, Y.-L. (2000) An estimation of turbulent kinetic energy and en-
ergy dissipation rate based on atmospheric boundary layer simimlarity theory. NASA/CR-
2000-212298
Han, O-S., Yashima, M., Matsuda, T., Matsui, H., Atsumi, M. & Ogawe, T. (2000). Behaviour
of ames propagating through Lycopdium dust clouds in a vertical duct. Journal of Loss
Prevention in the Process Industries, 13: 449-457.
Han, O-S., Yashima, M., Matsuda, T., Matsui, H., Atsumi, M. & Ogawe, T. (2001). A study of
ame propagation mechanisms in Lycopdium dust clouds based on dust particles be-
haviour. Journal of Loss Prevention in the Process Industries, 14: 153-160.
Hansen, O.R. & sheim, O. (1995). Simulation of explosion experiments using FLACS93 version 2.0.
Report Christian Michelsen Research, CMR-95-F20004 (Condential).
Hansen, O.R. & Storvik, I. (1993). FLACS93 version 1.0 validation report.Report Christian
Michelsen Research, CMR-93-F25052 (Condential).
Hansen, O.R. & Teigland, R. (1994). A multi-block extension of FLACS for BLAST calculations.
Report Christian Michelsen Research, CMR-94-F25053 (Condential).
Hansen, O.R. & Teigland, R. (1995). Validation of the BLAST option in FLACS.Report Christian
Hansen, O.R. & van Wingerden, K. (1993). Modelling of the effects of water spray on gas explosions
implementation in FLACS93.Report Christian Michelsen Research, CMR-93-F25048 (Con-
dential).
Hansen, O.R. (1994). Grid dependency study using FLACS.Report Christian Michelsen Research,
248 References
Hansen, O.R. (1997). Improvement of the waterspray model in FLACS.Report Christian Michelsen
Research, CMR-97-F30012 (Condential).
Hansen, O.R., Skjold, T. & Arntzen, B.J. (2004). DESC A CFD-tool for dust explosions. Inter-
national ESMG Symposium, Nrnberg, 16-18 March 2004.
Hattwig, M. & Steen, H. (2004). Handbook of explosion prevention and protection. Wiley-VCH Ver-
lag, Weinheim.
Hauert, F., Vogl, A. & Radant, S. (1994). Measurement of turbulence and dust concentration in
silos and vessels. Proceedings Sixth International Colloquium on Dust Explosions, Shenyang,
August 29 September 2, 1994: 71-80.
Hauert, F., Vogl, A. & Radant, S. (1996). Dust cloud characterization and the inuence on the
pressure-time histories in silos. Process Safety Progress, 15: _XXX_PageRange.
Hertzberg, M., Zlochower, I.A. & Cashdollar, K.L. (1992). Metal dust combustion: explosion
limits, pressures, and temperatures. Twenty-fourth Symposium (Int.) on Combustion: 1827-
1835.
Hinze, J.O. (1972). Turbulent uid and particle interactions. Progress in Heat and Mass Transfer, Vol-
ume 6, Proceedings of the International Symposium on Two-Phase Systems, Hetsroni, G., Side-
man, S. & Hartnett, J.P. (Eds.), Pergamon Press, Oxford: 433-452.
Hinze, J.O. (1975). Turbulence. Second Edition. McGraw-Hill, New York.
ISO 6184-1 (1985). Explosion Protection Systems Part 1: Determination of explosion indices of com-
bustible dusts in air. ISO.
Jarosinski, J. & Podlipski, J. (1999). Combustion mechanism of dust clouds in microgravity.
Fifth International Microgravity Combustion Workshop, Cleveland, Ohio.
Jarosinski, J., Podlipski, J. & Pu, Y.K. (1999). Visualisation of dust explosions under micro-
gravity conditions. Sevententh International Colloquium on the Dynamics of Explosions
and Reactive Systems (ICDERS), Heidelberg, Germany.
Kauffman, C.W., Srinath, S.R., Tezok, F.I., Nicholls, J.A. & Sichel, M. (1984). Turbulent and ac-
celerating dust ames. Twentieth Symposium (Int.) on Combustion: 1701-1708.
Kobayashi, H., Kawabata, Y. & Maruta, K. (1998). Experimental study on general correlation of
turbulent burning velocity at high pressure. Twenty-seventh Symposium (Int.) on Combustion:
941-948.
Kobayashi, H., Nakashimi, T., Tamura, T., Maruta, K. & Niioka, T. (1997). Turbulence mea-
surements and observations of turbulent premixed ames at elevated pressures up to 3.0
MPa. Combustion and Flame, 108: 104-117.
Koch, D.L. (1990). Kinetic Theory for a Monodisperse Gas-Solid Suspension. Physics of Fluids
A, 2 (10): 1711-1723.
Krause, U. & Kasch, T. (2000). The inuence on ow and turbulence on ame propagation
through dust-air mixtures. Journal of Loss Prevention in the Process Industries, 13: 291-298.
Kulick, J.D., Fessler, J.R. & Eaton, J.K. (1994). Particle response and turbulence in fully devel-
oped channel ow. Journal of Fluid Mechanics, 277: 109-134.
Kundu, P.K. and I.M. Cohen (2004). Fluid Mechanics. Amsterdam, Elsevier Academic Press,
Amsterdam.
Kuo, K.K. (2005). Principles of Combustion. Second Edition. John Wiley & Sons, New Jersey.
249
Langeland, T. (1993). FLACS INTERFACES le interface description.Report Christian Michelsen
Research, CMR-93-F40004, Bergen, Norway.
Langeland, T. (1993). FLACS Interfaces, File Interface Description .Report Christian Michelsen
Research, CMR Report CMR-93-F40004, Bergen, Norway.
Langeland, T. (1995). FLACS INTERFACES installation guide.Report Christian Michelsen Re-
search, CMR-95-F40003, Bergen, Norway.
Langeland, T. (1997). Flowvis Version 3.2, Users Guide.Report Christian Michelsen Research,
CMR-97-F50005, Bergen, Norway.
Lauder, B.E. & Spalding, D.P. (1974). The numerical computation of turbulent ows. Computer
Methods in Applied Mechanics and Engineering, 3: 269-289.
Law, C.K. (1993). _XXX_Title. In: Peters, N. & Rogg, B. (Eds.) , Reduced Kinetic Mechanisms for
Application in Combustion Systems, Springer, Berlin 1993.
Lee, J.H.S. (1988). Dust explosion parameters: their measurement and use. VDI-Berichte, 701:
113-122.
Lee, J.H.S., Pu, Y.K. & Knystautas, R. (1987). Inuence of turbulence on closed volume explo-
sion of dust-air mixtures. Archivum Combustionis, 7: 279-297.
Lipatnikov, A.N. & Chomiak, J. (2002). Turbulent ame speed and thickness: phenomenology,
evaluation, and application in multi-dimensional simulations. Progress in Energy and Com-
bustion Science, 28: 1-74.
Loth, E. (2000). Numerical approaches for motion of dispersed particles, droplets and bubbles.
Progress in Energy and Combustion Science, 26: 161-223.
Mannan, S. (2005). Lees loss prevention in the process industries.Third Edition. Elsevier Butter-
worth Heinemann, Amsterdam.
Meinkln, D. (1999). Oxide layer effects in metal particle combustion. Fifth International Micro-
gravity Combustion Workshop, 18-20 May, Cleveland, Ohio.
Metghalchi, M., Keck, J.C. (1980). Laminar burning velocity of propane-air mixtures at high
temperature and pressure. Combustion and Flame, 38: 143-154.
Mintz, K.J. (1993). Upper explosive limit of dusts: experimental evidence for its existence under
certain circumstances. Combustion and Flame, 94: 125-130.
Mitgau, P. (1996). Einu der Turbulenzlnge und der Schwankungsgeschwindigkeit auf die Verbren-
nungsgeschwindigkeit von Aerosolen. Abteilung Reaktionskinetik, Gttingen, Max-Planck-
Institut fr Strmungsforschung.
Motif (_XXX_Year). Motif 1.2 Users Guide Sun Microsystems. Inc. 2550 Garcia Avenue, Mountain
View, California 94043 U.S.A. Part No: 801-5363-10
Muhammadi, B. & Pirinneau, O. (1994). Analysis of the k-epsilon turbulence model. John Wiley &
Sons, Masson, Paris.
Nagy, J. & Verakis, H.C. (1983). Development and control of dust explosions. In the series: Occupa-
tional Safety and Health, 8, Marcel Dekker, New York.
Nagy, J., Conn, J.W. & Verakis, H.C. (1969). Explosion development in a spherical vessel. Report of
Investigation 7279, U.S. Department of the Interior, Bureau of Mines.
Nagy, J., Seiler, E.C., Conn, J.W. & Verakis, H.C. (1971). Explosion development in closed vessel.
Report of Investigation 7502, U.S. Department of the Interior, Bureau of Mines, Washington.
250 References
NASA (2005). ThermoBuild interactive tool using the thermodynamic database at NASA
Glenn Research Center.
NFPA 655 (2007). Standard for prevention of sulphur res and explosions 2007 edition. National
Fire Protection Association, Quincy, MA.
NFPA 68 (2007). Standard on explosion protection by deagration venting 2007 edition. National
Fire Protection Association, Quincy, MA.
NORSOK Standard Z-013 (2001). Risk and emergency preparedness analysis. Rev. 2, Norwegian
Technology Centre.
Patankar, Suhas V. (1980). Numerical Heat Transfer and Fluid Flow. Taylor & Francis, 1980
Peirano, E. & Leckner, B. (1998). Fundamentals of turbulent gas-solid ows applied to circulat-
ing uidized bed combustion. Progress in Energy and Combustion Science, 24: 259-296.
Popat, N.R., Catlin, C.A., Arntzen, B.J., Lindstedt, R.P., Hjertager, B.H., Solberg, T., Saeter, O. & Van den Berg, A.C. (1996).
Investigation to improve the accuracy of computational uid dynamic based explosion
models. Journal of Hazardous Material, 45: 1-25.
prEN14373 (2002). Explosion suppression systems. CEN, Brussels (draft March 2002). _XXX_-
Update
prEN14460 (2002). Explosion resistant equipment. CEN, Brussels (draft April 2002). _XXX_Update
prEN14491 (2002). Dust explosion venting protective systems. CEN, Brussels (draft June 2002).
_XXX_Update
prEN15089 (2004). Explosion isolation systems. CEN, Brussels (draft November 2004). _XXX_-
Update
Press, W.H., Teukolsky, S.A., Vetterling, W.T. & Flannery, B.P. (1992). Numerical recipes in C.
The art of scientic computing. Second edition. Cambridge. _XXX_Publisher_and_-
FORTRAN_Edition
Pu, Y.K. (1988). Fundamental characteristics of laminar and turbulent ames in cornstarch dust-air
mixtures. PhD Thesis, McGill University, Quebec.
Pu, Y.K., Jarosinski, J., Johnson, V.G. & Kauffman, C.W. (1990). Turbulence effects on dust explo-
sions in the 20-liter spherical vessel. Twenty-third Symposium (Int.) on Combustion: 843-849.
Pu, Y.K., Jarosinski, J., Tai, C.S., Kaufmann, C.W. & Sichel, M. (1988a). The investigation of the
feature of dispersion induced turbulence and its effect on dust explosions in closed vessels. Twenty-
second Symposium (Int.) on Combustion: 1777-1787.
Pu, Y.K., Li, Y.C., Kauffman, C.W. & Bernal, L.P. (1989). Determination of turbulence parameters in
closed explosion vessels. Twelvth International Colloquium on the Dynamics of Explosions and
Reactive Systems (ICDERS), Ann Arbor, Michigan, 23-25 July: 107-123.
Pu, Y.K., Mazurkiewicz, J., Jarosinski, J. & Kaufmann, C.W. (1988b). Comparative study of the
inuence of obstacles on the propagation of dust and gas ames. Twenty-second Symposium (Int.)
on Combustion: 1789-1797.
RASE (2000). The RASE project explosive atmospheres: risk assessment of unit operations and equip-
ment; report: methodology for the risk assessment of unit operations and equipment for use in
potentially explosive atmospheres. EU Project No. SMT4-CT97_2169. 17 March 2000.
Rogers, C.B. & Eaton, J.K. (1991). The effect of small particles on uid turbulence in a at-plate,
turbulent boundary layer in air. Physics of Fluids A, 3: 928-937.
251
Salvesen, H.-C. & Storvik, I.E. (1994). Subgrid modelling of drag forces for ow through louvres and
grating. Report Christian Michelsen Research, CMR-94-F25017 (Condential).
Salvesen, H.-C. (1995). Modelling of jet release of liqueed gas under high pressure. Report Christian
Salvesen, H.-C. (1996a). Subgrid modelling of ow through louvres and grating drag and deection
forces. Report Christian Michelsen Research, CMR-96-F20052 (Condential).
Salvesen, H.-C. (1996b). FLACS validation simulations of explosion experiments in 1:5 scale M24
module using new subgrid model for ow through louvres. Report Christian Michelsen Research,
Sand, I.. & Bakke, J.R. (1989). Wall-function boundary conditions in the solution of the Navier-
Stokes and the energy equations. Report Christian Michelsen Research, CMI-25110-3 (Con-
dential).
Senecal, J.A. & Beaulieu, P.A. (1998). KG: new data and analysis. Process Safety Progress, 17: 9-
15.
Shirolkar, J.S., Coimbra, C.F.M. & Queiroz McQuay, M. (1996). Fundamental aspects of mod-
eling turbulent particle dispersion in dilute ows. Progress in energy and combustion science,
22: 363-399.
Siwek, R. (1977). 20-L Laborapparatur fr die Bestimmung der Explosionskenngren brennbarer
Stube. Thesis, HTL Winterthur, Switzerland.
Siwek, R. (1988a). Zuverlssige Bestimmung explosionstechnischer Kenngren in der 20-
Liter-Laborapparatur. VDI Berichte, 701: 215-262.
Siwek, R., (1988b). Reliable determination of safety characteristics in the 20-litre apparatus. Pro-
ceedings of Conference on Flammable Dust Explosions, St. Louis, 2-4 November 1988.
Siwek, R., van Wingerden, K., Hansen, O.R., Sutter, G., Schwartzbach, Chr., Ginger, G. & Meili, R. (2004).
Dust explosion venting and suppression of conventional spray driers. Eleventh Interna-
tional Symposium on Loss Prevention, Prague, 31 May to 3 June 2004.
Skjold, T, Arntzen, B.J., Hansen, O.R., Taraldset, O.J., Storvik, I.E. & Eckhoff, R.K. (2005a).
Simulating dust explosions with the rst version of DESC. Process Safety and Environmental
Protection, 83: 151-160.
Skjold, T. & Hansen, O.R. (2005). The development of DESC A dust explosion simulation
code. International European Safety Management Group (ESMG) Symposium, 11-13 October
2005, Nrnberg, Germany, 24 pp.
Skjold, T. (2003). Selected aspects of turbulence and combustion in 20-litre explosion vessels.
Cand. Scient. Thesis, Department of Physics, University of Bergen, Norway.
Skjold, T. (2007). Review of the DESC project. Journal of Loss Prevention in the Process Industries,
20: 291-302.
Skjold, T., Arntzen, B.J., Hansen, O.J., Storvik, I.E. & Eckhoff, R.K. (2005b). Simulation of
dust explosions in complex geometries with experimental input from standardized tests.
Journal of Loss Prevention in the Process Industries, 19: 210-217.
Skjold, T., Larsen, . & Hansen, O.R. (2006b). Possibilities, limitations, and the way ahead for
dust explosion modelling. HAZARDS XIX, 28-30 March 2006, Manchester, UK: 282-297.
252 References
Skjold, T., Y.K. Pu, Arntzen, B.J., Hansen, O.J., Storvik, I.E., Taraldset, O.J. & Eckhoff, R.K. (2005c).
Simulating the inuence of obstacles on accelerating dust and gas ames. Poster presenta-
tion Twentieth International Colloquium on the Dynamics of Explosions and Reactive Systems
(ICDERS), 31 July to 5 August 2005, Montreal, Canada.
Skrbek, L. & Stalp, S.R. (2000). On the decay of homogeneous isotropic turbulence. Physics of
Fluids, 12: 1997-2019.
Sokolik, A. S., Karpov, V. P. & Semenov, E. S. (1967). Turbulent combustion of gases. Combus-
tion, Explosion, and Shock Waves, _XXX_Volume: 36-45 (Translated from Fizika Goreniya i
Vzryva, 3: 6176, 1967].
Squires, K.D. & Eaton, J.K. (1991). Preferential concentration of particles by turbulence. Physics
of Fluids A, 3: 1169-1178.
Steinberg, T.A., Wilson, D.B. & Benz, F. (1992). The combustion phase of burning metals. Com-
bustion and Flame, 91: 200-208.
Strehlow, R.A. (1979). Fundamentals of Combustion. Robert E. Kreiger publishing company, New
York.
Tai, C.S., Kauffman, C.W., Sichel, M. & Nicholls, J.A. (1988). Turbulent dust combustion in a
jet-stirred reactor. Progress Astronautics and Aerodynamics, 113: 62-86.
Tennekes, H. & Lumley, J.L. (1972). A First Course in Turbulence. The MIT Press, Cambridge.
Truesdell, G.C. & Elghobashi, S. (1994). On the two-way interaction between homogeneous
turbulence and dispersed solid particles. II: Particle dispersion. Physics of Fluids A, 6: 1405-
1407.
Tsinober, A. (2001). An informal introduction to turbulence. Kluwer Academic Publishers, Dor-
drecht.
Turns, S.R. (1996). An introduction to combustion: concepts and applications. McGraw-Hill, New
York.
Vagelopoulos, C.M., Egolfopoulos, F.N. & Law, C.K. (1996). Further considerations on the de-
termination of laminar ame speeds with the counterow twin-ame technique. Twenty-
fth Symposium (Int.) on Combustion: 1341-1347.
van der Wel, P.G.J. (1993). Ignition and Propagation of Dust Explosions. PhD Thesis, Delft Univer-
sity Press, Delft.
van Wingerden, K. (1996). Simulations of dust explosion using a CFD-code. Proceedings Sev-
enth International Colloquium on Dust Explosions, Bergen: 6.42-6.51.
van Wingerden, K., Arntzen, B.J. & Kosinski, P. (2001). Modelling of dust explosions. VDI-
Berichte, 1601: 411-421.
Veynante, D. & Vervisch, L. (2002). Turbulent combustion modelling. Progress in Energy and
Combustion Science, 28: 193-266.
Williams, F.A. (1986). Lectures on applied mathematics in combustion: Past contributions and
future problems in laminar and turbulent combustion. Physica D, 20D: 21-34.
Yuan, Z. & Michaelides, E.E. (1992). Turbulence modulation in particulate ows a theoretical
approach. International Journal of Multiphase Flow, 18 (5): 779-785.
Zevenbergen, J.F. (2004). Report on turbulence measurements in the 20-litre sphere: For partners
DESC project. The Explosion Group of Delft University of Technology.
253
Zhen, G. & Leuckel, W. (1995a). Inuence of transient injection induced turbulent ow on gas
and dust explosions in a closed vessel. Loss Prevention and Safety Promotion in the Process
Industries, 2: 257-268.
Zhen, G. & Leuckel, W. (1995b). The dynamic ow condition of dust dispersion induced turbu-
lence. Proceedings of Tenth Symposium of Turbulent Shear Flows, 3 (25): 7-12.
Zhen, G. & Leuckel, W. (1996). Determination of dust-dispersion-induced turbulence and its in-
uence on dust explosions. Combustion Science and Technology, 113-114: 629-639.
Zhen, G. & Leuckel, W. (1997). Effects of ignitors and turbulence on dust explosions. Journal of
Loss Prevention in the Process Industries, 10: 317-324.
Index
a1le, 202
abbreviations, 240
Aerodynamic roughness length, 77
air leak, 87
ASCII le, 200202
assembly, 45
average pressure, 124
axis show, 55
best practice, 206
beta model, 229
blank, 47
block, 103
Boundary conditions, 70
Boundary, Bernoulli, 70
Boundary, Eqchar, 70
Boundary, Euler, 70, 71
Boundary, Nozzle, 70, 71
Boundary, Plane Wave, 70
Boundary, Symmetry, 70
Boundary, Wind, 70, 72, 227
bugs and problems in CASD, 110
Burning velocity, 230
CASD, 32
casd command line options, 32
cc le, 147
CFLC, 67
CFLV, 67
change colour table, 166
cloud le, 140
cloud interface, 140
cn le, 148
cole, 198
colour scheme, 37
Combustion modelling, 229
comerge, 198
command input CASD, 35
contours, 173
database maintenance, 42
database, connect to , 41
database, creating, 41
dbfutil, 42
delete grid line, 57
delete instance, 51
delete subtree, 51
diffuse leak, 86
Dissipation, transport equation, 224
dose, 124
dumple, reading old, 111
Enthalpy, transport equation, 224
Equivalence ratio , 222
equivalent stoichiometric gas cloud, 87
error, 153
example, 20, 180
export, Flowvis , 163
exporting geometry CASD, 106
fan leak type, 87
Flame model, 229
ash, 195
Flowvis, 156
Fluid ow equations, 224
font settings, 179
Fuel lean mixture, 222
Fuel rich mixture, 222
Fuel, transport equation, 224
Fuel-oxidant ratio , 222
gas cloud, 83
gas cloud denition, 140
gas composition, 83
Gas Explosion Programme (GEP), 2
gas monitor region, 102
gas position and volume, 83
Gas Safety Program (GSP), 3
Gas thermodynamics, 222
geo2acs, 188
geometry appearance, 174
geometry import, 111, 188
geometry menu, 39
geometry, new, 40
geometry, open, 40
gexcon colour convention, 37
global objects, 43
gm, 192
grating, 100
Gravity constant, 75
INDEX 255
grid, 56, 192
grid guidelines, 58
grid smooth, 57
grid stretch, 57
Ground roughness, 77
heavy hydrocarbons, 111
hiding geometry, 174
hinged panel, 81
Ideal gas law, 223
ignition, 90
import grid, 58
import, Flowvis, 163
inactive panel, 81
Initial conditions, 75
Initial temperature, 76
instance, add, 46
interpolate, 179
invisible, 47
jet, 193
jet leak, 86
Jet sources, 230
job numbers, 5
leak, 85
leak buildup time, 87
leak excess area, 154
license terms, 180
local object, properties, 47
local objects, 43
LOD, 107
louvre panel, 95
macro le, 108
maintenance, 19
mark subtree, 51
Mass balance, 224
mass residual, 138, 153
material, 40
matrix, 47
Mixing of gases, 222
Mixture fraction, 222
Mixture fraction, transport equation, 224
modify page, 166
Momentum equation, 224
Monin-Obukhov length scale, 227
monitor panel, 79, 81
monitor points, 62
move grid line, 57
moving plot elements, 180
multiblock, 103
new Flowvis presentation, 160, 162
numercal schemes, 232
object create, 41
object open, 41
object, open, 48
online help, 180
optimize computer load, 117
options, 179
overlay panel, 81
panel drag coefcient, 83
panel maximum travel distance, 83
panel opening pressure difference, 82
panel porosity, 82
panel pressure, 124
panel sub sizes, 83
panel type imp, 82
panel weight, 82
panels, 80
particle traces, 175
Pasquill class, 78
pdf, 164
plastic panel, 81
plot domain, 171
plot specication, Flowvis, 170
Pool leakage le, 144
Pool setup le, 142
popout panel, 81
porcalc, 61
porsity calculation, 61
Portable Document Format, 164
position grid line, 56
Prandtl-Schmidt numbers, 226
pressure relief panels, 78
printing, Flowvis, 163
Progress variable, 222
project, new, 40
project, open, 40
Q8, 87
Q9, 87
r1le, 200
r3le, 201
range colours, 166
rdle, 197
rename, 48
rotate, 175
run manager, 15
run script, 117
scale object, 47
scenario, 61
scenario le, 132
script, 117
256 INDEX
select object, 47
shocked ow, 193
simulation volume, 56
single colours, 166
Speed of sound , 223
Stoichiometric mixture, 222
Stoichiometric reaction, 223
stopping simulations, 117
substitute job, 161, 165
substitute object, 47
substitute subtree, 51
suction, 87
support, 19
Toxic dose, 128
Toxic probabilty of death, 128
Toxic probit function, 128
toxic substances, 125
Turbulence model, 225
Turbulence viscosity, 225
Turbulent kinetic energy, transport equation,
224
two-phase ow, 195
unblank, 47
unspecied panel, 81
user dened species, 102
utilities, 188
value on curve, 178
variable appearance, 172
variables, 241
vector plot, 174
Velocity, friction , 227
ventilation time stepping, 136
verify porosity, 178
verify porosity option, 178
vessel burst, 138
visible, 47
Wall functions, 226
warning, 153
water spray, 91
Wind, 72, 77, 227
Wind buildup time, 73
Wind direction, 73
Wind speed, 72
windows service license manager, 13

Flacs v9 Manual

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Flacs v9 Manual

Încărcat de

Drepturi de autor:

Formate disponibile

FLACS v9.

, and the mean ow velocity U

, and the mean ow velocity U

. The default value is T

3.7.6.7 Ambient pressure

(upstream Mach no. corresponding to choking at the grating) is given by:

(). This relation is modelled

is the friction velocity. The Monin-

, the friction velocity. u

is generally given by:

depend on the grid size and the

{nd} order upwind and 2

Constant in the k equation; typically C

Root mean square of velocity ms

Dimensionless reaction rate -

S-ar putea să vă placă și