Case_Study_2

Case Study 2: Comparison between Epoxy Resin Productions from Algal or Soya Oil –
An LCA Based Problem Solving Approach
This chapter shows the calculations of marginal environmental impacts of replacing one
biomass feedstock with another. The calculations have been shown in the form of solving a
problem. The objective of the problem is to assess and compare the environmental impact
characterisations of epoxy resin productions from a first generation feedstock and a
lignocellulosic feedstock. Soya bean paste is the selected first generation biomass feedstock.
Algal paste is the selected lignocellulosic feedstock. The inventory analysis and impact
assessments have been shown for various parts of the systems. The results for any part can be
adapted to other studies. The chapter shows how to model the systems in the LCA software
GaBi.
Building on chapter 5 of the textbook, Monte Carlo simulation for uncertain parametric or
decision variable analysis has been shown in the context of problem solving approach,
demonstrating the importance and the powerfulness of the technique.
Epoxy resins are used in electronics, e.g. moulding transistors or printed circuit boards,
construction, automotive as laminates and marine construction industries. They form
continuous phase (matrix) of composite materials. The discontinuous phase also known as
reinforcement may consist of glass, carbon or Kevlar fibres.
1. Problem Statement
You are in charge of conducting an LCA study to compare between first generation and
lignocellulosic oily feedstocks’ environmental performances for the production of epoxy
resins. Your client is an epoxy resin (ER) plant owner based in the USA, who wants to
investigate the environmental impacts of epoxy resin production from soya oil and whether
epoxy resin production from algal oil is environmentally more viable compared to soya oil.
The result of the study is for the ER industry’s internal decision making for selecting the
environmentally more benign feedstock. The results of the study will only be published
internally.
At the end of the study of this chapter, you are asked to produce a technical report for the
board of directors of the company to show findings.
The company also confirms that they are interested in the marginal benefits in environmental
impacts between the two feedstocks. In addition to the primary impact characterisation
assessments, a comparison of land use between the two feedstocks is desirable.
The company recommends the use of GaBi software and Ecoinvent 2.0 databases for
inventory analysis for soya bean production and any part of the process and material (where
data are unavailable).
The goal of this study is to compare the primary impact characterisations between the two
biomass feedstock processing routes. The intention is to find environmentally more viable
routes in important primary impact categories and whether it is environmentally feasible to
use lignocellulosic for this production.
Biorefineries and Chemical Processes: Design, Integration and Sustainability Analysis, First Edition.
Jhuma Sadhukhan, Kok Siew Ng and Elias Martinez Hernandez.
© 2014 John Wiley & Sons, Ltd. Published 2014 by John Wiley & Sons, Ltd.
The system under consideration includes the cradle to gate system for 1 ton ER production.
The environmental impact categories can be determined for 1 ton of epoxy resin production
from soya paste and algal paste. The inventories and impact potentials can be reported for the
functional unit 1 ton of epoxy resin production from soya paste or algal paste.
The raw materials required by the main process blocks in the ER production process under
consideration are: hexane, steam and electricity for oil extraction; formic acid and hydrogen
peroxide (50% by weight in water solution) for epoxidation; phthalic anhydride and
trimethylamine for the curing processes. Figure 1 shows the chemical and energy
requirements of the ER production processes from soya paste as well as algal paste. Figure 2
shows the ER plant yields from the raw material pastes: soya and algae for the same
functional unit of 1000 kg ER production. Both figures follow the same naming convention.
Table 1 shows the chemical and energy consumptions by the two ER plant operations, each
producing 1000 kg ER. Further systems boundaries include the production of the raw
material and utility streams. Inventory data using GaBi/Ecoinvent can be collected for
comparisons between soya based and algal based ER production systems. All primary impact
categories must also be analysed for uncertainty analyses and probability distribution to find
their sensitivities.
Soya paste
or Algae paste Extracted oil Epoxidised oil
Oil extraction Epoxidation Curing Epoxy resin
Hexane to feedstock Formic acid: 0.224 Phthalic anhydride

weight ratio to be times the weight of 4.75 and 4.35 times
maintained at 10-14.4 oil the weight of soya
(soya oil and algal oil and algal oils,
respectively) Hydrogen peroxide respectively
Only 4% of hexane is (50% by weight):
consumed. 0.635 times the Triethanolamine
weight of oil 0.02 times the
Steam: 0.46 times the weight of oil
weight of feedstock
Electricity: 1.3 MJ times

the weight of feedstock
Figure 1 ER plant chemical and energy requirements.
Soya paste > Soya oil > Epoxidised oil > Epoxy resin
362 kg 175 kg 175 kg 1000 kg
Algae paste > Algal oil > Epoxidised oil > Epoxy resin
435 kg 188 kg 188 kg 1000 kg
Figure 2 ER plant yields.
Table 1 Chemical and energy consumption by ER plant operation for 1000 kg ER production.
Soya paste Algae paste

Hexane, kg 145 251
Steam, kg 167 200
Electricity, MJ 471 566
Formic acid, kg 39 42
Hydrogen peroxide (50%), kg 111 119
Phthalic anhydride, kg 831 818
Triethanolamine, kg 3.5 3.76
2. Cultivation System Environmental Impact Analysis
The USA based soya bean cultivation system extracted from the Ecoinvent 2.0 database
shows biogenic energy available 4.4 MJ m-2 land use or 19.73 MJ kg-1 soya paste. The results
of biogenic energy analysis for 362 kg soya paste production via cultivation (Figure 2) are
shown in Table 2. 100% yield of soya bean to soya paste and no requirement for additional
water for the paste preparation were assumed.
Table 2 Biogenic energy analysis for 362 kg soya bean cultivation.
Energy input, calorific value, in organic substance [Renewable energy resources] 7655.21 MJ
Energy input, gross calorific value, in biomass, primary forest [Renewable energy resources] 0.04 MJ
Energy input, kinetic (in wind), converted [Renewable energy resources] 0.63 MJ
Energy input, potential (in hydropower reservoir), converted [Renewable energy resources] 12.12 MJ
Energy input, solar, converted [Renewable energy resources] 0.01 MJ
2
Land use [Hemerobie ecoinvent] 1623.5 m
Waste heat output [Other emissions to fresh water] 12.64 MJ
Waste heat output [Other emissions to air] 512.71 MJ
Waste heat output [Fresh water] 0.11 MJ
Waste heat output [Other emissions to sea water] 0.00 MJ
Net biogenic energy in soya paste, 362 kg 7142.55 MJ
2 -2
Biogenic energy per m land use 4.40 MJ m
-1
Low heating value of soya paste 19.73 MJ kg
Table 3 shows the carbon dioxide balance, net carbon dioxide capture by 362 kg soya paste
produced, net carbon dioxide capture per kg of soya paste production, net carbon dioxide
capture per m2 land use and biogenic energy capture per kg of carbon dioxide capture.
Table 3 Carbon dioxide capture analysis by soya paste production.
Carbon dioxide input

Carbon dioxide [Renewable resources] 515.89 kg
Carbon, in organic matter, in soil [Non renewable resources] 0.00 kg
Carbon dioxide output
Carbon dioxide [Inorganic emissions to air] 31.44 kg
Carbon dioxide (biotic) [Inorganic emissions to air] 0.57 kg
Carbon dioxide, land transformation [Inorganic emissions to air] 0.01 kg
Net carbon dioxide capture by 362 kg soya paste 483.88 kg
-1
Net carbon dioxide capture per kg soya paste 1.34 kg kg
Net carbon dioxide capture by soya paste per m2 land 0.30 kg m
-2
-1
Biogenic energy capture per kg carbon dioxide capture 14.76 MJ kg
The following inventories for algae cultivation were obtained from the company: 1000 m2
land produce 3000 kg algae paste.
The other raw materials consumed per 1000 m2 land are:
• carbon dioxide: 5400 kg
• total N:P:K fertiliser (5:1:1): 192 kg
• electricity: 6537 MJ
3. Approach to the Solution of the LCA Based Decision Making Problem
3.1. Approach to the LCA Study
The inventories for hexane, formic acid, hydrogen peroxide, phthalic anhydride triethylamine
and heat and electricity productions need to be extracted. Along with these inventories, full
inventory data set for soya bean cultivation system need to be extracted using GaBi /
Ecoinvent database.
As the company is interested in the marginal benefits in environmental impacts between the
two feedstocks, this study aims to obtain the avoided inventories and emissions by the two
routes. More saving indicates environmentally more sustainable feedstock. Table 4 thus
shows comparisons of chemicals’, energy and CO2 emission savings between the two routes.
Hence, the soya paste system (362 kg soya paste conversion into 1000 kg ER production),
will save 106 kg hexane, 34 kg steam, 1043 MJ electricity, 3 kg formic acid, 8 kg hydrogen
peroxide (50%), 0.26 kg trimethylamine, 28 kg N:P:K (5:1:1) fertiliser and 484 kg carbon
dioxide (captured by soya beans).
Similarly, by using 435 kg algal paste for 1000 kg ER production, 13 kg phthalic anhydride
and 783 kg carbon dioxide (captured by algae) will be saved.
The savings are calculated from the differences in the chemical and energy consumptions by
the two ER plant operations for 1000 kg ER production, (from the data shown in Table 1 and
that given for algae cultivation by the company).
Table 4 Comparison of avoided chemical and energy consumptions and carbon dioxide
emissions.
Avoided inventory Avoided inventory

due to soya paste use due to algae paste use
Hexane, kg 106
Steam, kg 34
Electricity, MJ 95
Formic acid, kg 3
Hydrogen peroxide (50%), kg 8
Phthalic anhydride, kg 13
Triethanolamine, kg 0.26
Cultivation:
Carbon dioxide, kg 484 783
N:P:K fertiliser, kg 28 from GaBi/Ecoinvent
Electricity, MJ 948 from GaBi/Ecoinvent
Two cases for avoided inventories and primary impact characterisations can now be created,
as follows:
• Due to soy paste use
• Due to algae paste use
Each avoided primary impact characterisation by a feedstock can thus be determined.
Values of input flows to enter in the GaBi/Ecoinvent software
• Steam saving: 34 kg = 96 MJ (determined from enthalpy requirement to generate

steam at 5 atmospheric pressure and 175oC temperature from water at 25oC and 1
atmosphere)
• A wood based cogeneration plant generating 3.6 MJ of electricity and 4.33 MJ of heat
can be designed for 96 MJ of heat generation to meet the additional heat demand by
the algal system (Table 4). An associated electricity generation is 80 MJ.
• The balance of electricity is supplied from grid (USA): 963 MJ (From Table 4, grid
electricity supply = (948 + 95 – 80) MJ).
• 28 kg 5:1:1 N:P:K fertiliser supply for algae cultivation:
• Ammonium nitrate phosphate for 9 kg P2O5 or 3.98 kg phosphorous

production can be used. This provides 1.47 kg of nitrogen.
• Potassium nitrate 3.98 kg potassium or 4.8 kg K2O production can be used.

This provides 1.53 kg of nitrogen.
• The balance of nitrogen fertiliser required for algae cultivation is 17 kg

nitrogen, can be supplied from ammonium nitrate.
Once these bases are determined, the two models for avoided inventories and primary impact
characterisations can be built in the LCA software, GaBi. Adequate levels of inventory and
primary impact characterisation data have been shown for usability and adaptation. The
sensitivity and Monte Carlo simulation analyses results are shown in the end.
3.2. Model Basis
Avoided chemicals’ and energy consumptions and carbon dioxide emission by 362 kg soya
paste conversion into 1000 kg ER production are as follows.
1. 106 kg hexane, at plant
2. 0.26 triethanolamine, at plant
3. 3 kg formic acid, at plant
4. 8 kg hydrogen peroxide, 50% in H2O, at plant
5. ammonium nitrate phosphate, as 9 kg P2O5 at regional storehouse
6. ammonium nitrate, as 17 kg N, at regional storehouse
7. potassium nitrate, as 4.8 K2O, at regional storehouse
8. 80 MJ electricity and 96 MJ heat cogeneration from wood
9. 963 grid electricity
10. 484 kg carbon dioxide emissions avoided by soya bean cultivation.
The cradle to grave raw material and primary energy inventories of the above 1-9 material
and energy streams saved need to be aggregated and the overall primary impact
characterisations need to be estimated.
Avoided chemicals’ and energy consumptions and carbon dioxide emission by 435 kg algae
paste use for 1000 kg ER production are as follows.
1. 13 kg phthalic anhydride
2. inventories associated with 362 kg soya bean cultivation
3. 783 kg carbon dioxide emissions avoided by algae cultivation.
The cradle to grave raw material and primary energy inventories of the above 1-2 material
and energy streams saved need to be aggregated and the overall primary impact
characterisations need to be estimated.
3.3. Model Building in GaBi
Step 1: Create 2 new projects (select the ‘Projects’ folder and click on ‘new’ icon, top menu,
OR right-clicking in projects work area…) and name them ‘Epoxy resin plant – avoided
emission due to soya paste use’ and ‘Epoxy resin plant – avoided emission due to algae paste
use’. Go to the tab ‘ISO Documentation’ and describe the goal of the study, system function
and boundary, etc. (from the introduction). Activate the project.
Step 2: Create 2 new plans called ‘Epoxy resin plant – avoided emission due to soya paste
use’ and ‘Epoxy resin plant – avoided emission due to algae paste use’ in the relevant plan
folder (e.g. ‘Plans > Manufacturing’). Once you have named these plans, you can use some
processes already existing in the database by dragging them from the relevant ‘processes’
sub-folders:
• Processes > Ecoinvent > chemicals > organics > hexane, at plant
• Processes > Ecoinvent > chemicals > organics > formic acid, at plant
• Processes > Ecoinvent > chemicals > organics > phthalic anhydride, at plant
• Processes > Ecoinvent > chemicals > organics > triethanolamine, at plant
• Processes > Ecoinvent > chemicals > inorganics > hydrogen peroxide, 50% in H2O,
at plant
• Processes > Ecoinvent > wood energy > electricity, at cogeneration 6400kWth, wood,
allocation energy
• Processes > Ecoinvent > electricity > supply mix > electricity, low voltage, at grid
• Processes > Ecoinvent > agricultural means of production > mineral fertiliser >
ammonium nitrate phosphate, as P2O5, at regional storehouse
ammonium nitrate, as N, at regional storehouse
potassium nitrate, as K2O, at regional storehouse
• Processes > Ecoinvent > agricultural production > plant production > soya beans, at
farm
Screenshot showing an existing process from the database (from the main GaBi window, left
of the picture) dragged on to the DB Plan area on the right hand side. Click on the desired
object from the GaBi DB work area and drag it to the desired window while holding the left
button. This is shown in Figures 3-5.
Figure 3 Selection of ‘Processes > Ecoinvent > chemicals > organics > hexane, at plant’.
Figure 4 Double click on ‘Processes > Ecoinvent > chemicals > organics > hexane, at plant’,
to examine the inventories associated with the hexane production process.
Figure 5 Selection of ‘Processes > Ecoinvent > wood energy > cogeneration > electricity, at
cogeneration 6400kWth, wood, allocation energy’.
Step 3: Creating new processes: Obviously not inventory database of all processes exists in
the software platform, and you will have to create some when modelling systems for LCA
studies. They can be created directly on a plan, by right-clicking on the plan area and
choosing to create a new process; this would file the new process in the process root folder
(subsequently you should classify it in the relevant folder). Alternatively, you can create the
new process directly in the relevant database folder (with the ‘new’ icon or right-clicking in
the DB manager area at the correct process folder). The latter is the preferred approach.
To define a process you need to quantify the input and output flows to this process. To insert
a flow, drag and drop it from the relevant flows sub-folder to the inputs or outputs space of
the process sheet, as required. Then specify the amount (for some flows you can select the
quantity e.g. mass, volume, etc., units e.g. kg, tons, g, etc. and input other data e.g. the date)
and the type of flow under the ‘Tracked flows’ column: X denotes a valuable flow (material,
energy, product, etc.); * represents waste flows going to further treatment, and [void] is used
for elementary flows, either coming from nature (resources) or going into nature (emissions).
Step 4: Creating a new flow: Create a new flow ‘hexane flow’. When a relevant flow does
not exist in the GaBi DB, you must create it. It is recommended that you create it directly in
the relevant sub-folder (e.g. flows > valuable substances). Set ‘Mass’ as reference quantity.
Once you have finished with the process, save it and make sure it is filed in a relevant folder
(e.g. Processes).
Create a new process by right clicking on the DB Plan and name it ‘ER plant soya paste total
inventory’. Create a new flow ‘hexane flow’ as one of the inputs. Note that the new flow
‘hexane flow’ is in the Mass unit. Specify the amount of saving: 106 kg and track it. This is
shown in Figure 6.
Figure 6 Create a new process ‘ER plant soya paste total inventory’. Create a new flow
‘hexane flow’.
Figure 7 shows the creation of a new input flow for electricity generation from the CHP
plant.
Figure 7 Create a new input flow for electricity generation from the CHP plant.
A total of 9 new flows have been created to connect to the 9 process blocks. Note their
quantities are different, such as ‘Energy (net calorific value)’ quantity for ‘electricity flow’
and ‘electricity grid flow’ (Figure 8).
Figure 8 Creation of all 9 input flows for the cradle to grave inventory analyses of products
and services saved by the soya paste system.
Step 5: Linking processes: All processes and plans in GaBi need to be linked in order for the
mass balance to be performed. To link processes in a plan, first, select the ‘source process’
(the one originating the flow) by clicking on it. Then, click on the brown bar at the right of
the process box and hold the button while drawing a line (dragging) to the ‘sink process’ (the
one using the flow). This is shown in Figure 9.
1. Select right bar of the source process.
2. Hold to draw a line reaching the sink process.
Figure 9 Linking of ‘hexane, at plant’ source process block with the new sink process block
‘ER plant soya paste total inventory’ via ‘hexane flow’.
Figure 10 shows the completed cradle to gate system flowchart ‘Epoxy resin plant - avoided
emission due to soya paste use’. Note that GaBi only shows the mass flowrates on the DB
plan visually hence energy production blocks are showing 0 mass outputs.
Figure 10 Cradle to grave system flowchart ‘Epoxy resin plant - avoided emission due to
soya paste use’.
Fix the scaling factor. The ratio of various input mass flows will remain the same as they
have been tracked, shown in Figure 11.
Figure 11 Fixing the scaling factor in the new process ‘ER plant soya paste total inventory’
determines the output flows from the other 9 blocks in the ‘Epoxy resin plant - avoided
emission due to soya paste use’ flowchart.
The calculation has now already been done and you have to access the calculation. First save
the DB Plan: ‘Epoxy resin plant – avoided emission due to soya paste use’. Then work on the
balance. This is shown in Figure 12.
2
Figure 12 First save the DB Plan: ‘Epoxy resin plant – avoided emission due to soya paste
use’. Then work on the balance.
Once a process (and any other GaBi object) has been created in the database it can be used
any number of times in plans. In each new use, the flows related to the process will be scaled
up or down, according to the reference flow in the new plan; however, this is independent of
the flow values stored in the database, as other plans might be using the same process. Each
new use in a plan is called a process instance.
Instance: Double-clicking a process in the plan displays the properties of the process for the
current use. Any change in these properties (e.g. scaling factor, values of free parameters)
only affects the use in the current plan.
On the other hand, when opening a process (or other object) directly from the database, we
see the process details.
Details: right-clicking a process in the plan and choosing ‘details’ (or opening the process
from the database) displays the definition of the unit process in the database.
Step 6: Creating and navigating balances for life cycle inventory (LCI) and life cycle impact
assessment (LCIA): The ‘balance’ is the calculation of inputs and outputs to the studied
system (be it a single process, a plan representing a life cycle stage, or a full life cycle). In
order to calculate a balance, GaBi needs to know the reference flow according to which we
want it to calculate the burdens, i.e. one process or plan instance needs to be fixed at the level
at which we want to calculate the balance. Once the balance is calculated, it can show the
LCI and LCIA results.
To show the LCI mass results, you can simply ‘expand’ the rows of input flows and output
flows by double-clicking on the rows named ‘flows’ and the columns of life cycle
stages/items by double-clicking on the aggregated column, named after the plan you are
analysing. Double-clicking any shaded column (representing a plan) ‘expands’ the next level
of detail available (e.g. in the first plan, it shows the first level of sub-plans; for a sub-plan it
shows any lower levels of sub-plans and/or processes; etc.). You can shift to see balances of
other quantities (energy balance or contribution to various impact indicators) by selecting the
relevant quantity from the drop-down menu.
You can see all the impact indicator results (LCIA results) simultaneously and contribution of
each life cycle stage to these by clicking the “Quantity View” box (In GaBi impact indicator
results are called “environmental quantities”). You can see only those quantities that make a
substantial contribution to the LCI or LCIA by clicking on the “weak point analysis” icon.
You can see the relative contribution of each stage by selecting relative rather than absolute
values. In order to normalise the impact assessment results, you can use the Normalisation tab
and select a normalisation regime from the dropdown menu. To create a single weighted
environmental impact score, select the Evaluation tab and select a weighting set from the
drop down menu. (Note that this step is not recommended). To plot the results, select the
rows (flows or quantities) you want to represent and click on ‘Diagram’. However,
sometimes you might prefer to export the results to MS Excel and create a graph in Excel
(You can do so by copying the values you want to work with and pasting them in a
spreadsheet). The balance view has extensive functionality, and you are encouraged to
explore its possibilities by using the help options and the GaBi Manual. Figure 13 shows
significant LCI results in mass (kg).
Shows all available information on environmental and

technical quantities per plan / process in drop-down list.
Absolute
or relative
values.
Double click to expand rows to see individual flows and to
expand columns to see individual plans / processes.
Figure 13 Significant LCI results in mass (kg).
To show the LCI mass results, you can simply ‘expand’ the rows of input flows and output
flows (by double-clicking on the rows named ‘flows’) and the columns of life cycle
stages/items by double-clicking on the aggregated column, named after the plan you are
analysing. Double-clicking any shaded column (representing a plan) ‘expands’ the next level
of detail available e.g. in the first plan, it shows the first level of sub-plans; for a sub-plan it
shows any lower levels of sub-plans and/or processes; etc.. This is shown in Figure 13. The
balances of other quantities (energy balance or contribution to various impact indicators) can
be done by selecting the relevant quantity from the drop-down menu.
Following steps 1-6 and Figures 3-13, Figure 14 can be created to show the cradle to gate
system flowchart ‘Epoxy resin plant - avoided emission due to algal paste use’. The two
production processes involved here are ‘phthalic anhydride, at plant’ and ‘soya beans, at
farm’. Their inventories are set according to the values shown in Table 4. Phthalic anhydride
manufacturing plant inventory data are extracted for 13 kg of phthalic anhydride production
and 362 kg of soya beans’ cultivation is the basis for inventory data extraction.
Figure 14 Create another new plan ‘Epoxy resin plant - avoided emission due to algae paste use’ for
comparison of avoided emissions between soya paste and algae paste uses.
3.4. Impact Assessment Data Collection
Once the balance of the two flowcharts in the plans in Figures 12 and 14 is performed, the
individual as well as total avoided impacts estimated are collected. GaBi provides input and output
assessments separately. This is particularly relevant for the global warming potential impact over
100 years. Thus, in order to calculate the net avoided global warming potential, the carbon dioxide
input needs to be subtracted from the output carbon dioxide emission. Thus, the positive net value
of (input – output) implies carbon capture and negative value implies global warming potential
impact to the atmosphere.
Table 5 shows the avoided primary impact potentials from 362 kg soya paste conversion into 1000
kg ER production. The key areas of potential savings are shown in shades. These are global
warming potential over 100 years, marine aquatic ecotoxicity potential, human toxicity potential
and fresh water ecotoxicity potential. It is also shown that the maximum avoided impact potentials
are obtained by removing grid electricity system.
Table 4 shows 484 kg of biogenic carbon dioxide capture by 362 kg of soya bean cultivation (or
36.5% biogenic carbon content by weight in the soya paste). The net avoided global warming
potential calculated in Table 5, 503 kg CO2 equivalent, is due to the avoidance of additional
inventories for the algal paste system. However, this must be added with 484 kg of biogenic carbon
dioxide capture by 362 kg of soya bean cultivation, for the total avoided global warming potential
by the soya paste system. Thus, the total avoided global warming potential calculated is 987 kg CO2
equivalent.
However, Table 5 is not adequate to assess the differences in impact characterisations between the
soya paste and algal paste plants for ER production. The avoided primary impact potentials by the
435 kg algal paste conversion system for the production of 1000 kg ER must also be established.
These results are shown in Table 6. Significant savings are obtained from the toxicity potential
categories. The soya bean cultivation system is the hot spot for it.
Table 5 Avoided primary impact potentials by 362 kg soya paste conversion system for the production of 1000 kg ER.
Net saving Wood CHP Ammonium nitrate phosphate Ammonium nitrate Formic acid
Acidification potential kg SO2 equivalent 2.9336 0.0076 0.0935 0.4806 0.0274
Eutrophication Potential kg phosphate equivalent 0.7191 0.0015 0.0222 0.1777 0.0048
Freshwater aquatic ecotoxicity potential kg DCB equivalent 35.69 0.05 1.72 5.79 0.39
Global warming potential (GWP 100 years) kg CO2 equivalent 503.00 0.29 11.45 147.67 7.53
Human toxicity potential kg DCB equivalent 167.09 0.37 3.91 33.64 1.70
Marine aquatic ecotoxicity potential kg DCB equivalent 313678 96 8191 20155 2460
Ozone Layer Depletion Potential kg R11 Equivalent 7.02E-05 2.54E-08 1.35E-06 7.01E-06 1.37E-06
Photochemical ozone creation potential kg ethene equivalent 0.4345 0.0006 0.0083 0.0240 0.0043
Terrestric ecotoxicity potential kg DCB equivalent 5.0546 0.0210 0.1445 0.9799 0.0412
Abiotic depletion potential kg Sb equivalent 5.1542 0.0016 0.0812 0.4507 0.0820
Hexane Hydrogen peroxide Potassium nitrate Triethanolamine Grid electricity

Marine aquatic ecotoxicity potential kg DCB equivalent 32039 3848 1268 287 245334
Photochemical ozone creation potential kg ethene equivalent 0.3013 0.0045 0.0019 0.0005 0.0892
Table 6 Avoided primary impact potentials by 435 kg algal paste conversion system for the
production of 1000 kg ER.
Phthalic anhydride Soybeans

Net saving manufacturing cultivation
Acidification potential kg SO2 equivalent 0.6528 0.1418 0.5110
Eutrophication Potential kg phosphate equivalent 2.6218 0.0119 2.6098
Freshwater aquatic ecotoxicity potential kg DCB equivalent 4.1040 1.2446 2.8594
Global warming potential (GWP 100 years) kg CO2 equivalent 33.20 33.20
Human toxicity potential kg DCB equivalent 21.08 10.60 10.48
Marine aquatic ecotoxicity potential kg DCB equivalent 22458 7555 14903
Ozone Layer Depletion Potential kg R11 Equivalent 5.17E-06 8.84E-07 4.28E-06
Photochemical ozone creation potential kg ethene equivalent 0.0777 0.0432 0.0345
Terrestric ecotoxicity potential kg DCB equivalent 0.4190 0.3775 0.0415
Abiotic depletion potential kg Sb equivalent 0.6788 0.4437 0.2351
The total avoided global warming potential by algal system similarly has the avoided impacts due to
additional inventories for the soya paste plant, 33.20 kg CO2 equivalent (Table 6) and the biogenic
carbon dioxide capture by the algal cultivation process, 783 kg CO2 equivalent (Table 4). Thus, the
total avoided global warming potential by the algal system is 816.20 kg CO2 equivalent. Thus, the
resulting comparison of the avoided emissions between the soya and algal based epoxy resin
production plants is shown in Table 7. The soya based system provides greater avoided impact
potentials, except in eutrophication potential. As expected, first generation crops have more
profound leaching effects due to fertiliser / nutrient application. This could also be due to avoided
eutrophication potential by growing algae in closed bioreactor systems.
Table 7 Comparison of avoided emissions between soya and algal based epoxy resin production
plants.
Avoided impact Avoided impact Avoided impact
due to soya paste use due to algae paste use by soya system
Acidification potential kg SO2 equivalent 2.9336 0.6528 2.2808
Eutrophication Potential kg phosphate equivalent 0.7191 2.6218 -1.9027
Freshwater aquatic ecotoxicity potential kg DCB equivalent 35.69 4.10 31.5862
Global warming potential (GWP 100 years) kg CO2 equivalent 987.00 816.20 170.7960
Human toxicity potential kg DCB equivalent 167.09 21.08 146.0155
Marine aquatic ecotoxicity potential kg DCB equivalent 313678 22458 291220.2759
Photochemical ozone creation potential kg ethene equivalent 0.4345 0.0777 0.3568
Terrestric ecotoxicity potential kg DCB equivalent 5.0546 0.4190 4.6356
Abiotic depletion potential kg Sb equivalent 5.1542 0.6788 4.4754
The land use impact of soya based system is greater, 1623.5 m2 (obtained from GaBi), compared to
145 m2 obtained for algal system (on the basis of 1000 m2 land producing 3000 kg of algae), shown
in GaBi simulation snapshot in Figure 15. It is expected that first generation feedstocks would have
more land use impact than lignocellulosic feedstocks, by many folds. This study shows the practical
implication of this assumption. Hence, there are trade-offs between the land use impacts from soya
and primary impact potentials of algae systems. The lignocellulosic feedstocks are more often
preferred over first generation feedstocks, in general, to lower the land use impact.
Figure 15 Land use by soya based system.
The primary impact characterisations shown in Tables 5-6 and the land use data can be applied to
similar other systems.
3.5. Parametric Sensitivity Analysis
The grid electricity has the largest impact in the algal based ER production system (Table 5).
However, 1 MJ decrease in electricity input to the system has little effect in terms of reduction in
the primary impact characterisations. For LCA, it is more useful to have a measure of primary
impact characterisations due to unit electricity / unit mass of material used, shown in Table 8. Such
data will then show the effect on the primary impact characterisations due to reduction in unit
material or energy used by a system. This way, unit change in nitrogen (in the form of ammonium
nitrate) flowrate, the largest avoided impacts in acidification, eutrophication, freshwater aquatic
ecotoxicity, global warming, marine aquatic ecotoxicity and terrestric ecotoxicity potentials can be
shown. For human toxicity and abiotic depletion potentials, the largest benefit can be achieved by
reducing triethanolamine input to the system. For ozone layer depletion and photochemical ozone
creation potentials, the largest benefit per unit flowrate reduction can be obtained from the
reduction of hexane input. Thus, reducing the total electricity use, increasing the generation of
renewable electricity and lowering the requirements for ammonium nitrate, triethanolamine and
hexane must be investigated for algal system optimisation.
Table 8 Primary impact characterisations due to unit of electricity or unit mass of material used.
Wood CHP Ammonium nitrate phosphate Ammonium nitrate Formic acid

per MJ electricity per kg P2O5 per kg N per kg
Acidification potential kg SO2 equivalent 0.0001 0.0104 0.0283 0.0091
Eutrophication Potential kg phosphate equivalent 0.0000 0.0025 0.0105 0.0016
Freshwater aquatic ecotoxicity potential kg DCB equivalent 0.0006 0.1907 0.3407 0.1288
Global warming potential (GWP 100 years) kg CO2 equivalent 0.0036 1.2717 8.6864 2.5097
Human toxicity potential kg DCB equivalent 0.0046 0.4340 1.9787 0.5680
Marine aquatic ecotoxicity potential kg DCB equivalent 1.20 910.12 1185.60 820.05
Ozone Layer Depletion Potential kg R11 Equivalent 3.1704E-10 1.5001E-07 4.1233E-07 4.5821E-07
Photochemical ozone creation potential kg ethene equivalent 7.3853E-06 9.2101E-04 1.4112E-03 1.4365E-03
Terrestric ecotoxicity potential kg DCB equivalent 0.0003 0.0161 0.0576 0.0137
Abiotic depletion potential kg Sb equivalent 0.0000 0.0090 0.0265 0.0273
Hexane Hydrogen peroxide (50%) Potassium nitrate Triethanolamine Grid electricity

per kg per kg per kg K2O per kg per MJ electricity
Marine aquatic ecotoxicity potential kg DCB equivalent 302.25 481.04 264.25 1103.30 254.76
Photochemical ozone creation potential kg ethene equivalent 2.8420E-03 5.6518E-04 3.8695E-04 1.9593E-03 9.2593E-05
Single or multiple variable sensitivity analyses can be undertaken in a LCA based design problem.
When more than one independent or free parameter is involved in the design problem, the problem
is a multi-parametric or multi-variable problem. The problem defined for the plans ‘Epoxy resin
plant – avoided emission due to soya paste use’ and ‘Epoxy resin plant – avoided emission due to
algae paste use’ are examples of multi-variable problems.
The problem defined for the plan ‘Epoxy resin plant – avoided emission due to soya paste use’ has
9 variables, not all of them are independent or free variables, such as the two electricity flows, one
from the wood CHP system (F_CHP) and the other from the grid electricity system (F_grid). One of
the electricity flows can be independent, F_CHP, while the other, F_grid is a dependent parameter.
The two variables are related, shown in Equation 1. The dependent variable, F_grid, is related to the
independent variable, F_CHP, by Equation 1.
F_grid = 963 + (80 – F_CHP) (1)
Hence, with increasing CHP generation, electricity drawn from the grid reduces, while the total
additional electricity requirement by the 435 kg algal paste to ER production system (compared to
the equivalent soya based system) remains constant at (963+80) MJ.
Let’s assume that the other 7 additional material input flows required by the algal system are also
independent variables and examine the effect of their changes in flows from their base values on
each environmental impact characterisation. The symbols used for the various material flow
variables considered as independent variables are as follows.
1. hexane, at plant: F_H
2. triethanolamine, at plant: F_TE
3. formic acid, at plant: F_FA
4. hydrogen peroxide, 50% in H2O, at plant: F_HP
5. ammonium nitrate phosphate: F_P
6. ammonium nitrate: F_N
7. potassium nitrate: F_K
Table 9 shows the linear correlation coefficient for sensitivity analysis of each independent flow
from its base value shown in Table 4. Each one of them is examined to assess its effect on each
primary impact characterisation, when all other independent flows are kept constant. This approach
is called single variable sensitivity analysis.
Unlike the primary impact characterisations due to unit electricity or material requirement shown in
Table 8, the data in Table 9 account for the combined effects of total flowrates and impact
characterisation factors of individual materials. Thus, it can be seen that the largest avoided impacts
can be achieved by reducing hexane input. In global warming and terrestric ecotoxicity potentials,
reduction of ammonium nitrate is more beneficial. When the total flows are taken into account the
effect of reducing hexane has maximum savings in most of the categories due to its high flowrate.
Positive slope or linear correlation coefficient indicates that reduction in that flow helps to increase
the environmental impact saving, while negative slope such as for F_CHP, indicates that its
reduction (hence by increasing F_grid) increases the environmental impact. Similarly, higher
correlation coefficient or (slope) indicates higher saving potential.
Table 9 Linear single variable sensitivity analysis correlation coefficient of each independent electricity or material flow from the base value shown in
Table 4 on each primary impact characterisation, when all other independent flows are kept constant.
F_H F_TE F_FA F_HP F_P F_N F_K F_CHP

Acidification potential kg SO2 equivalent 0.2540 0.0013 0.0095 0.0097 0.0320 0.1640 0.0047 -0.0436
Eutrophication Potential kg phosphate equivalent 0.6010 0.0020 0.0068 0.0057 0.0309 0.2480 0.0030 -0.0085
Freshwater aquatic ecotoxicity potential kg DCB equivalent 0.2280 0.0017 0.0109 0.0572 0.0482 0.1630 0.0150 -0.0396
Global warming potential (GWP 100 years) kg CO2 equivalent 0.1930 0.0020 0.0157 0.0192 0.0229 0.2910 0.0088 -0.0371
Human toxicity potential kg DCB equivalent 0.2290 0.0642 0.0103 0.0180 0.0234 0.2020 0.0148 -0.0364
Marine aquatic ecotoxicity potential kg DCB equivalent 0.1020 0.0009 0.0079 0.0123 0.0261 0.0643 0.0040 -0.0650
Ozone Layer Depletion Potential kg R11 Equivalent 0.7570 0.0011 0.0196 0.0122 0.0192 0.0999 0.0073 -0.0070
Photochemical ozone creation potential kg ethene equivalent 0.6940 0.0012 0.0100 0.0104 0.0191 0.0553 0.0043 -0.0171
Terrestric ecotoxicity potential kg DCB equivalent 0.1840 0.0019 0.0084 0.0157 0.0287 0.1950 0.0127 -0.0461
Abiotic depletion potential kg Sb equivalent 0.5570 0.0021 0.0159 0.0143 0.0158 0.0875 0.0066 -0.0250
A multi variable sensitivity analysis can be undertaken by Monte Carlo Simulation discussed in the
following section. With Monte Carlo simulation, independent variables within their specified
standard deviations from their base values (Table 4) can be randomly selected during a simulation
run. All the primary impact characterisations, acidification, eutrophication, freshwater aquatic
ecotoxicity, global warming (over 100 years), human toxicity, marine aquatic ecotoxicity, ozone
layer depletion, photochemical ozone creation and terrestric ecotoxicity potentials (due to output
emissions) and abiotic depletion potential (due to input) are calculated for the selected set of values
of independent variables. At the end of Monte Carlo simulation runs, the chances of occurrence of
each impact characterisation for different standard deviations from its mean value are counted.
3.6. Monte Carlo Simulation Combined LCA
Intergovernmental Panel on Climate Change (IPCC) Guidelines for National Greenhouse Gas
Inventories; Geneva, Switzerland, 2006, Monte Carlo simulation has been recommended to estimate
the combined uncertainty lies in the IPCC derived GHGs inventory. The methodology can be
adapted according to the problem definition and for uncertainty lies in all associated primary impact
categories. Chapter 5 of the textbook shows the steps involved in uncertainty analysis using Monte
Carlo simulation. For multivariable approach it is obvious to take a stochastic simulation approach
to increase effectiveness of search in the entire space of solution. The Monte Carlo simulation has
three main steps:
1. selection of standard deviation and probability distribution function for each independent
variable;
2. Monte Carlo simulation runs;
3. estimation of probability distribution curve of each impact characterisation after a large number
of simulation runs.
The first two steps involve single variable sensitivity analysis. This is already shown in Tables 8-9
for the ER production problem. These two steps are essential to screen the independent variables for
which maximum impact reductions can be achievable.
LCA is data intensive. In addition, dispersed data set makes Monte Carlo simulation combined
LCA (MCLCA) computationally intensive. Additionally, there lies uncertainty in primary raw
material and energy flow data assimilation and eventually in inventory analysis. Large number of
Monte Carlo simulation runs ensures that the approximation can be made more accurate. As shown
in chapter 5 of the textbook, 5000 simulation runs, over the entire ±100% standard deviation in the
categories and presented over 50 clusters are recommended. Considering the data intensity of
MCLCA, the single variable sensitivity analyses shown in Tables 8-9 are thus essential to reduce
the number of uncertain free or independent variables. In the case of ‘Epoxy resin plant – avoided
emission due to soya paste use’, out of 8 independent variables, hexane and ammonium nitrate
flows can be selected out for the first round of MCLCA. If the chance or likelihood of reduction of
an impact potential is low (shown by narrow probability distribution curve), there is no need for
further investigation. For more significant impact reduction potentials (with wider probability
distribution curves), the next set of uncertain independent variables (e.g. total electricity
requirement) can be added to the list of variables for MCLCA.
Selection of standard deviation and probability distribution function for each uncertain and
independent variable
The standard deviation ( ) shows the normalised deviation of a dispersed dataset from its average
or mean value. Equation 2 shows the formula for calculating the standard deviation of n data points:
x1, x2, x3, … xn-1, xn, with an average of ̅ .
∑ ̅
√ (2)
Table 10 shows an explanation of the standard deviation calculation for hexane inventory for the
plan ‘Epoxy resin plant – avoided emission due to soya paste use’.
Assume that the average value of hexane flowrate for ‘Epoxy resin plant – avoided emission due to
soya paste use’ is 106 kg, equal to the base value, for 1000 kg ER production. For the sparse data
shown in Table 10, the standard deviation calculated using Equation 2 is 54 (n = 8).
Table 10 Hexane flowrate sparse data.
25 50 75 100 150 200 150 100
The sparse data distribution in Table 10 is shown in Figure 16. The mode of this data set is 100. The
median is 100 and the range is 175. Thus, from the base value, there is a scope for reduction of
hexane flowrate by 81 kg.
200
180
160
Hexane flowrate in kg
140
120
100
80
60
40
20
0
1 2 3 4 5 6 7 8
Figure 16 Sparse data distribution.
The uncertain and independent variables selected by screening by the single variable sensitivity
analyses are random varied within given ranges for Monte Carlo simulation. Values of these
variables can be generated using probability distribution function decided for each variable. The
simplest form of probability distribution function is the uniform probability distribution function.
Three other most common forms of probability distribution function are normal or Gaussian,
lognormal and triangular. Equations 3-5 show their respective correlations in terms of mean,
standard deviation and values of the variable (xi). The generic nature of the curves is shown in
Figure 17.
Normal or Gaussian probability distribution function =
̅
√
(3)
Lognormal probability distribution function =
̅
√
(4)
Triangular probability distribution function =
(5)
{ }
a = minimum value of x (for hexane flowrate sparse data given in Table 10, a = 25 kg)
b = modal value of x (for hexane flowrate sparse data given in Table 10, b = 100 kg)
c = maximum value of x (for hexane flowrate sparse data given in Table 10, c = 200 kg)
f(xi)
xi
f(xi)
a b c xi
Figure 17 Normalized or Gaussian; lognormal; triangular probability distribution functions.

Selection of standard deviation and probability distribution function of hexane mass flowrate
Using the hexane flowrate sparse data given in Table 10 in Equations 3-5, the Gaussian, lognormal
and triangular probability distribution functions are calculated shown in Table 11.
Table 11 Gaussian, lognormal and triangular probability distribution functions for sparse data given
in Table 10.
Sparse data points, x i , (hexane flowrate in kg) 25 50 75 100 100 150 150 200
Gaussian probability distribution function 0.00239 0.00429 0.00624 0.00732 0.00732 0.00532 0.00532 0.00164
Lognormal probability distribution function 33.2245 16.2135 10.6573 7.91347 7.91347 5.2021 5.2021 3.86304
Triangular probability distribution function 0 0.00381 0.00762 0.01143 0.01143 0.00571 0.00571 0
The Gaussian, lognormal and triangular probability distribution functions thus calculated in Table
11 are plotted with f(xi) in y-axis vs xi in x-axis, shown in Figures 18-20, respectively. Any of these
probability distribution functions can be used for the variable, hexane flowrate in kg; however, the
triangular distribution function captures the trend in sparse data set. GaBi provides Gaussian or
equal probability distribution functions for MCLCA.
probability distribution function
0.007
Normalised or Gaussian
0.006
0.005
0.004
0.003
0.002
0.001
25 45 65 85 105 125 145 165 185
Sparse data points (hexane flowrate in kg)
Figure 18 Gaussian probability distribution functions for the data set shown in Table 11.
Lognormal probability distribution function
30.001
25.001
20.001
15.001
10.001
5.001
0.001
25 45 65 85 105 125 145 165 185
Figure 19 Lognormal probability distribution functions for the data set shown in Table 11.
0.012
Triangular probability distribution function
0.01
0.008
0.006
0.004
0.002
0
25 45 65 85 105 125 145 165 185
Figure 20 Triangular probability distribution functions for the data set shown in Table 11.
Selection of standard deviation and probability distribution function of ammonium nitrate’s
nitrogen content in kg
Clearly, environmental benefits exist from the reduction of this fertiliser flowrate (Tables 8-9). The
following exercise problem can be solved to decide its standard deviation and probability
distribution function.
Exercise 1
Calculate the standard deviation for a set of sparse data on the additional mass flowrate (in kg) of
nitrogen input for algal ER production system, compared to soya ER production system: {5, 7, 10,
12, 13, 14, 15, 17}. Calculate the probability distribution functions using Equations 3-5 and show
the plots of the probability distribution functions with respect to the given sparse data points on the
additional mass flowrate (in kg) of nitrogen intake in the form of ammonium nitrate. Recommend a
rational probability distribution function for the given sparse data set.
Solution to Exercise 1
As environmental benefits exist from the reduction in nitrogen mass flowrate (in the form of
ammonium nitrate), the chosen data set is to show how much environmental benefits can be
obtained by the reduction in nitrogen mass flowrate from 17 kg in the base case down to 5 kg,
through 6 intermediate sparse data points. The mean of these data is 11.6 and the standard deviation
is 3.8.
Using the mean, standard deviation and given sparse data points in Equations 3-5, the probability
distributions for Gaussian and lognormal functions are calculated, shown in Table 12. There is no
modal point in the given data set hence the triangular probability distribution function can not be
applied to the given data set.
The plots of Gaussian and lognormal probability distribution functions with respect to the given
sparse data set (from Table 12) are shown in Figures 21-22. As Gaussian probability distribution
function is more usual than any others in MCLCA studies, this function is recommended as a
reasonably good guess for the given data set.
Table 12 Gaussian and lognormal probability distribution functions for sparse data given for
nitrogen (in the form of ammonium nitrate) flowrate.
Sparse data points, x i , (Ammonium nitrate (N) flowrate in kg) 5 7 10 12 13 14 15 17

Gaussian probability distribution function 0.02303 0.05008 0.09567 0.1043 0.09818 0.08626 0.07073 0.03866
Lognormal probability distribution function 60.8849 34.5967 19.1659 14.2194 12.4819 11.0674 9.89842 8.09057
0.11
probability distribution function 0.1

Normalised or Gaussian
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
5 7 9 11 13 15 17
Sparse data points (Ammonium nitrate (N) flowrate in kg)
Figure 21 Gaussian probability distribution functions for the data set shown in Table 12.
Lognormal probability distribution function
58
48
38
28
18
8
5 7 9 11 13 15 17
Sparse data points (Ammonium nitrate (N) flowrate in kg)
Figure 22 Lognormal probability distribution functions for the data set shown in Table 12.
3.7. Monte Carlo Simulation Runs
Once a standard deviation and an appropriate probability distribution function for each independent
variable are selected, these are user specified to carry out Monte Carlo simulation. A Monte Carlo
algorithm works on the principle of random number generations. In MCLCA case, a set of values of
chosen independent variables is randomly selected from their given standard deviations and
probability distribution functions during one simulation run. Then the LCIA is estimated.
Occurrence at a particular value of an environmental impact characterisation is noted. After a large
number of simulation runs the probability of occurrence in percentage of each environmental
impact characterisation are accounted.
The rule of thumb is the specification of a large number of simulation runs, e.g. 5000 as shown in
the IPCC Guidelines for National Greenhouse Gas Inventories; Geneva, Switzerland, 2006. This
way, uncertainty can be minimized and chances of occurrence or probability distributions of impact
potentials can be counted with adequate accuracy.
In the given case study on the ‘Epoxy resin plant – avoided emission due to soya paste use’, there
are two independent variables for Monte Carlo simulation, selected by using the single variable
sensitivity analyses, shown in Tables 8-9, hexane flowrate and nitrogen flowrate in the form of
ammonium nitrate. If their sparse data sets (containing 8 data points each) are known, there are 64
combinations. Thus, 5000 Monte Carlo simulation runs can repeat each combination 78 times on an
average. This large number of combinations allows generation of smooth and robust probability
distribution curves of impact potentials.
In an actual MCLCA study, such sparse data sets are unknown and rather a standard deviation and a
probability distribution function for each uncertain and independent variable are known or user
specified. Monte Carlo simulation algorithm allows random selection of a set of values of chosen
uncertain and independent variables using their standard deviations and probability distribution
functions. Thus, a large number of Monte Carlo simulation runs helps to have enough counts of
occurrences for various standard deviations from the mean value of each environmental impact
characterisation.
Probability distributions of each model predicted impact characterisation after a large

number of simulation runs
The two input specifications for MCLCA are the standard deviations and probability distribution
functions of uncertain and independent variables. The Gaussian or normalized probability
distribution function is the most powerful function to represent the nature of sparse data of
uncertain and independent variables for MCLCA study.
Some MCLCA packages such as GaBi require the standard deviation specification as a percentage
of the mean value. Table 13 shows a summary of the data input for MCLCA for the ‘Epoxy resin
plant – avoided emission due to soya paste use’ problem. The parametric specifications in GaBi
simulation is shown in Figure 23.
Table 13 Standard deviation as a percentage of mean value input for MCLCA for the ‘Epoxy resin
plant – avoided emission due to soya paste use’ problem in GaBi.
Base value Mean Standard deviation Standard deviation as a percentage of mean value
Hexane flowrate, kg 106 106 54 51
Nitrogen flowrate, kg 17 11.6 3.8 33
The other specifications are:
Probability distribution function for both variables = Gaussian distribution.
Number of Monte Carlo simulation runs = 5000
Number of clusters to show estimated results = 9

Deviation in estimated results from their base values = 25%
Selected impact characterisations are =
Freshwater aquatic ecotoxicity potential, kg DCB equivalent

Global warming potential (GWP 100 years), kg CO2 equivalent
Human toxicity potential, kg DCB equivalent
Marine aquatic ecotoxicity potential, kg DCB equivalent
Resource input flow, kg
Using the above specifications, the estimated reduction potentials using MCLCA in GaBi are
shown in Table 14. Thus, the reductions possible from the mean values to 10% Percentile values are
still not adequate to make an algal system competitive against a soya system (Table 7), shown as
follows. Maximum reduction potential in global warming is 52%.
Freshwater aquatic ecotoxicity potential, kg DCB equivalent 5.9

Global warming potential (GWP 100 years), kg CO2 equivalent 89.54
Human toxicity potential, kg DCB equivalent 29.03
Marine aquatic ecotoxicity potential, kg DCB equivalent 22860
Resource input flow, kg 3024
Table 14 Impact characterisations by percentiles using MCLCA in GaBi.
Base value Mean Standard deviation as a percentage of mean value 10% Percentile 25% Percentile Median 75% Percentile 90% Percentile
Freshwater aquatic ecotoxicity potential kg DCB equivalent 35.69 35.651 12.90% 29.79 32.594 35.615 38.755 41.507
Global warming potential (GWP 100 years) kg CO2 equivalent 518.46 518.41 13.70% 428.92 471.44 518.38 564.91 609.03
Human toxicity potential kg DCB equivalent 167.09 166.95 13.50% 138.06 152.02 166.67 182 195.99
Marine aquatic ecotoxicity potential kg DCB equivalent 313680 313510 5.65% 290820 301700 313410 325610 336200
Resource input flow kg 16232 16195 14.50% 13208 14644 16192 17799 19180
Table 15 Probability distributions of each impact characterisation.
Likelyhood of reduction by percentage (25 to 19)% (19 to 14)% (14 to 8)% (8 to 3)% (3 to -3)% (-3 to -8)% (-8 to -14)% (-14 to -19)% (-19 to -25)%
Freshwater aquatic ecotoxicity potential kg DCB equivalent 3.78% 7.66% 11.40% 16.20% 17.30% 15.20% 12.10% 7.14% 3.78%
Global warming potential (GWP 100 years) kg CO2 equivalent 4.10% 7.62% 11.50% 14.90% 16.70% 15.10% 11.40% 7.34% 4.40%
Human toxicity potential kg DCB equivalent 4.20% 8.02% 11% 15.30% 16.30% 14.90% 12.10% 7.16% 4.50%
Marine aquatic ecotoxicity potential kg DCB equivalent 0.02% 0.72% 6.30% 23.60% 38.40% 24.40% 5.96% 0.68% 0.02%
Resource input flow kg 4.84% 7.76% 11.30% 14.50% 15% 14.40% 11.40% 7.70% 4.66%
The chances or likelihoods of reduction by various percentage ranges in each impact
characterisation are shown in Table 15. The range from 25% reduction to 25% increase in impact
characterisations has been divided into 9 clusters according to specification. Positive percentage
ranges indicate reduction and negative percentage ranges indicate increase in each impact
characterisation. Each row in Table 15 shows the likelihoods of reductions by various percentage
ranges in each impact characterisation, totaling to 100 (however, only 9 clusters are shown, over
which the summation would be less than equal to 100).
Further observations can be made from the MCLCA results in Table 15, as follows.
1. The likelihood of impact characterisation reduction by upto about 14% from the mean value is
the maximum for the global warming potential category.
2. The likelihood of impact characterisation reduction by 19% from the mean value is more than the
likelihood of impact characterisation increase by 19% for all categories.
3. The likelihoods of varying the marine aquatic ecotoxicity potential beyond 8% from the mean
value are negligible.
4. The resource input flow directly controls all impact categories and less prominently the marine
aquatic ecotoxicity potential (the least sensitive impact category).
5. Minimizing the resource intake to the system within given constraints is likely to lower the
primary impact characterisations.
The probability distribution curve of each primary impact characterisation is shown in the snapshot
of GaBi window in Figure 24.
Figure 23 Parametric specifications for MCLCA study of Epoxy resin plant – avoided emission due
to soya paste use’ plan in GaBi.
Figure 24 Probability distribution curve of each impact characterisation in the snapshot of GaBi
window.
3.8. Combination of Values of Independent Variables for Various Percentage Reductions in

Impact Characterisations
By the MCLCA programme in GaBi, the values of the independent variables are generated
randomly and for each set of values of the independent variables, the LCA results are estimated. In
this section, the aim is to show the various combinations of values of the independent variables for
reducing the global warming potential from its base value, as an example. The independent
variables are the controlling variables for desired results. Therefore, their values need to be known
to reduce the impact characterisations.
Table 12 shows assumed sparse data set for the nitrogen (in the form of ammonium nitrate) mass
flowrate. For each value of nitrogen mass flowrate in Table 2, the value of hexane mass flowrate
can be determined, using the global warming potential reduction in kg CO2 equivalent per kg of
hexane flowrate (0.9197) shown in Table 8. Table 16 shows the evaluations of hexane mass
flowrates from various values of nitrogen mass flowrate for given reductions from the base value of
global warming potential. The reduced mass flowrate of hexane needed for 25% reduction in the
global warming potential from the base value of 518.46 kg CO2 equivalent (Table 14), for 5 kg
nitrogen input, is calculated as follows.
Reduction in the global warming potential in kg CO2 equivalent
Basis mass flowrate of nitrogen = 17 kg
Global warming potential reduction in kg CO2 equivalent per kg of nitrogen mass flowrate = 8.6864
(Table 8)
Reduction in the global warming potential in kg CO2 equivalent due to reduction in nitrogen mass
flowrate by (17-5) kg is
Reduced value of hexane mass flowrate for the reduction in the global warming potential by
(129.615 – 104.24) kg CO2 equivalent
kg
Table 16 Hexane mass flowrates from various values of nitrogen mass flowrate for given reductions
in the base value of global warming potential.
GWP (100 years) Hexane mass flowrate Nitrogen mass flowrate

reduction by kg kg
3 13
12 12
25% 31 10
60 7
78 5
18 15
27 14
19% 37 13
46 12
65 10
93 7
46 15
55 14
14% 65 13
74 12
93 10
8% 89 14
99 13
4. Summary
This chapter shows the calculations of marginal environmental impacts of replacing soya oil
feedstock by algal oil feedstock for the production of epoxy resins. The calculations have been
shown in terms of avoided impacts by the two feedstocks. Hence, environmentally more benign
feedstock will have more avoided impacts. The analysis shows that soya oil use as a feedstock is
environmentally more benign than algal oil feedstock; however, the former feedstock uses more
land as algae can be grown in enclosed photo bioreactor. It shows how independent variables can be
selected or prioritised to examine their effects on the environmental impacts. The various standard
deviation forms, ranges and how these can be decided to change the values of independent variables
for sensitivity analysis of impact potentials are shown. The results of Monte Carlo simulation are to
show the probability distribution curves of the impact characterisations for various percentage
standard deviations from their mean values. Of particular interest, is the need to identify important
inventory reductions to lower the impact characterisations from the lignocellulose derived products,
so that these impact characterisations are comparable against those from the first generation crops.
If such reductions in inventories are possible by advancing lignocellulose conversion technologies,
there will be reduction in not only environmental impacts, but also in land use impacts.
Readers are asked to work on the following for extending this research.
5. Write the pros and cons of various routes in terms of environmental impact characterisations.
6. Write recommendations (e.g. ranking of routes under various impact characterisations) with
evidences.
7. Identify the independent variables of epoxy resin production systems. Undertake and show the
dominance and contribution analyses results explained in chapter 5 of the textbook using
appropriate figures. Repeat the Monte Carlo simulations and report the probability distribution
results of LCIA, for ±25% standard deviation in free parameters, 5000 simulation runs, over the
entire ±100% standard deviation in the categories using 50 clusters, as shown in chapter 5 of the
textbook.

Case_Study_2

Uploaded by

Copyright:

Available Formats

Case_Study_2

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Case_Study_2

Uploaded by

Copyright:

Available Formats

Case Study 2: Comparison between Epoxy Resin Productions from Algal or Soya Oil –

An LCA Based Problem Solving Approach

Hexane to feedstock Formic acid: 0.224 Phthalic anhydride

Electricity: 1.3 MJ times

Figure 1 ER plant chemical and energy requirements.

Figure 2 ER plant yields.

Soya paste Algae paste

2. Cultivation System Environmental Impact Analysis

Table 2 Biogenic energy analysis for 362 kg soya bean cultivation.

Carbon dioxide input

The other raw materials consumed per 1000 m2 land are:

• carbon dioxide: 5400 kg

• total N:P:K fertiliser (5:1:1): 192 kg

3. Approach to the Solution of the LCA Based Decision Making Problem

3.1. Approach to the LCA Study

Avoided inventory Avoided inventory

• Due to soy paste use

• Due to algae paste use

Each avoided primary impact characterisation by a feedstock can thus be determined.

Values of input flows to enter in the GaBi/Ecoinvent software

• Steam saving: 34 kg = 96 MJ (determined from enthalpy requirement to generate

• 28 kg 5:1:1 N:P:K fertiliser supply for algae cultivation:

• Ammonium nitrate phosphate for 9 kg P2O5 or 3.98 kg phosphorous

• Potassium nitrate 3.98 kg potassium or 4.8 kg K2O production can be used.

• The balance of nitrogen fertiliser required for algae cultivation is 17 kg

3.2. Model Basis

1. 106 kg hexane, at plant

2. 0.26 triethanolamine, at plant

3. 3 kg formic acid, at plant

4. 8 kg hydrogen peroxide, 50% in H2O, at plant

5. ammonium nitrate phosphate, as 9 kg P2O5 at regional storehouse

6. ammonium nitrate, as 17 kg N, at regional storehouse

7. potassium nitrate, as 4.8 K2O, at regional storehouse

8. 80 MJ electricity and 96 MJ heat cogeneration from wood

9. 963 grid electricity

10. 484 kg carbon dioxide emissions avoided by soya bean cultivation.

2. inventories associated with 362 kg soya bean cultivation

3. 783 kg carbon dioxide emissions avoided by algae cultivation.

3.3. Model Building in GaBi

2. Hold to draw a line reaching the sink process.

Shows all available information on environmental and

3.4. Impact Assessment Data Collection

Hexane Hydrogen peroxide Potassium nitrate Triethanolamine Grid electricity

Phthalic anhydride Soybeans

3.5. Parametric Sensitivity Analysis

Wood CHP Ammonium nitrate phosphate Ammonium nitrate Formic acid

Hexane Hydrogen peroxide (50%) Potassium nitrate Triethanolamine Grid electricity

F_grid = 963 + (80 – F_CHP) (1)

1. hexane, at plant: F_H

2. triethanolamine, at plant: F_TE

3. formic acid, at plant: F_FA

4. hydrogen peroxide, 50% in H2O, at plant: F_HP

5. ammonium nitrate phosphate: F_P

6. ammonium nitrate: F_N

7. potassium nitrate: F_K

F_H F_TE F_FA F_HP F_P F_N F_K F_CHP

3.6. Monte Carlo Simulation Combined LCA

2. Monte Carlo simulation runs;

Table 10 Hexane flowrate sparse data.

25 50 75 100 150 200 150 100