Documente Academic
Documente Profesional
Documente Cultură
SYSTEMS
Doctor of Philosophy
by
Zhigang Shang
Declaration
No portion of the work referred to in this thesis has been submitted in support of an
application for another degree or qualification of this or any other university, or other
institution of learning.
Zhigang Shang
Acknowledgements
I would like to express my sincere gratitude to Professor Antonis Kokossis for his
guidance and encouragement throughout the course of this study. His patience and
understanding have been tremendous. I would like to thank him for spending long hours
on reviewing this thesis.
I would also like to thank Professor Robin Smith, as the head of the department, gave
me the opportunity to do my doctoral degree in the department.
Many thanks to all the staff and students in the department for their support and for
creating a pleasant atmosphere. I consider myself very lucky indeed for having been
able to work in such a friendly and dynamic environment. Big thanks to you for making
my stay in Manchester so wonderful.
I wish to thank member companies of the Process Integration Research Consortium for
funding the research.
I would like to thank my wonderful parents for their love and support throughout my
life. In spite of being thousands of miles away, they have been so close to me with their
prayers and support.
No words can express my gratitude to my beloved wife, Jun. I thank her for her love
and encouragement during this very demanding time.
ii
Abstract
This thesis provides systematic methodologies for the analysis and optimisation of total
site utility systems under operational variations. The methodologies address three major
problems: (i) the optimal design of total site utility systems, (ii) the debottlenecking and
planning optimisation of an existing site, (iii) the optimisation of total site maintenance
scheduling.
A set of new models is developed for boilers, condensing turbines and gas turbines.
These models are used for the analysis and optimisation of site utility systems in view
of operational variations. They enable for an accurate prediction of unit efficiencies and
embody the efficiency trends of realistic units in terms of their variation with capacity,
load and operating conditions.
In the design of site utility systems, it is often the case that strong interactions exist
between the site utility systems and site processes. A novel approach is proposed for the
synthesis and design of site utility systems integrated with site processes. The design
decisions are mainly concerned with the selection of the optimal steam levels, and the
determination of the layout of the utility system. The approach combines the benefits of
pinch analysis, thermodynamic analysis and mathematical optimisation techniques. It
easily identifies the interactions between the utility system and site processes and
greatly reduces the size and complexity of the optimisation problem.
iii
iv
Table of Contents
1.2
1.4
INTRODUCTION ..................................................................................................... 5
2.2
SySTEMS ......................................................................................................................... 5
2.3
2.4
INTRODUCTION ................................................................................................... 13
3.2
CONCLUSIONS ..................................................................................................... 29
INTRODUCTION ................................................................................................... 31
4.2
4.3
4.4
4.6
CONCLUSIONS ..................................................................................................... 58
INTRODUCTION .................................................................................................... 59
5.2
5.3
5.4
vi
5.8
6.1
6.2
6.3
6.4
6.5
7.2
vii
7.4
8.2
viii
List of Figures
10
10
Figure 2-3: The Site Composite Curves for Minimum Fuel Requirement (MFR) and
Minimum Utilities Cost (MUC)
11
Figure 2-4: The construction of the Site Utility Grand Composite Curve
12
15
15
Figure 3-3: Typical heat loss plot for boilers (Pattison and Sharma. 1980)
16
Figure 3-4: The BHM accounts for the effect of boiler size, load and operating
19
conditions on efficiency
Figure 3-5: The THM of steam turbines
21
22
24
Figure 3-8: Typical data on the electrical efficiency of gas turbines as a function of size
(Marechal and Kalitventzeff, 1998)
26
Figure 3-9: The GTHM accounts for the effect of gas turbine size and load on
28
efficiency
Figure 4-1: Different operation scenarios can be represented by sets of total site profiles
32
Figure 4-2: Schematic showing the steam level optimisation strategy
33
Figure 4-3: Transhipment network representation of the total site heat flow
35
Figure 4-4: Heat flow pattern of the temperature intervals for steam level i
36
Figure 4-5: Heat flow pattern in each temperature interval of the heat source cascade 41
Figure 4-6: Heat flow pattern in each temperature interval of the steam level cascade 42
ix
Figure 47: Heat flow pattern in each temperature interval of the heat sink cascade
42
47
Figure 49: The total site profiles for the site of Case 1
49
50
51
52
52
53
56
Figure 416: The resulting SUGCCs corresponding to the optimal steam levels of Case
57
62
63
65
66
67
69
70
71
72
74
75
Figure 512: The candidate BP steam turbines for the case of two scenarios
75
76
Figure 514: The effect of part load operation prevails over the increase of efficiency
77
78
79
Figure 517: Candidate VHP condensing turbines for the case of two scenarios
80
Figure 518: Candidate surplus steam condensing turbines for the case of two scenarios
81
92
94
96
97
99
100
105
107
Figure 63: Decomposing a complex turbine into different sizes of cylinders in each
107
expansion zone
108
Figure 65: The boiler and back-pressure turbine cycle for debottlenecking
109
Figure 66: The gas turbine and waste heat boiler cycle for debottlenecking
110
Figure 67: The gas turbine and waste heat boiler and condensing turbine cycle for
110
debottlenecking
Figure 68: The boiler and condensing turbine cycle for debottlenecking
111
111
112
121
122
126
128
129
131
135
136
137
141
152
155
xi
156
156
156
157
157
171
172
xii
List of Tables
48
Table 42: Candidate saturation temperatures for each steam level of Case 1
49
Table 43: Heat provided by the process heat sources in each temperature interval of
C~1
Table 44: Heat required by the process heat sinks in each temperature interval of Case
50
53
54
54
55
55
Table 410: Heat required of process sinks above temperature interval (1,1) for Case 2
55
Table 411: Surplus heat of process sources below temperature interval (I,J) for Case 2
55
Table 412: Optimal steam levels for single and mUltiple scenarios
57
76
89
91
91
92
Table 56: Steam amount across each expansion zone of Case 1 (t/h)
93
Table 57: Power outputs of possible SCT and BBPT cycles of Case 1
93
94
94
95
95
xiii
95
98
99
101
123
123
123
Table 6-4: Maximum power generation of the existing system of Case 1(MW)
124
124
Table 6-6: Capacities of simple turbines in every expansion zone of Case l(tIh)
125
125
127
127
127
128
130
130
132
132
Table 6-16: Power demands of the allocated turbines of the industrial case
133
Table 6-17: Utility cost data in different periods for the industrial case
133
Table 6-18: Maximum power generation of the existing system of the industrial case
(MW)
133
Table 6-19: Total steam flowrates across expansion zones for the industrial case (tIh)
134
Table 6-20: Capacities of existing turbines in every expansion zone for the industrial
case (tlh)
134
Table 6-21: Capacities of potential steam turbines for each period for the industrial case
(tlh)
135
Table 6-22: Capacities of GTWB cycles for different periods for the industrial case 136
Table 6-23: Optimal operation of all turbines of the industrial case (MW)
xiv
138
Table 6-24: Optimal operation of all boilers of the industrial case (tlh)
138
Table 6-25: Optimal operation of all gas turbines of the industrial case (MW)
138
152
153
153
153
xv
Chapter 1
Introduction
Chapter 1
Introduction
1.1
Problem Description
It is often the case that considerable changes exist in the chemical processes as a result
Most of the research conducted on the analysis and optimisation of total site utility
systems has given little attention to the operational variations and strong interactions
between the site utility system and site processes. Above all, no systematic methodology
for the analysis and optimisation of total site utility systems under operational variations
has been suggested.
Chapter i
introduction
p;:;o~
TIme
Feed specifications
c:> I
c:> I
Process
11
Process
31
UHeat
Process
21
Process n
c:>
~
TIme
TIme
Utility demands
Power
b::
TIme
In this thesis, three types of issues are addressed for total site utility systems:
1.2.1
In design situations the objective is to design the site utility system that will meet
fluctuating utility demands of site processes. The main decisions involved are the
selection of the steam levels with respect to their temperatures and pressures and the
determination of the configuration of the operating units with respect to type, number,
capacity and connections among the units. The best practical site utility system design
will feature a minimum total cost.
Introduction
Chapter 1
In debottlenecking and planning situations, the objective is to make sure the site utility
system satisfies the changing utility demands of site processes in a long-term horizon in
order to minimise the total cost. It involves the following two tasks: the selection of the
new units to be added to the current system, and the determination of the optimal
operational strategies of all units with respect to operating conditions and on/off status.
As both site process units and site utility units have to go through shutdown and start-up
changes, steam and power demands gradually increase or decrease. Strong interactions
exist between the site processes and the site utility system. In maintenance scheduling
situations the objective is to determine the optimal operation and maintenance schedules
of the total site. The total site maintenance scheduling approach simultaneously
considers the maintenance and operation of the site utility system and site processes so
that the total operating cost is minimised.
The procedures use total site analysis, thermodynamic analysis and mathematical
optimisation techniques. The work aims at using previous total site analysis tools and
developing new thermodynamic tools which can help engineers to scope and screen
promising design options. Mathematical optimisation techniques are then proposed to
find the optimum solution.
Introduction
Chapter 1
The next chapter presents a review of previous techniques for the analysis and
optimisation of total site utility systems. The hardware models for steam turbines, gas
turbines and boilers are introduced in chapter 3. A systematic methodology for the
design of total site utility systems is proposed in chapter 4 and chapter 5. Chapter 4
introduces the method for steam level optimisation and chapter 5 introduces the method
for configuration optimisation of site utility systems. A systematic methodology for
debottlenecking and planning optimisation of an existing site is presented in chapter 6.
Chapter 7 deals with maintenance scheduling problem for a total site. The last chapter
presents major conclusions of the work, along with the possible future research on the
topic.
Chapter 2
Chapter 2
Review of Previous Work
2.1
Introduction
Previous approaches to the synthesis and design of site utility systems will be
introduced first. This will be followed by a discussion on the available techniques for
the debottlenecking, and planning and scheduling of site utility systems. Finally, due to
their relevance to the concepts that will be developed in the following chapters, a
description of total site analysis approaches is provided.
A well-designed site utility system should be able to match the plant heat and power
variations and keep its total costs to a minimum. Design decisions are mainly concerned
with the selection of the optimum steam levels and the determination of the
configuration of the site utility system.
Several methods have been presented previously to address the problem of synthesis
and design of utility systems. These methods generally follow three basic approaches:
those based on thermodynamic targets, those based on pinch analysis targets, and those
based on optimisation techniques. Examples of the first group are papers presented by
Nishio et ai. (1980), EI-Masri and Magnusson (1984) and Chou and Shih (1987). The
traditional way of designing utility systems using thermodynamic methods is to
maximise the thermal efficiency of the system. Thermodynamic analysis reveals the
Chapter 2
thermal inefficiencies of the various subsystems. Once the inefficiencies have been
identified, heuristic rules are applied to obtain the design structure. Although there is no
question that thermodynamic targets and heuristics provide a good understanding of the
processes, they do not take into account the complex interactions that take place
between the subsystems, nor do they guarantee optimality. Furthermore, the capital cost
hasn't been taken into account.
Townsend and Linhoff (1983) explained the concept of "appropriate" heat engine and
heat pump placement in process networks. They also introduced procedures for
preliminary design, involving heat engine, and heat pump equipment selection and
performance assessment. Dhole and Linnhoff (1992) introduced the concept of "Total
Site Profiles" for the analysis of an entire total site. They used thermodynamic insights
to graphically represent a total site on a Carnot factor-enthalpy diagram. In order to
target the Minimum Cost of Energy Requirements (MCER), Marechal and Kalitventzeff
(1996) proposed a concept of integrated composite curves for the integration of utility
systems.
In order to address the problem of selecting the pressures of the steam mains, Morton
and Linnhoff (1984) proposed the use of Grand Composite Curves. By using total site
analysis method, Raissi (1994) studied the optimum placement of steam levels and
identified the two extreme cases of the Minimum Fuel Requirement (MFR) and the
Minimum Utilities Cost (MUC) for single operation scenarios. The target model of
steam turbines in this work cannot account for the efficiency variation with load,
operating conditions and capacity, and the exhaustive enumeration algorithm was used
to find the optimal solution. Mavromatis and Kokossis (1998a) proposed a new Turbine
Hardware Model (THM) for steam turbines and an exhaustive enumeration method to
search for the optimal levels. The cases they studied were only based on the shaft-work
target. However, it is very difficult to get the optimal levels in large problems by using
exhaustive enumeration methods because of the huge computation requirement. In
addition, neither Raissi nor Mavromatis has taken into account the boiler target model to
calculate the fuel cost in order to determine the minimum utility cost. Finally,
systematic methods for generating potential steam levels have not been accounted for.
Chapter 2
site utility systems. Most of the publications dealing with the synthesis and design of
utility systems have focused on fixed utility demands or specified header pressure
levels. Nishio (1977) appears to be the first to consider the problem of selecting optimal
header pressure levels and presented a direct search approach coupled with
simultaneous solution of the balance equations.
proposed an MILP method for the structural and parametrical optimisation of utility
systems under fixed steam and power demands. The work was subsequently extended to
account for the synthesis and design of chemical processes (Papoulias and Grossmann
1983c). Iyer and Grossmann (1998) have presented a multi-period MILP approach for
the synthesis of utility systems operating under multiple periods. However, steam level
optimisation is not considered in this work. Petroulas and Reklaitis (1984) used a
dynamic programming method to optimise the steam conditions as continuous variables
and a linear programming method for the optimum allocation of drivers with the
common objective of minimising the real work loss. A non-linear programming strategy
was applied by Colmenares and Seider (1989) for the design of utility systems
integrated with the chemical process. The interaction of steam level selection and steam
demand for each level is not accounted for in this work, and the steam raised by
chemical processes is not considered. A simulated annealing algorithm has been used by
Maia and Qassim (1997) for the synthesis of utility systems with variable utility
demands. Most recently, Mavromatis and Kokossis (1998b) have presented an MILP
approach for the optimal design of steam turbine networks. This approach is only
limited to the back-pressure steam turbine network design. An MINLP model for the
synthesis and design of utility plants has been presented by Bruno et al. (1998).
Wilkendorf et al. (1998) also proposed an MINLP model for the synthesis of complete
utility systems. In practice, however, there are several drawbacks in the approach. If all
the candidate options are included in the superstructure, the number of candidate
structures should be enormous and the size of the problem would become too large to be
handled even for moderate problems. The consideration of multiple operation scenarios
results in a further increase of the design options to an extent. Secondly, the formulation
of utility systems is inherently non-linear with respect to the efficiencies for the units,
which gives rise to complex models. Therefore it is essential to find a systematic
Chapter 2
methodology which can build a superstructure including all the promising alternatives
without being too large, and to develop new modelling methods for the units.
Above all, we need a systematic methodology for the synthesis and design of site utility
systems, capable of considering the realistic efficiency trends of units, operational
variations and the interaction between the site utility system and site processes. One of
the objectives of this thesis is to present a systematic optimisation approach for the
synthesis and design of total site utility systems under operational variations.
2.3
Previous
approaches
to
the
debottlenecking,
planning
and
A number of approaches have been reported for the debottlenecking and planning of site
utility systems. Iyer and Grossmann (1998) recently addressed the synthesis and
operational planning problem for utility systems that they formulated as an MILP
problem. A recent survey can be found in Iyer and Grossmann (1997) in which the
operational planning problem for utility systems is formulated as an MILP program.
The optimisation of steam production network under uncertainty has been reported by
Papalexandri and Pistikopoulos (1996). Nath and Holliday (1985) have proposed an
MILP model which can be used for the long term planning of plant utility systems. A
multi-period utility system model has been presented by Hui and Natori (1996) in which
an MILP model has been used to find out the best combination of new equipment to be
added. Yokoyama and Ito (1996) have proposed an MILP model for an operational
planning problem for a cogeneration system under a complex utility rate structure. A
planning method can be found in Ito et al. (1990) in which the operational policy of
each piece of constituent equipment has been decided so as to minimise the operational
cost. Makwana (1997) proposed a debottlenecking method for utility systems by using
top level analysis. The trend of assuming constant efficiencies for the gas turbines,
steam turbines and boilers appears to be a limiting assumption in these developments.
Furthermore, no systematic methodology has been addressed to target and scope for the
debottlenecking options for the utility systems under operational variations. An
alternative approach is reported by Yokoyama et al. (1994) in which an NLP+MILP
8
Chapter 2
model is used. The shortcoming is that it can only optimise the fixed structure and there
is only one steam header in the approach. The approach is often limited in the size of
problems that they can handle. Olsbu et ai. (1988) proposed an MINLP model for the
design and planning of power systems which accounts for variable production profiles
and the availability of the system over a given time horizon.
2.4
Dhole and Linhoff (1993) introduced a graphical representation of all process heat
sources and heat sinks in the site, that need to be connected to the utility system, termed
the "Total Site Profiles" (TSP) shown in Figure 2-1. It is used in this paper to represent
the chemical processes and extract the information required for the analysis of site
utility systems. By integrating the utility system, especially the steam system with the
total site profiles, the optimal steam level operating conditions and loads can be
9
Chapter 2
detennined to minimise the utility cost, or to minimise the fuel requirement. It should be
noted that in the total site profiles, the steam demand at each steam level is expressed as
heat load, namely, the amount of heat can be used for heating at the saturation
temperature. The Total Site Profiles provide site-wide targets for fuel and co-generation.
The targets can be used for screening possible design options of site utility systems.
HP
site heat
In order to understand the interactions between fuel demand, heat recovery and cogeneration, Raissi (1994) proposed "Site Composite Curves" (SCC). These are
constructed by shifting the Total Site Profiles towards each other (Figure 2-2).
HP
cw
10
Chapter 2
The Site Composite Curves provide targets for fuel and co-generation, in a similar way
to the original Total Site Profiles. More importantly, the Site Composite Curves can
visually represent the steam flow in the utility system in addition to heat flow between
processes and utilities.
Raissi (1994) discussed the trade-offs between the fuel requirement and the
cogeneration potential when the steam levels under selection for one operation scenario.
Two extreme cases of Minimum Fuel Requirement (MFR) as shown in Figure 2-3a and
Minimum Utilities Cost (MUC) as shown in Figure 2-3b were studied.
Increased
fuel
VHP
Cogeneration
MP
CW
Increased
cooling utility
a. MFR
b.MUC
Figure 2-3: The Site Composite Curves for Minimum Fuel Requirement (MFR) and Minimum
The costs of fuel and power determine the optimal steam levels for the MUC case. The
work boiler efficiency and the steam turbine efficiency are assumed constant. For the
utility system operating under operational variations, the steam load raised by VHP
boilers and steam amount passing steam turbines should not be constant. As the steam
raised by boilers varies for different steam level selection, as well as for different
operation scenarios, it requires a model for the boiler that is able to account for the
variation of steam boiler efficiency with the load and the size, as well as the operating
conditions. The shaftwork-targeting model of steam turbine should account for
11
Chapter 2
efficiency variation with load, capacity and operating conditions as well . If the models
that fail to address these effects are used to select the steam levels, the selection of
steam levels will not be the optimum.
VHP
HP
MP
IP
Figure 2-4: The construction of the Site Utility Grand Composite Curve
In order to provide the net steam balance for each steam header of site utility systems,
Raissi (1994) proposed a Site Utility Grand Composite Curve (SUGCC) (Figure 2-4) by
plotting the steam generation load towards the right and the steam use load towards to
the left for each steam level. The site-wide targets for fuel and co-generation can also be
identified by SUGCC in a similar way by TSP and SCC. Figure 2-4 illustrates steam
turbines placed on the Site Utility Grand Composite Curve. The characteristics of steam
turbines are easily identified.
12
Chapter 3
Hardware Models
Chapter 3
Hardware Models
3.1
Introduction
In the synthesis and design of the total site utility systems under operational variations,
the loads and operating conditions of the units vary for different operation scenarios.
Their efficiencies vary and the unit capacities influence the efficiencies. Most of the
publications dealing with the synthesis and design of site utility systems simplify the
problem with linear mass/energy balances whereby the units are assumed of constant
efficiencies. A new approach for the targeting models of the units is proposed in this
chapter. The targeting models account for the variation of efficiency with load and
capacity, as well as changes in the operating conditions.
In the first part of this chapter a new Boiler Hardware Model (BHM) is developed to
describe the performance of boilers. The BHM exploits the basic thermodynamic
principles relating to the operation of boilers. Next, the development of a new Gas
Turbine Hardware Model (GTHM) is proposed which is based on thermodynamic
principles, hardware data and engineering knowledge. The Turbine Hardware Model
(THM) proposed by Mavromatis and Kokossis (1998a) is introduced to describe the
performance of the back-pressure steam turbines. Finally, the THM is extended towards
a Condensing Turbine Hardware Model (CTHM) to describe the operation of
condensing turbines.
13
Chapter 3
Hardware Models
Most of the conventional applications dealing with the synthesis and design of site
utility systems have either simply focused on the balance calculations associated with
boilers or assumed the constant boiler efficiency. In order to address the problems of the
optimisation of total site utility systems under operational variations, a new boiler
model is proposed, termed the Boiler Hardware Model (BRM). The model makes use of
basic thermodynamic principles related to the operation of steam boilers. As a result, the
model is capable of accounting for the efficiency trends of realistic steam boilers.
3.2.1
Model Development
The BRM relies on the principle of the calculation of boiler efficiency. Figure 3-1
shows us the relation between steam load (Qsteam), heat losses (Qloss) and fuel
requirement (Qfuel). The numerical expression for this relation is:
(3-1)
(3-2)
where:
hI : the enthalpy of boiler inlet water
h2 : the enthalpy of steam raised by boiler
M: the steam load raised by boiler
14
Chapter 3
Hardware Models
T,
Steam
T.
H
~'
QIuoI
(3-3)
where:
Tin
Tout sat :
15
Chapter 3
Hardware Models
Cp
the specific heat of saturation water between Tin and Tout sat
~Tsat
: the specific heat load of the steam, namely, the heat that can be used for
(3-4)
It follows that:
(3-5)
Qjue/
(3-6)
35~----------------------------~
Q[oss
QSleam
30
(%) 25
20
15+-~--~--~~--~--~~--~--~~
o 10 20 30 40 50 60 70 80 90 100
Figure 3-3: Typical heat loss plot for boilers (Pattison and Sharma, 1980)
In order to define the boiler model, the Qloss needs to be known. The Qloss originates
mainly from two parts: the external boiler surface losses and the flue gas losses. Typical
16
Chapter 3
Hardware Models
data on the total heat loss can be found in the form of plot over a range of loads as
shown in Figure 3-3 (pattison and Sharma, 1980). The plot shows the heat loss percent
QIOSS
Q steam
~.
Mmax
where:
the steam heat load
Qsteam
M max
On the basis of the data as shown in Figure 3-3 (Passison and Sharma, 1980), the
following equation is regressed within the range between 10% and =100%
Qloss
~=a+b~,
Qsteam
10% 5 - max
- 5100%
M
M max
Mmax
of~:
max
M
(3-7)
Where a and b are regression parameters. The regression over the data provided by
Pattison and Sharma (1980) yields a=0.0126 and b=0.2156 for the plot in Figure 3-3. It
should be noted that both
~
M~
and
Qloss
Qsteam
~.
The
M~
expression estimates the heat loss within 2% throughout the operating range.
(3-8)
As a result of this expression, the operation of a boiler can be fully defined, once its size
max
17
Chapter 3
Hardware Models
The Equation (3-9) constitutes the core of the new boiler model. It relates the fuel
requirement of the boiler to the boiler size, the steam load and its operating conditions.
Given a set of expected conditions, the performance of the boiler is a function of its
size. Most notably, while the non-linear variation of the efficiency is accounted for, the
relation of the fuel requirement to the steam load is linear. The predicted boiler
efficiency from Equations (3-5) and (3-9) is:
MOlaX
11 = ---=-=----M
(3-10)
(1+b)--+a
M max
Equation (3-10) accounts for the variation of efficiency with load and capacity. If we
define the efficiency 11b as the ratio of the heat load of steam (the heat that can be used
for heating at the saturation temperature) to the heat of fuel, gives:
q MOlaX
11b=------=~----M-----
(3-11)
(CpllTsat +q)1+b)--+a)
MOlaX
M~X
for
different working conditions, the attributes of the new model are best revealed. As seen
in Figure 3-4, the new model accounts for the variation of efficiency Tlb with the effect
of load, capacity and operating conditions, as would be required by a realistic model. In
the remaining of the dissertation, the model will extensively be used for the
optimisation and analysis of total site utility systems in order to minimise the total
utility cost.
18
Chapter 3
Hardware Models
80
75
llb (%)
7f:-)(
P=88bar,
q=2005kj/kg
)(
70
~ Tin=144,
65
P=88bar,
q=1383kj.kg
60
Tin=144,
P=48bar,
q=2005kj/kg
55
50
45
40
0
20
60
40
80
100
Tin: 144,
P=48bar,
q=1651kj/kg
-(%)
M max
Figure 3-4: The BHM accounts for the effect of boiler size, load and operating conditions on
efficiency
3.2.2
Summary
The Boiler Hardware Model is based on basic thermodynamic principles and the typical
heat loss plot proposed by Pattisson and Sharma (1980). The thermodynamic principles
describe the operation of boilers and the relation between fuel requirement and steam
generation. The plot of heat loss against load represents typical operation of boilers.
Againist the published data (Pattisson and Sharma (1980 , the BHM provides estimates
within 2% error. In addition to size, load and operating conditions, the heat loss of a
boiler depends on a series of other factors, such as its type, technology and age. Hence,
it is not realistic to expect all boilers to fit in the same curve. To get plots for the
particular classes of boilers , it is sensible one revises the regression parameters to
accurately reflect on the particular case.
19
Chapter 3
Hardware Models
The Turbine Hardware Model (THM) presented by Mavromatis and Kokossis (1998a)
is only valid for back-pressure steam turbines. In order to address the optimisation and
analysis problem of total site utility systems, the THM is extended towards a
generalised Condensing Turbine Hardware Model (CTHM).
The THM is based on the Willans Line which depicts steam turbine performance as
shown in Figure 3-5. Where W max is maximum power output and n is incremental shaftwork generation against a unit of steam change. The THM relies on the size of the
steam turbine and describes the operation of the unit over its entire range. The turbine
shaftwork is given by (Mavromatis and Kokossis, 1998a):
6 1 (AUIIis
lJ
W =-5B
A)(M --M
1 max )
max
(3-12)
max
llis
_6~(1_
A
)1_1M
- 5 B
MlisM max ( 6 M )
where:
W
A, B
Ml is
M max
llis
20
(3-13)
Chapter 3
Hardware Models
Throttle
Flow
(M)
Willans line
wmax
Shaftwork (W)
The THM accounts for a maximum steam flowrate through a simple turbine, an
isentropic enthalpy change between the turbine inlet and outlet, and a set of regression
parameters. The isentropic enthalpy change and the regression parameters can be
calculated using saturated temperatures of the steam at the turbine inlet and outlet (see
Appendix A). The maximum steam flowrate is derived from the heat demands and
specific heat load at the turbine exhaust. The THM takes into account changes in
isentropic efficiency with the turbine load and working conditions.
It is assumed that the Willans Line applies to condensing turbines (Church, 1950). The
Condensing Turbine Hardware Model (CTHM) pursues similar calculations for the
shaftwork:
(3-14)
where:
W
A C , BC
Mi i.,
21
Hardware Models
Chapter 3
M m3x
Ae
1]/.1
1M
=~_1 (1 =:----)(1- - - - )
5 Be
~isMmax
max
(3-15)
The differences between the CTHM and theTHM relate to the regression parameters AC
and BC Typical data on the efficiency of condensing turbines are found from plots of
the maximum efficiency, as shown in Figure 3-6 (Peterson and Mann, 1985). The
regression parameters are derived from the data on Figure 3-6. By definition, the
maximum efficiency is:
rh.max
E max
= MI . M max
(3-16)
/.1
hence:
max
~i.\ Mmax
E
= __
(3-17)
77 ;s ,max
0.85 . . , . - - - - - - - - - - - - - - - - - - -
0.75
11 is , max
____ 28bar
0.65
41bar
0.55
62bar
X
0.45
83bar
+---------,,-------,------~
0.1
10
100
Emax(MW)
Figure 3-6: Typical data on the efficiency of condensing turbines
22
Chapter 3
Hardware Models
The curves in Figure 3-6 are represented by an expression of the following equation for
each inlet pressure:
max
Ml isM max
E
=__
=A + B
C
E max
(3-18)
l]is.max
The regression parameters AC and Be are derived as functions of the inlet pressure or the
respective saturation temperature T sat (see Appendix B):
(3-19)
Be
=be1 +be2Tsar
(3-20)
3.3.3 Summary
The Condensing Turbine Hardware Model follows similar principles with the THM. It
has the capacity to accurately consider the effect of turbine size, load and operating
conditions on the efficiency of typical condensing turbines.
3.4
Most of the previous applications dealing with the synthesis and design of site utility
systems assume constant efficiency for the gas turbine and disregard the effect of
capacity and partial loading. With a purpose to address the problems of the synthesis
23
Hardware Models
Chapter 3
and design of total utility systems under operational variations, a Gas Turbine Hardware
Model (GTHM) is proposed here. The GTHM makes use of the basic thermodynamic
principles, exploits available information for gas turbine equipment, and accounts for a
realistic model to use in synthesis and conceptual calculations.
3.4.1
Model Development
Figure 3-7 illustrates us the basic structure of a gas turbine cycle. The power output
(W), the total power loss of the gas turbine cycle (W IOSS ), the fuel flowrate (F\ the fuel
temperature (Tf) , the air flowrate ~), the air temperature (Ta) and the outlet
temperature rout are related by:
(3-21)
where,
Cpa
Cpf
cpg
MIf
W10s
Ft, Tt
r6
Combustor
Compressor
Turbine
24
Chapter 3
If we define
Hardware Models
Fa
(3-22)
Let us define
(3-23)
(3-24)
The parameters .1h and W 10ss generally depend on the size of the gas turbine. For a gas
turbine, the overall electrical efficiency increases with size. Typical data on the
electrical efficiency against the turbine size are shown in Figure 3-8 (Marechal and
Kalitventzeff, 1998). The curve explores the impact of the turbine size on the basis of a
full-load efficiency; It does not provide information on the resulting efficiency from a
partly loaded unit. The curve is regressed in the form:
(3-25)
25
Hardware Models
Chapter 3
o.~~----------------------------------------~
0.4
l1 emax
0.35
0.3
0.25
0.2
+-----..,.------r-----.----r-----.---~--____"
10
30
20
40
50
60
70
Figure 3-8: Typical data on the electrical efficiency of gas turbines as a function of size
w
11e
Ml Ff
(3-26)
It follows that:
wmax
11e.max
= Ml
F f .rnax
. (3-27)
It is assumed that:
(3-28)
26
Chapter 3
Hardware Models
According to Equations (3-24) and (3-28) the maximum power output is equal to:
(3-29)
(3-30)
By combining Equations (3-28), (3-29) and (3-30), the following expressions for the
parameters of the model are derived:
1
ilh =(1+n)-(Ml
j
g
Ag
(3-31)
F f ,max )
and
(3-32)
Substituting Equations (3-31) and (3-32) into the Equation (3-24) gives:
(3-33)
27
Tout
is given
Chapter 3
Hardware Models
Equations (3-30) and (3-33) relate the power output of the gas turbine to the size and the
load of the turbine, as well as its operating conditions. Equation (3-34) relates the outlet
temperature of the gas turbine to the size of the turbine and its operating conditions,
expressed through
r,
performance of the turbine can be estimated merely on the basis of its size. While the
non-linear variation of the efficiency is accounted for, the relation of the power output
to the load is linear. The predicted electrical efficiency can be deri ved from Equations
(3-26) and (3-33):
1
rtl' =-(1-
B8
A8
Ml fF
)1+n)-n
Ff. max
,max
(3-35)
f)
0.5
l1e
40MW
0.4
60MW
0.3
0.2
0.1
0
0
10
20
30
40
50
60
W(MW)
Figure 39: The GTHM accounts for the effect of gas turbine size and load on efficiency
By plotting the predicted efficiency against the fuel load Ff at various turbine size Ff,max,
Figure 3-9 shows the variation of the electrical efficiency with load and the effect of the
turbine size on the turbine efficiency as it would be expected by a reliable and realistic
model.
28
Chapter 3
Hardware Models
3.4.2 Summary
3.5 Conclusions
A set of hardware models CTHM, BHM and GTHM are proposed. The use of the
models for the analysis and optimisation of site utility systems will be explained in the
following sections. The models combine thermodynamic principles, engineering
knowledge and performance data for condensing turbines, boilers and gas turbines.
They enable for an accurate prediction of unit efficiencies and embody the efficiency
trends of realistic units in terms of their variation with capacity, load and operating
conditions. These effects are accounted for in a simple and linear fashion, the
importance of which will be illustrated in the subsequent chapters.
The hardware models CTHM, BHM and GTHM provide results of good accuracy, by
considering the dependence of the efficiency on the capacity and operating conditions.
In view of operational variations, the effect of the part-load operation is well accounted
for all these models.
29
Chapter 3
Hardware Models
With respect to the analysis and optimisation problem, the CTHM, BHM and the
GTHM models provide the basis for modelling the condensing turbines, boilers and gas
turbines respectively in a manner that the efficiency trade-offs of the various design
alternatives can be considered. The linear relations of the power output to the steam
load of condensing turbines, the steam load to the fuel requirement of boilers and the
power output to the fuel requirement of gas turbines are essential for simple
formulations for the optimisation of site utility systems.
30
Chapter 4
Chapter 4
Steam Level Optimisation
4.1
Introduction
In the design of process plants, the perfonnance of site utility systems directly
influences the operation of the plants, hence, the optimal design of site utility systems
often leads to significant savings. As there are strong interactions between the site
utility system and site processes, the design problem needs to consider the integration of
the site processes with the site utility system. The design usually involves two major
tasks:
1. the selection of steam levels with respect to their pressure or saturation
temperatures;
2. the development of a configuration for the operating units (ie, boilers, steam
turbines, gas turbines and the auxiliary units).
The placement of steam levels has a significant bearing on the utility demands, the
cogeneration potential of the processes and the operating conditions of the steam
turbines. The specific problem addressed in this approach assumes a given structure for
the chemical processes. Different operation scenarios are also given for the chemical
processes along with forecasts for prices of utilities over a finite number of time
periods. The different operation scenarios of the total site are described by the sets of
Total Site Profiles (TSP) shown in Figure 4-1. The process heat sources reject heat by
31
Chapter 4
raising steam at different levels; the process heat sinks absorb heat also at different
levels. The total amount of steam raised by the process heat sources doesn't usually
match the amount required by the process heat sinks. Because of thermodynamic
constraints and heat transfer constraints, auxiliary cooling and heating are required.
These are available by the cold utility and the VHP steam raised by the boiler. The
timing of demands changes the profiles of heat sources and sinks over time and the
duration of each time period is usually different.
The objective of this chapter is to find the optimum locations for the steam levels
considering the total site. Single operations will be discussed first and subsequently be
generalised for multiple operation scenarios. As discussed by Petroulas and Reklaitis
(1984), if auxiliary fuel-boilers are required, they should operate at the highest-pressure
level. The optimisation problem then needs to determine the temperature of the VHP
steam, the saturation temperature (or pressure) of each steam level, the auxiliary boiler
duty, the cooling utility demand and the shaft-work produced by the steam turbine
network for each expansion zone. By using the THM and BHM models, one is able to
target the overall fuel requirement, the cooling utility demand and the co-generation
potential. In order to obtain the optimal solution for minimum utility cost, we need to
identify the correct compromise between heat recovery and co-generation.
Scenario A
Scenario B
HP
MP
IP
LP
Figure 4-1: Different operation scenarios can be represented by sets of total site profiles
32
Chapter 4
4.2
The Strategy
Considering the drawbacks of the conventional techniques (see section 2.2) for the
selection of steam levels, the following strategy is proposed to determine the optimal
steam levels. The new strategy is shown in Figure 4-2. The strategy comprises two
stages:
1. The total site analysis is used to postulate a new transhipment network that describes
the heat integration of the steam levels with the process heat sources and heat sinks.
2. An optimisation model is developed based on the transhipment network and the
THM and BHM models. The model optimises by minimising the total annual cost
and is formulated as a multi-period Mixed Integer Linear Programming (Mll..P)
model.
Gal
G;
Analy~)-----t~~
and BHM
~r
m~)-----1~~
Optimisation
The transportation model determines the optimum transfer of commodities from sources
to destinations. The transhipment model has been widely used in the Operation
Research (Garginkel and Nemhauser, 1972) that deals with the optimum allocation of
resources and represents a variation of the transportation problem. Papoulias and
Grossmann (l983b) proposed a transhipment model for the synthesis of heat exchanger
33
Chapter 4
networks. In this work, a transhipment network representation is developed for the total
site in order to get the optimal steam levels.
The total site heat flows can be represented by total site profiles. As seen in Figure 4-3,
heat is regarded as a commodity to ship from process heat sources to steam levels and
from steam levels to process heat sinks through temperature intervals. These intervals
account for thermodynamic constraints in the transfer of heat. In particular the second
law of thermodynamics requires that heat flows only from higher to lower temperatures,
and therefore these thermodynamic constraints have to be accounted for in the network
model. This is accomplished by partitioning the entire temperature range into
temperature intervals. For the total site profiles the interval temperatures are the
temperatures of turning points (critical points) of each heat source and heat sink. These
are all candidate locations of the optimum steam levels. The selected temperatures are
listed in descending order. The optimal steam levels are selected from all potential
steam levels denoted by their saturation temperatures.
As shown in Figure 4-3, the points A,B,C,D,E,F... are the turning points. This
partitioning method guarantees the feasible heat transfer of heat in each interval, given
the minimum temperature approach L\T min. In this way as shown in Figure 4-3, the total
site heat flows are represented by the transhipment network. The network comprises
three cascades of temperature intervals:
The heat source cascade represents that heat flows from process heat sources to the
corresponding temperature interval, and then to the steam level in the same temperature
interval with residual going to the next lower temperature interval. For the heat sink
cascade, it can be considered that heat flows from steam level to the corresponding
temperature interval, and then to the process heat sinks in the same temperature interval
with residual going to the next lower temperature interval. The steam level cascade
represents that heat flows from process heat sources to the corresponding steam level,
34
Chapter 4
Steam LeveLOptimisation
and then to the process heat sinks in the same temperature interval with residual passing
a steam turbine to the next stearns level.
Heat source
cascade
Steam Levels
cascade
Heat sink
cascade
8 --
Cooling Utility
H
Figure 4-3: Transhipment network representation of the total site heat flow
It is assumed that the total number of stearn levels for the site is I. The levels are
labelled from the highest level (i=l) down to the lowest level (i=I). The temperature
range for each level is partitioned into J temperature intervals which are labelled from
the highest interval (j=I) down to the lowest interval (j=J). In this way, the entire
temperature range of the total site is partitioned into IxJ temperature intervals. The
intervals are labelled from the highest interval (i=l, j=l) down to the lowest interval
(i=I, j=J).
The heat flow pattern of the temperature intervals for stream level i can be illustrated as
shown in Figure 4-4. It is represented by the three heat cascades:
35
Chapter 4
Site heat
source
i, j=1
i, j=2
Heat souce
cascade
(]--.I
0--.1
~
JJ
H
H
n.
II
it
n:
~
~
i, j=J-1
Heat sink
cascade
Steam level
cascade
--------------
Site heat
sink
~
~
--------..
~
<l
i, j=J
Figure 4-4: Heat flow pattern of the temperature intervals for steam level i
36
Chapter 4
37
Chapter 4
4.4
Optimisation Model
Sets
IS = { i=I,2 .. .11 steam level}
I is the total number of steam levels.
1S = (j =1,2 ... 1 or jm=1,2 ... 1 I temperature intervals}
1 is the total number of temperature intervals for each steam level.
K = { k=I,2 ... Nk I operation scenarios}
Nk is the total number of operation scenarios.
V = { v=1,2 ... Nv I operating conditions of VHP steam }
Ny is the total number of candidate operating conditions of VHP steam.
Parameters
a,b
Ai,j, Bi,j: regression parameters of THM for the steam turbine using steam in interval
(i,j)
Ay,B y
38
Chapter 4
Ci,j,k
: total heat required by all process heat sinks in temperature interval (i, j) under
scenario k
CHk
: total heat required by all process heat sinks above temperature interval (1,1)
under scenario k
specific heat of saturation water between Tin and Tout sat
Cp
E1Si,j,jm: isentropic enthalpy change between the steam level in intervals (i, j) and the
steam level in interval (i+ 1, jm)
EISV v,j: isentropic enthalpy change between the VHP steam level at condition v and
the steam level in interval (1, j)
H
Hi,j,k
total heat provided by all process heat sources in temperature interval (i, j)
under scenario k
HLk
: total heat provided by all process heat sources below temperature interval (I, J)
under scenario k
Mimax
capacity of the steam turbine locating between the steam level i and the steam
level i+l
Mmax
MB max
qv
qi,j
Ti,j
lower bound temperature of interval(i, j). The intervals are labelled from the
highest interval (Tl,I) down to the lowest interval (TJ,J). The saturation temperature of
the steam level in interval (i, j) equals Ti,j
T kS
~Tsat
TVH
U/
U kc
UkP
W kd
39
Chapter 4
Binary variables
Yi,j
variable to denote the steam level in interval (i,j) is selected (Yi,j=1) or not
(Yi,j=<
ytij,k
variable to denote the steam turbine using steam in interval (i,j) works under
: variable to denote the operating condition v of the VHP steam level is selected
: variable to denote the VHP steam turbine works at operating condition v under
Continuous variables
RHi,j,k : residual heat flowing out of interval (i,j) of the heat source cascade under
scenario k
RCij,k : residual heat flowing out of interval (i,j ) of the heat sink cascade under
scenario k
Di,j,k
: heat flowing out of interval (i,j) of the heat source cascade to the steam level
: heat flowing into the steam level in interval (i,j) from a steam turbine under
scenario k
QSTv,j,k: heat flowing into the steam level in interval (l,j) from the VHP steam level at
operating condition v under scenario k
QSv,k
: heat flowing out of the VHP steam level at operating condition v to the VHP
: heat flowing out of the steam level in interval (i,j) passing a steam turbine to
: heat flowing out of the steam level in interval (i,j) to the heat sink cascade
under scenario k
FTj,j,jm,k: heat flowing out of the steam level in interval (i,j) passing a steam turbine to
the steam level in interval (i+l,jm) under scenario k
40
Chapter 4
Wj,k
: power output of the steam turbine locating between steam levels i and i+ 1
under scenario k
WVHk: power output of the VHP steam turbine under scenario k
: heat flowing into interval (1,1) of the heat sink cascade from the VHP steam
Cf :
cp,tot
4.4.1
Model Formulation
The previous sections explained the transhipment representation of the total site (as
shown in Figure 4-3) and the temperature intervals of the system (Figure 4-4). The heat
flows across the intervals of the heat source cascade, the steam level cascade and the
heat sink cascade can then be represented by Figures 4-5, 4-6 and 4-7 respectively.
RH,j-l.k
RHi-I.J.k
RH'j.k
Figure 4-5: Heat flow pattern in each temperature interval of the heat source cascade
41
Chapter 4
S'J.k
F'J.k
Figure 46: Heat flow pattern in each temperature interval of the steam level cascade
RC..I).k
E.J.k
Figure 47: Heat flow pattern in each temperature interval of the heat sink cascade
Given the sets, the parameters and the variables introduced above, the mathematical
model includes:
Heat balances for each temperature interval (i,j) in the process heat source cascade:
H i.J.k + RHH.J.k
=Di.J.k + RHi.J,k'
i E IS, J'
i :t 1, j
42
2, k E K
=1, k E
=1, k E
(4-1)
(4-2)
(4-3)
Chapter 4
Heat balances for each temperature interval (i,j) in the steam level cascade:
D"k
+S""k
=E."k
+F.I,j,Ok'
I,j,
I,j,
I,j,
iE IS,jE JS,kE K
(4-4)
Heat balances for each temperature interval (i,j) in the process heat sink cascade:
E;,j,k + RCi,j_I,k
i E IS, j ~ 2, k E K
i::l: 1, j = 1, k E K
(4-5)
(4-6)
(4-7)
Heat balances of the process heat sink cascade above the temperature interval (1,1):
CR k +QVk = LQCV,k' kE K
(4-8)
veV
For each steam level exactly one operating condition can be selected. Therefore,
~y"I,j" = 1,
~
jeJS
iE IS
(4-9)
(4-10)
If a steam level in interval (i, j) is not selected, the steam turbine cannot use steam from
the steam level, the following logical models apply:
Lyti,j,k ~ Yi,j'
keK
43
iE IS,jE JS
(4-11)
Chapter 4
To ensure that the input and output flowrates of each level are eliminated if the
corresponding level doesn't exist, the following inequalities must be included:
0,
ie IS, j e JS,ke K
(4-12)
:SO,
ie IS,je JS,ke K
(4-13)
0,
i e IS,j e JS,k e K
(4-14)
F;,j,k - UYi,j :S 0,
i e IS, j e JS, k e K
(4-15)
Si,j,k -UYi,j
Ei,j,k
~UYi,j
Dj,j,k -UYi,j
A single steam turbine is assumed for every expansion zone. The shaft-work of the
steam turbines employs the THM:
Wi,k
JEJS "nEJS
__ M.max t ..
FTI,j,jm,
. . k 6 I Y I,j,k
q',j
J]
'
ie IS,k e K
(4-16)
~
.. k = FI,j,
.. k'
L.J FTI,j,jm,
(4-17)
i ~ 2, J'm E JS, k E K
(4-18)
jmeJS
~
.. k = S.I,jm,
. k'
L.J IT1-1,j,jm,
jEJS
kE K
(4-19)
veV
44
Chapter 4
Lxbvk ~ Xv'
VE
(4-20)
V,k E K
(4-21)
keK
MB vk -Uxbvk
0,
vE
The heat balance for the VHP steam level is fonnulated as follows:
(4-22)
where:
LQSTv.j k = QSv.k'
VE
V,kE K
(4-24)
VE
(4-25)
jeJS
Lxt vk ~ xv'
keK
vE
V,kE K
(4-26)
(4-27)
lEV
c f = LU! Q!T/ H
(4-28)
keK
As the heat of cooling utility required by the process heat sources is RI,J.k + HLk under
scenario k, the annual cost of cooling utility is:
45
Chapter 4
CC = 'LU%(R1,J,k + HLk)TkS H
kEK
(4-29)
ct = LU:Wi,k T/ H,
iE IS
(4-30)
keK
= 'LU!WVHkT/H
(4-31)
= 'LU[W/T/ H - 'LC/-CVH
(4-32)
CVH
kEK
cp,tot
kEK
iE/S
The objective function used minimises the annual utility cost that includes the cost of
fuel and cooling utilities, as well as the cost of power. It is given by:
(4-33)
It should be noted that the proposed MILP model can also be used to find the optimal
steam levels for total site utility systems in order to minimise the fuel requirement. This
can be accomplished by replacing the above objective function by:
minMFR=C f
(4-34)
Normally, these two objectives define different steam levels and this can be illustrated
by Case Study 1 which will be introduced next.
46
Chapter 4
4.4.2
The above fonnulation consists of linear constraints of continuous and integer variables.
It comprises a multi-period Mixed-Integer Linear Programming (MILP) model. The
problem of synthesising a total site utility system given by the objective function and
the proposed set of constraints in Section 4.4.1, corresponds to a model whose
development requiring the following infonnation:
(a) Data on the total site profiles for each operation scenario.
(b) Specific heat load of VHP steam for each working condition; it is assumed that the
specific heat load of steam expanded through a turbine remains approximately
constant for all exhaust pressure values. The assumption is based on the observation
by Mavromatis and Kokossis (1998a), Raissi (1994) and Salisbury (1942).
(c) Cost correlations for the available utilities.
Transhipment superstructure
Optimisation
Optimal solution
The steps are presented in Figure 4-8. The total site profiles for each operation scenario
are generated from the procedure proposed by Dhole and Linhoff (1993). The
temperature range is partitioned following the propositions of Section 4.3. The
transhipment representation is developed as explained in Section 4.3. The optimisation
determines the optimal structure of the steam levels and the operating condition of the
47
Chapter 4
VHP steam. It also provides the steam load and power output of each steam turbine, the
heat duty of the boiler and the cooling utility demand for each operation scenario.
The methodology is illustrated with two different cases. Case 1 is a steam level
optimisation problem under a single operation scenario. The second case considers
multiple scenarios. The cost data of the available utilities are given in Table 4-1.
Cooling Water
Tout-Tin=20C
LfrV: 13856kWhlTon
Cost: O.0185$ffon
Cost: 223$ffon
Electricity
Cost: O.12$/kWh
4.5.1
The total site profiles are given in Figure 4-9. The steam system comprises a boiler, the
four steam levels shown on the TSPs, and a single cooling utility. It is assumed that a
single steam turbine is placed at each steam expansion zone. The very high-pressure
(VHP) steam is raised in the boiler house at 500C and 90 bar (Tsat =302C). The
temperature of inlet water is 245C. The specific heat load of the VHP steam is
0.557MWhlt at 500C and 90bar. The HP, MP and LP steam are raised against process
heat sources and consumed by process heat sinks. On the basis of the total site profiles,
the objective is to find the optimal placement of the intermediate steam levels HP, MP
and LP in order to minimise the utility cost and minimise the fuel requirement
respecti vel y.
48
Chapter 4
ToCi
Fuel _ _~
300
VHP
----~--------~~
HP
MP
LP
Cooling Utility
- - - '_ _ _ l
H(MW)
. H(MW)
Figure 4-9: The total site profiles for the site of Case 1
Based on the total site profiles and the temperature partition rule, the candidate
saturation temperatures for each level are obtained as shown in Table 4-2.
HPCOC)
MPeC)
LPCOC)
275
210
130
260
200
120
250
190
110
Table 4-2: Candidate saturation temperatures for each steam level of Case 1
The total heat provided by the process heat sources and total heat required by the
process heat sinks for each temperature interval (i,j) are obtained by targeting the heat
loads with the help of Figure 4-10 as shown in Table 4-3 and Table 4-4. The heat
requirement of the process sinks above the temperature interval (1,1) is 3MW. The
surplus heat of the process heat sources below the temperature interval (3,3) is 9MW.
49
Chapter 4
___
=====;::.;;.:;:::::====-==-
-+
=--t~.
-fic>==-------.-.-..
I.
H(MW)
H(MW)
na
3r<1 interval
H(I,j) MW
1sl interval
1sl level
18
2D <1leve l
10
3r<1 level
interval
Table 43: Heat provided by the process heat sources in each temperature interval of Case 1
Da
3 r <1 interval
C(I,j) MW
1sl interval
1sl level
2nd level
37
3 rd level
interval
Table 44: Heat required by the process heat sinks in each temperature interval of Case I
After the partition of the temperature intervals, the transhipment representation is set up
on the basis of Figure 4-11.
The proposed MILP model is applied to minimise the total utility cost. The model is
developed using the general algebraic modelling system (GAMS) (Brooke, etc., 1992)
and the optimisation has been conducted by employing the OSL solver. The
50
Chapter 4
302
270
260
250
210
200
190
VHP
Fue:..:..I_~
..
-=:- - +-,......+-
. .. .
HP
+
........ .......
L-=::;:~+:t~::::;;~:~
.,
- ........ _.
.................
....
-=<.:........-.
.~
~-----
MP
--+-
130
LP
120
110
Cooling Water
The optimal steam levels for the Minimum Utility Cost (MUC) case are shown in
Figure 4-12. The optimisation determines the saturation temperature of each steam
level, the heat load balance of each steam level, the heat load of each steam turbine, the
power output of each steam turbine, the heat duty and fuel requirement of the boiler and
the cooling utility requirement. The optimal temperature levels are 260C for HP, 200C
for MP and 110C for LP. The total utility cost is 68.5 k$/year, out of which the fuel
cost is 6618.7 k$/year, the cooling utility cost is 63.1 k$/year and the power saving is
6613.3 k$/year. The site pinch is located at the MP level.
The same MILP model can be used to minimise the fuel requirement. The optimal
steam levels for the Minimum Fuel Requirement (MFR) case are shown in Figure 4-13.
The optimal steam levels are 270C for HP, 200C for MP and 130C for LP. Even
though the MFR case has the same total site profiles as the MUC case, their optimal
steam levels are apparently different. The fuel cost for the MFR case is 4280 k$/year
51
Chapter 4
which is less than the MUC case. However, the power output is zero and there
IS
no
power saving.
302
Fuel
VHP
47_2MW
270
260
250
210
200
HP
- - - - - -/--
MP
190
130
120
LP
110
Cooling Water
~ 22~
302
Fuel 30MW
270 ---------------------------------
l5-~----r"
------~ --------- ------....I.----'r:::---
VHP
19MW _
210
200
190
14MW
17MW
19MW
130
120
110
-- ..._------------_.._----------:=+ ----
--.-----.--.--.---... --.----.-.-
,,-
MP
OMW
16~
Cooling Water
HP
7~
52
LP
Chapter 4
4.5.2
The site of Case 1 is now considered with three additional scenarios, scenarios B, C and
D. The total site profiles for each scenario are shown in Figure 4-14. Each scenario
spans for a period equal to one-fourth of the year.
T oe
'. -=r-=
--
T oe
'.
Scenario B
Scenario A
LP
H(MW)
H(MW)
H(MW)
H(MW)
T oe
Scenario D
Scenario C
H (MW )
H(MW)
H(MW)
Based on the total site profiles and the temperature partition, four candidate saturation
temperatures are obtained for each level as shown in Table 4-5.
HP(OC)
MP(OC)
LPeC)
280
210
140
270
200
130
260
190
120
250
180
110
53
Chapter 4
For each temperature interval, the total heat provided by the process heat sources and
the total heat required by the process heat sinks are obtained by targeting the heat loads
and using Figure 4-14. Results are summarised in Tables 4-6, 4-7, 4-8 and 4-9, where
H(i,j) represents the heat provided by the heat sources and C(i,j) represents the heat
required by the heat sinks. The heat requirement of process heat sinks above
temperature interval (1,1) for each scenario is shown in Table 4-10. The surplus heat of
the process heat sources below temperature interval (3,3) for each scenario is shown in
Table 4-11.
H(i,j) MW
1st interval
2na interval
3ra interval
4tn interval
1st level
15
2nd level
10
1.5
1.5
C(i,j) MW
1st interval
2na interval
3ra interval
4 tn interval
1st level
2 level
37
1.5
3rd level
1.5
level
nd
H(i,j)MW
1st interval
2nd interval
3rd interval
4 tn interval
1st level
15
2nd level
10
1.5
3rd level
1.5
C(i,j) MW
1st interval
2nd interval
3 interval
4tn interval
1st level
2nd level
12
1.5
3rd level
1.5
31
rd
54
Chapter 4
H(i,j)MW
1st interval
2nd interval
3rd interval
1st level
10
ha
rd
3 level
C(i,j) MW
1st interval
2 interval
3rd interval
1st level
2nd level
15
3ra level
13
2 level
nd
4th
4th
interval
interval
H(i,j) MW
1st interval
2na interval
3 interval
1st level
2na level
10
3ra level
C(i,j) MW
1st interval
2 interval
3 interval
1st level
2nd level
10
3ra level
10
nd
ra
ra
4th
4th
interval
interval
Scenario A
Scenario B
Scenario C
Scenario D
O(MW)
O(MW)
3(MW)
2MW
Table 4-10: Heat required of process sinks above temperature interval (1,1) for Case 2
Scenario A
Scenario B
Scenario C
Scenario D
9(MW)
9(MW)
5(MW)
7MW
Table 4-11: Surplus heat of process sources below temperature interval (1,1) for Case 2
55
Chapter 4
302
'----_.
VHP
Fuel
280- - - - - - - - - .
27()-- - .- +---=
260-'-'0
250 - -- - -.------ -.~-- -- ... -.-.. - .. + .. I -~----I~-+-~
HP
MP
.,
~
5
_OJ
==t~==t:::::=:::I==5t:::t:t:~:::::;;;:::::~
::::-:. : : :;=:= .=:::==::=:::::=:::::::~...-. :=:~~==::::====
....L.--+-L--_~--
- + -.---- .~ -.-. -
LP
Cooling Water
The Mll...P model is applied to minimise the total annual utility cost. The mathematical
model in GAMS involves 513 continuous variables, 88 binary variables and 628
constraints. The model employs the OSL solver.
The optimal saturation temperatures are 260C for HP, 200C for MP and 140C for
LP. Figure 4-16 shows the Site Utility Grand Composite Curves for the four scenarios.
The total utility cost is 150.6 k$/year, out of which the fuel cost is 5252.8 k$/year, the
cooling utility cost is 126.1 k$/year and the power saving is 5228.3 k$/year.
4.5.3
Summary
The results of Table 4-12 illustrate different optimal steam levels between single and
multiple scenarios. Figure 4-16 shows that the site pinch of scenario A is at the MP. The
site pinches of the other three scenarios are at the LP. The illustration explains that:
The optimal levels in considering isolated scenarios are generally different from the
case more scenarios are considered simultaneously
56
Chapter 4
Results may yield not only different levels but also different location of the site
pinch.
Scenario A
Scenario B
47,3MW
Q,5MW
VHP F===~~~~~~
2,5MW
HP
7MW
15MW
MP ~--------~------~-------
15MW
MP ----
19,5MW
58MW
2,2MW
17,5MW
17,5MW
LP J-----"- - - ' - - - - -
LP J - - - - - - ,
VLP
38 ,5MW
17,5MW
17,5MW
cw
cw
H
Scenario C
Scenario D
Q,86MW
HP
2,2MW
15MW
MP
'---- r - r - - -_
1, lMW
LP
22MW
15MW
CW
CW
Figure 4-16: The resulting SUGCCs corresponding to the optimal steam levels of Case 2
lIP (OC)
260
260
MP (OC)
200
200
LP (OC)
110
140
Table 4-12: Optimal steam levels for single and multiple scenarios
57
Chapter 4
4.6 Conclusions
As shown with the presented case studies, the proposed optimisation methodology is
powerful to address the needs of the preliminary steam levels design for the total site
system under operational variations. By exploiting engineering knowledge, the BHM
and THM models are capable of predicting the real efficiency trends of units, by
considering the dependency of the efficiency on load and operating conditions. The
application of the two models are particularly important in the case of multiple
operation scenarios, where the steam loads and the respective efficiencies may vary
significantly. By exploiting total site analysis techniques, a new transhipment network
is developed to represent the total site system. It can be used to describe the interaction
between the placement of steam levels and steam loads of site processes. Based on the
transhipment representation and combined with the BHM and THM models, a multiperiod MILP model is applied to minimise the total utility cost for the total site under
multiple operation scenarios. Major decision variables include the overall fuel
requirement, the cogeneration potential and the cooling utility demand. The MILP
model is a general model which can not only be used for the Minimum Utility Cost
(MUC) case but also for the Minimum Fuel Requirement (MFR) case.
58
Chapter 5
Chapter 5
Layout Screening and Optimisation
5.1
introduction
Once the steam levels are determined, the design can proceed with the development of
the best structure to produce utilities. This task comprises a large combinatorial
problem. Candidate systems involve layouts of the following units:
1. simple and/or complex back-pressure turbines.
2. simple and/or complex condensing turbines.
3. reheat cycles.
4. simple and/or regenerative gas turbine cycles.
5. boiler networks.
59
ChapterS
maintain high efficiency over the entire variation range. On the other hand, the optimum
trade-off between flexibility and capital cost needs to be identified. In order to evaluate
the alternative design options and distinguish amongst the associated efficiencies, the
effects of the unit size, as well as load and operating conditions on the unit efficiencies
need to be taken into account. These effects generally involve non-linear relations that
give rise to complex models and formulations.
The problem addressed in this chapter assumes given structures of existing chemical
processes and fixed steam levels whose optimisation has been separately addressed in
chapter 4. Also given are the power and steam demands at each level. The time
associated with each operation scenario can be different. Then the design problem is to
determine the structure of the site utility system to minimise the total cost, subject to the
satisfactory of the utility demands over the available operation horizon.
Steam is collected and distributed to chemical processes, steam turbines, reheat cycles
or to the next low pressure steam level through letdown valves.
Power can be generated by gas turbines, steam turbines and diesel engines or by
importing electricity from the utility grid.
60
Chapter 5
The gas turbine efficiencies are variable and depend on the gas turbine size, load and the
operating conditions. The exhaust gas may be used by a HRSG to generate steam.
5.2
The Strategy
Systematic techniques are required to address the optimal configuration design of total
site utility systems under operational variations. The strategy that is proposed in this
work is schematically shown in Figure 5-1.
The approach combines the benefits of total site analysis, thermodynamic analysis and
optimisation techniques. The total site analysis technology is used to screen and identify
all possible design options. The thermodynamic analysis is applied to reduce the size of
the optimisation problem. Since the energy cost of a utility system is an overwhelming
factor in the analysis of each year's cost, the thermodynamic efficiency is used to guide
the selection of the main utility structures. This strategy has been applied successfully in
utility system design (Chou and Shih, 1987) and heat exchanger network design
(Linnhoff and Turner, 1981). The proposed strategy comprises the following five
stages:
61
Chapter 5
1. The total site analysis is used to screen and identify all possible design options
c;;
site
ana~)-----I.~
G;;odynamiC
a~r----i.~
62
Chapter 5
5.3
There are enormous number of candidate structures which include steam turbine cycles,
condensing turbine cycles, simple gas turbine cycles, regenerative gas turbine cycles,
combined steam and gas turbine cycles with or without condensing turbine, diesel
drivers, and all of their djfferent combinations.
The synthesis and design problem is to find a site utility system that satisfies the
chemical processes ' varying heat and power requirements, subject to minimum energy
consumption and capital investment. The method discussed in this work proposes that
the process heat requirements are satisfied first and the heat and power requirements are
matched exactly. This is particularly desirable for those industties that use a very large
amount of thermal energy such as the petrochemkal , food processing and the paper &
pulp (Chou and Shih, 1987).
VHP
HP
CW
Figure 5-2: Identification of steam turbines of a site by using the SCC
The Site Composite Curves (SCC) reflect on the integration opportunities between
chemjcal production processes and the site utjlity system. The curves can represent the
steam flow in the utility system as well as the heat exchange between site processes and
utilities (Figure 5-2). The enclosed shaded area between the steam levels is proportional
to the potential for power cogeneration . The curves also reveal the heat recovery, fuel
63
ChapterS
conceptual tool to screen and target the possible design options for the site utility
system. Figure 5-2 illustrates a possible allocation of back-pressure turbines with the
use of the
see.
The sizes and positions of the turbines can be identified using the
curves.
5.4
Thermodynamic Analysis
The objective of the thermodynamic analysis is to screen out the infeasible and
inefficient options. As it will be shown later in the chapter, The size and complexity of
the optimisation problem is reduced dramatically.
Heat and power represent energy of different quality. The thermodynamic efficiency is
defined by a relationship that determines the ratio of the useful part of the energy to the
total fuel input. The thermodynamic efficiency of a typical utility unit is defined by:
(5-1)
Qfuel
64
Chapter 5
5.4.1
Figure 5-3 illustrates the total site profiles of a pl ant. The possi ble utility structures
which satisfy the heat and power demands of the site processes are identified. T he
thermodynamic efficiencies of the corresponding utility structures are provided.
VHP
cw
Figure 5-3: Total site profiles of a plant
5.4.1.1 Integration of the Boiler and Back-Pressure steam Turbine (BBPT) cycle
The stearn crossin g each expansion zone can be used by bac k-press ure steam turbines to
gene rate power (Figure 5-4). The VHF steam is raised by the VHF boiler. The
thermodynamjc efficiency is defined by the ratio of the useful energy (power outputs
an d steam heat loads to processes) over the fuel consumption :
wi + IQ/
(5-2)
Q;" eI,i
the steam turbine to processes and Q;',el,i is the net fuel heat input in boiler i.
65
Chapter 5
Q,
cw
Figure 5-4: Integration of the BBPT cycle
As shown in Figure 5-4, the ex haust heat of BP steam turbines can be used as process
heat, therefore the VHP steam load equals:
(5-3)
(5-4)
5.4.1.2 Integration of the Gas Turbine and Waste heat Boiler (GTWB) cycle
66
Chapter 5
F'
WB
GT
VHP
The gas turbine is integrated with the waste heat boiler to generate power, and the waste
heat boiler is used to raise steam. It is assumed the gas turbine works at full load (i.e. ,
maximum efficiency load). The thermodynamic efficiency of the GTWB cycle is
defined as the ratio of the useful energy (the power output of gas turbine and the steam
load raised by the waste heat boiler) over the fuel consumption. It is calcul ated by:
(5-5)
QG
and p f is the fuel consumption . On the basis of the GTHM, the fuel consumption is
given by:
(5-6)
67
Chapter 5
(5-7)
QG = 11,BQwaste
where
1]/8
(5-8)
The BCT cycle is shown in Figure 5-6. As the exhaust heat of the condensing turbine is
lost to cooling water, the thermodynamic efficiency is defined by the ratio of the useful
energy (the power output of the condensing turbine) and the fuel requirement. It is
calculated by:
BCT
11,
We
=-QB
(5-9)
fuel
where We is the power output of the condensing turbine and Q~e1 is the fuel
consumption.
It is assumed that the condensing turbine works at full load. Hence, on the basis of the
CTHM, the relation between the power output and the steam load crossing the
condensing turbine is:
(5-10)
68
Chapter 5
(5-11)
=-a qM
'fit
where q is the specific heat load of the steam and 'fIt is the thermodynamic efficiency of
the boiler.
Combining Equations (5-9), (5-10) and (5-11), the thermodynamic efficiency of the
BCT cycle is given by:
(5-12)
Boiler
VHP
VHP
+
VAC
5.4.1.4 Integration of the Gas Turbine, Waste heat Boiler and Condensing Turbine
(GTWBCT) cycle
The GTWBCT cycle is shown in Figure 5-7. When the gas turbine and waste heat boiler
cycle produces the amount of steam more than the steam requirements from the
processes, a condensing turbine can be integrated to the gas turbine and waste heat
boiler cycle to generate additional power by using the surplus heat. The thermodynamic
efficiency for the GTWBCT cycle is defined as the ratio of the useful energy (power
69
Chapter 5
outputs of the condensing turbine and the gas turbine) over the fue l requirement. It is
calcul ated as foll ows:
T]
GTWBCT
We + W
=---F J t1H
(5-13)
where W C and W G are the power outputs of the condensing turbine and the gas turbine
respecti vely. W C is calcul ated by applyin g the CTHM and W
GTHM.
F'
WB
Fa
GT
VHP
+
VAC
The SCT is defined as the condensing turbine using surplus steam from chemical
processes as shown in Figure 5-8. Condensin g turbines can be integrated to the site
utility systems to generate power with the use of surplus heat.
As the SCT uses surplus heat, it should be con sidered as the first pri ority choice fo r
power ge nerati on in the design probl em provided th at chemi cal processes have surplus
heat. In order to compare efficiencies of different SCT units, the thermodynami c
70
Chapter 5
efficiency of the SCT is defined as the ratio of the power output to the surplus heat of
the processes. It is calculated as follows:
surp _
llr
w surp
(5 -14)
Q surp
urp
Q sur
VHP
ew~
VAC
VAC
There is always an option to import power from a utility grid. In order to compare the
efficiencies of the site utility systems with imported power, a thermodynamic efficiency
for importing power is included based on the principle of the Top Level Analysis
(Makwanna, 1997). It is assumed that the cost of Qrucl fuel can buy Wimp! power. The
imp0l1ed power efficiency is defined as:
.1Jpt
rJ"
F
W impr
CF
=- - = _cP
Qfuel
(5-15)
Chapter 5
The concept defined above provides us an efficiency for importing power, which can be
directly used for comparison with the thermodynamic efficiencies of utility structures.
BBPT
GlWBCT
I
I SCT I
IP
---------------r-----~!--------+I----~I---------------~
Ll
Pc
Po
Power (MW)
The power requirement of the site determines the utility systems to consider. The
corresponding sizes of the units are identified and targeted with the use of the TEC; the
72
Chapter 5
inefficient options are screened out. Superstructures are developed only using the
options enabled by the TEC. Figure 5-9 explains the case. Assuming a power
requirement P A, the SCT and the BBPT cycles are only allowed. The other inefficient
options are excluded. Assuming the power requirement from PA to PB, additional
considerations include a gas turbine and waste boiler cycle. The other options are
similarly screened out. The power capacity of the gas turbine is equal to the difference
between the power requirement and PA. Additional power requirement from PB to Pc
introduces the GTWBCT cycle. The power capacity of the cycle is equal to the
difference between the power demand and PB. Extending the power requirement from
Pc to PD includes the option for the BCT cycle. The capacity in power of the cycle is
equal to the difference between the power requirement and Pc. Beyond the power
requirement PD, the residual power has to be imported from the utility grid.
particular based on energy efficiency. Alternative tools can be designed to replace the
focus on energy efficiency by a focus on fuel to power ratios or other economic criteria.
In all cases, one should ensure the screening tool considers second and third-best
options. Alternatively, the TEC procedure can be relaxed by shifting the cut-off line of
the synthesis screening and let synthesis options with slightly lower efficiencies to also
become part of the superstructure.
So far, the TEC is discussed on the basis of a single operation scenario. Changes in the
steam and power demands will affect the TEC and suggest different utility structures.
For mUltiple operation scenarios, a systematic methodology is presented that makes use
of superstructure development.
5.5
The TEC is constructed for each operation scenario. The curves are applied to identify
candidate structures and capacities of the utility units. A systematic method is presented
to generate the superstructure for the site utility system and optimise for the best option.
73
Chapter 5
5.5.1
The cogeneration potential available between the steam levels can be exploited through
a lot of combinations of turbine networks and layouts such as complex turbines,
multistage turbines, single turbines and so on. As discussed by Mavromatis and
Kokossis (1998b), both complex turbines and multistage turbines are equivalent to a
cascade of simple turbines, each taking up potential from a single expansion zone, as
shown in Figure 5-10. On the grounds of the equivalence, all possible combinations of
turbine layouts are reduced to a single superset of component cylinders as illustrated by
Figure 5-11. It is only this superset of design components that is required in order to
derive the optimum structure of back-pressure steam turbine network.
The sizes of component turbines for each scenario are determined at the thermodynamic
analysis stage. For multiple operation scenarios the number and sizes of the simple
component cylinders for each expansion zone are identified by using the discretisation
method proposed by Mavromatis and Kokossis (1998b). The method suggests that the
turbines are sized to match the loads of every scenario as well as all their possible
combinations.
74
Chapter 5
The case of two scenarios with the steam flows across an expansion zone is illustrated
in Figure 5-12. The first combination involves the selectjon of turbine TI sized to the
scenario B, while operating at part load under scenario A. This option features the
lowest capital cost, but lower part load efficiency for scenario A. Alternatively, turbine
T2 can be installed to size scenario A.
efficiency but requires the highest capital cost. In option 3, turbine T2 can be se lected,
along with turbine T3 sized to take up the remaining load for scenario B. The efficiency
for scenario B will be smaller than the first two options, but achieves the highe t
efficiency for scenario A and requires lower capital cost than the second option.
Scenario B
ScenarioA ~
14-- -
E}gg
HP
Q
MP
HP
MP
11
11
:f. . i.
""m
'~1""""';": ""'~"""""""''''''''''2?T~
'''''''''T' ' :
:' ""':['"'' ' ' ' ' ' ' ' '''''' 'Y
'r2
' ' ' ' :'
:
T3:
..
..
...
...
1 .
.
..
.
. ..
..
Option 1
Option 2
..
j
..
Q
Option 3
Figu re 5-12: The candidate BP steam turbines for the case of two scenarios
75
Chapter 5
The capacity of the gas turbine for each scenario is determined by using the TEC. For
multiple operation scenarios the number, sizes and types of candidate gas turbines of the
superset depend on the specific problem, as explained in the following section.
The types of the gas turbine cycles are concerned with simple and regenerative gas
turbine cycles. The major difference between the simple and regenerative gas turbine
cycles is the addition of a recuperator for heat exchange between the turbine outlet and
the compressor outlet as shown in Figure 5-13. Following Chou and Shih (1987), the
types of the gas turbines are screened by the characteristic value of power to heat ratio,
PIH. Table 5-1 shows the ratio corresponding to each gas turbine cycle.
Heat-exchanger
Air
Turbine
Compressor
PIH ratio
0.65
0.85
76
Power
Chapter 5
The GTHM relates the power output to the fuel load and the size of the gas turbine as
shown in Equation (3-33). As both the load and size of the unit need to be optimised,
straightforward modelling would result in an MINLP formulation. The discretisation
method follows next, however, reveals an MILP problem could be solved as the
economic analysis of the gas turbine operation suggests the discrete sizes and number of
the candidate gas turbines.
0.5
40MW
0.4
TJe
60MW
0.3
0.2
0.1
0
0
10
20
30
40
50
60
W(MW)
Figure 514: The effect of part load operation prevails over the increase of efficiency with gas
turbine size
As the electrical efficiency increases with size but decreases at part load, the highest
efficiency is obtained when a gas turbine is sized to operate at full load under the
specific power demands. This is shown in Figure 5-14. Hence, in terms of efficiency,
the optimum turbine size for each scenario is that which exactly matches the power
demand for each scenario. For multiple operation scenarios, the maximum efficient
sizes may not be the optimal sizes if capital cost is considered. Following the
discretisation scheme used for BP steam turbines, a discretisation method is proposed
whereby gas turbines are sized to match the power demand of every scenario as well as
all their possible combinations. The case of two scenarios with the power demands is
illustrated in Figure 5-15.
77
Chapter 5
There are three combinations and three component gas turbines are identified. The first
option features the lowest capital cost, but lower part load efficiency for scenario A. The
second option achieves the highest overall efficiency but requires the highest capital
cost.
In option 3, The efficiency for scenario B will be smaller than the first two
options, but achieves the highest efficiency for scenario A and requires lower capital
cost than the second option.
t4-
PB (MW)
PA (MW) ~
--+j
DO
PA
D=cJo
PB
G~
j /......... ..
..
n ....
~ ~
1...-_ _ _ _
: __
PA
PB P(MW)
.....
..
nr~
PA
PB P(MW)
Ootion 2
Option 1
..
:
PA
:
PB P(MW)
Option 3
Figure 5-15: Candidate gas turbines for the case of two scenarios
The BHM is used to describe the performance of each fired boiler and waste heat boiler.
It is reminded that the BHM relies on the size of the boiler M max The design model is
given by:
Qfuel
(5-16)
The above expression relates the fuel flowrate with the steam load and the boiler size.
As both the steam load and the size of the boiler need to be optimised, the model results
in an MILP formulation by converting the bilinear model to a linear model.
78
Chapter 5
Waste heat
...
Fuel
VHP
Figure 5-16: Superset of boilers
The superset of the boilers is shown in Figure 5-16. The waste heat is the waste heat
from gas turbine cycles. The number, sizes and fuel requirements of the boilers are
detennined by the optimisation.
The power output of the VHP condensing turbine for each scenario is determined with
the use of the TEes. In terms of efficiency, the optimum turbine size is the one that
exactly matches the power demand for a single scenario. For the mUltiple operation
scenarios, the maximum efficient sizes may not be the optimal sizes if the capital cost is
considered. The number and sizes of candidate condensing turbines of the superset for
multiple operation scenarios are determined by the discretisation scheme followed for
the gas turbines. The discretisation method is proposed for multiple operation scenarios
so that condensing turbines are sized to match the power demand of every scenario as
well as all their possible combinations.
The case of two scenarios is illustrated in Figure 5-17. There are three combinations and
three component condensing turbines. The first option features the lowest capital cost,
but lower part load efficiency for scenario A. The second option achieves the highest
overall efficiency but requires the highest capital cost. In option 3, The efficiency for
scenario B is smaller than the first two options, but achieves the highest efficiency for
scenario A and requires lower capital cost than the second option.
79
Chapter 5
VHP
VAH
Superset of VHP condensing turbines
. m~;,uu.T~ "~C~T
! uT'1~CT~iACTj
:
:
...:
'------'----'--+
PA
Option 1
PB P(M'N)
CT:1;
...
PA
:
...
PB P(MW)
Option 2
...:
PA
i:
...
PB P(M'N)
Option 3
Figure 5-17: Candidate VHP condensing turbines for the case of two scenarios
5.5.5
The surplus heat of the processes for each level is obtained by total site analysis. The
optimum turbine size for each scenario exactly matches the surplus heat load. The
number and sizes of candidate surplus steam condensing turbines of the superset for
multiple operation scenarios are determined by the discretisation scheme followed for
the back pressure steam turbines. The discretisation method is proposed for mUltiple
operation scenarios so that condensing turbines are sized to match the surplus heat loads
of every scenario as well as all their possible combinations. The case of two scenarios
with the surplus heat loads is illustrated in Figure 5-18. There are three combinations
and three component surplus condensing turbines are identified.
80
Chapter 5
IP
VAH
Superset of VHP condensing turbines
O~rp
0aurp P(MW)
Option 1
O~urp
Option 2
O ~urp
P[MW)
O~urp
0Burp
P(MW)
Option 3
Figure 5-18: Candidate surplus steam condensing turbines for the case of two scenarios
5.5.6
Reheat Cycles
In a reheat cycle, steam is first expanded to some intermediate pressure and then
reheated in the boiler. It next expands in the turbine to the exhaust pressure. Following
Chou and Shih (1987), the reheat cycle can improve the overall thermal efficiency only
if the thermal efficiency contributed by the reheat part is greater than that of the
remaining parts. The required large heat-exchange area and the increased complexity in
system design detract from the gain in efficiency due to reheating. The reheat cycle,
therefore, will be of interest to site utility system design only when a lot of heat is
exhausted to cooling water.
5.6
Optimisation Model
In thi s section , a mUlti -period MILP model is presented for the minimisation of capital
investment and operating cost. The model incorporates the BHM, THM, CTHM and the
GTHM models. The optimisation is a screening tool for the selected alternative design
81
Chapter 5
options by using the thermodynamic analysis, rather than for the exhaustive structures.
The binary variables account for the selection of units and their operation status at each
scenario. The continuous variables relate to the stream flowrates (steam, fuel), the
power outputs and the operating and capital costs.
The optimisation problem involves the following definitions for sets, parameters and
variables:
Sets
IB
IT
IC
VC
p
IG
K
Parameters
a, b
A ic, Bic
Ag, Bg
Az, B z
Cp
cpg
Cpa
ci
EISic
EIS z
F/,max
Ig
82
Chapter 5
M-/i
LB
M i!'max
r
P
AT:at
T/
cf1
U{
UfT,i
w/
em
Binary variables
Yi!e
Bo
y ib.k
yBT.e
Z.II
yBT.o
z.lI.k
Yi?,e
/7.0
ic.k
ygT.e
yGT.o
ig.k
83
Chapter 5
Continuous variables
eE!
(fiT!
C tot
C{
(;'01
M i~.k
M z.it.k
BT
M i~
: amount
under scenario k
Qih.k
B,J
QB.w
ih.k
QGT.w
ig,k
Fi{k
Wz.it,k
BT
W/
U
)'
Wi;]'
wi~I
5.6.1
Model Formulation
Given the parameters, sets and variables above, the design model includes consideration
for the following models:
84
Chapter 5
The above bilinear model is replaced by the following mixed integer linear model:
(5-18)
Constraints include:
B.o < X B
<U
L B Y ib.k
ib.k -
B.o
Y ib.k
'
rnax
ib E IB, k E K
(5-20)
The above logical constraint denotes X i:.k =0 while the boiler is not in operating status.
M~'rnax
_y~.eUB
<0,
,b
,b
-
ibE IB
(5-21)
(5-22)
The above logical constraints denote the boiler capacity should be zero if it is not
selected and the boiler capacity should be larger that its lower bound and smaller than
its upper bound.
(5-23)
The above logical constraint denotes the load of the boiler should be zero if it is off.
85
Chapter 5
(5-24)
The above constraint denotes the load of the boiler shouldn't be larger than its capacity.
(5-25)
The above logical constraint denotes the boiler can't operate if it is not selected.
The THM applied for the power output of a back-pressure steam turbine in zone z under
scenario k yields:
ZE
Z, it E IT, k E K
(5-26)
BT,o _
Yz,it,k
BT,e
Yz,it
<
0,
-
ZE Z'
,ltE IT ,kE
(5-27)
The load of the steam turbine shouldn't be larger than its capacity:
HT
M z.it.k
ZE
Z'
,It E
IT ,kE K
(5-28)
The CTHM applied for the power output of a condensing turbine under scenario k
yields:
86
Chapter 5
WicCTk
,
6 1 (ElS
=5B
if -
Ie
Aic
M.IeCT ,max
)(MCT
1 MCTmax
era)
,'c t - ic'
Yit t' ,
'6
'
iCE
[C,kE K
(5-29)
Yic,k
IC ,kE K
ICE
(5-30)
The load of the condensing turbine shouldn't be larger than its capacity:
CT
CT,oM CT max < 0
M ic.k - Yic,k
ic
-,
IC E
IC ,kE K
(5-31)
The GTHM applied for the power output of a gas turbine under scenario k yields:
ig E IG,k E K (5-32)
Where the maximum fuel flowrate F;{maX can be calculated by Equation (3-30).
GT.o _ yf?T,e
Y,g,k
19
<0
-,
ig E [G , k E K
(5-33)
The load of the gas turbine shouldn't be larger than its capacity:
Ff
,g.k
_yGT,OFf,max
'g.k
'g
<0
-,
87
igE
[G , kE K
(5-34)
Chapter 5
QiGT,w
g, k
[1f
1 1(
a a
Ag ] f
-CpT
+C p f T f +Ml f -( +n)Bg Ml f - F f ,max ) Figk, ,
igEIG,keK
Ig
(5-35)
The mass balance across each expansion zone z for scenario k involves the steam
through the turbines and the steam throttled through the let down valves (in case the
installation of a turbine is not cost effective). The mass balances give:
(5-36)
Z E Z, k e K
iteff
The VHP steam requirement is equal to the sum of the amount of steam across the first
expansion zone plus the amount of the steam through the VHP condensing turbines,
LMi!,k =M tk + LM;:k'
ibelB
kE
(5-37)
vceVC
~w.
+Wk
.J
pl,k
buy
dem
=Wk
kE K
'
(5-38)
piEP
(g) Costs
CB,f
= LU! Qi!:{T/ H
keK
ibelB
88
(5-39)
Chapter 5
(5-40)
The capital cost incurred for the installation of the equipment is:
cc,tot
LC
C
j
(5-41)
jei
The capital costs of the units are calculated using the functions presented by Bruno et
al. (1998). They are given in Table 5-2 along with linearized expressions for the boiler
cost. It is important to note that the discretisation methods (Section 5.5.2) treat the
capacities of the steam turbines, the gas turbines and the electric generators as
parameters in the optimisation model. Therefore, the nonlinear capital cost functions for
these units are applied.
Unit
4954p'77fp2
Nonlinear
Linear (9Mpa)
495384+13861F
Nonlinear
941Ffgu.7 )
Linear
6996+211.5Ffg
Steam turbine
Nonlinear
2237WstU.41
Nonlinear
952WgtU . /b
Nonlinear
176Wegu.49
Nonlinear
904FB u.Cll
89
Chapter 5
The optimisation model consists of linear constraints and integer variables, and
comprises a multi-period MILP model. The structure and the operation strategy are
optimised to minimise the total cost consisting of capital cost and operating cost. The
development of the MILP model requires the following information:
(a) steam level specifications
(b) data on total site profiles for each scenario
(c) power demand for each scenario
(d) cost correlations for the utilities
(e) capital cost correlations for the units
Programming (MINLP) formulation with a very large number of variables. The use of
total site analysis and thermodynamic analysis has reduced the problem into a
reasonably sized.
The outcome of the optimisation stage is a set of simple steam turbines and condensing
turbines. The simple turbines can be used to synthesise practical complex or multi-stage
90
Chapter 5
turbines. The synthesis of the complex turbines relies on the operation schedule of the
simple turbines as provided by the optimisation stage. For two cylinders to merge into a
complex unit, they both have to be loaded during the same scenario. Depending on
whether the steam flow through the upper cylinders of a complex turbine is larger or
smaller than that in the lower sections, the turbines can be of an extraction or induced
type.
Two case studies are selected to illustrate the capabilities of the methodology. The
operating conditions of the four steam levels, the vacuum header and the de aerator are
shown in Table 5-3. The steam used for heating can be returned as condensate. The
difference between the two case studies relates to the utility demands. The cost data of
the utilities used are given in Table 5-4. The capital costs are given in Table 5-2.
Unit
Operating Conditions
VHPHeader
HPHeader
MP Header
LPHeader
Vacuum Header
Deaerator
Demineralized Water
Temperature: 27C
LHV: 13856kWhffon
Cost: 0.24${fon
Cost: 223${fon
Table 54: Utility data
91
Electricity
Cost: O.I$/kWh
Chapter 5
5.8.1
Case Study 1
The Site Utility Grand Composite Curves (SUGCC) of Figure 5-19 reflect the steam
demand/generation of the site under three operation scenarios. The power demands for
the scenarios are given in Table 5-5. On the basis of the SUGCC, the aim is to find the
optimal configuration of the site utility system that satisfies the utility demands and
minimises the annual total cost.
Scenario B
ScenarlOC
ScenorloA
260
180
HP
MPI-----=-=30=-=-0-------1
MP
MP
LP
VHPI==="""""-~~-...,
VHP
HP
VHP
HP
130
30
LP
50
lP
VAC
VAC
VAC
60
H !h}
H !h}
HUh}
Scenario A
Scenario B
Scenario C
22
35
42
The SUGCC are given in Figure 5-19. The steam across each expansion zone is
obtained in Table 5-6. LP steam is in surplus and used by the condensing turbines.
Back-pressure turbines are installed in the steam expansion zones VHP-HP and HP-MP.
The power outputs for the possible SCT and BBPT cycles are calculated and shown in
Table 5-7.
92
Chapter 5
Scenario A
Scenario B
Scenario C
VHP-HP
180
220
260
HP-MP
130
260
300
30
50
60
MP-LP
LP-Vacuum header
Table 5-6: Steam amount across each expansion zone of Case 1 (tIh)
Scenario A
Scenario B
Scenario C
SCT(MW)
2.5
4.1
BBPT(MW)
13
22.9
26.7
Table 5-7: Power outputs of possible SCT and BBPT cycles of Case 1
The characteristic values of power to heat ratio of the site for the three scenarios are
presented in Table 5-8. The PIH values are lower than the characteristic value of the
simple gas turbine cycle. A simple gas turbine cycle is then employed instead of a
regenerative gas turbine cycle. By calculating the efficiency and the maximum power
output for every potential cycle, the TEC is constructed starting with the most efficient
option with the corresponding power capacity limit. The TECs for the three operation
scenarios are shown in Figure 5-20. On the basis of the TECs and the utility demands,
the promising candidate utility structures include the Boiler and Back-Pressure steam
Turbine (BBPT) cycles, the Surplus Condensing steam Turbine (SCT) cycles and the
Gas Turbine and Waste heat Boiler (GTWB) cycles. All the other design options are
screened out because of their lower efficiencies. By using the TECs, the power outputs
of the GTWB cycles are obtained for all scenarios as shown in Table 5-9.
93
Chapter 5
Scenario A
Scenario B
Scenario C
0.22
0.285
0.29
PIH ratio
Scenario 8
Sconarto A
,....
GTWB
I
,
,
I
IICT
~
I=q
seT
2.5
GTWIICT
'JMW
I.~ _1
6.5MN I
22.9MIN
"l.......................~......
PIIAW)
ScenaloC
B8PT
GT'M!
GlWIICT
IICT
_G'
..
seT
I--
~l-.
26,1WNI
f - - - - ---1 1OJMW ~.
Scenario A
Scenario B
Scenario C
6.51
7.98
10.32
5.8.1.3
On the basis of the steam amount across each expansion zone for each scenario,
candidate sizes of the BP steam turbines expressed as maximum steam capacities can be
obtained for each zone. The sizes are given in Table 5-10.
94
Chapter 5
BTl
BT2
BT3
BT4
BTs
VHP-HP
180
220
260
40
80
HP-MP
130
260
300
30
170
Similarly, for every scenario the surplus heat provided by the processes determines the
candidate sizes of the condensing turbines. These are given in Table 5-11. The simple
gas turbines are used. Expressed in terms of power capacities, the candidate sizes of the
gas turbines are given in Table 5-12.
Size (tJh)
CT l
CT2
CT3
CT4
CTs
30
50
60
10
20
Capacity (MW)
OTl
OT2
OT3
6.51
7.98
10.32
OT4
1.47
OTs
OT6
3.81
2.34
VHP boilers are fired by fuel and heated by the waste heat from the gas turbine cycles.
An HP waste heat boiler and an MP waste heat boiler recover surplus heat from the
processes. Without heat exhausted to cooling water, the reheat cycle is excluded as an
option. The superstructure of the site utility system is shown in Figure 5-21.
95
Chapter 5
VHP
HP
_ _- r - - - . - . - . - - - - - . -..- ....
MP
LP
~~~IDeaerator
L -_ _ _ .
'-------"
5.8.1.4 Optimisation
96
Chapter 5
operate between the HP and MP levels. They supply power by exploiting the
cogeneration potential. The condensing turbine is employed to generate power by using
surplus heat from the site processes. The back-pressure turbines and the condensing
turbine do not meet the power demands. The gas turbine is installed to supply the
remaining power. The total annual cost is 33.341 MM$. The annual fuel cost is 27.718
MM$ and the annual capital cost is 5.623 MM$. The capacities of the selected units are
given in Figure 5-22.
Max lO.32MN
VHP
Max6.91MW
HP
Max 19.78MN
LP
Max4.99MW
Max360tfrl
.&...,..._________
VAC.
UDeoeratO'
Figure 5-22: Optimal structure of Case 1
The optimal loads of the units are obtained under each scenario. These are given in
Table 5-13. The back-pressure turbine BTl (HP-MP) operates during A, but not during
Band C. Turbine BT3 (HP-MP) operates during Band C, but not during A. During A
turbine BTl (HP-MP) is more efficient than turbine BT3 (HP-MP) and during Band C
97
Chapter 5
turbine BT3 (HP-MP) is more efficient than turbine BTl (HP-MP). The HP waste heat
boiler shuts down during B and C because there is no waste heat during these periods.
All other units operate for all three scenarios.
Scenario A
Scenario B
Scenario C
BT3 (VHP-HP)
180tlh
220tlh
260tlh
4.36MW
5.64MW
6.91MW
BTl (HP-MP)
130tlh
Otlh
Otlh
8.38MW
OMW
OMW
BT3 (HP-MP)
Otlh
260tlh
300tlh
OMW
16.61MW
19.78MW
CT 3
30tlh
50tlh
60tlh
2.0MW
4.0MW
5.0MW
GT 3 (power output)
7.26MW
8.76MW
1O.32MW
180tlh
220tlh
260tlh
HP WB (steam load)
50tlh
Otlh
Otlh
LP WB (steam load)
30tlh
50tlh
60tlh
21O.3t1h
31O.5t1h
360.5t1h
Unit
5.S.1.5
As shown in Table 5-14, no steam turbines have the same operation schedule.
Therefore, no complex turbines can be synthesised.
98
Chapter 5
The SUGCC account for the steam demand/generation of the site under the three
operation scenarios of Case 1. These are shown in Figure 5-19. The power demands for
this case are given in Table 5-14.
Scenario A
Scenario B
Scenario C
12
25
30
Based on the TECs of Figure 5-20 and the power demands, the candidate utility
structures include the Boiler and Back-Pressure steam Turbine (BBPT) cycles and the
Surplus Condensing steam Turbine (SCT) cycles. The other power units are not
economic and are excluded. The superstructure is shown in Figure 5-23 and is smaller
than the one studied in Case 1. The candidate sizes of the turbines are apparently the
same.
VHP
HP
L---
MP
LP
~.~
...:...._. . ..-.lI....._ _.....&_....._
VAC .
....,.-.lI....._--l'--_ _.L
;Deoeralor
99
CONDo
Chapter 5
The MILP model is optimised against the total annual cost. The model involves 97
continuous variables, 60 binary variables and 157 constraints.
The optimum configuration is shown in Figure 5-24. It includes two BP steam turbines,
one condensing turbine, one VHP boiler, one HP waste heat boiler, one MP waste heat
boiler and the deaerator. One back-pressure turbine operates between the VHP and the
HP steam levels. Another one operates between the HP and the MP steam levels. The
condensing turbine generates power using surplus heat from the site processes. There is
no need for a gas turbine. The total annual cost is 30.751 M$. It is lower than Case 1
because of the lower power demand. The annual fuel cost is 26.325 M$ and the annual
capital cost is 4.426 M$. The capacities of the units are given in Figure 5-24. The
optimal loads for each scenario are given in Table 5-15.
VHP
t-::--,
~
Max 5.78MN
HP
Max19.78MN
MP
LP
Max5.0MN
..z...,..._ _ _ _....._ _ _ _
VAC.
The back-pressure turbine BT2 (VHP-HP) and the turbine BT3 (HP-MP) maintain
identical operation schedules as shown in Table 5-15. Therefore, the two turbines can be
synthesised as a single complex turbine.
100
Chapter 5
Unit
Scenario A
Scenario B
Scenario C
BT2 (VHP-HP)
180tlh
220tlh
220tlh
4.52MW
5.78MW
5.78MW
BT3(HP-MP)
119t1h
243t1h
300tlh
5.49MW
15.23MW
19.78MW
CT3
30tlh
50tlh
60tlh
2.0MW
4.0MW
5.0MW
180tlh
220tlh
260tlh
HP WB (steam load)
50tlh
Otlh
Otlh
LP WB (steam load)
30tlh
50tlh
60tlh
21O.3t1h
31O.5t1h
360.5t1h
5.8.3 Discussion
Although the SUGCC of the two studied cases are the same, the optimal configurations
of the site utility systems are different because of the different power demands. It
implies that different PIH ratios assume different optimal structures. By using the
proposed thermodynamic analysis, the reduced superstructure is obtained.
For Case 2, the MILP model is also optimised by integrating the gas turbine cycles into
the superstructure. The same optimal solution is obtained. However, the model involves
195 continuous variables, 92 binary variables and 242 constraints, which is much bigger
than the proposed method. If other possible design options are integrated with the
superstructure, the size of the optimisation problem should be even much bigger.
101
Chapter 5
5.9 Conclusions
A systematic methodology is presented for the optimal design of site utility systems
under operational variations. The methodology combines the benefits of total site
analysis, thermodynamic analysis and optimisation techniques. The approach accounts
for the interactions between the site utility systems and the site processes. The design
task is addressed in view of the anticipated variations in the process demands and the
effect of the unit capacities and varying loads on the efficiencies of the selected units.
These aspects normally give rise to highly complex and large problems. The proposed
methodology utilises total site analysis and thermodynamic analysis to reduce the size
and complexity of the design problem. The total site analysis is employed to screen and
identify all possible design options. A thermodynamic curve is proposed in this
approach. It is a useful tool to identify the promising candidate design options to be
included in the superstructure by screening among all alternative design options. The
size of the optimisation problem can be reduced by screening out the uneconomic
design options.
102
Chapter 6
Chapter 6
Debottlenecking and Planning Optimisation of an Existing Site
6.1
Introduction
Industrial plants are in need for debottlenecking technology and planning techniques to
accommodate with the best investment scheme and their process operations (Hirsheld
1987). There are additional incentives and challenges for site utility systems. First, one
has to assess the potential for the purchase of new equipment. The challenge has to
consider a dynamic environment with changes in utility demands and prices, changes in
technology and changes in markets and regulations. The site utility system addresses a
particularly dynamic market. Changes consider the operation of the chemical processes,
changes in the feed/product specifications and the modifications in the schedule of
production.
The problem assumes a given utility system. Also given are forecasts for the prices and
demands of utilities, the ambient conditions and the regulations over a finite number of
time periods. The problem identifies a number of time periods of different duration. The
heat requirements are satisfied in preference to the power requirements. A
superstructure is developed that considers existing processes and potential new
equipment. The candidates for debottlenecking are generated with the use of total site
103
Chapter 6
optimal operational strategies over the considered period. The units are possible to
tum on and off and the approach considers different operating levels in the plant.
the amount of fuel, the appropriate type of fuel, the amount of water and assesses
available options to import power.
Non-linear and bilinear models are converted to MILP models by using piecewise
linearisation and integer programming techniques. Non-linear cost models are used for
the steam and gas turbines.
1. Heat recovery system generators (HRSG) for recovering heat contained in gas
turbine or furnace gases and for generating super heated very high-pressure (VHP)
steam. Supplementary firing is allowed in these units.
2. Fuel fired units for generating VHP steam.
3. Waste heat units for recovering heat from process flue gases or from process units
such as chemical reactors.
The approach makes use of the BHM, so that capacities of the boilers can be
determined, and the boiler efficiency appears to be a variable of capacity, heat load and
operating conditions.
Gas and steam turbines are considered as potential electricity generators. Options for
steam turbines include:
1. Back-pressure steam turbines exhausting to lower pressure levels.
104
Chapter 6
Complex turbines are decomposed into simple turbines and the use of the THM and
CTHM enables the turbine efficiency to vary with the turbine capacity, its load, the
operating conditions and the type of exhaust.
6.2
Optimisation Strategy
cr;;i
site
ana~)-----I
~odynamiC~
Superstructure generation
(Existing system + promising debottlenecking options)
105
Chapter 6
The existing system is optimised for the maximum power output of the system over the
considered periods.
6.4
The Site Composite Curves account for integration between the site processes and the
site utility systems. Figure 6-2 shows the representation of a total site utility system that
uses the SCC. The shaded units are the candidates need to be installed; the others are
existing units. The capacities and the placement of the debettlenecking units are
possible to identify using the Sec.
In order to target the capacities of the potential steam turbines, one has to identify the
capacities of the existing complex turbines for each expansion zone. It is accomplished
by decomposing the complex turbines into sets of simple turbines with specific
capacities. A decompose strategy is presented following Chou and Shih (1987). Figure
6-3 illustrates a decomposition problem. The decomposition produces individual simple
turbines aJ, a2 and a3 which are of the same capacity. Similarly, simple turbines bi and
106
Chapter 6
b2 are of identical capacities. The decomposition strategy reduces the analysis at a level
where si mple turbines are only used.
VHP
cw
Figure 6-2: Identifi cation of debottlenecking turbines of a site by using the S
Figure 6-3: Decomposing a complex turbine into different sizes of cylinde rs in eac h ex pansion
zone
107
Chapter 6
The debottlenecking includes options for boiler and back-pressure steam turbine
(BBPf) cycles, gas turbine and waste heat boiler (GTWB) cycles, boiler and
condensing turbine (BCT) cycles, gas turbine, waste heat boiler and condensing turbine
(GTWBCT) cycles, diesel drivers and option to import power (IP) from a utility grid.
The thermodynamic analysis screens out the inefficient options and identifies sensible
scenarios to include in a superstructure representation. After the thermodynamic
analysis the size of the problem is reduced dramatically.
6.5.1
The basic idea is to employ the TEC first on the existing utility system and next with the
consideration of additional options.
BBPT
Exis1ing System
GlWBCT
IP
II~f1
PelOsi
PA
PB
PG1W8
1+1
PGTWBCT
I-~IPD
Pc
--CD
PE
Power (MW)
The construction of the TEC is based on the assumption the heat requirements from the
chemical processes are satisfied before the power demands are considered. Efficient
debottlenecking options are identified by comparing the thermodynamic efficiencies.
Given is the steam demand at each steam level. Then, the thermodynamic efficiency and
the maximum power output are calculated for the existing system. The maximum power
108
Chapter 6
IS
6.5.2
The superstructure is de eloped on the basis of the options suggested by the TEe. In
reference to Figure 6-4 for example, a power demand to P A would imply the existing
system is sufficient to satisfy it. Figure 6-5 explains scenarios with power requirements
above P A and up to P B , it is necessary to invest on a new BBPT cycle. The additional
power can be targeted wi th the hardware capacity. The capacity equals the difference
between power requirement and P A. It is inefficient to explore other debottleneckin g
options.
BBPT
Til
GlWB
EsIlng~
p.
P,
cw
Figure 6-5: The boiler and bac k-pressure turbine cycle for debottlenecking
Fi gure 6-6 exp lain cases of power requirements above P B and up to Pc. The gas turbine
and waste boiler c c le i required to consi der as an additional option . The capacity of
the gas turbine hould be consi dered from the difference between the power requirement
and P B .
109
Chapter 6
F'
WB
'l'
66PT
bistro~S!em
GTWB
GTWBCT
~
........ .
IP
P~~
P-
P-
P,
P,
Po
p.
Ft>wer (MW)
~",,"",,!;;=9 VHp:
HP
Figure 6-6: The gas turbine and waste heat boiler cycle for debottl enec king
Between Pc and Po as shown in Figure 6-7 , the additional power should be avail able
from a GTWBCT cycle. The power capacity of the cycle should equal the difference
between the power requirement and Pc.
WB
GT
n.
.........................................................
BBPT
Fuel
GTWBCT
BCT
-'~ I
~
HP
VHP
I:: : : : : .
Figure 6-7: The gas turbine and waste heat boiler and condensing turbine cyc le for
debottlenecking
110
Chapter 6
Similarly Figure 6-8 explains the picture between P D and P E. A BeT cycle should
produce the additional power. The power capacity equals the difference between the
power requirement and PD. Beyond P E, power should be imported from the utility grid.
r. . . . . . . . . . . . . . . . . . .
~~~;
GTWBCT
Fuel
BCT
F"";;;;;;;!;=i
WB
VHP
VHP!.
HP
Iff
Power (MW)
:..........................................................
C
::
VAC
Figure 6-8: The boiler and condensing turbine cycle for debottlenec king
Figure 6-9 shows that if the chemical processes generate surplus steam, the surplus
condensing turbines can be integrated to generate additional power using surplus steam
for diffe rent levels.
VHP
11
SSPT
E.omg $vslem
SCT
IP
~-[~}
P,
P,
Po wer IMW)
VAC
III
Chapter 6
So far, the TEC is discussed to identify the most efficient debottlenecking options for a
single period operation . The discrete scheme presented in Section 5 .5 is applied to
generate the debottlenecking superstructure.
6.5.3
The superstructure consists of existing units and new units. The superstructure
development is
schematically shown
in Figure 6-10.
The capacities
of the
debettlenecking options are identified with the use of the TEC for each single operation
period and the discretisation schemes. The superstructure is modelled over all operation
periods.
Period n.
Period 2
Pe riod 1
112
Chapter 6
The optimisation problem involves the definitions for sets, parameters and variables:
Sets
EB
IB
IT
NT
IC
NC
VC
NG
= {k I operation periods }
IG
Parameters
a, b
A ic , Bic
Ag, Bg
Az, Bz
113
Chapter 6
Cp
Cp
Cpa
Cp
ElSie
ElSz
max
F/'
Ig
Mlf
LB
M !.max
M Z.I
BT,'max : maximum steam load of BP steam turbine it in expansion zone z
M i;rmax : maximum steam load of condensing turbine ic
M ;,k
r
P
!:IT:al
T/
if
U~T,J
w/em
Binary variables
y!',~
yj~'O
yB.e
ib.k
114
Chapter 6
Yz.nI
BT.e
y:::;:~
y;;.e
yCl'.o
ie.k
y~T.e
yGT.o
ig.k
Continuousva~ks
cH!
cGT!
C lot
C{
Cot
M i!.k
M i!'rnax
M z.it.k
BT
M i~
M ;;:k
M ~.k
:amount of steam throttled through the let down valve of expansion zone z in
period k
QB,J
Qeb.k
B,J
QB.w
QB.w
QGT.w
ib.k
ib.k
eb.k
ig.k
115
Chapter 6
Fi:,k
WZ,il,k
BT
W/UY
Wi;:I
Wi~r
Wp~
6.6.1
Mathematical Formulation
Given the parameters, the sets and the variables above, the model includes consideration
for the following models:
Using the BHM, the fuel requirement of an existing boiler in period k is given:
The BHM is applied for the fuel requirement of a new boiler. As the boiler is optimised
both load and size, straightforward modelling with the BRM would result in the bilinear
tenn:
The above bilinear tenn is modelled instead as a mixed integer linear tenn:
116
Chapter 6
:,k equals M
< X
L B Y ibB ,0
,k -
B
<
ib ,k -
;!,max while
,0
Y ib ,k
'
ib
ib E IB, k E K
(6-5)
IB, k
(6-6)
The above constraint denotes the boiler capacity should be zero if it is not selected.
(6-8)
The above constraint denotes the load of the boiler should be zero if it is off.
(6-9)
The above constraint denotes the load of the boiler shouldn't be larger than its capacity.
(6-10)
The above constraint denotes the boiler can't operate if it is not selected.
117
Chapter 6
Using the THM, the power output of a back-pressure steam turbine in zone z in period k
yields:
BT
6 1
=--(EIS.
... ~ 5 B
-
W.I'I.
)(M BT
A:
:.it.k
BT.max
_'!"M
6
BT.O)
BT max
:.il
Y:.it.k'
ZE
Z, it E IT, k E K
Z,ll
(6-11)
Y:::: _y::
BT
M :.i1.1:
SO,
zE
BT.oM ;;.il
BT.max < 0
-,
Y:.il.k
Z't
IT ,kE K
,I E
(6-12)
(6-13)
CT
Wid
6 1
= ST(EI
ic -
Ie
Aic
)(M CT
CT.max
ic,k -
(6-14)
IC
Y LT.o -
y;:.e ~O,
(6-15)
cr _ M iccr.max Yie,k
cr ,0 <
0, I C
. E IC , k E K
M ie,/(
-
(6-16)
IC.~
118
Chapter 6
GT
WIg, k =-(Ml
f
Bg
GT.o -Yng
GT,e <0
Yig.k
,nge
NG'
,lge NG ,ke K
GT.o <0
' - F"max
Fig.1:
ig
Yig.k ,
(6-18)
. e I'1G , k E K
(6-19)
19
The waste heat from the gas turbine can be obtained by:
(6-20)
The mass balance across each expansion zone z in period k involves the steam through
the turbines and the steam throttled through the let down valves (in case the installation
of a turbine is not cost effective). The mass balances yield:
zE
(6-21)
Z,ke K
ilErr
The VHP steam requirement is the sum of the amount of stream across the first
expansion zone plus the amount of steam through the VHP condensing turbines:
"LMi!.k +ebEEB
"LM!,k = Mtk + vceVC
"LM;:k'
ibEIB
119
kE
(6-22)
Chapter 6
~w
pi.k
buy
+ Wk
= Wdi!m
k
'
kE K
(6-23)
pieP
(g) Costs:
The costs include the fuel cost of the utility system and the capital cost incurred for the
installation of the new equipment.
=~
~UIQ.BJr,sH+
k
ib.k k
keK
ibelB
~UfQBJT,sH
~
eb.k k
(6-24)
keK
ebeEB
C GT .!
~ U GT .! pi
r,s H
Ig.k k
(6-25)
keK
igelG
Cc,tot
= Lct
(6-26)
iel
The capital costs of the units are calculated using the functions presented by Bruno et
120
Chapter 6
(6-27)
The total annual cost consists of the capital cost and fuel cost.
The proposed model is an MILP problem and can be solved using a full space LP based
Branch and Bound method. A large number of the time periods however automatically
increases the size of the MILP model. The alternative approach followed in this work
solves the large MILP model using a decomposition algorithm such as the one proposed
by Iyer and Grossmann (1998). The algorithm is schematically shown in Figure 6-11.
The multi-period MILP is decomposed into a debottlenecking problem (Master
problem) and an operation-planning problem (Operation problem). Both problems are
formulated as MILPs. The master problem is a relaxed version of the original problem
in that it contains only a subset of its constraints and debottlenecking options. Its
solution provides a lower bound on the objective. The solution of the master problem is
used to set up the operation problem. The operation problem involves the
debottlenecking design options as they are fixed from the master problem. Its solution
provides an upper bound on the objective. Successive solution of these problems is
repeated until convergence criteria are met.
Master problem 1
~~"I
(MILP)
II---.-~
.1
Structure cuts
Tnal new
structure
121
Chapter 6
6.7.1
Case Study 1
Figure 6-12 shows the configuration of an existing system. The current system consists
of a main boiler (Bl), two complex steam turbines (Tl and T2) and the de aerator. The
capacities of the units are shown in Figure 6-12. There are four steam levels (VHP, HP,
MP and LP). The operating conditions of the levels and the de aerator are shown in
Table 6-1. The steam is used for heating and can be returned as condensate. The utility
system should satisfy the heat requirements in preference to the power requirements.
Letdown steam from higher levels is available. There is no option to sell power. Ten
operating periods are considered, each one occupying 0.1 year. The utility demands are
shown in Table 6-2. The cost data are given in Table 6-3. The objective is to find the
optimal debottlenecking options and the operational planning for the site utility system.
VHP
HP
MP
LP
,~Io-----
L -_ _ _-..J
,
i OeaeralOl
,
122
Chapter 6
Unit
Operating Conditions
VHPHeader
HPHeader
MPHeader
LP Header
Deaerator
Period
10
Power(MW)
25
35
50
55
45
60
55
60
63
56
VHP (tIh)
HP (tIh)
100
100
130
130
190
130
180
130
130
80
MP (tIh)
100
130
160
210
110
180
170
170
180
220
LP (tIh)
100
120
150
150
190
230
190
190
230
190
Electricity
Cost: O.1$IkWh
Cost: 223$ffon
123
Chapter 6
The existing system is optimised for maximum power over the operation periods. The
results are given in Table 6-4.
Period
10
Power
28.2
35.5
46.4
46.4
45.4
46.4
46.4
46.4
46.4
46.4
Table 6-4: Maximum power generation of the existing system of Case 1(MW)
The amount of steam for each expansion zone is calculated from the steam demand data
of Table 6-2. The amount of steam across each expansion zone is given in Table 6-5.
Period
10
VHP-HP
300
350
440
440
490
540
540
490
540
490
HP-MP
200
250
310
360
300
410
360
360
410
410
MP-LP
100
120
150
150
190
230
190
190
230
190
124
Chapter 6
Expansion zone
Tll
Tl2
T2l
T22
T 23
VHP-HP
80
90
50
70
150
70
150
90
HP-MP
150
MP-LP
Table 6-6: Capacities of simple turbines in every expansion zone of Case l(tlh)
The sizes of the potential steam turbines are determined by comparing the maximum
steam loads of the existing turbines and the total amount of steam across each expansion
zone. The capacities are given in Table 6-7.
Period
10
VHP-HP
50
100
100
50
100
50
HP-MP
50
100
50
50
100
100
MP-LP
40
80
40
40
80
40
Once the capacities are determined, efficiencies are calculated for the existing system
and the system with new units. The TECs are constructed for all periods. These are
given in Figure 6-13. On the basis of the TECs and the power demands, the options
should consider boiler and back-pressure turbine (BBPT) cycles and gas turbine and
waste heat boiler (GTWB) cycles. The capacities of the back-pressure turbines are given
in Table 6-7. The capacities of the GTWB cycles are given in Table 6-8.
125
Chapter 6
Period 1
'1.
Period 6
Exlltrog ...._
1].
ExJsIing....-
B6P!
,,,
,
,,
-j
2S.2MW
-,
'6._
-.'(MW]
' (MW]
Period 2
Period 7
Exlltrog ...._
'1.
1].
ExIo1IrQ ...._
asP!
,
,,,
,,
,
--j
19.7MW I
..
f--
~~ . ~
_2
' (MW]
!-TO-I
_
Period 3
'1.
ExIatIrQ ....'.."
G1W1I
, ,
i ~o_,
, G1W8CT
HI acT
:,
7.'l'lMW
'P
'6._
-- ~
3.6MW
_3
P(MW)
Period 8
'1.
EJdIIIrQ .........
46.4MW
,' ,~T
,H ,
""TO '
'P
6 .61MW
_8
P(MW]
P(MW]
Period 9
Period 4
1].
'1.
ExIltIrQ ...._
asP!
EJdIIIrQ .........
46. 4MW
4b.4MW
P(MW]
Period 10
PeriodS
asP!
,,
,
,,
,
--j
46._
.---4"'6.-_:=----,.1 0 ,,~
f- _._,
_5
~========~~~--~
_ '0 P(MW]
P(MW]
126
Chapter 6
Period
10
Capacity
3.58
5.81
5.61
The discretisation procedure (Section 5.5.1) is applied to determine the capacities and
number of the back-pressure turbines for debottlenecking. The capacities are given in
Table 6-9.
Zone
BTl
BT2
VHP-HP
50
100
HP-MP
50
100
MP-LP
40
80
From Table 6-8, the capacities of the gas turbines in periods 4 and 8 are close.
Therefore, the size 5.61 is dropped from the superset. The discretisation procedure
(Section 5.5.2) is used to determine capacities of the potential gas turbine and waste
heat boiler cycles. These capacities are given in Table 6-10. Hence, resulting
superstructure is generated and shown in Figure 6-14.
GTWB
GT I
GT2
GT3
Capacity
3.58
5.81
2.23
The MILP model minimises the total annual cost. The model is developed using GAMS
(Brooke et al . 1992) and the optimisation has been conducted by using the full space
search (Branch and Bound) as well as the proposed decomposition method (Section
6.6.2). The optimisation yields identical solutions. However the full space search
127
Chapter 6
appears less efficient. Table 6-11 summaries the size of the original problem and the
decomposition result with respect to the subproblems discussed earlier.
Max2 .23MW
VHP
HP
MP
LP
~I""""I-
Original problem
Debottlenecking
Operational planning
problem
problem
0-1 variables
170
70
160
Cont. variables
379
370
379
Constraints
479
419
479
128
Chapter 6
The optimal structure is shown in Figure 6-15. It consists of three new back-pressure
steam turbines, one new gas turbine and a new waste heat boiler. The back-pressure
turbines exploit the cogeneration potential and since they are unable to generate enough
power the gas turbine and waste heat boiler are employed to address the remaining
power demands. The total annual cost is 59.77 M$. The annual fuel cost is 56.95 M$
and the annual capital investment cost is 2.82 M$. The resulting optimal capacities are
given in Table 6-12.
VHP
HP
Max90t/h
Max 70tJh
MP
Max l5Ot1h
LP
COND.
129
Chapter 6
Unit
Capacity
VHP-HPBT2
2.88MW
HP-MPBT\
2.77MW
MP-LPBT2
8.20MW
5.81 MW
Boiler B2
100t/h
The optimal loads of all units are given in Table 6-13. Boilers Bl and B2 are operating
to raise steam in all periods because of the high steam demand from the site processes.
The steam turbines are employed to fully use the potential for cogeneration and generate
power to meet the demand of the site processes. The gas turbine is shut down unless the
power outputs of the steam turbines are unable to meet the demand for power.
Period
10
Til (t/h)
80
80
80
80
80
80
TI2 (t/h)
90
90
90
90
90
90
90
90
90
90
T2\ (t/h)
30
50
50
50
Tn (t/h)
70
70
70
70
70
70
70
70
T 23 (t/h)
87
120
150
150
140
150
150
150
150
150
BT 2 (VHP-HP) (t/h)
100
60
100
100
100
100
100
100
100
100
BT I (HP-MP) (t/h)
23
40
50
50
50
50
50
50
13
50
80
40
40
80
40
BJ (t/h)
200
250
340
340
390
440
440
390
440
390
B2 (t/h)
100
100
100
100
100
100
100
100
100
100
GT2 (MW)
4.13
5.73
5.81
1.73
1.87
130
Chapter 6
levels and the de aerator are shown in Table 6-14. The steam used for heating is returned
as condensate. Letdown steam from higher levels is also available.
Jc;D
.
Condenser
131
Chapter 6
The site utility system is connected with several chemical processes. There is no option
to sell power. The chemical processes are expanded in three periods and the utility
demands for each period are given in Table 6-15. The three periods span over 0.4, 0.3
and 0.3 year respectively. The power demands of the allocated turbines for each period
are given in Table 6-16. The cost data of the utilities are given in Table 6-17. The load
on the local boilers PI and P2 are 50tlh and 59.6 tIh respectively for all periods.
The current utility system can't satisfy the increasing heat and power demands of the
chemical processes. The optimum investment scheme for new units to be added into the
current system and the operational strategies are optimised so that the utility demands
are met at a minimum cost.
Operating parameters
Unit
VHPheader
Temperature: 550C
lIP header
Temperature: 236C
MP header
Temperature: 195C
LP header
Temperature: 144C
VLPheader
Temperature: 110C
Condenser level
Temperature: 50C
Deaerator
Period
Period 1
Period 2
Period 3
Electricity (MW)
168.6
239.1
338
171
181
220
MP steam (tIh)
133
183
267
LP steam (tIh)
108
159
170
54
94
120
132
Chapter 6
Period
Period 1
Period 2
Period 3
1.6
5.4
0.0
9.0
0.0
9.0
30.8
35.5
43.0
Table 6-16: Power demands of the allocated turbines of the industrial case
Period
Period 1
Period 2
Period 3
Demineralized water
0.47 $/ton
0.47 $/ton
0.47 $/ton
Electricity
O.I$/kWh
O.I$/kWh
O.I$/kWh
11.05 $/mmkcal
11.05 $/mmkcal
11.05 $/mmkcal
11.22 $/mmkcal
11.22 $/mmkcal
11.22 $/mmkcal
14.56 $/mmkcal
14.56 $/mmkcal
14.56 $/mmkcal
245 $/ton
245 $/ton
245 $/ton
16.1 $lMWh
. 16.1 $IMWh
16.1 $IMWh
Table 6-17: Utility cost data in different periods for the industrial case
Once the existing utility system is optimised, maximum power outputs and total steam
flowrates across expansion zones are determined for all operating periods. The results
are given in Table 6-18 and Table 6-19.
Period
Period 1
Period 2
Period 3
Power output
172
172
172
Table 6-18: Maximum power generation of the existing system of the industrial case (MW)
133
Chapter 6
Steam zone
Period 1
Period 2
Period 3
Zone 1
1204
1509
1852
Zone 2
991
1262
1538
Zone 3
748
939
1098
Zone 4
551
669
791
ZoneS
356
396
450
Table 6-19: Total steam flowrates across expansion zones for the industrial case (t/h)
For our case there is no surplus steam for the available levels. Therefore no condensing
turbines are considered for debottlenecking. To determine the capacities of the potential
back-pressure turbines, the capacities of the existing turbines are required. The
suggested capacities of the existing turbines in each steam zone are given in Table 6-20.
Tl
T2
T3
T4
T5
T6
Zone 1
215
420
568
Zone 2
215
400
214
48
Zone 3
112
270
214
48
Zone 4
214
48
400
Zone 5
80
48
400
Table 6-20: Capacities of existing turbines in every expansion zone for the industrial case (t/h)
The maximum steam loads across the existing turbines and the total steam across the
expansion zones determine the capacities of the potential steam turbines. The capacities
are given in Table 6-21.
134
Chapter 6
Period 1
Period 2
Period 3
Zone 1
306
649
Zone 2
162
433
709
Zone 3
104
295
455
Zone 4
121
Zone 5
Table 6-21: Capacities of potential steam turbines for each period for the industrial case (tlh)
The development ofTECs is shown in Figure 6-17. The debottlenecking options include
the boiler and back-pressure turbine (BBPT) cycles and the gas turbine and waste heat
boiler (GTWB) cycles.
'1.
1'1 ,
ExlSr.g .,.""""
Period 3
Perbd 2
Perbd 1
1'1,
BBPI
ExIling.,.".",
ExlSr.g .,.""""
88P!
GlWB
r I
I I
I I
H G1W8CT
~ ~T
I I
I I
:-, IP
In"""
InWi
H
168.6 _
1--1159 .4_
I I
I'~
239.1 _
P IMNj
--I
7.7Wi
172 MIl
PI""")
135
In
PIWi)
Chapter 6
Period 1
Period 2
Period 3
6.6
62.1
Table 6-22: Capacities of GTWB cycles for different periods for the industrial case
The synthesis superstructure is shown in Figure 6-18. The shaded units represent the
new units.
The MILP model minimises the total annual cost. The model consists of 166 binary
variables, 351 continuous variables and 462 constraints.
136
Chapter 6
The optimal structure is shown in Figure 6-19. The solution includes seven new bac kpressure steam turbines (NTI to NT7), two new gas turbines (OT 1 and OT2) and one new
waste heat boiler (NB). The capacities of the units are given in Figure 6-19. As shown
in the Figure, two small gas turbines are selected instead of a bigger gas turbine. The
result contradicts conventional wisdom that would advocate a bigger turbine. For the
low power demands , the smaller gas turbine has higher efficiency than a bi g partly
loaded gas turbine. Simjlarly, in expansion zones VHP-HP, HP-MP and MP-LP two
small steam turbines are preferred instead of a bigger turbine. In periods 1 and 2 there is
not steam between the LP level and the VLP level and a single turbine is se lected to
exploit the cogeneration potential in period 3.
Max S7.SMW
r.
Max20.3MW
Max 6.1MW
[::j
I
Make-up wale,
I
Figure 619: Optimal structure of the industrial case
137
Chapter 6
The optimal loads are given in Table 6-23, Table 6-24 and Table 6-25. In period 1, the
existing system satisfies all power demand. The new steam turbines and the gas turbines
are all shut down. In period 2, some of the new steam turbines and the smaller gas
turbine are active. In period 3, the power demand reaches its maximum value. All
turbines are in use generating power to satisfy the additional power demand.
Tl
Unit
Period 1 1.6
T4
T3
IT5
T2
29.0 67.S 71.9 9.0
T6
30.8
NT!
NT z NT3
0.0
0.0
NT4
0.0
NT'!
0.0
NT6
0.0
NT,
0.0
0.0
Period 2
5.4
0.0
67.S 63.2
0.0
35.5
0.0
57.0
0.0
20.1
0.0
IS.1
6.1
Period 3
0.0
0.0
67.S
67.0
9.0
43.0
26.6 57.0
7.S
20.3
5.2
IS.1
6.1
Table 6-23: Optimal operation of all turbines of the industrial case (MW)
Unit
PI
P2
Bl
B2
B3
NB
Period 1
50
60
170
425
550
Period 2
50
60
170
425
588
266
Period 3
50
60
170
425
588
610
Table 6-24: Optimal operation of all boilers of the industrial case (t/h)
Unit
GT!
GT 2
Period 1
0.0
0.0
Period 2
6.6
0.0
Period 3
6.6
55.5
Table 6-25: Optimal operation of all gas turbines of the industrial case (MW)
138
Cluzpter6
6.B Conclusions
139
Chapter 7
Chapter 7
Total Site Maintenance Scheduling
7.1
Introduction
Maintenance assumes an important part in the operation of a total site system. It reduces
the risk of capacity outrage and improves the availability of units. The maintenance
scheduling problem minimises the overall operating cost over the given operating
period subject to maintenance and system hardware and reliability constraints.
Scheduling methods have been proposed by Chattopadhyay et al. (1995) and Dapazo et
al. (1975), but they only considered power systems and assumed identical maintenance
needs for the system units. A typical chemical plant usually consists of several chemical
production processes that consume heat and power in order to make products. The
central utility systems supplies the heat and power of the processes that in tum have to
shutdown and start-up allowing, sometimes significant changes in the demands. Byproduct fuels of site process units can also be used to generate heat and power enabling
strong interactions between the site utility system and the processes. The optimal
maintenance schedule of the site utility system and the process units consequently
allows for a simultaneous consideration of the options, as it is schematically shown in
Figure 7-1.
This approach assumes given shutdown, start-up and maintenance periods for each unit
and applies a multi-period MILP model to develop the optimal maintenance schedule.
The optimisation further determines optimal switches for the fuel and economic
schemes for power import/export. The maintenance periods are allowed to be different,
the optimisation considers the maintenance of the chemical processes and the site utility
system, and the decisions are made to minimise the total operating cost.
140
Chapter 7
Time
7.2
Mathematical Model
Sets
EB
I
Ie
IG
IT
p
U
UF
141
Chapter 7
Parameters
a, b
CEPk
CESk
CFuk,k
Cp
Cp
Cpa
ci
CWk
DIp
D2p
D3p
EISit
Fj{max
Mil
MINu
MAXu
MINp
MAXp
qeb
yo
142
Chapter 7
T/
sal
I1Teb.1:
Binary variables
=0
Xu
Yu.k
Y1 p .k
Y2 p.k
Y!.k
y/~.o
= 1 denotes
=0
BT.o=o
YII.k
eT
Yic.k
eTk=o
YIC.
GT
Y ig.k
Yf~gT.k() =0
Zu.k
= 0 denotes the utility unit u starts its maintenance in period k, otherwise 2 u.k = 1
ZIp.k
=0 denotes the process unit p starts shutting down for maintenance in period k,
otherwise 21 pj
Z2p k
=1
Continuous variables
EDp.k
f
Fig.k
FWk
B
M "h.k
M eh.l.k
B
Mi:~
M it.l.k
BT.I
M ifBT.O
.I.k
143
Chapter 7
ft
M :.k
: amount of steam throttled through let down valves to steam level I in period k
QB.!
eb.k
QB.w
eb.k
QGT,W
ig,k
Quf,k
SDp,l,k
Wil~
W/
wi;,1'
Wi~r
Wkexp
UY
7.2.1
Objective Function
The objective function minimises the total operating cost. The cost includes the fuel
cost, the boiler feed water cost and the electricity cost / revenue. It is expressed by:
7.2.2
+ LCWkFWk + LCEPkW:UY
keK
keK
Performance Models
These include:
144
LCESkWkexP
keK
1r/
J
j
(7-1)
Chapter 7
(a) Boilers
The fuel requirement of each boiler in period k is provided by the Boiler Hardware
Model (Section 3.2):
(7-3)
The power output of each back-pressure steam turbine in period k is provided by the
Turbine Hardware Model (Mavromatis and Kokossis, 1998a):
BT
W'le
'.
6 1
= --(EIS
il
5B
/I
.4;,
M BT,max
.,
)(M BT
i/.k
It E
IT k
,
(7-4)
BT
M i/.1e -
BTMBT,max < 0
YiI.k
il
-,
It
IT , k E K
(7-5)
The power output of each condensing turbine in period k is given by the Condensing
Turbine Hardware Model (Section 3.3):
145
Chapter 7
CT
Wie,k
6_1_(ElS C
5 Be
ie
A;~
M CT,max
IC
)(M CT
ie,k
_.!..MCT,max
ic
CT)
Yic,k'
IC
IC k
,E
(7-6)
IC
CT
M ic,k
_MCT,max
ic
CT
< 0,
Yic,k -
'
~CE
IC ,kE K
(7-7)
The power output of each gas turbine in period k is given by using the Gas Turbine
Hardware Model (Section 3.4):
W GT
ig,k
1 (M!
= Bg
!
- F1,max
)1 + )F!
ig,k -
GT) ig E IG,kE K
nFf,max
ig
Yig,k'
(7-8)
Ig
!
Fig,k
F!
,max GT < 0
'
IG ,kE K
ig
Yig,k ,lgE
(7-9)
The waste heat from the gas turbine can be obtained by:
Q GT,,,,=[~caTa+C!Tf+M!
f
Ig,t
-(I+n)_I_(Mf
- Ag )]F.!
Bg
!
FI,max
Ig,k'
igEIG,kEK
Ig
(7-10)
The utility demands of the process units are formulated as general functions of the
operation conditions (i,e, start-up, shut down) as follows:
146
Chapter 7
(7-11)
ED p.k = !edp.k(Ylp.k,Y2p.k,Dlp,D2p,D3p),
(7-12)
pe P,ke K
(7-13)
Steam inputs at each team level include steam raised by boilers, steam exhausted from
BP steam turbines and steam throttled through let down valves. Steam outputs at each
steam level include steam loads to BP steam turbines and condensing turbines, steam
throttled through let down valves and steam loads to site processes. The mass balances
of all steam levels give:
i/Err
i/Err
ebEEB
Ie L,ke K(7-14)
peP
IIErr
eleCT
ige/G
peP
147
kE
(7-15)
Chapter 7
7.2.5
Maintenance Constraints
Within a given time period from M1Nu to MAXu, utility unit u needs to start its
maintenance:
MAXu
UE
(7-16)
k=MIN.
Within a given time period from MINp to MAXp, process unit p needs to shut down for
maintenance:
MAXp
pE P
(7-17)
k=MINp
After shutting down and maintenance, process unit p needs to start up for operation. The
start-up should start in a time period between MINp + Dlp+D2p and MAXp+ Dlp+D2p:
pE P
(7-18)
k=MIN p+D1p+D2p
Beyond the given time period from MINu to MAXu, utility unit u doesn't start its
maintenance:
Zu.k
=1,
148
(7-19)
Chapter 7
Beyond the given time period from MINp to MAXp, process unit p doesn't shut down for
maintenance:
(7-20)
Beyond the time period from MINp + DIp+D2p to MAXp+ Dlp+D2p, process unit p
doesn't need to start up for operation:
(7-21)
Yu.* = 1- k + LZu,j,
1:5 k :5 Du ,u E U
j=1
(7-22)
Yu,k = 1- Du + LZu,j'
j=k+I-D.
The constraints ensure that start up and shut down times develop within the appropriate
time, This means once time is allocated for shut down or start-up operation, the
operation complete without interruption,
149
Chapter 7
1 ~ k ~ Dl p ,pe P
(7-23)
I:
Yl p,l: = 1- Dl} +
LZl
p,} ,
DI p
N, peP
}=Ie+l-D1p
Y2 p ,A:
=l-k + LZ2
1 ~ k ~ D3 p ,pe P
p ,j,
j=1
Y2 P,A
L
=1- D3 }. +
(7-24)
Ie
~ Z2 P,}.,
J
D3 p
N, peP
j=le+l-D3 p
4. Logical constraints:
Before they shut down, process units cannot start up. Therefore,
k-1
pe P,ke K
(7-25)
j=1
After the maintenance of process units is completed, the units start up immediately:
iE U,u E U,k E K
(7-27)
(7-28)
150
Chapter 7
5. Co-ordination constraints:
There is a maximum number of units that can be maintained in each period. Therefore.
ke K
(7-29)
ueU
UMAX
6. Resource constraints:
A separate set of constraints ensure the resources assigned for maintenance is not
exceeding available limits:
=1,2, ...M
(7-30)
ueU
where:
Ru,m
RESm.k
=amount of m
=number of resources
7.3
th
The optimisation model that is used for maintenance scheduling is illustrated with a
case study. Figure 7-2 shows the configuration of a total site. The site consists of four
steam levels, four boilers, five steam turbines, the deaerator and one gas turbine. The
utility system serves four process units. The capacities of the utility units are given in
Figure 7-2. The operating conditions of steam levels are shown in Table 7-1. The steam
used for heating can be returned as condensate. The process units operate in normal
conditions and the utility demands from each unit are given in Table 7-2. The utility
151
Chapter 7
plant is interconnected with the utility grid. Cost data for the utilities are given in Table
7-3.
~ .
~- ....... -- ....-....... --- .. .
~--
VHP
HP
MP
:
!
I
I
I
I
I
l
I
!
---------------------------------r------- ----------r-------- .---------r--- - -------------)
Export/,mpn,9ctricitY
"
LP
.---.l.-.!..L.!.....L...,
COND.
Steam levels
Operating Conditions
VHP level
HP level
MP level
LP level
Thirty-six operating periods are considered, each occupying 10 days . The total time
hori zon is one year. Process and utility units all undergo maintenance. Process P 4 takes
longer times to start up and shut down and this all has been taken into account by the
model. It takes Process P 4 10 days to start up and 10 days to shut down . Maintenance
times are given in Table 7-4. No more than 5 units can be maintained at a given period.
152
Chapter 7
The optimisation simultaneously determines the optimal maintenance and the optimal
operating schedules for the utility system and the process plant.
Process unit
PI
P2
P3
P4
Power(MW)
20
65
35
100
40
240
300
40
60
400
MP steam (tJh)
50
200
LP steam (tIh)
150
100
Fuel used by B 1
Fuel used by B2
O.OI$IkWh
0.OO98$/KwH
Fuel Used by
Purchased
Sold
B3 andGT
electricity
Electricity
0.0 13$IkWh
O.l$IkWh
0.085$/kWh
Unit
BI
B2
B3
TI
T2
T3
T4
Ts
GT
PI
P2
P3
P4
Duration
20
20
20
20
20
10
20
10
20
10
10
10
30
The scheduling MILP model involves 1200 integer variables, 1320 continuous variables
and 2453 constraints.
The optimisation results are given in Figures 7-3 to 7-8. The minimum operating cost is
78.3 M$. The optimal maintenance and operation schedule is shown in Figure 7-3,
where shaded M-bars account for maintenance, I-bars for idle, St-bars for start-up and
Sh-bars for shutdown times. Processes PI, P 2 and P3 are shut down for maintenance in
periods 26, 29 and 27 respectively. Process P4 is shut down in period 25. It is
maintained during periods 26, 27 and 28 and starts up in period 29. Most utility units
are maintained during the time processes are maintained. For example, during periods
153
Chapter 7
26, 27, 28 and 29, PI, P2. P3 and P4 are down for maintenance. The optimal profile of
steam consumed by processes is given in Figure 7-4. Figure 7-5 explains the optimal
power consumption of processes. Figure 7-6 shows the optimal boiler operation and
Figure 7-7 gives the optimal turbine operation. The optimal power export profile is
shown in Figure 7-8. The power export is quite stable except for the periods where the
process and utility units are shut down for maintenance. No electricity import is
required during maintenance.
7.4
Conclusions
A multi-period MILP model is presented for the total site. The model can obtain a
minimum cost maintenance and operation schedule. The optimisation method
simultaneously considers the maintenance and the operation of the site processes and
the site utility systems. It is shown that interconnection mode of the utility system and
the process plant can lead to substantial changes in utility demand decisions and the
overall operating costs are reduced significantly. As the site processes start up and shut
down gradually. the start-up and shutdown models of site processes are incorporated in
the optimisation model. The model is capable of accounting for different maintenance
duration of operating units instead of same maintenance duration of the units proposed
by conventional methods. The practical maintenance constraints of industrial plants
have also been modelled in the proposed optimisation model.
154
ChapTer 7
Period
10
12
II
13
14
15
16
17
18
PI
P2
P3
P4
BI
1-
B2
B3
Tl
T2
T3
19
::w
T4
T5
GT
Periods
f/d'
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
!-
36
.-
PI
i-
P2
P3
Sh
P'+
Bl
B2
B3
TI
T2
T3
T-+
TS
GT
Idle
Shut
dow n
Slarl -up
155
Operation
Chapter 7
_ 700
~ 600
"0 500
c:
co 400
E
CI) 300
"0
E 200
co
CI) 100
en 0
,.....
a
~
18
12
24
-+-VHP st.
___ HP st.
MP st.
;\0(
30
LP st.
36
_ 300
-~
250
200
c:
co
150
E
CI)
"0 100
...CI)
50
~
0
a
a..
"0
\ 1
~ ~
~
12
18
24
36
30
700
J? 600
500
ell
~ 400
0
-=... 300
.!! 200
'0
In 100
0
e.
I
r. . . . . . . ,.,~...io, . . . . . . ".:.~
A ..
.....
~.~
I~
..... ., '\0
-+-B1
___ B2
Ii
B3
II"
HRSG
\ ~rv~
0
12
18
24
30
36
156
Chapter 7
120
c 100
0
:;::
...
80
cG)
60
C'::S
G)
C)
...
G)
40
Q.
20
::0
'.,. ~,
O\)()()ooOOOOOOOOOOOv
)(tOE
I
r
... ~
~~-
12
. - ...
I~ ~V / .. .. - . ....
24
18
30
-+-T1
- - T2
T3
T4
-.-T5
-+-GT
36
~
~
120
100
80
&.><
60
40
4)
...
o~
Il.
....
T\
J I \
20
18
12
24
30
157
36
Chapter 8
Chapter 8
Conclusions and Future Work
8. 1 Introduction
This chapter consists of two parts. The first part summarises the work presented in this
thesis. The second part gives some suggestions for future work.
8.2 Conclusions
An integrated approach has been proposed for the analysis and optimisation of total site
utility systems. The approach makes combined use of total site analysis, thermodynamic
analysis and mathematical optimisation techniques, in order to provide a comprehensive
solution to a multi-faced problem. A systematic methodology has been presented for the
design of total site utility systems, with particular emphasis on the anticipated
operational variations. In addition, the approach has proposed systematic optimisation
methods to address the debottlenecking and planning of site utility systems as well as
the maintenance scheduling of total sites.
8.2.1
A set of hardware models CTHM, BHM and GTHM are proposed. The use of the
models for the analysis and optimisation of site utility systems are explained in the
work. The models combine thermodynamic principles, engineering knowledge and
performance data for condensing turbines, boilers and gas turbines. They enable for an
158
Chapter 8
accurate prediction of unit efficiencies and embody the efficiency trends of realistic
units in terms of their variation with capacity, load and operating conditions. These
effects are accounted for in a simple and linear fashion, the importance of which are
illustrated in the work.
The hardware models CTHM, BHM and GTHM provide results of good accuracy, by
considering the dependence of the efficiency on the capacity and operating conditions.
In view of operational variations, the effect of the part-load operation is well accounted
for all these models.
With respect to the analysis and optimisation problem, the CTHM, BHM and the
GTHM models provide the basis for modelling the condensing turbines, boilers and gas
turbines respectively in a manner that the efficiency trade-offs of the various design
alternatives can be considered. The linear relations of the power output to the steam
load of condensing turbines, the steam load to the fuel requirement of boilers and the
power output to the fuel requirement of gas turbines are essential for simple
formulations for the optimisation of site utility systems.
It can be used to describe the interaction between the placement of steam levels and
steam loads of site processes. Based on the transhipment representation and combined
with the BHM and THM models, a multi-period MILP model is applied to minimise the
total utility cost for the total site under multiple operation scenarios. Major decision
variables include the overall fuel requirement, the cogeneration potential and the
cooling utility demand. The MILP model is a general model which can not only be used
159
Chapter 8
for the Minimum Utility Cost (MUC) case but also for the Minimum Fuel Requirement
(MFR) case.
8.2.2
160
Chapter 8
and prices of utilities, the approach determines investment schemes and schedules the
operation for maximum efficiencies.
A multi-period MILP model is presented for the total site. The model can obtain a
minimum cost maintenance and operation schedule. The optimisation method
simultaneously considers the maintenance and the operation of the site processes and
the site utility systems. It is shown that interconnection mode of the utility system and
the process plant can lead to substantial changes in utility demand decisions and the
overall operating costs are reduced significantly. As the site processes start up and shut
down gradually, the start-up and shutdown models of site processes are incorporated in
the optimisation model. The model is capable of accounting for different maintenance
duration of operating units instead of same maintenance duration of the units proposed
by conventional methods. The practical maintenance constraints of industrial plants
have also been modelled in the proposed optimisation model.
In the design problem of this thesis, it is assumed that the design conditions of site
processes are fixed. However, the design of site processes without simultaneously
161
Chapler8
considering the design of site utility systems may not be the optimal design in total site
context. There is a need for a methodology making simultaneous design of site
processes, heat exchanger network and site utility system under operational variations.
The simple models of site processes are applied in the total site maintenance scheduling
approach. More rigorous models of site processes need to be developed.
162
References
References
AI-Khamis, T.M., Vemuri, S., Lemonidis, L. and Yellen, 1., 1992, Unit maintenance
scheduling with fuel constraints. IEEE Transactions on Power System 7,933-939.
Brooke, A., Kendrick, D. and Meeraus, A., 1992, GAMS: A User Guide, Release 2.25,
The Scientific Press.
Bruno, J.c., Fernandez, F., Caste]]s F. and Grossmann I.E., 1998, A rigorous MINLP
model for the optimaJ synthesis and operation of utility plants. Chemical Engineering
Research & Design 76, 246-258.
Chattopadhyay D., Bhattacharya K. and Parikh J., 1995, A systems approach to leastcost maintenance scheduling for an interconnected power system. IEEE Transactions on
Power System 10,2002-2007.
Chen, L.N. and Toyoda, J., 1990, Maintenance scheduling based on two-level
hierarchical structure to equalize incremental risk. IEEE Transactions on Power System
5, 1510-1516.
Chou, c.c. and Shih Y. 1987, Thermodynamic approach to the design and synthesis of
plant utility system. Ind. Engng. Chem. Res. 26, 1100-1108.
Church, E.F., 1950, Steam Turbines, McGraw-Hill Book Company.
Cohen H., Rogers G.F.C. and Saravanamuttoo H.I.H., 1987, Gas turbine theory,
Harlow: Longman Scientific & Technical.
163
References
Colmenares, T.R. and Seider, W.D., 1989, Synthesis of utility systems integrated with
chemical processes. Ind. Eng. Chern. Res. 28, 84-93.
Dhole, V.R. and Linnhoff, B., 1992, Total site targets for fuel, co-generation, emissions
and cooling. Comput. Chern. Engng 17, s101-s109.
Dopazo J. F., and Merrill H. M., 1975, Optimal generator maintenance scheduling using
integer programming. IEEE Transactions on Power Apparatus and System PAS94,
1537-1545.
Garginkel R.S. and Nemhauser G.L., 1972, Integer Programming. Wiley, New York.
Hui, C.W. and Natori, Y., 1996, An industrial application using mixed integerprogramming technique: a multi-period utility system model. Computers Chern. Engng
20, sI577-s1582.
Ito K., Yokoyama R., Akagi S. and Matsumoto Y., 1990, Influence of fuel cost on the
operation of a gas turbine-waste heat boiler cogeneration plant. ASME Journal of Eng.
Iyer, R.R. and Grossmann I.E., 1997, Optimal Multiperiod operational planning for
utility systems. Computers Chern. Engng 21, 787-800.
164
References
Iyer, RR and Grossmann, I.E., 1998, Synthesis and operational planning of utility
systems for multiperiod operation. Computers Chern. Engng 22, 979-993.
Linnhoff, B. and Turner, I.A., 1981, Heat recovery networks: New insights yield big
savings, Chemical Engineering 88, 56-70.
Maia, L.O.A. and Qassim, RY., 1997, Synthesis of utility systems with variable
demands using simulated annealing. Computers Chern. Engng 21, 947-950.
Makwana, Y., 1997, Energy retrofit and debottlenecking of total sites. Ph.D. thesis,
Dept. of Process Integration, UMIST, Manchester, UK.
Marechal, F. and Kalitventzeff, B., 1996, Targeting the minimum cost of energy
requirements: a new graphical technique for evaluating the integration of utility
systems. Computers Chern. Engng 20, s225-s230.
Marechal, F. and Kalitventzeff, B., 1998, Process integration: selection of the optimal
utility system. Computers Chern. Engng 22, sI49-s156.
Moro, L.M. and Ramos, A., 1999, Goal programming approach to maintenance
scheduling of generating units in large scale power systems. IEEE Transactions on
Power System 14, 1021-1027.
165
References
Morton, RJ. and Linnhoff, B., 1984, Individual process Improvements in the context of
site-wide interactions. IChemE Annual Research Meeting, Bath, UK.
Nath, R. and Holliday, J., 1985, Optimizing a process plant utility system. Mechanical
Enginreering 44, 44-50.
Nishio, M., 1977, Computer aided synthesis of steam and power plants for chemical
complexes. Ph.D. thesis, The University of Western Ontario, London Canada.
Nishio, M, Itoh, J., Shiroko, K & Umeda, T., 1980, A thermodynamic approach to
steam and power system design. Ind. Engng Chern. Process Des. Dev. 19,306-312.
Pattison and Sharma, 1980, Selection of boiler plant and overall system efficiency.
Studies in Energy Efficiency in Buildings, British Gas.
166
References
Peterson, J.F. and Mann, W.L., 1985, Steam system design: how it evolves. Chemical
Petroulas, T. and Reklaitis, G.V., 1984, Computer aided synthesis and design of plant
utility systems. A.I.Ch.E. J. 30 (1), 69-78.
Raissi. K .. 1994. Total site integration. Ph.D. thesis, Dept. of Process Integration.
UMIST, Manchester, UK.
Salisbury, 1.K., 1942, The steam turbine regenerative cycles - an analytical approach.
Satoh, T. and Nara, K., 1990, Maintenance scheduling by using simulated annealing
method. IEEE PES Summer Meeting. Minneapolis, MN, USA.
Townsend, D.W. and Linnhoff, B., 1983, Heat and power networks in process design.
Part I: Criteria for placement of heat engines and heat pumps in process networks. Part
II: Design procedure for equipment selection and process matching. A. I. Ch.E. J. 29 (5),
742-771.
Wilkendorf, F., Espuna, A. and Puigjaner, L., 1998, Minimization of the annual cost for
complete utility systems. Chemical Engineering Research & Design 76, 239-245.
Yokoyama R., Ito K. and Matsumoto Y., 1994, Optimal sizing of a gas turbine
cogeneration plant in consideration of its operational strategy. ASME Journal of
Yokoyama R. and Ito, K., 1996, Operational strategy of a cogeneration system under a
complex utility rate structure. ASME Journal of Engineering for Gas Turbines and
167
Appendix A
Appendix A
Correlations of the THM
A.I. Correlationfor Ml is
aT
aT sa, as:
Sa/
Ml - - - - - is - 1854-1931q ill
(Ai)
Mavromatis and Kokossis (1998a) provided correlation equations for parameters A and
B. All these equations use the saturation temperature of steam at turbine inlet pressure
rat.
for wmax < 1.2MW
A = -0.0131 + 0.00117T sa'
168
(A2)
(A3)
Appendix A
(A4)
(AS)
169
Appendix B
Appendix B
Regression of Condensing Turbine Efficiency Data
The regression parameters used in the condensing turbine hardware model are derived
from typical efficiency data as shown in Figure 3-6. By definition the maximum
efficiency is:
llis.max
max
(Bl)
= b.H.... M max
hence:
_
MlisM
max
E max
=--
(B2)
1]is.max
These curves are regressed by an expression of the form for each inlet pressure:
max
Ml ;.,M max
E
=- =A
+B
max
(B3)
T/;s.max
Parameters AC and Be are extracted by regression for each inlet pressure. The plots of
parameters AC and BC against that saturation temperature corresponding to the inlet
pressure are given in Figure Bl and Figure B2 respectively. These parameters are in
tum approximated by the following expressions:
=-0.0896 + O.0013T
=1.1752 +0.OOO3T
170
sar
sQr
(B4)
(B5)
Appendix B
where the inlet saturation temperature is in DC. However, the above expressions don't
give accurate estimates of the maximum efficiency when the power output is below
1.5MW. More accurate estimates for the efficiency are obtained by conducting
regression below 1.5MW and above 1.5MW. The two segments' regression gives
efficiency estimates within 3% error. The corresponding expressions for the regression
parameters are:
for E max < I.5MW
(B6)
(B7)
(B8)
(B9)
AC
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
150
250
200
300
350
171
Appendix B
1.35
-r-------------------,
1.3
1.25
1.2 +------r-------,----...,..-----I
300
350
150
200
172
Appendix C
Appendix C
Regression of Gas Turbine Efficiency Data
The regression parameters used in the gas turbine hardware model are derived from
typical electrical efficiency data, such as the plot in Figure 3-8. The curve fitting of the
plot in Figure 3-8 yields the following expression:
wmax
(Cl)
The parameters Ag and Bg are obtained by conducting regression for the curve in Figure
3-8. The resulting values of Ag and Bg are 6.7571 and 2.4381 respectively. But the
above expressions do not give accurate estimates of the maximum efficiency when the
power output is below 6.9MW by using the parameters. More accurate estimates for the
efficiency are obtained by conducting regression below 6.9MW and above 6.9MW. The
two segments' regression gives efficiency estimates within 2% error. The corresponding
regression parameters are:
for W max < 6.9MW
g
=2.0836
(C2)
Bg
=3.1724
(C3)
=8.817
B g = 2.3905
(C4)
173
(C5)