You are on page 1of 74

The design of a bioenergy potential map and a

GIS-enabled optimization model for biofuel


production in India
by
Yan Yan
Submitted to the Department of Mechanical Engineering
in partial fulfillment of the requirements for the degree of

Master of Science in Mechanical Engineering

at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY

June 2019
Massachusetts Institute of Technology 2019. All rights reserved.

Signature redacted
Author ................................................ .......
Department of Mechanical Engineering
May 19, 2019

Signature redacted
C ertified by ............................
Ahmed Ghoniem
Professor
Thesis Supervisor
Signature redacted
Accepted by.
MASSACHUSETTS INSTITUTF Nicholas Iadjiconsantitou
OF TECHNOLOGY
Chairman, Department Committee on Graduate Theses
JUN 13 2019
LIBRARIES
ARCHIVES
2
The design of a bioenergy potential map and a GIS-enabled
optimization model for biofuel production in India
by
Yan Yan

Submitted to the Department of Mechanical Engineering


on May 19, 2019, in partial fulfillment of the
requirements for the degree of
Master of Science in Mechanical Engineering

Abstract
India has a remarkably large inventory of biomass, but the operation of biorefiner-
ies has been difficult due to the uncertainty in biomass availability and the lack of
biomass supply chain management in the industry. To help biorefinery operators make
strategic decisions, this study develops a model that optimizes and evaluates the eco-
nomic viability of bioenergy/biofuel production technologies in India. To enhance the
accuracy of the simulation, bioenergy potential is mapped out using geospatial infor-
mation system, and biomass transportation distances are calculated using geospatial
network analysis. The optimization model is built as a mixed integer linear program-
ming problem to solve for the optimal locations, number and scale of the biorefinery,
and biomass flows to it. Biosynthetic natural gas (bioSNG) production technology,
GoBiGas, is used to test the model. The levelized cost of bioSNG and the ensuing
cost sensitivity analysis concludes that this particular technology configuration is not
sustainably profitable in India, but a less capital intensive technology might be able
to demonstrate otherwise.

Thesis Supervisor: Ahmed Ghoniem


Title: Professor

3
4
Acknowledgments
This work would have not been possible without the funding and support from MIT
Tata center for Technology and Design. I want to thank all Tata center staff, mentors
and students for adding so much to my MIT experience.
I want to thank Professor Ghoniem for always being patient and supportive, pro-
viding me insightful guidance and the flexibility of defining my own research topic.
I also want to extend my gratitude to every member of RGD lab. I could have not
asked for a better group of colleagues, friends and mentors.
I also want to thank all the staff at MIT GIS lab for providing me tremendous
help in collecting and processing my data.
Finally I want to thank my family for their unconditional love and support for
me.

5
6

-Im T-- "M


Contents

1 Introduction 15
1.1 India bioenergy outlook and challenges . . . . . . . . . . . . . . . 15

.
1.2 Key steps of biorefinery supply chain . . . . . . . . . . . . . . . . 18

.
1.2.1 Harvesting and collection . . . . . . . . . . . . . . . . . . . 18

.
1.2.2 Preprocessing . . . . . . . . . . . . . . . . . . . . . . . . . 18

.
1.2.3 Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
.
1.2.4 Transportation . . . . . . . . . . . . . . . . . . . . . . . . 19
.

1.3 Optimization modeling for biorefinery operation . . . . . . . . . . . 20

2 Bioenergy potential mapping in India 23


2.1 Agricultural residues mapping . . . . . . . . . 24
.

2.2 Animal manure mapping . . . . . . . . . . . . 26


.

2.3 Municipal solid waste mapping . . . . . . . . 29


.

2.4 Mapping result summary . . . . . . . . . . . . 31


.

3 Optimization of biofuel production in Punjab 33


3.1 Agricultural residue in Punjab . . . . . . . . . 33
.

3.2 Biomass supply chain model . . . . . . . . . . 34


.

3.3 Indirect gasification for bio-synthetic natural gas (bioSNG) production 36


3.3.1 GoBiGas system . . . . . . . . . . . . 36
.

3.3.2 Capital and Operating costs of GoBiGas 38


3.4 M ethodology . . . . . . . . . . . . . . . . . . 43
.

3.4.1 Base maps generation . . . . . . . . . . 44


.

7
3.4.2 GIS origin-destination algorithm . . . . . . . . . . . . . . . . . 50
3.4.3 Optimization formulation . . . . . . . . . . . . . . . . . . . . 50

4 Results 55
4.1 Total cost breakdown . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Environmental impact . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3 Levelized cost of bioSNG . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.3.1 Future Indian natural gas market price projection . . . . . . . 59
4.3.2 Levelized gas cost calculation . . . . . . . . . . . . . . . . . . 60
4.4 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

5 Conclusion and future work 67

A Tables 69

8
List of Figures

1-1 India's total installed power capacity by source in 2019 . . . . . . . . 16

1-2 Annual emissions of different pollutants in India (left), and the contri-
butions of different activities toward annual emission levels(right) . . 17

1-3 Installed biomass power generation capacity in India from 2008 to 2017
[1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

1-4 Biomass transportation costs using different transportation media [2] 20

1-5 Publications on biomass supply chain optimization from 1997 to 2018


by different regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2-1 The user interface of NREL Biofuels Atals . . . . . . . . . . . . . . . 23

2-2 Biomass power generation potential map by MNRE Biomass Knowl-


edge Portal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2-3 Percentages of utilized and burned biomass surplus in India [3] . . . . 25

2-4 Mapping of agricultural residue based bioenergy . . . . . . . . . . . . 26

2-5 Animal manure based bioenergy production and distribution in India.


The energy potential of a. all the manure products (GJ), b. cattle
manure(MJ) , c. chicken manure (MJ), d. sheep manure (MJ), e. pig
manure (MJ), f. goat manure (MJ) . . . . . . . . . . . . . . . . . . . 29

2-6 Indian's municipal solid waste composition . . . . . . . . . . . . . . . 30

2-7 Energy potential of annual municipal solid waste produced in India in


m egajoule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

9
2-8 Type of biomass waste that has higher energy potential at each district.
Yellow regions have higher livestock manure ; blue regions have higher
municipal solid waste production, and green regions have higher burned
agricultural residues . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3-1 Locations and the amount of burned crop residue in Punjab, 2018 . 34
3-2 Year-round PM2 .5 concentration in some major cities in India 14] 34
3-3 The cost of collection, processing and storing per ton of straw-based
agricultural residue . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3-4 Gap between natural gas production and demand in India from 2000
to 2016 [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3-5 Illustration of the flows in the gasification section of the GoBiGas plant
of the bed material (blue), product gas ash (green), and coarse flue gas
ash (orange) [6] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3-6 Overall system layout of GoBiGas [7] . . . . . . . . . . . . . . . . . . 38
3-7 Linear piecewise breakdown of GoBiGas capital and operating costs . 40
3-8 Capital cost breakdown of GoBiGas in India and Sweden . . . . . . . 42
3-9 Annual operational cost breakdown of GoBiGas in India and Sweden 43
3-10 GIS-enabled optimization model building processes . . . . . . . . . . 43
3-11 Agricultural residue supply spots that are proximate to the road network 44
3-12 Different classes of roads included in OpenStreetMap data . . . . . . 45
3-13 Road network adopted for transporting biomass in Punjab . . . . . . 45
3-14 Suitability filtering maps for determining optimal biomass storage cen-
ters: a.regions with burned biomass, b. rice plantation area, c. prox-
imity to road network, d. proximity to biomass suppliers, e. land cover
m ap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3-15 Final suitability mapping products: a. compiled suitability map, b.
regions with suitability score greater than 39, c. suitable locations
that have area greater than 5000 m 2. . . . . . . . . . . . . . . . . . . 48
3-16 Potential bioSNG plant locations (highlighted in blue) in Punjab . . 49

10
3-17 Authorized city gas distribution regions in Punjab (highlighted in blue)
and their urban population density . . . . . . . . . . . . . . . . . . . 50

4-1 Locations of optimal biomass suppliers, storage centers and bioSNG


plan t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4-2 Percentage of each cost component towards the total cost . . . . . . . 56
4-3 Emissions from bioSNG production in comparison to burning biomass
in the field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4-4 European Union Allowance (EUA) price from 2017 December to 2022
D ecember [8] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4-5 World natural gas prices . . . . . . . . . . . . . . . . . . . . . . . . . 60
4-6 Future India domestic natural gas price projection . . . . . . . . . . . 60
4-7 Contributions of natural gas market price, carbon credit and tariff
towards the levelized costs of bioSNG as the plant's lifetime increases 62
4-8 Capital cost curves generated from varying reactor system cost and
auxiliary equipment and service cost . . . . . . . . . . . . . . . . . . 64
4-9 Change in levelized cost of bioSNG attributed to changes in capital
cost and dem and . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

11
12
List of Tables

2.1 Predictor variables that have direct impact on livestock population [9] 27
2.2 The total annual energy potential derived from different types of biomass 32

3.1 Capital costs and scale factors of a 20 MW GoBiGas plant . . . . . . 39


3.2 operating costs and scale factors of the 20 MW GoBiGas plant . . . 39
3.3 Cost distinctions between Indian and Sweden power plants . . . . . . 42

4.1 Breakdown down of the total cost . . . . . . . . . . . . . . . . . . . . 56


4.2 Emission of pollutants from open field straw burning [10] . . . . . . . 57
4.3 Emissions from the process of bioSNG production [11] . . . . . . . . . 58

A. 1 Investment costs and scale factors for all process sections of GoBiGas
plan t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
A.2 Auxillary equipment, project costs and their corresponding scale fac-
tors of GoBiGas plant . . . . . . . . . . . . . . . . . . . . . . . . . . 70

13
14
Chapter 1

Introduction

Mapping out bioenergy potential and minimizing bioenergy/biofuel production costs


are crucial for unlocking the great energy potential of biomass waste in India. As
such, this study strives to achieve these objectives by developing a visualization tool
of the biomass distribution in India using the geographical information system, and
then creating a mixed integer linear programming (MILP) optimization model to
determine the optimal locations, scales and numbers of biomass suppliers, storage
centers and biorefineries.
This first chapter introduces the background and motivation of the project. Chap-
ter two describes the design process of the bioenergy potential mapping tool. Chapter
three and four demonstrate the formulation and testing of the optimization model.
The last chapter concludes the project and explains how it can be improved and
extended in the future.

1.1 India bioenergy outlook and challenges

The large population base of India makes its energy consumption grow rapidly. It is
estimated that the country will overtake China as the world's largest energy market
by mid 2020s 112]. Conventional or fossil fuel resources are limited, so the government
has started to focus more on the development of renewable energy. As for March 2018,
the total renewable energy installed power capacity in India was 69 GW, consisting

15
mostly of wind and solar power [13], as seen in figure 1-1. Such an energy landscape
indicates there is still great renewable energy potential awaiting to be explored in the
country.

Nuclear

Wind 50%

Solar 30%

Biomass
15%
L

Small Hydro 5%

Figure 1-1: India's total installed power capacity by source in 2019

Although biomass energy only accounts for 15% of the total installed renewable
power capacity, it plays an important role in India's energy market. Around 80% of
the energy consumed in rural communities for cooking and heating is from burning
biomass such as fuel wood, crop residues, and livestock manure [14]. However, this
traditional way of utilizing biomass is neither efficient nor environmental friendly.
As shown in figure 1-2, traditional biofuel burning is a major cause of emissions,
making India the third largest CO 2 emitter in the world [15] [16]. A better way to
utilize biomass is to convert it into heat, electricity,and biofuels using new thermal,
thermochemical or biochemical conversion technologies [17]. Carefully studied and
controlled biomass reactions in these systems ensure a much more efficient and cleaner
biomass conversion.
After realizing that converting biomass into clean biopower and biofuel can both
fulfill the growing energy demand, and reduce GHG emissions, the government started
to dedicate a significant amount of financial and political support to bioenergy and
biofuel development. In 2018, the Ministry of New and Renewable Energy (MNRE)
approved Rs.170 Crores (24 million USD) to assist biomass based cogeneration projects
[18], and set the target of reaching 20% blending of bioethanol in petrol and 5% blend-

16
S02
Pollutants Total emissions (Gg/yr)
CO
SO 2 4170
CO 62,000 C02
CO 2 1,860,000 PM2.5
PM 25 4290
OM
OM (organic matter) 2120 0
OC (organic carbon) 1150 0C
BC (black carbon) 400 BC
IOM (inorganic oxidized matter) 1765 1GM

0% 20% 40% 60% 80% 100%


relative contribution(%) of different sources
m traditional biofuclsa *fossil fuels u forcst burning u crop waste open buming

Figure 1-2: Annual emissions of different pollutants in India (left), and the contribu-
tions of different activities toward annual emission levels(right)

ing of biodiesel in diesel by 2030 [191. With these incentives, the installed biomass
power generation capacity increased to 9.5 GW in 2017 [11, as shown in figure 1-
3. However, the sustainable operation of these installed biomass power/cogeneration
plants have been challenging. According to Natarajan and Pelkonen's report, more
than 20% (1200MW) of the total installed biomass power generating capacity in
India was either temporarily or permanently shut down in 2015 [20].

12000 - - - - - - - - - - --............................................................................

10000 - - - - - - - - 9.533-
--.......................................................................
9,024

6000 - -....................................................... 5,606 - -. -.


5,148

4,01 . 4,280

2,453

2 000 1.-.-- -. . . . . --.. ---


-

200 2010 2011 2012 2013 2014 2015 2016 2017

Figure 1-3: Installed biomass power generation capacity in India from 2008 to 2017
[1]

Unsecured biomass supply and unorganized supply chains are two biggest chal-
lenges to the sustainability and profitability of biorefineries. Agricultural based

17
biomass is often only produced in certain time periods, but biomass plants have
to operate throughout the year. If reliable biomass supply is not secured at the
beginning of the new cropping season, the plant is likely to fail to operate at full
scale in the coming year. The locations of suppliers are also critical. The low energy
density and scattered distribution of biomass makes its transportation costly. If not
planned strategically, the cost of transportation alone can make the whole operation
unprofitable.

1.2 Key steps of biorefinery supply chain

The supply chain of biomass involves various activities such as harvesting, handling,
baling, transport, preprocessing, storage, biofuel production and distribution to de-
mand zones. The subsections below detail some main steps of biomass supply chain.

1.2.1 Harvesting and collection

The collection of agriwaste is more challenging than other types of biomass, because it
is strictly limited to a short window during which the crop seasons transition. During
that period of time, farmers are often occupied with crop harvesting and don't want to
participating in collecting the residues. Therefore it is important to recruit personnel
to perform this task and train them to use the right harvesting tool, e.g. baler for
rice straw and shredder for cotton stalks.

1.2.2 Preprocessing

To reduce the cost of transportation, biomass is often preprocessed to enhance its


energy density and the efficiency of the following conversion process. There are dif-
ferent types of preprocessing approaches. Ensiling is a thermal-chemical process that
produces silage for the following anaerobic digestion process. Pelletization is the
process of compressing and molding biomass into the form of a pellet. Agricultural
residues are often pelletized before being transported to a gasification plant. Tor-

18
refaction is a mild thermal treatment of biomass at a temperature around 200-300
Co. After torrefaction, biomass turns into a charcoal form [21]. The low moisture
content and high carbon density of torrefied biomass can greatly extends its shelf-life
and transportability.

1.2.3 Storage

Biomass is often stored at a collection storage center until it is needed at the plant. To
reduce mass loss and control the quality of stored biomass, it is favorable to establish
sheltered storage at locations that have humid climate. The size and location of the
storage center is determined by the scale of the plant and biomass distribution in the
region.

1.2.4 Transportation

Several transportation modes can be employed to transport biomass, such as roads,


railroads, waterways or a combination of several. A previous study has demonstrated
that transportation using a barge is the most cost efficient method, as shown in fig-
ure 1-4, but it is limited to areas that have access to the waterway [21. For inland
regions, railroad or road transportation are more practical. Railroad companies often
charge tariffs on their customers for utilizing the facility. This is why the trans-
portation costs of using a single or multiple rail cars start at nonzero in figure 1-4.
Although the cost of using the railroad is higher than using the road at the ini-
tial stage, the lower per ton-kilometer cost of railroad transportation makes it more
economical as the transportation distance increases. Despite of the cost efficiency
of railroad transportation, the limited accessibility and network extension makes it
less flexible than road transportation. In other words, there is not a transportation
method that works well everywhere. The local environment and the existing network
infrastructure determine which transportation mode would be optimal at a specific
location.

19
40

-
35 .Or_
_
. 30
225
20
1
0 In
91

04
0 100 200 300 400 500
Distance Traveled (in km)
-.- Truck -- CSXT: Single Car
-a- CSXT: Unit Train 65 Cars
-++- CSXT: Unit Train 90 Cars -*- Barge

Figure 1-4: Biomass transportation costs using different transportation media [2]

1.3 Optimization modeling for biorefinery operation

The challenging economic viability of bioenergy systems has attracted a growing num-
ber of researchers to look into this field. Among different types of models created for
optimizing the biorefinery operation, the majority of models focus on minimizing the
production cost of bioenergy or biofuel, while the rest explore how to minimize the
system's social and environmental footprints, or how to fulfill different criteria si-
multaneously. [21]. Mathematical programming is the most common and effective
approach for building an optimizing model. Depending on the characteristics of opti-
mization variables, objectives and constraints, models can be categorized into linear

programming, integer programming, mixed integer linear programming and nonlinear

programming. As most of real-world problems are complicated nonlinear problems,


linear programming is not applicable in most situations. On the other hand, a non-
linear model is a more accurate representation, but is much more difficult to solve.
Fortunately, some nonlinear problems can be broken down into linear pieces and
solved as a mixed integer linear programming (MILP) problem which is allowed to
have a mixture of integer and non-integer variables. The binary integer variables of
a MILP model are used to make decisions regarding facility scale and location, while

20
8

26
05

R4
C4 -4

.3

0)If611 I II IflIII I\
If)"""P1
1"
11\iili1
w US * Canada n Europe m India m Other

Figure 1-5: Publications on biomass supply chain optimization from 1997 to 2018 by
different regions

non-integer variables indicate the biomass flow. To further enhance the accuracy of
biomass transportation route and distance, the eographical information system (GIS)
is used to incorporate real-life geospatial information, such as the road and railroad
network, land cover and population into the model [221.
The development and adoption of biorefinery supply chain and operation opti-
mization models is lacking in India. As shown in figure 1-5, among 99 publications
on biomass supply chain design and optimization published between 1997 to 2018,
only 4% of them are based on Indian cases, while the majority of the rest are based
in Europe and North America. Therefore, the goal of this study is to fill this void in
the research and practice of optimizing biorefinery profitability in India.

21
22
Chapter 2

Bioenergy potential mapping in India

Being able to visualize the production and distribution of biomass is the first step
toward planning and optimizing bioenergy/biofuel production. Many countries with
large biomass inventory have dedicated national research resources on mapping out
biomass distribution in their national territories, such as the Biofuels Atlas devel-
oped by NREL and Biomass Inventory Mapping and Analysis Tool developed by
the Government of Canada. The user interface of NREL's Biofuel Atlas is shown in
figure 2-1.

T..OR AllL.06.
I U' -"1"-
Fedcks

somUUU A
*

sw AS

BItion-ton S6UdY 2016 446 ~


IoErvagy Sites

Neureal Go,

06tW Pow Plat


Atternatlw Fual Slati
Vehicle Dsrofy
lefrelatwWur and lloundarla

Figure 2-1: The user interface of NREL Biofuels Atals

Figure 2-2 is a biomass power potential map from the Biomass Knowledge Portal
website created by the MNRE. Its resolution, however, is too coarse to be utilized for

23
intrastate or district-level biorefinery operation planning. Therefore, in this chapter,
a more refined and comprehensive visualization tool of bioenergy potential in India
is introduced.

LNOJ ASMIL

L A
RADESEH t

A ALAND

L P KER. AD NICOBAR I ON

M2 I1 to -A

~oiteoU
--tot
2
s

to 1
a1
toa 0

Figure 2-2: Biomass power generation potential map by MNRE Biomass Knowledge
Portal

2.1 Agricultural residues mapping


India has the second largest agricultural-based economy with year-round cultivation
in the world. According to the World Bank data, close to 60% of the land in India is
under agricultural use, with rice and wheat as the two most prevalent crops [231. The
majority of agricultural residues are utilized for different aspects of farming, such as
being used as animal fodder, manure and cooking fuels, but the surplus biomass in
the country can still mount to 234 million ton/year. The estimated annual bioenergy
potential from these surplus biomass is 3.74 EJ, equivalent to 12% of India's primary
energy consumption in 2016. However, most of these biomass had not been utilized

24
efficiently. 64% of them were burned in the field according to the data collected in
2014 13], as shown in figure 2-3.

64% a utilized crop residue


721 28% - burned crop residue surplus

U36% ,iuhurned crop residue surplus

Figure 2-3: Percentages of utilized and burned bioniass surplus in India [3]

The distribution of these burned agricultural residues is mapped out using re-
motely sensed terrestrial observations from NASA's Moderate Resolution Imaging
Spectroradiometer (MODIS) instruments. NASA MODIS instruments have been used
extensively for generating a suite of global land, ocean and atmospheric dataset for
science community [24]. Among this suite, MCD45A1 and MCD64A1 burned area
products map out the spatial extent and the approximate date of biomass burning.
To improve the accuracy and reliability of the burning data, especially active fires
at a small scale, scientists reprocess the data into different collections using refined
algorithms. The most recent burned area product collection is generated using the
most updated burned area mapping algorithm by the Giglio et al , and it is used to
reprocess the raw burning area data in this study 124].
After generating the burning area map, the production of crops on per kilometer
of land and the ratio between crop grain and residue productions are used to convert
the burning area to the amount of agricultural residue produced [25]. Because the
majority of burned residues are rice and wheat, the residue production rates of these
two crops are adopted at all the points.
The final product is configured into an interactive website application using the
software ArcGIS online. The interface of final product is shown in figure 2-4. Each
dot on the map represents a burning spot. The size of the circle indicates the amount

25
of bioenergy that could be generated from the local residues. The time scale bar
on the bottom of the map allows users to scroll through the year and observe the
bioenergy production and distribution in each month.

Inar biommofass wse idannual podulcion addsrbtinm pV e r

c.*-rsd -e ft C-" whk InsnsnsWorM- "te W4 e A* F'


-

VWL fOALPlyOPH ,ML.adNMC, 0, 11 M&WG a

Ghanbhd

-'t61r. :7 KISTAN F

- ISA CIF N-

FERE. Es. Ga.min F

Figure 2-4: Mapping of agricultural residue based bioenergy

According to the map, the highest burning rate occurs in May and October,
corresponding to the transition periods between Kharif and Rabi cropping seasons in
India, and most of the burning takes place in northern and middle regions of India.

2.2 Animal manure mapping

Aside from having high production rate of crops, India also has a large livestock
population. According the 19th India Livestock Census, the total livestock population
in the country is more than 512 million, with its bovine and buffalo populations as the
highest in the world, and goat and sheep populations coming second and third (141.

26
Unlike crops, livestock population cannot be detected by remote satellite sensing, so
the raw data for animal manure mapping is from the Indian livestock census data.
The resolution of the available census data only goes up to district level, so a series
of reprocessing steps are conducted.

i Create a suitability map to filter out regions unsuitable for raising


livestock
Certain regions on the map are unsuitable for raising livestock, such as permanent
water bodies and populated urban areas, so a suitability map is created using
MODIS land cover and population density map to filter out these regions. Then
population densities of livestock are calculated by dividing the census data with
suitability-corrected area.

ii Identify predictor variables that have impact on livestock population


density
Certain environmental factors have direct impact on livestock population densi-
ties, as described by Robinson et. al [9]. These factors are defined as predictor
variables and are listed out in Table ii.

Table 2.1: Predictor variables that have direct impact on livestock population [91
Anthropogenic Human population density (consensus model between Worldpop, Landscan and GPW4)
Travel time to cities of 50,000 people
Topography Elevation(GTOPO30)
Slope(GTOPO30)
Vegetation 10 Fourier-derived variables from Normalized Difference Vegetation Index from MODIS
Length of growing period
Green-up and senescence (annual cycle 1 and 2)
Cropping intensity
Forest cover
Climate 14 Fourier-derived variables from Day Land Surface Temperature (MODIS)
Precipitation

iii Choose sampling points with a balance of data coverage and quality
The correlation between existing livestock density and predictor variables can be
learned by a machine learning model to predict the population density at a smaller
scale. To achieve this, a group of sampling points are selected from the available
dataset with a balance between high coverage and preference for points with more

27
detailed census data. The same selection approach is adopted by Nicolas et. al
[26]. The predictor variable values of all the sampling points are recorded, then
70% of the sampling points are categorized as the training data, while the rest
30% as testing data.

iv Train and test random forest model using sampling points


The training and testing sampling points are input into a random forest model
built using the random forest package in R. Random forest is an ensemble learning
method that is used to perform categorization or regression analysis on large
datasets. Compared to other traditional regression models, this method has the
advantage of avoiding overfitting [27].

v Predict population densities at all grid cells


The random forest model trained using sampling points are used to predict the
population at each 100 km 2 grid cell based on the values of predictor variables at
each cell.

vi Convert livestock population density to bioenergy


To convert the livestock population densities to bioenergy potential, the density
data is multiplied with the production rates and net calorific values of animal
manures provided by ECN.TNO biomass and waste database [28].

The distributions of bioenergy extracted from different types of livestock manures


are shown in figure 2-5.

28
a. b.

Legsd Legend
bea,_Nvestvck_cowMbed
.ss.NVGJ "Ol_maftre-vMJM
b e.-- IVeA e
aC... 0 -. T-0.. ,cmb+n.

a %5.o
.0.. -,
7 0.03
4-g+. .nj....-V.3.

.2.M 24-0..,

120...5.52..

- . 3- 7.-6
- 5.5 - 0..W6
515-06-42-056

m she~nm"e~VR
-S5.7-3.47

C. 'I d.
6~

Legend
he"e_ 56ces.33ed
A
~ ,-
3..suesc...mn.eV.MJ
... 0 -725.3 &..4. -tI.."
248.hc.es30

.-- ,
-5.47 .15.00
31.48 - 56-0

e. f.

~1%
Le..W 2 153..3sLj..,Gs.JVJWJ
-

- 053. 4 91- - 5.40 -1-4-0

335. 515 5.3-d 215f 41 13556.41


-555-5323-a 31147-51".W
n55.o5 55..3 &J-0 514- t35.
m -, 13.72055.3
212-103 -32435 I 15.48-23.43
. 33-.3 650,175 Z15..02630.3
035.3-tat-"

Figure 2-5: Animal manure based bioenergy production and distribution in India.
The energy potential of a. all the manure products (GJ), b. cattle manure(MJ) , c.
chicken manure (MJ), d. sheep manure (MJ), e. pig manure (MJ), f. goat manure
(Mi)

2.3 Municipal solid waste mapping

As one of the most populated countries, India produces a large quantity of municipal
solid waste annually. As shown in figure 2-6, 58% of municipal solid waste produced
in India is biomass based (food, green and wood). Therefore, it is mapped out as a
type of biomass waste in this study.

29
m Food and green
1% 15% * Glass
2%
2 Metal
8% . Paper and carboard
m Plastic
10% * Rubber and leather
m Wood
3% * Other
4%

Figure 2-6: Indian's municipal solid waste composition

The random forest learning method that was used to predict the livestock pop-

ulation distribution in the previous subsection can also be used to predict human
population density, and this task was completed by an open source spatial demo-

graphic data platform called World Pop. The municipal solid waste production is

then predicted by multiplying the population density with waste production per day

per capita data provided by the World Bank. According to the World Bank's global
waste management report, the production rate in India was 0.57 kg/capita/day in
2018, and this number is predicted to increase to 0.7 kg/capita/day by 2025 [29].

The compiled MSW distribution map is shown in figure 2-7. Darker areas on the
map indicate regions with higher MSW energy potential. Those are also the re-

gions with high population densities, such as the top three populated states in the

country: Uttar Pradesh (828people/km2), Bihar (1102people/km 2 ), and West Bengal


(1029people/km 2).

30
P;L AAU 0 1 rI

.41
* I
N3ANW
M" WM

4, 7TAf

,0

U~~N I ~ I~ NOM.
WI PAO VM

Figure 2-7: Energy potential of annual municipal solid waste produced in India in
megajoule

2.4 Mapping result summary

The bioenergy potential of burned agriwaste, animal manure, and municipal solid
waste are shown in Table 2.4. Figure 2-8 shows which type of bioenergy has the
higher potential in each district. According to the map, the majority of districts
have higher animal or municipal solid waste than burned agricultural residues, which
explains the booming interest in anaerobic digestion to produce biogas and bioCNG
at household and industrial scales. Punjab is the only state that has higher burned
agriculture residue than two types of wastes, making it an ideal region to study
the optimization of biofuel production that uses agricultural residues, which will be
discussed full-fledgedly in the next chapter.

31
Table 2.2: The total annual energy potential derived from different types of biomass
Type of biomass waste Energy potential (TJ)
Burned agriwaste 7 x 105
Livestock manure 8.3 x 106
Municipal solid waste 2.4 x 106

WM WW"" MWOAMb M
"M .w

-7
KI
STAN

41
Th,

ti

Figure 2-8: Type of biomass waste that has higher energy potential at each district.
Yellow regions have higher livestock manure ; blue regions have higher municipal solid
waste production, and green regions have higher burned agricultural residues

32
Chapter 3

Optimization of biofuel production in


Punjab

3.1 Agricultural residue in Punjab


According to the national policy for management of crop residue, the crop residue
production in Punjab was 51 Mt in 2018, ranked number 2 in the country [30]. Rice
and wheat are two dominant crops in the area, so more than 80% of the crop residue
are cereal based. Due to the fast crop season rotation, many farmers choose to burn
rice and wheat stubbles in the field after harvesting.
Based on Jain et al. and the Intergovernmental Panel on Climate Change (IPCC)
coefficient, more than 40% of the paddy straw is burned annually, making Punjab the
second largest carbon emitting state in India [31]. The haze caused by the burning
severely pollute the air in nearby regions. The national capital, Delhi, which is in
proximity to Punjab, recorded an annual average of 98 pm/m 3 of PM2 .5 concentra-
tion, which is twice as much as the Indian standard and ten times as much as the
WHO standard [32]. Up to 20% of the particulate concentration is caused by the
agricultural residue burning, as shown in figure 3-2.
If these burned agricultural residue in Punjab could be efficiently collected and
converted into energy, both waste management and air pollution issues could be
alleviated.

33
Punjab burning spots
with straw weight (kg)
strawamt
-A-
1139 - 164861
* 164862 - 362621
A,
* 362622 - 655113
't 4 * 655114 - 1834060
* 1834061 -5364110
* 5364111 - 12401200
* 12401201 - 23618700

Figure 3-1: Locations and the amount of burned crop residue in Punjab, 2018

200
180 'i very unhealthy

160
140 * unhealthy
120

100

80

60
16%
I unhealthy for
sensitive
groups

., 16% * moderate
40
7%/a 13%
14% 18%
20
* good
0
Delhi Delhi Delhi Fall Delhi Kolkata Kolkata Kolkata Fall Kolkata
Spring Summer Winter Spring Summer Winter

MDiesel N Gasoline 1 Road Dust OCoal * Biomass burning -] Other

Figure 3-2: Year-round PM2 .5 concentration in some major cities in India [4]
3.2 Biomass supply chain model
The low energy density of biomass has always been a big hurdle to cost efficient
biomass supply. In India, the challenge is heightened by the compact timing of crop
34
harvesting and planting. If a company wants to collect enough biomass for a year-
round plant operation, the supply chain of biomass should be carefully planned out
ahead the harvesting season. In order to ensure the supply chain model created
in this study is practical in Indian communities, the operation model of a biomass
management company in Punjab (PRESPL) is studied and replicated.

In this model, the identification of biomass suppliers and the recruitment of labor
initiate 4-5 months before the actual harvesting season. Locals are employed and
trained to be biomass collectors. Meanwhile, storage and preprocessing centers are
built based upon the locations of biomass suppliers and potential plant locations.
Once harvesting takes place, employers cut and bale straws, then transport them to
storage centers. At a storage center, the straws will be further cut down and densified.
The preprocessed biomass is transported to the plant for the rest of the year. The
costs of collecting, processing and storing per ton of straw are shown in figure 3-3

100

10
1
W1

0 t 0.1
I-I

0.01
op C

Figure 3-3: The cost of collection, processing and storing per ton of straw-based
agricultural residue

35
3.3 Indirect gasification for bio-synthetic natural gas
(bioSNG) production

As biofuel production technologies growing mature, a variety of biofuels are entering


the world energy market. Among all the options, bio-synthetic natural gas (BioSNG)
is chosen as the fuel of interest, as India is short of domestic natural gas production.
As shown in the figure 3-4, the natural gas production in india has been lower than its
consumption level since 2004, and the gap has been broadening since then. To fulfill
the demand, the Indian government has been relying heavily on LNG imports. If the
existing biomass waste could be converted into synthetic natural gas, the dependence
on import could be largely mitigated. This idea has been practiced in many European
countries, so it would be interesting to explore whether this approach would be a
feasible in India.

70

60

50

40

30

20

10

Marketed gas production Natural gas consumption - LNG net imports

Figure 3-4: Gap between natural gas production and demand in India from 2000 to
2016 [5]

3.3.1 GoBiGas system

For the production of bioSNG, two conversion pathways have been commercially
tested: biochemical conversion using anaerobic digestion and thermochemical conver-
sion using gasification. These two methods utilize different feedstock. Biochemical

36
pathway is better at converting manure-based biomass, while thermochemical path-
way performs better at converting cellulosic biomass. Because Punjab has a large
quantity of agricultural residue, a thermochemical conversion technology called Go-
BiGas is adopted for the model.
Gothenburg Biomass Gasification Project (GoBiGas) was the first commercial
bioSNG plant that injected synthetic natural gas into Sweden's national grid. Al-
though the plant is currently shut down due to funding issue, the abundant perfor-
mance and cost data of the plant makes it a suitable subject for research. The plant
used thermal indirect gasification, during which the heat required to sustain endother-
mic gasification reactions was provided from outside the gasifier [33]. As shown in
figure 3-5, a riser column is used to transfer the heat from biomass combustion into
the gasification chamber using circulated bed materials. With this configuration, the
flue gas and the raw product gas are not mixed, so there is minimal N2 present in the
product gas, making gas cleaning and upgrading in later stages much more efficient.
In figure 3-6, the raw gas is shown to be cleaned thoroughly after exiting the gasifier,
as impurities like tar, H2 S and C02 can easily poison and deactivate methanation
catalysts.
The final output from GoBiGas is 97% of CH4 , 1.9% of H 2 and 1.1% of C02, and
the cold gas efficiency of the overall process is calculated as

_L HVSNG x VSNG
71CG = . X 100% (3.1)
LHVio X hbio

where,

LHVSNG = 44MJ/m 3 (3.2)

LHVio = 16MJ/kg (3.3)

According published experimental and simulation results, the cold gas efficiency of
GoBiGas was 65% [34] [7].

37
5

Limestone
~~~'J
Raw gas Flue gas
3 W--
PRE
A
9I
>1 8'

S + K2 CO3 : s

Product gas ash Fly ash


Coarse flujgas ash
I
1 Gasifier 5 Post combustion chamber
2 Product gas cooler 6 Convection path and flow reversal space
3 Product gas filter 7 Convection path and flow reversal space
4 Combustion chamber 8 Flue gas filter

Figure 3-5: Illustration of the flows in the gasification section of the GoBiGas plant
of the bed material (blue), product gas ash (green), and coarse flue gas ash (orange)
[6]

SODOMs 20 MW 1. Fuel handling system


2. Gasification and Combustion
3. Primary product gas cleaning
3.1. Product gas cooler
.3 3.2. Product gas filter
5 3.3. Precoating and particle recirc.
-Z 3.4. Tar scrubber
3.4 3.5. product gas fan
4. Flue gas system
2 4.1. Flue gas cooling
4.2. Flue gas filter and fan
4.3. Ash handling system
5. Tar adsorption
3.3 6. Compressor
7. Olefin hydrogenation
3.5 8. H2Sscrubber
0__
__ J4.3 9. Water gas shift reaction
10. Pre methanation
77 9 10 11. C02 scrubber
12. Methanation
113 13. Drying and odorization
8

4 12

Figure 3-6: Overall system layout of GoBiGas [7]

3.3.2 Capital and Operating costs of GoBiGas

As one of the first commercialized bioSNG plant, the cost data of the 20MW GoBiGas
has been meticulously recorded [7]. The detailed breakdown of its costs are shown

38
in Table A.1 and Table A.2. After summing up all the components, the capital and
annual operating cost of a 20MW GoBiGas plant are listed out in Table 3.1 and
Table 3.2, respectively. Both tables include values of scale factors for different cost
components. A scale factor (f) is a coefficient that helps determines the cost ratio of
two plants with different capacities using equation 3.4, where C1 and C2 denote the
costs at plant capacity V and V2, respectively.

= (V) (3.4)
C2 V2

Table 3.1: Capital costs and scale factors of a 20 MW GoBiGas plant


Cost Scale factor(f) Capital cost of 20 MW planlt (Million USD)
Reactor systems (from Table A.1) (RE) 0.68 26
Auxillary Equipment and Project Costs (from Table A.2) (AUX) 0.44 126
Total 152

Table 3.2: operating costs and scale factors of the 20 MW GoBiGas plant

Cost Scale factors (f) Annual operating cost (USD/ MWh)


Peronnel (PR) 0.10 20
Maintenance(MA) 0.67 10
Consumables and waste products(CS) 1.00 14.5
Electricity 4
Rapeseed Oil Methyl Esters(RME) / 3.5
Activated carbon/aromatic hydrocarbons removal / 0.9
Other(OR) 5.9
Other costs 0.67 2.9
Total 61.7

Reactor systems (RE) and auxiliary equipment (AUX) are two major capital cost
components. The first component consists mostly the manufacturing and purchasing
costs of reactors and chemicals. The second cost component is more miscellaneous.
Expenditure on land, non-reactor facilities, engineering services, and other auxiliary
costs are all included. The total cost of auxiliary equipment and services is actually
higher than the 'cost of reactor systems, but it has a lower scale factor that makes the
cost lower for larger plants.

39
Operating cost consists expenditures associated with personnel(PR), maintenance(MA),
consumables(CS) and other operation-relevant costs (OR). Unlike capital cost, which
only needs to be made in the first year of operation, operating cost recurs annually.

The capital and operating costs of GoBiGas at scale (V) are calculated using the
costs of the 20 MW GoBiGas pilot plant and scale factors, that are listed out in Table
3.1 and 3.2.

CapEx = RE2oMW x ( 1)0.68 + AUX 20Mw x ( )0.44 (3.5)


20 20
OpEx = PR20MW x (2)"1 + MA20MW X ( 20.
(-VC2M+ 0. (1, OR.6M
CS20W X ( 0)+ x (V) 067(
OR20MW X 6)
20 20 20 20

The resulting economy of scale curves are nonlinear, meaning that as the scale
of the plant increases, the per unit production cost is reduced. Such nonlinearity
can significantly complicate the optimization model, so the curves are separated into
linear pieces as shown in figure 3-7, and the cost equations are broken into piecewise
linear equations shown in equation 3.7 and 3.8.

500
450 - CapEx
44.-+- piecewise linear CapEx
annual OpEx
400- ---+--- piecewise linear OpEx
y3=0.07x+163
350
-

300
-

250
-

y2=0.12x+91
1 200-

150
-

100- y1=0.26x+37

50 y=b.Olx+l.75
50

0 500 1000 1500 2000 2500 3000 3500 4000 4500


biomass consumption [ton/day]

Figure 3-7: Linear piecewise breakdown of GoBiGas capital and operating costs

40
0.26x + 3 7 0 < x < 400
CapEx = 0.12x + 91 400 < x < 1500 (3.7)

0.07x + 163 1500< x

OpEx = 0.01x + 1.75 (3.8)

where,
X = tons of biomass consumed per day
As the original GoBiGas was established in Sweden, to transfer this technology
from Sweden to India, the plant costs should be adjusted according to the social-
economic differences between these two countries. The Energy Sector Management
Assistance Program (ESMAP) administered by the World Bank did a comparative
study on the power sector equipment prices in India and developed countries like
US. The study showed that the most significant cost reductions come from the lower
prices of manufacturing, steel, piping, civil and engineering services in India [35].
Since these findings are applicable to different power systems, they are used to adjust
the GoBiGas cost in India. Feasible cost reductions are listed out in Table 3.3. Most
of the reductions take effect on the costs of auxiliary equipment and services due
to the cheaper prices of raw materials and services in India. The cost of reactor
systems, however, does not get noticeably decreased because a lowered investment in
this section is likely to compromise the plant's performance.
All the cost components in these two countries are plotted comparatively in fig-
ure 3-8 and 3-9. Even though the auxiliary costs are reduced significantly in India,
they are still the largest portion of the capital cost. Similarly as the capital cost, the
plant's operating cost is reduced in India. The main reason for the reduction is that
manpower and consumables like electricity are at a lower cost in India. Rapeseed oil
methyl ester (RME) is a tar scrubbing agent used in the original GoBiGas configura-
tion, and according to a report from Chalmers University of Technology, RME could
be replaced by self-cleaning heat exchangers or scrubber agent distilled from inherent

41
tar products [7]. This new approach is adopted in this cost model, so the cost of
RME is completely eliminated from the operational cost.

Table 3.3: Cost distinctions between Indian and Sweden power plants

Cost components India Sweden India: Sweden


Structural steel (USD/ton) 604 737 1: 1.22
Gas compressor (kUSD) 1 1350 1380 1: 1.02
Intrumentation and control (kUSD) 1 210 240 1: 1.14
Civil(kUSD) 1 930 1260 1 :1.35
Engineering service (kUSD) 1 660 1680 1: 2.54
Fuel handling system ( kUSD) 2 17 38.2 1: 2.25
Piping and mechanical equipments (million USD) 2 164 311 1: 1.90
Cost data from a 25 MW aeroderivative gas turbine plant
2 Cost data from a 300 MW coal powered plant

Services
Piping, mechanical equipment and insulation
Structural steel
Civil
Auxiliary equipment
Drying and odorization
Methanation U
C02 scrubber
Premethanation
Water-gas shift reaction
H2S scrubber
Olefin hydrogenation U
Compressor
Tar adsorption (AC filter) I
Flue gas system IL
Primary product gas cleaning
Gasification and combustion
Fuel handling system
0 10000 20000 30000 40000
Capital expenditure (thousand $)
a India

Figure 3-8: Capital cost breakdown of GoBiGas in India and Sweden

42
Maintenance
Personnel

-
Activated carbon/BTX removal

*
Rapeseed oil methyl ester (RME)
electricity
Other consumables

&
Other operational costs

0 5 10 15 20 25
Operation cost ($/MWh)
m India(US$/MWh) m Sweden (US$/MWh)

Figure 3-9: Annual operational cost breakdown of GoBiGas in India and Sweden

3.4 Methodology

To ensure that the optimization model is representative of real-life situations, es-


pecially on the transportation section, geospatial data on biomass availability, road
network (excluding railroad in this case), and locations of candidate storage cen-
ters and biorefineries are collected and compiled in a GIS software called ArcGIS.
The origin-destination (OD) cost matrix analysis function in ArcGIS can output the
shortest travel distance between any selected points on the map, so it is used to cal-
culate the optimal distances between each pair of connectable locations, e.g. each
biomass supply location to each potential storage center, then each storage center
to each potential plant. The distance matrices are then input into the optimization
model along with the plant's cost curves to derive the optimal plant location, scale
and total cost. This overall process has also been adopted by some other biomass
supply chain optimization models [36] [341.

Biomass
availability Optimization
GIS origin- using Julia and
Road destination Gurobi solver
network algorithm Transportation

Candidate plant Distance matrix


locations

Population

Figure 3-10: GIS-enabled optimization model building processes

43
3.4.1 Base maps generation

Biomass supply locations

Biomass supplier locations are determined with locations of burned agricultural residues.
The only difference between the two maps is that supplier locations exclude burning
spots that are more than 1km away from the road network. This is based on the
assumption that it becomes impossible for collectors to manually move biomass to
the designated pick up points if the farm is more than 1km away from the nearest
road.

Figure 3-11: Agricultural residue supply spots that are proximate to the road network

Road network

The commercial value of road network data makes it difficult to acquire complete
road network data online. Fortunately, OpenStreetMap, a crowd-sourcing mapmaking
platform, provides open-source data on the global road network. Although the quality
of the data is not as high as commercial products, it is sufficient for the current
model. The data includes different classes of roads that are listed out in figure 3-12.
Roads that are categorized as minor or very small roads are eliminated, because it
is implausible to have heavy-duty trucks to drive on these small-scale roads. After
cleaning up the data, the road network in Punjab appears as in figure 3-13.

44
code layer fclaas Deseription OSwlgs
51lx roads Major roads
5111 roads motorway Motorway/freeway highway=motorway
5112 roads trunk Important roads, typically divided highway=trunk
5113 roads primary Primary roads, typically national. highway=primary
5114 roads secondary Secondary roads, typically regional. highway=secondary
5115 roads tertiary Tertiary roads, typically local. highway=tertiary
512x roads Minor Roads
5121 roads unclassified Smaller local roads highway=unclassified
5122 roads residential Roads in residential areas highway=residential
5123 roads living-street Streets where pedestrians have priority highway=living-street
5124 roads pedestrian Pedestrian only streets highway=pedestrian
513x roads Highway links (sliproads/ramps)
5131 roads motorway-link Roads that connect from one road to another highway=motorwayjink
5132 roads trunk link of the same of lower category. highway=trunklink
5133 roads primary-link highway=primary-link
5134 roads secondary link highway=secondary-link
514x roads Very small roads
5141 roads service Service roads for access to buildings, parking highway=service
lots, etc.
5142 roads track For agricultural use, in forests, etc. Often highway=track without
gravel roads. tracktype specification
5143 roads track-gradel Tracks can be assigned a "tracktype" from 1 ... with tracktype=gradel
5144 roads track-grade2 (asphalt or heavily compacted) to 5 (hardly ... with tracktype=grade2

Figure 3-12: Different classes of roads included in OpenStreetMap data

Figure 3-13: Road network adopted for transporting biomass in Punjab

Candidate collection/storage center locations

If every single point in Punjab is proposed as a candidate site for biomass storage
center, the processing time needed for distance generation and optimization will be
extensively long. To reduce the model size and processing duration, sites that are most

45
-0 Rqq

suitable for establishing storage centers are screened out using an original suitability
mapping approach that is described in details below.
To generate a suitability map, the top criteria for a biomass storage center are
listed out first:

1. Established on a piece of unutilized yet accessible land

2. Adjacent to the existing road network

3. Adjacent to biomass suppliers

To fulfill the first criteria, regions that are occupied by the natural land cover and
urban establishment (figure 3-14.e), crop land (figure 3-14.a, figure 3-14.b) are elim-
inated. Even though figure 3-14.a and 3-14.b both reflect the crop land, 3-14.a are
strictly areas that have demonstrated signs of burning of difference kinds of agricul-
tural residues, while 3-14.b include all the regions that have rice plantation, whether
the crops are burned or not. To ensure that biomass can be easily transported to and
from the storage center, locations that are proximate to the road network are given
higher priority, as shown in figure 3-14.c.
Aside from the adjacency to road network, proximity to biomass supply locations
are just as crucial. To evaluate this, a smoothly surface curve is fitted over each
biomass supply location. The surface value is highest at the location of the point
and diminishes with increasing distance from the point, reaching zero at the search
radius distance from the point. The search radius r is determined by equation 3.9,
where SD is the standard distance of input points from the enter, Dm is the weighted
median of these distances and n is the number of points [371.

1 2
r = 0.9 min (SD, Dm)n- (3.9)
ln(2)

The volume underneath the surface curve equals to the amount of straw available
at the point. These smooth surface curves are called the kernel surfaces and adjacent
surfaces always 'overlay with each other. Adding these overlaid values together gives
rise to the kernel density of biomass at each output grid cell. Point with high kernel

46
.. -* W b. C. 1V
1 m 1 High: 10

10 10 Low: 1

d. e.
and u3- WupndisPennemant Sni and KefWalef I
High 10

Low .
3
M
ForseWhrubIandslapM/Penan WeIadsJtiai
GrasidsiCmpianMNaturd

Croplandstfaen
vegetaioln m oan

Figure 3-14: Suitability filtering maps for determining optimal biomass storage cen-
ters: a.regions with burned biomass, b. rice plantation area, c. proximity to road
network, d. proximity to biomass suppliers, e. land cover map.

47
4 -

-
121

26-29 *d
M 30 -32

Figure 3-15: Final suitability mapping products: a. compiled suitability map, b.


regions with suitability score greater than 39, c. suitable locations that have area
greater than 5000 m 2

density indicates there is more biomass in its neighborhood, and vice versa. Figure 3-
14 a-c have a suitability value assigned to each cell grid. A cell with value 10 means
it's a optimal site under that specific criteria, while 0 means the point is a least
suitable site. Although figure 3-14.e is not directly labeled with suitability values, red
and green regions correspond with value 0 while blue region is 10. The final suitability
map is generated by adding all the filter layers together, as shown in figure 3-15.a, and
the most suitable locations have the highest suitability value. Figure 3-15.b filters out
regions with values between 39 and 40, and these regions are aggregated into points
that have areas greater than 5000 m 2 in figure 3-15.c.

Candidate plant locations

Candidate plant locations selection is more straightforward. Sites are selected at


intersections of road network and existing pipelines so that feedstock can be eas-
ily transported to the plant and the produced bioSNG can be injected directly into
pipelines. There are two natural gas pipelines built in Punjab. The major one is
the Bhatinda-Jammu-Srinagar gas pipeline operated by GSPL India Gasnet Limited.
The second pipeline is the Dadri-Bawana-Nangal Gas Pipelines operated by GAIL.
Because the GAIL pipeline does not run through major cities in Punjab, the opti-
mization focuses on the Bhatinda-Jammu-Srinagar pipeline.

48
Figure 3-16: Potential bioSNG plant locations (highlighted in blue) in Punjab

Demand

The amount of bioSNG needs to be produced is determined by the demand of nat-


ural gas at authorized city gas distribution (CGD) regions in the state. As shown
in figure 3-17, six districts in Punjab have been proposed as city gas distribution
areas. The state government's goal is to set up compressed natural gas (CNG) fill-
ing stations and provide cooking gas to households at those regions. The population
that needs to be served is determined by the total population at CGD areas with
population density greater than 1300 people/km 2 , because extending infrastructure
to less populated areas is not cost efficient. The total population eligible for city gas
connection at Bathinda, Tarn Taran, and Jalandhar is around 2 million. According
to India NSSO's report on Energy Sources of Indian Households for Cooking and
Lighting, consumption of LPG per month per capita is 1.926 kg [38]. Using the gross
calorific values LPG and NG, the consumption of natural gas per month per capita
is estimated to be 2.38 n3

49
Punjab regions
with population
density >1300
AOL
Per-km2
1300 -120.0
-1.2091 - 2027.10
-k
2D27.11 - 250.80
20.81 - 3289.70
S329.71 - 407190
4071.91 - 4887.70
487.71 - 5747.00
5747.01 - 6740.50
0740.51 - 802040
SD20.41 - 98699.50

Figure 3-17: Authorized city gas distribution regions in Punjab (highlighted in blue)
and their urban population density

3.4.2 GIS origin-destination algorithm

All the information collected from base maps are input into the GIS software ArcMap.
As aforementioned, the Origin-Destination (OD) cost matrix analysis function in the
software is used to measures the distances between assigned origins and destinations.
In the first round of the analysis, biomass supply locations are featured as origins,
while candidate storage centers are destinations. In the second round of the analysis,
candidate storage centers are origins, and potential plant locations are destinations.
Each round of the analysis generates a distance matrix that contains the distances
between all the origins and destinations points.

3.4.3 Optimization formulation

The objective of the optimization model is to minimize the production cost of bioSNG.
The production cost consists of five factors: feedstock cost (CB), collection cost(Cc),
transportation cost(CT), storage and preprocessing cost(Cs) and plant cost (Cp).

min (CB CC+CT-+ CS+Cp) (3.10)

50
Feedstock cost is calculated by multiplying the market price of biomass residue(cb)
with the total amount of biomass being utilized. fij is the amount of biomass trans-
ported between biomass supply location(i)and storage center (j). There are cer-
tain constraints on the amount of biomass being transported. The total transported
biomass needs to fulfill the demand (D). All the biomass transported out of location
i should not exceed its biomass inventory (b ). Similarly, all the biomass transported
into storage center j cannot go beyond its storage capacity (Si).

CB Cbfi,j (3.11)

f ,j = D (3.12)

f , < bi (3.13)

f < S3
5 (3.14)

The labor and tool cost required for harvesting per ton of agricultural residue is
summed into cc. Multiplying this unit cost with total biomass weight gives rise to
the total collection cost.

Cc ccfij (3.15)

Transportation cost consists of a variable and a fixed component. Variable trans-


portation cost T,) is made up of vehicle fuel cost and drivers' wages, so it increases
with the transportation distance. Fixed transportation cost(Tf), on the other hand,
is the initial investment made for purchasing transportation vehicles, and does not
vary with traveled distance.

CT = TV + Tf (3.16)

The variable cost of transporting one ton of biomass is different between from
supply locations to storage centers (tvi) and from storage centers to the plant(tv 2 ).

51
The reason is that biomass is preprocessed at the storage center and has higher
density afterwards, thus allowing twice as much biomass to be fitted onto the truck.
Therefore, t4 2 can be reduced to half of ti. In eqn. 3.17, subscript k indicates the
variable is associated with plant k. Matrices dij and dj,k contains the distance values
determined by the OD analysis.

TV = E tvjfjdij + tv1fjkdjk (3.17)


i j i k

The transportation of biomass to storage centers has a limited time window due to
the short transition period between crop seasons. However, the transportation from
storage centers to the bioSNG plant does not need to be constrained within the same
time period, and could be carried out throughout the year. Therefore, right after the
harvesting season, all the vehicles can be assigned to transport biomass from supply
locations to storage centers, and then be used to transport between storage centers
and the plant after the peak collection season. This allows the same set of vehicles
to be used for both stages of transportation, so no fixed transportation investment is
needed for the second stage (tf2 = 0).

1ffj + tf2 f,k (3.18) Tf = t


i i j k

The cost of storage center is assumed to linearly proportional to the amount of


biomass stored at the site. c, is the sum of all the cost items included in figure x.

C, = cD (3.19)

The cost of the plant consists of its capital cost(Pc) and operating cost(Po).
Operating cost is linearly proportional to the amount of consumed biomass(qk). Po,,v
is the variable operating cost, while pof is the fixed value. Since not all the candidate
plant locations will be selected, a binary variable, Ok, is introduced. During the

52

I- I 1-11-1-1--l ".-w- " I-..",-"- II-q, -Mmmm""M"


optimization process, Ok is 1 for optimal plant locations, and 0 for the rest.

CP = PC + PO (3.20)

Po = (po,v qk + Po,fOk) (3.21)


k

Ok E (0, 1) (3.22)

To reflect the piecewise linearity of the plant's capital cost, the scale of the plant is
expressed by three consecutive intervals. A, and A2 are the lower and upper bounds
of the plant at a small scale (1 = 1), A2 and A 3 are the lower and upper bounds of
the plant at a medium scale (1 2), and finally A3 and A4 are the lower and upper
bounds of the plant at a small scale (1 = 3).

Alok,1 < qk,1 < A2Ok,1 (3.23)

A2ok,2 qk, 2 < A3ok,2 (3.24)

A3ok,3 < qk,3 < A4ok,3 (3.25)

Binary variable Ok,j is used to indicate whether a plant of scale 1 is decided to be


established at the candidate plant locaiton k or not. If Ok,j equals to one, it means
location k is a optimal plant location, and otherwise if Ok,I equals to zero. In order to
ensure that no more than one plant will be established at each location, the sum of
scale binary variables(ok,l, Ok,2, Ok,3) is always lower or equal to one.

Ok,I = Ok (3.26)

Ok,j E (0, 1), 1 = 1, 2,3 (3.27)

Ok E (0, 1) (3.28)

The variable capital costs (pl,,) are the coefficients of total capital cost equation shown

53
below, and the fixed capital costs (pef) are the constants.

PC= (pqk + P',fOk,l) (3.29)


I k

Considering mass balancing, the amount of biomass processed at bioSNG plants


should equal to the amount of biomass being transported from suppliers to stor-
age centers, and from storage centers to plants. Above all, all the transported and
processed biomass should always equal to the demand.

fj,k - qk,l (3.30)

qk,l = D (3.31)
k 1

54
Chapter 4

Results

4.1 Total cost breakdown

The bioSNG production cost optimization model is formulated using a dynamic pro-
gramming language called Julia and solved with a commercial optimization solver
Gurobi. The locations of optimal biomass suppliers, storage centers and the plant
output from the model are shown in figure 4-1. Based on the scale result, one 80 MW
plant and two storage centers is the most cost efficient setup. Different components
of the total cost are listed out in Table 4.1. The capital cost and the fixed transporta-
tion cost in the table are annualized costs, which are computed by dividing the total
initial investment by an annualization factor, At,,. A 5% discount rate(r = 5%), 20
years of plant operation (t = 20) and 7 years of truck lifetime (t = 7) are adopted for
calculating At,,.

- 1r)-t +
At,r =r (4.1)
r
CapExa = CapEx/At,r (4.2)

Tf,a = Tf/At,r (4.3)

Figure 4-2 shows that annual operation cost and annualized capital cost of Go-
BiGas are the two largest cost components, making transportation and other costs

55
appear insignificant in comparison.

~
0tJp
code
strg
plarI Wae"

PunbJ_ .. bumngspoWaU743N
d
mtrawjaon W ,r..
0 12431 - 107 668697 C.

S10 656494 - 176 33262


+ 176 332629 - 207.40313
+ 207 406314 -278 416728

Figure 4-1: Locations of optimal biomass suppliers, storage centers and bioSNG plant

Table 4.1: Breakdown down of the total cost


Cost Million USD
Feedstock cost (CB) 2.49
Collection cost(Cc 6.54
Annualized fixed transportation cost (Tfa) 2.7
Annual variable transportation cost (TV) 0.5
Preprocessing and storage cost(Cs) 2.5
Annualized plant capital cost(Cp,Capa) 13.7
Annual plant operation and maintenance cost(C,op,) 8.4

Feedstock cost
7%
Annual
operation cost
23%

&storage Cost
7%
Preprocessing

Annualized
transportation
Annualized cost
capital cost 8%
37%

Figure 4:-2: Percentage of each cost component towards the total cost

56
4.2 Environmental impact

Burning of biomass in the open field is notoriously detrimental to the environment,


especially to air quality, so it is interesting to evaluate how much emission can be saved
by converting biomass waste to bioSNG. Table 4.2 lists out the emission intensities
of certain pollutants during open-field straw burning. Most of these emissions are
minimized during bioSNG production, but the transportation of biomass, consumed
electricity and the usage of bioSNG could still lead to emissions, as illustrated by
Table 4.3. The values in Table 4.2 and 4.3 are used to calculate annual emissions of
CO 2 and other pollutants from burning or converting 244510 ton of biomass, which
is the amount required to fully operate the 80MW plant. One thing to note is that
when calculating the emission of C02, the emission from burning bioSNG as cooking
fuel at households is included, while the emission from burning alternative cook fuel,
e.g. LPG, is considered for open-field biomass burning case. The emissions of other
pollutants during bioSNG conversion are mostly caused by the usage of diesel trucks.

Table 4.2: Emission of pollutants from open field straw burning [10j
Name of pollutant Unit Emission level
C02 g/kg 1460
CH4 g/kg 1.2
N 20 g/kg 0.07
CO g/kg 34.7
NOx g/kg 3.1
SO 2 g/kg 2
Total particulate matter (TPM) g/kg 13
PM2 .5 g/kg 12.95
PM10 g/kg 3.7

As shown in figure 4-3, C02 is reduced by 60% from converting burned agricul-
tural residues into bioSNG, while other pollutants are reduced by an average of 98%,
meaning that the pollution from trucks' diesel exhaust is trivial compared to biomass
burning.
The emission reduction could be a source of revenue to the plant if the saved C0 2,e
is sold as carbon credit, which is a form of reward to industries that contributes to

57
Table 4.3: Emissions from the process of bioSNG production [11]
Procedure Emission Intensity(gCO /MJ bioSNG)
Pre-processing electricity 4.34
Transportation 0.65
Plant electricity 10.43
RME combustion 1.3
Gas compression into pipeline 1.65
Household usage 63

C02

PM 2.5

aPM 10

TPM

NOx

NMHC

CO

1.E+00 1.E+02 1.E+04 1.E+06 l.E+08 1.E+10 L.E+12


Annual emission intensity (g)

m Conversion into BioSNG m Burning in the field

Figure 4-3: Emissions from bioSNG production in comparison to burning biomass in


the field

GHG mitigation. India has always been an active player in the carbon market. In
2008, the waste disposal units, plantation companies, chemical plants and municipal
corporations in India produced more than 30 million of carbon credits, which ac-
counted for 20% of the global carbon trading [39]. Since a majority of the carbon
credits produced in India were purchased by European companies in the past, it is
assumed that the carbon credits generated from the biosNG plant will be sold at the
price of European Union Allowance (EUA). Intercontinental Exchange (ICE) predicts
that EUA price will rises to $30/ton by 2022 [8], as shown in figure 4-4. This price is
used to calculate the annual carbon credit income of the plant.

58
35
30
0 25
. 20
<<15

10
5

Year \ month

Figure 4-4: European Union Allowance (EUA) price from 2017 December to 2022
December [8]

4.3 Levelized cost of bioSNG

4.3.1 Future Indian natural gas market price projection

In order to determine the profitability of the plant, the levelized cost of the produced
bioSNG needs to be compared with the future market price of domestic natural gas in
India. As India relies greatly on natural gas import, its domestic natural gas price is
strongly influenced by major gas distributers like the Henry Hub. This correlation is
confirmed in figure 4-5, where the Indian natural gas prices almost overlaps with the
Henry Hub prices that were released a quarter earlier, so the Henry Hub price is used
to predict future Indian natural gas price. In figure 4-6, the future Indian natural gas
price are plotted out based on the Henry Hub price prediction provided by Alberta
Energy Regulator [40]. The orange and red lines correspond with the ceiling and the
floor of future gas prices, while the black line remains as the base price. The levelized
cost of bioSNG is compared with the future base price.

59
Pricing time scale for India

c(0 11 N% b

\'&O\
12 12

10 10

8 8

6 6

4 4

2 2

0 0

Pricing time scale for other regions

- India domestic -Henry Hub - Alberta Hub


- Russian --- modeled India -UK NBP

Figure 4-5: World natural gas prices

9
8
7
6

04

3~
2

0
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
-base -high -low

Figure 4-6: Future India domestic natural gas price projection

4.3.2 Levelized gas cost calculation

The levelized cost of energy is defined as the net present value of the unit-cost of
produced energy over the lifetime of a generating asset. It helps developers and
investors to understand the economic potential of a project over the long-term. In

60
this case, the levelized cost of bioSNG is calculated as:

LCOE - -It + A_ + n Et(


E~ (1 + r)t (I + r)(

where,

It ,Jnvestment cost = C+ + Cc + Cs + CPca, + Tf (t 1)

I ,Investment cost = CB + Cc + Cs (t > 1)

Mt ,Maintenance cost CPOPe

F ,Fuel cost =T

CB = Feedstock cost

Cc = Collection cost

Cs = Storage center cost

CP,Cap = Total capital cost


COpe = Annual operating cost

Tf = Fixed transportation cost

T = Variable transportation cost

The reason that the investment cost in the first year is different from the later
years is that initial investments like plant establishment cost and truck purchases
only need to be made in the first year. On the contrary, collection, feedstock, fuel,
storage center setup, operation and maintenance costs recur annually. As the value
of money deflates with time, the value of future expenditure and generated energy
(Et) are corrected with the discount rate r.

The levelized costs of bioSNG with different plant lifetime projections are calcu-
lated and plotted out in figure 4-7. Assuming that the plant starts its operation at
the beginning of 2019, the vertical axis corresponds with the year when the operation
stops. The lengths of horizontal bars represent the levels of levelized costs, and each
color fraction of a bar indicates a source of revenue that could be used to sustain

61
2039
2028
2027
2026
2025
2024
2023
2022
2021
2020
2019
0 0.5 1 1.5 2 2.5 3 3.5 4
$/M3
2039
2028
2027
2026
2025
2024
2023
2022
2021
2020
2019
0% 20% 40% 60% 80% 100%
mNatural gas market price * Carbon Credit m Shortfall (tariff)

Figure 4-7: Contributions of natural gas market price, carbon credit and tariff towards
the levelized costs of bioSNG as the plant's lifetime increases

the production of bioSNG. For example, the black fraction of bars is amount of the
levelized cost that could be covered by selling bioSNG at the market price of natural
gas, and the red fraction is the amount of cost that has to be covered by tariff or
other sources of external support.

As the number of years in operation increases, the levelized cot of bioSNG is


reduced drastically, especially in the first five years. After that, the price reduction
gradually levels off, until there is no significant cost difference between operating the
plant for 10 years (to 2028) and operating for 20 years (to 2039). Despite of the
gradual reduction of levelized cost, at least 50% of the cost is still in shortfall at

62
the end of 20 years of operation. Even though the India government is known to
distribute funding to biomass plants, it is unreasonable to assume that the plant can
sustain itself by relying so heavily on tariff.

4.4 Sensitivity analysis

As the plant is shown to be not sustainably profitable in the previous section, this
section examines what parameters have the most significant impact over the final
cost and how much the cost can be varied by the changes in these parameters. Since
the capital cost constitutes a great portion of the total cost, it is chosen as the
main subject of this analysis. As shown in section 3.3.2, capital cost consists of two
variables: reactor system cost and auxiliary equipment and service cost. Figure 4-8
demonstrates how the capital cost curve changes as each of these variables increases
or decreases by 20%, and new capital cost curves are linearized again to be inputted
into the optimization model.
Another parameter considered for the analysis is demand. As the demand in-
creases, the scale of the plant should increase accordingly, thus the cost of production
can be driven down by the economy of scale.
The result of sensitivity analysis is shown in figure 4-9. The relationships are not
linear because as one parameter varies, the selection of supplier and site locations
could completely change. Nevertheless, consistent trends are demonstrated by data
points on all the lines. Reactor and auxiliary capital costs are positively correlated
with the levelized cost of bioSNG. As they each reduces 20%, the levelized cost de-
creases around 1.7-2.5%. Auxiliary-cost has twice as much impact on the final cost,
meaning that minimizing the expense on auxiliary equipment and service is a bet-
ter cost reduction strategy than adopting less expensive reactors. Demand, on the
other hand, has a negative correlation with the cost. If the demand increases 20%,
the levelized cost of bioSNG could be reduced 2.5%, but if the demand decreases
20%, the cost increases 3%. In this case, the nonlinearity comes from the economy
of scale. As the levelized cost of bioSNG is at most reduced by 3% with 20% change

63
reactor -20%
4 auxilay -20%
230ne
50 - b obandine
- reator+20%
-00auxary +20%

300

200
-

1004-

0
'

0 500 1000 1500 2000 2500 3000 3500 4000 4500


biomass consumption [ton/day]

reactor cost reduced by20% reactor cost Increase by 2%

=0.07x+153.28 400 y3-0.08x+172.32


300y3.007ll3.76
200 =0.llx+SB.77 - 0 0.13x+95.31
110o 0.24x+3531 100 y1=0.28x+37.78

0 1000 2000 3000 4000 WOO 0 1000 2000 3000 4000 5000
biomass consusr
ption ton/day] biomass consumption [ton/day)
ili auary decrease by 20%cO oastiry hcr*@*@ by 20%

300 y3=0.07x+1.39.76 400 y3-0.0x+185.84'

200 2=0.11x+77.1 30 y2-0.13x+104.98

10 y=.22x+04 y1 -0.29x+42.62

00 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
biomass consumption Iton/day] biomass consumption [ton/day

Figure 4-8: Capital cost curves generated from varying reactor system cost and aux-
iliary equipment and service cost

64
4
3
0

CI0 .

0-

Percentage change in cost parameters (%)


-Demand -Reactor cost -Auxilary capital cost

Figure 4-9: Change in levelized cost of bioSNG attributed to changes in capital cost
and demand

in the examined parameters, the plant is still unable to operate profitably with these
changes.

65
66
Chapter 5

Conclusion and future work

The first objective of this study was to visualize the availability and distribution
of bioenergy potential in India. This was achieved by mining and reprocessing data
from remote sensing satellite images and census statistics, and then building a website
application to display the generated data. This application provides more refined and
comprehensive illustration of Indian bioenergy potential than any other existing data
sources.
After successfully visualizing the availability of bioenergy in India, it was impera-
tive to determine the economic feasibility of bioenergy/biofuel production system in
the country, so a detailed operation optimization study was carried out using the agri-
cultural residue in Punjab as the feedstock and a Sweden bioSNG production plant
(GoBiGas) as the target technology. The optimization model was programmed as a
mixed integer linear programming problem. Optimal locations of suppliers, storage
centers and plants, plant number and scale and biomass flow patterns are output from
the model. After conducting levelized cost calculation and sensitivity analysis on the
optimization result, it is concluded that GoBiGas is not a profitable technology to be
operated in India. However, this is not representative of all biomass systems. It will
be interesting to run the model with a less capital intensive technology configuration,
such as the heat pipe reformer that is being developed at FAU [41] or the anaerobic
digestion technology, and observe how the result will change.
Currently, the model assumes all the input values are fixed and certain, but this

67
assumption is invalid in real life. There are a multitude of uncertainties associated
with the real-life supply chain, e.g. the uncertainty in biomass availability, road
condition, demand and etc. All these variations combined can result in considerable
deviation from the modeled result. Therefore, uncertainty analysis should be carried
out in the future using techniques like Monte Carlo simulation, so the output is not
just a single value, but instead a probability distribution plot.
Among different sources of uncertainties, the uncertainty in biomass availability
is the most critical, because if a company plans its operations based on the projected
availability data but later finds out the biomass is not here, the whole plant's operation
could paralyze. To enhance the data accuracy, the modeled data should be reinforced
with the field data. More specifically, the farmers should be able to input their
inclination to sell or keep their biomass into the potential mapping platform. The
communication could be done through pre-season interviews or texting. Supplier
locations with positive responses from farmers would be given higher priority during
the optimization.
The ultimate goal for this optimization model is to transforms it into an interac-
tive user interface where the user can input their interested plant configuration, plant
cost and feedstock. The platform will then run the analysis and output the results
online. This is challenging currently because running geospatial analysis and opti-
mization on a large dataset usually requires advanced hardware, software and digital
storage from the local user, but with the emergence of CyberGIS, this task could be a
feasible advancement. CyberGIS is a new software framework that combines cyberin-
frastructure with geospatial and other types of big data analyses. The research in this
field was initiated in 2017 at UIUC with NSF support, aiming to tackle global-scale
problems like renewable energy implementation, population growth and emergency
management. This is a great opportunity to incorporate this new research direction
into our existing model, and enhance its usability and impact in the future.

68
Appendix A

Tables

Table A.1: Investment costs and scale factors for all process sections of GoBiGas
plant

Process systems Cost GoBiGas, kUSD Scale factor low7Scale factor Scale factor high
I Fuel handling system 5544
1.1 External fuel feeding system 0.5 0.6 0.7
1.2 Internal fuel feeding system, including lock hoppers 0.4 0.5 0.6
2 Gasification and combustion 3244
2.1 Reactors and refractory 0.6 0.7 0.8
2.2 Condensate treatment andn steam generation / 0.5 0.6 0.7
3 Primary product gas cleaning 2615
3.1 Product gas cooler / 0.6 0.7 0.8
3.2 Product gas filter / 0.6 0.7 0.8
3.3 Precoating and particle handling system, including bed material storage and feed / 0.5 0.6 0.7
3.4 Product gas scrubber / 0.6 0.7 0.8
3.5 Product gas fan and secondary cooler / 0.6 0.7 0.8
4 Flue gas system 2082
4.1 Flue gas cooler / 0.7 0.8 0.9
4.2 Flue gas filter and flue gas fan 0.7 0.8 0.9
4.3 Ash handling system / 0.7 0.8 0.9
5 Tar adsorption (AC filter) 1168
5.1 Activated carbon beds / 0.7 0.8 0.9
5.2 Regeneration system / 0.5 0.6 0.7
6 Compressor 3804 0.6 0.7 0.8
7 Olefin hydrogenation 997 0.6 0.7 0.8
8 H2S scrubber 1007 0.6 0.7 0.8
9 Water-gas shift reaction 582 0.6 0.7 0.8
10 Premethanation 567 0.6 0.7 0.8
11 C02 scrubber 1933 0.6 0.7 0.8
12 Methanation 2135 0.6 0.7 0.8
13 Drying and odorization 547 0.6 0.7 0.8
Total cost of process systems 26225

69
Table A.2: Auxillary equipment, project costs and their corresponding scale factors
of GoBiGas plant

Auxillary equipment and project costs Cost GoBiGas, kUSD Scale factor low Scale factor Scale factor high
A Auxillary equipment 16117
A.1 Flare / 0.6 0.7 0.8
A.2 Hot water system / 0.4 0.5 0.6
A.3 Instrumentation and constrol system (DCS) / 0.3 0.4 0.5
A.4 Power distribution / 0.4 0.5 0.6
A.5 Electrical and instrument installation / 0.3 0.4 0.5
A.6 Compressed air system / 0.5 0.6 0.7
A.7 Fire protection system / 0.5 0.6 0.7
A.8 Inert gas system / 0.5 0.6 0.7
A.9 Safety and security / 0.3 0.4 0.5
A.10 Laboratory and sampling system / 0.2 0.3 0.4
B Civil 24190
B.1 Ground preparation / 0.3 0.4 0.5
B.2 Foundations / 0.3 0.4 0.5
B.3 Buildings, including lights / 0.4 0.5 0.6
B.4 Explosion protection walls / 0.4 0.5 0.6
B5 HVAC / 0.5 0.6 0.7
C Structural steel 5280 0.4 0.5 0.6
D Piping, Mechanical equipment, and insulation 29330 0.5 0.6 0.7
E Services 50636
El Engineering / 0.2 0.3 0.4
E2 Construction Serivices and conunissioning / 0.2 0.3 0.4
E3 Start-up / 0.2 0.3 0.4
E4 Other project servieves / 0.2 0.3 0.4
Total cost 125554 / /
/

70
Bibliography

[1] "IRENA renewable capacity statistics 2018,"


https://www.irena.org/publications/2018/Mar/Renewable-Capacity-Statistics-
2018, 2018.

[2] D. Gonzales, E. M. Searcy, and S. D. Eksioglu, "Cost analysis for high-volume


and long-haul transportation of densified biomass feedstock," TransportationRe-
search Part A: Policy and Practice, vol. 49, p. 48-61, 2013.

[3] M. Hiloidhari, D. Das, and D. Baruah, "Bioenergy potential from crop


residue biomass in india," Renewable and Sustainable Energy Reviews, vol. 32,
p. 504-512, 2014.

[4] Z. Chowdhury, M. Zheng, J. J. Schauer, R. J. Sheesley, L. G. Salmon, G. R.


Cass, and A. G. Russell, "Speciation of ambient fine organic carbon particles
and source apportionment of PM2.5 in indian cities," Journal of Geophysical
Research, vol. 112, no. D15, 2007.

[5] S. Cornot-Gandolphe, "India's vision to a gas-based economy drivers and chal-


lenges," GEDIGAZ Insights, 2017.

[6] H. Thunman, M. Seemann, T. B. Vilches, J. Maric, D. Pallares, H. Strdm,


G. Berndes, P. Knutsson, A. Larsson, C. Breitholtz, and et al., "Advanced bio-
fuel production via gasification - lessons learned from 200 man-years of research
activity with chalmers' research gasifier and the gobigas demonstration plant,"
Energy Science and Engineering, vol. 6, no. 1, p. 6-34, 2018.

[71 H. Thunman, C. Gustavsson, A. Larsson, I. Gunnarsson, and F. Tengberg, "Eco-


nomic assessment of advanced biofuel production via gasification using cost data
from the gobigas plant," Energy Science and Engineering, 2019.

[81 "EUA futures," ICE, https://www.theice.com/products/197/EUA-


Futures/data?marketId=5115270,2019.

[9] T. P. Robinson, P. K. Thornton, G. Kruska, A. Notenbaert, G. Cecchi, M. Her-


rero, M. Epprecht, S. Fritz, L. You, G. Conchedda, and et al., "Global livestock
production systems," Food and Agriculture Organization of the United Nations
(FAO) and InternationalLivestock Research Institute (ILRI), p. 152, 2011.

71
[10] B. Gadde, S. Bonnet, C. Menke, and S. Garivait, "Air pollutant emissions from
rice straw open field burning in india, thailand and the philippines," Environ-
mental Pollution, vol. 157, no. 5, p. 1554-1558, 2009.

[111 A. Alamia, I. Magnusson, F. Johnsson, and H. Thunman, "Well-to-wheel analysis


of bio-methane via gasification, in heavy duty engines within the transport sector
of the european union," Applied Energy, vol. 170, p. 445-454, 2016.

[12] "India to surpass china as world's largest energy growth market: Bp," The
Economic Times, https://economictimes.indiatimes.com/industry/energy/oil-
gas/india-to-surpass-china-as-worlds-largest-energy-growth-market-
bp/articleshow/68010822.cms, Feb 2019.

[13] "India energy," https://www. export.gov/article?id=India-Energy,Oct 2018.


[14] N. Ravindranath, H. Somashekar, M. Nagaraja, P. Sudha, G. Sangeetha,
S. Bhattacharya, and P. Abdul Salam, "Assessment of sustainable non-plantation
biomass resources potential for energy in india," Biomass and Bioenergy, vol. 29,
no. 3, p. 178-190, 2006.

1151 T. Boden, G. Marland, and R. Andres, "Global, regional, and national fossil-fuel
co2 emissions," Carbon Dioxide Information Analysis Center (CDIAC) Datasets,
2017.

1161 "UNFCCC national inventory submissions 2018,"


https://unfccc.int/process/transparency-and-reporting/reporting-and-review-
under-the-convention/greenhouse-gas-inventories-annex-i-parties/national-
inventory-submissions-2018, 2018.

[17] S. Canzana, "Biomass energy conversion," Sustainable Growth and Applications


in Renewable Energy Sources, 2011.

1181 "Scheme to support promotion of biomass based cogeneration in sugar mills and
other industries in the country (up to march 2020), file no.3/141/2017-cpg,"
Ministry of New and Renewable energy, 2018.

1191 "National policy on biofuels 2018," Ministry of New and Renewable Energy, 2018.
[20] K. Natarjan and P. Pelkonen, "Exploiting the unexploited biomass energy in
india through finnish CHP solutions," 2015 4th InternationalConference on In-
formatics, Environment, Energy and Applications, vol. 82, 2015.
[21] N. Z. Atashbar, N. Labadie, and C. Prins, "Modeling and optimization of biomass
supply chains: A review and a critical look," IFA C-PapersOnLine,vol. 49, no. 12,
p. 604-615, 2016.

[221 H. Ghaderi, M. S. Pishvaee, and A. Moini, "Biomass supply chain network design:
An optimization-oriented review and analysis," Industrial Crops and Products,
vol. 94, p. 972-1000, 2016.

72
[23] "Agricultural land (percentage of land area)," The World Bank,
https://data.worldbank.org/indicator/ag.lnd.agri.zs,
2016.

[241 L. Giglio, L. Boschetti, D. P. Roy, M. L. Humber, and C. 0. Justice, "The


collection 6 modis burned area mapping algorithm and product," Remote Sensing
of Environment, vol. 217, p. 72-85, 2018.

[25] P. Purohit and D. Fischer, Second-generation biofuel production in India: Sus-


tainability and cost considerations. 2014.

[26] G. Nicolas, T. P. Robinson, G. R. W. Wint, G. Conchedda, G. Cinardi, and


M. Gilbert, "Using random forest to improve the downscaling of global livestock
census data," Plos One, vol. 11, no. 3, 2016.

[27] F. R. Stevens, A. E. Gaughan, C. Linard, and A. J. Tatem, "Disaggregating


census data for population mapping using random forests with remotely-sensed
and ancillary data," Plos One, vol. 10, no. 2, 2015.

[28] "Database for biomass and waste phyllis 2 by ECN. TNO," https://phyllis.nl/,
2019.

[29] S. Kaza, L. Yao, P. Bhada-Tata, and F. Van Woerden, "What a waste 2.0: A
global snapshot of solid waste management to 2050," The World Bank, 2018.

[30] "National policy for management of crop residues (NPMCR)," Ministry of Agri-
cultural of Government of India, 2014.

[31] N. Jain, A. Bhatia, and H. Pathak, "Emission of air pollutants from crop residue
burning in india," Aerosol and Air Quality Research, vol. 14, no. 1, p. 422-430,
2014.

[321 S. Bhuvaneshwari, H. Hettiarachchi, and J. Meegoda, "Crop residue burning


in india: Policy challenges and potential solutions," International Journal of
Environmental Research and Public Health, vol. 16, no. 5, p. 832, 2019.

[331 J. Ahrenfeldt, B. Jorgensen, and T. Thomsen, "BioSNG potential assessment:


Denmark 2020," Rational Laboratoryfor Sustainable Energy, DTU, 2010.

[341 A. Singlitico, I. Kilgallon, J. Goggins, and R. Monaghan, "Gis-based optimiza-


tion model for the smart design of a nationwide bio-sng production system for
ireland," 25th European Biomass Conference and Exhibition, 2017.

[35] D. Pauschert, "Study of equipment prices in power sector," ESMAP Technical


Paper, 2009.

[36] T. Lin, L. F. Rodriguez, Y. N. Shastri, A. C. Hansen, and K. Ting, "Gis-enabled


biomass-ethanol supply chain optimization: model development and miscanthus
application," Biofuels, Bioproducts and Biorefining, vol. 7, no. 3, p. 314-333,
2013.

73
[371 "ArcGIS pro: How kernel density works," https://pro.arcgis.com/en/pro-
app/tool-reference/spatial-analyst/how-kernel-density-works.html,2018.

[381 Energy Sources of Indian Households for Cooking and Lighting. 2012.

[39] J. Joseph, "Carbon credit and how you can make money from it," The Rediff
Interview, https://www.rediff.com/money/2008/feb/05interl.html, 2008.

[40] "Natural gas prices |henry hub price," Alberta Energy Regulator,
https://www. aer. ca/providing-information/data-and-reports/statistical-
reports/henry-hub, 2019.

[41] J. Karl, "Biomass heat pipe reformer-design and performance of an indirectly


heated steam gasifier," Biomass Conversion and Biorefinery, vol. 4, no. 1,
p. 1-14, 2013.

74

You might also like