ArticlePDF Available

Analyzing uncertainty in a comparative life cycle assessment of hand drying systems

Authors:

Abstract and Figures

Purpose The goal of this study is to evaluate and compare the environmental impact (with a focus on global warming potential) of five hand drying systems: hands-under (HU) dryers, high-speed hands-under (HSHU) dryers, high-speed hands-in (HSHI) dryers, cotton roll towels, and paper towels. Another objective is to incorporate uncertainty into this comparative life cycle assessment (LCA) as a means of understanding the statistical robustness of the difference between the environmental impacts of the hand drying systems. Methods We conducted a life cycle assessment in accordance with the ISO 14040/14044 standards using data primarily from publicly available reports. As part of the study, we performed a parameter uncertainty analysis for multiple scenarios to evaluate the impact of uncertainty in input data on the relative performance of products. In addition, we conducted a probabilistic scenario analysis of key drying system parameters in order to understand the implications of changing assumptions on the outcomes of the analyses. Results and discussion The scope of the analyses enabled us to draw robust conclusions about the relative environmental performance of the products. We can say with a high degree of confidence that the high-speed dryers have a lower impact than paper towels and cotton roll towels. Differentiating the performance of the hand dryers requires being more specific about framing assumptions. Under certain conditions, the HSHI dryer is expected to have a lower impact than the HU and HSHU dryers. However, under other conditions, one cannot say that the HSHI dryer is clearly better than the other dryers. We cannot differentiate the performance between the HU dryer, cotton roll towels, and paper towels. Conclusions This work demonstrates the importance of going beyond traditional uncertainty analyses for comparative LCAs that are used for assertions of relative product environmental impact. Indeed, we found instances where the conclusions changed as a result of using the probabilistic scenario analysis. We outline important elements that should be included in future guidance on uncertainty analyses in comparative LCAs, including conducting parameter and scenario uncertainty analyses together and then using the outcomes to guide selection of parameters and/or choices to analyze further.
Content may be subject to copyright.
UNCERTAINTIES IN LCA
Analyzing uncertainty in a comparative life cycle assessment
of hand drying systems
Jeremy R. Gregory &Trisha M. Montalbo &Randolph E. Kirchain
Received: 10 July 2012 /Accepted: 26 May 2013 /Published online: 19 June 2013
#Springer-Verlag Berlin Heidelberg 2013
Abstract
Purpose The goal of this study is to evaluate and compare the
environmental impact (with a focus on global warming po-
tential) of five hand drying systems: hands-under (HU) dryers,
high-speed hands-under (HSHU) dryers, high-speed hands-in
(HSHI) dryers, cotton roll towels, and paper towels. Another
objective is to incorporate uncertainty into this comparative
life cycle assessment (LCA) as a means of understanding the
statistical robustness of the difference between the environ-
mental impacts of the hand drying systems.
Methods We conducted a life cycle assessment in accor-
dance with the ISO 14040/14044 standards using data pri-
marily from publicly available reports. As part of the study,
we performed a parameter uncertainty analysis for multiple
scenarios to evaluate the impact of uncertainty in input data
on the relative performance of products. In addition, we
conducted a probabilistic scenario analysis of key drying
system parameters in order to understand the implications
of changing assumptions on the outcomes of the analyses.
Results and discussion The scope of the analyses enabled us
to draw robust conclusions about the relative environmental
performance of the products. We can say with a high degree
of confidence that the high-speed dryers have a lower impact
than paper towels and cotton roll towels. Differentiating the
performance of the hand dryers requires being more specific
about framing assumptions. Under certain conditions, the
HSHI dryer is expected to have a lower impact than the
HU and HSHU dryers. However, under other conditions,
one cannot say that the HSHI dryer is clearly better than
the other dryers. We cannot differentiate the performance
between the HU dryer, cotton roll towels, and paper towels.
Conclusions This work demonstrates the importance of go-
ing beyond traditional uncertainty analyses for comparative
LCAs that are used for assertions of relative product envi-
ronmental impact. Indeed, we found instances where the
conclusions changed as a result of using the probabilistic
scenario analysis. We outline important elements that should
be included in future guidance on uncertainty analyses in
comparative LCAs, including conducting parameter and
scenario uncertainty analyses together and then using the
outcomes to guide selection of parameters and/or choices
to analyze further.
Keywords Life cycle assessment .Hand drying systems .
Parameter uncertainty .Scenario analysis
1 Introduction
Characterizing the relative environmental impact of everyday
life activities (and the products that enable them) has been a
staple of life cycle assessment since the inception of the field.
The relative environmental impact of hand drying systems is a
clear example of this. Interest by the public at large is made plain
by broad coverage of this topic in the media (Koerner 2008;
Clarren 2007; Skoczen 2009;Adams2007;Watson2007).
Interest within the technical and academic community is clearly
evidenced by no less than nine studies that target this very topic,
including a streamlined life cycle assessment (LCA) conducted
for Airdri Ltd. and Bobrick Washroom Equipment that com-
pares a standard warm air dryer to paper towels (Environmental
Resources Management 2001), a hand dryertowel comparison
produced by myclimate and commissioned by Dyson in
Switzerland (Wettstein 2009), a comparison between cotton roll
Responsible editor: Andreas Ciroth
Electronic supplementary material The online version of this article
(doi:10.1007/s11367-013-0606-0) contains supplementary material,
which is available to authorized users.
J. R. Gregory (*):T. M. Montalbo :R. E. Kirchain
Materials Systems Laboratory, Engineering Systems Division,
Massachusetts Institute of Technology, 77 Massachusetts Ave, Rm.
E38-424, Cambridge, MA 02139, USA
e-mail: jgregory@mit.edu
Int J Life Cycle Assess (2013) 18:16051617
DOI 10.1007/s11367-013-0606-0
towels and paper towels commissioned by Vendor (Schryver
and Vieira 2008), and some calculations made by the Climate
Conservancy for Salon (Clarren 2007). More comprehensive life
cycle assessments that comply with the ISO 14040 and 14044
life cycle assessment standards (International Organisation for
Standardisation 2006) are also available. These include a study
for the European Textile Services Association that compares
cotton roll towels to paper towels (Eberle and Möller 2006),
another investigating multiple types of tissue products for
Kimberly-Clark (Madsen 2007), and a third for Excel Dryer
that compares its XLERATOR® hand dryer to a standard warm
air dryer and paper towels (Dettling 2009). Dyson has also
conducted a life cycle assessment of its Dyson Airblade
hand dryer in accordance with the PAS 2050 standard (British
Standards Institute 2008) in order to obtain a Carbon Reduction
Label from the Carbon Trust (Dyson 2010).
Among all these studies, the one by myclimate (Wettstein
2009) is the most comprehensive in the scope of hand drying
systems considereda high-speed hands-in dryer (the Dyson
Airbladehand dryer), a standard warm air dryer, cotton roll
towels, and paper towels. It does not include the hands-under
variant of high-speed dryers, however. By contrast, the report
conducted for Excel Dryer includes a high-speed hands-under
dryer (the XLERATOR® hand dryer) but does not consider a
high-speed hands-in dryer or cotton roll towels. Because of the
studiesdiffering functional units, assumptions, and data, life
cycle assessment outcomes cannot be easily compared. We
conducted this study in order to address this gap. Thus, the
primary goal of the study is to evaluate and compare the various
hand drying systemsincluding both variants of high-speed
hand dryersfrom the different studies by placing the systems
on a consistent basis.
A second and equally important objective is to incorpo-
rate uncertainty into this comparative LCA as a means of
understanding the statistical robustness of the difference
between hand drying system environmental impacts. Lloyd
and Ries (2007) provide definitions for the types of uncer-
tainty in their analysis of the LCA uncertainty literature.
Parameter uncertainty is derived from uncertainty in ob-
served or measured values. In an LCA context, it generally
refers to uncertainty in the input data used to create life cycle
inventories. Scenario uncertainty is related to the choices
made in framing the LCA and constructing scenarios. This
may be driven by inherent variability in geographic locations
or situations in the analysis, or it may be due to methodolog-
ical decisions around issues such as scope and allocation for
which there is no clear direction. Lloyd and Ries also discuss
model uncertainty, which relates to the structure and mathe-
matical relationships in models. In addition, there is uncer-
tainty in the impact factors used to translate the life cycle
inventory to life cycle impact.
In the literature, relatively few studies have been reported
that explore the role of uncertainty in comparative life cycle
assessments (Hong et al. 2010;Huijbregts1998;Huijbregts
et al. 2003;deKoningetal.2010). Of these, all have evaluated
the impact of parameter uncertainty. As a representative ex-
ample, Huijbregts (1998) quantified the implication of uncer-
tainty in the mass of the functional unit, the rate of recycling,
and many of the associated inventory items on the ability to
resolve the environmental performance of two alternative roof
gutter systems. Methods typically involve defining a baseline
set of conditions (or scenarios) around which a Monte Carlo
simulation or other analysis is conducted; ratios or differences
of impacts between products being compared are then calcu-
lated from analytical results. Indeed, current international
product carbon footprint standards suggest this type of analy-
sis (World Resources Institute and World Business Council for
Sustainable Development 2011). While these approaches are
important for the consideration of parameter uncertainty, they
only provide insight for a given scenario. Huijbregts first
pointed out this limitation and demonstrated an approach to
address it by examining four specific scenarios (two end-of-
life allocation rules and two future global warming potential
(GWP) reference scenarios). Later, Huijbregts et al. (2003)
demonstrate a more expansive comparative assessment that
includes both parameter and scenario uncertainty. Specifically,
in comparing two insulation alternatives for a Dutch home,
Huijbregts et al. evaluated aggregate parameter, scenario, and
model uncertainty across 32 specific scenario and model con-
ditions. Finally, de Koning et al. explored scenario uncertainty
due to analyst choices, but only insofar as uncertainty due to the
choices of other analysts is embedded in available databases
and secondary data. Instead, the focus of de Koning is on
demonstrating the importance of the decision framing scenario
(e.g., evaluating alternatives within the firm or comparing
products across firms) on the ability to resolve alternatives.
There is limited attention paid in LCA standards on how to
comment on the significance of the difference between prod-
uctsenvironmental impacts. The ISO 14044 standard recom-
mends that An analysis of results for sensitivity and uncertainty
shall be conducted for studies intended to be used in compara-
tive assertions intended to be disclosed to the public(Interna-
tional Organisation for Standardisation 2006). While this state-
ment is important, there is no further guidance in the ISO 14044
standard on how to conduct sensitivity and/or uncertainty anal-
yses to support comparative assertions. The International Ref-
erence Life Cycle Data System (ILCD) Handbook (European
Commission-Joint Research Centre-Institute for Environment
and Sustainability 2010), the PAS 2050 (British Standards
Institute 2008), and the recently released Product Life Cycle
Accounting and Reporting Standard from the GHG Protocol
(World Resources Institute and World Business Council for
Sustainable Development 2011) each have sections discussing
uncertainty. Although these documents contain definitions of
uncertainty similar to those provided by Lloyd and Ries (2007),
the guidance is limited in that the focus is on qualitative
1606 Int J Life Cycle Assess (2013) 18:16051617
characterizations of data quality and quantitative calculations of
uncertainty in input data (i.e., parameter uncertainty) with min-
imal discussion about how to conduct a full uncertainty analysis.
For instance, the ILCD Handbook discusses the importance of
sensitivity analysis (like the ISO 14044 standard) and states that
scenario analysis and uncertainty calculations are the methods
to support the sensitivity analysis. However, there is limited
guidance in the handbook on the differences between sensitivity
analysis, scenario analysis, and uncertainty analysis. This is
almost certainly part of the motivation for calls from the litera-
ture for more guidance on uncertainty analyses in standards
(Draucker et al. 2011).
We will not completely address this gap of meaningful guid-
ance on uncertainty analyses for comparative LCAs in this paper.
However, we believe that the analyses conducted as part of this
case study on hand dryers can illuminate the importance of
elements that should be included in such guidance and will add
to the body of knowledge in this area. We conduct parameter
uncertainty analyses for several scenarios following the approach
proposed by Huijbregts and implied in the standards. In addition,
we go beyond these traditional analyses by conducting further
probabilistic scenario analyses of key parameters in order to
understand the implications of changing key assumptions about
the characterization of uncertainty in the parameters and correla-
tion of parameters among compared alternatives on the outcomes
of the analyses. We use these analyses to make recommendations
for important elements that should be included in future guidance
on uncertainty analyses in comparative LCAs.
2 Methods
2.1 Goal and scope
The goal of this study is to compare the environmental impacts of
a broad range of hand drying systems using a consistent basis and
to analyze uncertainty in order to comment on the statistical
significance of the differences between the systems. The study
was conducted in accordance with the ISO 14040 and 14044
standards (International Organisation for Standardisation 2006),
including critical review. A report that includes the full details of
the study and the critical review comments is available
(Montalbo et al. 2011), but we have included information on
key data and methodological assumptions here and have sum-
marized other important details in the Electronic Supplementary
Material available online. This paper builds on the original report
by providing a more comprehensive uncertainty analysis, pre-
senting a more precise evaluation of the ability to resolve alter-
natives, and using this case study as a platform to discuss the
issues associated with comparative life cycle assessment.
The scope of the study includes five hand drying systems,
detailed in Table 1, which describes the method used by each
product to dry hands and references the primary source of
product material and performance data (the actual data are in
the Electronic Supplementary Material). In addition to the
dryers and towels themselves, packaging is considered in all
cases, as well as dispensers in the case of the towel systems
and a waste bin and bin liners for the paper towel system.
Drying a single pair of hands represents the functional
unit. Since the electric hand dryers (hands-under (HU), high-
speed hands-under (HSHU), and high-speed hands-in
(HSHI)) dry numerous pairs of hands over their lifetimes,
their environmental impacts are allocated across all these
pairs of hands. The same holds true for the cotton roll towels,
towel dispensers, waste bin, bin liners, and packaging used
by these products (see Section 1 in the Electronic Supple-
mentary Material for details on allocations per functional
unit).
The system boundaries encompass all life cycle phases,
from cradle to grave, along with transportation between and
within each phase. Details on the system boundary and the
life cycle stages are included in Section 1 of the Electronic
Supplementary Material. All systems, with the exception of
paper towels, are assumed to be manufactured in China.
Table 1 Overview of hand drying systems evaluated
Hand drying system Drying method Primary source of product
material and performance data
Hands-under (HU) dryer User places hands under the dryer nozzle and
warm air blows onto the hands to dry them
Generic dryer in Dettling (2009)
High-speed hands-under
(HSHU) dryer
Same as HU, but air is blown at high speeds Excel XLERATOR® dryer in
Dettling (2009)
High-speed hands-in (HSHI)
dryer
User places hands into dryer and air blows at high
speeds onto hands to dry them
Dyson Airbladedryer with
aluminum cover (Blower (2011)
Dyson Ltd., personal communication)
Cotton roll towels (CRT) User pulls cotton towel from roll in dispenser and
dries hands by rubbing on towel
Generic cotton towel in Eberle and
Möller (2006)
Paper towels (PT) User pulls paper towels from dispenser and dries
hands by rubbing on towels
Kimberly-Clark 100 % virgin paper
towel in Madsen (2007)
Int J Life Cycle Assess (2013) 18:16051617 1607
Upstream processes such as the mining of ore or the extrac-
tion and refining of petroleum for vehicle fuel are included
within system boundaries. Only the energy required to man-
ufacture dryers or towels is accounted for in the calculation
of manufacturing phase impactproduction of capital
equipment is considered outside the scope for this phase
due to limitations on data availability (it is included in
upstream processes that are part of the ecoinvent database).
The use phase takes place in the USA. For the electric
hand dryers, use phase impact is solely due to the production
and distribution of electricity required for operation. The use
phase for cotton roll towels encompasses not only the use of
the towel inside a washroom, but also a cleaning step which
takes place at a laundry facility. Finally, at the end-of-life, all
product types are transported to a nearby waste facility where
they are incinerated or sent to a landfill. With the exception
of the cardboard packaging, there is no clear evidence that
these products are commonly recycledor in the case of
cotton and paper towels, compostedin the USA.
2.2 Life cycle inventory and impact assessment
We obtained data used to generate drying system life cycle
inventories from a variety of existing sources outlined in
Section 2 in the Electronic Supplementary Material (the pri-
mary sources are summarized in Table 1). Baseline assump-
tions used to generate hand drying system life cycle invento-
ries are listed in Table 2.TheElectronic Supplementary Ma-
terial provides details on the sources for the data, but the hand
drying time data are particularly noteworthy. Hand drying
time is characterized in two ways and will be analyzed in the
uncertainty analyses. The first is drying-driven usage, where a
user dries his hands to either to a defined dryness standardin
this case NSF Protocol P335 (NSF 2007)(thebaselinedry
times are based on tests run using this standard)or to a
users personal dryness comfort level. This assumes users
are using the products as recommended. The second is time-
driven usage, where users employ the dryer for a prescribed
amount of time, regardless of the dryness of the hands. This
assumes users are not as concerned about having completely
dry hands. The times in Table 2are based on measurements
conducted in accordance with the NSF protocol (representing
drying-driven usage), but the uncertainty analyses will also
present results using drying times reported by hand dryer
manufacturers and time-driven usage scenarios (and are listed
in Section 2.1.3 of the Electronic Supplementary Material).
The only significant difference between the measured and
reported dry times is that the HSHU reported dry time is
12 s as compared to the 20 s measured dry time.
The importance of the hand drying time characterization
has led us to use three scenarios in the uncertainty analyses:
(1) baseline scenario assumptions including hand dryer mea-
sured dry times based on the NSF Protocol P335; (2) the same
baseline scenario assumptions except for dryer dry times
based on reported times from manufacturers (detailed in
Section 2.1.3 in the Electronic Supplementary Material); and
(3) the same baseline scenario assumptions and reported dry
times and a consistent printed wiring board (PWB) unit pro-
cess data used for all three hand dryers. The original data
sources for the three dryers had used different sets of unit
process data: the generic unit process, electronic component,
active, unspecified,was used to represent the control and
optics assemblies in original studies on the HSHU and the HU
dryers, whereas a more specific unit process, printed wiring
board, through-hole, lead-free surface,was specified for the
HSHI dryers inventory; the latter process has a lower impact.
The PWB has a significant impact on the production impact of
the dryers, and thus, the choice of the PWB can be important.
We chose to defer to the judgment of the original LCA
analysts in their selection of inventory data for the PWB and
have used those in our baseline analyses and scenario 2.
However, we believe it is unlikely the PWBs in the three
dryers would be significantly different, which is why we have
analyzed the impact of changing this assumption in scenario 3.
We obtained background inventory data for intermediate
flows from the ecoinvent v2.1 database (Frischknecht et al.
2007). The majority of the data were used directly from the
database, but in a few cases, we needed to make some mod-
ifications to ecoinvent datasets because of a lack of existing
inventory data. These cases and the associated data are listed
in Section 2.2 of the Electronic Supplementary Material.
We conducted all life cycle analyses using a combination of
models in SimaPro LCA software and Microsoft Excel with a
Crystal Ball extension. The specific instances for each software
tool are described in the Uncertainty analysis methodology
section below.
In the full report, we used IMPACT 2002+ (Hischier et al.
2010) to calculate life cycle impact. However, due to re-
source limitations, uncertainty analyses were only conducted
using the GWP life cycle impact assessment methodology
(Hischier et al. 2010), which incorporates the impact of
gaseous emissions according to their potential to contribute
to global warming based on the 100-year characterization
factors published in 2007 by the Intergovernmental Panel on
Climate Change (2007). Thus, only results using GWP are
presented here and the conclusions should not be generalized
to other impact measures without further analysis.
2.3 Uncertainty analysis methodology
We conducted two types of analyses to study the impact of
uncertainty on the outcomes of our comparative LCA: a pa-
rameter uncertainty analysis and a probabilistic scenario anal-
ysis. The former is consistent with the approach implied in the
standards, but the latter has not been explicitly outlined in the
literature or standards. A scenario analysis is presumably a
1608 Int J Life Cycle Assess (2013) 18:16051617
study of the impact that various scenarios have on the out-
comes of the LCA (it is not explicitly defined in the standards).
This is typically done using analyst-defined, discrete scenarios
(where a scenario is simply a defined set of parameters and
framing assumptions). We have extended this by defining
ranges for specific types of parameters that are particularly
important when defining scenarios and used probabilistic tech-
niques to explore thousands of potential scenarios.
This type of scenario analysis is not to be confused with a
scenario uncertainty analysis. Various sources in the litera-
ture described in the Introductionhave defined scenario
uncertainty as being about choices made in framing the LCA
(typical examples include allocation methods or boundary
decisions) and hence are not typically analyzed in a param-
eter uncertainty analysis. Our scenario analysis includes
some of these issues, but it also includes analysis of param-
eters that would be studied in a sensitivity or scenario anal-
ysis because of their high impact or high profile.
We did not characterize uncertainty in the models or in the
impact factors due to a lack of information on how the
uncertainty should be defined. Because these are not includ-
ed in our analysis, the method underestimates actual uncer-
tainty and therefore overestimates the ability to resolve dif-
ferences between alternatives. Future work should investi-
gate the significance of this limitation.
2.3.1 Parameter uncertainty analysis
In our uncertainty analyses, parameter uncertainty was char-
acterized using the pedigree matrix approach implemented in
the ecoinvent database (Frischknecht et al. 2007) because it
represents the most widely used methodology in LCA for
characterizing uncertainty. Details on the approach and the
specific scores used in this analysis are provided in Section 3
of the Electronic Supplementary Material. A lognormal dis-
tribution was used for all input data because it always
Table 2 Assumptions used to generate hand drying system life cycle inventories for the baseline analysis
Drying system Hands-under
dryer
High-speed
hands-under dryer
High-speed hands-in
dryer
Cotton roll towels Paper towels
Functional unit 1 pair of dry hands
Lifetime usage 350,000 pairs of dry hands over 5 years (Blower (2011) Dyson Ltd., personal communication; Dyson Airblade
Technical Specifications (2011)
Mass (+ manufacturing scrap)
per dryer or towel
6.4 kg (0.9 kg)
(Dettling 2009)
9.4 kg (1.12 kg)
(Dettling 2009)
14.8 kg (1.43 kg)
(Blower (2011)
Dyson Ltd., personal
communication)
16.2 g (2.2 g) (Eberle
and Möller 2006)
1.98 g (0.08 g)
(Dettling 2009)
Manufacturing location China China China China USA
Manufacturing energy per
dryer or towel
156 MJ electricity
(Dettling 2009)
156 MJ electricity
(Dettling 2009)
146 MJ electricity
(Dettling 2009)
431 kJ electricity,
507 kJ gas (Eberle
and Möller 2006)
14.7 kJ electricity,
24.4 kJ gas
(Dettling 2009)
Use location USA
Use intensity 31 s at 2,300 W 20 s at 1,500 W 12 s at 1,400 W 1 towel (pull) + laundry 2 towels
+ 1.5 s at 1,150 W + 1.5 s at 750 W + 0 s at 0 W
+ 406 s at 0.4 W + 429 s at 1 W + 439 s at 1 W
End-of-life scenario 76.7 % of cardboard recycled
19 % of remaining waste incinerated with energy recovery
81 % of remaining waste landfilled with methane capture and conversion to electricity (Franklin Associates 2011;
United States Environmental Protection Agency 2009)
Transportation
Raw material to plant 250 km via truck
Plant to warehouse 10,500 km via ocean freighter + 2,600 km via freight train + 24 km via truck (excluding paper towels)
Warehouse to washroom 1,760 km via truck
Washroom to laundry and back 100 km via truck (cotton towels only)
Washroom to waste facility 100 km via truck
Additional lifecycles Packaging Packaging Packaging Packaging, dispenser Packaging,
dispenser, waste
bin, bin liners
Packaging per dryer or towel 0.45 kg cardboard
(Dettling 2009)
0.27 kg cardboard
(Dettling 2009)
2.94 kg cardboard
(Blower (2011)
Dyson Ltd., personal
communication)
0.08 g polyethylene
(Eberle and Möller
2006)
0.18 g cardboard
(Dettling 2009)
Int J Life Cycle Assess (2013) 18:16051617 1609
remains positive and is consistent with the data available in
the ecoinvent database and the pedigree matrix approach.
We conducted parameter uncertainty analyses using
pairwise Monte Carlo simulations in SimaPro. The difference
in these results was evaluated using a one-way analysis of
variance on the means and through a pairwise comparison
indicator. The latter involved analyzing a pair of products
simultaneously and calculating the ratio of impacts for the
two products in each simulation (the ratio is known as the
comparison indicator (CI) as proposed by Huijbregts (1998)).
The pairwise analysis ensures a correlated analysis. That is,
values selected in each simulation for parameters that are
common within and among the analyzed pair will be the same.
(There may be instances where this correlation constraint is
unrealistic (e.g., it may be reasonable to expect that materials
common to the two products in the analysis come from differ-
ent sources), and thus, the correlation may create an appearance
of more overlap in environmental impact than actually exists.)
A result of the analysis is a cumulative frequency distribution of
comparison indicators (F(CI)) from all simulations. From the
F(CI), we can derive the fraction of simulations in which the
modeled impact of one product exceeded the other. Because the
current analysis excludes some forms of uncertainty (i.e., model
andimpactfactor),wehavechosentoidentifydifferenceinthe
two products at two conservative threshold values of 90 and
95 %. That is, a comparison indicator of greater than or equal to
one is observed in more than 90 or 95 % of the trials.
We conducted pairwise analyses of all product combina-
tions (ten total combinations) for a given set of conditions (or
scenario) using 1,000 simulations in each comparison. These
comparisons were conducted for three different scenarios (de-
scribed in the previous subsection) for a total of 30 analyses.
2.3.2 Probabilistic scenario analysis
A typical uncertainty analysis for a comparative LCA will
include a stochastic analysis of parameter uncertainty for a
selected number of discrete scenarios, which is consistent with
the parameter uncertainty analysis we have described here.
While this is important, the analysis represents a small sample
of the overall scenario space. That is, there will likely be
countless scenarios that can be derived from various combina-
tions of methodological choices and values for parameters that
are evaluated in scenarios. For this paper, we have attempted to
explore the scenario space more broadly by using Monte Carlo
analysis to randomly sample values for a set of drying system
parameters, such as use intensity or electric grid mix, and
repeat this process thousands of times to create a wide range
of scenarios. This enables us to comment on the robustness of a
claim that one product has a lower environmental impact than
another by stating how often a product is observed to have a
lower modeled impact across a wide range of scenarios.
Our probabilistic scenario analysis was implemented using
Microsoft Excel with Crystal Ballfor the Monte Carlo method
and the same inventory and unit process data used in the
parameter uncertainty assessment in SimaPro. Unlike the
parameter uncertainty analyses, all five products can be ana-
lyzed simultaneously for this spreadsheet model. The param-
eters varied in the analysis are shown in Table 3along with
their baseline values, ranges, and distributions; all other pa-
rameter values remained the same as in the baseline scenario.
All of the parameters (with the exception of the composting
assumption which is binary) have a uniform distribution be-
cause we are assuming that all parameter values and scenarios
are equally likely (in the absence of alternative information).
The use of a statistical distribution on the parameter values
and the Monte Carlo analysis enable us to generate numerous
combinations of scenarios across the scenario space.
The set of parameters shown in Table 3was selected because
they are parameters for which the selection of a representative
value was difficult or infeasible due to a lack of solid data or
because they represented variation in geographical scope (i.e.,
use grid mix and municipal solid waste incineration fraction). In
addition, a number of the parameters have a strong influence on
drying system environmental performance, as dictated by sensi-
tivity analyses detailed in the full report. Details on the motiva-
tion for the specific ranges for each parameter are provided in
Section4oftheElectronic Supplementary Material, along with
details about correlation assumptions for the parameters. Corre-
lation assumptions in a scenario analysis are not straightforward,
but can be critical (see de Koning et al. 2010). Due to this
importance, we conducted multiple analyses in which hand dryer
use intensity is either correlated or uncorrelated (correlated use
intensity means that deviations from the use intensity are the
same for all products in a given simulation; uncorrelated means
that deviations for all products are independent). We also
changed the usage pattern to be either drying-driven or time-
driven and whether the PWB unit process is consistent or incon-
sistent in each scenario uncertainty analysis. While we have
attempted to select sensitivity analysis ranges which covered
the majority of reasonable and significant cases, it is possible
that some cases are excluded. As such, the conclusions drawn
here should not be generalized to other cases without further
study.
We conducted six sets ofscenario uncertainty analyseswith
multiple combinations of these three framing assumptions
(dryer use intensity correlation, usage pattern, and PWB unit
process consistency). Each scenario uncertainty analysis in-
volved 20,000 Monte Carlo simulations. The differences in
the means of the resulting simulations were tested using the
SteelDwass variant of the KruskalWallis test which exam-
ines significance of difference in sample means while control-
ling for alpha across the entire set of comparisons (details are
provided in Section 5 of the Electronic Supplementary Mate-
rial). Differences in the sample medians were evaluated by a
1610 Int J Life Cycle Assess (2013) 18:16051617
sign test on each combination of samples. Additionally, in
each simulation, we calculated a comparison indicator for the
relative impact of two products in the same manner as the
parameter uncertainty analysis. We used the same 90 and
95 % frequency levels as thresholds for evaluating whether
one product has a lower environmental performance than
another across the scenario space.
3 Results
3.1 Deterministic
The deterministic GWP results associated with drying a single
pair of hands for the three scenarios described in Section 2.2 are
showninFig.1. The figures illustrate that there are differences in
overall expected impact among the drying systems, but in some
cases, this can depend on the scenario under consideration. They
also show that the use phase is expected to be the driving factor
for the GWP of electric hand dryers (due to the electricity
consumption during use) and the cotton roll towels (from wash-
ing the towels), whereas manufacturing is expected to be the
driving factor for the paper towel GWP. If reported dry times are
used (scenario 2, results in Fig. 1b), the HU dryer impact
decreases 3 % (relative to the baseline results with measured
dry times in Fig. 1a)to17.3gCO
2
eq (due to a dry time decrease
from 31 s down to 30 s) and the HSHU dryer decreases 34 % to
5.4gCO
2
eq(duetoadrytimedecreasefrom20sdownto12s);
the other three dryer system impacts remain unchanged. If re-
ported dry times and a consistent PWB unit process are used
(scenario 3, results in Fig. 1c), the HU dryer impact decreases
6 % (relative to the baseline results with measured dry times in
Fig. 1a)to16.8gCO
2
eq and the HSHU dryer decreases 40 % to
4.9 g CO
2
eq; once again, the other three dryer system impacts
remain unchanged. Although it may appear that there are clear
differences among the products, with this analysis alone, one
cannot comment on the robustness of comparisons between the
products, particularly when making comparisons across the three
scenarios.
Contribution analyses have highlighted the key drivers of
environmental impact for the hand drying systems for global
warming potential (Montalbo et al. 2011). For hand dryers,
environmental impact is driven by the use phase energy
consumed in the active use of the hand dryer. Within the
production phase (including materials and manufacturing),
key contributors are the housing materials, electricity used in
production, and the printed wiring boards for the controls
and optics assembly.
Table 3 Parameter ranges and distributions for probabilistic scenario analysis
Independent parameters Drying systems Baseline Range Distribution
Lifetime usage High-speed hands-in 350,000 150550 K pairs of hands over 5 years Uniform
Lifetime usage High-speed hands-under 350,000 150550 K pairs of hands over 5 years Uniform
Lifetime usage Hands-under 350,000 150550 K pairs of hands over 5 years Uniform
Lifetime usage Cotton roll towel (dispenser) 350,000 150550 K pairs of hands over 5 years Uniform
Lifetime usage Paper towel (dispenser and bin) 350,000 150550 K pairs of hands over 5 years Uniform
Number of reuses Cotton roll towel 103 70130 launderings and reuses Uniform
Manufacturing grid mix High-speed hands-in CN average 0.0191.44 kg CO
2
eq/kWh Uniform
Manufacturing grid mix High-speed hands-under CN average 0.0191.44 kg CO
2
eq/kWh Uniform
Manufacturing grid mix Hands-under CN average 0.0191.44 kg CO
2
eq/kWh Uniform
Manufacturing grid mix Cotton roll towels CN average 0.0191.44 kg CO
2
eq/kWh Uniform
Manufacturing grid mix Paper towels US average 0.0111.22 kg CO
2
eq/kWh Uniform
Use grid mix High-speed hands-in US average 0.0161.32 kg CO
2
eq/kWh Uniform
High-speed hands-under
Hands-under
Cotton roll towels
Use intensity High-speed hands-in 12 s Drying-driven:50 to +25 % of measured
baseline; time-driven:530 s
Uniform
High-speed hands-under 20 s
Hands-under 31 s
Use intensity Cotton roll towels 1 towel 12 towels Uniform
Use intensity Paper towels 2 towels 13 towels Uniform
Municipal solid waste
incineration fraction
All 19 % 0100 % Uniform
Compost Paper towels, cotton roll towels No Yes, No Binary
CN China, US United States
Int J Life Cycle Assess (2013) 18:16051617 1611
For cotton roll towels, the use phase (i.e., washing the
towels) accounts for over half of the total impact, followed
by transportation and manufacturing and then by materials.
Between material production (cotton fibers) and processing
(spinning, weaving, and de-sizing), no single step dominates
the GWP production impact of a cotton towel roll. These
results indicate that all life cycle phases, with the exception
of end-of-life, are important to consider when assessing the
life cycle impact of cotton roll towels.
For paper towels, the manufacturing phase makes up over
half of the impact for global warming potential and water
consumption, followed by the materials production phase and
transportation. It is noteworthy that paper towels are the only
product for which product end-of-life has any significant im-
pactspecifically in global warming potential (caused by deg-
radation of paper towels in landfills). Electricity and the natural
gas used in the paper towel production are the primary contrib-
utors to the production GWP, making up approximately three
quarters of the impact, followed by pulp manufacturing.
3.2 Parameter uncertainty
The outcomes of the parameter uncertainty analyses are
shown in Table 4. (Outcomes from the one-way analysis of
variance test, in particular, and the resulting significance of
differences in sample means are reported in Table 28 of
Section 5 of the Electronic Supplementary Material). These
outcomes are consistent with, but generally less stringent than
the comparison indicator results discussed here. Results from
pairwise comparisons are listed for three different scenarios in
Table 4. In some instances, one drying system is better than
another in all simulations and across all three scenarios. For
example, the high-speed hands-in dryer outperformed the
hands-under dryer in every simulation in all three scenarios.
In fact, both high-speed dryers are almost always better than
the other three products (the one exception being the HSHU
and cotton roll towels whose similarity cannot be rejected at
90 % (or 95 %) confidence when the dryer is evaluated at the
measured dry time). When accounting for uncertainty in input
data, we can clearly say with confidence that the high-speed
dryers have a lower GWP than the other drying systems for
these three scenarios. However, the results are much less clear
when comparing the two high-speed dryers (in only one
scenario can the similarity of the two products be rejected at
90 or 95 % confidence) or when comparing cotton roll or
paper towels and the hands-under dryer.
3.3 Probabilistic scenario analysis
The resulting GWP frequency distributions for the 20,000
iterations of the scenario uncertainty analysis are presented
in Fig. 2for both drying-driven (see Fig. 2a) and time-driven
(see Fig. 2b) usage patterns. Dryer GWP distributions associ-
ated with time-driven usage are similar because the dryers
themselves are differentiated only by their respective power
ratings: dry times are the same for all dryers. By contrast, the
distributions of the HSHU and HU dryer systems have a much
wider spread than that of the HSHI dryer system when drying-
driven usage is considered due to the broader range of dry
times for the first two systems. Statistical tests on the signif-
icance of difference for the central tendency of these results
(KruskalWallis test on means and sign test of medians) are
presented in Tables 29 and 30 in Section 5 of the Electronic
Supplementary Material. These results are consistent with, but
0
5
10
15
20
Hands-under
dryer
High-speed
hands-under
dryer
High-speed
hands-in dryer
Cotton roll
towels
Paper towels
Global warming potential [g CO2 eq]
End-of-Life
Use
Transportation
Manufacturing
Materials
(a)
0
5
10
15
20
Hands-under
dryer
High-speed
hands-under
dryer
High-speed
hands-in dryer
Cotton roll
towels
Paper towels
Global warming potential [g CO2 eq]
End-of-Life
Use
Transportation
Manufacturing
Materials
(b)
0
5
10
15
20
Hands-under
dryer
High-speed
hands-under
dr
y
er
High-speed
hands-in dryer
Cotton roll
towels
Paper towels,
virgin
Global warming potential [g CO2 eq]
End-of-Life
Use
Transportation
Manufacturing
Materials
(c)
Fig. 1 Global warming potential for drying a single pair of hands under
three scenarios: ascenario 1: baseline assumptions, measured dry times;
bscenario 2: baseline assumptions, reported dry times; cscenario 3:
consistent PWB unit process, reported dry times
1612 Int J Life Cycle Assess (2013) 18:16051617
generally less stringent (i.e., identify more statistically signif-
icant differences) than the comparison indicator results
discussed here.
While the frequency distributions of drying system GWP in
Fig. 2clearly overlap, it is important to look at the comparison
indicator results in order to understand the impact of correla-
tion. Comparison indicator distributions for the GWP of dif-
ferent drying systems relative to that of the HSHI dryer are
shown in Fig. 3given both drying-driven and time-driven
usage patterns. The HSHI dryer was chosen as the point of
comparison because it has the lowest impact in the baseline
deterministic analyses. The results indicate that for drying-
driven dry times, the comparison indicator distribution is
almost entirely above one. By contrast, this distribution is
shifted to the left for time-driven dry times. Overall, however,
the GWP of the high-speed hands-in dryer is still lower than
that of any given drying system in over 92 % of the iterations.
The results in Figs. 2and 3assume correlated use phase grid
mix and use intensity for the hand dryers. The HSHI dryer
system almost always has the lower impact for a given scenario
due to this key assumption of correlated usage (i.e., when the
high-speed hands-in dryer is used at low intensity, the other
dryers are as well, leading the high-speed hands-in dryer to be
consistently favored as it has the shorter dry time (in the case of
drying-driven usage)) and the lower power rating. There are a
few instances, however, when other drying methods have a
lower impact. In scenarios where the HSHU dryer has a high
lifetime usage, the HSHI dryer has a low lifetime usage, and the
use phase takes a low-carbon grid mix, the HSHU system has
the lower impact. In another scenario, the longer HSHI hand
dryer dry times associated with time-based usage in-
crease the frequency at which the other drying systems
will have a lower impact than the HSHI dryer system.
Results from a range of different correlation, use intensity,
and unit process framing assumptions are presented in Table 5
using the comparison indicator metric as the basis for evaluat-
ing the frequency at which one products GWP is lower than
that of another. It is particularly important to explore alternative
correlation assumptions because the three hand dryers are
inherently different and may elicit different user behavior.
The six sets of framing assumptions are essentially meta-
scenarios for each set of simulations that explore 20,000 spe-
cific scenarios. (The results in Figs. 2a and 3a were generated
using the meta-scenario 1, whereas the results in Figs. 2b and
3b were generated using the meta-scenario 4. Additionally,
they only represent drying system impact relative to the HSHI
dryercorresponding to the top four rows in Table 5.)
The results involving the HSHI dryer as a comparator
indicate that removing the correlation between dryer use
intensities increases the frequency at which the GWPs of
the HSHU and HU dryer systems are less than that of the
HSHI dryer system, particularly for time-driven usage. The
comparison indicators for cotton roll towel and paper towel
systems with the HSHI dryer are unaffected by usage corre-
lation. Reconciling the PWB unit process for the HSHU and
the HU dryers also increases the number of Monte Carlo-
generated scenarios that result in their GWPs undercutting
the HSHI dryer GWP; in fact, these lead to comparison
indicator frequencies below the 90 % threshold. Time-
driven usage patterns (which are longer for the HSHI prod-
uct) also decrease the differentiation between the HSHI
product and all other alternatives. In fact, under time-driven
assumptions, the HSHI and HSHU cannot be declared dif-
ferent for a more stringent 95 % threshold (and is only
declared different for one framing assumption (4) for the less
stringent 90 % threshold). For the HSHU system, the use of
Table 4 Results of parameter uncertainty analyses. Numbers in bold are below the 90 % threshold and, therefore, do not meet the prescribed
significance level to be declared different
Drying system A Drying system B Frequency GWP
A
>GWP
B
(F(CI1))
1. Baseline assumptions,
measured dry times (%)
2. Baseline assumptions,
reported dry times (%)
3. Consistent PWB unit
process, reported dry
times (%)
High-speed hands-in dryer High-speed hands-
under dryer
0.0 15 35
High-speed hands-in dryer Hands-under dryer 0.0 0.0 0.0
High-speed hands-in dryer Cotton roll towels 0.2 0.2 0.2
High-speed hands-in dryer Paper towels 0.0 0.0 0.0
High-speed hands-under dryer Hands-under dryer 0.0 0.0 0.0
High-speed hands-under dryer Cotton roll towels 13 0.0 0.5
High-speed hands-under dryer Paper towels 1.0 0.0 0.0
Cotton roll towels Hands-under dryer 7.8 10 12
Cotton roll towels Paper towels 1.2 1.2 1.2
Paper towels Hands-under dryer 35 41 46
Int J Life Cycle Assess (2013) 18:16051617 1613
the same PWB unit process and time-driven usage assump-
tions impacts the GWP of the alternatives enough so that in
half the Monte Carlo-generated scenarios, its impact is less
than that of the HSHI dryer system. In fact, this comparison
is the only example where significance testing cannot reject
the null hypothesis that the median impacts of these two
products are the same (see Section 5 of the Electronic Sup-
plementary Material). This result is notably different than the
results from the first set of framing assumptions where the
HSHI dryer impact is almost always lower than the HSHU
dryer impact.
The comparison of the HSHU dryer with other dry-
ing systems is consistent with HSHI dryer results: the
HSHU dryer exceeds the 90 % threshold (i.e., can be
declared significantly different) for all sets of framing
assumptions for the cotton roll towels and paper towels
(although it fails a more stringent 95 % threshold for
paper towels). However, it falls below the threshold
(i.e., cannot be declared different) when compared to
the HU dryer in uncorrelated time-driven use intensity
meta-scenarios (numbers 5 and 6 in Table 5). The
cotton roll towels, HU dryer, and paper towels all show
significant overlap in impacts across all sets of framing
assumptions. We cannot say with confidence that any
one of them has a lower impact than another based on
this scenario uncertainty analysis.
4 Discussion
After reviewing the outcomes of the parameter uncertainty
analysis and probabilistic scenario analysis, it is valuable to
assess how much confidence we have in asserting the differ-
ence between product environmental impacts (i.e., the sta-
tistical robustness of the comparison). The deterministic re-
sults in Fig. 1, although temptingly different for the five
products, are clearly insufficient to draw meaningful conclu-
sions because they represent a single scenario and do not take
into account uncertainty in the input data. The outcomes of
the parameter uncertainty analysis listed in Table 4indicate
that for the three scenarios considered, we have statistical
confidence that the high-speed dryers have lower GWPs than
the other drying systems (the slight exception being HSHU
dryer compared to cotton roll towels for measured dry
times), but that the results are inconclusive when comparing
the two high-speed dryers or when comparing paper towels
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0 10203040
Frequency
GWP [g CO2 eq]
(a)
HSHI
HSHU
Cotton Roll
Towels
Paper
Towels
HU
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0 10203040
GWP [g CO2 eq]
High-speed hands-
in dryer
High-speed hands-
under dryer
Hands-under dryer
Cotton roll towels
Paper towels
(b)
Fig. 2 GWP frequency
distributions given (a) drying-
driven and (b) time-driven
usage patterns. Both plots are
based on correlated use
intensities and different printed
wiring boards for the hand
dryers (in accordance with
meta-scenarios 1 and 4,
respectively, from Table 5)
0.00
0.10
0.20
0.30
0.40
0246
Frequency
GWP CI
(a)
HSHU
HU
Cotton Roll
Towels
Paper
Towels
0.00
0.10
0.20
0.30
0.40
0246
GWP CI
High-speed hands-
under dryer
Hands-under dryer
Cotton roll towels
Paper towels
(b)
Fig. 3 GWP comparison
indicator (CI) frequency
distributions (calculated relative
to high-speed hands-in dryer)
given (a) drying-driven and (b)
time-driven usage patterns. The
dashed line indicates a
comparison indicator value of
one as a reference. Both plots
are based on correlated use
intensities and different printed
wiring boards for the hand
dryers (in accordance with
meta-scenarios 1 and 4,
respectively, from Table 5)
1614 Int J Life Cycle Assess (2013) 18:16051617
and the hands-under dryer. When we combine this with the
results of the probabilistic scenario analysis in Table 5, the
conclusion that the high-speed dryer impacts are lower than
the impacts of cotton roll towels and paper towels is strength-
ened. However, in two of the six meta-scenarios with time-
driven and uncorrelated use intensity, we cannot say with
confidence that any of the three hand dryer systems have
significantly different impacts because all comparison indi-
cator frequencies are below the 90 % threshold value. Fur-
thermore, we cannot say with confidence that the HU dryer,
cotton roll towels, or paper towels impacts are distinguish-
able from each other based on the fact that all three products
had comparison indicator frequencies below the 90 % thresh-
old across all six meta-scenarios in the scenario uncertainty
analysis when compared with one another.
Based on this information, we can say with a high degree
of statistical confidence that the high-speed dryers have a
lower impact than paper towels and cotton roll towels across
a broad set of scenarios. Differentiating the performance of
the hand dryers compared to one another requires being more
specific about framing assumptions. For drying-driven use
(when consumers use the products as recommended), the
HSHI dryer has a lower impact than the HU dryer in nearly
all cases and a lower impact than the HSHU in the majority
of cases. However, for time-driven use (when consumers are
not as concerned about having equivalently or completely
dry hands), one cannot say that the HSHI dryer is clearly
better than the other dryers. Additionally, we cannot confi-
dently differentiate performance between the HU dryer, cot-
ton roll towels, and paper towels across this scenario space.
This last statement is important to consider. This analysis
does not mean that these products are equivalent in any given
scenario. It only means that when there is considerable
scenario ambiguity and, therefore, the scenario space is
large, it is not possible to reach a single definitive conclu-
sion. If the decision-maker is able to further refine the sce-
nario space (e.g., credibly establish the use location or the
drying habits of the user population), further resolution is
plausible. The case of comparing cotton roll towels and
paper towels is particularly illustrative of this point because
the results of the parameter uncertainty analysis (see Table 4)
indicated that cotton roll towels had a statistically significant
lower impact than paper towels for three specific scenarios.
However, the probabilistic scenario analysis indicated that
there was significant overlap in impacts across all six meta-
scenarios and one could not assert that one product clearly
had a lower impact than the other across a broad scenario
space. This motivates the importance of both going beyond a
typical parameter uncertainty analysis with a few scenarios
to a broader exploration of scenario uncertainty and to al-
ways revisit results with the ultimate decision-maker to
ensure that further information cannot be brought to bear or
that the decision space cannot be further divided.
This work demonstrates the importance of conducting un-
certainty analyses for comparative LCAs that are used for
assertions of relative product environmental impact, as recom-
mended in the several standards. Indeed, there are several
recommendations that are included in the standards and are
supported by this case study. First, broad assertions of relative
impact (e.g., Product Xhas a lower environmental impact than
Table 5 Probabilistic scenario analysis results for six sets of framing
assumptions. In the descriptions of the six sets of framing assumptions,
drying-drivenand time-drivenrefer to the ranges of use intensity
times for the hand dryers (see Table 3), correlated useand
uncorrelated userefer to whether the dryer use intensities are
correlated, and different PWBand same PWBrefer to whether
the unit process used for the printed wiring board in the hand dryers is
the same for all products or different. Numbers in bold are below the
90 % threshold and, therefore, do not meet the prescribed significance
level to be declared different
Drying
system A
Drying
system B
Frequency GWP
A
>GWP
B
(F(CI1))
1. Drying-driven,
correlated use,
different PWB
(%)
2. Drying-driven,
uncorrelated use,
different PWB
(%)
3. Drying-driven,
uncorrelated use,
same PWB
(%)
4. Time-driven,
correlated use,
different PWB
(%)
5. Time-driven,
uncorrelated use,
different PWB
(%)
6. Time-driven,
uncorrelated
use, same PWB
(%)
HSHI dryer HSHU dryer 1.3 4.5 14 7.3 37 50
HSHI dryer HU dryer 0.2 0.2 1.2 1.0 19 27
HSHI dryer Cotton roll towels 0.0 0.0 0.0 1.3 1.3 1.3
HSHI dryer Paper towels 0.4 0.5 0.4 6.5 6.2 6.4
HSHU dryer HU dryer 1.0 1.6 0.5 4.1 26 26
HSHU dryer Cotton roll towels 1.1 1.2 0.8 2.7 2.8 1.9
HSHU dryer Paper towels 7.8 7.6 6.4 9.4 9.0 7.8
Cotton roll towels HU dryer 65 65 67 85 85 87
Cotton roll towels Paper towels 59 59 60 59 59 60
Paper towels HU dryer 57 58 59 77 78 79
Int J Life Cycle Assess (2013) 18:16051617 1615
product Y.) should only be made if the claim can be supported
by uncertainty analyses that demonstrate that the claim is robust
across all of these analyses. Second, if the uncertainty analyses
reveal significant overlap in the distribution of impacts associ-
ated with the alternatives, then assertions of relative product
impact need to be stated alongside a clear definition of key
framing assumptions (i.e., those assumptions that change the
relative impact of the products). Finally, given the uncertainty in
calculated environmental impact values and the variation across
equally plausible scenarios, we discourage the use of quantify-
ing relative impact (e.g., Product Xhas a Z% lower environ-
mental impact than product Y.) unless it is accompanied by a
specific confidence level (such as those listed in Tables4and 5).
While these broad recommendations in the standards are
useful, there is almost no specific guidance on conducting
uncertainty analyses for comparative LCAs. This case study
has illuminated several issues that should be included in such
guidance. First, although it is useful to aggregate the uncer-
tainty of multiple parameters in the parameter uncertainty
analysis, it will almost always be meaningful to conduct
further uncertainty analyses with specific parameters held
constant as a means of gaining insight on the impact of key
parameters on outcomes. This is analogous to conducting
parameter uncertainty analyses using multiple scenarios (as
was done here and in other work), but there should be
analytical justification for the selection of parameters that
should be analyzed further. An example of this can be seen in
Mattila et al. (2011). Second, the literature discusses a sep-
aration between parameter and scenario uncertainty, but in
reality, there is overlap between the two in the implementa-
tion of uncertainty analyses because many choices (related to
scenario uncertainty) manifest themselves as changes in
parameters. Thus, parameter and scenario uncertainty should
be analyzed together in an aggregate manner where possible
and then analytical methods should be used to determine
which parameters and/or choices should be analyzed further.
Of course, there are some choices that cannot be aggregated,
such as the use of different life cycle impact assessment
methods, and these will still need to be analyzed separately.
As with any LCA, our analysis has several limitations. One
limitation centers on data collection. We received data on the
HSHI dryer directly from a manufacturer (which meant higher
data quality and lower uncertainty), whereas we relied heavily
on data from previously published studies for other hand drying
systems, and in some cases, we did not have strong sources for
key pieces of information (such as observed use intensity of the
hand dryer systems). These challenges are common in compar-
ative LCAs and motivate the need for expanded uncertainty
analyses such as the ones we have presented here. In-depth
uncertainty analyses are the strongest way to understand the
implications of data limitations and other important assumptions
(that are part of all LCAs) on comparative assertions of envi-
ronmental impact. Additionally, due to resource limitations,
uncertainty analyses were only conducted for GWP impacts.
The results presented here should not be generalized to other
impacts without further study. Finally, this paper focuses on
uncertainty analysis among the various systems. A complete
assessment of uncertainty should also include sensitivity analy-
sis to isolate the main drivers of impact (and possibilities for
improvement).
Acknowledgments This work was commissioned and funded by Dyson.
References
Adams C (2007) Whats better for the environment, electric hand dryers or
paper towels? http://www.straightdope.com/columns/read/2721/whats-
better-for-the-environment-electric-hand-dryers-or-paper-towels
British Standards Institute (2008) Publicly Available Specification
2050:2008: specification for the assessment of the life cycle green-
house gas emissions of goods and services. London, UK
Clarren R (2007) Public bathroom dilemma: paper or air? Salon. http://
www.salon.com/2007/11/12/bathroom_dilemma/
de Koning A, Schowanek D, Dewaele J, Weisbrod A, Guinée J (2010)
Uncertainties in a carbon footprint model for detergents; quantify-
ing the confidence in a comparative result. Int J Life Cycle Assess
15(1):7989
Dettling J (2009) Comparative environmental life cycle assessment of
hand drying systems: the XLERATOR hand dryer, conventional
hand dryers and paper towel systems. Quantis US, Boston, US.
http://www.exceldryer.com/PDFs/LCAFinal9-091.pdf
Draucker L, Kaufman S, ter Kuile R, Meinrenken C (2011) Moving forward
on product carbon footprint standards. J Ind Ecol 15(2):169171
Dyson (2010) The Dyson Airbladehand dryer receives industry first
Carbon Reduction Label. http://www.dysonairblade.com/news/
newsArticle.asp?newsID=116
Dyson Airblade Technical Specifications (2011) http://www.dysonairblade.
co.uk/specification/default.asp?product=AB01-AIRBLADE
Eberle U, Möller M (2006) Life cycle analysis of hand-drying systems.
A comparison of cotton towels and paper towels. Öko-Institut e.V.,
Freiburg, Germany. http://www.etsa-europe.org/news/documents/
ETSALCAbrochureoncottonrolltowelsJuly06EN.pdf
Environmental Resources Management (2001) Streamlined life cycle assess-
ment study. Oxford, UK. http://www.airdri.com/script/AirdriBobrick.pdf
European Commission-Joint Research Centre-Institute for Environ-
ment and Sustainability (2010) International Reference Life
Cycle Data System (ILCD) Handbookgeneral guide for life
cycle assessmentdetailed guidance. Luxembourg. http://
lct.jrc.ec.europa.eu/pdf-directory/ILCD-Handbook-General-
guide-for-LCA-DETAIL-online-12March2010.pdf
Franklin Associates (2011) Documentation for the paper calculator.
http://www.papercalculator.org/
Frischknecht R, Jungbluth N, Althaus H-J, Doka G, Dones R, Heck T,
Hellweg S, Hischier R, Nemecek T, Rebitzer G, Spielmann M,
Wernet G (2007) Overview and methodology. ecoinvent report no.
1. Swiss Centre for Life Cycle Inventories, Dübendorf, Switzerland.
http://www.ecoinvent.org/fileadmin/documents/en/01_Overview
AndMethodology.pdf
Hischier R, Weidema B, Althaus H-J, Bauer C, Doka G, Dones R,
Frischknecht R, Hellweg S, Humbert S, Jungbluth N, Köllner T,
Loerincik Y, Margni M, Nemecek T (2010) Implementation of life
cycle impact assessment methods. ecoinvent report no. 3, v2.2.
Swiss Centre for Life Cycle Inventories, Dübendorf, Switzerland.
http://www.ecoinvent.org/fileadmin/documents/en/03_LCIA-
Implementation-v2.2.pdf
1616 Int J Life Cycle Assess (2013) 18:16051617
Hong J, Shaked S, Rosenbaum RK, Jolliet O (2010) Analytical uncertainty
propagation in life cycle inventory and impact assessment: application
to an automobile front panel. Int J Life Cycle Assess 15(5):499510
Huijbregts M (1998) Part II: dealing with parameter uncertainty and
uncertainty due to choices in life cycle assessment. Int J Life Cycle
Assess 3(6):343351
Huijbregts MAJ, Gilijamse W, Ragas AMJ, Reijnders L (2003) Evalu-
ating uncertainty in environmental life-cycle assessment. A case
study comparing two insulation options for a Dutch one-family
dwelling. Environ Sci Technol 37(11):26002608
Intergovernmental Panel on Climate Change (2007) Climate Change
2007: the physical science basis. Contribution of Working Group I
to the Fourth Assessment Report of the Intergovernmental
Panel on Climate Change, Cambridge, UK. http://www.ipcc.ch/
publications_and_data/publications_ipcc_fourth_assessment_
report_wg1_report_the_physical_science_basis.htm
International Organisation for Standardisation (2006) ISO/FDIS 14044
Environmental management-life cycle assessment-requirements
and guidelines
Koerner B (2008) Electric hand dryers vs. paper towels. Slate. http://
www.slate.com/id/2193740/
Lloyd SM, Ries R (2007) Characterizing, propagating, and analyzing
uncertainty in life-cycle assessment: a survey of quantitative ap-
proaches. J Ind Ecol 11(1):161179
Madsen J (2007) Life cycle assessment of tissue products. environmen-
tal resources management. http://www2.kimberly-clark.com/pdfs/
LifeCycleAssessment.pdf
Mattila T, Kujanpää M, Dahlbo H, Soukka R, Myllymaa T (2011)
Uncertainty and sensitivity in the carbon footprint of shopping bags.
J Ind Ecol 15(2):217227. doi:10.1111/j.1530-9290.2010.00326.x
Montalbo T, Gregory J, Kirchain R (2011) Life cycle assessment of
hand drying systems. Massachusetts Institute of Technology,
Cambridge, USA. http://msl.mit.edu/publications/HandDryingLCA-
Report.pdf
NSF (2007) NSF Protocol P335: hygienic commercial hand dryers.
Ann Arbor, MI
Schryver AD, Vieira M (2008) LCA of two different hand drying
systems: a comparative life cycle assessment of non-reusable
paper and reusable cotton towels PRé Consultants bv, Amersfoort,
The Netherlands. http://www.vendorinternational.com/upload/
Engelse%20plaatjes/VendorReport_shortversion.pdf
Skoczen S (2009) Sustainability showdown: paper towels vs electric
dryer. http://www.sixlinks.org/People/Steven-Skoczen/Blog/82/
Sustainability-Showdown-Paper-Towels-Vs-Electric-D
United States Environmental Protection Agency (2009) Municipal Sol-
id waste generation, recycling, and disposal in the United States:
facts and figures for 2008. http://www.epa.gov/osw/nonhaz/
municipal/msw99.htm
Watson J (2007) Greener dryer better: LCA of hand dryers vs. paper
towels. http://www.treehugger.com/clean-technology/greener-
dryer-better-lca-of-hand-dryers-vs-paper-towels.html
Wettstein D (2009) Ecobalance of emissions with climatic effect
resulting from hand drying. myclimate, Zurich, Switzerland.
http://www.climatop.ch/tl_files/factssheet/2011/E_Factsheet_
Dyson_Handdryer_2011_2013_v2.pdf
World Resources Institute and World Business Council for Sus-
tainable Development (2011) Greehouse gas protocol: product
life cycle accounting and reporting standard. http://www.
ghgprotocol.org/files/ghgp/public/Product-Life-Cycle-Accounting-
Reporting-Standard_041613.pdf. Accessed 2011
Int J Life Cycle Assess (2013) 18:16051617 1617
... Cotton roll towels are advantageous only in terms of water consumption. Globally, the lower impacts are expected when choosing novel types of hand dryers, like high-speed hand dryers and, especially, airblade-type hand dryers (Gregory et al. 2013). Joseph et al. (2015) carried out an LCA on paper roll towels and conventional hand dryers, by analyzing four impact categories: global warming, ecosystem quality, human health and resources. ...
... From the environmental point of view, each scenario will be evaluated in terms of GHG emissions, following multiple waste treatment options and different electric grid mixes. In addition, in terms of carbon emissions, this paper wants to go beyond the results of the LCAs cited above (Budisulistiorini 2007;Gregory et al. 2013;Joseph et al. 2015;Montalbo et al. 2011), since the present work considers multiple scenarios for electric energy production and waste management, a variable number of users and the effect of the variability of other parameters on the final results, which can lead to different conclusions with respect to a classical LCA. Furthermore, the model allows estimating the annual cost associated with each option and evaluating the potential impact of possible changes in the cost structure of each scenario (e.g., increasing cost of energy, alternative energy mixes, increasing costs for waste treatment). ...
... These results are confirmed by the work of Joseph et al. (2015), which estimated that the global warming potential of hand dryers is about 60% lower than that of paper towels. The results of Joseph et al. (2015) are very close to those obtained by Gregory et al. (2013), especially concerning high-speed hand dryers (e.g., airblade-type). Budisulistiorini (2007) got to the same conclusions in terms of the environmental impacts of electrical hand dryers compared to paper towels. ...
Article
Full-text available
Due to the high density of users hosted everyday, public buildings are important producers of waste and emissions of greenhouse gases (GHGs). Public restrooms play an underrated role in waste generation and GHG emissions, especially if paper towels are used as the hand-drying method. The choice of the hand-drying method (i.e., paper towels vs. electrical hand dryers) also affects the economic balance of a public institution, involving costs for the purchase of hand-drying materials/apparatuses, energy and waste disposal. The present paper aims at evaluating the economic and environmental impact of the introduction of electrical hand dryers (alternative scenario) in place of paper towels (reference scenario) in a public building. The paper presents a solid methodology, based on a numerical experiment approach, to identify a decision criterion for establishing the economical convenience of adopting the alternative scenario in public restrooms. Key factors affecting the choice between the proposed alternatives are presented and discussed in a dedicated sensitivity analysis. From the environmental point of view, this study evaluates the impact of each scenario in terms of GHG emissions, related to multiple waste treatment options and different electric grid mixes. Based on the experimental assumptions, the method allowed concluding that the alternative scenario becomes economically convenient when the number of daily usages ( N ) is > 57 ± 4. The environmental convenience of the alternative scenario is visible even at N < 10. The method here described can be successfully used to support strategic decisions for cost optimization and environmental mitigation in institutional buildings.
... The electric dryer results six positive indicators, among others organic breathing, inorganic breathing, ozone layers, ecotoxicity, acidification eutrophication, and fossils fuel [5]. Moreover, research about environmental impacts with focus on global warming potential explained in life cycle assessment (LCA) concludes that high speed dryers 2 have a lower environmental impact and low global warming potential compared with other kinds of method dryer [6]. The research aims to make hand dryer with sanitizer to support sustainability and ergonomic for user and environmentally friendly, easy to recycle if it not used again. ...
... However, as has been profusely reported in the literature [46][47][48], it has been very difficult to adopt the traditional ABC model because of the costs incurred during its implementation and maintenance over time [47]. Another disadvantage in applying ABC models in long-term scenarios, is the high level of variability associated with driver behavior [49][50][51][52]. ...
Article
Full-text available
In capital-intensive industries, physical assets and maintenance activities play a relevant and strategic role in terms of providing operational continuity and business sustainability. As a result, maintenance support structures are highly complex and sophisticated. Therefore, maintenance capacity planning must be addressed using reliable techniques to assure the adequate service levels and availabilities of critical assets at the minimum opportunity cost. There has been relatively limited research on how to determine and optimize the maintenance support structure (human resources) in such organizations. This paper proposes a novel technique for dimensioning and optimizing maintenance capacity that combines Time-Driven Activity-Based Costing and Life-Cycle Costing with the Weibull function-based reliability model. Following the main principles of the Design Science Research we propose a sophisticated but simple artifact. Through this model, it is possible to compute maintenance costs and assess both used and idle capacities, considering the behavior over time of the failure rates and the reliability of critical assets within a plant. To demonstrate how the proposed methodology addresses the problem, the model was applied in a real medium-sized Chilean comminution plant and a sensitivity analysis was performed, particularly, to evaluate the relevance of appropriate maintenance workforce planning.
... Spearman's rank correlation coefficients (Hamby, 1994) was used as the measure of sensitivity that allows ranking different uncertain inputs based on their level of contributions to the variance of modelled output. Since we had no information on the probability profile of the waste coefficients, we assumed a uniform distribution; therefore assuming that all coefficient values are equally likely (Gregory et al., 2013). ...
Article
Full-text available
In 2015, the United Nations defined the Sustainable Development Goals (SDG), which include a target (12.3) on food waste. The target requires “by 2030, to halve per capita global food waste at the retail and consumer levels and to reduce food losses along production and supply chains, including post-harvest losses”. The target has increased awareness about the food waste problem and boosted research in food waste quantification. Nevertheless, there is a lack of studies that adopt a systematic approach to account for food waste providing disaggregated values per food supply chain stage and per food groups. Such an approach could support policy makers in prioritizing interventions for food waste reduction. To fill this gap, this paper presents a high-level top-down approach to food waste accounting in the European Union. The study aims to support the understanding of the mass flows associated with food production, consumption, and waste, addressing different food groups along the food supply chain. The method for accountin is the mass flow analysis. According to the results, cereals, fruit, and vegetables as the food groups are responsible for the highest amount of food waste, and the consumption stage to be responsible for the largest share of food waste for most food groups. This work highlights the need for further primary research on food waste generation in the EU. Ultimately, this would allow to robustly estimate the food waste generated at EU level, and establishing a more accurate baseline to track the progress towards SDG target 12.3.
... An alternative is use of the so-called fuzzy reasoning [24,25]. Through this technique, the decision-making processes can be carried out using incomplete, subjective, and fuzzy information [26][27][28]. ...
Article
Full-text available
The concept of life cycle sustainability assessment provides an interdisciplinary space to discuss the main challenges in addressing sustainability from a long-term perspective. However, uncertainty related to cost parameters constitutes one of the main challenges in LCC. This uncertainty prevents managers from precisely anticipating the exact values of important variables. These variables include the costs of spare parts inventories, which in some cases can constitute high percentages of the total cost of a product or service. We propose an activity-based costing model that handles uncertainty using triangular numbers, along with a matrix representation, and Weibull functions to incorporate the projected equipment reliability into the model. A case study based on a spare parts distribution center, which serves the mining industry in Northern Chile, is presented. The results obtained are compared with a traditional costing method. The advantage of using the proposed approach over the traditional one is evident. Future research is directed towards the implementation of a larger number of inventory policies, along with experimentation with other types of fuzzy numbers.
... All input variables were assumed to be independently distributed of each other. Following similar works on the wine sector (Ponstein et al., 2019;Falcone et al., 2016;Meneses et al., 2016), the second analysis step consisted of employing a Monte Carlo simulation (10,000 iterations), which is the predominant method used in LCA studies (Gregory et al., 2013;Henriksson et al., 2015;Pryshlakivsky and Searcy, 2017;Ross and Cheah, 2017;Song et al., 2018). This method generates random variables selected by the imposed distribution functions for each input parameter. ...
Article
As wine supply chains become increasingly globalized, sustainability issues take on ever greater importance. This is the first study to analyse the environmental sustainability aspect of greenhouse gas (GHG) emissions from a global wine supply chain perspective, covering just over 90% of Finland's wine imports. Lacking substantial domestic production capacity, virtually all wine consumed in Finland is imported. Finland is comparable to its Nordic neighbours, Sweden and Norway, in this respect. The Life Cycle Assessment (LCA) methodology was combined with sensitivity and scenario analyses to investigate GHG emissions implications from prospective policy changes. Our results spotlight differences related to wine production in the eight main wine producing countries for the Finnish market (Australia, Chile, France, Germany, Italy, Spain, South Africa, and the United States), related logistics, and all packaging types for wine used in Finland (glass bottle, bag-in-box, PET bottle, beverage carton, and pouch). We found an average value of 1.23 kg CO2e for 0.75 L wine consumed in Finland, ranging from 0.59 kg CO2e for French wine in a bag-in-box packaging to 1.92 kg CO2e for Australian wine in a glass bottle. After identifying the main GHG emission hotspots in the wine supply chain, our scenario analyses highlight the effects of reducing glass bottle weight, moving away from glass packaging toward bag-in-box, increasing bulk wine export volumes to Finland, and following the European Commission's Energy 2020 strategy which targets increasing energy efficiency by 20 percent.
Article
The increasing market demand for milk powder has not only promoted the production capacity of milk powder, but also increased the impact on the environment. Therefore, it is very important to study the relationship between the environmental impact of milk powder spray drying (MPSD) system and system-related parameters and identify the key parameters to improve the efficiency of the sustainable improvement of the system. Treed Gaussian Process (TGP) and Standardized Regression Coefficients (SRC)methods are used to analyze the sensitivity of the system to environmental impacts. The results show that the inlet air temperature of the drying tower has the greatest impact on the environment of the system, accounting for about 82%, followed by the atomization pressure and the feed pump speed, accounting for about 9% and 8% respectively. Moreover, not only the environmental performance of the system should be improved, but also the quality of milk powder should be guaranteed when optimizing the parameters such as the inlet air temperature of drying tower. This study can help the manufacturers of milk powder and related equipment to determine the priority of improving the system from the perspective of environmental protection.
Chapter
To design environmentally responsible chemical products and processes, the impacts they have across their entire lifetime need to be considered—from production to end of life and beyond. This requires understanding all of the flows of material and energy involved, as well as all of the effects each of these flows can have on the many aspects that affect human health and the environment. Life cycle assessment (LCA) has been developed as a method to help systematically calculate and consider all of these flows and impacts. As the first chapter within the second part of this book focusing on the technical foundation, it provides an introduction through the four phases of LCA covering both the technical theory and the practical application. It introduces some of the key data inventories and impact assessment methods used by professionals today and provides an example that guides the reader through the basics of completing an LCA. As a method, LCA supports the first guiding principle of eco-efficiency, and the concept of net present value is introduced to help calculate and optimize this during the design process.
Article
Full-text available
Introduction Most LCA studies are comparative and to an increasing extent the effects uncertainty are included in LCA results. This raises the question how the best option from a set of product alternatives can be selected when the product scores are uncertain. The starting point of this article is a set of Monte Carlo results for a number of alternative products. Indicators for single product alternatives First we discuss different ways of expressing results for product alternatives separately. This includes a discussion of centrality (mean, median, geometric mean, etc.) and dispersion (standard deviation, standard error, confidence interval, etc.). Indicators of difference for two product alternatives A critical review of approaches to single out the superior option on case of a comparison of two is given. This includes familiar approaches such as $$t$$ t tests, but also lesser known ones such the Bhattacharyya coefficient and Cohen’s $$d$$ d . All approaches are defined, discussed, and illustrated with one consistent, downloadable, example. More than two product alternatives The findings for two products are generalized for the multi-product situation. In particular, the issue of inflation of type I errors in multiple comparisons is discussed. Discussion Two main questions are identified: (1) What is the probability that a randomly selected specimen of product A performs better than a randomly selected specimen of product B? (2) How much will a randomly selected specimen of product A perform better than a randomly selected specimen of product B? These two options can both be relevant, but existing approaches for distinguishing product alternatives address one of these two only, or they even turn out to answer a different, less relevant, question. A proposal for a new indicator that addresses both questions simultaneously is offered and its use is illustrated.
Article
Berry fruits such as black currants are known to have health-promoting effects as they contain high amounts of phytochemicals (polyphenols, anthocyanins) and dietary fibers concentrated in the peel and seeds. So far, the berry pomace by-product obtained from the production of berry juice is mainly subject to incineration and less recognized by companies as an opportunity to create additional value. Material Flow Cost Accounting is a powerful tool when it comes to the identification of resource inefficiencies in production systems and can provide useful information as an ad-hoc analysis supporting investment decisions, such as the extension of the product portfolio. This study investigates the material efficiency potential of a black currant juice production in Germany while considering potentials for cost reduction, revenue generation, and carbon dioxide mitigation resulting from the avoidance and upcycling of food waste. We used Material Flow Cost Accounting in combination with carbon footprinting and probabilistic scenario analysis to examine the subject. The analysis showed that, in terms of the global warming potential, the environmental performance of the main product, black currant juice, benefits from the upcycling of the pomace to a marketable product. From the economic side, we could not demonstrate a reasonable amortization of the additional investment costs. However, the scenario and uncertainty analysis revealed promising optimization strategies which would lead to an increase in profit of about 13%: first, additional use of apple pomace and second, the production of high-quality, organic pomace to achieve a selling price of at least €2.00/kg as a functional food supplement or in pharmaceutical and cosmetic applications.
Article
Full-text available
Background, aim, and scope Uncertainty information is essential for the proper use of life cycle assessment (LCA) and environmental assessments in decision making. So far, parameter uncertainty propagation has mainly been studied using Monte Carlo techniques that are relatively computationally heavy to conduct, especially for the comparison of multiple scenarios, often limiting its use to research or to inventory only. Furthermore, Monte Carlo simulations do not automatically assess the sensitivity and contribution to overall uncertainty of individual parameters. The present paper aims to develop and apply to both inventory and impact assessment an explicit and transparent analytical approach to uncertainty. This approach applies Taylor series expansions to the uncertainty propagation of lognormally distributed parameters. Materials and methods We first apply the Taylor series expansion method to analyze the uncertainty propagation of a single scenario, in which case the squared geometric standard deviation of the final output is determined as a function of the model sensitivity to each input parameter and the squared geometric standard deviation of each parameter. We then extend this approach to the comparison of two or more LCA scenarios. Since in LCA it is crucial to account for both common inventory processes and common impact assessment characterization factors among the different scenarios, we further develop the approach to address this dependency. We provide a method to easily determine a range and a best estimate of (a) the squared geometric standard deviation on the ratio of the two scenario scores, “A/B”, and (b) the degree of confidence in the prediction that the impact of scenario A is lower than B (i.e., the probability that A/B<1). The approach is tested on an automobile case study and resulting probability distributions of climate change impacts are compared to classical Monte Carlo distributions. Results The probability distributions obtained with the Taylor series expansion lead to results similar to the classical Monte Carlo distributions, while being substantially simpler; the Taylor series method tends to underestimate the 2.5% confidence limit by 1-11% and the 97.5% limit by less than 5%. The analytical Taylor series expansion easily provides the explicit contributions of each parameter to the overall uncertainty. For the steel front end panel, the factor contributing most to the climate change score uncertainty is the gasoline consumption (>75%). For the aluminum panel, the electricity and aluminum primary production, as well as the light oil consumption, are the dominant contributors to the uncertainty. The developed approach for scenario comparisons, differentiating between common and independent parameters, leads to results similar to those of a Monte Carlo analysis; for all tested cases, we obtained a good concordance between the Monte Carlo and the Taylor series expansion methods regarding the probability that one scenario is better than the other. Discussion The Taylor series expansion method addresses the crucial need of accounting for dependencies in LCA, both for common LCI processes and common LCIA characterization factors. The developed approach in Eq. 8, which differentiates between common and independent parameters, estimates the degree of confidence in the prediction that scenario A is better than B, yielding results similar to those found with Monte Carlo simulations. Conclusions The probability distributions obtained with the Taylor series expansion are virtually equivalent to those from a classical Monte Carlo simulation, while being significantly easier to obtain. An automobile case study on an aluminum front end panel demonstrated the feasibility of this method and illustrated its simultaneous and consistent application to both inventory and impact assessment. The explicit and innovative analytical approach, based on Taylor series expansions of lognormal distributions, provides the contribution to the uncertainty from each parameter and strongly reduces calculation time.
Article
Full-text available
Background, aim, and scopeA new trend driven by climate change concerns is the interest to label consumer products with a carbon footprint (CF) number. Here, we present a study that examines the uncertainty in the estimated CFs of a liquid and a compact powder detergent and how the uncertainty varies with the type of comparison one wishes to make. Materials and methodsA simplified CF model for detergents, encompassing all life cycle stages, has been used for the calculation of CFs. The CFs for the two detergents were compared under three different cases: (1) a situation where most of life cycle assessment (LCA) system is similar, (2) a situation where the LCA background systems may be different but certain choices with regard to system boundaries are standardized, and (3) a situation where the LCA background system, choices of system boundaries, and foreground system may also be different. Uncertainty in the CFs was calculated for each of the three comparison situations using a stepwise sensitivity/uncertainty analysis approach. ResultsThe stepwise approach makes it possible to obtain reliable uncertainty estimates without the need to have very good uncertainty descriptions for every input parameter. Only a few input parameters were found to drive the uncertainty of the CF values. For case 1, the uncertainties in the difference between the CF of the ultraliquid and compact powder products are very small. The CF of compact powder is always larger than that of the ultraliquid product. In case 3, the uncertainties become much larger, such that in 23% of the cases, a CF comparison would wrongly indicate that the compact powder product has a lower CF than the ultraliquid product. Case 2 falls between the extremes of cases 1 and 3. DiscussionOne of the challenges of developing user-friendly CF methods based on the ISO 14040 framework is to ensure a high level of comparability of CF values, such that misleading or oversimplified conclusions can be avoided. Our analysis shows how the uncertainty margins around the calculation of a CF for a set of given products will broaden as the assessment moves from an “internal” comparison to a comparison with data from third parties where there is no specific information how these data have been obtained. CF calculations based on internal comparisons can lead to very clear distinctions between products and illustrate the utility of a CF tool to optimize the environmental performance of products using difference analysis. ConclusionsCF calculations for products can only provide a fair comparison if the LCA background system used for the two products is the same and exactly the same choices in the foreground system are made. In practice, this would require consultation and agreement on specific product category rules. Recommendations and outlookSimplification is needed for a wider adoption of uncertainty analysis in CF and LCA. This article introduces some first steps towards such simplification, but more work is needed both on the theoretical and practical aspects of simplified uncertainty analyses.
Article
Carbon footprints for several shopping bag alternatives (polyethylene, paper, cotton, biodegradable modified starch, and recycled polyethylene) were compared with life cycle assessment. Stochastic uncertainty analysis was used to study the sensitivity of the comparison to scenario and parameter uncertainty. On the basis of the results, we could give only a few robust conclusions without choosing a waste treatment scenario or limiting the parameter space. Given the scenario of current waste infrastructure in Finland, recycled polyethylene bags seem to be the most preferable (−7 to 24 g CO2 eq./bag) and biodegradable bags the least preferable (38 to 60 g CO2 eq./bag) option. In each analyzed waste treatment scenario, a few parameters dominated the uncertainty of results. Most of these parameters were downstream of the shopping bag manufacturing (consumer behavior, landfill conditions, method of waste combustion, etc.). The choice of waste treatment scenario had a greater effect on the ranking of bags than parameter uncertainty within scenarios. This result highlights the importance of including several scenarios in comparative life cycle assessments.
Article
Life-cycle assessment (LCA) practitioners build models to quantify resource consumption, environmental releases, and potential environmental and human health impacts of product systems. Most often, practitioners define a model structure, assign a single value to each parameter, and build deterministic models to approximate environmental outcomes. This approach fails to capture the variability and uncertainty inherent in LCA. To make good decisions, decision makers need to understand the uncertainty in and divergence between LCA outcomes for different product systems. Several approaches for conducting LCA under uncertainty have been proposed and implemented. For example, Monte Carlo simulation and fuzzy set theory have been applied in a limited number of LCA studies. These approaches are well understood and are generally accepted in quantitative decision analysis. But they do not guarantee reliable outcomes. A survey of approaches used to incorporate quantitative uncertainty analysis into LCA is presented. The suitability of each approach for providing reliable outcomes and enabling better decisions is discussed. Approaches that may lead to overconfident or unreliable results are discussed and guidance for improving uncertainty analysis in LCA is provided.
Article
Results of product assessments are often criticised as to their handling of uncertainty. Therefore, it is necessary to develop a comprehensive methodology reflecting parameter uncertainty in combination with uncertainty due to choices in the outcome of LCAs. This paper operationalises the effect of combined parameter uncertainties in the inventory and in the characterisation factors for global warming and acidification for the comparison of two exemplary types of roof gutters. For this purpose, Latin Hypercube sampling is used in the matrix (inventory) method. To illustrate the influence of choices, the effect on LCA outcomes is shown of two different allocation procedures in open-loop recycling and three time horizons for global warming potentials. Furthermore, an uncertainty importance analysis is performed to show which parameter uncertainties mainly contribute to uncertainties in the comparison and the separate environmental profiles of the product systems. These results can be used to prioritise further data research.
Article
The evaluation of uncertainty is relatively new in environmental life-cycle assessment (LCA). It provides useful information to assess the reliability of LCA-based decisions and to guide future research toward reducing uncertainty. Most uncertainty studies in LCA quantify only one type of uncertainty, i.e., uncertainty due to input data (parameter uncertainty). However, LCA outcomes can also be uncertain due to normative choices (scenario uncertainty) and the mathematical models involved (model uncertainty). The present paper outlines a new methodology that quantifies parameter, scenario, and model uncertainty simultaneously in environmental life-cycle assessment. The procedure is illustrated in a case study that compares two insulation options for a Dutch one-family dwelling. Parameter uncertainty was quantified by means of Monte Carlo simulation. Scenario and model uncertainty were quantified by resampling different decision scenarios and model formulations, respectively. Although scenario and model uncertainty were not quantified comprehensively, the results indicate that both types of uncertainty influence the case study outcomes. This stresses the importance of quantifying parameter, scenario, and model uncertainty simultaneously. The two insulation options studied were found to have significantly different impact scores for global warming, stratospheric ozone depletion, and eutrophication. The thickest insulation option has the lowest impact on global warming and eutrophication, and the highest impact on stratospheric ozone depletion.
What’s better for the environment, electric hand dryers or paper towels?
  • C Adams