Sunteți pe pagina 1din 164

Techno-enviro-economic evaluations of biomass

gasification for the production of ammonia from


synthesis gas utilizing multi-scale modeling

Submitted in partial fulfilment of the requirements


of the degree of
Doctor of Philosophy
of the
Indian Institute of Technology Bombay, India
and
Monash University, Australia
by

Pratham Arora
Supervisors:
Prof. Sanjay Mahajani (IIT Bombay)
Prof. Anuradda Ganesh (IIT Bombay)
Prof. Andrew Hoadley (Monash University)

The course of study for this award was developed jointly by


Monash University, Australia and the Indian Institute of Technology, Bombay
and was given academic recognition by each of them.
The programme was administrated by The IITB-Monash Research Academy

2017
Declaration

I declare that this written submission represents my ideas in my own words and where others’
ideas or words have been included, I have adequately cited and referenced the original sources. I
also declare that I have adhered to all principles of academic honesty and integrity and have not
misrepresented or fabricated or falsified any idea/data/fact/source in my submission. I understand
that any violation of the above will be cause for disciplinary action by the Institute and can also
evoke penal action from the sources which have thus not been properly cited or from whom proper
permission has not been taken when needed.

Notice 1

Under the Copyright Act 1968, this thesis must be used only under the normal conditions of
scholarly fair dealing. In particular no results or conclusions should be extracted from it, nor should
it be copied or closely paraphrased in whole or in part without the written consent of the author.
Proper written acknowledgement should be made for any assistance obtained from this thesis.

Notice 2

I certify that I have made all reasonable efforts to secure copyright permissions for third-party
content included in this thesis and have not knowingly added copyright content to my work without
the owner’s permission.

Pratham Arora

IITB Roll No.: 114174002

Monash ID no.: 23758856

Date: 23rd March 2017


अज्ञानतिमिरान्धस्य ज्ञानाञ्जनशालाकया

चक्षुरुन्िीमलिं येन िस्िै श्रीगुरवे निः

Salutation to the noble Guru, who has opened


the eyes blinded by darkness of ignorance
with the collyrium-stick of knowledge.
DEDICATED……

TO MY PARENTS

for their love, faith, patience and understanding


Abstract
The world is steadily moving toward greener alternatives of established chemical production
processes. Process modeling and simulation play a vital role in conceiving, designing and
optimizing these greener processes. In many cases, these simulation tools are limited to a few
project design stages, thus limiting their true potential. This thesis presents the benefits of process
modeling and simulation at different levels of process development. The work focuses on ammonia
production, which sustains global agricultural and mining activities. Ammonia production has
traditionally been based on large-scale plants. The thrust toward large-scale production in order to
gain economic advantages has overshadowed the benefits that could be derived from small-scale
production plants. Additionally, the ammonia industry is a major consumer of global fossil fuels,
and in the process of providing ammonia, also burdens the planet with greenhouse gases. In order
to ensure sustainable ammonia production, this thesis highlights the techno-economic advantages
that result from small-scale ammonia plants based on biomass feedstock. The use of biomass as a
potential feedstock can help neutralize the carbon footprint of ammonia production. To establish
the feasibility of such a process, this thesis presents the application of process simulations at
different modeling levels.

A Computational Fluid Dynamics (CFD) model was used to understand the flow patterns inside a
dual fluidized bed gasifier. This elevated the understanding of the hydrodynamics of the gasifier
freeboard, which is neglected by the conventional two-phase methodology. The CFD simulation
was utilized to perform a Residence Time Distribution (RTD) analysis of the reactor. Four different
tracer approaches namely the frozen velocity approach, the snapshot approach, the data sampling
approach and the transient approach, were compared. The RTD analysis formed the basis of a
steady-state compartment model that was developed in ASPEN Plus simulation software. The
ASPEN Plus gasifier model decoupled the pyrolysis, gasification, and combustion sections of the
gasifier to affect a better comprehension of the process and the results. The model predicated
satisfactory results upon validation. Additionally, the model could also be used to predict the
output for different biomass feedstocks.

ASPEN Plus is used to simulate the biomass gasification process as well as downstream gas
conditioning that leads to ammonia production. Apart from major constituent gases, the model
accounted for tar production in the gasifier process as well as for its subsequent removal. Three
different configurations of the biomass-to-ammonia process, namely, Autothermal Reforming
(ATR), Steam Methane Reforming (SMR), and CO2 Reforming (CR) have been modeled and
compared. The configurations differ on the basis of the reforming of the methane and tars, which
are present in the syngas. The output of the ASPEN Plus simulation is fed into an MS Excel-based
framework, which performs Life Cycle Costing (LCC) and Life Cycle Assessment (LCA) for the
simulated flowsheet. Finally, these outputs are utilized by an MS Excel-/VB-based Multi-
Objective Optimization (MOO) framework in order to optimize both LCC and LCA for changing
process parameters.

Three biomass feedstocks, namely, wood, straw pellets, and bagasse, which are available in
Australia, India, and Brazil, respectively, have been compared in order to understand the effect of
biomass composition, the supply chain, and the national economic and environmental scenario on
the viability of the biomass-to-ammonia process. The MOO results that were derived for changing
flowsheet configurations as well as for different biomass feedstocks that were grown at different
locations worldwide are expected to highlight a more rigorous analysis of the biomass-to-ammonia
process. Such an analysis will provide decision makers with consistent and comparable data to
objectively judge the viability of the proposed process. This analysis distinguishes this thesis from
past studies that focused on ammonia production from biomass, and presented their LCC and LCA
results for a single process configuration and for a single feedstock. The results predicted that
different biomass feedstocks, process configurations, and geographical regions have their own
advantages. The results provide a benchmark for comparison of other biomass-to-ammonia
processes on a techno-enviro-economic scale. The methodology can be employed as an example
in the development of different sustainable chemical production processes.
Table of Contents

Table of Contents ........................................................................................................................... i

List of Figures ............................................................................................................................... iv

List of Tables ............................................................................................................................... vii

Nomenclature ............................................................................................................................... ix

Chapter 1: Introduction ............................................................................................................... 1

1.1 Context ............................................................................................................................ 1

1.2 Biomass as feedstock ...................................................................................................... 2

1.3 Ammonia production ..................................................................................................... 4

1.4 Biomass gasification ....................................................................................................... 6

1.5 Research objectives and major contributions ............................................................. 9

1.6 Organization of thesis .................................................................................................. 11

Chapter 2: Literature review ..................................................................................................... 12

2.1 Introduction .................................................................................................................. 12

2.2 Gasifier modeling ......................................................................................................... 12

2.3 Biomass-to-ammonia process ...................................................................................... 16

2.4 Multi-objective optimization ....................................................................................... 19

2.5 Summary ....................................................................................................................... 23

Chapter 3: Gasifier Modeling .................................................................................................... 25

3.1 Introduction .................................................................................................................. 25

3.2 CFD model .................................................................................................................... 27

3.3 RTD analysis ................................................................................................................. 31

3.4 ASPEN Plus model ....................................................................................................... 37

3.5 Results and discussion.................................................................................................. 41

3.6 Conclusions ................................................................................................................... 45

i
Chapter 4: Process modeling ..................................................................................................... 46

4.1 Introduction .................................................................................................................. 46

4.2 ICI Leading Concept Ammonia process .................................................................... 46

4.3 Biomass-to-ammonia process ...................................................................................... 49

4.4 Energy Integration ....................................................................................................... 53

4.5 Process economics ........................................................................................................ 56

4.5.1 Capital cost ............................................................................................................ 56

4.5.2 Operating cost ....................................................................................................... 58

4.6 Results ........................................................................................................................... 60

4.7 Conclusions ................................................................................................................... 63

Chapter 5: Life Cycle Assessment ............................................................................................. 65

5.1 Introduction .................................................................................................................. 65

5.2 Life cycle assessment .................................................................................................... 65

5.3 Results ........................................................................................................................... 73

5.4 Conclusions ................................................................................................................... 79

Chapter 6: Multi-Objective Optimization ................................................................................ 81

6.1 Introduction .................................................................................................................. 81

6.2 Methodology ................................................................................................................. 81

6.3 Multi-Objective Optimization ..................................................................................... 83

6.4 Results ........................................................................................................................... 85

6.5 Conclusions ................................................................................................................... 87

Chapter 7: Discussion ................................................................................................................. 89

7.1 Introduction .................................................................................................................. 89

7.2 Multi-scale approach.................................................................................................... 89

7.3 Process economics ........................................................................................................ 91

ii
7.4 Life cycle assessment .................................................................................................... 93

7.5 Multi-objective optimization ....................................................................................... 94

7.6 Future Work ................................................................................................................. 97

Chapter 8: Conclusions .............................................................................................................. 99

8.1 Future Work ............................................................................................................... 102

Appendix A: Review of different RTD approaches. .............................................................. 103

Appendix B: Pyrolysis correlation constants ......................................................................... 109

Appendix C: ICI Leading Concept Ammonia process .......................................................... 111

Appendix D: Syngas Reforming .............................................................................................. 118

Appendix E: Economic modeling parameters ........................................................................ 120

Heat Integration .................................................................................................................... 120

Costing Parameters ............................................................................................................... 122

Appendix F: Life cycle inventories .......................................................................................... 123

Appendix G: LCA results validation....................................................................................... 126

Appendix H: LCA methodology .............................................................................................. 127

References .................................................................................................................................. 131

List of publications.................................................................................................................... 143

Acknowledgments ..................................................................................................................... 144

iii
List of Figures
Figure 1.1: Characteristics of different types of gasifiers (E4tech, 2009). ..................................... 7
Figure 1.2: Economic viability range of different types of gasifiers (E4tech, 2009). .................... 8
Figure 1.3: Multi-scale modeling methodology for the biomass-to-ammonia process. ................. 9
Figure 1.4: Organization of the thesis. .......................................................................................... 11
Figure 2.1: Typical biomass-to-ammonia process flowsheet. ...................................................... 16
Figure 2.2: Pareto front depicting dominated and non-dominated points. ................................... 20
Figure 2.3: Crossover in genetic algorithms. ................................................................................ 21
Figure 3.1: Dual fluidized bed gasifier ......................................................................................... 26
Figure 3.2: Compartment Model Methodology ............................................................................ 27
Figure 3.3: Fluidized bed solid volume fraction. .......................................................................... 29
Figure 3.4: RTDs obtained from different approaches (a) Frozen velocity approach, (b) Snapshot
approach, (c) Data sampling approach and (d) Transient approach. ............................................ 32
Figure 3.5: RTD (tracer) profile at different reactor surfaces (a) as observed and (b) normalised.
....................................................................................................................................................... 33
Figure 3.6: RTD obtained for middle section of the fluidized bed reactor ................................... 34
Figure 3.7: Velocity vectors in the middle section of the bed. ..................................................... 35
Figure 3.8: Compartment model for the fluidized bed. ................................................................ 36
Figure 3.9: Comparison of RTD obtained from the CFD model, and the RTD obtained from
MATLAB compartment model..................................................................................................... 36
Figure 3.10: ASPEN Plus compartment model. ........................................................................... 37
Figure 3.11: The effect of gasification temperature on the quality of the syngas. ....................... 44
Figure 4.1: Effect of economies of scale on ammonia production processes (Arkley et al., 1988).
....................................................................................................................................................... 48
Figure 4.2: Flowsheet for the biomass-to-ammonia process. ....................................................... 51
Figure 4.3: Biomass-to-ammonia process configuration: ATR, SMR and CR. ........................... 53
Figure 4.4: Electricity generation cycle and process GCC and demonstrating the automation of the
steam production curve. The LHS graph has greater waste heat available and produces more steam
compared to RHS. ......................................................................................................................... 54
Figure 4.5: Biomass-to-ammonia process grass-root level cost. .................................................. 61
Figure 4.6: Ammonia production cost build-up. ........................................................................... 62

iv
Figure 4.7: Comparison of the production prices of ammonia in various studies ........................ 63
Figure 5.1: LCA framework. ........................................................................................................ 66
Figure 5.2: Biomass-to-Ammonia process system boundary. ...................................................... 69
Figure 5.3: ReCiPe methodology (Goedkoop et al., 2009)........................................................... 72
Figure 5.4: GWP for different configurations of biomass-to-ammonia process. ......................... 75
Figure 5.5: Ozone depletion potential (kg CFC-11 eq) for different configurations of biomass-to-
ammonia process. .......................................................................................................................... 78
Figure 5.6: Single score endpoint indicator for biomass-to-ammonia process............................. 79
Figure 6.1: MOO methodology..................................................................................................... 82
Figure 6.2: Evolution of Pareto front with progress in generations. ............................................. 83
Figure 6.3: The Pareto plots that are obtained for the biomass-to-ammonia process, along with the
variation in ammonia production .................................................................................................. 86
Figure 6.4: The Pareto plots that are obtained for the biomass-to-ammonia process along with
variation in the import of electricity. ............................................................................................ 86
Figure 7.1: Sensitivity analysis for biomass feedstock price for the Wood ATR configuration. . 91
Figure 7.2: Sensitivity analysis for biomass feedstock price for the Straw ATR configuration. . 92
Figure 7.3: GWP comparison of biomass-to-ammonia process with conventional ammonia
production. .................................................................................................................................... 95
Figure 7.4: Global Pareto for the biomass-to-ammonia process................................................... 96
Figure 7.5: Sensitivity analysis for PSA H2 recovery. .................................................................. 97
Figure 7.6: Sensitivity analysis for CCS of PSA off-gas. ............................................................. 98
Figure A1: RTD obtained after freezing the velocity field at 20, 25 and 30s, respectively. ...... 103
Figure A2: RTD obtained from the multiple snapshot approach. ............................................... 104
Figure A3: RTD obtained by data sampling approach. .............................................................. 105
Figure A4: RTD obtained from the full transient simulation. .................................................... 106
Figure A5: Tracer flow in (a) frozen velocity profiles and (b) transient systems. ...................... 107
Figure A6: Velocity profile at different points in the geometry. ................................................ 108
Figure C1: ICI Leading Concept Ammonia process flowsheet (Kroschwitz and Seidel, 2000). 111
Figure C2: ASPEN Plus flowsheet for ICI Leading Concept Ammonia process. ...................... 112
Figure C3: Heat exchanger network for the ICI plant. ............................................................... 114
Figure C4: Grand Composite Curve for the ICI process ............................................................ 115

v
Figure C5: GCC for the ICI process (including air preheating) ................................................. 115
Figure C6: Heat Exchanger design made using Aspen Energy analyser .................................... 116
Figure C7: Revised heat exchanger network for the ICI plant. .................................................. 117
Figure E1: Heat integration working methodology. ................................................................... 120
Figure H1: Technology matrix (A) and intervention matrix (B) ................................................ 128
Figure H2: Calculation of scaling vector .................................................................................... 129
Figure H3: Characterization factors of ReCiPe .......................................................................... 130

vi
List of Tables
Table 1.1: Bioenergy potential of different biomass categories (Hoogwijk et al., 2003). .............. 1
Table 1.2: GHG emissions for fossil fuel-based ammonia production in different world regions
(Kool et al., 2012). .......................................................................................................................... 2
Table 2.1: Comparison of literature for ammonia production from biomass. .............................. 19
Table 2.2: Comparison of different genetic algorithms (Konak et al., 2006). .............................. 22
Table 3.1: CFD simulation parameters. ........................................................................................ 30
Table 3.2: Proximate and ultimate analysis of biomass feedstocks. ............................................. 38
Table 3.3: Reaction kinetics for homogenous and heterogenous reactions. ................................. 40
Table 3.4: Simulation results and validation................................................................................. 41
Table 3.5: Simulation results from different biomass feedstocks. ................................................ 42
Table 4.1: Reaction kinetics for the shift reactor, the methanator reactor and the ammonia synthesis
reactor. .......................................................................................................................................... 50
Table 4.2: Operating parameters for the biomass-to-ammonia process. ...................................... 52
Table 4.3: Operating cost parameters for the biomass-to-ammonia process. ............................... 59
Table 4.4: Economic parameters for biomass-to-ammonia process. ............................................ 60
Table 4.5: Flowsheet results and economic predictions for different variants of biomass-to-
ammonia process. .......................................................................................................................... 61
Table 5.1: Overview of consequential and attributional LCA (Thomassen et al., 2008). ............ 67
Table 5.2: Summary of mass and energy flows that were considered in the LCA of the biomass-
to-ammonia process. ..................................................................................................................... 68
Table 5.3: Overview of different LCA perspectives (De Schryver, 2011). .................................. 73
Table 5.4: LCIA of the biomass-to-ammonia process for different process configurations. ....... 77
Table 6.1: Decision variables and their range for biomass-to- ammonia process optimization. .. 83
Table 6.2: Constraints for biomass to ammonia optimization. ..................................................... 84
Table A1: RTD characteristics (frozen velocity approach). ....................................................... 103
Table A2: RTD characteristics (snapshot approach) ................................................................. 105
Table A3: RTD characteristics (snapshot approach) .................................................................. 105
Table A4: RTD characteristics (snapshot approach) .................................................................. 106
Table B1: Wood pyrolysis correlation constants for Equation 3.6 (Abdelouahed et al., 2012;
Dufour et al., 2009a). .................................................................................................................. 109

vii
Table B2: pyrolysis correlation constants for Equation B1 ........................................................ 110
Table C1: Stream table for ICI Leading Concept Ammonia process. ........................................ 113
Table D1: List of reactions considered for syngas reforming..................................................... 119
Table E1: Film heat transfer coeffients for different process streams. ....................................... 121
Table E2: Costing parameters for different pieces of equipments. ............................................. 122
Table E3: Utility cost parameters for the biomass-to-ammonia process. ................................... 122
Table F1: Ecoinvent inventories used for wood-to-ammonia process........................................ 123
Table F2: Ecoinvent inventories used for bagasse-to-ammonia process .................................... 124
Table F3: Ecoinvent inventories used for straw-to-ammonia process ........................................ 125
Table G1: Comparison of LCA results from MS Excel model and SimaPro software. ............. 126

viii
Nomenclature
Symbols

ΔTLMK Log mean temperature difference


Anetwork Heat exchanger area
A Technology matrix
B Intervention matrix
CA Corrosion allowance
CBM Bare module cost
CBMO Bare module cost at base-case conditions
CGR Grassroots level cost
COL Cost of operating labor
COM Cost of manufacture
Cp Equipment cost
CRM Cost of raw materials
CUT Utilities
CWT Waste treatment
D Vessel diameter
d Process vector for normalizing all process flows
f Scale factor for different components
FCI Fixed cost of investment
Fm Material factor
Fp Pressure factor
𝑔⃗ Gravity (m/s2)
g Process emission vector resulting from normalized process flows
hi Heat transfer coefficients (W/m2 K)
Ii Inflation indices
KGS Gas/solid momentum exchange coefficient
KƟS Diffusion coefficient for granular energy (kg/m s)
M Process matrix
m Normalised process matrix
p Pressure (Pa)

ix
P Gauge pressure
q Steam duty
s Scaling vector
Si Scales of the equipment
T Temperature
tmin Thickness of the vessel
v Velocity (m/s)
Yi Mass yield of different pyrolysis products

Greek Symbols

Volume fraction
Density (kg/m3)
Stress tensor (Pa)
Granular temperature (m2/s2)
𝐼̿ Unity matrix
Collision dissipation of energy (kg/s3m)
Transfer rate of kinetic energy (kg/s3m)

Acronyms

$/GJ U.S. Dollars Per Gigajoule


$/t U.S. Dollars Per Ton
1,4-Dcb 1,4 Dichlorobenzene
AER Adsorption Enhanced Reforming
ANFO Ammonium Nitrate/Fuel Oil
ANN Artificial Neural Network
ASU Air Separation Unit
ATR Autothermal Reforming
BFB Bubbling Fluidized Bed
C10H8 Naphthalene
C2H4 Ethene
C2H6 Ethane
C6H6 Benzene

x
C6H6O Toluene
C7H8 Phenol
CaCO3 Calcium carbonate
CaO Calcium oxide
CCS Carbon Capture and Storage
CEPCI Chemical Engineering Plant Cost Index
CFB Circulating Fluidized Bed
CFC Chlorofluorocarbon
CFD Computational Fluid Dynamics
CH4 Methane
CHP Combined Heat and Power
CO Carbon monoxide
CO2 Carbon dioxide
COM Cost of Manufacturing
COP 21 Conference of the Parties Twenty-First Session
CR CO2 Reforming
CSTR Continuous Stirred Tank Reactors
DFBG Dual Fluidized Bed Gasifier
DMOEA Dynamic Multi-Objective Evolutionary Algorithm
EJ Exajoule
EJy-1 Exajoule per Year
FCI Fixed Cost Of Investment
Fe Iron
FICFB Fast Internally Circulating Fluidized Bed
GA Genetic Algorithms
GCC Grand Composite Curve
GHG Greenhouse Gas
GJ/t Gigajoule per Ton
GWP Global Warming Potential
H2 Hydrogen
HEN Heat Exchanger Network
HP High Pressure

xi
ICI Imperial Chemical Industries
IEA International Energy Agency
IPCC Intergovernmental Panel on Climate Change
ISO International Organization for Standardization
KTGF Kinetic Theory of Granular Flows
LCA Life Cycle Assessment
LCC Life Cycle Costing
LCI Life Cycle Inventory
LCIA Life Cycle Impact Assessment
m2 a Square Meter*Year
MINLP Mixed Integer Nonlinear Programming
MJ/t Megajoule per Ton
MOGA Multi-Objective Genetic Algorithm
MOO Multi Objective Optimisation
MW Megawatt
N Nitrogen
N2 Nitrogen
NMVOC Non Methane Volatile Organic Carbon Compound
NPGA Niched Pareto Genetic Algorithm
NPV Net Present Value
NSGA Non-dominated Sorting Genetic Algorithm
NSGA Non-Dominated Sorting Genetic Algorithm
O2 Oxygen
Odt Oven Dry Ton
Odt/d Oven Dry Ton per Day
P Phosporus
PAES Pareto-Archived Evolution Strategy
PESA Pareto Envelope-Based Selection Algorithm
PFR Plug Flow Reactors
PM Particulate Matter
POX Partial Oxidation
PSA Pressure Swing Adsorption

xii
PSD Particle Size Distribution
QET Quasi-Equilibrium Temperature
RDGA Rank-Density Based Genetic Algorithm
RER Europe
RKS-BM Redlich Kwong Soave Boston Mathias
RME Rape Seed Methyl Ester
ROW World Average Excluding Europe
RTD Residence Time Distribution
RWGA Random Weighted Genetic Algorithm
SMR Steam-Methane Reforming
SO2 Sulfur Dioxide
SOO Single Objective Optimization
SPEA Strength Pareto Evolutionary Algorithm
tpa Ton per Annum
tpd Ton per Day
U235 Uranium-235
UDF User-Defined Function
UNFCCC United Nations Framework Convention on Climate Change
VB Visual Basic
VEGA Vector Evaluated Genetic Algorithm
WBGA Weight-Based Genetic Algorithm

xiii
Chapter 1: Introduction

1.1 Context
At the 2015 United Nations Climate Change Conference (COP 21), 195 countries agreed to
reduce their Greenhouse Gas (GHG) emissions “as soon as possible” and to initiate steps to
keep the global warming “to well below 2 °C above preindustrial levels” (UNFCCC, 2015).
This commitment by world leaders emphasizes the dangers of the world continuing with its
present rate of consumption of fossil fuels. Achieving these targets will not be possible unless
there is a reduction in the use of fossil fuel. The IEA has estimated a total primary energy
consumption of 567 EJ for the year 2013; eighty one percent of this consumption was sustained
by fossil fuels (IEA, 2015). With little action to date, the target of limiting global warming to
2 ºC becomes tougher and more expensive with every passing year. In view of the growing
global population and its ever increasing needs, alternative green routes to fossil fuel-based
processes must be adopted. One such potential alternative is the use of biomass, which has been
used in energy applications since earliest civilizations. The growth of biomass acts a carbon
sink, hence, the majority of emissions from its consumption would potentially be carbon-
neutral. The future energy potential from biomass feedstocks has been projected to be between
33 and 1130 EJ per annum (Hoogwijk et al., 2003). The contribution comes from a variety of
biomass, which are shown in Table 1.1.

Table 1.1: Bioenergy potential of different biomass categories (Hoogwijk et al., 2003).
Biomass category Potential bioenergy supply EJy-1(1018 Jy-1)
Biomass production on surplus agricultural 0–988
land
Biomass production on degraded land 8–110
Agricultural residues 10–32
Forest residues 10–16
Animal manure (dung) 9–25
Tertiary residues (organic wastes) 1–3
Total 33–1130

Industrial processes contribute up to 21 percent of global greenhouse gas emissions (Edenhofer


et al., 2014). The emissions are a result of the burning of fossil fuels for energy and for
chemical, metallurgical, and mineral transformations. One such industrial process is ammonia
production, which consumes fossil fuels both as a feedstock and as an energy provider. With a

1
global production of 146 million tons (US Geological Survey, 2016), ammonia forms the
backbone of the nitrogen based fertilizer industry. The average global energy consumption
from the production of ammonia is estimated to be 52.6 MJ/t NH3 (Kool et al., 2012). Natural
gas supplies 70.7 percent of this energy consumption and the remaining is supplied by oil and
coal. The use of fossil fuels is inherently coupled with GHG emissions. The energy
consumption and GHG emissions that are related to ammonia production in different world
regions has been shown in Table 1.2 (Kool et al., 2012).

Table 1.2: GHG emissions for fossil fuel-based ammonia production in different world
regions (Kool et al., 2012).
ton CO2 eq./ton NH3 MJ/ton NH3
Western Europe 2.34 41.6
North America 2.55 45.5
Russia and Central Europe 3.31 58.9
China and India 5.21 64.3
Rest of the world 2.45 43.7
World average 3.45 52.6

Biomass stands as a strong candidate for the replacement of fossil fuels due to its wide
availability and carbon-neutrality. Biomass gasification can yield hydrogen-rich syngas, which
is a suitable feedstock for ammonia production. Although the scale of biomass-fed processes
is generally relatively small because it is based on renewable feedstock, these processes have
the potential of producing ammonia with a low carbon footprint. This thesis assesses the
different leading technologies for biomass gasification in the generation of ammonia from
syngas feedstock. This thesis contains techno-economic evaluations of the identified processes,
and the overall energy and CO2 penalties. These evaluations would identify the best options
with respect to different biomasses, locations and processes.

1.2 Biomass as feedstock


Ammonia production from biomass is seen as an alternative to traditional carbon-intensive
ammonia production processes. Biomass is a generic term that encompasses a variety of
feedstocks. In order to avert the food-versus-fuel debate, this thesis has focused on second-
generation biomass feedstocks, which can be divided into dedicated energy crops, primary
residues, secondary residues, and tertiary residues (Eisentraut, 2010). Dedicated energy crops

2
include short rotation coppices such as eucalyptus, and perennial grasses such as miscanthus.
Primary residues are a result of the harvesting of crops and timber. Crop-processing facilities
such as sugar mills are a source of secondary residues. Tertiary residues consist of post
consumption wastes such as municipal solid waste. Dedicated energy crops are generally labor
intensive and require specialized machines; they are usually available throughout the year.
Primary residues do not normally require any specific machines, but the feedstock is widely
scattered. Secondary feedstocks are localized but may be used for other purposes such as
generation of heat and electricity. Both primary and secondary residues are limited by harvest
cycles. Tertiary residues require special treatment for removal of many unwanted wastes. A
representative biomass from each category, namely, dedicated energy crop, primary residue,
and secondary residue was selected in this thesis and tested for its potential for ammonia
production through gasification. Tertiary residues were not included in this thesis. Eucalyptus
(wood), straw pellets, and bagasse were chosen as prospective biomass feedstocks in order to
represent energy crops, primary residue, and secondary residues, respectively.

Numerous techno-enviro-economic studies that project the benefits of replacing fossil fuels
with biomass feedstocks have been performed. These studies play an influential role in the
conception of new processes. The emphasis of these studies is on the discovery of all
environmental anomalies that are related to a process, before their physical realization.
However, the majority of biomass feedstock-based studies are focused on a single geographical
location (Patel et al., 2016; Cambero and Sowlati, 2014). The economic and environmental
performances of a process in one geographical location are not necessarily relevant to other
parts of the world. Additionally, there is a significant variation in the economic and
environmental scenarios globally. This thesis aims to reflect the difference in pricing and
emissions for any biomass-based green process, which emerge from a variation in the
geographical locations. Three geographical locations were chosen for the selected biomass
feedstocks. Eucalyptus biomass is widely grown in Australia. Additionally, there is also an
impetus to grow trees such as eucalyptus in agricultural lands in order to counter the problem
of dryland salinity (Rutovitz, 2004). India was chosen as a preferred location for straw. India
produces around 97 million tons of straw per annum, of which 22 million tons is left unused
after meeting local demand (Singh et al., 2016). Brazil, being the largest sugarcane producing
country, is a convenient choice for bagasse feedstock. Every year 169 million tons of bagasse
is produced in Brazil (Hofsetz and Silva, 2012).

3
1.3 Ammonia production
Ammonia is often regarded as one of the few chemicals that have changed the course of the
20th century. As a key component of nitrogen fertilizers, ammonia has sustained global
agriculture, and as a raw material in ammonium nitrate explosives, it has become indispensable
to the mining industry. Development of high-pressure turbo-compressors in the late 1960s
facilitated the building of high-pressure ammonia units with production capacities greater than
1000 tons per day (tpd). These large-scale plants steadily dominated ammonia production
worldwide, because they could cater for the demand for ammonia. Due to energy-efficiency
improvements, the energy requirements in these plants have come down from 60 GJ/tNH3 to
28 GJ/tNH3 in the last 50 years. The process arrangement, however, has remained largely
unaltered (Kroschwitz and Seidel, 2000). This phenomenon has overshadowed the other
advantages that may come from small-scale ammonia production. The production of ammonia
from a world-scale ammonia plant often exceeds its local consumption as a fertilizer or for
mining activities. A typical 1000 tpd ammonia plant can meet the ammonia requirements of
many small countries. For landlocked countries that are located at a long distance from the
ports, the freight cost doubles or even triples the cost of ammonia-based products, and thus
limits consumption (UNIDO, 1998). This problem can be solved by decentralized ammonia
plants that have smaller capacities. Smaller plants tend to take advantage of the skid-mounted
“serial” installation concept in which the installations, in a nearly completed form, are
transported from the building site. This counters the absence of necessary industrial
infrastructure in a remote location. Additionally, the smaller plants have shorter construction
times, higher capacity utilization, and lower engineering costs. Ease of training and a quicker
and lesser number of start-ups/shut-downs are other advantages (UNIDO, 1998).

Ammonia production from biomass is usually proposed only at the production level of a world-
scale ammonia plant. Smaller plants have to suffer the effects of economies of scale. However,
biomass has a low energy density when compared to fossil fuels. Transport over large distances
can result in more energy being consumed as transport fuel when compared to the energy stored
in the biomass. 100 km has been recommended as the practical limit for the transport of
biomass (Hamelinck et al., 2005). Biomass supply is also seasonal and each biomass has a
different harvesting time. This poses a significant challenge in obtaining a large-scale supply
of biomass throughout the year. The operation of large-scale biomass plants is thus restricted
to certain specific geographical locations in which a large amount of biomass is economically
available. In such a scenario, small-scale biomass-based plants may be more practical because

4
they can be based on locally available biomass. Additionally, they can cater to local ammonia
demands effectively. A potential application could be found in the mining industry. Ammonia-
based mining explosives such as ANFO are conventionally produced at a small scale due to
safety concerns and relatively small demands. The demand for such explosives originates from
remote mining locations. A small-scale biomass-to-ammonia plant can, thus, easily meet the
ammonia requirements in the production of explosives.

This thesis tries to establish the viability of small-scale ammonia production from biomass. A
scale of 120 tpd of biomass intake is selected to buffer a sporadic biomass supply, while
simultaneously accommodating the local ammonia demand. A conservative scale is chosen to
assist the applicability of the process concept globally. This capacity would require around
5000 hectares of land for eucalyptus cultivation in Australia (Rutovitz, 2004). The required
straw in India is estimated to be available in 7600 hectares of land (Singh et al., 2016). An
average sugar mill in Brazil, which processes more than two million tons of bagasse per annum,
can easily meet the bagasse requirement for ammonia production at the chosen scale (Walter
et al., 2014). Another rationale behind selecting this scale of ammonia production for the
biomass-to-ammonia plant is that any small-scale ammonia production facility would generally
be larger than this scale and benefit from increased economies of scale. Thus, the scale
considered in this thesis would form a limiting case.

Conventionally, ammonia is produced at a large scale, with production capacities being greater
than 1000 tpd. Scaling down the conventional ammonia flowsheet would burden the
economics. This thesis builds upon the ICI Leading Concept Ammonia® process, which
substantially contrasts with the conventional highly integrated large-scale ammonia plant
(Arkley et al., 1988). It is a small-capacity natural gas-based ammonia plant that can compete
with modern large-capacity natural gas-based ammonia plants in terms of energy consumption
and specific investment. In practice, ammonia production plants with a capacity of less than
600 tpd are considered small scale (UNIDO, 1998).The process energy consumption of the
ICI process is 29.3 GJ/t NH3 for a natural gas feedstock, which compares well with 28 GJ/t
NH3 for state-of-the-art large-scale technology (Kroschwitz and Seidel, 2000; Hicks et al.,
1990). Furthermore, the technology competes well with large-scale productions in terms of
economics (Arkley et al., 1988). Primary features of the ICI plant include milder primary
reformer conditions, use of gas-heated reformer, an isothermal CO shift reactor, Pressure
Swing Absorption (PSA) for CO2 removal, and a lower pressure ammonia synthesis loop
(Hicks et al., 1990).

5
1.4 Biomass gasification
Biomass gasification is a thermo-chemical process that converts biomass into syngas, char, and
tars on the basis of the operating conditions. The syngas mainly consists of hydrogen and
carbon-monoxide and can be utilized for a variety of energy and chemical applications. Tar is
a complex mixture of more than two hundred condensable hydrocarbons. It is a viscous liquid
that is an undesirable product of the gasification process. Char is carbon-rich solid product that
is a result of the gasification process. A variety of conversion technologies have been proposed,
developed, and demonstrated for syngas production from biomass through gasification. The
differences in these technologies are due to,

 The feeding of the biomass and the type of pretreatment of biomass that is required.
 The type of oxidant used.
 The gasification working conditions (temperature and pressure).
 Heat integration between exothermic and endothermic reactions.

Based on these characteristics, biomass gasifiers can be broadly classified into updraft,
downdraft, bubbling fluidized bed, circulating fluidized bed, dual fluidized bed, entrained flow,
and plasma gasifiers. The characteristics of the different types of gasifiers are described in
detail in Figure 1.1 (E4tech, 2009). Additionally, different biomass gasification configurations
are proposed as being optimal in different capacity ranges. The scales of operation of different
types of gasifiers, in which they are likely to be the most economical, is shown in Figure 1.2
(E4tech, 2009). In the selected capacity range of this thesis, fluidized bed gasifiers are expected
to be the most economical.

6
Updraft Gasifier
•The biomass is fed in at the top of the gasifier, and the air, or steam intake is at the bottom; hence, the
biomass and gases move in opposite directions.
•Some of the resulting char falls and burns to provide heat.
•The methane and tar-rich gas is expelled from the top of the gasifier, and the ash falls from the grate
for collection at the bottom of the gasifier.

Downdraft Gasifier
•The biomass is fed in at the top of the gasifier and the air, and oxygen or steam intake is also at the top
or at the sides; hence, the biomass and gases move in the same direction.
•Some of the biomass burns, falling through the gasifier throat to form a bed of hot charcoal, through
which the gases have to pass (a reaction zone).
•This ensures a fairly high quality syngas, which is expelled from the base of the gasifier; the ash is
collected under the grate.

Entrained Flow Gasifier


•Powdered biomass is fed into a gasifier with pressurized oxygen and/or steam.
•A turbulent flame at the top of the gasifier burns some of the biomass, providing large amounts of
heat at a high temperature (1200–1500 °C), for fast conversion of biomass into very high quality
syngas.
•The ash melts onto the gasifier walls and is discharged as molten slag.

Bubbling fluidized bed (BFB)


•A bed of fine inert material sits at the bottom of the gasifier. Air, oxygen or steam is blown upwards
through the bed fast enough (1–3 m/s) to agitate the material.
•Biomass is fed in from the side, where it mixes and combusts or forms syngas, which is expelled
upwards.

Circulating fluidized bed (CFB)


•A bed of fine inert material has air, oxygen or steam blown upwards through it fast enough (5–10m/s)
to suspend the material throughout the gasifier.
•Biomass is fed in from the side and it is suspended, following which it combusts and provides heat, or
reacts to form syngas.
•The mixture of syngas and particles are separated using a cyclone, and the material is returned into
the base of the gasifier.

Dual Fluidized bed gasifiers


•This system has two chambers: a gasifier and a combustor.
•Biomass is fed into the CFB/BFB gasification chamber and converted to nitrogen-free syngas and char
by using steam.
•The char is burnt in air in the CFB/BFB combustion chamber, heating the accompanying bed particles.
•This hot bed material is then fed back into the gasification chamber, providing the indirect reaction
heat.

Plasma gasifier
•Untreated biomass is dropped into the gasifier, where it comes into contact with an electrically
generated plasma, usually at atmospheric pressure and temperatures of 1,500–5,000 °C.
•Organic matter is converted into very high quality syngas and inorganic matter is vitrified into inert
slag.
•Note that plasma gasification uses plasma torches. It is also possible to use plasma arcs in a
subsequent process step for syngas clean-up

Figure 1.1: Characteristics of different types of gasifiers (E4tech, 2009).

7
Figure 1.2: Economic viability range of different types of gasifiers (E4tech, 2009).

Among fluidized bed gasifiers, the Dual Fluidized Bed Gasifier (DFBG) configuration appears
to be promising due to its potential to supply clean syngas, in spite of the fact that it uses air as
an oxidising agent instead of oxygen. One such gasifier is the 8 MW CHP gasifier at Gussing,
Austria, which is manufactured by Repotec (E4tech, 2009). The gasifier has been successfully
operating for heat and power applications since 2002. In a dual bed gasifier, the gasification
and combustion take place in different vessels; the gasification is primarily carried out using
steam, and combustion is carried out using air. The heat transfer takes place through the
circulation of the hot bed material between the two beds. The syngas produced is not
significantly diluted with nitrogen, because steam is the sole gasifying agent. The methane
content, however, is slightly higher. The heat produced from combustion of char is enough to
sustain the gasification process, and no external heat source is required. Furthermore, the flue
gases that exit the combustor can be used for preheating of air, superheating of steam, drying
of biomass, and other heating applications. Other advantages of dual bed gasifiers when
compared with conventional gasifiers according to Pfeifer et al. (2011) are:

• Low tars due to steam gasification.


• No requirement of oxygen to obtain nitrogen-free syngas.
• Compact design and easy feeding of biomass.
• A wide range of feedstocks can be gasified.
• A bed material like olivine can be used as a catalyst.

8
1.5 Research objectives and major contributions
This thesis explores the viability of small-scale ammonia production from biomass. To achieve
this goal, this thesis utilizes various process simulation tools at different modeling levels.
Computational Fluid Dynamics (CFD) software ANSYS Fluent is used to simulate a fluidized
bed biomass gasifier. This simulation is used to procure a Residence Time Distribution (RTD)
curve, which in turn can be used to propose a compartment model for the otherwise complex
biomass gasifier. This simplified compartment model can be easily integrated in the simulation
software, ASPEN Plus. ASPEN Plus is used to simulate the biomass gasification process as
well as downstream gas conditioning that leads to ammonia production. The outputs of ASPEN
Plus simulation were used to perform Life Cycle Costing (LCC) and Life Cycle Assessment
(LCA) for the simulated flowsheet. The multi-scale modeling methodology is illustrated in
Figure 1.3.

Figure 1.3: Multi-scale modeling methodology for the biomass-to-ammonia process.

A compartment model for DFBG was conceptualized with the help of different simulation
platforms. A CFD model was used to understand the flow patterns inside a fluidized bed. This
elevated the understanding of the hydrodynamics of the gasifier freeboard, which is neglected
by the conventional two-phase methodology. The CFD simulation was utilized to perform RTD
analysis of the reactor. There does not appear to be any previous study of this kind in the
published literature. Four tracer approaches were investigated, namely, the frozen velocity
approach, the snapshot approach, the data sampling approach, and the transient approach were
compared. The RTD analysis formed the basis of a steady-state compartment model that was
developed in ASPEN Plus simulation software. The ASPEN Plus gasifier model decoupled
the pyrolysis, gasification, and combustion section of the gasifier to affect a better
comprehension of the process and results. Additionally, the model could also be used to predict
the output for different biomass feedstocks.

9
The downstream process, which consisted of syngas conditioning and ammonia production,
was modeled in ASPEN Plus. The ASPEN Plus simulation data was used to analyze the
economics of the biomass-to-ammonia process. The economic performance of a plant was
calculated on the basis of the initial capital investment and the annual operation costs. The
proposed route leverages the inherent benefits of small-scale ammonia production and the use
of the dual fluidized bed biomass gasification technologies. The predicted economics were
found to be comparable to large-scale biomass-to-ammonia production plants. This novel
amalgamation of a relatively new gasification method that is particularly suited to biomass at
a small scale as the front end of the plant, with a largely over-looked ICI process for the tail-
end of the plant aids the economics of the biomass-to-ammonia process.

The technology needs to be environmentally sustainable in order to be successful. An LCA was


carried out in order to access the environmental impacts of the biomass-to-ammonia process.
The Life Cycle Inventory (LCI) data was imported from the ASPEN Plus flowsheets and the
ecoinvent database (Weidema et al., 2011).. The LCA was performed by using Excel
worksheets and was linked to ASPEN Plus. The LCA helped to identify the most significant
environmental impacts in different cases of interest.

The data obtained from ASPEN Plus simulation was used to predict the LCC and LCA results.
These outputs are utilized by an MS Excel-/VB-based Multi-Objective Optimization (MOO)
framework to optimize both LCC and LCA. The MOO results derived for changing flowsheet
configurations as well as for different biomass feedstocks grown at different locations
worldwide are expected to highlight a more rigorous analysis of the biomass-to-ammonia
process. This analysis distinguishes this thesis from past studies focusing on ammonia
production from biomass, which presented their LCC and LCA results for a single process
configuration and for a single feedstock. Different biomasses are expected to behave differently
in terms of emissions (due to different feedstock composition) and plant economics (due to
different feedstock price). Such an analysis provides the decision makers with consistent and
comparable data to objectively judge the viability of the proposed process. Thus, this thesis
establishes the advantages of process modeling and simulation at different levels, namely, the
phase level, the process unit level, the plant level, and the environmental effects level for the
conception of the biomass-to-ammonia process.

10
1.6 Organization of thesis
The thesis has been divided into eight chapters. This first chapter describes the rationale behind
this thesis. It also briefly describes the research methodology that has been used to analyze the
biomass-to-ammonia process. Chapter 2 provides an exhaustive review of the relevant
literature on biomass-to-ammonia processes. It also presents the different modeling and
simulation studies that have been carried out in order to understand the characteristics of
gasifiers. A brief overview of the MOO studies of bioenergy systems has also been presented.
The gasifier modeling that was carried out at different levels through the use of CFD, RTD
analysis, MATLAB, and ASPEN Plus simulations is explained in Chapter 3. The next chapter,
Chapter 4, describes the simulation of the complete biomass-to-ammonia process as well as the
different process variants. The heat-integration methodology and process LCC are also
elaborated on here. Chapter 5 provides the LCA methodology and the results that were derived
for different flowsheet configurations as well as different feedstocks. The MOO framework
that was used in order to optimize the process economics as well as environmental emissions
is elucidated in Chapter 6. Chapter 7 presents a discussion on the results that were predicted by
the MOO analysis, and the effects of different biomasses and different locations on the
biomass-to-ammonia process. Chapter 8 concludes the research with recommendations for
future work. A summary of the chapter contents is shown in Figure 1.4.

Figure 1.4: Organization of the thesis.

11
Chapter 2: Literature review

2.1 Introduction
This chapter presents a review of literature relevant to the present study of ammonia production
from biomass feedstock. The biomass feedstocks under consideration and the selection of a
suitable gasifier technology have been explained in the previous chapter. This chapter
establishes the need for the three main aspect of the thesis namely the gasifier model, the
biomass-to-ammonia process and multi-objective optimization of the biomass-to-ammonia
process. Modeling of the dual fluidized bed gasifier is critical to this study for prediction of the
quality and quantity of syngas under different operating conditions and for different feedstocks.
A gasifier model is also vital for energy integration of the biomass gasification process with
the downstream gas conditioning and ammonia production. Section 2.1 describes the different
modeling approaches that have been successfully used for modeling of the biomass gasification
process in general and dual fluidised bed gasification in particular. It also establishes a need
for a compartment model for fluidized bed gasification.

Section 2.2 provides a thorough review of different biomass feedstock-based ammonia


production processes proposed by different researchers worldwide. The emphasis is to justify
the need for yet another techno-enviro-economic study for ammonia production from biomass.
Section 2.3 highlights the advantages of MOO that would assist in-depth understanding of the
biomass-to-ammonia process. The chapter ends with a summary highlighting the different
research gaps that the present thesis is trying to tackle.

2.2 Gasifier modeling


Computer modeling is an efficient tool for the optimization of gasifier operations. It is also a
potent tool that can be used for the recommendation of novel process variations. The biomass
gasification simulation models can be grouped under four major methodologies, namely,
equilibrium model, kinetic model, CFD model, and Artificial Neural Network (ANN) model
(Basu, 2010). These modeling approaches are often mutually exclusive and are used on a stand-
alone basis.

Equilibrium models are characterized by the ease of their formulation and wide applicability.
These models are based on the assumption that the reacting system reaches its most stable of
all compositions. In other words the Gibbs Free Energy is minimized. However at the normal

12
gasification temperatures of 700–1000°C, thermodynamic equilibrium is not achieved (Puig-
Arnavat et al., 2010). In order to counter the deviation from experimental data, which arises
from this assumption, researchers use a Quasi-Equilibrium Temperature (QET) approach
(Puig-Arnavat et al., 2010). In this approach, the simulation software calculates the system
equilibrium for a temperature that is lower than the actual operating temperature. The QET can
be defined either for individual reactions or for the whole reactor on the basis of the flowsheet
configuration. The results predicted by the equilibrium models are independent of reactors.
Their advantage lies in their ability to predict the thermodynamic viability; however, their
accuracy in predicting differences from the different types of gasifiers is limited. Schuster et
al. (2001) performed one of the earliest simulation for the dual fluidised bed gasifier. Using an
equilibrium model, they tried to quantify the effect of gasifier temperature and biomass oxygen
content on gasification efficiency. However, their model underestimated the methane content
in the product gas. He at al. (2012) proposed an ASPEN Plus model with a QET of 300 ºC, for
modeling the MIUN dual bed gasifier. The model predicted the CO content to be higher than
H2 content at low steam-to-biomass ratio. Doherty et al. (2013) utilized the QET approach
along with empirical correlations for pyrolysis, for modeling the FICFB gasifier in ASPEN
Plus. An equilibrium model for the same gasifier was also proposed by Kaiser et al. (2003)
using the IPSEpro software. The authors extended the simulation for power generation and
performed an exergy analysis for the whole process.

Kinetic models are preferred over equilibrium models, when the design of the gasifier and other
physical parameters need to be accounted for. Kinetic models find their application in the
modeling of fluidized bed reactors, because the process is kinetically limited. Ideally, reaction
kinetics must be solved simultaneously with bed hydrodynamics. This, however, makes the
simulation very complex and, thus, the hydrodynamics is sometimes neglected. A semi-
detailed kinetic model for gasification in a DFBG using ASPEN Plus has been presented by
Abdelouahed et al. (2012). The initial pyrolysis was considered to be instantaneous and was
modeled with the help of empirical correlations. The homogenous and heterogeneous
gasification reactions were modeled on the basis of chemical kinetics. The dense phase region
and the freeboard of the reactor were assumed to be Continuous Stirred Tank Reactors (CSTR)
and Plug Flow Reactors (PFR), respectively. Another model for the DFBG, which uses an
equation-oriented steady-state software, IPSEPro, is presented by Proell et al. (2007).

The performance of a fluidized bed gasifier is very strongly dictated by hydrodynamics.


Nonlinear interactions between gases and the solid particle movement gives rise to very

13
complex hydrodynamics (Loha et al., 2014b). Conventionally, fluidized beds have been
modeled by using two-phase models (or three-phase models) (Levenspiel, 1999). In two-phase
models, the dense bed of the fluidized bed is divided into two phases, namely, the bubble phase
and the emulsion phase. The hydrodynamics of the fluidized bed is characterized by using
correlations in order to approximate the mixing and reactivity within each phase rather than by
solving the momentum equation for each phase. Thus, the applicability of these models is
constrained by the validity of the assumptions (Loha et al., 2014b). Recent studies have
predicted that these models underestimated the process parameters like gas flowrates and
residence times when compared with the CFD results (Loha et al., 2014b; Atsonios et al.,
2015). Moreover, the two-phase model can only be utilized for the modeling of the dense bed
of a fluidized bed. The freeboard is normally modeled crudely by assuming a plug flow
behavior. The splash zone, which is formed above the dense zone, is also neglected. It is noted
that the presence of a splash zone makes the demarcation of a dense zone and freeboard difficult
(Timmer, 2008). It has also been reported that catalytic destruction of tar occurs in the splash
zone, in the case of co-gasification of biomass and plastics (Wilk and Hofbauer, 2013). Kaushal
et al. (2011) proposed a model for DFBG by using two-phase theory. The bed material
concentration in the freeboard (including in the splash zone) was modeled by using an
exponential function that decays with increasing height. They neglected the complex gas
mixing pattern in the splash zone. Shahrivar et al. (2015) have modeled the dense zone of a
fluidized bed by using a complex counter-current combination of CSTR’s and PFR’s on the
basis of the two-phase theory. However, the freeboard is modeled as a simple PFR. It may be
concluded that there is a need for a better understanding of the mixing pattern in the freeboard,
especially, in the splash zone.

The CFD-based models, with improved modeling of complex gas-solid interactions are capable
of providing solutions which are close to real life observations (Basu, 2010). The results are
obtained by solving the mass, momentum and energy balance equations over a control volume.
There are two main approaches to CFD modeling: the Eulerian–Lagrangian approach and the
Eulerian–Eulerian approach. In the Eulerian–Langrangian approach, the gas phase is
considered to be a continuum and is defined by using the Navier–Strokes equation. The solid
is considered a discrete phase and is modeled using Newton’s law of motion. The temperature
and gas concentrations are calculated using mass and energy balances for each particle. This
improves the accuracy, but is computationally expensive. In the Eulerian–Eulerian approach,
the solid phase is also modeled as a continuum and requires a low degree of computational

14
ability (Liu et al., 2013). The major drawbacks of any CFD simulation are that it is
computationally intensive and time-consuming. Thus, an optimization or a rigorous sensitivity
analysis of the DFBG system would require either very large computational resources or an
enormous amount of time. Several researchers have studied the DFBG using CFD simulations.
Nguyen et al. (2012) developed a 2-D CFD simulation model for their pilot dual bed gasifier,
utilizing the Eulerian–Eulerian approach. They observed a good hydrodynamic similarity
between CFD and lab scale model. However, the simulations were only performed for a total
gasification time of 30 s. A 3-D CFD model for the BFB of the FICFB gasifier was prepared
by Thapa et al. (2014) in Barracuda VR 15 software utilizing the Eulerian–Lagrangian
approach. Their CFD model also included the reaction kinetics for the gasification reactions.
A similar study was done by Loha et al. (2014a) for 3-D modeling of a BFB. Liu et al. (2015)
performed a 3-D simulation for the complete dual bed gasifier. The CFD model needed 120
hours to simulate 100 s of real time, using a workstation with an Intel i7 CPU @3.50 GHz.
Their simulation results were found to be in agreement with the experimental outputs. The
model was further utilized to predict the effect of solid particle size distribution on gasifier
performance.

Complex systems such as biomass gasification may also be solved using Artificial Neural
Networks. The ANN approach mimics biological nervous systems such as the brain in order
to process information. Its main functional units are interconnected processing elements known
as neurons, which work in coherence to solve specific problems. The system learns by example,
which is similar to the way humans learn. An ANN is usually designed for a specific
application. It provides a numerical solution rather than an analytical one (Basu, 2010). The
ANN model needs to be trained with real-time data, which requires a magnitude of operational
data. If the input data differs substantially from the training data, the model might give poor
results. The strength of this modeling approach lies in its being able to predict the output of a
real-time gasification process rather than in the understanding or optimizing of the process. An
ANN model for fluidized bed gasifiers has been used by Puig-Arnavat et al. (2013).

In this thesis, an attempt has been made to connect models at different scales. A compartment-
based methodology has been proposed for effective modeling of DFBG systems by
amalgamating the reaction kinetics approach and the CFD modeling approach. CFD modeling
will be used to predict hydrodynamics, which along with reaction kinetics would be used to
model the DFBG system. The use of CFD simulation in hydrodynamics will help overcome
the short-comings of the two-phase approach. Moreover, since the thesis is aimed at steady-

15
state simulations, the time required for the modeling of different gasifier variants would be
considerably reduced. It will also be shown that such a system would also have the ability to
predict syngas output for different feedstocks.

2.3 Biomass-to-ammonia process


The syngas produced from biomass gasification is a rich source of hydrogen and carbon
monoxide, which can be converted to hydrogen via the water-gas shift reaction. The hydrogen
source may be further utilized for ammonia production. Many authors in the recent past have
acknowledged the potential of ammonia production from biomass through the gasification
route. Analogous flowsheets have been proposed by different researchers. The basic flowsheet
starts with biomass gasification. The syngas produced is cleansed of impurities and
particulates. The syngas is also reformed if it contains a considerable amount of methane.
Thereafter, the shift reaction is employed to convert carbon monoxide to carbon dioxide. The
syngas is then stripped of carbon dioxide using a Pressure Swing Adsorption (PSA) or
absorbing solvents. A methanator is then used for removal of any trace carbon monoxide and
carbon dioxide. Finally, the syngas enters into a high pressure ammonia loop which leads to
the production of pure ammonia. The flowsheets is quite similar to conventional large-scale
ammonia production plants, with the biomass gasification replacing the reforming of natural
gas (Figure 2.1). The flowsheet can be simulated in various process modeling softwares such
as ASPEN plus, ASPEN HYSYS etc. The mass balance and the energy balance calculations in
simulation software may be effectively used for predicting the LCC and LCA for the biomass-
to-ammonia process.

Figure 2.1: Typical biomass-to-ammonia process flowsheet.

16
A considerable number of researchers have reported the results of a LCA of bioenergy systems
in order to advise policy making. LCA methodology is seen as a reliable and efficient means
to interpret the sustainability of bioenergy and other bio-based products. However, variations
arising from data uncertainties, incompleteness of included impacts and methodological
choices, sometimes make it difficult to draw general conclusions (Singh et al., 2013). The past
LCA studies were mostly limited to net energy output and global warming. Cherubini and
Sromann (2011) state that while 90 percent of the LCA studies include global warming, only
20–40 percent of LCA studies estimate other impact categories such as acidification and
eutrophication. A similar trend is noticed for studies concentrating on the biomass-to-ammonia
process. This stems from the fact that climate change has always been considered a major
global issue. However recent studies have predicted other impacts such as land use, water use,
and biodiversity. Furthermore, the use of genetically modified organisms cannot be neglected
for bioenergy systems (Wiloso et al., 2012; Voet et al., 2010). It has also been predicted that
the water footprint of biomass-based systems is much larger than other energy systems.
Gerbens-Leenes et al. (2009) predict the water footprint of biomass based systems to vary from
24-143 m3/GJ, as compared to 0.1 m3/GJ for natural gas, 0.2 m3/GJ for coal and 0.3 m3/GJ for
solar thermal energy. According to the IPCC (2011), the environmental impacts can be divided
into the global ones such as global warming, acidification etc. and local ones such as soil
quality, biodiversity etc. Recent LCA studies use standard life-cycle impact assessment
methods such as ReCiPe, EDIP, LIME, TRACI, and the like (Curran, 2012). These impact
assessment methods may be limited to endpoint or midpoint indicators or might predict both.
In LCA, impact categories based on midpoints are more common in characterization models,
given that relationships between emissions and environmental stressors are often well
established. In contrast, impact categories based on specific endpoints are less common,
because these endpoints require extensive observations, risk analyses, and data to relate them
to emissions of different substances, and these relationships can vary with region. The
following section provides an overview of LCA and LCC of biomass-to-ammonia processes
as proposed by different authors.

Ahlgren et al. (2008) published a Life-Cycle Assessment of ammonium nitrate fertilizer


production from biomass. They considered two biomass feedstocks, namely, cereal straw and
Salix. In their cradle-to-gate analysis of the process, they highlighted the effect of a biomass
feedstock on the global warming potential, acidification potential and eutrophication potential.
They have predicted a reduction of 30 percent and 22 percent in the global warming impact,

17
for the Salix and straw feedstocks, respectively. The eutrophication potential was predicted to
be higher for biomass-to-ammonia system. No significant improvement was noticed in the
predicted acidification potential for biomass-based systems when compared to conventional
natural-gas-based system. Tuna et al. (2014) did an economic analysis for a similar system and
reported an ammonia production cost in the range of $ 970-1342/t NH3 for a 87 tpd ammonia
plant.

Gilbert et al. (2014) did an economic analysis for 1200 tpd ammonia plant based on round
wood feedstock. They predicted that the cost for the production of ammonia from biomass
would be competitive to the present market price of ammonia. They also report a greenhouse
gas reduction of 65 percent when compared to conventional natural gas fed ammonia plants.
Their proposed plant would, however, require 1,121,000 tons per annum (tpa) of biomass
feedstock. This is greater than the world’s largest biomass power plant. A similar techno-
enviro-economic study including carbon capture was carried out by Tock et al. (2014) for a
1187 tpd ammonia plant. They predicted a need for a carbon tax to make their process
competitive with the conventional process. Their simulation predicts a decrease of $ 72/t NH3
in the ammonia production cost, for a carbon tax of $ 35/t CO2. The GWP is estimated to be 20
percent of the conventional natural-gas-based ammonia production. They also predicted the
endpoint indicators using IMPACT 2002 impact assessment method. Furthermore, if a part of
biogenic carbon emissions are captured and stored, the GWP of the biomass-to-ammonia
process can be as low as -1.7 kg CO2/ kg NH3. Another study carried out by Andersson and
Lundgren (2014) reported a cost of approximately $ 750–1200/t NH3 for a 700 tpd biomass-
based ammonia plant. In their case study, they assumed ammonia production via integrated
gasification in an existing paper and pulp mill. The authors conclude that the considered
capacity range was too small to be economically viable.

Sarkar et al. (2011) and Bartels (2009) studied the effect of scale on the price of ammonia
produced from biomass. Both predicted a steep rise in the ammonia production price with a
decreasing plant scale. Sarkar et al. (2011) predicted that an optimum biomass-to-ammonia
plant would require 990,000 tpa of dry biomass. The majority of the studies mentioned above
indicate that ammonia production from biomass is feasible only at the production level of a
world-scale ammonia plant. Smaller plants have to suffer the effects of economy of scale. The
economics and environmental results for biomass-to-ammonia process from different studies
are summarised in Table 2.1.

18
Table 2.1: Comparison of literature for ammonia production from biomass.
Study Feedstock Capacity (tpd of NH3 Country Ammonia production GHG emissions
production) cost ($/ NH3) (t CO2/ t NH3)
Ahlgren et al. (2008) Straw 310 Sweden - 0.70
Salix 0.55
Bartels (2009) Biomass 10 U.S. 1519 -
1022 488
Sarkar et al. (2011) Wood chips 2000 tpd biomass intake Canada 2060 .006
Gilbert et al. (2013) Roundwood 1200 Poland/U.K. 500 0.67
Tuna et al. (2014) Wood 87 Sweden 970–1342 -
Anderson et al. (2014) Wood 700 Sweden 750–1150 -
Tock et al. (2015) Wood 1187 Switzerland 968 -1.79 (including CCS)
Thornley et al. (2016) Wood chips - U.K - 0.613
Gilbert et al. (2013) Natural gas 1200 U.K. 500 1.9

A wide variation in ammonia production cost as well as the associated GHG emissions is
observed in different studies of ammonia production from biomass. The present thesis will
objectively try to understand the economic and environmental variation that result from
different feedstocks, countries and economic assumptions. Since, majority of the studies
reported in Table 2.1 were carried out under different sets of economic and environmental
assumptions, it would be difficult to quantify the effect of biomass feedstock and location on
the ammonia production cost and the associated GHG emissions. Furthermore, the focus of
this thesis would be to evaluate the feasibility of the biomass-to-ammonia process on a small
scale, which has been overlooked by majority of studies centred on the biomass-to-ammonia
process. A scale of 120 tpd of biomass intake is selected to buffer the sporadic biomass supply,
while simultaneously accommodating the local ammonia demand. A conservative scale is
chosen to assist the applicability of the process concept globally. Another rationale behind
selecting a scale of 65 tpd of ammonia production for the biomass-to-ammonia plant is that any
small scale ammonia production facility would generally be larger than this scale and benefit
from the economies of scale. Thus, the scale studied in this thesis would form a limiting
economic case.

2.4 Multi-objective optimization


In process design, the aim is to propose the best process possible in terms of efficiencies, in
particular, positive economics and low environmental impacts. Sometimes, these objectives are
conflicting or interfering, such that improvement in one objective leading to a decline in the
other. Maximizing carbon sequestration while simultaneously minimizing production costs is
a typical problem faced by both chemicals and energy industries (Tock and Maréchal, 2012a).
In the past, such systems were optimized by placing various weights on different objectives

19
and optimising the combined single objective. Multi Objective optimisation (MOO) on the
other hand, gives a range of optimal solutions called Pareto-optimal solutions for conflicting
objectives, in which each individual solution is said to be non-dominated by the other solutions.
This means that if a solution is better in one objective, it is inferior in the other objective
(Figure 2.2). No solution on a Pareto curve is thus better than any other point with respect to
all the objectives. Thus, the Pareto optimal solution presents a pool of equally good points, and
the user can choose the optimal points best catering to their needs.

Figure 2.2: Pareto front depicting dominated and non-dominated points.

Many algorithms have been proposed to predict solutions for MOO problems. A
comprehensive review of the different algorithms has been presented by Deb (2001), Leyland
(2002) and Konak et al. (2006). The advantages of using Genetic Algorithms (GA’s) have been
suggested by Deb (2001) and Harkins (2012). The GA’s being derivative free are expected to
converge at only global optimum solutions. Apart from GA’s, other MOO methods include
simulated annealing algorithm, particle swarm optimisation, global criterion, value function
and goal programing methods. A comparison of different MOO methods is beyond the scope
of the present work, but can be found in Rangaiah (2009). Genetic algorithms are based on
Charles Darwin’s theory about evolution. The solution evolves with time. The algorithm begins
with an initial set of solutions (represented by chromosomes) called a population. Solutions
from one population are taken and used to form a new population based on parameters provided
by the user (Figure 2.3). The motive is to create a new population that is better than the old
one. Solutions which are selected to form new solutions (offspring) are selected according to
their fitness - the more suitable they are, the more chances they have to reproduce. This is
repeated until some condition (for example number of populations or improvement of the best
solution) is satisfied.

20
Figure 2.3: Crossover in genetic algorithms.

The parameters provided by the user are the population size, the maximum generations, the
crossover probability and the mutation probability. The population size represents the number
of solutions that will be created after each generation. The maximum generations limits the
number of generation and this parameter is made large enough to ensure that the final
population represents the converged Pareto front. The crossover probability indicates a ratio of
how many couples (chromosomes) will be picked for mating (Figure 2.3) and the mutation
probability is basically a measure of the likeness that random elements of your chromosome
will be flipped into something else to bring in new solutions.

A number of GA’s are available in literature namely, Vector Evaluated Genetic Algorithm
(VEGA), Multi-objective Genetic Algorithm (MOGA), Niched Pareto Genetic Algorithm
(NPGA), Weight-based Genetic Algorithm (WBGA), Random Weighted Genetic Algorithm
(RWGA), Nondominated Sorting Genetic Algorithm (NSGA), Strength Pareto Evolutionary
Algorithm (SPEA), improved SPEA (SPEA2), Pareto-Archived Evolution Strategy (PAES),
Pareto Envelope-based Selection Algorithm (PESA), Region-based Selection in Evolutionary
Multiobjective Optimization (PESA-II), Fast Non-dominated Sorting Genetic Algorithm
(NSGA-II), Rank-Density Based Genetic Algorithm (RDGA), and Dynamic Multi-objective
Evolutionary Algorithm (DMOEA). A review of different GA’s has been presented in
Table 2.2 (Konak et al., 2006). The NSGA II algorithm has been effectively used for the
optimization of economics and environmental emissions of energy systems (Harkin, 2012).
Some of the advantages of NSGA II algorithm include preservation of solution diversity, a
maximum complexity of O(MN2) and use of elitism to preserve Pareto optimal solutions.
Nandasana et al. (2003) have used the NSGA II code for optimizing an industrial steam
reformer. Inamdar et al. (2004) have also demonstrated the suitability of NSGA II algorithm
for optimizing profits while minimizing energy costs for a crude distillation unit. A
comprehensive review of use of NSGA II algorithm for optimal design of chemical processes
with multiple economic and environmental objectives has been presented by Lee et al. (2009).
The NSGA II code was found to deliver satisfactory performance in different design problems.

21
The purpose of this thesis was not to research new MOO software and NSGA II is recognized
as a very robust solver, even if not the fastest. Thus, this thesis uses a framework developed by
Sharma et al. (2012), which utilizes the NSGA-II methodology (Deb et al., 2002). The MOO
framework results in a Pareto plot with the global optimum for LCC and LCA, for a given set
of decision variables and constraints. These results provide a range of process parameters,
which can help minimize the economic and environmental loads. By allowing the flowsheet
configurations and the design variables within these configurations to be optimised, a more
rigorous analysis of the biomass-to-ammonia process can be obtained.

Table 2.2: Comparison of different genetic algorithms (Konak et al., 2006).


Algorithm Fitness assignment Advantages Disadvantages
VEGA Each subpopulation is evaluated Straightforward implementation Tend to converge to extreme of
with respect to a different each objective
objective
MOGA Pareto ranking Simple extension of single Usually slow convergence,
objective GA problems related to niche size
parameter
WBGA Weighted average of normalized Simple extension of single Difficulties in nonconvex
objectives objective GA objective function space
NPGA No fitness assignment, Very simple selection process Problems related to niche size
tournament selection with tournament selection parameter, extra parameter for
tournament selection
RWGA Weighted average of normalized Efficient and easy implement Difficulties in nonconvex
objectives objective function space
PESA No fitness assignment Easy to implement, Performance depends on cell
computationally efficient sizes, prior information needed
about objective space
PAES Pareto dominance is used to Random mutation hill-climbing Not a population based
replace a parent if offspring strategy, easy to implement, approach, performance depends
dominates computationally efficient on cell sizes
NSGA Ranking based on non- Fast convergence Problems related to niche size
domination sorting parameter
NSGA-II Ranking based on non- Single parameter, well tested, Crowding distance works in
domination sorting Efficient objective space only
SPEA Raking based on the external Well tested, no parameter for Complex clustering algorithm
archive of non-dominated clustering
solutions
SPEA-2 Strength of dominators Improved SPEA, make sure Computationally expensive
extreme points are preserved fitness and density calculation
RDGA The problem reduced to bi- Dynamic cell update, robust More difficult to implement
objective problem with solution with respect to the number of than others
rank and density as objectives objectives
DMOEA Cell-based ranking Includes efficient techniques to More difficult to implement
update cell densities, adaptive than others
approaches to set GA
parameters

22
The inclusion of LCA into MOO was first proposed by Azapagic and Clift (1999). Since then
this approach has found its application in different branches of process systems engineering.
However, the use of MOO coupled with LCA in chemicals production from biomass is a
relatively new research area (Calva, 2016). A few studies are available for MOO of the
production of chemicals from biomass via a thermo-chemical pathway. Gerber et al. (2011)
optimized the economics and LCA impact for cogeneration of SNG and power from
lignocellulosic biomass using the MOO methodology. An economic and environmental
optimization of a bio-refinery has been demonstrated by Santibanez-Aguilar et al. (2011). Tock
and Marechal (2012b) undertook a MOO for the co-production of hydrogen and electricity
from lignocellulosic biomass. They brought forward the trade-off between hydrogen and
electricity generation and CO2 capture using the MOO philosophy. Wang et al. (2013)
proposed an optimal design for a hydrocarbon biorefinery via gasification and Fischer-Tropsch
synthesis. They utilized a MINLP-based MOO approach to understand the trade-off between
Net Present Value (NPV) and GWP of the proposed process. You et al. (2012) presented the
optimization of economic, environmental and social objectives for the cellulosic ethanol supply
chain. There does not appear to have been any previous study that focuses upon MOO of
ammonia production from biomass. As discussed in the previous section, all studies focussing
upon ammonia production from biomass present their LCC and LCA results for a single
process configuration.

2.5 Summary
The review of literature indicated the absence of MOO studies in the domain of ammonia
production from biomass. Another research gap lies in the quantification of the effect of
biomass feedstock and location on the economics and life-cycle emissions of the biomass-to-
ammonia process. Different biomasses cultivated at different locations worldwide are expected
to behave differently in terms of both life-cycle emissions and plant economics. Additionally,
processes based on biomass feedstock are usually small-scale and are limited by biomass
supply. In order to ensure sustainable ammonia production, this thesis highlights the techno-
economic advantages that result from small-scale ammonia plants based on biomass feedstock.
Such an analysis will provide the decision makers with consistent and comparable data to
objectively judge the viability of the proposed process. To meet the above objectives a robust
model for the gasifier is vital. This thesis proposes a kinetics based compartment model for a
dual fluidized bed gasifier in ASPEN Plus software. The novelty of this model lies in the use

23
of CFD simulations, which enables the understanding of the hydrodynamics of a fluidized bed.
This approach would help in establishing a quasi-steady-state model for an inherently dynamic
system. Additionally, the gas-mixing pattern in the freeboard (splash zone), which is neglected
by the two-phase models, can also be taken into account. The modeling approach for the
complete biomass-to-ammonia process is explained in detail in the following chapters.

24
Chapter 3: Gasifier Modeling

3.1 Introduction
Biomass is a renewable energy source that is useful in a wide range of applications such as
heat, power, fuel and chemicals. Biomass gasification is considered to be an important
technology for the future production of biomass-based products. The different biomass
gasification technologies are grouped under updraft gasification, downdraft gasification,
bubbling fluidized bed gasification, circulating fluidized bed gasification, dual fluidized bed
gasification and entrained flow gasification. The strengths and weaknesses of each technology
have been discussed in the previous chapter. This thesis focuses on the state-of-the-art, Dual
Fluidized Bed Gasifier (DFBG) (Figure 3.1). The choice of the gasifier stems from its
successful operation for over a decade in the domain of biomass gasification, which has a few
success stories (Knoef, 2012). As the name suggests, the DFBGs operate by utilizing two
interconnected fluidized beds. A Bubbling Fluidized Bed (BFB) is used to gasify the biomass,
which results in gases, tars and char. The bed material and char then flow due to gravity to the
Circulating Fluidized Bed (CFB), in which the char is combusted, which results in the reheating
of the bed material. This reheated bed material is supplied to the gasifier and it provides the
heat that is needed for the gasification of the biomass. The advantages of DFBG’s are the
production of nitrogen-free syngas, a low production of tar, and flexibility in the type of
feedstock used.

25
Figure 3.1: Dual fluidized bed gasifier

A multi-scale approach has been utilized in order to study the operation of the DFBG. The
8 MW Combined Heat and Power (CHP) gasifier at Gussing, Austria was selected as the
prototype for this study. Gasifier syngas is the main gasification output which is utilized as a
feedstock for ammonia production. Thus, the study of the behavior of the BFB in which the
syngas is produced is of the prime interest in this thesis. This thesis utilizes ANSYS Fluent
CFD software in order to understand the hydrodynamic behavior of a BFB. The CFD
simulation for a BFB is inherently unsteady. Therefore, it would be computationally very
expensive to model the integrated DFBG process, which consists of the gasifier, the combustor,
cyclones and seals. The CFD model was utilized to predict a steady-state model for the gasifier
to be employed within the ASPEN Plus model, which is a steady-state simulation of the DFBG
process. The char combustor is also modelled in ASPEN Plus to close the energy balance.

26
The link between CFD simulation and the reactor model inside ASPEN Plus is based on a
Residence Time Distribution (RTD) study that is based on the gasifier CFD simulation. The
RTD curve along with the flow characteristics of the tracer was utilized to formulate a
compartment model in ASPEN Plus. The ASPEN Plus gasifier model decoupled the pyrolysis,
gasification and combustion section of the gasifier for better comprehension of the process and
the results. An advantage of such a compartment model is that it can easily be integrated with
the flowsheet simulation of the downstream process that is modeled in ASPEN Plus. The
methodology is depicted in Figure 3.2.

Figure 3.2: Compartment Model Methodology

3.2 CFD model


CFD provides an efficient mechanism that enables the understanding of the intricate details of
any real-world system. CFD simulations were performed in ANSYS Fluent 14 software by
using a geometry that was similar to the gasifier section of the DFBG (Kaushal et al., 2011).
Figure 3.3 depicts the 2D geometry and the solid volume fractions at different time instants. A
Eulerian–Eulerian approach, in which both solid and fluid phases are modeled as a continuum
was used. A per phase standard k-ε turbulence model along with continuity and momentum
equations was used to model the fluid mechanics of both the gaseous and the solid phases. The
closures are based on the Kinetic Theory of Granular Flows (KTGF), which was proposed by
Ding and Gidaspow (1990). The governing equations 3.1–3.5 are described below:

Continuity equation:
𝜕
(𝛼𝑔 𝜌𝑔 ) + ∇(𝛼𝑔 𝜌𝑔 ⃗⃗⃗⃗⃗)
𝑣𝑔 = 0 (3.1)
𝜕𝑡
𝜕
(𝛼𝑠 𝜌𝑠 ) + ∇(𝛼𝑠 𝜌𝑠 ⃗⃗⃗⃗)
𝑣𝑠 = 0 (3.2)
𝜕𝑡

27
Momentum equation:
𝜕
𝑣𝑔 + ∇(𝛼𝑔 𝜌𝑔 𝑣⃗𝑔2 ) = −𝛼𝑔 ∇𝑝 + ∇𝜏̿𝑔 + 𝛼𝑔 𝜌𝑔 𝑔⃗ + 𝐾𝑔𝑠 (𝑣⃗𝑔 − 𝑣⃗𝑠 )
(𝛼𝑔 𝜌𝑔 ⃗⃗⃗⃗⃗) (3.3)
𝜕𝑡
𝜕
𝑣𝑠 + ∇(𝛼𝑠 𝜌𝑠 𝑣⃗𝑠2 ) = −𝛼𝑠 ∇𝑝 − ∇𝑝𝑠 + ∇𝜏̿𝑠 + 𝛼𝑠 𝜌𝑠 𝑔⃗ + 𝐾𝑔𝑠 (𝑣⃗𝑔 − 𝑣⃗𝑠 )
(𝛼𝑠 𝜌𝑠 ⃗⃗⃗⃗) (3.4)
𝜕𝑡

Granular Temperature
3 𝜕
𝑣𝑔 𝑠 )] = (−𝑝𝑠 𝐼 ̿ + 𝜏̿𝑔 ): ∇𝑣⃗𝑠 + ∇(𝑘𝜃𝑠 ∇𝜃𝑠 ) − 𝛾𝜃𝑠 + ∅𝑔𝑠
[ (𝛼𝑠 𝜌𝑠 𝜃𝑠 ) + ∇(𝛼𝑔 𝜌𝑔 ⃗⃗⃗⃗⃗𝜃 (3.5)
2 𝜕𝑡

where
Volume fraction
Density (kg/m3)
Stress tensor (Pa)
Granular temperature (m2/s2)
𝐼̿ Unity matrix
Collision dissipation of energy (kg/s3m)

Transfer rate of kinetic energy (kg/s3m)


g Gravity
Gas/solid momentum exchange coefficient
Diffusion coefficient for granular energy (kg/m s)
p pressure (Pa)
v velocity (m/s)
The Syamlal-O’Brien drag (1989) model was used for the calculation of the momentum
exchange between the solid and the gaseous phases. The geometry boundary conditions and
the models used in the simulation are detailed in Table 3.1. The governing equations were
solved using the finite volume approach. A square mesh with a mesh spacing of 0.005 m was
used. A similar grid-spacing has been used by Utikar and Ranade (2007) and Nguyen et al.
(2012) for comparable systems. The Phase Coupled SIMPLE (PC-SIMPLE) algorithm was
used to solve the pressure-velocity coupling. The simulations were run with a time step size of
0.001 s and a maximum of 100 iterations per time step. The convergence criteria for residuals
were set to 0.0001. The bed was patched with solids up to a height of 0.8 m and a solid volume
fraction of 0.6 was used to initialize the simulation. These conditions would change as the
simulation achieves quasi steady state. The inlet velocity was fixed at 0.5 m/s, which is about
twice the minimum fluidization velocity. In normal operations, the fluidization velocity varies
from two to five times the minimum fluidization velocity (Schmid et al., 2012). The bubble

28
profiles were found to be similar to bubble profiles reported by Loha et al. (2014a) for their 3D
simulation of a BFB. An experimental validation of the CFD model is beyond the scope of the
present thesis and would be taken as a future work.

Figure 3.3: Fluidized bed solid volume fraction.

29
Table 3.1: CFD simulation parameters.
Geometry and models
Width 0.67 m
Height 6m
Grid size 0.005 x 0.005 m
Bed height 0.8/1.2 m
Boundary conditions
Inlet Velocity
Outlet Pressure
Walls Zero shear
Initial conditions
Velocity 0.5 m/s
Minimum fluidization velocity 0.25 m/s
Initial packing fraction of solids 0.6
Maximum packing fraction of solids 0.61
Material properties
Gas density 1.2 kg/m3
Solid density 2660 kg/m3
Particle diameter 500 µm
Models
Granular viscosity Syamlal-O’Brien (1989)
Granular Conductivity Syamlal-O’Brien (1989)
Granular bulk viscosity Lun et al. (1984)
Solids pressure Lun et al. (1984)
Radial distribution Lun et al. (1984)
Frictional viscosity Schaeffer (1987)
Boundary conditions Johnson and Jackson (1987)
Spatial Discretization
Gradient Green-Gauss Node Based
Momentum Second Order Upwind
Volume Fraction QUICK
Granular Temperature Second Order Upwind
Turbulent Kinetic Energy Second Order Upwind
Turbulent Dissipation rate Second Order Upwind

30
3.3 RTD analysis
In the CFD simulation, a quasi-steady state was noticed after a simulation time of 15 s. The
tracer analysis was performed after a simulation time of 20 s. A tracer with zero diffusivity was
introduced as a user-defined scalar into the gas phase. A surface monitor was used to report the
mass-weighted average at the outlet. The E-curve was then calculated using the concentration
profile. Tests were performed using both the pulse and the step tracer, leading to an E-curve
and F-curve, respectively. The pulse tracer was found to be more reliable because the step
tracers resulted in unrealistic fluctuations in the E-curve, which was derived from the numerical
differentiation of the F-curve. For RTD analysis, four approaches, namely, frozen velocity,
snapshot, data sampling, and transient were compared. The four approaches were tested to
understand the trade-off between the accuracy of the RTD and the computational time needed.

In the frozen velocity approach, the simulation for the flow field was frozen at a particular time
instant. A pulse tracer was then introduced into the frozen flow field and the response was
observed at the outlet. A similar approach has been used by Daggupati et al. (2011) for RTD
calculations of steady-state systems. The second approach is the snapshot approach, which
could partially incorporate the effect of changing flow fields as the tracer travels through the
fluidized bed. This approach is based on a study conducted by Buwa and Ranade (2003) to
calculate the mixing time of bubble columns. This approach stands as a trade-off between
accuracy and computational load. In this approach, the flow fields are saved for different time
instants, as shown in Figure 3.3. After the introduction of the tracer, instead of keeping the flow
fields static, as in the previous approach, the flow fields are regularly updated with the
previously stored results. This approach was implemented using the User-Defined Function
(UDF) in Fluent. The frequency of change in snapshots was kept constant for one particular
RTD study. Three such studies were performed, with a change in the snapshots every 5 s, 2 s,
and 0.5 s.

The third approach was the data sampling approach. In this approach, transient simulation was
performed for a duration of 5 s of simulation time and all the transport properties were averaged
over this duration. The tracer was introduced in the averaged out stagnant flow fields and the
tracer concentration was observed at the outlet. In the last approach, which is the transient
approach, the tracer equation was solved in parallel with the different governing equations of
the fluidized bed simulation. This approach was computationally the most expensive and
required 15 days of computation on an Intel® Core™ i7-3770 3.4GHz processor that utilized

31
eight cores. The RTDs that were obtained from the different approaches are shown in
Figure 3.4.

0.4 0.4
b 0.5 sec
a
2 sec
0.3 0.3
5 sec
E(t)

E(t)
0.2 0.2

0.1 0.1

0 0
0 10 20 0 10 20
t (s) t (s)

0.4 1
c d
0.8
0.3
0.6
E(t)
E(t)

0.2
0.4
0.1 0.2

0.0 0
0 10 20 0 10 20
t(s) t(s)

Figure 3.4: RTDs obtained from different approaches (a) Frozen velocity approach, (b)
Snapshot approach, (c) Data sampling approach and (d) Transient approach.

The RTD obtained from the full transient approach is expected to be the most accurate due to
its ability to encompass unsteady flow dynamics. The other three short-cut techniques were
found to be reasonably good for the prediction of the mean residence distribution time, but fail
to account for the flow behavior of the fluidized bed. The reason is the rapidly changing
velocity profiles in the fluidized bed. It is noted that the gas phase velocity in the bubbles is an
order of magnitude higher than the gas phase velocity in the rest of the dense bed. The rapid
rising of the bubbles is neglected when the flow fields are frozen. Even in the case of snapshot
techniques, this phenomenon could not be accounted for due to very high bubble-rise speeds.
Although computationally expensive, the transient approach appears to be the best in achieving
a qualitatively accurate RTD. An elaborate analysis of different RTD approaches can be found
in Appendix A.

32
In order to understand the mixing pattern of the fluidized bed reactor in a better manner, the
RTD pulse was studied at three surfaces at varying reactor heights. The response to the pulse
input, that is, tracer concentration, was first recorded just above the solid fluidized bed, then at
the middle of bed (in the freeboard region,) and, lastly, at the exit of the bed. The tracer
concentration profiles at these surfaces are presented in Figure 3.5. However, it should be noted
that the time delays in achieving these tracer profiles have been omitted in Figure 3.5(b) in
order to achieve a better understanding of the variation in tracer profiles at different reactor
heights.

3
a
2.5

2
E(t)

1.5

0.5

0
0 2 4 6 8 10 12 14 16 18 20
t(s)

3
b
2.5

1.5 Outlet
E(t)

Middle Bed
1 Above Dense Zone

0.5

0
0 1 2 3 4 5
t (s)

Figure 3.5: RTD (tracer) profile at different reactor surfaces (a) as observed and (b)
normalised.

33
The tracer profile, when measured just above the bed, points toward a CSTR-like behavior,
with a recycle. The fluctuations can be attributed to the recycling as well as the transient nature
of the reactor. Additionally, the similarity between the tracer profiles that were measured at the
middle of the bed and at the outlet point toward a PFR-like behavior in the upper-half portion
of the bed (Figure 3.5). The RTDs predict a complex behavior for the remaining middle portion
of the bed. In order to gain a better understanding of the mixing patterns in the middle portion
of the bed, a tracer was patched just above the solid dense bed (Figure 3.6). The fully transient
simulation was continued and the tracer profile was studied at the middle of the bed. The tracer
profiles that were obtained for tracer injections are shown in Figure 3.6. The tracer profile
points toward a system that consists of two streams of reactors. This observation defies the
common assumption that associates the entire freeboard with plug flow behavior. Thus, better
models are required to predict the flow regimes in the freeboard of a fluidized bed.

0.9
0.8
0.7
0.6
0.5
E(t)

0.4
0.3
0.2
0.1
0
0 2 4 t (sec) 6 8 10

Figure 3.6: RTD obtained for middle section of the fluidized bed reactor

Different compartment models that comprised the ideal reactor types of PFRs (with no axial
dispersion) and CSTRs were modeled in MATLAB® R2010a in order to match the different
RTDs (complete RTD and middle zone RTD) obtained from the CFD simulation. The dense
zone of the fluidized bed was modeled as a CSTR with recycle, followed by a PFR in order to
account for the time delay. The middle portion of the bed, which starts from the end of the
dense zone and ends at the mid-point, exhibited a complex behavior, shown in Figure 3.6. It
was noticed that on exiting the dense bed, the gas bubbles formed wakes just above the bed
(Figure 3.7). This divided the gas flow into parallel streams that rose along the walls and the
centerline. A similar velocity profile was also observed by Dou et al. (2008) for a CFD
simulation of a fluidized bed. Thus, the middle portion of the bed was modeled as two parallel

34
streams of reactors. One stream consisted of one CSTR and the other consisted of a PFR
followed by two CSTRs (of equal size). The combined stream was introduced into PFR to
account for the time delay. The top portion of the bed was modeled as a single PFR, because
the flow pattern remained undisturbed in the top portion of the bed. This phenomenon can be
seen in Figure 3.5.

Figure 3.7: Velocity vectors in the middle section of the bed.

The complete compartment model for the fluidized bed has been presented in Figure 3.8. The
Levenberg–Marquardt algorithm (Moré, 1978) was employed to acquire the best range of
compartment model parameters, by minimizing the sum of the square of errors for RTDs that
were obtained by CFD simulation and MATLAB simulation. The NLINFIT optimization tool
in MATLAB was used to estimate the model parameters. The residence time of all the reactors
(9 reactors); the percentage of recycle for the dense zone; and the percentage of split for the
middle bed were estimated using MATLAB in order to achieve the best result. The percentage
standard error for a 95 percent confidence interval was less than two percent for all coefficients.
The MATLAB algorithm utilizes an asymptotic normal distribution in the computing of the
confidence intervals. A comparison of the RTDs that were obtained from the CFD simulation
and the MATLAB code is presented in Figure 3.9. The mean residence times for the RTDs
from the CFD model and the MATLAB compartment model were 10.89 s and 10.85 s,
respectively. The variance of the two models was calculated to be 1.84 s2 and 2.51 s2,
respectively.

35
Figure 3.8: Compartment model for the fluidized bed.

0.9

0.8 MATLAB
Compartment
0.7 Model

0.6 RTD from CFD


simulation
0.5
E(t)

0.4

0.3

0.2

0.1

0
0 5 10 15 20
t(s)

Figure 3.9: Comparison of RTD obtained from the CFD model, and the RTD obtained from
MATLAB compartment model.

36
3.4 ASPEN Plus model
The MATLAB compartment model that was based on the CFD model was implemented in
ASPEN Plus simulation software. The ASPEN Plus model (Figure 3.10) helped in the
formulation of a steady-state model for complex gasifier operations. The Redlich Kwong Soave
Boston Mathias (RKS-BM) property method was employed for thermodynamic property
calculations. The pyrolysis, gasification and combustion section of the DFBG were decoupled
for a better understanding of biomass devolatization and heat integration between the
combustor and the gasification chamber. The biomass, char, and ash were modeled as non-
conventional solids without Particle Size Distribution (PSD). The proximate and ultimate
analysis of the different biomass feedstocks that were studied are listed in Table 3.2.

Figure 3.10: ASPEN Plus compartment model.

37
Table 3.2: Proximate and ultimate analysis of biomass feedstocks.
Proximate analysis (weight %) Wood Straw
Bagasse
(Eucalyptus) Pellets
Moisture (as-received) 15.00 6.80 7.00
Fixed Carbon (dry basis) 16.20 23.00 16.00
Volatile Matter (dry basis) 83.80 73.00 83.10
Ash (dry basis) 1.16 3.90 0.90
Ultimate Analysis (weight %) (dry basis)
Carbon 47.41 45.24 46.88
Hydrogen 6.12 5.87 6.14
Nitrogen 0.20 0.48 0.29
Chlorine 0.05 - -
Sulfur 0.02 - -
Oxygen 45.00 44.66 45.78
Ash 1.15 3.75 0.89

Initially, simulations were formulated and validated for a woody (eucalyptus) biomass, since
the majority of the data that is available in the literature is for a woody biomass. The simulation
begins with the drying of the biomass. After the biomass dries, it is sent to the pyrolysis reactor.
The heat for pyrolysis is provided by the recycled bed material (assumed to be silica sand). The
pyrolysis is assumed to be instantaneous. The experimental data that is reported by Dufour et al.
(2009a) for woody biomass is used to predict the yields of gas, tar and char for biomass
pyrolysis and to predict the gas and char composition. The data is applicable to a temperature
range of 700 ºC to 1000 ºC and to particle heating rates that vary from 20 ºC/s to 40 ºC/s. These
heating rates are in agreement with experimental measurements for fluidized bed gasifiers
(Abdelouahed et al., 2012). A necessary simplification is that tars, which are a mixture of more
than 200 compounds have been represented by only four compounds, namely, benzene (C6H6),
phenol (C7H8), toluene (C6H6O), and naphthalene (C10H8). Correlations have been formulated
to give the mass yields of each compound (Abdelouahed et al., 2012) and the char composition,
on the basis of the pyrolysis reactor temperature. The correlations are quadratic equations that
are represented in Equation 3.6.

38
𝑌𝑖 = 𝑎𝑇 2 + 𝑏𝑇 + 𝑐 (3.6)

where Yi is the mass yield of different pyrolysis products that are based on an anhydrous
biomass input (or an elemental composition in the case of char), T is the temperature of the
pyrolysis reactor and a, b and c are constants that are reported in the Appendix B.

The pyrolysis reaction occurs instantaneously on the arrival of the biomass in the fluidized bed
gasifier. It is thereafter mixed with steam, which acts as a gasifying and fluidizing medium.
The product stream is separated as input to the gasifier and the combustor. The bed material
and a majority of the char are sent to the combustor. The char is burned in the combustor, which
heats up the bed material. A cyclone, which is modeled using a splitter, is used to separate the
flue gases and the hot bed material. Fifteen percent of the bed material is also purged at this
stage. This assumption, although higher than that which is experienced in practice, helps to
account for heat loses and maintains a desirable amount of heat transfer between the gasifier
and the combustor. All the reactors were modeled as adiabatic reactors and, hence heat losses
to the surroundings are not accounted.

The syngas and tars produced from pyrolysis are sent to the gasification reactor. Based on the
data reported in literature (Thapa et al., 2014; Doherty et al., 2013), five percent of the char
produced in the pyrolysis products is sent to the gasifier and the rest is sent to the combustor.
The compartment model developed in MATLAB formed the basis of the gasifier model. In the
dense zone, heterogeneous tar-cracking reactions take place. The gases then enter the freeboard
that comprises the middle and the top portions of the bed. Here, homogenous gasification
reactions take place. The reactions that were considered were the water-gas shift reaction, the
water-gas reaction, the methanation reaction, the steam-methane reforming reaction, and the
tar-reforming reactions. Hydrogen and steam are reported to play an inhibiting role in the
steam-methane reforming reaction in the gasification atmosphere (Dufour et al., 2009b). Thus,
appropriate kinetics must be chosen in the case of a reducing environment. The reactions
kinetics used in the modeling of the gasifier is listed in Table 3.3.

39
Table 3.3: Reaction kinetics for homogenous and heterogenous reactions.

Reaction Kinetics Reference

Heterogeneous reaction

61000 (Abdelouahed et al.,


𝐶10 𝐻8 → 9𝐶 + 1⁄6 𝐶6 𝐻6 + 7⁄2 𝐻2 𝑟 = 21.11 𝑒𝑥𝑝 (− ) 𝐶𝐶10 𝐻8
𝑅𝑇 2012)

61000 (Abdelouahed et al.,


𝐶6 𝐻6 + 𝐻2 𝑂 → 3𝐶 + 2𝐶𝐻4 + 𝐶𝑂 𝑟 = 21.11 𝑒𝑥𝑝 (− ) 𝐶𝐶6 𝐻6
𝑅𝑇 2012)

61000 (Abdelouahed et al.,


𝐶7 𝐻8 + 𝐻2 → 𝐶6 𝐻6 + 𝐶𝐻4 𝑟 = 21.11 𝑒𝑥𝑝 (− ) 𝐶𝐶7 𝐻8
𝑅𝑇 2012)

263000 (Abdelouahed et al.,


𝐶𝐻4 → 𝐶 + 2𝐻2 𝑟 = 10−2 𝑒𝑥𝑝 (− ) 𝑝𝐶𝐻4
𝑅𝑇 2012)

Homogenous reactions

263000
𝐶𝑂2 + 𝐻2 → 𝐶𝑂 + 𝐻2 𝑂 𝑟 = 1.26 ∗ 105 𝑒𝑥𝑝 (− ) 𝐶𝐶𝑂2 𝐶𝐻2 (Basu, 2010)
𝑅𝑇

12560
𝐶𝑂 + 𝐻2 𝑂 → 𝐶𝑂2 + 𝐻2 𝑟 = 2780 𝑒𝑥𝑝 (− ) 𝐶𝐶𝑂 𝐶𝐻2𝑂 (Basu, 2010)
𝑅𝑇

211000 1.48 −0.11 −0.91


𝐶𝐻4 + 𝐻2 𝑂 → 𝐶𝑂 + 3𝐻2 𝑟 = 4.7 ∗ 104 𝑒𝑥𝑝 (− ) 𝑝𝐶𝐻4 𝑝𝐻2𝑂 𝑝𝐻2 (Dufour et al. 2009b)
𝑅𝑇

49884
𝐶 + 𝐻2 𝑂 → 𝐶𝑂 + 𝐻2 𝑟 = 2 ∗ 108 𝑒𝑥𝑝 (− ) 𝐶𝐻2𝑂 (Inayat et al., 2009)
𝑅𝑇

148995
𝐶 + 2𝐻2 → 𝐶𝐻4 𝑟 = 120 𝑒𝑥𝑝 (− ) 𝐶𝐻2 (Inayat et al., 2009)
𝑅𝑇

350000 1.6 (Abdelouahed et al.,


𝐶10 𝐻8 → 9𝐶 + 1⁄6 𝐶6 𝐻6 + 7⁄2 𝐻2 𝑟 = 3.4 ∗ 1014 𝑒𝑥𝑝 (− ) 𝐶𝐶10𝐻8 𝐶𝐻−0.5
𝑅𝑇 2
2012)

443000 1.3 −0.4 0.2 (Abdelouahed et al.,


𝐶6 𝐻6 + 𝐻2 𝑂 → 3𝐶 + 2𝐶𝐻4 + 𝐶𝑂 𝑟 = 4 ∗ 1016 𝑒𝑥𝑝 (− ) 𝐶𝐶6𝐻6 𝐶𝐻2 𝐶𝐻2𝑂
𝑅𝑇 2012)

247000 (Abdelouahed et al.,


𝐶7 𝐻8 + 𝐻2 → 𝐶6 𝐻6 + 𝐶𝐻4 𝑟 = 1.04 ∗ 1012 𝑒𝑥𝑝 (− ) 𝐶𝐶7 𝐻8 𝐶𝐻0.5
𝑅𝑇 2
2012)

40
3.5 Results and discussion
The results that were obtained were compared with the data that was available for the actual
gasifier operation (E4tech, 2009). The comparison is presented in Table 3.4.

Table 3.4: Simulation results and validation.


Variable Literature This Model
(E4tech, 2009) (Wood Feedstock)
Biomass input (Dry) (t/hr) 1.68 1.68
Output Gas (Dry basis)
H2 (%) 38–45 41.7
CO (%) 20–25 19.9
CO2 (%) 20–23 18.7
CH4 (%) 9–12 8.8
C2H4 (%) 2–3 1.8
C2H6 (%) 0.5 -
C6H6 (g/Nm3) 8 3.2
C7H8 (g/Nm3) 0.5 0.7
C10H8 (g/Nm3) 2 0.1
Steam-to-Fuel Ratio (kg/kg dry biomass) 0.2–0.8 0.4
Gas yield (kg/kg dry biomass) 0.9–1.1 0.85
Gasifier Temperature (°C) 850–900 891
Combustor Temperature (°C) 950–1000 965
Bed material circulation (kg/kg dry biomass) ~50 (olivine) 34 (silica sand)

The gas composition is within the range reported in the literature. The bed material circulation
is lower than the one used in experiments because olivine was used as the bed material in the
experiments, and this thesis assumes sand as the bed material. The results were compared with
a simplified compartment model in which the dense zone is assumed to be a CSTR and the
complete freeboard is assumed to be a PFR (the other aspects of the gasifier model were not
changed.) The results of the simplified model differed mainly in terms of a higher tar yield,
which increased by a factor of two, from 6.38 g/Nm3 to 11.2 g/Nm3. The steam conversion
increased from 26 percent to 32 percent and the syngas outlet temperature (from the gasifier)
decreased from 891.1 °C to 858.1 °C. Hofbauer and Rauch (2000) have reported a tar yield of
3.8 g/Nm3 and a water conversion of 18 percent for their lab-scale model of DFBG (the steam-
to-biomass ratio was 0.4 kg/kg dry biomass.) The water-gas shift reaction that reached
equilibrium buffered any variation in the composition of the final gas. The application of the
CFD-based compartment model would, therefore, become more significant in the case of a

41
shorter freeboard, which would inhibit the water-gas shift reaction from reaching equilibrium.
Additionally, simplified models may not be the best choice for studies that focus on the tar
production behavior.

The ASPEN Plus model can be modified to predict the gasification results for different biomass
species. This is done by adopting the pyrolysis mass yields of different biomass feedstocks.
High temperature pyrolysis mass yields for birch wood (another hardwood similar to
eucalyptus), wheat straw pellets, and bagasse have been reported by Zanzi et al. (2002). These
mass yields, although not as elaborate as those reported by Dufour et al. (2009a), can be used
to formulate a linear correlation (shown in the Appendix B) between the mass yields and the
temperature, in a typical temperature range of 800 to 1000°C. The results that were obtained
by using these correlations are presented in Table 3.5. Although, the actual gasification results
would also depend on the physical characteristics of the biomass feedstock, such parameters
would be difficult to account for in a simulation model. Since the CHP gasifier has been
developed for the wood feedstock, it is not possible to validate the results for other biomass
feedstocks. Nonetheless, this model can play a constructive role in predicting the output of
syngas in the gasification of different biomass feedstocks.

Table 3.5: Simulation results from different biomass feedstocks.


Variable Straw pellets Bagasse Birch wood
Biomass input (Dry) (t/hr) 1.68 1.68 1.68
Output Gas (Dry basis)
H2 (%) 39.85 43.68 39.24
CO (%) 22.62 22.73 24.75
CO2 (%) 21.75 19.46 20.39
CH4 (%) 9.19 6.53 9.7
C2H4 (%) 2.2 2.37 2.67
C2H6 (%) 0.15 0.14 0.12
3
C6H6 (g/Nm ) 1.48 1.02 2.87
C7H8 -
C10H8 -
Steam-to-Fuel Ratio (kg/kg dry biomass) 0.4 0.4 0.4
Gas yield (kg/kg dry biomass) 0.92 0.87 0.86
Gasifier Temperature (°C) 867.7.8 887.9 876
Combustor Temperature (°C) 903.6 905.3 913
Bed material circulation (kg/kg dry biomass) 34 24 34

42
The results for birch (another wood) feedstock are very close to the previous model results for
wood and the experimental data. However, in the case of birch, the char produced was not
enough to heat up the bed material. A small amount (15 percent) of syngas that was produced
had to be recycled, in order to achieve the required heat integration. Recycling of syngas or the
addition of fuel to the combustor is a practice that is also reported in literature (Pröll et al.,
2007; Francois et al., 2012). Moreover, the results that are derived from these alternate
pyrolysis correlations point to the efficacy and the robustness of this model. Since, the CHP
gasifier has been used primarily for woody feedstock, little data is available for different
feedstocks. Syngas variation with a change in the feedstock has been reported by Schmid et al.
(2012); Kitzler et al. (2012); and Pfeifer et al. (2011) for a lab-scale setup of the original CHP
gasifier. All the studies report that the variation of the biomass feedstock does not have a
significant effect on the composition of the syngas. The different biomass feedstocks reported
in these two studies include varieties of woody biomass, wood pellets, bark, bagasse, palm
fronds, and wood/straw mixtures. It has been reported that in wood/straw mixtures, there is an
increase in the yield of syngas. A similar trend has been predicted by simulations as well.
However, the ash-melting behavior of the straw ash must be ascertained in order to avoid any
melting of ash in the gasifier. For the bagasse feedstock, a lower water conversion has been
reported (Kitzler et al. 2012). The simulation results are in agreement with this observation.
The experiments report a lower tar yield for both, the straw/wood mixture and the bagasse
feedstock (Schmid et al., 2012; Kitzler et al. 2012). The simulation results also predict a lower
benzene yield for straw and bagasse feedstocks under similar conditions. The simulation
predicts similar gas yields for bagasse and wood feedstock, which is analogous to the
experimental findings (Kitzler et al. 2012). These results signify that the ASPEN Plus model
can effectively capture the relative gas output trends from changing biomass feedstocks.

Similar to birch wood, in the case of the bagasse feedstock there was also insufficient char
produced for full energy integration. Thus, the syngas also had to be recycled in order to
provide sufficient heat to the combustor. It might be noticed that the temperature difference
between the gasifier and combustor is quite small for the bagasse feedstock. The results depict
the gasifier outlet temperature, which is greater than the dense bed (and pyrolysis) temperature
due to an exothermic water-gas shift reaction that reaches equilibrium in the freeboard. Similar
temperature profiles have been noticed by Thapa et al. (2014) in his CFD simulations for the
BFB.

43
The bed material flowrate strongly dictates the temperatures of the gasifier reactor and the
combustor reactor. An increase in the bed circulation rate would result in a decrease in the
combustor temperature because, now, the heat generated by the combustor is absorbed by a
larger quantity of bed material. This leads to a further small decrease in the pyrolysis
temperature as well as in the gasification temperature, which results from the change in the
composition of the pyrolysis products. A sensitivity analysis was performed in order to study
the effect of a change in the gasifier temperature on the gas quality and quantity. The gasifier
temperature was in turn manipulated by changing the bed material flowrate from 20 to 60 kg
bed material per kg dry biomass for the wood feedstock. The minimum gasifier temperature
achieved at bed material flowrate of 60 kg bed material per kg dry biomass was 880 oC. The
effect of an increase in the temperature on the gas composition is shown in Figure 3.11.

0.5 1800
Gas mole fraction (Dry)

Gas flowrate (Nm3/h)


0.4
1750
0.3
1700
0.2
1650
0.1

0 1600
860 880 900 920 940
Gasifier temperature (oC)
H2 CO CO2
CH4 H2O Gas flowrate

Figure 3.11: The effect of gasification temperature on the quality of the syngas.

An increase in the gasifier temperature results in an increase in the concentration of H2 and


CO, and a simultaneous decrease in the yield of CO2 and CH4. This phenomenon can be
attributed to enhanced biomass pyrolysis, which results in a greater production of H2 and CO
at a higher pyrolysis temperature. However, the water-gas shift reactions at a high temperature
consume more CO. This explains the increase in the concentration of H2 when compared to

44
concentration of CO, as the temperature rises. These results are consistent with the
experimental results that are reported in literature (Pfeifer et al., 2004). In Figure 3.11, a
decrease in moisture or an increase in water conversion is predicted, which is in agreement
with the literature. The extents of the water-gas reaction and the water-gas shift reaction
increase with an increase in the temperature and, thus consumes more steam. A higher pyrolysis
temperature increases the percentage of gaseous products, which results in a greater yield of
syngas. However, this is accompanied by a decrease in the gas calorific value due to the
decrease in the content of CO and CH4. Pfeifer et al. (2004) have predicted a similar trend for
gas yields and calorific values in their experimental setup. The results of this analysis establish
the validity of the DFBG simulation model in predicting the syngas output for varying biomass
feedstocks as well as the effect of changing the process conditions.

3.6 Conclusions
A kinetics based compartment model for a DFBG has been developed using the ASPEN Plus
software. The novelty of this model lies in the use of CFD simulations, which enables the
understanding of the hydrodynamics of a fluidized bed. This approach has helped in
establishing a quasi-steady-state model for an inherently dynamic system. Additionally, the
gas-mixing pattern in the freeboard (splash zone), which is neglected by the two-phase models,
can also be taken into account. The CFD model for a fluidized bed was used to carry out RTD
studies by using four different tracer approaches. The results from a fully transient simulation
were found to be most appropriate for the formulation of a compartment model that was based
on RTD. The other approaches failed to account for the hydrodynamics that is linked to the
tracer flow. The RTD was used to predict a compartment model in MATLAB software. An
ASPEN Plus simulation model was developed by expanding on the MATLAB compartment
model. The decoupling of pyrolysis, gasification and combustion reactors enabled the
understanding of the characteristics of the process. The model results and sensitivity analysis
are found to be in good agreement with reported industrial data. An important aspect of this
model is its ability to predict syngas output for different biomass feedstocks, namely, wood
(eucalyptus), bagasse, and straw. A significant benefit of the compartment model is its easy
amalgamation with any downstream syngas process, which would be required in the modeling
of the biomass-to-ammonia process.

45
Chapter 4: Process modeling

4.1 Introduction
This chapter focuses on the modeling and simulation of small-scale ammonia production from
biomass. The scale of biomass-to-ammonia production is much smaller than conventional
ammonia plants and the effect of the economies of scale is expected to make the plant less
appealing. To overcome this effect, this design incorporates features from the ICI Leading
Concept Ammonia® process. This chapter highlights the process features of the ICI process,
which suit small-scale ammonia plants. The complete biomass-to-ammonia process was
simulated in ASPEN Plus software for three feedstocks. Furthermore, three process
configurations of the biomass-to-ammonia process were modeled on the basis of the
technology that was employed for the reforming of syngas. Heat integration was performed to
improve further the performance of the process. The results from the ASPEN Plus flowsheet
were used to predict the economics of different process variants. The chapter ends with a
discussion on the economic performance of different feedstocks and process configurations for
the proposed biomass-to-ammonia process.

4.2 ICI Leading Concept Ammonia process


Based on the fundamental research work of Fritz Haber in the year 1909, Carl Bosch and his
engineering team refined the ammonia synthesis process to technical operability about a
century ago. Since then there has been no fundamental change in the synthesis reaction itself.
However, the entire process has continued to evolve and has led to a drastic reduction in the
overall energy requirement. At present, KBR, Linde, ICI, and Udhe are the main technology
providers. Although the process flow for all the providers is similar, each incorporates certain
distinct process features that make them unique. These plants operate at a typical capacity of
1000 tpd or more of ammonia production. However, there are a few inherent disadvantages of
these large world-scale ammonia plants. Appl (1993) reports that world-scale ammonia plants
require an “extremely large commitment of capital and feedstock and, in the case of the latter,
there may sometimes be doubts about long-term availability and reliability of supply. Another
unpleasant feature is the lack of flexibility, which makes it difficult to respond to fluctuating
demand by turning down the production rate without losing efficiency or stressing equipment.
The high integration complicates start-up and shut-down, thus drawing out the time needed
and leading to considerable expenditure of natural gas without producing ammonia. Large-

46
scale equipment and machinery need more elaborate maintenance and turn-around
schedules.”

The ammonia production being considered in this work is at a much lower scale when
compared to these world scale ammonia plants that have production capacities of greater than
1000 ton per day (tpd). Thus, it is important to select a base-case design that is competent
enough at a small-scale production of ammonia. The ICI Leading Concept Ammonia process
is an ammonia production process that substantially contrasts with the conventional highly
integrated large scale ammonia plant. It is a small-capacity ammonia plant that can compete
with modern large-capacity plants in terms of energy consumption and specific investment.
The flow sheet energy consumption of the ICI process is 29.3 GJ/t NH3 for a natural gas
feedstock. This energy consumption compares well with 28 GJ/t NH3 for state-of-the-art large-
scale technology (Kroschwitz and Seidel, 20000. Primary features of the plant are (Hicks et al.,
1990):

• Milder primary reformer conditions,

• A gas-heated reformer,

• An isothermal CO shift reactor,

• Pressure Swing Absorption (PSA) for CO2 removal, and

• A lower pressure ammonia synthesis loop.

The PSA also helps in maintaining the H2:N2 ratio as well as in the removal of a portion of CO
and CH4. The isothermal shift and PSA lead to robust plant operations and makes the process
less sensitive to catalyst performance and plant upsets. The plant design is modular, eliminating
any unwanted interactions and interdependencies. This gives the plant the flexibility to operate
efficiently at reduced feed rates. Additionally, start-up and shutdown steps in small-scale plants
are easier and faster, thus reducing operation costs (Arkley et al., 1988). The cost
competitiveness of the ICI ammonia production process can be understood with the help of
Figure 4.1 (Arkley et al., 1988), which depicts the effect of economies of scale on the ICI
ammonia production process as well as on the conventional ammonia production process.

47
Figure 4.1: Effect of economies of scale on ammonia production processes (Arkley et al.,
1988).

To gain a thorough understanding of the ICI methodology, the process was modeled with the
ASPEN Plus simulation software. The model was validated against the plant data that was
reported in the open literature (UNIDO, 1998; Hicks et al., 1990; Appl, 1993). An error of
2.1 percent was realized in the final ammonia production rate. The ASPEN Plus flowsheet
along with the associated stream data is presented in Appendix C. The plant simulation revealed
niche heat sinks for all heat sources, making the plant more self-sufficient vis-à-vis. its heating
requirements. An attempt was made to optimize the original Heat Exchanger Network (HEN)
in order to increase the efficiency of the plant. It was found that the optimum HEN was very
complex and involved considerable stream splitting, which may be hard to realise in practice.
The optimization of the HEN for the ICI ammonia production process has been explained in
Appendix C. It was therefore confirmed that The HEN for the ICI plant has been heat
integrated by matching streams within specific units and thus achieve a balance between energy
efficiency and capital cost. The same stream matching philosophy has been applied for the
biomass-to-ammonia process, where surplus energy is converted to steam for the production
of electricity.

48
4.3 Biomass-to-ammonia process
This section describes the proposed flowsheet for the biomass-to-ammonia process, which was
developed in ASPEN Plus simulation software. The feeding of the biomass to the dual bed
gasifier was chosen as the simulation starting point. A compartment-based reactor model
(described in the previous chapter) was used for the modeling of the dual fluidized bed gasifier.
The gasifier model has been validated against the plant data for the CHP gasifier. Three
biomass feedstocks namely eucalyptus (wood), straw pellets and bagasse were simulated for
syngas production. The bed material parameters were same as reported in Table 3.4 in
Chapter 3. The gasifier and the combustion chambers were modeled as separate reactors and
the heat integration between the two was made possible by the circulation of the bed material.
The output syngas has a high hydrogen content, which is a result of the steam that is used as
the gasifying agent. The use of steam as the gasifying agent also results in a higher methane
content, which needs to be reformed.

The downstream process (syngas conditioning) has been modeled on the basis of the ICI
Leading Concept Ammonia process. The syngas from the gasifier is reformed for the removal
of methane and tars. The reformers were modeled as equilibrium reactors for faster and easier
convergence. An analysis of modeling approaches for the reforming of tars has been provided
in Appendix D. The reforming of syngas is followed by gas cleaning through the use of filters,
scrubbers and coolers for the removal of particulates and inorganic impurities such as halides,
alkali and sulfur compounds. It is assumed that hot gas cleaning technology is used for syngas
cleaning, due to the better efficiency of such technology (Tock et al., 2010), but it should be
noted that the technology is still in its developmental stage. After the cleaning of the syngas, it
is then compressed using a three-stage compressor. The compressed gas then enters an
isothermal shift reactor for conversion of CO to CO2 by the water-gas shift reaction. This thesis
uses a single shift reactor as opposed to the two shift reactors used in conventional plants. This
can be justified, because the absolute carbon-monoxide slip would be small due to the smaller
flowrates and can be taken care of by the PSA and the methanator downstream (Appl, 2007).
Thus, an entire reactor vessel in the form of an additional shift reactor can be avoided.
Furthermore, the heat produced in the shift reactor is used to generate steam which is used as
a reactant in the process for reforming and in the shift reactor.

The desaturator follows the shift reactors and reduces the moisture content of the syngas. The
gas then enters a PSA for the removal of CO2, residual CO, CH4 and inerts. The PSA can also

49
be used to remove excess nitrogen, if any. Thus, the PSA plays a critical role in maintaining
the appropriate H2-to-N2 ratio. The hydrogen recovery of the system was considered to be 87
percent. Also, the PSA is the only place where inerts are removed from the process. The PSA
off-gas also acts as a low calorific value fuel (with an approximate calorific value of 2.4 MJ/kg).
After this, the cleansed syngas enters a methanator for the removal of carbon oxides, which are
poisonous to the ammonia synthesis catalysts. They are converted into methane by the
methanator reactor. After this, a molecular sieve drier is used to strip the gas of any remaining
moisture content. The syngas now enters the Haber–Bosch synthesis loop. Based on the ICI
process, the ammonia synthesis loop is operated at a relatively low pressure (Figure 4.2). This
helps in achieving compressor efficiencies that are similar to the large-scale ammonia plants
(Arkley et al., 1988). The process uses a tube-cooled counter-current reactor that contains a
low pressure ammonia synthesis catalyst and operates at temperatures of up to 400 ºC (Appl,
1993). The product stream is flashed to produce the ammonia product, which is refrigerated by
using a simple refrigeration cycle. The reactors, namely, the shift reactor, the methanator
reaction, and the ammonia synthesis reactor were modeled using the reaction kinetics that are
available in literature. The reaction kinetics used for these systems is reported in Table 4.1. To
prevent the accumulation of methane, a purge (2.6 %) is introduced in the synthesis loop. The
purge is recycled and mixed with the syngas that enters the reformer. The complete flowsheet
for the biomass-to-ammonia process can be seen in Figure 4.2. The operating parameters for
major unit operations are shown in Table 4.2. These parameters are based on the ICI process
and are found to have delivered better results in small-scale operations.

Table 4.1: Reaction kinetics for the shift reactor, the methanator reactor and the ammonia
synthesis reactor.
Reaction Kinetics Reference
Shift Reactor
CO + H2O --> CO2 + H2 52800 Choi and
𝑟 = 3.99 ∗ 103 𝑒𝑥𝑝 (− ) 𝐶𝐶𝑂
𝑅𝑇 Stenger (2003)
Methanator reactor
3 H2 + CO --> CH4 + H2O 77070 Hayes et al.
𝑟 = 8.79 𝑒𝑥𝑝 (− ) 𝑃𝐶𝑂 −0.87 𝑃𝐻2 1.27 𝑃𝐻2𝑂 −0.13
𝑅𝑇 (1985)
Synthesis reactor
3 H2 + N2 --> 2 NH3 87090 Araujo and
𝑟 = 1.49 ∗ 106 𝑒𝑥𝑝 (− ) 𝑃𝑁2 𝑃𝐻2 1.5 𝑃𝑁𝐻3 −1
𝑅𝑇 Skogestad
(2008)
2 NH3 --> 3 H2 + N2 198464
𝑟 = 2.15 ∗ 1016 𝑒𝑥𝑝 (− ) 𝑃𝐻2 −1.5 𝑃𝑁𝐻3
𝑅𝑇

50
Figure 4.2: Flowsheet for the biomass-to-ammonia process.

51
Table 4.2: Operating parameters for the biomass-to-ammonia process.
Parameter Nominal Range
Steam-to-biomass ratio 0.4 0.3–1
Shift temperature (°C) 265 220–400
PSA H2 recovery (%) 87 80–95
Synthesis reactor pressure(Bar) 82 80–300
Synthesis temperature (°C) 380 300–500
H2/N2 ratio 3 2–4
Synthesis loop purge (%) 2.6 2–10

Three different configurations of the biomass-to-ammonia process namely Autothermal


Reforming (ATR), Steam-Methane Reforming (SMR) and CO2 Reforming (CR) have been
modeled and compared (Figure 4.3). The configurations differ on the basis of the reforming of
the methane and tars that are present in syngas. In the ATR configuration, the reforming
happens in the presence of air. The oxygen is used to oxidize a part of the syngas in order to
provide heat for the endothermic reforming reactions. The air input also provides the nitrogen
that is needed for ammonia production. The functioning of an ATR is analogous to a secondary
reformer in a conventional ammonia plant. In the SMR configuration, steam is used for syngas
reforming. The nitrogen that is needed for ammonia production is added before the ammonia
synthesis loop, using an ASU. A portion of the combustor flue gas from the DFBG is
introduced to the reformer in the CR configuration. The CO2 and N2 in the flue gas, act as a
reforming agent and a nitrogen source, respectively. The flue gas is filtered of particulates
(resulting from unfiltered fly-ash) and other impurities, before its use as an oxidizing agent.
The excess nitrogen in the flue gas that enters the reformer is removed with the help of the
PSA. The heat that is required for the reforming in the SMR and the CR configurations is
provided by the combustion of an external fuel. Since the ATR consumes a part of the syngas
to provide heat for the endothermic reforming reactions, the final ammonia yield is lower in
the ATR configuration than in the SMR and CR configurations.

52
Figure 4.3: Biomass-to-ammonia process configuration: ATR, SMR and CR.

4.4 Energy Integration


The three flowsheet configurations and the HEN were based on the ICI design philosophy. The
ICI process for ammonia production from natural gas uses a heat exchange reformer, in which
heat from the cooling of secondary reformer syngas is utilized to heat the primary reformer.
However, the absence of primary reforming in the biomass-to-ammonia flowsheet created a
scope for additional heat integration. Sensible heat could be potentially extracted from the
cooling of the syngas outlet from SMR/ATR/CR. The ICI flowsheet utilized this heat for
primary reforming or gas-heated reforming. Additional heat could also be extracted from the
ammonia synthesis loop (also available in the ICI plant) and the cooling of the combustor flue

53
gas. The PSA off-gas can also be used as a low calorific value fuel. A potential application is
the preheating of air in the ATR configuration.

The extra heat can also be effectively used for the production of electricity. The Grand
Composite Curve (GCC) of all the plant configurations reflects the scope for utility steam and
electricity generation. Electricity, if taken from a grid, would add to the production cost and
carbon footprint of the ammonia produced. Electricity is utilized mainly to run the compressors.
The ammonia production cost could be reduced if the waste heat (~5-7 MW) available is
utilized for electricity production. The flowsheets of all the process configurations were altered
to incorporate electricity production from waste heat. The heat available in the plant was used
to produce superheated steam at 40 bar, ~410 0C. A back pressure turbine was used to generate
electricity from this steam. A part of the let-down steam was used in the gasifier and the drying
of the biomass. A condensing turbine was employed to produce additional electricity from the
remaining steam. The process diagram for electricity production is shown in Figure 4.4.

1200 1000
1000 800
Temperature (0C)
Temperature (0C)

GCC
800 GCC 600
600 Steam
400
Production
400 Steam
200
Production
200
0
0 0 5 10 15
0 5 10 15
Heat Flow (MW)
Heat Flow (MW)

Figure 4.4: Electricity generation cycle and process GCC and demonstrating the automation
of the steam production curve. The LHS graph has greater waste heat available and produces
more steam compared to RHS.

54
In order to automate the heat-integration process, two spreadsheets are prepared in ASPEN
Plus software. The first spreadsheet imports the duty and temperatures of various heat sources
and sinks that are available in the flowsheet. Based on the imported data, a GCC is prepared,
and the steam generation potential is calculated (Smith, 2005). The steam that is generated is
used to produce electricity through the electricity-generation process that is shown in
Figure 4.4. The second spreadsheet imports the data for the biomass-to-ammonia process as
well as electricity generation process. An updated GCC is now prepared to calculate additional
utility requirement, namely, the required cooling duty. This helps in closing the GCC. Based
on the biomass-to-ammonia process, the electricity production process, and the calculated
utility requirements, hot and cold composite curves that represent the entire process are
formulated. Figure 4.4 depicts the automated change in the GCC and the steam composite curve
for two different process configurations. On the left hand side graph maximum waste heat is
available and more steam is produced compared to the right hand side graph. The hot and cold
composite curves are used to estimate the total heat exchanger area that would be required for
the biomass-to-ammonia process. This is done by dividing the composite curves into discrete
enthalpy intervals. A change in the slope of either the hot or the cold composite curve defines
a new enthalpy interval. The heat-integration methodology has been depicted in Appendix E.
The enthalpy intervals along with overall heat transfer coefficients for individual streams are
used to predict the overall heat transfer area by using Equation 4.1 (Smith, 2005):
1 𝑞𝑖,𝑘 𝑞𝑗,𝑘
𝐴𝑁𝑒𝑡𝑤𝑜𝑟𝑘 = ∑𝐼𝑁𝑇𝐸𝑅𝑉𝐴𝐿𝑆
𝑘
𝐾
[∑𝐻𝑂𝑇
𝑖
𝑆𝑇𝑅𝐸𝐴𝑀𝑆 𝐼
+ ∑𝐶𝑂𝐿𝐷
𝑗
𝑆𝑇𝑅𝐸𝐴𝑀𝑆 𝐽
] (4.1)
∆𝑇𝐿𝑀𝑘 ℎ𝑖 ℎ𝑗

where,
𝐴𝑁𝑒𝑡𝑤𝑜𝑟𝑘 = heat exchanger area for vertical heat transfer in the entire network
∆𝑇𝐿𝑀𝑘 = log mean temperature difference for the interval, k;
𝑞𝑖,𝑘 = stream duty on hot stream, i, in enthalpy interval, k;
𝑞𝑗,𝑘 = stream duty on cold stream, j, in enthalpy interval, k;
ℎ𝑖 , ℎ𝑗 = film transfer coefficients for hot stream, I, and cold stream, j, (including wall and
fouling resistances)
I = total number of hot streams in enthalpy interval, k;
J = total number of cold streams in enthalpy interval, k;
K = total number of enthalpy intervals;
The heat exchanger area thus calculated is used to estimate the cost of HEN for the flowsheet.
The overall heat transfer coefficients that are used for different process streams are provided
in Appendix E.
55
4.5 Process economics
In comparing different processes for the manufacture of ammonia, the profitability of the
operation is bound to be one of the deciding factors. As stated earlier, this thesis looks at three
different configurations for small-scale ammonia production from biomass and predicts the
ammonia production cost for all the configurations. One of the main objectives in optimizing
the process is to improve the economic benefits.

4.5.1 Capital cost

The economic correlations that are proposed by Turton (2009) were used to size various pieces
of equipment. The bare module equipment cost was calculated on the basis of the results of the
simulation. The bare module cost is the sum of the direct and indirect costs of the piece of
equipment that is designed for ambient pressures and is made of common material such as
carbon steel. The correlation can be understood with the help of Equation 4.2.

𝑙𝑜𝑔10 𝐶𝑝0 = 𝐾1 + 𝐾2 𝑙𝑜𝑔10 𝑋 + 𝐾3 [𝑙𝑜𝑔10 𝑋]2 (4.2)

where Cp is the equipment cost, X is the area, volume or duty of the process equipment and Ki
are the specific constants for different pieces of equipment. The values for specific constants
for different pieces of equipment have been presented in Appendix E. The correlations can only
be used for common equipment. In the case of specialized equipment such as the gasifier and
the ASU, the cost was taken directly from previous studies that predict the cost of similar
equipment (Bartels, 2009; E4tech, 2009). A capacity exponent of 0.67 was used for the costs
of equipment that are available at a different scale. Chemical Engineering Plant Cost Index
(CEPCI) (Economic indicators, 2016) was used to account for the effects of inflation. The
relation can be seen in Equation 4.3

𝑆 0.67 𝐼
𝐶𝑝,2 = 𝐶𝑝,1 (𝑆2 ) 2
(𝐼 ) (4.3)
1 1

where Si are the scales of the equipment, and Ii are the inflation indices for different years. In
the proposed process, the majority of the equipment operates at pressures much higher than
ambient. Higher pressure requires an increase in the thickness of the steel. The relationship
between the cost of the piece of equipment and the operating pressure is estimated using
Equation 4.4.

𝑙𝑜𝑔10 𝐹𝑝0 = 𝐶1 + 𝐶2 𝑙𝑜𝑔10 𝑃 + 𝐶3 [𝑙𝑜𝑔10 𝑃]2 (4.4)

56
where Fp is the pressure factor that is to be used to calculate the bare module cost; P is the gauge
pressure; and Ci are constants. The pressure vessels, however, are treated differently because
of risk of injury upon failure (Morgan, 2013). Whose pressure factors can be estimated using
Equation 4.5.

(𝑃+1)𝐷
+𝐶𝐴
2(850−0.6(𝑃+1))
𝐹𝑝,𝑣𝑒𝑠𝑠𝑒𝑙 = (4.5)
𝑡𝑚𝑖𝑛

where P is the operating pressure; D is the vessel diameter; CA is the corrosion allowance,
which is assumed to be 0.00315 m; and tmin is the thickness of the vessel, which is assumed to
be .0063 m.

Often the equipment are manufactured from materials that are more expensive than carbon
steel. In such cases, material factors need to be applied to the equipment costs. A conservative
approach has been taken towards the materials of construction with assumption of stainless
steel SS 316 used for the fabrication of all equipments. However, in practice, only equipment
that operate at a high temperature would be constructed from stainless steel. The bare module
costs are calculated from the equipment cost by applying the pressure factor as well as the
material factors (Equation 4.6.)

𝐶𝐵𝑀 = 𝐶𝑝0 (𝐵1 + 𝐵2 𝐹𝑚 𝐹𝑝 ) (4.6)

Where CBM is the bare module cost and B1 and B2 are equipment specific fitting costs, which
include direct and indirect costs. Once the bare module costs are calculated, they can be
summed to calculate the grass-roots level costs by including the contingency costs, other
miscellaneous costs, and the cost of auxiliary facilities. The contingency costs are assumed to
be 15 percent of the bare module cost and the fees are assumed to be three percent. The
auxiliary costs were assumed to be 50 percent of the bare module cost for the base case
condition, that is, the condition in which ambient pressure and carbon steel were used as the
construction material. This is because the auxiliary costs are not affected by the condition of
the equipment. The grassroots level cost can be calculated from Equation 4.7.

𝐶𝐺𝑅 = 1.18 ∑𝑛𝑖=1 𝐶𝐵𝑀,𝑖 + 0.50 ∑𝑛𝑖=1 𝐶𝐵𝑀,𝑖


0
(4.7)

Where CGR is the total grass roots level costs, CBM is the bare module cost at process conditions
and CBM0 is the bare module cost at ambient pressure with carbon steel as the fabrication
material.

57
The Pressure Swing Adsorption cost made a significant contribution to the total capital cost.
The PSA system was assumed to have six beds and the beds were sized on the basis of an
algorithm that was suggested by Ribeiro et al. (2010). The algorithm begins with assuming a
diameter for the bed. The input velocity is calculated and checked for feasibility on the basis
of the area and the input volumetric flowrate. An adsorption time of 500 s was assumed. The
input carbon dioxide molar flowrate was identified from the simulation, and the number of
moles to be adsorbed per cycle was calculated from the flowrate. The adsorption isotherms that
were reported by Jee et al. (2001) were used to predict the weight of the required adsorbent.
Due to the limitations of the effects of mass transfer and temperature, the effective use of the
adsorbent was assumed to be 50 percent. The volume of the bed was calculated on the basis of
the density and the porosity of the bed, which, in turn, yielded the bed height and column
height.

Cryogenic ammonia storage was modeled on the basis of an approach suggested by Morgan
(2013). Thirty days of storage at -33 ºC and ambient pressure was considered in this approach.
Boil-off of ammonia that results from heat transfer through the storage walls is a normal
phenomenon. A simple compression refrigeration cycle was created in ASPEN Plus to recover
the escaping ammonia vapors. The latent heat of vaporization of ammonia and the heat transfer
though the walls of the storage tank were utilized to calculate the ammonia boil-off. A boil-off
was found to be close to the typical boil-off value of 0.04 percent. The air separation unit cost
and the gasifier system cost were taken directly from literature (Bartels, 2009; E4tech, 2009)
and the equipment costs were calculated using Equation 4.3.

4.5.2 Operating cost

The manufacturing cost is the sum of the costs of all resources that are consumed in the process
of making a product. The manufacturing costs can be divided into three main categories,
namely, the direct cost, which includes the raw material, labor, and maintenance costs; fixed
costs which take into account depreciation, taxes, and insurance; and general expenses, which
represents overheads, and research and development. The empirical correlations that were used
to calculate the cost of manufacturing are reported in Table 4.3 (Turton, 2009).

58
Table 4.3: Operating cost parameters for the biomass-to-ammonia process.
Cost Item Value used
1. Direct Manufacturing Costs
a. Raw materials CRM
b. Waste treatment CWT
c. Utilities CUT
d. Operating labor COL
e. Direct supervisory and clerical labor 0.18*COL
f. Maintenance and repairs 0.04*FCI
g. Operating supplies 0.006*FCI
h. Laboratory charges 0.1*COL
i. Patents and royalties 0.01*COM
2. Fixed Manufacturing Costs
a. Local taxes and insurance 0.014*FCI
b. Plant overhead costs 0.59*COL+ 0.02*FCI
3. General Manufacturing Expenses
a. Administration costs 0.177*COL+ 0.006* FCI
b. Distribution and selling costs 0.02*COM
c. Research and development 0.05*COM
COM=0.093*FCI + 2.225*COL +
Total Costs
1.09*(CRM+CWT+CUT)

The biomass input is fixed at 120 tpd for all feedstocks and flowsheet variations. The
assumptions that were used in the calculation of the operating costs and the final ammonia
production cost of different feedstocks have been reported in Table 4.4. The utility costs are
reported in Appendix E. The number of operators that were needed per shift for the running of
the plant was assumed to be five and thus, the total operating labor required was twenty-eight.
It has been reported, that the ICI ammonia production process required three operators per shift
for normal plant operation (Arkley et al., 1988). Two additional operators have been added to
take care of gasification and biomass handling. The electricity consumption by the ASU was
approximated on the basis of the correlations that were suggested by Hu et al. (2010). Location
factors were utilized in order to convert the capital cost investment estimates for different
countries (Intratec solutions, 2016). The factors used for Australia, India, and Brazil were 1.0,
0.65 and 1.04, respectively, in comparison with the US. Production of ammonia in India
benefits from lower capital and lower operating costs. The economic parameters such as project
life, plant availability and discount rate were based on the data provided by the project sponsor,
which is based on their experience with ammonia production plants.

59
Table 4.4: Economic parameters for biomass-to-ammonia process.
Eucalyptus Straw Bagasse
(Australia) (India) (Brazil)
Project life (years) 20 20 20
Plant availability (%) 90 90 90
Discount rate (%) 15 15 15
Feedstock Price($/t) 100a 30b 20c
Electricity Price ($/MWh) 150d 65e 100f
Labor Cost ($/annum) 90000g 15000h 10500i
Location factor 1j 0.65j 1.04j
a
Feedstock gate price varies from $ 70 to $ 150 per oven dry ton (Rutovitz, 2004). A
representative price of $100 has been chosen. Delivered cost is made up of 25 percent
production cost of crops, 50 percent harvesting and 25 percent transport.
b
Delivered cost is made up of 50 percent biomass handling and storage charges, 25
percent farmer’s compensation, and 25 percent processing charges (Singh and Gu, 2010).
c
Based on estimates reported by (Santos et al., 2016).
d
Based on estimates reported by Australian Energy Market Operator (2016)
e
Electricity supply from grid for biomass-based plants (Jana and De, 2015).
f
Based on Brazilian electricity cost reported by (Khatiwada et al., 2016)
g
(Rimos et al., 2014)
h
Average salary of a shift operator.
i
(Santos et al., 2016)
j
Based on data reported by Intratec solutions (2016)

4.6 Results
The flowsheet results and economic analysis of the biomass-to-ammonia process for different
feedstocks and process configurations is presented in Table 4.5. The ammonia production cost
build-up for different process variants is illustrated in Figure 4.5 and Figure 4.6. All the costs
are reported in terms of 2016 US dollars.

60
Table 4.5: Flowsheet results and economic predictions for different variants of biomass-to-
ammonia process.
Wood Wood Wood Straw Straw Straw Bagasse Bagasse Bagasse
ATR SMR CR ATR SMR CR ATR SMR CR
Operating conditions

Feed (kg/hr) 5000 5000 5000 5000 5000 5000 5000 5000 5000

Gasification temperature(ºC) 876.20 875.85 876.20 864.54 864.53 861.90 888.69 885.28 885.28

Combustion temperature (ºC) 956.23 953.76 956.23 921.33 921.33 918.30 902.77 898.55 898.55

Gasifier Syngas yield (kg/hr) 6487.20 6496.89 6487.20 6605.17 6614.88 6602.32 6178.68 6176.33 6176.33

Reforming temperature (ºC) 827.30 899.90 861.39 878.48 900.00 821.02 989.12 803.52 804.40

Flowsheet output

Tars (g/Nm3) 17.06 17.18 17.06 3.64 3.64 3.79 2.31 2.37 2.37

Electricity Production (MWhr) 1.49 1.84 2.23 2.31 2.02 1.92 1.58 1.67 1.87

Electricity required from grid 1.65 1.89 1.84 1.52 1.88 1.88 1.27 1.86 1.63
(MWhr)
Ammonia Production (Kg/hr) 2393.83 2703.68 2716.49 2283.57 2640.23 2631.90 2095.34 2411.24 2439.97

Economic predictions

Capital Cost ($) 5.72E+07 6.53E+07 5.79E+07 3.70E+07 4.22E+07 3.72E+07 5.77E+07 6.51E+07 5.80E+07

Operating Cost ($) 1.69E+07 1.85E+07 1.77E+07 4.46E+06 5.26E+06 4.89E+06 8.69E+06 1.02E+07 9.34E+06

7.00
Ammonia plant grass-root level cost (Million $)

6.00

5.00

4.00 ASU
Electricity generation
3.00
Ammonia Loop
2.00 Gas Conditioning
Gasifier
1.00

0.00

Figure 4.5: Biomass-to-ammonia process grass-root level cost.

61
1600

1400
Annual investment
Research and development
1200
Ammonia Production Cost ($/t NH3)

Distribution and selling costs

1000 Administrative costs


Taxes and insurance

800 Plant Overhead


Patents and royalities

600 Laboratory charges


Operating supplies

400 Maintenance
Labour cost

200 Catalysts and utilities


Electricity

0 Biomass Feedstock

Figure 4.6: Ammonia production cost build-up.

The results predict that the use of wheat straw feedstock in India lead to the most economical
biomass-to-ammonia process. The reason is the lower capital cost in the Indian manufacturing
scenario, which is attributed to a low location factor. The capital cost contributes 35 percent,
45 percent, and 52 percent of the total ammonia production cost for the wood, straw, and
bagasse feedstock. The capital cost contribution to the total ammonia production cost is the
highest for the bagasse feedstock because of the lower ammonia production per unit of biomass
input and a slightly higher location factor. The lower ammonia production in the bagasse
feedstock is a result of a lower syngas yield from bagasse gasification. The labor cost
contribution is lower for straw and bagasse feedstocks due to lower wages in India and Brazil.
The ATR plant configuration has the lowest ammonia output for all three feedstocks, because
a part of the syngas is consumed in order to generate heat for autothermal reforming. The SMR
plant configuration has a higher output, which is also accompanied by higher capital and
operating costs. The use of an ASU contributes to a higher capital cost (Figure 4.5), and the
higher syngas flowrates lead to higher compression costs (operating costs). It is observed that
although the output from the SMR plant is 14 percent higher than that from the ATR plant, a

62
higher total production cost of the SMR plant normalizes the final ammonia production cost
per unit of ammonia. Thus it is seen that both the ATR and SMR configurations have similar
ammonia unit production costs. The CR plant is the most economic for all three feedstocks due
to the absence of the ASU, which is costly. Additionally, in the CR configuration, syngas is
not combusted to provide heat for the reforming reactions. The biomass feedstock contribution
is different for the three different process configurations because the production costs are
normalized by the total ammonia production costs.

The results above were compared with other studies that undertake the economic analysis of
ammonia production from biomass (Figure 4.7). The comparison highlighted the economic
advantages that are a result of the process concepts of the ICI plant. Furthermore, any small-
scale ammonia production facility would generally be larger than the scale adopted in this
thesis and would benefit from increased economies of scale.

4000 Sarkar et al. (2011)


Ammonia Price ($/t NH3)

3500 Bartels (2008)


3000 Gilbert et al.(2014)
2500 Tock et al. (2014)
2000 Anderson et al. (2014)
1500 This study(Wood)
1000 This study (Straw)
500 This study (Bagasse)
0
10 100 1000
Ammonia Production Scale (tpd)

Figure 4.7: Comparison of the production prices of ammonia in various studies

4.7 Conclusions
A thorough analysis of the economic viability of the biomass-to-ammonia process has been
performed. Three biomass feedstocks namely wood, straw pellets and bagasse, which are
available in Australia, India and Brazil, respectively have been compared to understand the
effect of biomass composition and the national economic scenario on the viability of the
biomass-to-ammonia process. Each feedstock has been further modeled for three process
configurations. The wood feedstock yielded the maximum possible ammonia production. The

63
LCC was found to be the lowest in the Indian scenario, followed by Brazil and Australia. Lower
capital costs and operating costs in India, made the straw feedstock economically favorable. It
was also noticed that the incorporation of the small-scale features of the ICI process and the
use of dual fluidized bed gasifiers in the biomass-to-ammonia process have helped to reduce
the effect of the economies of scale. This innovative combination of new and old technologies
at the front-end and tail-end of the plant, respectively can help in the identification of
applications for biomass-to-ammonia concept, which might otherwise be constrained by a lack
of supply of biomass feedstock.

64
Chapter 5: Life Cycle Assessment

5.1 Introduction
Conventional ammonia production that is based on hydrocarbon feedstock is known to be
energy intensive and tends to make a substantial contribution to the global greenhouse gas
emissions. Therefore, an environmentally benign feedstock in the form of biomass is proposed
in this thesis as an alternative. Biomass, when used as a feedstock for ammonia production is
expected to yield a considerable reduction in environmental impacts. This thesis chapter
undertakes a cradle-to-gate Life Cycle Assessment (LCA) for ammonia production from
biomass through the gasification route. The cradle-to-gate approach encompasses the product
stages, which begin with the production of raw material and ends with the factory gate (finished
product). The use of the product and the subsequent disposal of the product are omitted in this
approach. Three different biomass feedstocks namely wood, straw and bagasse are compared
for their environmental sustainability by using different environmental indicators.
Furthermore, these feedstocks are modeled for cultivation in three different geographical
regions. This analysis would assist in understanding the biomass-specific environmental
impacts as well as geography-specific environmental impacts.

5.2 Life cycle assessment


The environmental impacts of the biomass-to-ammonia process are estimated using the LCA
framework. This framework involves four major steps (ISO, 2006):

1. determining the goals and scope of the LCA;


2. compiling an inventory of energy and mass inputs and outputs across all relevant
life cycle stages;
3. evaluating relevant environmental impacts that are associated with the life-cycle
inputs and outputs; and
4. interpreting the results, which will lead to a more informed decision making and
improvement of results.
These four mandatory stages are governed by ISO norms (ISO, 2006). The stages are presented
in Figure 5.1.

65
Figure 5.1: LCA framework.

The goal and scope definition step is used to state the intended applications of the study, the
audience of the study, and the methods (and assumptions) that are undertaken. In the Life Cycle
Inventory (LCI) analysis, all the mass and energy flows in the different process stages are
tabulated. The inventories are characterized in the impact assessment stage in order to predict
various midpoint and endpoint impacts. Any additional results are evaluated in the
interpretation stage to predict the impacts of the process, while stating the limitations and
providing future recommendations. The flows between the different LCA stages suggest that
the whole process requires several iterations to reconcile the goals, data requirements, available
inventories, impact categories, and characterization models.

The goal and scope definition step provides the necessary details and transparency that make
an LCA study useful to its intended audience. The goal of this thesis chapter is to identify the
environmental “hot spots” in the life cycle of ammonia production from a biomass feedstock
through the gasification route. The purpose of this thesis is to understand and compare the
environmental impacts that result from different process stages and different process
configurations. The thesis also aims to minimize the overall environmental impact of the
process through the application of MOO, which will be explained in the next chapter. Another
important decision to be made is the choice between attributional and consequential LCA. The
aim of attributional LCA is to describe the environmental impacts that can be attributed to a
particular system or product. A consequential LCA aims to describe the environmental impacts
that are induced by a product in a surrounding environment. An example of the consequential
LCA can be the effect on food supply systems as a result of land use for biofuels. The main

66
characteristics of both the approaches are listed in Table 5.1 (Thomassen et al., 2008). The
LCA study in this thesis will focus on the attributional LCA of the biomass-to-ammonia
process. The selection of the functional unit is another important parameter. The purpose of the
functional unit is to quantify the identified functions in a more precise manner, which will
facilitate mathematical analysis. A functional unit serves as a reference to which the inputs and
outputs of our life-cycle system are normalized. The present thesis has adopted a functional
unit of production of 1 kg of ammonia from biomass feedstock. This choice of the functional
unit was made to facilitate a comparison between the present thesis and other studies on
ammonia production.

Table 5.1: Overview of consequential and attributional LCA (Thomassen et al., 2008).
Characteristic Attributional LCA Consequential LCA
Synonym Status quo Change-oriented
Data Average historical Marginal future
Knowledge required Physical mechanisms Physical and market mechanisms
Functional unit Represents static situation Represents change in volume
System boundaries Static processes Affected processes by change in demand
System expansion Optional Obligatory
Quality Sensitive to uncertainties Higher sensitivity to uncertainties
A cradle-to-gate LCA study was carried out with the help of MS Excel worksheets for different
biomass-to-ammonia process variants. The LCI data was extracted from the ASPEN Plus
flowsheets for the biomass-to-ammonia process as well as the ecoinvent databases (Weidema
et al., 2011). The ASPEN Plus flowsheets provided the material and energy balances for major
process unit operations. Ecoinvent databases were used to supply data for background
processes such as biomass cultivation, transportation, electricity production and the like. A
summary of the mass and energy flows that were considered in the LCA is reported in
Table 5.2. A system boundary for the biomass-to-ammonia process is shown in Figure 5.2.
Figure 5.2 displays all the major process stages and the inputs and outputs (from both nature
and the techno-sphere, that is, man-made world) that are considered in the LCA. Except for
biomass production, all the other stages have been modeled in ASPEN Plus. The ecoinvent
inventories that were used for biomass production, transportation, catalysts and other
background processes have been reported in Appendix F. CO2 emissions and intake during
cultivation have been treated as carbon cycles. CO2 emissions that originate from biomass have
been termed as biogenic and have not been considered in the analysis.

67
Table 5.2: Summary of mass and energy flows that were considered in the LCA of the
biomass-to-ammonia process.
LCI elements considered Process stage Reference quantity
Wood/straw/bagasse production Biomass production From ASPEN Plus model
Biomass transportation Biomass production 100 km transport for wood
and straw feedstock
Electricity Gasification, compression, From ASPEN Plus model
refrigeration and steam turbines
Infrastructure Whole plant 5.79*10-6 p/Nm3 syngas
Silica sand Gasification 9.5 kg/t biomass
Fuel (Natural gas) Gasification 0.16 MJ/t biomass
Charcoal Gasification 5.35 kg/t biomass
Calcium carbonate Gasification 5.76 kg/t biomass
Fly ash disposal Gasification 9.71 kg/t biomass
Bottom ash Gasification From ASPEN Plus model
Reforming catalyst Reforming From ASPEN Plus model
Dolomite Reforming From ASPEN Plus model
Fuel (natural gas) Reforming From ASPEN Plus model
Rape methyl ester Gas cleaning 3.95 kg/t biomass
Sodium hydroxide Gas cleaning 0.16 kg/t biomass
Sulfuric acid Gas cleaning 2.78 kg/t biomass
Zinc oxide Gas cleaning 0.51 kg/t biomass
Shift catalyst Shift reactor From ASPEN Plus model
Adsorbent PSA From ASPEN Plus model
Methanation catalyst Methanator From ASPEN Plus model
Synthesis catalyst Synthesis reactor From ASPEN Plus model
Waste water Gasification and desaturator From ASPEN Plus model
Fly ash disposal Gasification 60 % to incineration and 40 %
to landfill
Auxiliary transport (chemicals, Whole process 100 km by road and 600 km
catalysts, adsorbents and bed by rail
material)
Landfill (Waste chemicals, Waste disposal From ASPEN Plus model and
catalysts, adsorbents and bed correlations
material)
Landfill transportation Waste disposal 10 km by road

68
Figure 5.2: Biomass-to-Ammonia process system boundary.

The model for the inventory analysis that was developed by Heijungs and Suh (2002) has been
used in this thesis. The approach involves the use of matrix algebra to reconcile the various
unit process inventories that are taken from ASPEN Plus and the ecoinvent databases. This
thesis considers a total of nineteen hundred different types of emissions that are a result of
different unit processes. Each unit process is represented by a matrix column; the rows indicate
the process inputs and the outputs including the emissions. The matrix columns that represent
different unit processes together form the process matrix, M. The process matrix is divided into
a technology matrix (A) that represents the mass and energy flows among unit processes, and
an intervention matrix (B) that represents flows from and into the environment such as CO2
emissions. The combined inventories for all the relevant processes are combined to form a
technology matrix and an intervention matrix as shown in Equation 5.1.

𝐴
𝑀 = (𝐵) (5.1)

The final output from the complete process is represented by vector, m. The vector q maybe
further divided two vectors: vector, d, which represents the functional unit or the intended
process output for normalizing all process flows; and vector, g, which represents the total
process emissions that are a result of normalized process flows (Equation 5.2).

𝑑
𝑚 = (𝑔) (5.2)

69
The relationship between the technology matrix (A) and the final demand vector (d) can be
established with the help of a scaling vector, s, as shown in Equation 5.3. Since both the
technology matrix and the final demand matrix would be known, the scaling vector can be
calculated using Equation 5.4.

𝐴𝑠 = 𝑑 (5.3)

𝑠 = 𝐴−1 𝑑 (5.4)

The scaling vector is then used to calculate the process emissions pertaining to the intended
process output or the functional unit (Equation 5.5).

𝑔 = 𝐵𝑠 (5.5)

To summarize, the technology matrix is solved for a particular process configuration (or the
selected functional unit) and the related intervention matrix gives the relevant emissions for
the same process configuration (or the selected functional unit). The multi-functionality of the
unit processes has been solved by using system-expansion (substitution) methodology. This
methodology involves adding stand-alone unit processes to simplify algebraic operations in
matrices.

The environmental emissions that are calculated for the complete process are used in the Life
Cycle Impact Assessment (LCIA). According to ISO standards (ISO 2006), LCIA consists of
both mandatory and optional elements. They can be summarized as:

 Selection of impact categories, category indicators and characterization models that


correspond to the goal and the scope of the LCA
 Assignment of elementary flows in the inventory to the selected impact categories.
This is also known as classification.
 Calculation of category indicator results. This is also known as characterization.
 Normalization, which means the calculation of the magnitude of the category indicator
results that are relative to some reference value.
 Grouping, which means the sorting and, possibly, the ranking of the impact categories
in the order of their importance to your decisions.
 Data quality analysis, which helps in better understanding of the reliability of the
collection of indicator results.

70
Among these steps, the last three, which are normalization, grouping, and data quality analysis
are optional. When selecting impact categories, it is important to understand the difference
between midpoints and endpoints. The main difference lies in the extent of the cause and effect
chain that is considered. For example, consider an ozone depleting chemical such as CFC-11.
When CFC-11 is emitted, it can cause depletion of the stratospheric ozone layer, leading to a
thinning of this layer over certain parts of the globe. When this happens, a larger amount of
ultraviolet light can pass through the ozone layer, which can lead to higher rates of skin cancer
and eye damage in exposed humans, crop damage on exposed land, and degradation of plastics.
All of these specific adverse consequences are considered endpoint impacts, because they are
the ultimate consequence of the emission. Stratospheric ozone depletion is considered to be the
midpoint impact, because it is a common stressor that is caused by the emission that ultimately
leads to one or more endpoint impacts. All the emissions that contribute to an impact category
must be converted to a common unit, which is known as a category indicator; for example, the
CO2 equivalent for Global Warming Potential. Different characterization models may be
employed for the conversion of emissions to a particular category indicator.

According to IPCC (2011), the environmental impacts can be divided into global ones such as
global warming, acidification and the like; and local ones such as soil quality, biodiversity, and
such others. Recent LCA studies use standard LCIA methods such as ReCiPe, EDIP, LIME,
TRACI, and the like (Curran, 2012). These impact assessment methods may be limited to
endpoint or midpoint indicators or might predict both. In LCA, impact categories that are based
on midpoints are more common in characterization models, given that relationships between
emissions and environmental stressors are often well established. In contrast, impact categories
that are based on specific endpoints are less common, given that these endpoints require
extensive observations, risk analyses, and data to relate them to emissions of different
substances; these relationships can also vary with region.

The present thesis has used the ReCiPe impact assessment methodology (Goedkoop et al.,
2009). This methodology predicts eighteen midpoint indicators and three endpoint indicators,
which are based on the LCI data. . In this thesis, impact categories that are based on midpoints
have been given emphasis because relationships between emissions and environmental
stressors are better established for midpoint indicators. In contrast, impact categories that are
based on specific endpoints are less common, given that these endpoints require extensive
observations, risk analyses, and data to relate to them to emissions of different substances;
these relationships can also vary with region. Thus the ReCiPe methodology with the widest

71
range of midpoint indicators became a natural choice. Some advantages of ReCiPe when
compared to other LCIA methods include the use of impact mechanisms with global scope,
wherever possible and non-inclusion of emissions from future extractions in the impact
assessment. The impact categories and the category indicators for the ReCiPe methodology are
shown in Figure 5.3 (Goedkoop et al., 2009).

Figure 5.3: ReCiPe methodology (Goedkoop et al., 2009).

Further, each of these midpoint and endpoint indicators can be predicted on the basis of three
different perspectives. They are:

 Individualist: The focus is on short-term interests with the belief that technology can
avoid many problems in future.
 Hierarchist: This is a consensus model and is often regarded as being the default model.
 Egalitarian: The focus is on long-term interests and is based on precautionary principle
thinking
The assumptions undertaken in these perspectives are presented in Table 5.3 (De Schryver,
2011). This thesis will perform calculations on the basis of the Hierarchist perspective, which
72
is based on common policy principles and average time-frame. Additionally, since the
Hierarchist perspective is widely used, it assists in the comparison of the LCA results with
other LCA studies. The different perspectives are expected to yield different results based on
the underlying assumptions. For example, the GWP considers a time horizon of 20 years, 100
years and 500 years for individualist, hierarchist and egalitarian perspectives, respectively. This
alters the characterisation factors for different chemicals leading to GWP, eg. the charactization
factor for methane is 72, 25 and 7.6 for individualist, hierarchist and egalitarian perspectives,
respectively.

Table 5.3: Overview of different LCA perspectives (De Schryver, 2011).


Individualist Hierarchist Egalitarian
Vision on nature Considers nature robust Considers nature tolerant Considers nature vulnerable
Level of knowledge Only considers certain Considers likely effects Considers all known effects
(proven) effects
Time horizon Emphasizes present and Balanced time perspective Current and future effects
short-term effects are considered equal
Vision on society Economic output is Developments within Equality and social driven
market driven limits of nature
Manageability Adaptive management Preventive and Controlling and limited
style comprehensive management style
management style
The LCA results that were calculated from the MS Excel-based framework were validated
against the results that were generated by the commercial LCA software Simapro. An error of
less than two percent was noticed between the two results. The comparison has been presented
in Appendix G. The MS Excel sheets were used in this study instead of the SimaPro software
because of the ease of inclusion and operation of MS Excel with ASPEN Plus software.
Moreover, the MOO code is also developed in MS Excel. Use of MS Excel provided a coherent
platform for LCC, LCA and MOO calculations. The working methodology for the MS Excel
sheets has been demonstrated in Appendix H.

5.3 Results
The LCA results for the biomass-to-ammonia process were found to be better than fossil fuel-
based processes in most of the midpoint category indicators such as the GWP and the Ozone
Depletion Potential. The GWP sources for the different process configurations are shown in
Figure 5.4. Electricity consumption was found to be the major GHG emitter in all the cases.
GHG emissions from centralized electricity production are 1.06, 1.51, and 0.28 kg CO2

73
eq./kWh for Australia, India, and Brazil, respectively. This explains the strong variation in the
results with respect to electricity production between these countries. Coal-based electricity
production is the major contributor to GHG emissions from electricity production in India. It
is also seen that straw production leads to maximum GHG emissions, followed by wood and
bagasse. The use of fertilizers is a major contributor to GHG emissions from straw production.
These emissions could be further reduced if the final ammonia production is utilized in the
manufacturing of fertilizers. The transport emissions in the case of bagasse feedstocks are not
taken into account under the assumption that the biomass-to-ammonia plant would be located
near the sugarcane processing facility. The consumption of charcoal and Rape Seed Methyl
Ester (RME), which are utilized in syngas cleaning, is included in the analysis.

74
0
0.2
0.6
0.8
1

0.4

0
1
0
1

0.2
0.4
0.6
0.8
0.2
0.4
0.6
0.8
Electricity Production Electricity Production Electricity Production

Bagasse Production Straw Production Transportation


Charcoal Auxillary Transport Eucalyptus cultivation
RME Charcoal Charcoal
Wood ATR

Straw ATR
FICFB Gasification RME RME

Bagasse ATR
Reforming catalyst FICFB Gasification FICFB Gasification
Zinc oxide Reforming catalyst Reforming catalyst 0
1

0.2
0.4
0.6
0.8

0
1

0.2
0.4
0.8
0.6

0
1

0.2
0.4
0.6
0.8
Electricity Production
Electricity Production Electricity Production
Fuel Production (NG)
Fuel Production (NG) Straw Production
Auxillary Transport
Bagasse Production Fuel Production (NG)
Eucalyptus cultivation
Charcoal Transport

75
Wood SMR

Charcoal

Straw SMR
RME Charcoal

Bagasse SMR
RME
FICFB Gasification RME FICFB Gasification
Reforming catalyst FICFB Gasification

0
1

0.2
0.4
0.6
0.8
0
0.2
0.4
0.6
0.8
1

0
0.2
0.4
0.6
0.8
1

Electricity Production Electricity Production Electricity Production


Fuel Production (NG) Fuel Production (NG)
Straw Production
Bagasse Production Auxillary Transport
Fuel Production (NG)
Charcoal Eucalyptus cultivation
Transport
Wood CR

Bagasse CR
Straw CR

RME Charcoal
Charcoal(kg)
FICFB Gasification RME
RME

Figure 5.4: GWP for different configurations of biomass-to-ammonia process.


Reforming catalyst FICFB Gasification
FICFB Gasification
The LCA results for all the category indicators are presented in the Table 5.4. The LCA results
for the biomass-to-ammonia process were compared with the LCA data that was reported for
the production of conventional fossil fuel-based ammonia. Inventories are available in the
ecoinvent database for ammonia production (world average, except Europe) from natural gas
by steam methane reforming (SMR); and ammonia production from heavy naphtha, and coal
by partial oxidation (POX). Table 5.4 shows that the biomass-to-ammonia process is predicted
to have a higher eutrophication potential (freshwater and marine) and ecotoxicity (terrestrial
and freshwater). The terrestrial ecotoxicity, stems mainly from biomass cultivation, whereas
electricity production and biomass cultivation are the major contributors to the freshwater
ecotoxicity. Eutrophication and acidification were the highest for the straw feedstock. The
consumption of coal-based electricity in India led to the higher acidification and eutrophication
potentials in India. The agricultural land occupation potential was significantly higher for
wood-based ammonia production. Eucalyptus production requires separate land for cultivation,
whereas both straw and bagasse are by-products of established agricultural processes. Different
factors contributed to the ozone depletion potential of the three feedstocks. The ozone depletion
potential of the wood feedstock results mainly from transportation and eucalyptus cultivation.
In the case of the straw feedstock, the major contributors to the ozone depletion potential were
straw cultivation, transportation and electricity production. Electricity production and bagasse
production were the main contributors to the ozone depletion potential in the bagasse
feedstock (Figure 5.5).

76
Table 5.4: LCIA of the biomass-to-ammonia process for different process configurations.
Impact category Unit Rest of World Rest of World Wood Wood Wood Straw Straw Straw CR Bagasse Bagasse Bagasse
(SMR) (POX) ATR SMR CR ATR SMR ATR SMR CR
Climate change kg CO2 eq 1.90E+00 3.03E+00 1.02E+00 1.29E+00 1.24E+00 1.57E+00 1.83E+00 1.78E+00 2.80E-01 4.40E-01 4.06E-01

Ozone depletion kg CFC-11 eq 2.22E-07 5.98E-07 2.53E-08 2.22E-08 2.25E-08 4.89E-08 7.38E-08 7.73E-08 1.87E-08 2.47E-08 2.79E-08

Terrestrial acidification kg SO2 eq 8.26E-03 1.90E-02 6.52E-03 5.46E-03 6.22E-03 1.09E-02 1.00E-02 1.15E-02 3.84E-03 3.74E-03 4.15E-03

Freshwater eutrophication kg P eq 1.33E-04 1.85E-04 5.85E-05 4.87E-05 5.54E-05 5.40E-04 4.58E-04 5.41E-04 9.21E-05 8.07E-05 9.39E-05

Marine eutrophication kg N eq 4.00E-04 5.83E-04 1.81E-04 1.56E-04 1.72E-04 6.90E-03 6.05E-03 6.13E-03 5.34E-04 4.55E-04 4.84E-04

Human toxicity kg 1,4-DB eq 4.02E-01 7.15E-01 3.57E-01 3.04E-01 3.30E-01 6.70E-01 6.26E-01 6.95E-01 3.33E-01 2.97E-01 3.12E-01

Photochemical oxidant formation kg NMVOC 3.73E-03 6.16E-03 4.63E-03 4.00E-03 4.39E-03 4.85E-03 4.59E-03 5.22E-03 2.43E-03 2.25E-03 2.40E-03

Particulate matter formation kg PM10 eq 2.44E-03 6.02E-03 1.96E-03 1.66E-03 1.87E-03 4.25E-03 3.85E-03 4.49E-03 1.05E-03 1.01E-03 1.13E-03

Terrestrial ecotoxicity kg 1,4-DB eq 3.02E-04 8.64E-04 5.63E-04 4.92E-04 5.01E-04 3.10E-02 2.72E-02 2.73E-02 4.72E-03 3.91E-03 4.07E-03

Freshwater ecotoxicity kg 1,4-DB eq 1.61E-02 1.29E-02 3.61E-03 3.07E-03 3.38E-03 3.01E-02 2.70E-02 3.07E-02 1.62E-02 1.49E-02 1.79E-02

Marine ecotoxicity kg 1,4-DB eq 1.30E-02 2.03E-02 3.22E-03 2.74E-03 3.04E-03 2.30E-02 2.08E-02 2.41E-02 1.38E-02 1.23E-02 1.49E-02

Ionising radiation kBq U235 eq 7.69E-02 2.59E-01 4.10E-04 3.54E-04 3.78E-04 5.52E-02 5.75E-02 6.39E-02 2.81E-02 2.59E-02 3.14E-02

Agricultural land occupation m2a 1.67E-02 2.43E-02 2.51E+00 2.21E+00 2.22E+00 2.06E-01 1.80E-01 1.84E-01 2.60E-01 2.26E-01 2.38E-01

Urban land occupation m2a 4.09E-03 7.99E-03 5.32E-02 4.48E-02 4.98E-02 1.63E-02 1.93E-02 2.08E-02 2.83E-03 2.51E-03 2.62E-03

Natural land transformation m2 4.77E-04 1.17E-03 2.29E-04 2.01E-04 2.02E-04 1.18E-04 1.69E-04 1.80E-04 1.54E-04 1.57E-04 1.90E-04

Water depletion m3 5.48E-02 5.83E-02 4.61E-03 3.35E-03 3.56E-03 3.61E-02 3.08E-02 3.15E-02 3.52E-02 2.95E-02 3.40E-02

Metal depletion kg Fe eq 8.19E-02 7.38E-02 5.85E-02 5.40E-02 5.45E-02 6.11E-02 5.64E-02 6.27E-02 4.64E-02 3.93E-02 4.58E-02

Fossil depletion kg oil eq 8.38E-01 1.15E+00 2.06E-01 2.77E-01 2.69E-01 3.05E-01 3.68E-01 4.11E-01 4.03E-02 9.48E-02 1.05E-01

77
Wood ATR Straw ATR Bagasse ATR
2.5E-08 2.5E-08 2.5E-08

2E-08 2E-08 2E-08

1.5E-08 1.5E-08 1.5E-08

1E-08 1E-08 1E-08

5E-09 5E-09 5E-09

0 0 0
RME
Rail Transport
Road Transport

Charcoal(kg)
Eucalyptus cultivation

RME

Charcoal

RME
Straw cultivation

Charcoal(kg)
Water

Water

Sodium Hydroxide
Road Transportation

Bagasse production
Electricity Production

Electricity Production

Electricity Production
Figure 5.5: Ozone depletion potential (kg CFC-11 eq) for different configurations of
biomass-to-ammonia process.

The incineration of fly ash from the gasifier is a major contributor to human toxicity. The
incineration of fly ash is a common practice in Europe; however, its application may vary with
location. The majority of photochemical oxidant formation, particulate matter formation, and
ionizing radiation resulted from electricity consumption, biomass cultivation, and
transportation. The water depletion potential of bagasse (which comes from the cultivation of
sugarcane which has a high water requirement) is similar to other biomass feedstocks namely
wood and straw. This can be understood by the allocation of water requirement among various
products of sugarcane cultivation. Bagasse as a waste material would not normally have the
water allocated to it. Thus, a mere 1.5 % of the total inputs to the sugarcane cultivation are
attributed to bagasse production. The midpoint indicators can be combined into a single score
endpoint indicator by normalizing all the impact categories. Single score endpoint results for
different process variants that were normalized using world average emissions are shown in
Figure 5.6.

78
3.5
Single score indicator (mpt)
3
2.5
2
1.5
1
0.5
0
Rest of Rest of Wood Wood Wood Straw Straw Straw Bagasse Bagasse Bagasse
World World ATR SMR CR ATR SMR CR ATR SMR CR
(SMR) (POX)

Figure 5.6: Single score endpoint indicator for biomass-to-ammonia process.

These results highlight the potential of LCA in understanding the sustainability aspects of
renewable energy systems. The single score endpoint indicators confirm the environmental
superiority of the biomass-to-ammonia production when compared to conventional fossil fuel
based ammonia production. The single score indicator for the straw feedstock is largely
attributed to electricity consumption. The results for the GWP and the ozone depletion
potential underline the variation experienced in LCA results, when the feedstock and the
process configurations are changed. The electricity consumption from the grid was found to be
a major contributor to many impact categories. The key to a sustainable biomass-to-ammonia
process is the reduction in the import of electricity from the grid. This reduction is possible if
a part of the syngas from biomass gasification is used for in-house electricity production.
However, this would result in a lower yield of ammonia and higher cost of production of
ammonia. This trade-off is presented in detail in the following chapters. Additionally, studying
the results for a single flowsheet configuration for a process variant might be misleading. To
overcome this, a MOO of the different process variants is performed in the next chapters.

5.4 Conclusions
A cradle-to-gate LCA for the different variants of the biomass-to-ammonia process has been
performed. The use of bagasse feedstock in Brazil was found to have the lowest carbon
footprint. This is attributed to renewables-based electricity production and the absence of
transportation for the bagasse feedstock in Brazil. GHG-intensive electricity production and
straw production in India make the GWP of the straw-to-biomass process similar to the GWP
of the fossil fuel based ammonia production. The GWP of the wood-to-ammonia process was

79
mid-way between straw and bagasse. The single score endpoint indicators confirm the
reduction in emissions that result from the biomass-to-ammonia process when compared to
conventional fossil-fuel based ammonia production.

80
Chapter 6: Multi-Objective Optimization

6.1 Introduction
The literature review pointed toward an absence of MOO studies in the domain of ammonia
production from biomass. The chapter focuses upon MOO of different variants of biomass-to-
ammonia process. The focus of the MOO was to minimize the ammonia production costs as
well as the environmental emissions that are associated with biomass-based ammonia
production. Different biomasses feedstocks and flowsheet configurations are expected to
behave differently in terms of emissions and plant economics. Such an analysis will provide
the decision makers with consistent and comparable data to objectively judge the viability of
the proposed biomass-to-ammonia process.

6.2 Methodology
Process design often involves optimization of two opposing objectives. In many of these cases,
improvement in one objective leads to a decline in others and vice versa. A similar situation
arises when the economics and environmental impacts of the ammonia process are optimized
simultaneously. Optimization of such systems has typically been performed by assigning
certain weights to different objectives and optimizing the combined single objective by using
a derivate-based optimization method. However, such an approach would need several runs of
the single objective optimization problem to be solved in order to obtain a set of Pareto-optimal
solutions. An alternate approach, Multi-Objective Optimization (MOO), has the ability to yield
a range of Pareto-optimal solutions, in a single run. Since, such an approach does not require
derivative information, it is expected to converge eventually to a global optimum, if evolved
for a substantial number of generations.

The Non-dominated Sorting Genetic Algorithm (NSGA-II), which was developed by Deb et al.
(2002), has been used and linked to the ASPEN Plus simulation software and the LCC/LCA
worksheets, to optimize economic and environmental objectives. For this purpose the Excel-
based MOO framework that was developed by Sharma et al. (2012) has been modified to
incorporate heat integration, economic, and environmental assessment.

The search starts with random generation of a set of decision variables (parents) by an Excel-
based VB interface, after which the set is sequentially supplied to the ASPEN flowsheet for
evaluation. All the parameter values that affect the economic and environmental performance

81
of the processes are then extracted. Based on the extracted values, the ammonia production
costs and the LCA results are calculated. The genetic algorithm then makes use of this
information to generate the next set of potential solutions (offspring) that are to be sequentially
evaluated by the ASPEN simulation. The best individuals among the combined parent and
offspring population are then selected to produce the next set of offspring. This procedure
continues until a termination criterion (for example, maximum number of generations) is met.
The methodology is depicted in Figure 6.1.

Figure 6.1: MOO methodology.

The ASPEN Plus flowsheets that do not converge due to infeasible input are neglected by the
code. The ASPEN Plus simulations that converge with the status “Success” or “Warning” are
selected for creation of future solutions. If constraints are also added, the solution that meets
the constraints is given preference over the solutions that defy the constraints. In general, the
solutions are selected on the basis of their non-domination score, that is, the proximity of the
solutions to the Pareto front. In the case of solutions with a similar non-domination score, the
solutions are sorted on the basis of their crowding distance value. The crowding distance value
represents the density of solutions in the neighborhood of one solution and helps introduce
diversity in the non-dominated solutions. Figure 6.2 depicts the improvement in the MOO
solutions with increasing generations, with little change for last ten generations.

82
1600

Ammonia Production Cost ($/t NH3) 1550

1500

1450

1400 1st Generation


10th Generation
1350 20th Generation
30th Generation
1300
0.8 1 1.2 1.4 1.6
Global Warming Potential (t CO2 eq./t NH3)

Figure 6.2: Evolution of Pareto front with progress in generations.

6.3 Multi-Objective Optimization


MOO was carried out for the three different biomass feedstocks, namely, wood, straw, and
bagasse. The objectives of MOO were to minimize the GWP of the biomass-to-ammonia
process, and to minimize the cost of ammonia production. The decision variables and the
constraints for the MOO are given in Table 6.1 and Table 6.2, respectively. A population size
of 40 was simulated for 50 generations. A crossover probability of 0.8 and a mutation
probability of 0.05 were selected as MOO parameters, which were used by the genetic
algorithm NSGA-II.

Table 6.1: Decision variables and their range for biomass-to- ammonia process optimization.
Decision Variable Range(Biomass)
Bed material flowrate (𝑘𝑔/𝑘𝑔 𝑜𝑓 𝑏𝑖𝑜𝑚𝑎𝑠𝑠) 23–50
Steam-to-biomass ratio in the gasifier (𝑘𝑔/𝑘𝑔 𝑜𝑓 𝑏𝑖𝑜𝑚𝑎𝑠𝑠) 0.4–1
Percentage of char that goes to the gasifier (%) 5–20
Gas conditioning pressure (𝑏𝑎𝑟) 20–40
Steam flowrate to the shift reactor (𝑘𝑔/ℎ𝑟) 5000–10000
Ammonia purge fraction 0.026-0.100
Air to the ATR (𝑘𝑔/ℎ𝑟) 2400–5000
Reformer temperature for SMR (°C) 800–1000
Flue gas recycling for the CR (%) 0–50
Syngas percentage for electricity co-production (%) 0–30

83
Table 6.2: Constraints for biomass to ammonia optimization.
Variable ≥ Or ≤ Value
Ammonia production (kg/hr) ≥ 1800
Reformer temperature (ºC) ≥ 800
Reformer temperature (ºC) ≤ 1100
Gasifier temperature (ºC) ≥ 800
Combustor temperature (ºC) ≥ 20 + Gasifier temperature

The bed material flowrate influences the temperature of the gasifier and the combustor and,
hence, the syngas compositions. The steam-to-biomass ratio dictates the gas composition by
shifting the equilibrium composition of the water-gas shift reaction. The percentage of char
that goes to the gasifier would influence the carbon division between the gasifier and the
combustion chamber, and the temperature profile of the gasifier as well. The gas conditioning
pressure plays an important role in determining the equipment cost and the electricity that is
required for the compression of the syngas. The moisture content in the syngas is controlled by
the flowrate of the steam to the shift reactor. The ammonia purge fraction controls the mass
flowrate (hence, the compression energy) in the ammonia synthesis loop. It also controls the
amount of syngas that is recycled back to the gasifier. The air that goes to the ATR (in the ATR
configuration) performs a dual role of providing oxygen for partial oxidation of syngas and
providing a nitrogen source for ammonia production. The reformer temperature (in the SMR
configuration) influences the electricity production through the cooling of the syngas. The flue
gas recycle (in the CR configuration) provides heat for reforming as well as for nitrogen
requirements. It also influences the load on the PSA. In the previous chapter, it was shown that
the biomass-to-ammonia process requires an external source of electricity. The electricity
consumption has a major impact on the LCA and on the cost of ammonia production. The
import of electricity from the grid can be reduced by combusting a percentage of syngas for
the production of electricity through the steam cycles. Thus, electricity can be produced, but at
the cost of a reduction in the output of ammonia.

The operational range for the different decision variables correspond to the normal operational
range that is experienced in actual plant operations. A constraint was put on minimum
ammonia production to provide a bound for the trade-off between ammonia production and
electricity generation. In the absence of this bound the Pareto front would also include results
where majority of syngas is being utilized for electricity production alone, whereas this thesis
is aimed at ammonia production. The temperature constraints for the reformer and the gasifier

84
are dictated by the reaction kinetics for the lower temperature bound and material of
construction for the upper temperature bound. The gasifier and the combuster temperature are
also constrained to maintain sufficient temperature difference for a favourable heat transfer.

6.4 Results
The results of the MOO are illustrated in Figure 6.3 and Figure 6.4. The wood, straw, and
bagasse exhibited different behaviors for their respective Pareto plots. The straw feedstock was
found to be the most economical, but had the least significant savings in terms of GWP. The
lower capital cost added to lower labor and feedstock costs, which led to a significant reduction
in the straw-based ammonia plant. A high GWP results from the coal-based electricity
production in India and from the GHG emissions that are related to the production of straw
pellets. The bagasse feedstock benefits from the hydro-electric power and biomass-based
electricity production in Brazil, which has a very low carbon footprint. A lower syngas yield
from bagasse leads to a low ammonia output and an increase in the cost of the production of
ammonia. Additionally, Brazil has the highest location factor, which increases the capital cost.
Eucalyptus wood does not have the positive attributes that both straw and bagasse cases have.
It suffers from both high production costs as well as carbon-intensive electricity production.
The difference in the slopes of the Pareto plots for the three different feedstocks can be
attributed to the GHG emissions that are linked to the centralized electricity production and to
the contribution of the cost of this electricity to the cost of ammonia production. Any increase
in the production of electricity would lead to a proportional decrease in the production of
ammonia. The decrease in the output of ammonia would increase the cost of the production of
ammonia. The effect is the maximum in the case of eucalyptus, followed by bagasse and straw.

The ATR process configuration has the lowest ammonia yield, because a part of the syngas is
used to heat the reformers. An external fuel is used to heat the reformers in the SMR and CR
configurations, which results in higher GWP in these two process configurations. It is observed
that although the output from the SMR plant is higher than that from ATR plant, a higher total
production cost of the SMR plant normalizes the final ammonia production cost per unit of
ammonia. Thus, it is seen that both the ATR and SMR configurations have similar ammonia
unit production costs. The CR plant is the most economic for all three feedstocks due to the
absence of the ASU, which is an expensive equipment item. This phenomenon has been
illustrated with the help of Figure 6.3 and Figure 6.4.

85
Figure 6.3: The Pareto plots that are obtained for the biomass-to-ammonia process, along
with the variation in ammonia production

Figure 6.4: The Pareto plots that are obtained for the biomass-to-ammonia process along with
variation in the import of electricity.
86
Figure 6.3 and Figure 6.4 point at a trade-off between electricity production and ammonia
production. The Pareto plots are dictated by two decision variables, namely, the percentage of
syngas used for electricity production and the gas conditioning pressure. The linear trend of
Pareto results may be attributed to the increase in the production of electricity and a
simultaneous decrease in the production of ammonia. The lower region of the Pareto plots is
dictated by the variation in the gas conditioning pressure. The biomass-to-ammonia process
consumes the lowest amount of electricity (and results in the lowest emission) at the highest
pressure of 40 bar. At a pressure of 40 bar, the combined electricity consumption of the two
compressors is the lowest. Additionally, the highest pressure is limited by the steam
regeneration loop, which is operated by using the heat from the isothermal shift that takes place
at 260 °C. Lowering the gas conditioning pressure leads to a slight decrease in the cost of
ammonia production, which occurs due to the reduction in the pressure factor of the
equipments. However, this is accompanied by an increase in GWP, which is a result of the
higher electricity consumption that sustains the higher compression load. All other decision
variables converge to the minimum possible values. A low bed material circulation rate helps
in increasing the combustor temperature and the flue gas temperature. A high flue gas
temperature provides the scope for the generation of a greater amount of electricity. Similarly,
a lower steam-to-biomass ratio and a low percentage of char that goes to the gasifier helps in
maintaining a higher temperature in the system. A low purge from the ammonia loop helps to
reduce the recycle volumes and, hence, the equipment sizes. A low air-to-ATR ratio (in the
ATR configuration) results in lower compression costs and lower equipment costs for the PSA.
The lowest air flowrate is dictated by the minimum nitrogen flowrate that is needed to maintain
the ratio of 3:1 for H2:N2 in the ammonia synthesis loop. The gasifier flue gas recycle to the
reformer in the CR configuration stabilized at around 30 percent of the total flue gas that was
produced by the gasifier. This percentage fulfilled the minimum nitrogen requirement and had
the lowest compression costs.

6.5 Conclusions
Three biomass feedstocks, namely, wood, straw pellets, and bagasse which are available in
Australia, India, and Brazil, respectively, have been compared in order to understand the effects
of the biomass composition and the national economic and environmental scenarios on the
viability of the biomass-to-ammonia process. LCC was found to be the lowest in the Indian
scenario, followed by Brazil and Australia. Lower capital costs and operating costs in India

87
made the straw feedstock economically favorable. However, the GHG-intensive electricity
production in India makes the GWP of the Indian straw-to-biomass process similar to the GWP
of the fossil fuel based ammonia production. The use of bagasse feedstock in Brazil was found
to have the lowest GWP. This is attributed to renewables-based electricity production and the
absence of transportation for the bagasse feedstock in Brazil. Wood-based ammonia production
was found to be the most expensive due to the high operating costs in Australia. The GWP of
the wood-to-ammonia process was mid-way between straw and bagasse. The MOO results
show that the GWP of the biomass-to-ammonia process may be reduced if some amount of the
syngas that is produced from gasification is used in the production of electricity. This would
reduce the GHG emissions that are related to the import of electricity. However, this would
lead to a reduced output of ammonia and an increase in its cost. The trade-off between ammonia
production and electricity production varied for the three feedstocks on the basis of the break-
up of both LCC and LCA.

88
Chapter 7: Discussion

7.1 Introduction
This chapter presents a critical discussion of the process modeling approach that has been
adopted in this thesis. The chapter begins by highlighting the advantages that are derived from
a multi-scale approach, which was used to model the biomass-to-ammonia process. Thereafter,
a sensitivity analysis of the process economics is presented. An analysis of the LCA
methodology along with the possible improvements is also presented. The advantages offered
by the MOO are then discussed. The chapter ends with a comparison of the LCA results for the
biomass-to-ammonia process with fossil fuel-based ammonia production and, thereby
establishing the advantages that are offered by different biomass-to-ammonia process variants.

7.2 Multi-scale approach


The thesis illustrates the application of multi-scale modeling in sustainable chemical
production. CFD software ANSYS Fluent is used to simulate a fluidized bed biomass gasifier.
This simulation is used to procure an RTD curve, which in turn can be used to propose a
compartment model for the otherwise complex biomass gasifier. This simplified compartment
model can be easily integrated with the simulation software, ASPEN Plus. ASPEN Plus is used
to simulate the biomass gasification process as well as downstream gas conditioning that leads
to ammonia production. The output of ASPEN Plus simulation is fed into an MS Excel-based
framework, which performs LCC and LCA for the simulated flowsheet. Finally, these outputs
are used by an MS Excel-/VB-based Multi-Objective Optimization (MOO) framework in order
to optimize both the LCC and the LCA for changing flowsheet configurations.

There might be some apprehension about taking such a detailed approach. The final results of
the thesis in the form of the MOO of different process variants do not reflect any contribution
that result from the rigorous gasifier model. The model, nonetheless, facilitated the prediction
of the composition of the syngas from different biomass feedstocks. However, these results
could have been achieved using a simpler model. A much greater contribution of this model is
its ability to predict tars. Tar concentration should be less than 1 ppmV in order to prevent
coking/deactivation of the catalyst during the production of chemicals from gasification-
derived syngas (Asadullah, 2014). Thus, prediction of tar production and its subsequent
removal is an important aspect of any biomass gasification based chemical production process.
The accuracy of tar predictions, however, would not have any significant effect on the LCC

89
and LCA of the biomass-to-ammonia process. Nonetheless, accurate analysis of the tar
handling system is critical for a successful physical realization of the biomass-to-ammonia
process.

A compartment-based model for the gasifier also facilitated efficient heat integration. The flue
gas from the combustor, which is normally used for biomass drying, can be utilized for
electricity production using steam turbines. The let-down steam from the turbine is instead used
for biomass drying. The same steam cycle is also used to produce the steam that is required by
the gasification process. The air preheating for the PSA off-gas furnace and the ATR reactor is
also performed using the excess heat that is available in the biomass-to-ammonia process.

A higher steam-to-biomass ratio is expected to increase the hydrogen concentration in the


syngas as well as the total yield of syngas. Thus, a higher steam-to-biomass ratio would be
desirable in a biomass-to-ammonia process because such a process is expected to yield a higher
ammonia output. However, the MOO analysis predicted that a lower steam-to-biomass ratio
would be beneficial to the biomass-to-ammonia process. It is understood that the advantage
that may be derived from a syngas of a better quality was offset by effective heat integration
of the biomass-to-ammonia process. The utilization of the steam that is produced by the
biomass-to-ammonia process in electricity production is predicted to yield better results in
terms of both economics and environmental performance. These results highlight the benefits
of heat integration, which is an outcome of a rigorous ASPEN plus gasifier model.

Additionally, the ASPEN Plus model can also be utilized for possible implementation of the
Adsorption Enhanced Reforming (AER). In AER, a CO2-adsorbing bed material such as
CaO/CaCO3 is used. The CO2 is transferred from the gasifier to the combustor through periodic
carbonation of CaO and calcination of CaCO3. The difference between the partial pressure of
CO2 in the reactor and the equilibrium partial pressure of CO2 drives the reactions. The removal
of CO2 inside the gasifier reduces the CO2 and CO contents downstream, and the gas obtained
in this manner is rich in H2. The use of such bed materials has experimentally yielded a syngas
with an H2 content that was as high as 75 percent (Göransson et al., 2011). The ASPEN Plus
model may also be utilized for the study of gasification in the presence of steam and air. The
air content may be adjusted to fulfill the nitrogen requirement. Such an approach can help
eliminate the need for reforming and for the ASU. However, the modeling of these techniques
would have to be validated with experiments.

90
7.3 Process economics
One of the objectives of this thesis is the evaluation of small-scale ammonia production through
the incorporation of the ICI process and the use of dual fluidized bed gasification. This
innovative combination of new and old technologies at the front end and tail end of the process
can help in the commercialization of the biomass-to-ammonia concept, which would otherwise
be constrained by the inadequate supply of biomass feedstock. Additionally, a conservative
approach has been taken with many of the parameters that were used to determine the
economics. A sensitivity analysis for the production cost of ammonia was carried out and it
was found that the biomass price and discount rate had a major impact on the ammonia
production costs. The sensitivity analysis results are presented in Figure 7.1 and Figure 7.2.

1700
Wood ATR

Percentage change in ammonia production cost


1600 18
Ammonia Production Cost ($/t NH3)

1500 8

w.r.t. base-case (%)


1400
-2
1300

1200 -12

1100
Biomass price $ 50/odt -22
1000 Biomass price $ 100/odt
-32
900 Biomass price $ 150/odt

800 -42
0 5 10 15 20 25
Discount rate (%)

Figure 7.1: Sensitivity analysis for biomass feedstock price for the Wood ATR configuration.

91
900 26
Straw ATR

Percentage change in ammonia production cost


850
16
Ammonia Production Cost ($/t NH3) 800

750 6

w.r.t. base-case (%)


700
-4
650
-14
600

550 -24
Biomass price $ 50/odt
500
Biomass price $ 100/odt
-34
450 Biomass price $ 150/odt
400 -44
0 5 10 15 20 25
Discount rate (%)

Figure 7.2: Sensitivity analysis for biomass feedstock price for the Straw ATR configuration.

The results predict that a variation of 1 percent in the discount rate alters the ammonia
production cost by approximately $ 25.19/t NH3 and $ 17.25/t NH3 in the Wood ATR and the
Straw ATR process configurations. A discount rate of 5 percent (biomass price being $ 100/odt)
can reduce the ammonia production cost by 17.5 percent and 23 percent in the Wood ATR and
Straw ATR configurations, respectively. This result may be attributed to the higher capital cost
that is experienced in the case of wood. Additionally, at higher discount rates larger future cash
flows are required to provide a positive net present value. This leads to an increase in the
ammonia production cost. A higher discount rate is usually chosen for new processes to
compensate investors for greater financial risk. Similarly, a variation of $1 in the price of
biomass led to a variation of approximately $2.27/t NH3 in the Wood ATR system and
approximately $0.97/t NH3 in the Straw ATR system. A biomass feedstock price of $50/odt
(discount rate being 15 percent) can reduce the ammonia production price by 8.3 percent and
6.8 percent for the Wood ATR and the Straw ATR configurations, respectively. This difference
in sensitivity results from different feedstock prices for wood and straw. A lower biomass
feedstock price as well as a lower discount rate can further assist in making the biomass-to-
ammonia process economical.

A conservative scale for the biomass-to-ammonia process was chosen to assist the applicability
of the process concept globally. Another rationale behind selecting this scale of 120 odt/d of

92
biomass intake for the biomass-to-ammonia plant was that any small-scale ammonia
production facility would, generally, be larger than this scale and would benefit from the
economies of scale. Thus, this study would form a limiting case. If the scale is increased to 600
odt/d of biomass intake, the ammonia production cost (Wood ATR process) would reduce from
$1370/t NH3 to $1080/t NH3. Such a process would require five dual bed gasifiers operating in
parallel. Replacing the dual bed gasifiers with entrained flow gasifiers is also expected to
improve the economical performance of the biomass-to-ammonia process at larger scales.
Another advantage that may be derived at larger scales would be a better electricity generation
system by employing gas turbines. However, the application of a large-scale biomass-to-
ammonia facility would be limited by the availability of the biomass feedstock.

7.4 Life cycle assessment


A cradle-to-gate LCA for the different variants of the biomass-to-ammonia process has been
performed in this thesis. Variations that arise from data uncertainties, the incompleteness of
impacts, and the methodological choices sometimes make it difficult to draw general
conclusions from the LCA. In this thesis, impact categories that are based on midpoints have
been given emphasis because relationships between emissions and environmental stressors are
better established for midpoint indicators. In contrast, impact categories that are based on
specific endpoints are less common, given that these endpoints require extensive observations,
risk analyses, and data to relate to them to emissions of different substances; these relationships
can also vary with region.

This thesis relies upon ecoinvent databases for the majority of LCI data. The assumptions
adopted in this thesis are consistent with the assumptions adopted by the ecoinvent database.
Since the ecoinvent database primarily focuses on the European conditions, some of the
assumptions in the LCA might not translate to real impacts. One example is the assumption
that fly ash that is generated in the gasification process will be disposed of by municipal
incineration, which is based on the ecoinvent database. This incineration of fly-ash is
responsible for ~50 percent of the emissions that lead to human toxicity. This practice of
incineration of fly ash was not found to be common in the geographical locations that are
considered in this thesis. The human toxicity potential can, possibly, be reduced to half if the
fly-ash is disposed of in landfills or recycled for construction applications. Even if the fly ash
is incinerated, the ecoinvent database would appear to treat all fly ash in a similar manner, such

93
that fly ash that results from coal combustion is considered to be identical to fly ash from
biomass gasification.

Another important aspect of the LCA is the selection of inventories. This thesis utilizes
ecoinvent inventories that are normalized for Australian conditions, in the wood-to-ammonia
scenario. For the straw-to-ammonia scenario, only the electricity production inventory was
available for India. World average inventories were utilized for all the other process inputs and
outputs. In the bagasse-to-ammonia scenario, inventories were available for bagasse production
and electricity production in Brazilian conditions. The remaining process inputs and outputs
were modeled using world average inventories (Appendix F). Since, electricity was found to
be the chief contributor to the GWP and to the single score endpoint indicator, the assumptions
mentioned above are expected to yield reliable results in the comparison of ammonia
production from three different feedstocks. This novel comparison of three different biomass
feedstocks that are grown at different locations leading to the same output differentiates this
thesis from the previous analysis of chemical production from biomass through the thermo-
chemical pathway. The scope of available LCI data for different world regions is expected to
increase in the future. This enhancement of LCI data would further assist in a rigorous
comparison on the basis of LCA. Thus, this thesis may be used as an example for a thorough
analysis of any biomass-based system. Different biomass feedstocks, process configurations,
and geographical regions have their own advantages.

7.5 Multi-objective optimization


In order to gain a better understanding of the trade-off that exists between economics and
environmental emissions of the biomass-to-ammonia process, an MS Excel-/VB-based MOO
framework was utilized. The MOO framework resulted in a Pareto plot that had the global
optimum for LCC and LCA for a given set of decision variables and constraints. These results
provide a range of process parameters that can help minimize the economic and environmental
loads in the process. Additionally, these economic and environmental hotspots may be
enhanced or replaced in alternate flowsheet configurations in order to obtain better results. This
analysis distinguishes these results from past studies that presented their LCC and LCA results
for a single process configuration of the biomass-to-ammonia process.

The GWP for the biomass-to-ammonia process from the MOO analysis was compared with the
GWP that was reported for the production of conventional fossil fuel-based ammonia.

94
Inventories are available in the ecoinvent database for ammonia production from natural gas
by steam methane reforming (SMR), and ammonia production from heavy naphtha and coal
via partial oxidation (POX). These inventories are available for Europe (RER), World average
excluding Europe (ROW) and Australian normalized. Inventories for ammonia production by
the steam methane reforming for European conditions are also available in the Agri-footprint
database (Blonk Consultants, 2014) and an industrial database (Boustead, 2005). The results
are presented in Figure 7.3.

3.50
Global Warming Potential (kg CO2 eq./ kg NH3)

3.00

2.50

2.00

1.50

1.00

0.50

0.00

Figure 7.3: GWP comparison of biomass-to-ammonia process with conventional ammonia


production.

The results predict a reduction in GWP if biomass is used as a feedstock for ammonia
production. Bagasse is expected to have the lowest GWP, followed by wood and straw. The
GWP of straw-based ammonia production is predicted to be similar to the GWP of natural gas-
based ammonia production, when there is no electricity cogeneration. Thus, the cogeneration
of electricity is advisable in the case of straw. The use of bagasse feedstock in India may lead
to a slight reduction in both economics (due to a lower feedstock price) and environmental
emissions (due to the absence of transportation and lower GHG emissions that are related to

95
bagasse production). However, since the majority of the emissions in the Indian scenario stem
from electricity consumption, the GHG emissions that are related to the biomass-to-ammmonia
process would remain significantly high. The ATR configuration is expected to have the least
environmental impact in all three feedstocks, followed by the CR configuration, and the SMR
configuration.

The individual MOO results for different process configurations of the biomass-to-ammonia
process have facilitated the predicting of a global MOO solution (Figure 7.4). The global MOO
highlights the biomass feedstock (and geography) and process configurations that are favorable
to the achieving of the desired process performance, in terms of economics and emissions.
Ammonia is a bulk chemical and is used in the production of many other chemicals. The Pareto
results shown in Figure 7.4 can be effectively used to reduce emissions that are related to any
ammonia-based chemical production. However, this global Pareto can be further improved
with the inclusion of additional biomass feedstock.

Global Pareto for


biomass-to-ammonia
process

Figure 7.4: Global Pareto for the biomass-to-ammonia process.

It is concluded that the feedstock as well as the geographical location under consideration
would have a substantial effect on the predicted economics and emissions reductions of any
proposed biomass-based process. Changing the feedstock alone might not be good enough for

96
a sustainable process. It is predicted that a sizable amount of emissions can result from
background processes such as electricity production. Thus the methodology proposed in this
thesis provides a framework for feasibility analysis of novel biomass-based processes.
Different biomass feedstocks, process configurations, and geographical regions have their own
advantages.

7.6 Future Work


An enhancement to the proposed biomass-to-ammonia process would include a more rigorous
modeling of the complex PSA process. A more comprehensive PSA simulation is expected to
yield a better estimation of the PSA equipment cost that may lead to more favorable economics.
Additionally, the H2 recovery from the PSA has a major influence on the operating cost and
GWP of the biomass-to-ammonia. A higher H2 recovery would lead to a higher ammonia
production, and a lower ammonia price. However, a lower H2 recovery would assist in
additional electricity production by increasing the calorific value of the PSA blowdown which
is used as a fuel gas. The resultant decrease in electricity consumption from the grid reduces
the CO2 emissions associated with the electricity production. However, this excess electricity
production comes at the expense of H2 which could have been used to produce more ammonia.
A sensitivity analysis of changing process economics and GWP with changing PSA recovery
for the Wood ATR scenario is reported in Figure 7.5.

Figure 7.5: Sensitivity analysis for PSA H2 recovery.

97
It is also interesting to observe the impact of Carbon Capture and Storage (CCS) on the
biomass-to-ammonia process. It was assumed that from the PSA, an additional stream of CO2
(with purity greater than 95 percent) could be generated at no additional cost. This CO2 stream
is then compressed to 100 bar pressure. A sensitivity analysis was performed to understand the
effect of amount of CO2 sequestrated from the PSA off-gas on the economical and
environmental effects of the biomass-to-ammonia process (Figure 7.6). An incremental cost of
~$ 35/t NH3 was observed for the Wood ATR case. A higher CO2 sequestration would increase
both the capital cost (compressor) as well as the operation cost (electricity consumption).
However as predicted previously by Tock et al. (2014), the capture of this biogenic CO2 can
make the biomass-to-ammonia process carbon negative.

Figure 7.6: Sensitivity analysis for CCS of PSA off-gas.

However, the CCS results also require a better modeling of the PSA process, for a more
accurate prediction of the incremental costs for CO2 sequestration. Given the complexity
involved in the PSA process, a rigorous modeling of the PSA process is beyond the present
scope. Nevertheless, these results bring to light the potential CO2 emissions reductions that
could be achieved from a biomass-to-ammonia process, with CCS.

98
Chapter 8: Conclusions
Producers of industrial chemicals are becoming increasingly aware that it is their responsibility
to develop methods that enable their products to have a low negative impact on the
environment. With the maturing of technology, numerous modeling and simulation platforms
are available for the improvement of old models as well as for the conception of new ones.
However, the application of these modeling and simulation tools has often remained limited to
certain process aspects, which undermines the potential of these tools. In this study, a
systematic framework for conception, design, and optimization of greener alternatives for
chemical production through the use of multiple simulation platforms is demonstrated. This
study focuses on ammonia production, which sustains the global agricultural and mining
activities. The majority of ammonia produced around the globe, relies on fossil fuels (mainly
natural gas) as the primary feedstock and also contributes to greenhouse gas emissions. The
present study assesses the viability of small-scale ammonia production from biomass, which is
a renewable energy source. A scale of 70 tpd of ammonia production was selected to buffer the
sporadic biomass supply, while the local demand for ammonia was simultaneously
accommodated. The scale is much smaller than conventional ammonia plants, and the effect of
the economies of scale was expected to make the plant less economical. In order to overcome
these effects, the design incorporated features from the ICI Leading Concept Ammonia®
process. This process incorporated significant changes in terms of heat integration and process
flow in order to suit small-scale ammonia plants.

The use of the dual fluidized bed gasifier technology also contributed to a cleaner syngas and
easier energy integration. In order to predict and optimize the operation of the dual fluidized
bed gasifier effectively, a compartment-based reactor modeling approach was used to model
the gasifier. This model served as a trade-off between accuracy of results and modeling
complexity involved. Instead of utilizing the conventional two-phase model, a CFD model for
the gasifier was developed in ANSYS Fluent. This helped to account for the gas-mixing pattern
in the freeboard (splash zone), which is neglected by the two-phase models. The CFD model
for a fluidized bed was used to carry out RTD studies by using four different tracer approaches.
For the formulation of a compartment model that was based on RTD, the results from a fully
transient simulation were found to be most appropriate. The other approaches failed to account
for the hydrodynamics that is linked to the tracer flow. The RTD was used to predict a
compartment model in MATLAB software. It is believed that, this is the first study of its kind
to perform an RTD analysis of a gasifier by using a CFD model. The compartment model

99
coupled with experimentally measured pyrolysis yields of biomass was used to simulate the
gasifier model in ASPEN Plus. The advantage of this methodology was that it facilitated the
modeling of gasification of different biomass feedstocks. The model results were validated for
woody feedstock. Based on this validation, predictions were made for gasification of other
biomass feedstocks such as straw pellets and bagasse.

The gasifier model and the downstream gas conditioning that led to ammonia production were
modeled in the ASPEN Plus simulation software. Apart from major constituent gases, the
ASPEN Plus model also accounted for tar production in the gasifier process as well as its
subsequent removal. The representation of the tars by four typical compounds adds to the
reliability of this study. Typically the other simulation studies focussing upon biomass-based
chemicals tend to neglect these compounds. The ASPEN Plus flowsheet downstream of the
gasifier also incorporates kinetics-based modeling of the reactors. The simulation takes into
account utilities such as the storage and refrigeration of ammonia as well as the electricity
generation cycle from the excess heat that is available in the biomass-to-ammonia process.
Moreover, different flowsheet configurations have been proposed to study the viability of
different biomass-to-ammonia production routes.

The output of ASPEN Plus simulations was fed into an MS Excel-based framework, which
performs LCC and LCA for the simulated flowsheet. The LCC yields the breakdown of the
capital and the operating costs. The LCC highlighted the factors that play a dominant role in
fixing the cost of ammonia production. Additionally, the costs were also found to be sensitive
to the chosen discount rate and the biomass feedstock price. The LCA on the other hand helped
to identify environmental hotspots for different environmental impact categories. The
consumption of electricity was a major contributor to the global warming potential, and the
transportation of biomass was found to be a major emitter to the ozone depletion potential.
When compared to conventional fossil fuel-based ammonia production, considerable emission
reductions are achieved by the biomass-to-ammonia process.

In order to gain a better understanding of the trade-off that exists between the economics and
the environmental emissions of the biomass-to-ammonia process, an MS Excel-/VB-based
MOO framework was utilized. The MOO framework results in a Pareto plot that has the global
optimum for LCC and LCA. These results provide a range of process parameters, which can
help minimize the economic and environmental loads in the process. Additionally, these
economic and environmental hotspots may be enhanced or replaced in alternate flowsheet

100
configurations in order to obtain better results. This analysis distinguishes this study from past
studies, which presented their LCC and LCA results for a single process configuration of the
biomass-to-ammonia process. Three biomass feedstocks, namely, wood, straw pellets, and
bagasse, which are available in Australia, India, and Brazil, respectively, have been compared
in order to understand the effect of biomass composition, the supply chain, and the national
economic and environmental scenario on the viability of the biomass-to-ammonia process. The
MOO results that were derived for changing flowsheet configurations as well as for different
biomass feedstocks grown at different locations worldwide are expected to highlight a more
rigorous analysis of the biomass-to-ammonia process. Additionally, the individual MOO
results for different process configurations of the biomass-to-ammonia process have facilitated
in predicting a global MOO solution. Such an analysis will provide decision makers with
consistent and comparable data to objectively judge the viability of the proposed process.
Wood feedstock produced the maximum amount of ammonia possible. The LCC was found to
be the lowest in the Indian scenario, followed by Brazil and Australia. Lower capital costs and
operating costs in India rendered the straw feedstock economically favorable. However, the
GHG-intensive electricity production and straw production in India make the GWP of the
straw-to-biomass process similar to the GWP of fossil fuel-based ammonia production. Use of
bagasse feedstock in Brazil was found to have the lowest GWP. This advantage is attributed to
renewables-based electricity production and the absence of transportation for the bagasse
feedstock in Brazil. Wood-based ammonia production was found to the most expensive due to
the high operating costs in Australia. The GWP of the wood-to-ammonia process was mid-way
between straw and bagasse. The MOO results show that the GWP of the biomass-to-ammonia
process may be reduced if some amount of the syngas that is produced from gasification is used
for electricity production. This would reduce the GHG emissions that are related to the import
of electricity. However, this would lead to a reduction in the output of ammonia and an increase
in its unit cost. The trade-off between ammonia production and electricity production varied
for the three feedstocks, based on what the major contributors to LCC and LCA were. The
results elucidate the effect of the biomass feedstocks and the geographical entity under
consideration on the economic and environmental suitability of any biomass-based system.
This fills in the research gap in the quantification of the effect of biomass feedstocks and
location on the economics and life cycle emissions of a biomass-based system.

It can be concluded that a biomass-based ammonia production process can help counter the
environmental emission of a conventional ammonia production process. The incremental

101
production cost could be justified in the case of remote, small-scale requirements. Apart from
establishing the viability of a small-scale biomass-to-ammonia plant, this study also establishes
the advantages of process modeling and simulation at different levels, namely, the phase level,
the process unit level, the plant level, and the environmental effects level. The proposed
methodology can be used as an example in the development of different sustainable chemical
production processes.

8.1 Future Work


In order to enhance this study further, the author recommends utilization of a three-dimensional
CFD model of the gasifier rather than only a cross-section, in order to encompass the complete
gasifier design. The aim of using the three-dimensional model is to produce a syngas
composition that maximizes the production of ammonia, while minimizing the cost of the
production. Such a study would provide a feedback loop between the gasifier geometry and the
synthesis gas output. This would also widen the scope for optimization of the whole process,
by allowing for incorporation of greater number of decision variables. The number of decision
variables undertaken in the present thesis are limited because of the assumptions undertaken.
However, the scope of decision variable would increase with increase in complexity of the
biomass-to-ammonia simulation model. An additional enhancement to the proposed biomass-
to-ammonia process would include a more rigorous modeling of a complex PSA process. A
more comprehensive PSA simulation is expected to yield a better estimation of the PSA process
parameters that may lead to more favorable economics. Another value-addition of such PSA
modeling could be the study of the biomass-to-ammonia process including CCS, in which the
PSA off-gas, being rich in carbon dioxide, could be sequestrated. CCS is expected to make the
biomass-to-ammonia process carbon negative at only a small additional cost.

These results predict that electricity consumption would be a major contributor to both
economics as well as environmental emissions. Application of technologies that would result
in a reduction in electricity consumption would make the biomass-to-ammonia process more
competitive when compared to natural gas feedstock. Application of pressurized gasification
of biomass, which is still at a development stage, can help to reduce electricity consumption.
This study reiterates that different biomass feedstocks, process configurations, and
geographical regions have their own niche economic and environmental advantages. Thus,
capitalizing on the advantages that result from different factors must be emphasized in order to
enhance the viability of the proposed green processes.

102
Appendix A: Review of different RTD approaches.
The calculation of the Residence Time Distribution (RTD) using different approaches is
explained in this section. In the first approach, the flow fields were frozen for a given time
instant. A pulse tracer with zero diffusivity was introduced from the inlet for a duration of
0.005 s. Thereafter, only the tracer equation was solved, keeping the volume fraction and the
velocity profiles constant. The tracer concentration was studied at the outlet and this data was
used to calculate the E-curve. The RTD obtained separately at three different time instants of
20 s, 25 s and 30 s are shown in Figure A1. It can be seen that the RTD obtained is a function
of the time instant at which the flow fields are frozen. The mean residence time, variance and
skewness (different moments) of the three different RTDs are shown in Table A1.

0.4
0.35
0.3
0.25
20 sec
E(t)

0.2
0.15 25 sec
0.1 30 sec
0.05
0
0 5 10 15 20
t (sec)

Figure A1: RTD obtained after freezing the velocity field at 20, 25 and 30s, respectively.

Table A1: RTD characteristics (frozen velocity approach).


Time instant Mean residence time (s) Variance (s2) Skewness
20 s 10.54 2.15 1.78
25 s 10.88 2.58 2.04
30 s 10.80 1.85 1.59

The variation in RTDs at different time instants shows that the transient behaviour cannot be
avoided. Thus, the flow and other equations need to be solved simultaneously with the tracer
equation. However, this is computationally expensive. A novel multiple snapshot approach

103
has been proposed by Buwa and Ranade (2003) to calculate the mixing time of bubble columns.
In this approach, the transient simulations are solved and the results are stored at regular
intervals. The tracer is introduced in the similar way to the previous approach. However instead
of keeping the flow fields static, as in the previous approach, the flow fields are regularly
updated with the previously stored results. This takes care of the dynamic behavior of the
fluidized bed. This approach is proposed as a trade-off between accuracy and the computational
load. Specialized UDFs (User Defined Functions) were written in Fluent for application of this
approach. This approach leads to a significant reduction in the computational time when
compared to a full transient solution. To implement this approach, the velocity profiles and
volume fractions for the fluid and solid phases were saved after every 1 second for 12
subsequent seconds. These saved snapshots were used to change the velocity and volume
fraction profiles at regular intervals, as explained by the multiple snapshot approach. In this
study, the multiple snapshot approach is employed by changing the snapshots every 5, 2 and
0.5 s. The results are shown in Figure A2.

0.35

0.3

0.25

0.2 0.5 sec


E(t)

0.15 2 sec
5 sec
0.1

0.05

0
0 10 20
t (sec)

Figure A2: RTD obtained from the multiple snapshot approach.

The RTDs obtained by using multiple stored snapshots varied significantly. The mean
residence time, variance and skewness for the snapshot approach results are reported in
Table A2. It was also noted that the order of change of snapshots also had a significant effect
on the final RTD curve. In these cases the snapshots were changed in the same order with only
the interval between the snapshots being varied.

104
Table A2: RTD characteristics (snapshot approach)
Snapshot changing Mean residence time (s) Variance (s2) Skewness
interval
0.5 s 11.56 5.22 2.05
2s 10.52 2.12 1.07
5s 10.98 3.87 2.51

In the third approach, an attempt was made to average out the flow properties. In this approach,
the transient simulation was performed for a duration of 5 s of simulation time and all the
transport properties were averaged over this duration. The tracer was introduced in the
averaged out stagnant flow fields and the tracer concentration was noticed at the outlet. This
approach helped in removal of discrete velocity variation due to presence to bubbles, which
was much more evident in the earlier two approaches. The tracer profile and the RTD
characteristics are shown in Figure A3 and Table A3 respectively.

0.4

0.3
E(t)

0.2

0.1

0.0
0 5 10 15 20
t(s)

Figure A3: RTD obtained by data sampling approach.

Table A3: RTD characteristics (snapshot approach)


Simulation start time Mean residence time (s) Variance (s2) Skewness
20 s 10.03 5.07 1.39

105
In the final approach, all the equations of the transient simulation were solved simultaneously
with the tracer equation. The simulations were run with a time step size of 0.001 s with a
maximum of 100 iterations per time step. The convergence criteria for residuals were
maintained at 0.0001. The tracer introduction was similar to the previous approaches. This
approach was found to be very expensive computationally. It took 15 days of computation on
an Intel® Core™ i7-3770 3.4GHz processor utilizing four cores. The RTD obtained from full
transient simulation is shown in Figure A4. The mean residence time, variance and skewness
for the transient approach results are reported in Table A4.

0.9
0.8
0.7
0.6
0.5
E(t)

0.4
0.3
0.2
0.1
0
0 5 10 15 20
t(s)

Figure A4: RTD obtained from the full transient simulation.

Table A4: RTD characteristics (snapshot approach)


Simulation start time Mean residence time (s) Variance (s2) Skewness
20 s 10.89 1.84 1.83

The RTDs obtained by full transient simulation are found to vary significantly when compared
to the former approaches. Although the mean residence time and variance are quite similar to
the values calculated in the former approaches, the shape of the RTD varied considerably. To
understand the cause for this variation, the tracer flow through the bed was studied for all four
approaches. It was found that the tracer remained much more concentrated in the transient
approach when compared to the other three approaches. This pattern of tracer flow is shown in
Figure A5. This phenomenon might be attributed to the rapidly changing velocity profiles in

106
the fluidized bed. It is noted that the gas phase velocity in the bubbles is an order of magnitude
higher when compared to the gas phase velocity in the rest of the dense bed. The rising bubbles
are themselves quite fast with ~4–5 bubbles passing though the bed every simulation minute.
When the velocity profiles are frozen for certain time intervals in the first two approaches, this
rapid change in the gas phase velocity because of bubble flow is neglected. As a consequence,
the tracer tends to travel on the frozen velocity profiles for a much greater time than that it does
in fully transient simulations. The variations in velocity with time, at five different points in
the geometry (three in the centerline and two along the wall) are shown in Figure A6. This also
explains the smoothness of RTD curves derived from the former two approaches, for they fail
to capture the extremely fast transient behavior of the gas phase in the solid bed. It should be
noted that experiments would be helpful in validating the stated results. However, such
experiments are out of scope of the present study.

Figure A5: Tracer flow in (a) frozen velocity profiles and (b) transient systems.

107
Figure A6: Velocity profile at different points in the geometry.

108
Appendix B: Pyrolysis correlation constants
The wood pyrolysis correlation constants for Equation 3.6 are reported in Table B1.

Table B1: Wood pyrolysis correlation constants for Equation 3.6 (Abdelouahed et al., 2012;
Dufour et al., 2009a).
Syngas composition a(*10-5) b(*10-2) c
CH4 -4.341 10.12 -51.08
H2 1.362 -2.517 12.19
CO -3.524 9.770 -24.93
CO2 3.958 -9.126 64.02
C2H4 -6.873 14.94 -76.89
C2H6 0.826 -2.105 13.38
C6H6 -3.134 7.544 -42.72
C7H8 -0.453 0.687 1.462
C6H6O 1.508 -3.662 22.19
C10H8 -0.854 1.882 -9.851
H2O 5.157 -11.86 84.91
Char Elemental Composition
C -2.50 6.59 50.87
H 1.75 -4.62 26.99
O 0.75 -2.34 22.14

Linear pyrolysis correlation constants for birch wood, bagasse and straw are reported in
Table B2. The linear correlation used is displayed in equation B1.

𝑌𝑖 = 𝑚𝑇 + 𝑐 (B1)

where Yi is the mass yield of different pyrolysis products that are based on a anhydrous biomass
input, T is the temperature of the pyrolysis reactor and m and c are constants which are reported
in the Table B2.

109
Table B2: pyrolysis correlation constants for Equation B1
Syngas composition m C
Birch Wood
H2 1.02E-04 -9.88E-02
CH4 -5.40E-06 9.13E-02
C2H4 -2.57E-04 3.33E-01
C2H6 -1.55E-05 1.98E-02
C6H6 -5.10E-05 8.58E-02
CO2 1.42E-04 -3.16E-02
CO 5.46E-04 -1.17E-01
Straw Pellets
H2 1.09E-04 -1.01E-01
CH4 -1.14E-04 2.07E-01
C2H4 -2.06E-04 2.64E-01
C2H6 -2.26E-05 2.88E-02
C6H6 -9.05E-05 1.15E-01
CO2 -8.80E-04 1.22E+00
CO 1.71E-03 -1.52E+00
Bagasse
H2 1.33E-04 -8.37E-02
CH4 -1.19E-04 1.62E-01
C2H4 -2.57E-04 2.57E-01
C2H6 -1.58E-05 1.58E-02
C6H6 -1.74E-05 2.26E-02
CO2 -9.92E-04 1.01E+00
CO 1.66E-03 -9.10E-01

110
Appendix C: ICI Leading Concept Ammonia process
The ICI Leading Concept Ammonia plant was modeled in Aspen Plus to gain a better
understanding of the process. The simulation was based on equilibrium reactors and used
Redlich Kwong Soave Boston Mathias (RKS-BM) as the property package. Gibbs or
equilibrium reactors were used to model the flowsheet reactors. These reactors work on the
principle of minimizing the Gibbs free energy, thus, predicting the equilibrium composition.
The choice of this type of reactor model is based on the fact that the undergoing reactions are
known to reach close to equilibrium (Appl, 2007). The process flowsheet is shown in Figure C1
(Kroschwitz and Seidel, 2000). The ASPEN Plus simulation as well as the material balance for
different streams is shown in Figure C2 and Table C1, respectively.

Figure C1: ICI Leading Concept Ammonia process flowsheet (Kroschwitz and Seidel, 2000).

111
Figure C2: ASPEN Plus flowsheet for ICI Leading Concept Ammonia process.

112
Table C1: Stream table for ICI Leading Concept Ammonia process.
INLET GHR-IN REF-OUT SHFT-IN SHIFTOUT PSA-IN PSA-OUT SYNGAS TO-REAC REAC- AMM- PRODUCT
OUT PUR
Temperature (C) 45 220.4 962.5 386 265 55 65.3 85.3 54 380 35 -30
Pressure (bar) 41 40 37.7 37.1 37 35 35 33.8 82 82 75 75
Vapor Frac 1 1 1 1 1 1 1 1 1 1 1 0
Mole Flow (kmol/hr) 640.00 2691.01 4988.99 4988.99 4988.99 3989.95 2570.79 2541.81 11783.52 10686.45 335.69 1096.87
Mass Flow (kg/hr) 10315.13 44599.03 82828.38 82828.38 82828.38 64828.86 22665.46 22284.10 118543.05 118543.05 3497.67 18723.48
Mole Frac
CH4 0.958 0.232 0.004 0.004 0.004 0.005 0.004 0.006 0.029 0.032 0.035 0.002
H2O 0 0.632 0.282 0.282 0.203 0.004 0.007 0 0 0 0 0
H2 0.018 0.09 0.357 0.357 0.436 0.545 0.736 0.739 0.678 0.593 0.661 0.001
CO 0 0 0.083 0.083 0.004 0.005 0.002 0 0 0 0 0
CO2 0 0 0.043 0.043 0.122 0.153 0 0 0 0 0 0
N2 0.006 0.035 0.226 0.226 0.226 0.283 0.248 0.251 0.258 0.233 0.26 0.001
O2 0 0 0 0 0 0 0 0 0 0 0 0
AR 0 0.003 0.004 0.004 0.004 0.005 0.004 0.004 0.018 0.019 0.021 0.002
NH3 0 0.004 0.001 0.001 0.001 0 0 0 0.018 0.122 0.023 0.994
C2H6 0.018 0.004 0 0 0 0 0 0 0 0 0 0

113
One of the main innovations of the ICI process is the use of a Gas heated Reformer (GHR) where
heat from the secondary reformer supports the reactions in the primary reformer thus eliminating
the need for a furnace. Apart from this niche heat sinks have been located for all heat sources
making the plant more self-sufficient vis-à-vis. its heating requirements. The plant produces a
small amount of HP steam (60bar) and all the compressors rely on electricity. The major heat
integration can be understood with the help of Figure C3.

Figure C3: Heat exchanger network for the ICI plant.

It is seen that the heat integration has been localised to a great extent. Although, it might have led
to a loss of efficiency, it has greatly contributed to minimising the start-up and shut-down time,
optimising the number of heat exchangers and modularisation of the whole plant.

In order to better understand the heat exchanger network, a heat integration was carried out using
first principles and a minimum temperature difference (dTmin) of 20 ºC. The Grand Composite
Curve thus obtained can be seen in Figure C4.

114
Figure C4: Grand Composite Curve for the ICI process

The ASPEN Plus simulation predicts a potential for about 8000 KW of HP steam (60 bar). This
goes well with the GCC. However, a significant quantity of heat is lost in this process. It must be
noted that the GCC doesn’t take into account air preheating for the secondary reformer and the
refrigeration unit. As per the ICI philosophy air is heated by the PSA off-gas. However if the air-
preheat is also taken into account, the GCC would change and can be seen in Figure C5.

Figure C5: GCC for the ICI process (including air preheating)
115
It can be seen that the air-preheating can be done with the heat available in the process but at the
expense of steam that was produced earlier. In order to see its applicability a heat exchanger
network was designed. The heat exchanger network so designed could be seen in Figure C6 and
Figure C7.

Figure C6: Heat Exchanger design made using Aspen Energy analyser

116
Figure C7: Revised heat exchanger network for the ICI plant.

It can be seen that the optimum heat exchanger network is very complex and involves considerable
stream splitting which may be hard to realise in practice. Furthermore, it results in much less LP
steam production. On the positive side, the PSA off-gas serves as a side fuel and can be used for
other processes in the vicinity of the plant. It can thus be proposed that the heat exchanger network
for the ICI plant is very intricately designed, reaching a trade-off between efficiency and capital
cost. Additionally, the original heat exchanger network decouples the process steps and eliminates
the interactions and interdependencies. This further gives the plant the flexibility to operate
efficiently at reduced feed rates. As a consequence care was taken not to alter the original heat
exchanger network for the development of biomass-to-ammonia process.

117
Appendix D: Syngas Reforming
ASPEN Plus software provides various modules for modeling of reacting systems. Two of the
major modeling approaches may be termed as equilibrium approach and kinetic approach. The
equilibrium approach is based on stability, wherein the system reaches its most stable
configuration based on the reactants and the reactions involved. The kinetic approach is based on
the reactivity and measures how fast the reactants are moving towards equilibrium for given
reaction conditions. While the former approach is faster, it lacks accuracy since it is extremely
difficult to approach the exact equilibrium composition in commercial reactors. The latter
approach is computationally expensive and coupled with convergence issues (resulting from
complex integrations). However, the latter approach is much closer to real-world situations and
can be used for the accurate estimation of reactor sizes. The study compared the two approaches
for reforming of tars and methane. The comparison was done for ATR and SMR process
configurations. For the SMR approach, the reactor temperature was employed to predict the
equilibrium composition of syngas. The ATR reactor was modeled as an adiabatic reactor and the
air inlet to the reactor was used to control the reactor temperature as well as the equilibrium
composition.

A kinetic modeling of SMR and ATR reactors, a compartment modeling approach was used. The
SMR reactor was divided into two reactors in series. The first reactor was used to model the
dolomite, which is essential to reduce the tar concentration below 2000 g/Nm3 to avoid the
poisoning of Ni catalyst downstream (Srinivas et al., 2013). The second reactor modeled the
reforming of methane and tars over a Ni catalyst. The heat needed for sustaining the endothermic
reaction was provided from outside. A similar methodology was used to model the ATR reactor.
The first reactor represented partial oxidation of syngas to reach the desired temperature. The
second reactor was used to model the catalytic reforming of methane and tars. No heat was
supplied from outside for the ATR configuration. The reactions considered for both configurations
are shown in Table D1. Reaction kinetics proposed by Srinivas et al. (2013) have been used for all
the reactions. Complete conversion was considered for hydrocarbons of molecular weight greater
than methane.

118
Table D1: List of reactions considered for syngas reforming.
Tar reforming on dolomite (SMR)
2 C7H8 + 21 H2O --> 7 CO2 + 29 H2 + 7 CO
C6H6 + 2 H2O --> 1.5 C + 2.5 CH4 + 2 CO
C10H8 + 4 H2O --> C6H6 + 4 CO + 5 H2

Partial oxidation of syngas (ATR)


CO + 0.5 O2 --> CO2
C6H6 + 4.5 O2 --> 6 CO + 3 H2O
C10H8 + 7 O2 --> 10 CO + 4 H2O
H2 + 0.5 O2 --> H2O
CO + H2O --> CO2 + H2
C10H8 --> 7.38 C + 0.275 C6H6 + 0.97 CH4 + 1.235 H2
C6H6 + 2 H2O --> 1.5 C + 2.5 CH4 + 2 CO
C + H2O --> CO + H2
C7H8 + 3.5 O2 --> 7 CO + 4 H2
2 C7H8 + 21 H2O --> 7 CO2 + 29 H2 + 7 CO
C7H8 + H2 --> CH4 + C6H6

Syngas reforming on Ni catalyst (ATR and SMR)


2 C7H8 + 21 H2O --> 7 CO2 + 29 H2 + 7 CO
C7H8 + H2 --> CH4 + C6H6
C10H8 --> 7.38 C + 0.27 C6H6 + 0.97 CH4 + 1.235 H2
C6H6 + 2 H2O --> 1.5 C + 2.5 CH4 + 2 CO
C + H2O --> CO + H2
C10H8 + 4 H2O --> C6H6 + 4 CO + 5 H2
CO2 + H2 --> CO + H2O
CH4 + H2O --> 3 H2 + CO
The results obtained for both equilibrium and kinetic approach were compared. The relative
difference between equilibrium and kinetic models for the ATR and SMR configurations were
found to be 3.9 percent and 2.2 percent respectively, in terms of final ammonia production. Since
the deviation between the equilibrium and kinetic approach was not found to be very significant,
it was thus decided to model the reformer reactors using the equilibrium approach. This would
assist in faster and easier convergence of the ASPEN Plus simulations. Given the fact, that the
MOO would involve solving of thousands of simulations, even a small decrease in convergence
time would be helpful in reducing the MOO computational load.

119
Appendix E: Economic modeling parameters

Heat Integration
In order to automate the heat-integration process, two spreadsheets are prepared in ASPEN Plus
software. The first spreadsheets import the duty and temperatures of various heat sources and sinks
that are available in the flowsheet. Based on the imported data, a GCC is prepared, and the steam
generation potential is calculated (Figure E1 (a)). The steam that is generated is used to produce
electricity. The second spreadsheet imports the data for the biomass-to-ammonia process as well
as electricity generation process. An updated GCC is now prepared to calculate additional utility
requirement namely the required cooling duty. This helps in closing the GCC (Figure E1 (b)).
Based on the biomass-to-ammonia process, the electricity production process, and the calculated
utility requirements; hot and cold composite curves that represent the entire process are
formulated. The hot and cold composite curves are utilized to estimate the total heat exchanger
area. This is done by dividing the composite curves into discrete enthalpy intervals. A change in
the slope of either the hot or the cold composite curve defines a new enthalpy interval
(Figure E1(c)).

1000 1000
a b
Temperature (ºC)

900
Temperature (ºC)

900
800 800
700 700
600 600
500 500
400 400
300 300
200 200
100 100
0 0
0 10 20 0 2 4 6
Heat (MW) Heat (MW)

1000
Temperature (ºC)

800 c
600
Hot Composite Curve
400
Cold Composite Curve
200
0
0.00 2.00 4.00 6.00 8.00 10.00 12.00
Heat (MW)

Figure E1: Heat integration working methodology.


120
The film transfer coefficients of various streams were used for the calculation of heat exchanger
area for the steam/electricity generation cycle. The film heat transfer coefficients (hi) are calculated
from fluid heat transfer coefficients (h) and fouling factors (hf). The relationship is shown in
Equation E1. The influence of wall coefficients has been neglected in this study. The heat transfer
coefficients and fouling factors reported by Harkins (2012) have been used in this study (Table E1).

1 1 1
= + (E1)
ℎ𝑖 ℎ ℎ𝑓

Table E1: Film heat transfer coefficients for different process streams.
Duty hi (W/m2 K) h (W/m2 K) hf (W/m2 K)
Boiler feed water 2429 3464 8124
Steam Generation 2884 4472 8124
Steam superheating/desuperheating 1905 2353 10000
Air heater 317 353 3162
Flue gas/syngas cooling 188 200 3162
Flue gas condensing region 4330 8660 8660
Ammonia condensing 2228 3000 8660
Cooling water 1907 3464 4243
Compressors intercoolers 3429 8660 5678
Refrigeration 4330 8660 8660

A general cost function reported by Girardin et al. (2009) was utilized for calculation of heat
exchanger costs for the steam/electricity generation. The approach is based on the number of heat
exchangers anticipated (N) and the total heat exchanger area (A). The costing calculations were
performed in 2006 US dollars as per Equation E2.

Chex = 7038*N*(A/N)0.7948 (E2)

121
Costing Parameters
The costing parameters utilized for costing of various pieces of equipment as reported in Chapter
4 are presented in Table E2.

Table E2: Costing parameters for different pieces of equipments.


Equipment Modeled as K1 K2 K3 Capacity, Units
Compressors Reciprocating compressor 2.2897 1.3604 -0.1027 Fluid power, KW
Drive (electric, totally enclosed) 1.956 1.7142 -0.2282 Shaft power, KW
Reactors Vertical process vessels 3.4974 0.4485 0.1074 Volume, m3
Heat exchangers Floating head heat exchanger 4.8306 -0.8509 0.3187 Area, m2
Heat exchangers Air cooler 4.0336 0.2341 0.0497 Area, m2
Water pump Reciprocating pump 3.8696 0.3161 0.122 Shaft power, KW
Ammonia storage Tank (API-fixed roof) 4.8509 -0.3973 0.1445 Volume, m3
Back pressure turbine Axial gas turbine 2.7051 1.4398 -0.1776 Fluid power, KW
Condensing turbine Gas/liquid expander 2.2476 1.4965 -0.1618 Fluid power, KW
Drive (turbine) Steam turbine 2.6259 1.4398 0.1776 Shaft power, KW

The utility costs assumed by this study are presented in Table E3.

Table E3: Utility cost parameters for the biomass-to-ammonia process.

Component Unit Cost Unit

Natural gas price 5.5 $/GJ

Reformer Catalyst 25.44 $/kg

Shift Catalyst 17.64 $/kg

Methanation Catalyst 22.05 $/kg

Synthesis catalyst 26.79 $/kg

Adsorbent 9.00 $/kg

Bed material 172.90 $/ton

Ash disposal 23.52 $/ton

Boiler feed water 3 $/ton

122
Appendix F: Life cycle inventories
Ecoinvent inventories used for wood-to-ammonia process are reported in Table F1.

Table F1: Ecoinvent inventories used for wood-to-ammonia process.


LCI element (Unit) Ecoinvent equivalent
Eucalyptus cultivation, in m3 Wood chips, mixed, u=120%, at forest/RER U/AusSD S
Road Transport, in tkm Transport, lorry >28t, fleet average/CH U/AusSD U/Link S
Electricity, in kWh Electricity, low voltage, New South Wales/AU S
Infrastructure, in p Synthetic gas plant/CH/I U/AusSD S
Rail transport, in tkm Transport, freight, rail/AU S
Fuel (Natural gas), in MJ Natural gas, burned in industrial furnace >100kW/RER U/AusSD
S
Sand (Bed material), in kg Silica sand, at plant/DE U/AusSD S
Charcoal, in kg Charcoal, at plant/GLO U/AusSD S
Calcium carbonate, in kg Calcium carbonate > 63 microns, production, at plant EU-27 S (of
project ELCD)
Rape methyl ester, in kg Rape methyl ester, at esterification plant/RER S
Sodium Hydroxide, in kg Sodium hydroxide, 50% in H2O, production mix, at plant/RER
U/AusSD S
Sulphuric acid, in kg Sulphuric acid, liquid, at plant/RER U/Adapted/AU S
Zinc oxide, in kg Zinc oxide, at plant/RER U/AusSD S
Nickel catalyst (Reforming), in kg Nickel, 99.5%, at plant/GLO U/Adapted/AU S
Dolomite, in kg Dolomite, at plant/RER U/AusSD S
Copper catalyst(Shift), in kg Copper oxide, at plant/RER U/AusSD S
Activated carbon (adsorbent), in kg Activated carbon, at plant/RER Economic (of project Agri-
footprint - economic allocation)
Magnetite (synthesis catalyst), in kg Magnetite, at plant/GLO U/AusSD S
Fly ash inceration, in kg Fly ash and scrubber sludge {GLO}| market for | Alloc Def, S
Waste landfill, in kg Inert waste, for final disposal {GLO}| market for | Alloc Def, S
Waste water, in kg Treatment, sewage, from residence, to wastewater treatment,
class 2/CH U/AusSD S

123
Ecoinvent inventories used for bagasse-to-ammonia process are reported in Table F2.

Table F2: Ecoinvent inventories used for bagasse-to-ammonia process


LCI element (Unit) Ecoinvent equivalent
Bagasse production, in kg Bagasse, from sugarcane {BR}| cane sugar production with
ethanol by-product | Alloc Rec, S
Road Transport, in tkm Transport, freight, lorry 16-32 metric ton, EURO6 {GLO}|
market for | Alloc Def, S
Electricity, in kWh Electricity, low voltage {BR}| market for | Alloc Def, S
Infrastructure, in p Synthetic gas factory {CH}| construction | Alloc Def, S
Rail transport, in tkm Transport, freight train {RoW}| market for | Alloc Def, S
Fuel (Natural gas), in MJ Heat, district or industrial, natural gas {RoW}| heat production,
natural gas, at industrial furnace >100kW
Sand (Bed material), in kg Silica sand {GLO}| market for | Alloc Def, S
Charcoal, in kg Charcoal {GLO}| market for | Alloc Def, S
Calcium carbonate, in kg Calcium carbonate > 63 microns, production, at plant EU-27 S (of
project ELCD)
Rape methyl ester, in kg Rape methyl ester, at esterification plant/RER S
Sodium Hydroxide, in kg Sodium hydroxide, without water, in 50% solution state {GLO}|
market for | Alloc Def, S
Sulphuric acid, in kg Sulfuric acid {GLO}| market for | Alloc Def, S
Zinc oxide, in kg Zinc oxide {GLO}| market for | Alloc Def, S
Nickel catalyst (Reforming), in kg Nickel, 99.5% {GLO}| market for | Alloc Def, S
Dolomite, in kg Dolomite {GLO}| market for | Alloc Def, S
Copper catalyst(Shift), in kg Copper oxide {GLO}| market for | Alloc Def, S
Activated carbon (adsorbent), in kg Activated carbon, at plant/RER Economic (of project Agri-
footprint - economic allocation)
Magnetite (synthesis catalyst), in kg Magnetite {GLO}| market for | Alloc Def, S
Fly ash inceration, in kg Fly ash and scrubber sludge {GLO}| market for | Alloc Def, S
Waste landfill, in kg Inert waste, for final disposal {GLO}| market for | Alloc Def, S
Waste water, in m3 Wastewater, average {GLO}| market for | Alloc Def, S

124
Ecoinvent inventories used for straw-to-ammonia process are reported in Table F3.

Table F3: Ecoinvent inventories used for straw-to-ammonia process


LCI element (Unit) Ecoinvent equivalent
Straw production, in kg Straw {GLO}| market for | Alloc Def, S
Road Transport, in tkm Transport, freight, lorry 16-32 metric ton, EURO6 {GLO}|
market for | Alloc Def, S
Electricity, in kWh Electricity, low voltage {IN}| market for | Alloc Def, S
Infrastructure, in p Synthetic gas factory {CH}| construction | Alloc Def, S
Rail transport, in tkm Transport, freight train {RoW}| market for | Alloc Def, S
Fuel (Natural gas), in MJ Heat, district or industrial, natural gas {RoW}| heat production,
natural gas, at industrial furnace >100kW
Sand (Bed material), in kg Silica sand {GLO}| market for | Alloc Def, S
Charcoal, in kg Charcoal {GLO}| market for | Alloc Def, S
Calcium carbonate, in kg Calcium carbonate > 63 microns, production, at plant EU-27 S (of
project ELCD)
Rape methyl ester, in kg Rape methyl ester, at esterification plant/RER U/AusSD S
Sodium Hydroxide, in kg Sodium hydroxide, without water, in 50% solution state {GLO}|
market for | Alloc Def, S
Sulphuric acid, in kg Sulfuric acid {GLO}| market for | Alloc Def, S
Zinc oxide, in kg Zinc oxide {GLO}| market for | Alloc Def, S
Nickel catalyst (Reforming), in kg Nickel, 99.5% {GLO}| market for | Alloc Def, S
Dolomite, in kg Dolomite {GLO}| market for | Alloc Def, S
Copper catalyst(Shift), in kg Copper oxide {GLO}| market for | Alloc Def, S
Activated carbon (adsorbent), in kg Activated carbon, at plant/RER Economic (of project Agri-
footprint - economic allocation)
Magnetite (synthesis catalyst), in kg Magnetite {GLO}| market for | Alloc Def, S
Fly ash inceration, in kg Fly ash and scrubber sludge {GLO}| market for | Alloc Def, S
Waste landfill, in kg Inert waste, for final disposal {GLO}| market for | Alloc Def, S
Waste water, in m3 Wastewater, average {GLO}| market for | Alloc Def, S

125
Appendix G: LCA results validation
The LCA results calculated from the MS Excel based framework were validated against the results
that were generated by the commercial LCA software Simapro for a sample problem. An error of
less than 2 percent was noticed between the two results. The comparison is shown in Table G1.
The slight difference in results is attributed to the extent of significant figures used in calculations.

Table G1: Comparison of LCA results from MS Excel model and SimaPro software.
Impact category Excel model Simapro
Climate change (kg CO2 eq) 7.75E-01 7.75E-01
Ozone depletion (kg CFC-11 eq) 4.20E-08 4.20E-08
Terrestrial acidification (kg SO2 eq) 3.29E-03 3.29E-03
Freshwater eutrophication (kg P eq) 5.32E-05 5.30E-05
Marine eutrophication (kg N eq) 5.95E-04 5.95E-04
Human toxicity (kg 1,4-DB eq) 8.69E-02 8.60E-02
Photochemical oxidant formation (kg 2.68E-03 2.68E-03
NMVOC)
Particulate matter formation (kg PM10 eq) 1.04E-03 1.04E-03
Terrestrial ecotoxicity (kg 1,4-DB eq) 4.15E-04 4.11E-04
Freshwater ecotoxicity (kg 1,4-DB eq) 3.38E-03 3.30E-03
Marine ecotoxicity (kg 1,4-DB eq) 2.82E-03 2.82E-03
Ionising radiation (kg U235 eq) 5.41E-02 5.37E-02
Agricultural land occupation (m2a) 1.95E+00 1.95E+00
Urban land occupation (m2a) 2.66E-02 2.67E-02
Natural land transformation (m2) 1.72E-04 1.73E-04
Water depletion (m3) 4.61E-03 4.61E-03

Metal depletion (kg Fe eq) 4.19E-03 4.18E-03


Fossil depletion (kg oil eq) 3.23E-02 3.22E-02

126
Appendix H: LCA methodology
The spreadsheets for calculation of LCA results which is discussed in Chapter 5 are presented in
this appendix. An input for technology matrix (A) (shown in orange) and intervention matrix (B)
(shown in green) is presented in Figure H1. The input into the matrix A are linked to the ASPEN
Plus flowsheets and are updated automatically. The actual matrix A consists of 42 rows and 42
columns, to encompass all the process stage, process inputs and process outputs. The matrix B
consists of 1900 rows to represent the different emissions for 42 different process stages
(represented by columns).

127
Figure H1: Technology matrix (A) and intervention matrix (B)

Figure H2 depicts the calculation of scaling vector, s using the inverse of Matrix A and demand
matrix, d (which represents the functional unit). The demand matrix depicts a production of 1 kg
128
of refrigerated ammonia which is the functional unit used in all LCA calculations. The scaling
vector is then used to calculate the process emissions pertaining to the intended process output or
the functional unit

Vector d Vector s

Drying and chipping of Biomass


Biomass Transportation
d Final demand vector

Eucalyptus cultivation

FICFB Gasification
s Scaling vector
Eucalyptus logs at Farm(m3) - 0.011083138 0.011083138 -0.011083138 0 0
Eucalyptus logs at Plant (wet ton)(120% m) - 0.004583976 0 0.004583976 -0.004583976 0
Dried Eucalyptus(dry ton)(5% m) - 0.002083626 0 0 0.002187807 -0.002187807
Syngas after gasification(ncm) - 0.000416646 0 0 0 3.086942773
Syngas after reforming (ncm) - 0.000416646 0 0 0 0
Syngas after cooling and cleaning (ncm) - 0.000416646 0 0 0 0
Syngas after compression(ncm) - 0.000416646 0 0 0 0
Syngas after shift and desaturator(ncm) - 0.000416646 0 0 0 0
Syngas after PSA (ncm) - 0.000416646 0 0 0 0
-1
Syngas after methanator (ncmh) - 0.000416646 0 Matrix A
0 0 0
Ammonia Produced (kg) - 0.000416646 0 0 0 0
Referigerated Ammonia (kg) 1.00 0.000416646 0 0 0 0
Water(Kg) - 1.753822107 0 0 0 0
Steam (kg) - 0.749962386 0 0 0 -0.749962386
Electricity (kwh) - 0.705198482 0 0 -0.015692306 0
Fuel(MJ) - 0.000344983 0 0 0 -0.000344983
Bed Material(Kg) - 0.019790674 0 0 0 -0.019790674
Charcoal(kg) - 0.011145274 0 0 0 -0.011145274
CaCO3(kg) - 1.19994E-20 0 0 0 -0.011999398

Figure H2: Calculation of scaling vector

The normalized process emissions are multiplied with their respective characterization factors for
the Hierarchist ReCiPe methodology (Figure H3). The emissions are then added for calculation of
different mid-point impact categories such global warming potential, ozone depletion potential and
the like. Normalization and weighing is used for calculation of end point indicator.

129
Water 1-Methylfluorene

1-Propanol

Water 1-Propanol
1-Pentanol

Water 1-Pentanol

1-Pentene

Water 1-Pentene
1-Butanol
Water 1-Butanol
Air

Air

Air

Air
climate change (kg CO2 eq) 0 0 0 0 0 0 0 0 0
ozone depletion (kg CFC-11 eq) 0 0 0 0 0 0 0 0 0
terrestrial acidification (kg SO2 eq) 0 0 0 0 0 0 0 0 0
freshwater eutrophication (kg P eq) 0 0 0 0 0 0 0 0 0
marine eutrophication (kg N eq) 0 0 0 0 0 0 0 0 0
human toxicity (kg 1,4-DB eq) 1 0 0 0 0 0 0 0 0
photochemical oxidant formation (kg NMVOC) 1 0 0 0 0 1.65 0 0.948 0
particulate matter formation (kg PM10 eq) 0 0 0 0 0 0 0 0 0
terrestrial ecotoxicity (kg 1,4-DB eq) 0 0 0 0 2.7E-05 0 0 0.0001 1.8E-05
freshwater ecotoxicity (kg 1,4-DB eq) 0 0 0 0 0.0162 0 0 4E-05 0.00264
marine ecotoxicity (kg 1,4-DB eq) 0 0 0 0 0.00012 0 0 8E-05 2.1E-05
ionising radiation (kg U235 eq) 0 0 0 0 0 0 0 0 0
agricultural land occupation (m2a) 0 0 0 0 0 0 0 0 0
urban land occupation (m2a) 0 0 0 0 0 0 0 0 0
natural land transformation (m2) 0 0 0 0 0 0 0 0 0
water depletion (m3) 0 0 0 0 0 0 0 0 0
metal depletion (kg Fe eq) 0 0 0 0 0 0 0 0 0
fossil depletion (kg oil eq) 0 0 0 0 0 0 0 0 0

Figure H3: Characterization factors of ReCiPe

130
References
Abdelouahed L., Authier O., Mauviel G., Corriou J., Verdier G. and Dufour A., ‘Detailed
Modeling of Biomass Gasification in Dual Fluidized Bed Reactors under Aspen Plus’, Energy &
Fuels, 26(6), 3840-3855, 2012.

Ahlgren S., Baky A., Bernesson S., Nordberg K., Norén O. and Hansson P.A., 'Ammonium nitrate
fertiliser production based on biomass - Environmental effects from a life cycle perspective.',
Bioresource Technology. 99, 8034–8041, 2008.

Andersson J. and Lundgren J., ‘Techno-economic analysis of ammonia production via integrated
biomass gasification’, Applied Energy, 130, 484-490, 2014.

Appl M. Ammonia: Principles and Industrial Practice, Wiley Online Library, 2007.

Appl. M., ‘Modern ammonia technology: where have we got to, where are we going?’, Nitrogen,
44 , 1993.

Araújo A. and Skogestad S., ‘Control structure design for the ammonia synthesis process’,
Computers & Chemical Engineering, 32(12), 2920–2932, 2008.

Arkley K. I., Halstead J. L. and Pinto, A., ‘Modernization of ammonia industry by small economic
plants’,.AIChE Ammonia plant safety and related issues,18, 58-65, 1988.

Asadullah M., ‘Biomass gasification gas cleaning for downstream applications: A comparative
critical review’, Renewable and Sustainable Energy Reviews, 40, 118–132, 2014.

Atsonios K., Zeneli M., Nikolopoulos A., Nikolopoulos N., Grammelis P. and Kakaras E,.'
Calcium looping process simulation based on an advanced thermodynamic model combined with
CFD analysis.', Fuel, 153, 370–381, 2015.

Australian Energy Market Operator, AEMO, http://www.aemo.com.au/, visited on 01 April 2016.

Azapagic A. and Clift R., ‘Life cycle assessment and multiobjective optimisation’, Journal of
Cleaner Production, 7(2):135–143, 1999.

131
Bartels J., ‘A feasibility study of implementing an ammonia economy’, Ph.D. Thesis, Iowa State
University, 2009.

Basu P., Biomass Gasification and Pyrolysis: Practical Design and Theory, Academic Press,
2010.

Blonk Consultants, Agri-footprint Description of data V1.0, Blonk Consultants, 2014.

Boustead I., Eco-profiles of the European Plastics Industry: Ammonia, PlasticsEurope, 2005.

Buwa V. V and Ranade V. V.,'Mixing in Bubble Column Reactors : Role of Unsteady Flow
Structures.', 81, 402–411, 2003.

Calva C.A.C., ‘Economic, Environmental and Social Optimization of Forest based Biomass
Supply Chains for Bioenergy and Biofuels’, Ph.D. Thesis, University of British Columbia,
Vancouver, 2016.

Cambero C. and Sowlati T, ‘Assessment and optimization of forest biomass supply chains from
economic, social and environmental perspectives - A review of literature’, Renewable and
Sustainable Energy Reviews, 36, 62–73, 2014

Cherubini F. and Strømman A. H., 'Life cycle assessment of bioenergy systems: State of the art
and future challenges.', Bioresource Technology, 102, 437–451, 2011.

Choi Y. and Stenger H.G. 2003, ‘Water gas shift reaction kinetics and reactor modeling for fuel
cell grade hydrogen,’ Journal of Power Sources, 124(2), 432–439, 2003.

Curran M. A., Life Cycle Assessment Handbook a Guide for Environmentally Sustainable
Products, Wiley, Hoboken, N.J., 2012.

Daggupati S., Mandapati R.N., Mahajani S.M., Ganesh A., Pal A. K., Sharma R.K. and
Aghalayam P., ‘Compartment Modeling for Flow Characterization of Underground Coal
Gasification Cavity’, Industrial & Engineering Chemistry Research, 50(1), 277–290, 2011.

Deb K. Multi-Objective Optimization Using Evolutionary Algorithms, John Wiley & Sons, 2001.

132
Deb K., Pratap A., Agarwal S. and Meyarivan T., ‘A fast and elitist multi-objective genetic
algorithm: NSGA-II’, IEEE Transactions on Evolutionary Computation, 6(2), 182-197, 2002.

De Schryver, A.M., ‘Value choices in life cycle impact assessment’, Ph.D. Thesis, Radboud
University, Nijmegen, 2011.

Ding J and Gidaspow D, ‘A Bubbling Fluidization Model Using Kinetic Theory of Granular Flow’,
AIChE Journal, 36(4), 523–538, 1990.

Doherty W., Reynolds A. and Kennedy D., Aspen plus simulation of biomass gasification in a
steam blown dual fluidised bed, A. Méndez-Vilas (Ed.), Materials and processes for energy:
communicating current research and technological developments, Formatex Research Centre,
2013.

Dou B., Dupont V. and Williams P.T., ‘Computational Fluid Dynamics Simulation of Gas-Solid
Flow during Steam Reforming of Glycerol in a Fluidized Bed Reactor,’ Energy & Fuels, 22(6),
4102–4108, 2008.

Dufour A., Girods P., Masson E., Rogaume Y. and Zoulalian A., ‘Synthesis gas production by
biomass pyrolysis: Effect of reactor temperature on product distribution’, International Journal of
Hydrogen Energy, 34(4), 1726-1734, 2009a.

Dufour A., Valin S., Castelli P., Thiery S., Boissonnet G., Zoulalian A. and Glaude P.,
‘Mechanisms and Kinetics of Methane Thermal Conversion in a Syngas’, Industrial &
Engineering Chemistry Research , 48(14), 6564-6572, 2009b.

E4tech, Review of Technologies for Gasification of Biomass and Wastes, NNFCC, 09/008, 2009

Economic indicators. Chemical Engineering, 123(3), 92, 2016.

Edenhofer O., Pichs-Madruga R., Sokona Y., Minx J.C., Farahani E., Susanne K., Seyboth K.,
Adler A., et al. Climate Change 2014: Mitigation of Climate Change, Working Group III
Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change,
IPCC, 2014.

133
Eisentraut A., Sustainable Production of Second-Generation Biofuels, IEA, 2010.

Francois J., Abdelouahed L., Mauviel G., Patisson F., Mirgaux O., Rogaume C., Rogaume Y.,
Feidt M. and Dufour A., ‘Detailed process modeling of a wood gasification combined heat and
power plant’, Biomass and Bioenergy, 51, 68-82, 2013.

Gerber L., Gassner M. and Maréchal, F., ‘Systematic integration of LCA in process systems
design: Application to combined fuel and electricity production from lignocellulosic biomass’,
Computers and Chemical Engineering, 35(7), 1265–1280, 2011.

Gerbens-Leenes P.W., Hoekstra A.Y., van der Meer Th., ‘The water footprint of energy from
biomass: A quantitative assessment and consequences of an increasing share of bio-energy in
energy supply’, Ecological Economics, 68(4), 1052-1060, 2009. Gilbert P., Alexander S.,
Thornley P. and Brammer J. ‘Assessing economically viable carbon reductions for the production
of ammonia from biomass gasification.’ Journal of Cleaner Production 64, 581–589, 2014.

Girardin L., Bolliger R. and Marechal F., ‘On the use of process integration techniques to generate
optimal steam cycle configurations for the power plant industry’, Chemical Engineering
Transactions, 18, 171-176, 2009

Goedkoop M., Heijungs R., De Schryver A., Struijs J. and van Zelm R., ReCiPe 2008. A LCIA
method which comprises harmonised category indicators at the midpoint and the endpoint level.
First edition Report I: Characterisation, 2009.

Göransson K., Ulf S., Jie h. and Wennan Z., ‘Review of syngas production via biomass DFBGs’,
Renewable and Sustainable Energy Reviews, 15(1), 482-492, 2011

Hamelinck C. N., Suurs R. and Faaij A. P. C. ‘International bioenergy transport costs and energy
balance.' Biomass and Bioenergy 29, 114–134, 2005.

Harkins T., ‘Multi-Objective Optimisation of CCS using simulation, heat integration and cost
estimation’, Ph.D. Thesis, Monash University, Melbourne, 2012.

Hayes R.E., Thomas W.J. and Hayes K.E, ‘A study of the nickel-catalyzed methanation reaction’,
Journal of Catalysis, 92, 312–326, 1985.
134
He J., Göransson K., Söderlind U. and Zhang W. ‘Simulation of biomass gasification in a dual
fluidized bed gasifier.’ Biomass Conversion and Biorefinery, 2(1), 1-10, 2012.

Heijungs R. and Suh S. The Computational Structure of Life Cycle Assessment. Springer, 2002.

Hicks T., Pinto A. and Moss, M., ‘Operational flexibility of the LCA process’, Ammonia Plant
Safety (and Related Facilities), 152–160, 1990.

Hofbauer, H. and Rauch, R., ‘Stoichiometric water consumption of steam gasification by the
FICFB-gasification process,” Progress in Thermochemical Biomass Conversion, Innsbruck,
Austria, 2000.

Hofsetz K. and Silva M.A, ‘Brazilian sugarcane bagasse: Energy and non-energy consumption’,
Biomass and Bioenergy, 46, 564–573, 2012

Hoogwijk M., Faaij A., Broek R., Berndes G., Gielen D. and Turkenburg, W., ‘Exploration of the
ranges of the global potential of biomass for energy’, Biomass and Bioenergy, 25(2), 119–133,
2003.

Hu Y., Li H. and Yan J., Integration of Evaporative Gas Turbine with Oxy-Fuel Combustion for
Carbon Dioxide Capture, International Journal of Green Energy, 7(6), 615-631, 2010

IEA, Key World Energy Statistics 2015, IEA, 2015.

Inamdar S. V., Gupta S. K. and Saraf D. N., ‘Multi-objective Optimization of an Industrial Crude
Distillation Unit Using the Elitist Non-Dominated Sorting Genetic Algorithm’, Chemical
Engineering Research and Design, 82(5), 611-623, 2004.

Inayat A., Ahmad M., Mutalib M., and Yunus M., ‘Kinetic modeling of biomass steam gasification
system for hydrogen production with CO2 adsorption’, Proceedings of International Conference
for Technical Postgraduates, Kula Lumpur, Malaysia, 2009.

Intratec Solution, Free Location Factor for Cost Estimation, http://www.intratec.us/free-


tools/location-factor, visited on1 April 2016.

135
IPCC. Special Report on Renewable Energy Sources and Climate Change Mitigation. Cambridge
University Press, 2011.

Jana K. and De S., ‘Techno-economic evaluation of a polygeneration using agricultural residue -


A case study for an Indian district’, Bioresource Technology, 181, 163–173, 2015.

Jee J., Kim M. and Lee C., ‘Adsorption Characteristics of Hydrogen Mixtures in a Layered Bed: 
Binary, Ternary, and Five-Component Mixtures’, Industrial & Engineering Chemistry Research,
40(3), 868-878, 2001.

Johnson P. C. and Jackson R., 'Frictional-collisional constitutive relations for granular materials,
with application to plane shearing', Journal of Fluid Mechanics, 176, 67–93, 1987.

Kaiser S., Löffler G., Bosch K. and Hofbauer H., ‘Hydrodynamics of a dual fluidized bed gasifier.
Part II: Simulation of solid circulation rate, pressure loop and stability’, Chemical Engineering
Science, 58(18), 4215–4223, 2003.

Kaushal P., Proell T. and Hofbauer H., ‘Application of a detailed mathematical model to the
gasifier unit of the dual fluidized bed gasification plant’, Biomass and Bioenergy, 35(7), 2491-
2498, 2011.

Khatiwada D., Leduc S., Silveira S. and McCallum I., ‘Optimizing ethanol and bioelectricity
production in sugarcane biorefineries in Brazil’, Renewable Energy, 85, 371–386, 2016.

Kitzler H., Pfeifer C. and Hofbauer H. ‘Gasification of different kinds of non woody biomass in a
100 KW dual fluidized bed gasifier’, Proceedings of 21st International Conference on Fluidized
Bed Combustion, 760–766, 2012.

Knoef H., Handbook biomass gasification, BTG Biomass Technology Group, 2012.

Konak A., Coit D.W. and Smith A.E., ‘Multi-objective optimization using genetic algorithms: A
tutorial’, Reliability Engineering and System Safety, 91(9), 992–1007, 2006.

Kool A., Marinussen M. and Blonk H., LCI data for the calculation tool Feedprint for greenhouse
gas emissions of feed production and utilization Dry Milling Industry, Blonk Consultants, 2012.

136
Kroschwitz J. and Seidel A., Ammonia, Kirk-Othmer Encyclopedia of Chemical Technology, John
Wiley & Sons, New York, 2000.

Lee E. S-Q., Chapter 10: Optimal Design of Chemical Processes for Multiple Economic and
Environmental Objectives, G.P. Rangaiah (Ed.), Multi-objective Optimisation: Techniques and
Applications in chemical engineering, Singapore: World Scientific, 301-338, 2009

Levenspiel O., Chemical Reaction Engineering, Wiley, 1999.

Leyland G.B., ‘Multi-objective optimisation applied to industrial energy problems’, Ph.D. Thesis,
EPFL, 2002.

Liu H., Cattolica R.J., Seiser R. and Liao-heien, C., ‘Three-dimensional full-loop simulation of a
dual fluidized-bed biomass gasifier’, Applied Energy, 160, 489–501, 2015.

Loha C., Chattopadhyay H. and Chatterjee P.K., ‘Three dimensional kinetic modeling of fluidized
bed biomass gasification’, Chemical Engineering Science, 109, 53–64, 2014a.

Loha C., Gu S., De Wilde J., Mahanta P. and Chatterjee P.K., ‘Advances in mathematical modeling
of fluidized bed gasification’, Renewable and Sustainable Energy Reviews, 40, 688–715, 2014b.

Lun C. K. K., Savage S. B., Jeffrey D. J. and Chepurniy, N., 'Kinetic theories for granular flow:
inelastic particles in Couette flow and slightly inelastic particles in a general flowfield', Journal of
Fluid Mechanics, 140, 223–256, 1984.

Moré J.J., The Levenberg-Marquardt algorithm: implementation and theory. Numerical analysis,
Springer, 105–116, 1978.

Morgan E., ‘Techno-Economic Feasibility Study of Ammonia Plants Powered by Offshore Wind’,
Ph.D. Thesis, University of Massachusetts, 2013.

Nandasana A. D., Ray A. K.and Gupta S. K., ‘Dynamic Model of an Industrial Steam Reformer
and Its Use for Multiobjective Optimization’,Industrial & Engineering Chemistry Research, 42
(17), 4028-4042, 2003.

137
Nguyen T. D. B., Seo M. W., Lim Y. I., Song B. H. and Kim, S. D., 'CFD simulation with
experiments in a dual circulating fluidized bed gasifier.', Computers & Chemical Engineering, 36,
48–56, 2012.

ISO 14040:2006, Environmental management - Life cycle assessment - Principles and framework,
International Organization for Standardization, Geneva, 2006.

Patel M., Zhang X. and Kumar A., ‘Techno-economic and life cycle assessment on lignocellulosic
biomass thermochemical conversion technologies: A review’, Renewable and Sustainable Energy
Reviews, 53, 1486–1489, 2016.

Pfeifer C., Rauch R. and Hofbauer H., ‘In-Bed Catalytic Tar Reduction in a Dual Fluidized Bed
Biomass Steam Gasifier,’ Industrial & Engineering Chemistry Research, 43(7), 1634–1640, 2004.

Pfeifer C., Koppatz S. and Hofbauer H., ‘Steam gasification of various feedstocks at a dual
fluidised bed gasifier: Impacts of operation conditions and bed materials’, Biomass Conversion
and Biorefinery, 1(1), 39-53, 2011.

Pröll T., Rauch R. and Aichernig C.,‘Fluidized Bed Steam Gasification of Solid Biomass -
Performance Characteristics of an 8 MWth Combined Heat and Power Plant.’, International
Journal of Chemical Reactor Engineering, 5(1), 2007.

Puig-Arnavat M., Bruno J. and Coronas A., ‘Review and analysis of biomass gasification models’,

Puig-Arnavat, M., Hernández J., Bruno J. and Coronas A., ‘Artificial neural network models for
biomass gasification in fluidized bed gasifiers’, Biomass and Bioenergy, 49, 279-289, 2013.

Rangaiah G.P., Chapter 1: Introduction, G.P. Rangaiah (Ed.), Multi-objective Optimisation:


Techniques and Applications in chemical engineering, Singapore: World Scientific, 1-25, 2009

Ribeiro A., Santos J. and Rodrigues A., ‘PSA design for stoichiometric adjustment of bio-syngas
for methanol production and co-capture of carbon dioxide’, Chemical Engineering Journal,
163(3), 355-363, 2010.

138
Rimos, S., Hoadley, A. F. A. and Brennan, D. J. ‘Determining the economic consequences of
natural gas substitution’, Energy Conversion and Management, 85, 709-717, 2014.

Rutovitz J., NSW Bioenergy Handbook, Mark Ellis & Associates for the Department of Energy,
Utilities and Sustainability, 2004.

Santibañez-Aguilar J.E.., González-Campos J. B., Ponce-Ortega J. M., Serna-González M. and El-


Halwagi M. M., ‘Optimal planning of a biomass conversion system considering economic and
environmental aspects’, Industrial and Engineering Chemistry Research, 50(14), 8558–8570,
2011.

Santos V.E.N., Ely R.N., Szklo A.S. and Magrini A., ‘Chemicals, electricity and fuels from
biorefineries processing Brazil’s sugarcane bagasse: Production recipes and minimum selling
prices’, Renewable and Sustainable Energy Reviews, 53, 1443–1458, 2016.

Sarkar S., Kumar A. and Sultana A.,’ Biofuels and biochemicals production from forest biomass
in Western Canada’, Energy, 36(10), 6251-6262, 2011.

Schaeffer D. G.,'Instability in the evolution equations describing incompressible granular flow',


Journal of Differential Equations, 66, 19–50, 1987.

Schmid J.C., Wolfesberger U., Koppatz S., Pfeifer C. and Hofbauer H., ‘Variation of feedstock in
a dual fluidized bed steam gasifier-influence on product gas, tar content, and composition’,
Environmental Progress & Sustainable Energy, 31(2), 205–215, 2012.

Schuster G., Löffler G., Weigl K. and Hofbauer H., ‘Biomass steam gasification – an extensive
parametric modeling study’, Bioresource Technology, 77(1), 71-79, 2001.

Shahrivar H.H., Sheikhi A. and Sotudeh-Gharebagh R., ‘On the flow direction effect in sequential
modular simulations: A case study on fluidized bed biomass gasifiers,’ International Journal of
Hydrogen Energy, 40(6), 2552–2567, 2015.

Sharma S., Rangaiah G.P. and Cheah K.S., Multi-Objective Optimisation Using MS Excel with
an Application to Design of a Falling-Film Evaporator System, Food and Bioproducts Processing,
90(2), 123-134, 2012
139
Singh A., Pant D. and Olsen S. I., Life Cycle Assessment of Renewable Energy Sources, Springer
London, 2013.

Singh J. and Gu S., ‘Biomass conversion to energy in India-A critique’, Renewable and
Sustainable Energy Reviews., 14(5), 1367–1378, 2010.

Singh R., Srivastava M. and Shukla A., ‘Environmental sustainability of bioethanol production
from rice straw in India: A review’, Renewable and Sustainable Energy Reviews, 54, 202–216,
2016.

Smith R., Chemical Process: Design and Integration, John Wiley & Sons, 2005.

Srinivas S., Field R.P. and Herzog H.J., ‘Modeling tar handling options in biomass gasification,’
Energy and Fuels, 27(6), 2859–2873, 2013.

Syamlal M. and O’brien T. J., 'Computer simulation of bubbles in a fluidized bed', AIChE
Symposium Series, 85, 22–31, 1989.

Thapa R. K., Pfeifer C. and Halvorsen B.M., ‘Modeling of reaction kinetics in bubbling fluidized
bed biomass gasification reactor’, International Journal of Energy and Environment, 5(1), 35-44,
2014.

Thomassen M. A., Dalgaard R., Heijungs R. and De Boer I., 'Attributional and consequential LCA
of milk production', International Journal of Life Cycle Assessment, 13, 339–349, 2008.

Timmer K., ‘Carbon conversion during bubbling fluidized bed gasification of biomass’, Ph.D.
Thesis, Iowa State University, 2008.

Tock L. and Maréchal F., ‘H2 processes with CO2 mitigation: Thermo-economic modeling and
process integration’, International Journal of Hydrogen Energy, 37(16), 11785–11795, 2012a.

Tock L. and Maréchal F., ‘Co-production of hydrogen and electricity from lignocellulosic
biomass: Process design and thermo-economic optimization’, Energy, 45(1), 339–349, 2012b.

140
Tock L., Gassner M. and Maréchal F. ‘Thermochemical production of liquid fuels from biomass:
Thermo-economic modeling, process design and process integration analysis, Biomass and
Bioenergy, 34(12), 1838–1854, 2010.

Tock L., Maréchal F. and Perrenoud M. ‘Thermo-environomic evaluation of the ammonia


production.’ Canadian Journal of Chemical Engineering, 93(2), 356–362, 2014.

Tuna P., Hulteberg C. and Ahlgren S., ‘Techno-Economic Assessment of Nonfossil Ammonia
Production’, Environmental Progress and Sustainable Energy, 33, 1390-1297, 2014.

Turton R., Analysis, Synthesis, and Design of Chemical Processes, 3rd Edition, Prentice Hall,
2009.

UNFCCC. Conference of the Parties (COP), Adoption of the Paris Agreement. Proposal by the
President, Paris Climate Change Conference, COP 21, 2015.

UNIDO, International Fertilizer Development Center. Fertilizer Manual, Springer; 1998.

Utikar R. P. and Ranade V. V., 'Single jet fluidized beds: Experiments and CFD simulations with
glass and polypropylene particles', Chemical Engineering Science, 62, 167–183, 2007.

US Geological Survey, Mineral Commodity Summaries 2016, US Geological Survey 202, 2016.

Voet E., Lifset R. J. and Luo L., 'Life-cycle assessment of biofuels, convergence and divergence.'
Biofuels, 1, 435–449, 2010.

Walter A., Galdos M.V., Scarpare F.V., Leal M.R.L.V., Seabra J.E.A., da Cunha M.P., Picoli
M.C.A. and de Oliveira C.O.F., ‘Brazilian sugarcane ethanol: Developments so far and challenges
for the future’, Wiley Interdisciplinary Reviews: Energy and Environment, 3(1), 70–92, 2014.

Wang B., Gebreslassie B.H. and You F., ‘Sustainable design and synthesis of hydrocarbon
biorefinery via gasification pathway: Integrated life cycle assessment and technoeconomic
analysis with multiobjective superstructure optimization’, Computers and Chemical Engineering,
52, 55–76, 2013.

141
Weidema B.P., Bauer Ch., Hischier R., Mutel Ch., Nemecek T. Reinhard J., Vadenbo C.O., and
Wernet, G, The ecoinvent database: Overview and methodology, Data quality guideline for the
ecoinvent database version 3, 2013.

Wilk V. and Hofbauer H., ‘Co-gasification of plastics and biomass in a dual fluidized-bed steam
gasifier: Possible interactions of fuels’, Energy and Fuels, 27(6), 3261–3273, 2013.

Wiloso E.I., Heijungs R. and De Snoo G.R., ‘LCA of second generation bioethanol: A review and
some issues to be resolved for good LCA practice,’ Renewable and Sustainable Energy Reviews,
16(7), 5295–5308, 2012.

You F., Tao L., Graziano D.J. and Snyder S.W, ‘Optimal design of sustainable cellulosic biofuel
supply chains: Multiobjective optimization coupled with life cycle assessment and input–output
analysis’, AIChE Journal, 58(4), 1157–1180, 2012.

Zanzi R., Sjöström K. and Björnbom E., ‘Rapid pyrolysis of agricultural residues at high
temperature.’ Biomass and Bioenergy, 23(5), 357-366, 2002.

142
List of publications
 Arora, P., Hoadley, A. F. A., Mahajani, S., Ganesh, A., 2017. Multi-Objective
Optimization of biomass based ammonia production: Potential and perspective in different
countries, Journal of Cleaner Production, 148, 363-374.
 Arora, P., Hoadley, A. F. A., Mahajani, S., Ganesh, A., 2017. Compartment model for a
dual fluidized bed biomass gasifier. Chemical Engineering Research and Design, 117, 274-
286.
 Arora, P., Hoadley, A. F. A., Mahajani, S., Ganesh, A., 2016. Small scale ammonia
production from biomass: A techno-economic perspective. Industrial & Engineering
Chemistry Research, 55(22), 6422-6434.
 Arora, P., Hoadley, A. F. A., Mahajani, S., Ganesh, A., 2016. Multi-level Modelling of
Sustainable Chemical Production; From CFD to LCA. Computer Aided Chemical
Engineering, 38, 499-504.
 Arora, P., Hoadley, A. F. A., Mahajani, S., Ganesh, A., 2015. Modelling and Optimisation
of Dual Fluidisation Bed Gasifiers for Production of Chemicals. Chemical Engineering
Transactions, 45, 1111.
 Sharma, I., Arora, P., Hoadley, A. F. A., Mahajani, S., Ganesh, A., 2016. Greener
Alternatives for Ammonia Production. In Proceedings of ECOS 2016, Slovenia.
 Arora, P., Hoadley, A. F. A., Mahajani, S., Ganesh, A., 2015. Residence Time Distribution
Study for Bubbling Fluidised Beds using Computational Fluid Dynamics, In Proceedings
of APCChE 2015 Congress and Chemeca 2015, Melboune.
 Arora, P., Hoadley, A. F. A., Mahajani, S., Ganesh, A., 2014. Innovative Method of
Production of Ammonia from Biomass. In Proceedings of 22nd European Biomass
Conference and Exhibition, Hamburg, 1078 – 1085.

143
Acknowledgments
I would like to express my gratitude and thanks to my supervisors; Prof. Andrew Hoadley, Prof.
Sanjay Mahajani and Prof. Anuradda Ganesh, for their constant guidance and supervision. During
the highs and lows of my PhD journey, they have always been there to provide the much needed
support. I sincerely acknowledge the constructive suggestions provided by Prof. Santanu
Bandyopadhyay, Prof. Anand B. Rao, Prof. Yogendra Shastri, Prof. Sankar Bhattacharya and Prof.
Akshat Tanksale during the annual research progress committee meetings. I also Prof. Vivek
Ranade (NCL), Dr. Trent Harkins (Monash) and Dr. Laurence Tock (EPFL) for providing me
guidance at different junctions of my PhD journey.

I would also like to acknowledge the financial and academic support provided by Orica Ltd.
through the IITB-Monash research academy (Project ID: IMURA 0221 (A)). In particular, I would
like to acknowledge the support from Mr. Ajay Joshi, Dr. Greg Rigby and Dr. Milinda Ranasinghe
at Orica.

I am thankful to the staff at IITB-Monash research academy namely; Prof. Mohan


Krishnamoorthy, Prof. Murali Sastry, Anasuya ma’am, Kuheli ma’am, Mamta ma’am, Nancy
ma’am, Sheba, Krishna Warrier, Jayasree, Laya, Priyanka, Kiran, Rahul Krishna, Rahul and
Bharat, for their timely and much needed assistance. I would like to thank Adrian Gertler, Mrs. Jill
Crisfield and Mrs. Lilyanne Price for helping me during my stay at Monash.

I am also thankful to my lab mates; Ishan, Bhoja, Yughabala, Sravan, Ganesh, Praveen, Nilam,
Sonal, Kapish, Sanchit, Sminu, Sudip, Rohidas and Detke; for all the help and support in my PhD
journey. I am also thankful to my friends Sushant, Nirmal, Indrajit, Karan, Sanket, Sravan,
Mayank, Nitin, Tamal, Tushar, Niraj, Ganesh and Rohit for the eventful time I had during my stay
at Bombay. Last but not the least; I would like to thank my parents for the never ending stream of
encouragement and love that I was fortunate enough to receive

Pratham Arora

144