uncertainty in life cycle inventory data

Addressing uncertainty in LCI

data with particular emphasis on

variability in upstream

supply chains

By Christoph Koffler, Martin Baitz, Annette Koehler

PE INTERNATIONAL | January 2012

Whitepaper

1

ContentContentContentContent

Nomenclature .................................................................................................................... 1

1 Quantifying uncertainty in Life Cycle Inventories ............................................... 2

2 Aspects of data uncertainty due to variability in supply chains .......................... 4

2.1 Influence of varying import/production country for same technology ................ 5

2.2 Influence of varying technology in the same country ......................................... 7

2.3 Coefficients of variation ................................................................................... 9

3 Summary ....................................................................................................... 10

Annex A - known technology and unknown country of origin ............................................... I

Annex B - unknown technology and known country of origin ............................................. XI

Credits

Cover graphics: i-stockphoto

NomenclatureNomenclatureNomenclatureNomenclature

AP Acidification Potential

EP Eutrophication Potential

GWP Global Warming Potential

LCA Life Cycle Assessment

PED Primary Energy Demand (non-renewable)

POCP Photochemical Ozone Creation Potential

2

1111 Quantifying uncertainty in Quantifying uncertainty in Quantifying uncertainty in Quantifying uncertainty in Life Life Life Life

Cycle Inventories Cycle Inventories Cycle Inventories Cycle Inventories

Uncertainty in LCA can be split into two parts:

• Data uncertainty (the uncertainty of

the modeled, measured, calculated,

estimated data within each unit pro-

cess as such).

• LCI model uncertainty (uncertainty in-

troduced in the results of a life cycle

inventory analysis due to the cumula-

tive effects of model imprecision, in-

put uncertainty and data variability).

Uncertainty in LCA is usually related to meas-

urement errors determination of the relevant

data, e.g., consumption or emission figures.

Since the ‘true’ values (especially for back-

ground data) are often unknown, it is virtually

impossible to completely avoid uncertain data

in LCA. These uncertainties then propagate

through the model and show in the final re-

sult. Small uncertainties in input data may

have a large effect on the overall results, while

others will be diminished along the way. This

article addresses PE International’s recom-

mendations for addressing the quantification

of uncertainty in an LCA study, and how this

can be done practically and with reasonable

accuracy.

Quantifying the uncQuantifying the uncQuantifying the uncQuantifying the uncertainty of primary ertainty of primary ertainty of primary ertainty of primary

data points on company specific processes data points on company specific processes data points on company specific processes data points on company specific processes

can be relatively straight forward and easy can be relatively straight forward and easy can be relatively straight forward and easy can be relatively straight forward and easy

for a company to calculate using the mean for a company to calculate using the mean for a company to calculate using the mean for a company to calculate using the mean

value and its standard deviation over a cevalue and its standard deviation over a cevalue and its standard deviation over a cevalue and its standard deviation over a cer-r-r-r-

tain number of tain number of tain number of tain number of data pointsdata pointsdata pointsdata points.... The number of The number of The number of The number of

data points and their ddata points and their ddata points and their ddata points and their date and mode of ate and mode of ate and mode of ate and mode of

measurement should then be documented measurement should then be documented measurement should then be documented measurement should then be documented

for full transparency. for full transparency. for full transparency. for full transparency. 1111

1Unfortunately, this information is often not dis-

closed for reasons of confidentiality, so most

But quantifying the uncertainty in the back-

ground systems (hundreds of upstream pro-

cesses including mining, extraction, refining,

etc.) and then performing error propagation

calculation is typically neither practical nor

feasible due to cost and time constraints in an

industrial setting. In addition, one should be

wary of data with seemingly precise uncer-

tainty values to each inventory flow, as these

are usually best estimates rather than having

been calculated with the accuracy that those

values imply.

A common rule of thumb estimates that the

best achievable uncertainty in LCA to be

around 10%. This was supported by a 2005

Ph.D. thesis on the forecast of environmental

impacts in the design of chemical equipment

(Kupfer, T. Ph.D. 2005). Nevertheless, the ac-

tual degree of uncertainty can vary signifi-

cantly from study to study.

The overarching question that really needs to

be answered therefore is:

How robust is my overall result when taking

into account the combined data and LCI model

uncertainties?

The effort to come up with a reasonable esti-

mate can be significantly reduced by following

a two-step approach:

1) Understand the model structure and its

dependencies

Keep it simple at first and start by setting up

your model with the values you have. Then try

to develop an understanding of the most rele-

vant aspects of your LCA model, i.e., of those

life cycle phases, contributors, or data points

industry data is based on yearly averages with

little or no indication of the variance.

3

that have the largest impact on your result.

This is usually done by a contribution or ‘hot

spot’ analysis and a subsequent sensitivity

analysis. Both of these functions are available

to GaBi users in the LCA balance sheet

through the Weak Point Analysis and the GaBi

Analyst.

Here is an example: the contribution or ‘hot

spot’ analysis of an energy-using product may

show that the use phase is dominating the life

cycle greenhouse gas emissions, closely fol-

lowed by the production of a printed circuit

board and logistics. Sensitivity analyses may

then show that the parameters that influence

these contributors the most are the split be-

tween online and stand-by mode during use,

the amount of precious metals in the circuit

board, and the distance from the Asian pro-

duction facility to the local distribution center.

2) Test the robustness of your model results

The next step then is to focus your efforts on

estimating the level of uncertainty of each of

the identified key parameters. Do some more

research to establish upper upper upper upper and lower and lower and lower and lower

boundsboundsboundsbounds for the relevant parameters. Theoreti-

cal min/max values, literature values, etc. can

provide additional insights here. The higher

the uncertainty, the larger these intervals will

be. You may even be able to find data that

allows for the cacacacalculation of a standard dlculation of a standard dlculation of a standard dlculation of a standard de-e-e-e-

viationviationviationviation or confidence intervalsconfidence intervalsconfidence intervalsconfidence intervals.

You can then assess the combined effect of

these uncertainties using the MonteMonteMonteMonte----Carlo Carlo Carlo Carlo

simulationsimulationsimulationsimulation available in the GaBi Analyst. By

defining uncertainty intervals around your key

parameters, the Monte-Carlo simulation is

able to produce a statistical estimate (mean produce a statistical estimate (mean produce a statistical estimate (mean produce a statistical estimate (mean

value) of the end result (e.g., X kg of COvalue) of the end result (e.g., X kg of COvalue) of the end result (e.g., X kg of COvalue) of the end result (e.g., X kg of CO2222

equivalents) as well as its standard deviequivalents) as well as its standard deviequivalents) as well as its standard deviequivalents) as well as its standard devia-a-a-a-

tiontiontiontion across all simulation runs. To do this, it

simply draws random numbers from the de-

fined intervals and calculates a single result

using that set of numbers. By repeating this

procedure a multitude of times (10,000 runs is

usually a good choice), it will produce a prob-

ability distribution based on 10,000 individual

results. The lower the standard deviatThe lower the standard deviatThe lower the standard deviatThe lower the standard deviation ion ion ion

associated with it, the more associated with it, the more associated with it, the more associated with it, the more certain certain certain certain or ‘or ‘or ‘or ‘prprprpre-e-e-e-

cisecisecisecise’ your result is.’ your result is.’ your result is.’ your result is. If the upper and lower

bounds as well as the probability distribution

in between were chosen correctly, then tttthe he he he

resulting mean value is also closer to the resulting mean value is also closer to the resulting mean value is also closer to the resulting mean value is also closer to the

‘real’ value‘real’ value‘real’ value‘real’ value, i.e., more , i.e., more , i.e., more , i.e., more ‘accurate’‘accurate’‘accurate’‘accurate’ than the

value you get when doing a simple balance

calculation based on your basic parameter

settings (see below).

If you want to make the assessment even

more robust towards any additional, unknown

uncertainties, you may increase the asceincrease the asceincrease the asceincrease the ascer-r-r-r-

tained intervalstained intervalstained intervalstained intervals around your key parameters

by a certain ‘‘‘‘safety factorsafety factorsafety factorsafety factor’’’’ of, e.g., +/- 10 %,

+/- 20 %, etc. If these additional uncertainties

do not affect the standard deviation across

the 10,000 runs, it is again an indicator of the

robustness of your results.

4

2222 Aspects of data uncertainty due Aspects of data uncertainty due Aspects of data uncertainty due Aspects of data uncertainty due

to variability in supply chainsto variability in supply chainsto variability in supply chainsto variability in supply chains

While chapter 1 addressed data and LCI model

uncertainty assuming that the practitioner

has been able to select the most appropriate

or ‘representative’ datasets for the product

system under study, this chapter will attempt

to quantify relevant aspects of variability in

background data due to its technological and

geographical representativeness.

As already stated in the previous chapter, +/-

10 % uncertainty seems to be the minimum

overall uncertainty, even if the model is set up

with high quality data containing low errors.

The model’s degree of representativeness

regarding supply chains and technology

routes depends on the specific situation under

consideration. It varies due to specific supplier

companies, geographical / national import

situations, etc.

How well How well How well How well the background data the background data the background data the background data matchesmatchesmatchesmatches the the the the

specific situationspecific situationspecific situationspecific situation at handat handat handat hand can only be acan only be acan only be acan only be an-n-n-n-

swered by doing a primary data collswered by doing a primary data collswered by doing a primary data collswered by doing a primary data collection ection ection ection

for for for for each each each each specific supply situation and specific supply situation and specific supply situation and specific supply situation and then then then then

compare it with the average situationcompare it with the average situationcompare it with the average situationcompare it with the average situation rerererep-p-p-p-

resented by resented by resented by resented by thethethethe background databackground databackground databackground data....

The background data as such may be very

precise and of extreme high representative-

ness within the situation where it was once

set up. The aim of this chapter is to estimate-

possible variations in background data due to

the mismatch between the average and actual

supply chain in a specific situation. To do so,

two types of possible misrepresentation in-

troduced by the user of the data are assessed:

• the influence of varying the im-

port / production country, and

• the influence of varying the technolo-

gy route in the same country to sup-

ply the same material or substance.

The analysis focuses on chemical products and

their intermediate products.

Disclaimer:Disclaimer:Disclaimer:Disclaimer:

The following analysThe following analysThe following analysThe following analyseeees s s s areareareare specific to the specific to the specific to the specific to the

products and datasets available in the products and datasets available in the products and datasets available in the products and datasets available in the 2006200620062006

releases of the releases of the releases of the releases of the GaBi databasesGaBi databasesGaBi databasesGaBi databases, Service Pack , Service Pack , Service Pack , Service Pack

17171717. The results cannot be generalized to . The results cannot be generalized to . The results cannot be generalized to . The results cannot be generalized to

other products other products other products other products or data sources.or data sources.or data sources.or data sources.

5

2.12.12.12.1 Influence of varying import/production Influence of varying import/production Influence of varying import/production Influence of varying import/production country for same technologycountry for same technologycountry for same technologycountry for same technology

The following chemical substances were analyzed regarding their variability with regard to their ge-

ography.

TableTableTableTable 2: Chemical substance datasets available for various countries in GaBi2: Chemical substance datasets available for various countries in GaBi2: Chemical substance datasets available for various countries in GaBi2: Chemical substance datasets available for various countries in GaBi

Acetic acid from methanol Hydrogen (Steamreforming fuel oil s)

Acetone by-product phenol methyl styrene (from Cumol) Hydrogen (Steamreforming natural gas)

Adipic acid from cyclohexane Maleic anhydride (MA) by-product PSA (by oxidation of

xylene)

AH-salt 63% (HMDA via adipic acid) Maleic anhydride from n-butane

Ammonium sulphate by-product caprolactam Methyl methacrylate (MMA) spent acid recycling

Benzene (from pyrolysis gasoline) Methyl methacrylate (MMA) from acetone and hydrogen

cyanide

Benzene (from toluene dealkylation) Methylene diisocyanate (MDI) by-product hydrochloric

acid, methano

Benzene by-product BTX (from reformatee) Phenol (toluene oxidation)

Caprolactam from cyclohexane Phenol from cumene

Caprolactam from phenol Phosphoric acid (wet process

Chlorine from chlorine-alkali electrolysis (amalgam) Phthalic anhydride (PAA) (by oxidation of xylene)

Chlorine from chlorine-alkali electrolysis (diaphragm) Propylene glycol over PO-hydrogenation

Chlorine from chlorine-alkali electrolysis (membrane) Propylene oxide (Cell Liquor)

Ethanol (96%) (hydrogenation with nitric acid) Propylene oxide (Chlorohydrin process)

Ethene (ethylene) from steam cracking Propylene oxide by-product t-butanol (Oxirane process)

Ethylbenzene (liquid phase alkylation) p-Xylene (from reformate)

Ethylene glycol from ethene and oxygen via EO Toluene (from pyrolysis gasoline)

Ethylene oxide (EO) by-product carbon dioxide from air Toluene by-product BTX (from reformate)

Ethylene oxide (EO) by-product ethylene glycol Toluene by-product styrene

Hexamethylene diamine (HMDA) via adipic acid Toluene diisocyanate (TDI) by-product toluene diamine,

hydrochloric acid (phosgenation)

Hydrochloric acid by-product methylene diisocyanate (MDI) Xylene mix by-product benzene (from pyrolysis gasoline)

6

These routes were analyzed (as available)

concerning boundary conditions in various

countries like:

Australia (AU), Belgium (BE), China (CN), Ger-

many (DE), Spain (ES), France (FR), Great Brit-

ain (GB), Italy (IT), Japan (JP), Netherlands (NL),

Norway (NO), Thailand (TH), Unites States (US)

The following figure shows the resulting max-

imum variations of all analyzed materials and

substances for selected impact categories. The

respective technologies are kept constant and

only the country of origin is varied. The figure

shows the maximum variability across the

various chemicals analyzed, as well as the

90 % and 10 % percentiles.

Two cases were calculated for each route,

assuming that the actual location of the sup-

plier is unknown in a given LCA project: choos-

ing the data set with the lowest burden while

the one with the highest burden would have

been appropriate (‘choose min’; relative er-

ror = (min-max)/max) and vice versa (‘choose

max’; relative error = (max-min)/min). The

resulting values are therefore the relative

‘‘‘‘worstworstworstworst----case case case case errorserrorserrorserrors’’’’ based on the considered

data sets.

Figure 1: Figure 1: Figure 1: Figure 1: Maximum Maximum Maximum Maximum relative relative relative relative errorserrorserrorserrors regarding randomly chosen geographregarding randomly chosen geographregarding randomly chosen geographregarding randomly chosen geographyyyy

Figure 1 shows that when assuming that the

technology route for a certain substance is

known and the specific country of origin route

is not, the maximum uncertainty of the relat-

ed impacts is between between between between ----65656565 % and +189% and +189% and +189% and +189 % for % for % for % for

80808080 % of all chemical substances % of all chemical substances % of all chemical substances % of all chemical substances for which

different country-specific data sets are availa-

ble in the GaBi database.

Some of the analyzed substances seem to be

highly sensitive concerning their geographic

reference. The per-substance results can be

found in Annex A. These individual errors can

PED AP EP GWP POCP

10% percentile -21% -65% -56% -41% -59%

choose min -68% -95% -79% -82% -93%

choose max 209% 1870% 380% 461% 1288%

90% percentile 27% 189% 129% 70% 143%

-200%

-100%

0%

100%

200%

300%

400%

500%

7

be applied to specific studies to estimate the

sensitivity of the overall result.

Summarized it can be said that when taking

the background information of the GaBi Mas-

terDB in to account, the sensitivity concerning

country of origin seems to be more relevant

for process chains where energy and the re-

spective emissions in energy supply are domi-

nating the impacts. However, in selected cases

country specific emission or efficiency of the

synthesis as such and differences in country

specific upstream supply are also relevant.

2.22.22.22.2 Influence of Influence of Influence of Influence of varying varying varying varying technology in the same countrytechnology in the same countrytechnology in the same countrytechnology in the same country

The following chemical substances were analyzed regarding their variability with regard to their

technology route in the same country.

Table 2: Chemical substance datasets available for various technology routes in GaBiTable 2: Chemical substance datasets available for various technology routes in GaBiTable 2: Chemical substance datasets available for various technology routes in GaBiTable 2: Chemical substance datasets available for various technology routes in GaBi

Chlorine from chlorine-alkali electrolysis diaphragm Ethylene-t-Butylether from C4 and bio ethanol

Chlorine from chlorine-alkali electrolysis membrane Hexamethylene diamine via Adiponitrile

Chlorine from chlorine-alkali electrolysis amalgam Hexamethylene diamine via adipic acid

Acetic acid from vinyl acetate Hydrochloric acid primary from chlorine

Acetic acid from methanol Hydrochloric acid by-product allyl chloride

Acrylamide catalytic hydrolysis Hydrochloric acid by-product chlorobenzene

Acrylamide enzymatic hydration Hydrochloric acid by-product epichlorohydrine

AH salt 63% HMDA from adipic acid Hydrochloric acid by-product Methylene diisocya-

nate

AH salt 63% HMDA from acrylonitrile Hydrogen Cracker

Ammonium sulphate by-product acetone cyanhydrin Hydrogen Steamreforming fuel oils

Ammonium sulphate by-product Caprolactam Hydrogen Steamreforming natural gas

Benzene from pyrolysis gasoline Maleic anhydride from n-butane

Benzene from toluene dealkylation Maleic anhydride by-product phthalic anhydride

Benzene by-product BTX Maleic anhydride from benzene

Benzene by-product ethine Methyl methacrylate from acetone and hydrogen

cyanide

Butanediol from ethine, H2 Cracker, allotherm Methyl methacrylate spent acid recycling

Butanediol from ethine H2 Steam ref. natural gas,

autotherm

Oleic acid from palm oil

Chlorodifluoroethane from 1,1,1-Trichloroethane Oleic acid from rape oil

Chlorodifluoroethane by-product Dichloro-1-

fluoroethane

Phenol by toluene oxidation

Dichlorpropane by-product epichlorohydrin Phenol by-product acetone

Dichlorpropane by-product dichlorpropane Phosphoric acid (54%)

Ethanol catalytic hydrogenation with phosphoric acid Phosphoric acid (100%)

Ethanol hydrogenation with nitric acid Propylene oxide Cell Liquor

Ethylene glycol by-product Ethylene oxide Propylene oxide Chlorohydrin process

8

Ethylene glycol from Ethene and oxygen via EO Propylene oxide Oxirane process

Ethylene glycol from Ethyleneoxide Toluene from pyrolysis gasoline

Ethylene oxide by-product carbon dioxide Toluene by-product BTX

Ethylene oxide by-product ethylene glycol via

CO2/methane

Toluene by-product styrene

Ethylene oxide by-product ethylene glycol via

CO2/methane with CO2 use

Xylene from pyrolysis gasoline

Ethylene-t-Butylether from C4 Xylene from reformate

The following figure shows the resulting max-

imum errors across all analyzed materials and

substances for selected impact categories.

Here, the respective countries of origin are

kept constant and only the technology route is

varied. The figure shows the maximum errors

across the various chemicals analyzed, as well

as the 90 % and 10 % percentiles.

Again, two cases were calculated for each

country, assuming that the actual technology

route of the supplier is unknown in a given

LCA project: choosing the technology-specific

data set with the lowest burden while the one

with the highest burden would have been

appropriate (‘choose min’; relative er-

ror = (min-max/max)) and vice versa (‘choose

max’; relative error = (max-min)/min). The

resulting values are therefore again the rela-

tive ‘worst‘worst‘worst‘worst----case case case case errorerrorerrorerrors’s’s’s’ possible based on the

available data sets.

Figure 2: Maximum Figure 2: Maximum Figure 2: Maximum Figure 2: Maximum relative relative relative relative errorserrorserrorserrors regarding randomly chosen regarding randomly chosen regarding randomly chosen regarding randomly chosen technologytechnologytechnologytechnology

PED AP EP GWP POCP

10% percentile -34% -57% -61% -71% -66%

choose min -96% -94% -93% -96% -96%

choose max 2409% 1596% 1332% 2609% 2731%

90% percentile 52% 132% 156% 248% 197%

-200%

-100%

0%

100%

200%

300%

400%

500%

9

Figure 2 shows that when assuming that the

country of origin for a certain substance is

known and the specific technology route is

not, the relative error of the related impacts is

between between between between ----71717171 % and +% and +% and +% and +248248248248 % for 80% for 80% for 80% for 80 % of all % of all % of all % of all

chemical substanceschemical substanceschemical substanceschemical substances for which different

technologies are available in the GaBi data-

base. When comparing the values to the ones

in chapter 2.1, it seems fair to state that it is

worse to not know the specific technology

route than the country of origin since all val-

ues are higher for the latter.

Yet again, some of the analyzed substances

seem to be highly sensitive concerning the

choice of technology. The per-substance re-

sults can be found in Annex B. These individu-

al values can then be applied to specific stud-

ies to estimate the sensitivity of the overall

result towards them.

2.32.32.32.3 Coefficients of variationCoefficients of variationCoefficients of variationCoefficients of variation

As seen in chapter 2.1 and 2.2, the maximum

relative error can easily reach several orders of

magnitude for the ‘choose max’ cases. These

numbers can be misleading, though, since

they heavily depend on the magnitude of the

respective denominator, i.e., the minimum

values. A more unbiased way to look at the

variability across the evaluated datasets is to

calculate the coefficientcoefficientcoefficientcoefficientssss of variationof variationof variationof variation across

the absolute indicator results, which is defined

as the standard deviation divided by the standard deviation divided by the standard deviation divided by the standard deviation divided by the

modulus of the modulus of the modulus of the modulus of the mean valuemean valuemean valuemean value. Due to the use

of the modulus, the coefficient is always a

positive value.

The following table displays the maximum

coefficients of variation across chemical pro-

duction datasets for each impact category

separately. Again, knowing the country of knowing the country of knowing the country of knowing the country of

origin but not knowing the specific tecorigin but not knowing the specific tecorigin but not knowing the specific tecorigin but not knowing the specific tech-h-h-h-

nology route can be considered worsenology route can be considered worsenology route can be considered worsenology route can be considered worse than

the opposite case. The coefficients of variation

are significantly higher for the latter case.

ImpactImpactImpactImpact known technology / unknown country known technology / unknown country known technology / unknown country known technology / unknown country

of originof originof originof origin

unknown technology / known country unknown technology / known country unknown technology / known country unknown technology / known country

of originof originof originof origin

PEDPEDPEDPED 32% 88%88%88%88%

APAPAPAP 92% 98%98%98%98%

EPEPEPEP 63% 123%123%123%123%

GWPGWPGWPGWP 47% 89%89%89%89%

POCPPOCPPOCPPOCP 86% 132%132%132%132%

10

3333 SummarySummarySummarySummary

The report at hand tried to answer two ques-

tions: first, how do I assess the uncertainty of

my LCI model with the GaBi software (chapter

1), and second, how large are the uncertainties

across different data sets assuming that either

the country of origin or the technology route

is not known.

While it is known that the LCI model uncer-

tainty can hardly be kept below 10% once the

most appropriate datasets have been chosen,

the uncertainty around this choice can be

significantly higher. For most of the consid-

ered datasets, the relative error is between -

75% and +250%, while the coefficient of varia-

tion is roughly between 90% and 130%.

Based on these results, the following conclu-

sions can be made:

1. Worry about the appropriate choice of da-

tasets before you worry about uncertainty on

elementary flow level. Especially the selection

of the most representative technology route

has a large influence on the resulting envi-

ronmental profile. The most ‘certain’ dataset

can introduce a massive error to your model if

it is not representative to the process / prod-

uct at hand.

2. When the most representative datasets

have been identified and deployed, worry

about the accuracy of your model structure

and parameter settings. Here the described

functionalities of the GaBi Analyst can help

you understand the dependencies and assess

the overall effect on your results.

PE INTERNATIONAL AG

Hauptstraße 111 - 113

70771 Leinfelden - Echterdingen

Germany

Phone +49 711 341817 - 0

Fax +49 711 341817 - 25

www.pe-international.com

Authors

Dr. Christoph Koffler

Dr. Martin Baitz

Dr. Annette Koehler

Yijian Wu

About PE INTERNATIONAL

PE INTERNATIONAL is one of the world’s most experienced sustainability software, content and stra-

tegic consulting firms. With 20 years of experience and 20 offices around the globe, PE INTERNA-

TIONAL allows clients to understand sustainability, improve their performance and succeed in the

marketplace. Through market leading software solutions, Five Winds Strategic Consulting Services

and implementation methodologies PE INTERNATIONAL has worked with some of the world‘s most

respected firms to develop the strategies, management systems, tools and processes needed to

achieve leadership in sustainability.

I

Annex A - known technology and unknown country of origin

PED uncertainty for known technology and unknown country of origin

PEDPEDPEDPED (min(min(min(min----max)/maxmax)/maxmax)/maxmax)/max (max(max(max(max----min)/minmin)/minmin)/minmin)/min

Chlorine from chlorine-alkali electrolysis (diaphragm) -68% 209%

Chlorine from chlorine-alkali electrolysis (membrane) -52% 108%

Toluene by-product styrene -47% 89%

Phosphoric acid (wet process) -45% 83%

Caprolactam from phenol -39% 64%

Ammonium sulphate by-product caprolactam -37% 59%

Hydrochloric acid by-product methylene diisocyanate (MDI) -35% 54%

Maleic anhydride (MA) by-product PSA (by oxidation of

xylene)

-32% 47%

Hydrogen (Steamreforming natural gas) -21% 27%

Maleic anhydride from n-butane -21% 26%

Acetone by-product phenol, methyl styrene (from Cumol) -17% 21%

p-Xylene (from reformate) -16% 19%

Chlorine from chlorine-alkali electrolysis (amalgam) -16% 18%

Phenol from cumene -13% 15%

Acetic acid from methanol -12% 14%

-68%-52%-47%-45%-39%-37%-35%-32%-21%

-21%-17%-16%-16%-13%-12%-12%-10%

-9%-9%-9%-8%-8%-8%-8%

-8%-7%-6%-6%-5%-5%-5%-5%-5%-4%-4%-4%-3%-3%-3%

-2%-2%-1%

209%108%89%83%64%59%54%47%27%

26%21%19%18%15%14%14%11%9%9%9%9%9%9%9%

8%8%6%6%6%6%5%5%5%4%4%4%3%3%3%

2%2%1%

-200% -100% 0% 100% 200% 300% 400% 500%

Chlorine from chlorine-alkali electrolysis (diaphragm)

Chlorine from chlorine-alkali electrolysis (membrane)


Phosphoric acid (wet process)

Caprolactam from phenol

Ammonium sulphate by-product caprolactam

Hydrochloric acid by-product methylene diisocyanate (MDI)

Maleic anhydride (MA) by-product PSA (by oxidation of xylene)

Hydrogen (Steamreforming natural gas)

Maleic anhydride from n-butane

Acetone by-product phenol, methyl styrene (from Cumol)

p-Xylene (from reformate)

Chlorine from chlorine-alkali electrolysis (amalgam)

Phenol from cumene

Acetic acid from methanol

Hydrogen (Steamreforming fuel oil s)

Methylene diisocyanate (MDI) by-product hydrochloric acid, methano

Ethene (ethylene) from steam cracking

Toluene (from pyrolysis gasoline)

Phthalic anhydride (PAA) (by oxidation of xylene)

Toluene diisocyanate (TDI) by-product toluene diamine, hydrochloric acid (phosgenation)

Propylene oxide by-product t-butanol (Oxirane process)

Xylene mix by-product benzene (from pyrolysis gasoline)

Methyl methacrylate (MMA) spent acid recycling

Adipic acid from cyclohexane

Benzene (from pyrolysis gasoline)

Benzene by-product BTX (from reformatee)

Propylene glycol over PO-hydrogenation

Ethanol (96%) (hydrogenation with nitric acid)

Ethylene glycol from ethene and oxygen via EO

Benzene (from toluene dealkylation)

Toluene by-product BTX (from reformate)

Propylene oxide (Cell Liquor)

Propylene oxide (Chlorohydrin process)

Ethylene oxide (EO) by-product carbon dioxide from air

Methyl methacrylate (MMA) from acetone and hydrogen cyanide

Caprolactam from cyclohexane

Hexamethylene diamine (HMDA) via adipic acid

Ethylene oxide (EO) by-product ethylene glycol via CO2/methane

Phenol (toluene oxidation)

AH-salt 63% (HMDA via adipic acid)

Ethylbenzene (liquid phase alkylation)

(min-max)/max

(max-min)/min

II


Hydrogen (Steamreforming fuel oil s) -12% 14%

Methylene diisocyanate (MDI) by-product hydrochloric

acid, methano

-10% 11%

Ethene (ethylene) from steam cracking -9% 9%

Toluene (from pyrolysis gasoline) -9% 9%

Phthalic anhydride (PAA) (by oxidation of xylene) -9% 9%

Toluene diisocyanate (TDI) by-product toluene diamine,


-8% 9%

Propylene oxide by-product t-butanol (Oxirane process) -8% 9%

Xylene mix by-product benzene (from pyrolysis gasoline) -8% 9%

Methyl methacrylate (MMA) spent acid recycling -8% 9%

Adipic acid from cyclohexane -8% 8%

Benzene (from pyrolysis gasoline) -7% 8%

Benzene by-product BTX (from reformatee) -6% 6%

Propylene glycol over PO-hydrogenation -6% 6%

Ethanol (96%) (hydrogenation with nitric acid) -5% 6%

Ethylene glycol from ethene and oxygen via EO -5% 6%

Benzene (from toluene dealkylation) -5% 5%

Toluene by-product BTX (from reformate) -5% 5%

Propylene oxide (Cell Liquor) -5% 5%

Propylene oxide (Chlorohydrin process) -4% 4%

Ethylene oxide (EO) by-product carbon dioxide from air -4% 4%

Methyl methacrylate (MMA) from acetone and hydrogen

cyanide

-4% 4%

Caprolactam from cyclohexane -3% 3%

Hexamethylene diamine (HMDA) via adipic acid -3% 3%

Ethylene oxide (EO) by-product ethylene glycol via

CO2/methane

-3% 3%

Phenol (toluene oxidation) -2% 2%

AH-salt 63% (HMDA via adipic acid) -2% 2%

Ethylbenzene (liquid phase alkylation) -1% 1%

III

AP uncertainty for known technology and unknown country of origin

APAPAPAP (min(min(min(min----

max)/maxmax)/maxmax)/maxmax)/max

(max(max(max(max----min)/minmin)/minmin)/minmin)/min










Maleic anhydride (MA) by-product PSA (by oxidation of xylene) -63% 170%



Toluene diisocyanate (TDI) by-product toluene diamine, hydro-

chloric acid (phosgenation)

-58% 140%




-95%-94%-90%-89%-80%-79%-76%-68%-66%

-63%-63%-61%-58%-58%-57%-57%-56%-55%-55%-51%-49%-49%-49%-46%

-46%-45%-44%-42%-40%-40%-40%-39%-39%-36%-35%-34%-31%-31%-26%

-14%-11%

-7%

1870%1442%880%797%394%383%311%212%197%

170%170%153%140%136%135%132%127%124%121%106%98%96%95%85%

85%82%80%73%67%67%66%63%63%55%53%52%46%44%35%

16%12%8%

-200% -100% 0% 100% 200% 300% 400% 500%














Phenol from cumene





























(min-max)/max

(max-min)/min

IV

APAPAPAP (min(min(min(min----

max)/maxmax)/maxmax)/maxmax)/max

(max(max(max(max----min)/minmin)/minmin)/minmin)/min










CO2/methane

-46% 85%














Methyl methacrylate (MMA) from acetone and hydrogen cya-

nide

-26% 35%

Methylene diisocyanate (MDI) by-product hydrochloric acid,

methano

-14% 16%



V

EP uncertainty for known technology and unknown country of origin

EPEPEPEP (min(min(min(min----max)/maxmax)/maxmax)/maxmax)/max (max(max(max(max----min)/minmin)/minmin)/minmin)/min







Maleic anhydride (MA) by-product PSA (by oxidation of

xylene)

-66% 194%











-79%-79%-78%-74%-74%-72%-66%-60%-57%

-55%-53%-53%-52%-52%-50%-48%-46%-46%-45%-45%-43%-42%-40%-39%

-37%-37%-36%-36%-32%-32%-19%-18%-18%-17%-15%-15%-14%-13%

-9%

-7%-4%-1%

380%369%361%291%285%253%194%151%132%

123%114%113%108%106%101%94%85%85%82%80%74%74%68%65%

59%58%56%55%47%47%23%23%23%21%18%17%16%15%10%

8%4%1%

-200% -100% 0% 100% 200% 300% 400% 500%










Phenol from cumene

































(min-max)/max

(max-min)/min

VI




Toluene diisocyanate (TDI) by-product toluene diamine,


-45% 80%










methano

-32% 47%




Methyl methacrylate (MMA) from acetone and hydrogen

cyanide

-18% 23%






CO2/methane

-13% 15%





VII

GWP uncertainty for known technology and unknown country of origin

GWPGWPGWPGWP (min(min(min(min----max)/maxmax)/maxmax)/maxmax)/max (max(max(max(max----min)/minmin)/minmin)/minmin)/min



















-23% 30%

-82%-74%-70%-68%-65%-51%-47%-44%-43%

-36%-35%-31%-30%-27%-25%-25%-23%-22%-18%-18%-17%-17%-15%-15%

-15%-13%-12%-12%-12%

-9%-9%-9%-8%-8%-7%-5%-5%-4%-4%

-3%-2%0%

461%289%234%217%189%102%88%79%75%

57%53%44%43%37%33%33%30%28%22%22%21%20%18%18%

17%15%14%13%13%10%10%9%9%9%8%5%5%4%4%

3%2%0%

-200% -100% 0% 100% 200% 300% 400% 500%














Phenol from cumene





























(min-max)/max

(max-min)/min

VIII
















CO2/methane

-9% 10%








methano

-4% 4%

Methyl methacrylate (MMA) from acetone and hydrogen cya-

nide

-4% 4%



Caprolactam from cyclohexane 0% 0%

IX

POCP uncertainty for known technology and unknown country of origin

POCPPOCPPOCPPOCP (min(min(min(min----max)/maxmax)/maxmax)/maxmax)/max (max(max(max(max----

min)/minmin)/minmin)/minmin)/min


















-93%-89%-83%-82%-73%-72%-71%-70%-60%

-55%-53%-52%-48%-47%-46%-43%-43%-42%-40%-40%-38%-37%-37%-35%

-29%-26%-25%-24%-19%-19%-19%-16%-15%-13%-11%-10%-10%-10%

-8%

-5%-5%-4%

1288%804%483%464%267%262%249%230%152%

121%111%108%94%87%84%76%74%73%67%66%63%59%59%53%

42%35%33%32%24%24%23%20%18%15%13%12%11%11%8%

6%5%4%

-200% -100% 0% 100% 200% 300% 400% 500%

















Phenol from cumene


























(min-max)/max

(max-min)/min

X

POCPPOCPPOCPPOCP (min(min(min(min----max)/maxmax)/maxmax)/maxmax)/max (max(max(max(max----

min)/minmin)/minmin)/minmin)/min












-25% 33%




methano

-19% 24%








Ethylene oxide (EO) by-product ethylene glycol via CO2/methane -10% 11%



Methyl methacrylate (MMA) from acetone and hydrogen cyanide -5% 5%


XI

Annex B - unknown technology and known country of origin

PED uncertainty for unknown technology and known country of origin


JP: Hydrogen -96% 2409%

DE: Hydrochloric acid -91% 1004%

DE: Benzene -63% 171%

NO: Hydrogen -63% 171%

IT: Toluene -62% 166%

DE: Toluene -57% 134%

DE: Propylene oxide -55% 125%

US: Propylene oxide -54% 117%

US: Phosphoric acid -41% 70%

DE: Phosphoric acid -41% 69%

DE: Ethylene-t-Butylether (ETBE) -40% 66%

IT: Xylene -36% 56%

IT: Maleic anhydride -32% 48%

DE: Ethylene glycol -32% 48%

DE: Xylene mix -32% 47%

FR: Xylene mix -32% 46%

NL: Toluene -31% 45%

-96%-91%-63%-63%-62%-57%-55%-54%-41%-41%-40%-36%-32%-32%-32%-32%-31%-31%-29%-28%-28%-28%-28%-28%-27%-26%-24%-23%-21%-21%-19%-18%-18%-17%-17%-16%-16%-13%-13%-12%-12%-11%-11%-11%-11%-11%-10%

-9%-9%-8%-8%-7%-7%-6%-6%-1%-1%-1%

2409%1004%171%171%166%134%125%117%70%69%66%56%48%48%47%46%45%45%41%39%39%39%38%38%37%34%32%30%27%26%23%22%21%21%20%19%18%15%15%14%14%13%12%12%12%12%11%10%10%9%8%7%7%7%6%1%1%1%

-500% 0% 500% 1000% 1500% 2000% 2500% 3000%

JP: Hydrogen

DE: Hydrochloric acid

DE: Benzene

NO: Hydrogen

IT: Toluene

DE: Toluene

DE: Propylene oxide

US: Propylene oxide

US: Phosphoric acid

DE: Phosphoric acid

DE: Ethylene-t-Butylether (ETBE)

IT: Xylene

IT: Maleic anhydride

DE: Ethylene glycol

DE: Xylene mix

FR: Xylene mix

NL: Toluene

GB: Maleic anhydride (MSA)

US: Hydrogen (highly pure)

DE: Phenol

FR: Benzene

DE: Acetic acid

GB: Benzene

IT: Benzene

NL: Benzene

US: Toluene

US: Benzene

DE: Methyl methacrylate (MMA)

DE: Dichlorpropane

DE: Hexamethylene diamine (HMDA)

DE: Ethylene oxide (EO)

NO: Chlorine from chlorine-alkali electrolysis

DE: Acrylamide

FR: Hydrogen

FR: Hexamethylene diamine (HMDA)

GB: Hydrogen

DE: Oleic acid

NL: Hydrogen

DE: AH salt 63%

IT: Hydrogen

DE: Ethanol (96%)

DE: Hydrogen

NL: Chlorine from chlorine-alkali-electrolysis

US: Ethylene oxid (EO)

JP: Chlorine from chlorine-alkali electrolysis

DE: Chlor aus Chlor-Alkali-Elektrolyse

US: Chlorine from chlorine-alkali electrolysis

AU: Chlorine from chlorine-alkali electrolysis

FR: Chlorine from chlorine-alkali electrolysis

DE: Chlorodifluoroethane (HCFC 142b)

BE: Chlorine from chlorine-alkali electrolysis

ES: Chlorine from chlorine-alkali electrolysis

GB: Chlorine from chlorine-alkali electrolysis

IT: Chlorine from chlorine-alkali electrolysis

US: Hydrogen

DE: Butanediol

DE: Ammonium sulphate

DE: Epichlorohydrin

(min-max)/max

(max-min)/min

XII


GB: Maleic anhydride (MSA) -31% 45%

US: Hydrogen (highly pure) -29% 41%

DE: Phenol -28% 39%

FR: Benzene -28% 39%

DE: Acetic acid -28% 39%

GB: Benzene -28% 38%

IT: Benzene -28% 38%

NL: Benzene -27% 37%

US: Toluene -26% 34%

US: Benzene -24% 32%

DE: Methyl methacrylate (MMA) -23% 30%

DE: Dichlorpropane -21% 27%

DE: Hexamethylene diamine (HMDA) -21% 26%

DE: Ethylene oxide (EO) -19% 23%

NO: Chlorine from chlorine-alkali electrolysis -18% 22%

DE: Acrylamide -18% 21%

FR: Hydrogen -17% 21%

FR: Hexamethylene diamine (HMDA) -17% 20%

GB: Hydrogen -16% 19%

DE: Oleic acid -16% 18%

NL: Hydrogen -13% 15%

DE: AH salt 63% -13% 15%

IT: Hydrogen -12% 14%

DE: Ethanol (96%) -12% 14%

DE: Hydrogen -11% 13%

NL: Chlorine from chlorine-alkali-electrolysis -11% 12%

US: Ethylene oxid (EO) -11% 12%

JP: Chlorine from chlorine-alkali electrolysis -11% 12%

DE: Chlor aus Chlor-Alkali-Elektrolyse -11% 12%

US: Chlorine from chlorine-alkali electrolysis -10% 11%

AU: Chlorine from chlorine-alkali electrolysis -9% 10%

FR: Chlorine from chlorine-alkali electrolysis -9% 10%

DE: Chlorodifluoroethane (HCFC 142b) -8% 9%

BE: Chlorine from chlorine-alkali electrolysis -8% 8%

ES: Chlorine from chlorine-alkali electrolysis -7% 7%

GB: Chlorine from chlorine-alkali electrolysis -7% 7%

IT: Chlorine from chlorine-alkali electrolysis -6% 7%

US: Hydrogen -6% 6%

DE: Butanediol -1% 1%

DE: Ammonium sulphate -1% 1%

DE: Epichlorohydrin -1% 1%

XIII

AP uncertainty for unknown technology and known country of origin

APAPAPAP (min(min(min(min----max)/maxmax)/maxmax)/maxmax)/max (max(max(max(max----min)/minmin)/minmin)/minmin)/min












IT: Xylene -60% 148%






-94%-90%-82%-76%-76%-73%-71%-71%-70%-69%-63%-60%-54%-53%-52%-47%-43%-42%-41%-39%-39%-39%-38%-37%-36%-35%-35%-33%-33%-32%-32%-32%-31%-30%-26%-25%-24%-21%-18%-18%-17%-16%-14%-14%-13%-13%-12%-11%-10%-10%

-8%-7%-5%-5%-3%-2%-1%-1%

1596%867%467%320%312%271%250%245%232%226%174%148%116%114%107%90%75%72%70%65%64%64%62%59%57%54%54%50%49%48%47%46%44%43%36%34%32%26%22%22%20%19%17%16%15%14%14%12%11%11%9%7%5%5%3%2%1%1%

-200% 0% 200% 400% 600% 800% 1000% 1200% 1400% 1600% 1800%


JP: Hydrogen

IT: Toluene

DE: Benzene

US: Propylene oxide

DE: Toluene

GB: Hydrogen

DE: Propylene oxide


NL: Hydrogen


IT: Xylene

US: Toluene

GB: Benzene

US: Benzene

IT: Benzene


NO: Hydrogen

IT: Hydrogen

US: Phosphoric acid

DE: Phosphoric acid

FR: Xylene mix

FR: Hydrogen


DE: Hydrogen


DE: Acetic acid

DE: Phenol

NL: Toluene

DE: Xylene mix

DE: Ethylene glycol

DE: Dichlorpropane


NL: Benzene

FR: Benzene








DE: Ethanol (96%)

DE: Oleic acid





DE: AH salt 63%


DE: Acrylamide



US: Hydrogen



DE: Epichlorohydrin

DE: Butanediol

(min-max)/max

(max-min)/min

XIV

APAPAPAP (min(min(min(min----max)/maxmax)/maxmax)/maxmax)/max (max(max(max(max----min)/minmin)/minmin)/minmin)/min











DE: Phenol -33% 50%















DE: Ethanol (96%) -14% 17%






DE: AH salt 63% -10% 11%





US: Hydrogen -5% 5%





XV

EP uncertainty for unknown technology and known country of origin



















-93%-91%-87%-86%-81%-79%-79%-67%-67%-64%-62%-61%-61%-59%-58%-57%-55%-53%-51%-47%-46%-44%-44%-42%-42%-40%-40%-39%-37%-32%-32%-30%-26%-24%-24%-23%-23%-21%-19%-19%-18%-16%-15%-14%-13%-12%-11%

-9%-6%-6%-5%-5%-4%-2%-2%-1%0%0%

1332%961%664%619%439%382%372%205%199%175%163%157%156%144%139%130%124%111%102%90%84%80%77%74%71%67%66%64%59%48%46%44%36%32%31%30%29%27%24%23%22%19%18%16%14%13%12%10%7%6%5%5%4%2%2%1%0%0%

-200% 0% 200% 400% 600% 800% 1000% 1200% 1400% 1600%



DE: Propylene oxide

IT: Toluene

JP: Hydrogen

DE: Toluene

US: Propylene oxide

DE: Ethylene glycol

IT: Xylene

GB: Benzene

NO: Hydrogen

DE: Benzene


DE: Dichlorpropane

GB: Hydrogen

IT: Benzene


FR: Xylene mix

DE: Acetic acid

DE: Xylene mix

US: Toluene


US: Benzene

NL: Toluene

US: Phosphoric acid

DE: Phosphoric acid

FR: Benzene

NL: Hydrogen

NL: Benzene

DE: Oleic acid




FR: Hydrogen

DE: Phenol


DE: Ethanol (96%)










IT: Hydrogen

DE: AH salt 63%

DE: Hydrogen


US: Hydrogen


DE: Acrylamide




DE: Epichlorohydrin

DE: Butanediol

(min-max)/max

(max-min)/min

XVI



















DE: Phenol -24% 31%


DE: Ethanol (96%) -23% 29%











DE: AH salt 63% -9% 10%

DE: Hydrogen -6% 7%


US: Hydrogen -5% 5%






DE: Epichlorohydrin 0% 0%

DE: Butanediol 0% 0%

XVII

GWP uncertainty for unknown technology and known country of origin

-96%-91%-88%-86%-83%-82%-80%-79%-75%-74%-73%-71%-71%-70%-68%-66%-65%-64%-63%-63%-60%-59%-55%-54%-53%-53%-52%-51%-48%-47%-42%-42%-40%-38%-38%-29%-27%-26%-25%-23%-21%-21%-19%-19%-18%-17%-16%-14%-13%-13%-10%-10%

-7%-5%-4%-3%-2%0%

2609%968%758%620%495%445%411%372%299%282%266%251%245%234%216%195%183%175%168%167%150%143%122%117%113%112%108%105%91%90%74%73%65%61%60%40%38%34%33%30%27%26%24%24%22%21%19%17%15%15%11%11%8%5%4%3%2%0%

-500% 0% 500% 1000% 1500% 2000% 2500% 3000%

JP: Hydrogen


NO: Hydrogen

IT: Toluene

DE: Toluene

DE: Benzene

DE: Propylene oxide

NL: Hydrogen

DE: Hydrogen

GB: Hydrogen

FR: Hydrogen

US: Propylene oxide

IT: Hydrogen

DE: Oleic acid

IT: Xylene



FR: Xylene mix

US: Hydrogen

DE: Xylene mix

GB: Benzene

IT: Benzene

NL: Toluene

US: Toluene

NL: Benzene

FR: Benzene

US: Benzene


DE: Acetic acid

DE: Ethylene glycol

US: Phosphoric acid

DE: Phosphoric acid

DE: Dichlorpropane








DE: Acrylamide

DE: Ethanol (96%)


DE: Phenol


DE: AH salt 63%






DE: Epichlorohydrin

DE: Butanediol






(min-max)/max

(max-min)/min

XVIII




















US: Hydrogen -63% 168%















NO: Chlorine from chlorine-alkali elec-

trolysis

-38% 61%

FR: Chlorine from chlorine-alkali elec-

trolysis

-38% 60%







XIX

DE: Ethanol (96%) -21% 26%


DE: Phenol -19% 24%


DE: AH salt 63% -17% 21%

JP: Chlorine from chlorine-alkali elec-

trolysis

-16% 19%


US: Chlorine from chlorine-alkali elec-

trolysis

-13% 15%

NL: Chlorine from chlorine-alkali-

electrolysis

-13% 15%

AU: Chlorine from chlorine-alkali elec-

trolysis

-10% 11%



IT: Chlorine from chlorine-alkali elec-

trolysis

-5% 5%

GB: Chlorine from chlorine-alkali elec-

trolysis

-4% 4%

ES: Chlorine from chlorine-alkali elec-

trolysis

-3% 3%


BE: Chlorine from chlorine-alkali elec-

trolysis

0% 0%

XX

POCP uncertainty for unknown technology and known country of originPOCP uncertainty for unknown technology and known country of originPOCP uncertainty for unknown technology and known country of originPOCP uncertainty for unknown technology and known country of origin

-96%-96%-87%-80%-78%-76%-76%-73%-73%-72%-67%-67%-66%-66%-63%-61%-55%-54%-54%-53%-53%-52%-51%-50%-47%-44%-42%-40%-39%-39%-36%-34%-31%-29%-28%-26%-23%-22%-21%-21%-20%-18%-15%-13%-11%

-9%-9%-9%-9%-9%-8%-8%-8%-8%-3%-2%-1%-1%

2731%2381%669%403%351%311%310%276%272%260%202%199%194%193%169%158%121%118%116%114%112%108%105%101%90%80%72%66%65%64%56%51%44%40%38%35%29%29%27%27%25%22%18%15%12%10%10%10%10%9%9%9%9%8%4%2%1%1%

-500% 0% 500% 1000% 1500% 2000% 2500% 3000%

DE: Oleic acid

JP: Hydrogen


IT: Toluene

DE: Toluene


DE: Ethylene glycol

DE: Acetic acid

NL: Hydrogen



DE: Benzene


GB: Hydrogen


IT: Xylene

US: Toluene

NL: Toluene


NL: Benzene

US: Benzene

GB: Benzene

US: Propylene oxide

IT: Benzene

FR: Xylene mix

DE: Xylene mix

DE: Propylene oxide

US: Phosphoric acid

DE: Phosphoric acid



FR: Benzene


DE: Hydrogen

US: Hydrogen

NO: Hydrogen

DE: AH salt 63%





DE: Epichlorohydrin


FR: Hydrogen


DE: Acrylamide

DE: Ethanol (96%)







DE: Dichlorpropane


DE: Phenol

DE: Butanediol

IT: Hydrogen

(min-max)/max

(max-min)/min

XXI

POCPPOCPPOCPPOCP (min(min(min(min----max)/maxmax)/maxmax)/maxmax)/max (max(max(max(max----min)/minmin)/minmin)/minmin)/min































NO: Chlorine from chlorine-alkali elec-

trolysis

-36% 56%


FR: Chlorine from chlorine-alkali elec-

trolysis

-31% 44%


US: Hydrogen -28% 38%


DE: AH salt 63% -23% 29%

JP: Chlorine from chlorine-alkali elec-

trolysis

-22% 29%

NL: Chlorine from chlorine-alkali-

electrolysis

-21% 27%

XXII






US: Chlorine from chlorine-alkali elec-

trolysis

-11% 12%


DE: Ethanol (96%) -9% 10%

AU: Chlorine from chlorine-alkali elec-

trolysis

-9% 10%

IT: Chlorine from chlorine-alkali elec-

trolysis

-9% 10%



ES: Chlorine from chlorine-alkali

electrolysis

-8% 9%

GB: Chlorine from chlorine-alkali

electrolysis

-8% 9%


BE: Chlorine from chlorine-alkali

electrolysis

-3% 4%

DE: Phenol -2% 2%


IT: Hydrogen -1% 1%

uncertainty in life cycle inventory data

Documents

quantifying uncertainty

aspects of data uncertainty

quantification of uncertainty

data variability

lci data

estimated data

relevant data

input data