Renewable Energy 30 (2005) 109–130

www.elsevier.com/locate/renene

Life cycle assessment of a solar thermal

collector: sensitivity analysis, energy
and environmental balances
Fulvio Ardente, Giorgio Beccali, Maurizio Cellura
,
Valerio Lo Brano
Dipartimento di Ricerche Energetiche ed Ambientali (DREAM),
Università degli Studi di Palermo, Viale delle Scienze, 90128 Palermo, Italy
Received 9 March 2004; accepted 5 May 2004

Abstract

Starting from the results of a life cycle assessment of solar thermal collector for sanitary
warm water, an energy balance between the employed energy during the collector life cycle
and the energy saved thanks to the collector use has been investigated. A sensitivity analysis
for estimating the effects of the chosen methods and data on the outcome of the study was
carried out. Uncertainties due to the eco-profile of input materials and the initial assump-
tions have been analysed.
Since the study is concerned with a renewable energy system, attention has been focused
on the energy indexes and in particular the ‘‘global energy consumption’’. Following the
principles of Kyoto Protocol, the variations of CO2 emissions have also been studied.
# 2004 Elsevier Ltd. All rights reserved.

Keywords: Life cycle assessment (LCA); Solar thermal collector; Energy payback time; Sensitivity
analysis

1. Introduction

The life cycle assessment (LCA) is a useful tool to estimate the effective energy
and environmental impacts related to products or services. However, the results of
LCA do not represent ‘‘exact’’ and ‘‘precise’’ data, but are affected by a multitude
of uncertainty sources.

Corresponding author. Tel.: +39-91-236-131; fax: +39-91-484-425.
E-mail address: mcellura@dream.unipa.it (M. Cellura).

0960-1481/$ - see front matter # 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2004.05.006
110 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

The reliability of LCAs strictly depends on complete and sharp data that unfor-
tunately are not always available [1]. ISO 14040 recommends to investigate all
those parameters that could heavily influence the final eco-profile [2]. Because Life
Cycle Inventory (LCI) results are generally used for comparative purposes, the
quality of data is essential to state whether results are valid or not [3–5]. Regarding
data quality, LCA studies should include: time-related coverage, geographical
coverage, technology coverage, precision, completeness and representativeness of
data consistency and reproducibility of methods used throughout the LCA, sources
of the data and their representativeness, uncertainty of the information [2].
The international standards give little practical guidance on how to manage such
information. In addition to previously listed parameters, other sources of uncer-
tainty are [6]:

– Data inaccuracy (due to errors and imperfection in the measurements);

– Data gaps or not representative data;
– Structure of the model (as simplified model to represent the functional relation-
ships);
– Different choices and assumptions;
– System boundaries definition;
– Characterisation factors and weights (as those used in the calculation of poten-
tial environmental impacts);
– Mistakes (unavoidable in every step of LCA).

Furthermore, the global environmental balance of a product is strictly related to

the service life (‘‘Period of time after installation during which all essential properties
of an item meet or exceed the required performance’’ [7]) and durability (‘‘Capability
of an item to perform its required function over a period of time’’ [7]) concepts. The
durability is certainly a key element since LCA takes the life cycle of the material
into account, which includes its use over a number of years: by increasing the
length of the service life, the use of resources is improved as much as specific
impacts are reduced. Design concepts, aiming to improve the environmental per-
formance of a product, should include the design for durability and the design for
longevity including, for example, concepts of reparability, maintainability and
upgradability [8,9]. However, even the durability assessment implies many pro-
blems and uncertainties as: non-reproducibility and traceability of field tracking
studies, subjectivity of expert opinion, length of accelerated tests and natural
weathering, relevance of stress test, required quality and quantity of knowledge for
modelling [10].
Moreover, the study of uncertainty sources is itself affected by uncertainty. It is
necessary to distinguish uncertainty, which arises due to the lack of the knowledge
about the true value of a quantity, from variability that is attributable to the natu-
ral heterogeneity of values [6]. Uncertainty could be reduced by more precise and
accurate measurements while variability is entailed into processes. Details con-
tained in the normal LCI study do not often allow distinguishing uncertainty from
variability. Consequently, in this study, they will be jointly considered.
 F. Ardente et al. / Renewable Energy 30 (2005) 109–130 111

Starting from the results of the LCA applied to a solar thermal collector [11–13],
a sensitivity analysis (SA) has been carried out. This is a systematic procedure for
estimating the effects on the outcome of a study of the chosen methods and data [14].
SA can be applied with either arbitrarily selected ranges of variation, or variations
that represent known ranges of uncertainty.
SA is an important element of judgement for the corroboration or the refutation
of the scientific hypotheses embedded into a model. This is particularly critical
when both model parameters and available data are affected by uncertainties (as
occurs in LCAs). However, SA can also be used to direct the research priorities by
focusing on the parameters which mostly determine the uncertainty of the model.
The results presented in this work are extracted from the case study ‘‘CS2’’ per-
formed within the works of Task 27—Subtask C of International Energy Agency
(IEA) about ‘‘Performance, durability and sustainability of advanced windows and
solar components for buildings’’. The study follows three main steps:

1. Estimation of the energy and environmental balances: being concerned with a

renewable energy system and following the principles of Kyoto Protocol, the
attention of this study has been focused on the energy and CO2 indexes. In
particular, the energy and the emission payback times have been calculated.
2. Study of the uncertainty sources: the eco-profile has been studied in detail. The
study focused on the main uncertainty sources in order to estimate their influence
on final results. Data quality regarding raw materials and the main LCA initial
assumptions (as system boundaries or impacts allocation) has been analysed.
3. Incidence of uncertainties on the environmental indexes: following the previous
considerations, the uncertainties of a parameter have been depicted as a variation
range of the energy and environmental indexes. Analogously, a scenario analysis
to describe the incidence of the different assumptions has been carried out.

2. Synthesis of LCA results

The energy and environmental analyses have been performed on the results of an
LCA applied to a passive1 solar collector for domestic warm water demand [11].
The studied functional unit (FU) includes: the absorbing collector, the water tank
(180 l capacity) and the support for the roof fastening. The following life-cycle
phases have been investigated:

– Production and delivery of energy and raw materials;

– Production process;
– Installation;
– Maintenance;
– Disposal;
– Transports occurring during each step.

1
Passive collectors are those whose thermal fluid circulation occurs naturally.
112 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

Fig. 1. Synthesis of solar collector eco-profile.

The analysis has shown a global primary energy consumption of about 11.5 GJ
mainly related to the manufacture of the absorbing collector (32.2%) and of the
water tank (38.7%). Other significant contributions are: maintenance (11.6%),
manufacture of the support (9.6%), and transport of raw materials (3%). The main
environmental impacts arise from air emissions. Details of the collector eco-profile
are shown in Fig. 1.

3. The energy and environmental balances

The energy saving referred to the use of solar collector has been calculated con-
sidering the average temperatures and solar inputs of the city of Palermo (Italy,
v
38 latitude) [12,15,16]. The yearly average energy harvest is close to 6.2 GJ per
 F. Ardente et al. / Renewable Energy 30 (2005) 109–130 113

year. This is an end-energy quantity (meaning the energy effectively saved by the
final consumers) and it has to be converted into primary energy2. The useful pri-
mary energy saving ‘‘Euseful’’ is estimated to be 6.6 GJ per year [12].
3.1. The energy payback time

The energy payback time is a key indicator to evaluate the eco-performance of a

renewable energy source. It is an indicator generally used in economic studies to
state the time to recover an initial investment. The energy payback time (EPT) can
be likewise defined as the time necessary for a solar equipment to collect the energy
(valued as primary) equivalent to that used to produce it [12]:
LCAenergy
EPT ¼ ð1Þ
Euseful  Euse
where LCAenergy, primary energy consumed during all the life-cycle phases (GJ);
Euseful, yearly useful saved energy (GJ per year); Euse, energy employed during the
use of the renewable system (GJ per year).
In passive collector systems, the water circulation occurs naturally, without
employment of pumps; consequently, the term ‘‘Euse’’ is null. The payback time
related to the studied equipment results lower than 2 years. This value shows the
great energy convenience of such technology.
3.2. Emission payback time
Knowing the yearly ‘‘Euseful’’, we have calculated the yearly emission saving
(EMS-i). It represents the emissions that the auxiliary system would produce to
deliver as much energy as that saved by means of the solar collector. The EMS
depends on the typology of the employed auxiliary heater. The global impacts dur-
ing the life cycle and the emission saving are summarised by the emission payback
time (EMPT). It is defined as the time during which the avoided emissions due to
the employment of the solar plant are equal to those released during the pro-
duction and use of the renewable plant itself. It is possible to calculate the EMPT
relative to the pollutant ‘‘i’’ as [13]:
EMi
EMPT-i ¼ ð2Þ
EMS-i  EMuse-i
where EMi, global emissions of generic pollutant ‘‘i’’ related to the production,
assembly, transport, maintenance and disposal of the solar plant (kgi); EMS-i,
yearly emission saving of generic pollutant ‘‘i’’ (kgi per year); EMuse-i, yearly emis-
sion of pollutant ‘‘i’’ related to the use of the renewable plant (kgi per year).

2
Supposing the renewable system supported by a conventional gas boiler, the conversion factor (from
the end energy to the primary one) has been assumed to be 1.06 MJPRIM/MJEND [12]. The conversion
factor includes every energy loss occurring during extraction, refining and distribution of gas to users.
The efficiency of the domestic gas boiler has been not considered.
114 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

The EMuse is caused by the use of the conventional energy that the plant needs
to work (mainly the electricity used by pumps).
The global warming potential (GWP) related to the collector life cycle is
721 kgeq CO2 [11]. Considering a domestic gas boiler, it is assumed a specific global
warming factor of 65:7  103 kgeq CO2 per MJ of useful heat [17]. The yearly
CO2 eq emission saving is estimated to be 407 kgeq CO2 [13]. Similar to the energy
payback time, even the CO2-payback time resulted lower than 2 years.

4. Sensitivity analysis of the study

The sensitivity analysis (SA) was performed on the input data and assumptions
of LCA in order to estimate those responsible for the greatest eco-profile uncer-
tainties [18]. The calculation has been performed adopting a linear model and the
following elements were analysed:

– Eco-profile of input materials;

– Eco-profile of electricity;
– Transport of raw materials, installation, maintenance and disposal steps.

4.1. Sensitivity analysis upon input materials

The embodied energy of materials represented about 80% of the overall con-
sumption [11]. Consequently, the SA focused upon input materials. Table 1 lists
the percentage incidence3 of each material on the overall energy balance. We
choose to investigate those materials whose incidence on the total energy require-
ment is greater than 1%. The study refers to various environmental databases
[17,19,20]. When possible, data have been compared to a study performed by spe-
cialised companies (e.g. regarding aluminium, steel and plastic products).
Galvanised steel is the main constituent of the functional unit (FU), representing
more than half of the overall employed mass. Having also a great specific value of
embodied energy, galvanised steel is responsible for about 37% of the overall
energy consumption. In the collector LCA, it has been assumed that this material
involves the consumption of about 38 MJ/kg of embodied energy and the emission
of 2:4 kgCO2 =kg [19]. More detailed research of the scientific literature has shown a
great variability of these impacts depending on the production process. In parti-
cular, an overestimation of the embodied energy has been observed. The values cal-
culated for 11 different production sites have shown the variability of the embodied
energy from 27.3 to 37.9 MJ/kg and the variability of CO2 emission from 1.8 to
2.8 kgCO2 =kg [21]. Considering an average eco-profile of galvanised steel, it has
been calculated that galvanised steel causes primary energy consumption of
3.43 GJ (with a variation range from 2.67 to 4.27 GJ) and emission of 250 kgCO2
(with a variation range from 204 to 313 kgCO2 ).

3
The ‘‘incidence’’ percentage has been calculated as the ratio between the energy consumption related
to a material divided by the global energy consumption.
 F. Ardente et al. / Renewable Energy 30 (2005) 109–130 115

Table 1
Incidence of embodied energy of materials on the global energy balance
Material Incidence (%)
Galvanised steel 37.2
Thermal fluid 12.6
Copper 9.8
Steel 9.3
Rigid PUR 7.6
Aluminium 5.0
Cardboard 2.0
Magnesium 1.6
Epoxy dust 1.4
Glass 1.3
HDPE 0.6
LDPE 0.6
Brass 0.1
Flexible PUR 0.03
PVC 0.02

The collector employs about 12.5 kg of thermal fluid composed of a mixture of

water and propylene glycol to avoid freezing problems during the cold season. The
fluid mixture can vary from 20% (employed in warm areas) to 50% of glycol (sites
with low temperature in the winter season). The eco-profile of Fig. 1 has been cal-
culated considering a 50% mixture. This choice can result in an over-estimation of
the global energy consumption. Fig. 2 shows the eco-profile of thermal fluid con-
sidering a mixture of demineralised water and glycol in variable percentage [19,20].
Considering as reliable the average condition (correspondent to a 35% glycol mix),
it has been calculated that the thermal fluid causes primary energy consumption of
1.1 GJ (with a variation range of 0.467 GJ) and emission of 26 kgCO2 (with a
variation range of 11 kgCO2 ).
In the solar collector, copper is used for the production of the absorbing plate
and the pipes for the fluid circulation. The study of the copper eco-profile has

Fig. 2. Embodied energy and CO2 emissions for variable percentages of glycol in the thermal fluid.
116 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

shown a great uncertainty and variability of the energy and environmental data
due to differences in the production process (ascribable to the use of heat for melt-
ing and electricity in the electrolysis) and to the ratio of reused copper scraps. An
average copper eco-profile has been calculated on the basis of various references
[17,20]. Under this hypothesis, copper parts involve the primary energy consump-
tion of 1 GJ (with an uncertainty of 0.23 GJ) and the emission of 62 kgCO2 (with
an uncertainty of 18 kgCO2 ).
Aluminium has a sensible incidence in the global energy balance mainly due to
the high specific energy consumption related to its production. The SA regarding
this material has been referred to data representative of the average production in
Europe [22]. A great variability has been detected depending on the percentage of
recycled material. Fig. 3 shows the variation of the aluminium sheet eco-profile
(calculated on the basis of Italian energy mix) related to the recycled fraction.
When using 30% recycled material, the embodied energy related to aluminium is
584 MJ (with a variation range from 111 to 787 MJ) and the CO2 emission is 33 kg
(with a variation range from 5 to 45 kgCO2 ).
A further detail is necessary for stainless steel parts. The production of the solar
collector implies the use of 29 kg of this material. Detailed research has shown a
shortage of environmental information concerning this material. Lacking infor-
mation regarding the Italian stainless steel eco-profile, it has been computed
as normal steel in the LCA [11]. It has also been assumed to refer to average
European data related to cold rolled austenitic (grade 304) stainless steel [23]. With
respect to normal steel, the production of stainless steel requires a larger amount of
energy (62 MJ/kg with an increase of about 75%) and larger environmental
impacts (emission of CO2 is more than doubled with a specific factor of
6:2 kgCO2 =kg). Assuming to refer to these new specific impact values, the initial
eco-profile has been sensibly modified. It has been estimated that stainless steel
causes the consumption of 1.8 GJ (causing the energy consumption to increase of
770 MJ) and the emission of 180 kgCO2 (causing the CO2 emissions to increase of
94 kgCO2 ). In contrast to other materials, it has not been possible to compare the

Fig. 3. Embodied energy and CO2 emissions for variable percentages of recycled aluminium.
 F. Ardente et al. / Renewable Energy 30 (2005) 109–130 117

stainless eco-profile with other references and to calculate a sensitivity variation

range. It is also important to point out that the use of data regarding normal steel
instead of stainless steel drastically changes the final eco-profile. This example
stresses a key question in the LCA: in a generic LCA study, a clear description of
limits and assumptions is necessary, particularly, when we need to compare
replaceable products.

4.1.1. Synthesis of data quality

The data quality related to input materials has been investigated throughout. In
particular, Table 2 lists:

– Age of data or period which data refer to;

– Geographical coverage: location which data refer to (when possible eco-profiles
have been adapted to Italian case study by using the Italian energy mix);
– Technological level: describes the functional unit and the whole process to pro-
duce it;
– Completeness of data (underlying possible gaps);
– Representativeness: represents the extent to which a set of measurements taken
in a space–time domain reflects the actual conditions in the same or different
space–time domain;
– Transparency of the study (considering whether the study boundaries, inputs and
outputs are clearly described);
– Relevance: summarises whether materials have a great incidence on the study or
not;
– References.

When possible, uncertainties of the previous indexes have been summarised in

the energy and emission variation ranges. Actually, the width of these ranges can
be related to uncertainties as much as to the natural variability in the process.
However, as previously described, having no way to distinguish uncertainties from
inner variability, they are managed together. Following the experience of experts, a
qualitative judgement on the basis of the previous indexes has been formulated.
In the light of this approach, the collector eco-profile has been recalculated
involving data considered more representative, transparent and reliable. It has been
calculated that the life cycle of the solar collector implies the global primary con-
sumption of 11 GJ and the release of 700 kg of CO2. Due to the variability of
input materials eco-profile, it is estimated that the energy consumption could
change from 8.9 to 13 GJ, while CO2 emission could vary from 580 to 820 kg
(Fig. 4).
In turn, the variations of these indexes caused by each material have been stud-
ied, taking all the others fixed to their reference values. Fig. 5 shows the incidence
of each input material to the global FU eco-profile. Regarding galvanised steel, it
is possible to observe that, modifying its eco-profile, the global energy consump-
tion changes from 10.2 to 11.8 GJ and the CO2 emission changes from 655 to
Table 2 118
Data quality of input materials
Material Age Geographical Technology Completeness Representa- Energy CO2 emis- Transparency Relevance Data Reference
coverage coverage of data tiveness variation sion vari- of data in the study quality
range ation range indicator
(MJ/kg) (kgCO2 =kg)
Galvanised 1994– Average of 11 Hot-dip gal- Low air emis- Data could 27.3–37.9 1.8–2.8 Medium Very high Good [19,21]
steel 1995 sites vanised steel sion details, be assumed
coil not clear data as European
about energy average
consumption
Stainless not Average of Cold rolled Low detail Data could 62.1 6.2 Very low Very high Medium [23]
steel speci- the most austenitic be assumed
about the input
fied important stainless materials and as European
European steel coil the energy con-
average. No
producers (grade 304) sumption other simi-
lar study
available
which can
be com-
pared to
Thermal 1998 Data adapted Mix demi- All the main Average 17–41 0.4–1.0 Medium High Good [19,20]
fluid to Italian case neralised energy and value from
study water and environmental two Eur-
propylene impacts are opean stu-
glycol shown dies, recal-
F. Ardente et al. / Renewable Energy 30 (2005) 109–130

culated with
Italian
energy mix
Copper 1980– Estimation Copper with All the main Average 57–91 3.3–5.9 Low High Low [17,20]
1990 from other 40–50% of energy and value from
European stu- reused environmental two Eur-
dies adapted scraps impacts are opean stu-
to Italian case shown dies, but
study one is very
old and not
complete,
and the
other is not
transparent
Rigid PUR 1997 Average of Poly- All the main Representa- 105–118 3.4–3.8 High High Very [17,19]
European urethane energy and tive of the good
producers. foam used environmental European
Data adapted as thermal impacts are average.
to Italian case insulation shown Adapted
study with Italian
energy mix
Aluminium 2000 Average of Aluminium All the main Average 28–198 1.3–11.3 Very high High Very [22]
(data 70–90% of the cold rolled energy and European good
from European sheet environmental data about
1992 to producers impacts are primary alu-
1994) shown minium.
Estimation
of recycling
process.
From 30 to
F. Ardente et al. / Renewable Energy 30 (2005) 109–130

100% recy-
cled alu-
minium
119
120 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

Fig. 4. Summary of energy and emission global variation ranges.

762 kgCO2 . These results show that the impacts related to input materials could sen-
sibly change the FU eco-profile. It is possible to make the following considerations:

– The global energy consumption can vary by about 20% from its reference
value of 11.0 GJ;
– CO2 emission can vary by about 17% from the reference value of 700 kgCO2 ;
– The variation ranges are not symmetric and depend on asymmetric ranges of
input materials (Fig. 5);
– Galvanised steel, the dominant material, is also the component whose uncer-
tainty is greatest. It is responsible for 40% of the global uncertainty on the

Fig. 5. Sensitivity analysis due to input materials.

 F. Ardente et al. / Renewable Energy 30 (2005) 109–130 121

energy consumption and for 50% of that related to the CO2 emission;
– PUR, magnesium and cardboard are responsible for a low incidence in the vari-
ation range (less than 5%). The great variability of the eco-profiles of aluminium,
copper and thermal fluid causes large variation ranges (from 10% to 20%).

4.2. Sensitivity analysis of electricity eco-profile

Electricity is the only energy source directly employed during the product manu-
facturing. The primary global consumption due to electricity is estimated to be
540 MJ (considering the hypothesis of medium voltage electricity produced with
the average Italian mix [17]). About 5% of global impacts are ascribable to the
power energy use. To compute the uncertainty related to this energy source, five
different scenarios have been considered:

Scenario 1 Electricity referred to the medium voltage electricity (base scenario)

[17];
Scenario 2 Electricity referred to the low voltage electricity [17]. The eco-profile of
low voltage electricity takes into account the energy losses for distri-
bution and transformation;
Scenario 3 Electricity referred to the average Italian energy mix (no detail for the
voltage) [20];
Scenario 4 Electricity referred to the regional case study. Data have been estimated
on the basis of the regional electricity production mix (97.04% thermo-
electric, 2.94% hydroelectric and 0.02% wind farms) [20,24];
Scenario 5 Electricity referred to the average European energy mix (no detail for
the voltage) [19].

The primary energy consumption and the main air emissions (CO2, CO, SOx,
NOx, dust) have been the two criteria to compare the different scenarios. It is poss-
ible to observe that (Fig. 6):

– The energy consumption could vary from 540 to 610 MJ while carbon dioxide
emission from 22 to 40 kgCO2 ;
– The incidence of the production process into the global energy consumption has
small variation (from 5% to 6%) while incidence into the CO2 emission may
change from 3% to 7%;
– The greatest energy and CO2 impacts are those related to scenario 4.

Regarding the emission values, in particular the carbon dioxide, the quality of
data used in scenario 4 is good. In fact, these values come from direct measure-
ments that the regional power company makes on the main power plants. We
could make the following comments:

– The greatest variations occur in scenario 3 that, in particular, overestimates SOx

and NOx.
122 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

Fig. 6. Comparison of electricity scenarios.

– In general, the small variations affecting energy and CO2 values suppose a good
reliability of electrical eco-profiles.
– On the contrary, data regarding other pollutants have very large variations.
However, this does not necessarily mean low data reliability but could be related
to the different energy mix. For example, lower impacts affect the Scenario 5 that
supposes a greater percentage of renewable energy sources.

4.3. Sensitivity analysis of transports

The transport of raw materials to the production site causes about 3% of global
environmental impacts. The analysis focused on the transported masses and dis-
tances. Investigated transports need diesel trucks exclusively.
Uncertainty grows regarding the transport conditions. Details are missing about
the transport of some materials and in particular regarding plastic components
(coming from north Italy) and the glass (produced abroad). External companies
market these products and we lack detailed data regarding the freights.
Furthermore, the company acquires great stocks of metals in different periods.
The metals are then stored and later used on demand for many different products.
It is not possible to determine the exact number of travels related to the production
of the solar collector, thus we choose to define six different transports scenarios.
Initially, it has been assumed to employ medium capacity trucks exclusively. The
consumptions refer to Italian database [17] (Scenario 1) or to other databases [20]
 F. Ardente et al. / Renewable Energy 30 (2005) 109–130 123

(Scenario 2). Both scenarios refer to average conditions of roads and traffic and a
50% load factor.
Scenarios 3 and 4 suppose that long distance transport occurs entirely or partially
by means of high capacity trucks. Regarding the other three scenarios, it has been
assumed they use low capacity trucks for the regional transports and extra-regional
transport with medium trucks (Scenario 5) or assuming to neglect extra-regional
transport (Scenario 6). In particular, the last scenario assumes all the supplies
within the regional context. Data concerning the fuel consumption come from the
Italian database [17]. From their comparison, it possible to observe that (Fig. 7):

– The lower the capacity of the trucks, the larger is the related impact. In parti-
cular, scenario 5 has the highest impact while scenario 3 has the lowest;
– The incidence of transport on the global energy and CO2 balances varies from
2.5% to 5%;
– Comparison between different databases (Scenarios 1 and 2) shows that emission
could have sensible variations;
– Extra-regional transport has a considerable weight on the global impacts of
transport. Their incidence varies from 50% (Scenario 1) to 25% (Scenario 3);
– The hypothesis of employing only regional purchasers would decrease the collec-
tor impacts by up to 1%.

Fig. 7. Comparison of different transport scenarios.

124 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

4.4. Sensitivity analysis of the installation phase

Following the installation procedures carried out by the selling company, it has
been estimated that the global incidence of this life cycle step is less than 2% of the
global environmental impacts. In particular, the effective electricity necessary to
fasten the support to the roof is negligible. The only significant contributions are
those related to the transport of collectors from factory to selling point and,
finally, to the purchaser’s home. However, distance and transport conditions are
very changeable parameters. Transports from factory to selling points are supposed
to occur by 28,000 kg capacity trucks. About the final destination, it is assumed:

Scenario 1 Transport by means of 3500 kg van for a global distance of 15 km;

Scenario 2 Transport by means of 3500 kg van for a global distance of 30 km;
Scenario 3 Transport by means of 3000 kg rigid truck for a global distance of
15 km;
Scenario 4 Transport by means of 3000 kg rigid truck for a global distance of
30 km.

Data regarding 3.5 ton truck refer to Italian data [17], while data regarding
3000 kg truck refer to average European transports [20]. It is possible to observe
that (Table 3):

– Data regarding the two different trucks have the same order of magnitude. The
scenarios calculated referring to 3500 kg truck have larger impacts;
– The incidence of installation process on the collector eco-profile is generally low
(regarding the energy balance, it varies from 1% to 2%, while the regarding the
CO2 balance, it varies from 1% to 3%).

4.5. Sensitivity analysis of maintenance phase

The LCA has shown a non-negligible influence of the maintenance in the energy
and emission balances. In particular, maintenance processes are responsible for
about 10% of the cumulative energy requirements. This amount is mainly due to
the substitution of some collector parts such as the magnesium anode and the elec-

Table 3
Sensitivity analysis of transport conditions during installation phase
Environmental impact Scenario 1 Scenario 2 Scenario 3 Scenario 4
Energy consumption (MJ) 109 175 145 247
CO2 (kg) 6.3 10.2 10.9 19.4
CO (g) 26.6 46.2 13.8 20.7
NOx (g) 46.2 61.2 116.3 201.5
SOx (g) 3.2 4.4 36.4 70.7
Dust (g) 7.1 12.4 8.6 15.6
VOC (g) 8.1 12.8 3.3 3.3
 F. Ardente et al. / Renewable Energy 30 (2005) 109–130 125

trical resistance (subjected to corrosion and fouling problems), sealing, gaskets and
thermal fluid.
Propylene glycol is the component responsible for the main environmental
impacts during the maintenance phases. The high temperatures reached during the
hot season can cause the fluid to evaporate. A security valve in the boiler is
designed to decrease the pressure into pipes and avoid damage to the collector.
Furthermore, during a long operational period, the thermal properties of the fluid
could be modified.
Following the previous consideration about products’ durability, a more detailed
study about the efficiency of the fluid over the years would be necessary. Actually,
the company does not have direct measurements, and consequently the problem
has been investigated supposing different scenarios:
Scenario 1 Two4 maintenance cycles with only 20% fluid refilling;
Scenario 2 Three maintenance cycles with only 20% fluid refilling;
Scenario 3 Two maintenance cycles with total fluid substitution;
Scenario 4 Three maintenance cycles with total fluid substitution.
Fig. 8 shows the results. From scenario 1 to scenario 4, CO2 emissions are dou-
bled while the energy consumption is trebled. This suggests that the incidence of
the maintenance on the global energy balance varies from 5% to 10% and incidence
on carbon dioxide balance varies from 4% to 8%. It is possible to observe that the
maintenance steps involve significant impacts even with the partial substitution of
the fluid (Scenarios 1 and 2). In every scenario, the contribution of maintenance to
LCA results is never negligible.
4.6. Sensitivity analysis of disposal and recycling
Regarding the FU disposal, no data are available. In fact, the firm started pro-
duction of solar collectors a few years ago and, consequently, the collectors sold
have not yet reached their ‘‘end-life’’. Data regarding disposal come from estima-
tions. Six different scenarios have been analysed (Fig. 9). It is important to under-
line that the disposal scenarios take into account only the impacts related to the
transport.
The first end-life scenario supposes the disposal to the nearest landfill (Scenario
1). The results show a negligible contribution of this hypothesised process to the
global eco-profile (less than 0.2%). This very low value could be related to the
neglected impacts regarding the landfill management and to the assumption that
transport occurs by large trucks (such those used for the normal waste collection).
The calculation has been repeated assuming to dispose of the collector by low
capacity truck (Scenario 2). This assumption will involve the energy consumption
of 250 MJ (2.2% of the global energy demand) and the release of 18 kgCO2 (2.4% of
the overall CO2 emission).

4
The number of maintenance cycles has been established considering that the technicians operate
every 4/5 years and supposing 15 years of collector service life [11].
126 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

Fig. 8. Maintenance—sensitivity analysis.

The environmental impacts would rise further if the distance covered was
increased. In scenario 3, we suppose that the collector, after its service life, is
brought to the factory and disassembled before disposal. The global distance can
vary from 50 to 100 km and, consequently, the energy consumption could vary
from 250 to 500 MJ (till 5% of the global energy demand) and CO2 emission could
vary from 18 to 35 kgCO2 (5% of the overall CO2 emission).
Regarding the possible recycling scenarios, no data are available. Following the
principle of design for recycling, the possibility to recycle or reuse some collector
parts has been considered. The only recycling that sometimes occurs is concerned
with the use of steel scraps to produce smaller parts (as bolts or connections) and
their successive employment inside the collector or other products. However, it has
been not possible to precisely measure this recycled flow. In scenario 4, collector
bolts (0.6 kg of iron) have been supposed as produced from production steel
scraps. This assumption would involve negligible reduction of impacts (about 0.2%

Fig. 9. Effects on the collector eco-profile of different disposal and recycling scenarios.
 F. Ardente et al. / Renewable Energy 30 (2005) 109–130 127

of the environmental impacts and, in particular, a reduction of 18 MJ and

1.5 kgCO2 ).
Considerable reduction of impacts could be obtained with the reuse/recycle of
some parts, as the selective glass (Scenario 5) or the support (Scenario 6). How-
ever, these assumptions should follow the durability considerations as the decay of
selective properties or corrosion problems. The scenario analysis shows that
(Fig. 9):

– The previous assumptions regarding disposal underestimate the effective environ-

mental impacts. From a superficial observation of the scenarios, one could
believe that the environmental impacts of disposal are only related to transport.
Actually, there are many other environmental problems that have been not
included in the energy and environmental balances (such as waste management,
use of soil, landfill management, contamination of soil–water and air, resource
depletion, etc.);
– Following scenarios 2 and 3, the incidence of disposal on the global impacts
could vary from 2% to 5%;
– On the other hand, the adoption of reuse and recycling processes could sensibly
reduce the overall impacts (till 5% of energy consumption and 6% of CO2 emis-
sions). However, these results come from hypotheses that should first be verified
and adapted to productive and economic requirements.

5. Synthesis of main results

Considering the variability due to the eco-profiles of materials, it has been esti-
mated that:

– The global primary energy consumption can vary from 8.9 to 13.0 GJ, with a
variation range of about 20% from the referring value of 11.0 GJ;
– The CO2 emission can vary from 581 to 815 kgCO2 , with a variation range of
about 17% from the referring value of 700 kgCO2 .

Regarding the contribution of each life cycle step to the global energy consump-
tion and the CO2 emission, a scenario analysis has been employed, obtaining the
following results:

– The incidence of transport on the global energy and CO2 balances varies from
2.5% to 5%. A considerable incidence is related to extra-regional transport;
– The incidence of the production process on the global energy consumption has
small variation (from 5% to 6%), while incidence into CO2 emission varies from
3% to 7%;
– The incidence of installation process on the global energy balance varies from
1% to 2%. Regarding the CO2 balance, the incidence varies from 1% to 3%;
– The contribution of maintenance to LCA results is significant. Its incidence on
the global energy balance varies from 5% to 10%. On carbon dioxide balance,
128 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

the incidence varies from 4% to 8%. It has been observed that even the partial
substitution of thermal fluid involves significant impacts;
– The analysis of disposal scenarios has shown that its incidence on the global
impacts could vary from 2% to 5%. Considerable reductions of impacts could be
obtained with the reuse of some parts (till 5% of energy consumption and 6% of
CO2 emissions).

Including the variability related to raw material eco-profiles and the uncertain-
ties due to the other life cycle steps, it has been estimated that the variation range
can be extended as following:

– Energy consumption range from 8 to 15 GJ;

– CO2 emission range from 500 to 900 kg.

6. Effects of uncertainties into global energy and environmental balances

The previous sections have shown a great variability of the collector eco-profile.
However, these variations have a small influence on the energy and emission pay-
back times that in any case are enclosed within 1.2 and 2.3 years.
The payback times strictly depend on the yearly values of energy and emission
saving. The estimated values of Section 3 refer to local climate conditions and
assuming that the collector at the top of its efficiency. Actually, the analysis should
also involve the study of the energy performance decay. Complex phenomena,
involving many time dependent processes, can negatively influence the collected
useful energy (as fouling or obstruction problems in pipes and heat exchanger, the
corrosion of steel parts, the decay of coatings employed to improve the absorbing
properties of surfaces, the breaking up of polyurethane foam and generally the
reduction of insulating properties, presence of dirt on glass or other surfaces)
[10,25]. Supposing the energy performances decrease up to 40%, the payback times
have been calculated considering that primary energy saving could vary from 6.6 to
3.8 GJ and the CO2 emission saving from 400 to 250 kg (Fig. 10).

Fig. 10. Uncertainty on payback times.

 F. Ardente et al. / Renewable Energy 30 (2005) 109–130 129

The upper curves of Fig. 10 refer to the highest eco-profile impacts while the
lower curves regard the lowest impacts. It is possible to observe that, even in pessi-
mistic scenarios, the payback times are always lower than 4 years. The environ-
mental convenience of a collector is strictly related to the above mentioned
concepts of durability and service life. However, a comparison between the life
cycle length and the payback time shows a great environmental convenience of this
renewable plant.

7. Conclusions

The LCA studies have generally an intrinsic uncertainty related to various fac-
tors (i.e. difficulty in the survey of data, lack of detailed information sources, data
quality, etc.). Consequently, it is more important for the experts to evaluate the
order of magnitude of input–output flows ascribable to the product than to trace
an ‘‘exact’’ eco-profile of products.
This problem, commonly noticed in every LCA, has been strongly noticed in our
case study. Regarding the solar thermal collector, we have observed a strong
dependence of the FU eco-profile on the input materials. They globally imply
about 70–80% of the environmental impacts. The environmental impacts of
material have been supposed to be enclosed within a variation range. These inter-
vals have been realised on the basis of data coming from environmental databases,
LCA tools and, in general, from European environmental studies. The analysis of
data quality has been based on many parameters such as geographical coverage,
technological level, representativeness, etc. Results have shown a great uncertainty
regarding aluminium, copper, thermal fluid and galvanized steel, the dominant
material.
Even the other life cycle steps (transports, installation and maintenance) cause
large impacts. The production process affects the eco-profile for about 5% of
impacts (excepting some air pollutants released during cutting and welding steps).
The LCA results have been synthesised into two indexes: the energy and environ-
mental payback times. The great energy and environmental convenience of this
equipment are shown by very low values of payback times (lower than 2 years).
Including the variability related to raw material eco-profiles and the uncertain-
ties due to the other life cycle steps, it has been estimated that the variation range
can be extended as following: energy consumption from 8 to 15 GJ, CO2 emission
from 500 to 900 kg. Joining concepts of durability and supposing a loss of
efficiency up to 40%, it has been estimated that, even in pessimistic scenarios, the
energy and emission payback times are lower than 4 years. These results permit to
state a positive qualitative judgement regarding the environmental performances of
the collector that is not sensibly influenced by all the study uncertainties.

References
[1] Zhang HC, Yu SY. A quantitative approach in environmental conscious product design support.
Proceedings of International Symposium on Electronics and the Environment, 11–13 May, Dan-
vers, MA. 1999, p. 280–5.
130 F. Ardente et al. / Renewable Energy 30 (2005) 109–130

[2] ISO14040. Environmental management—life cycle assessment—principles and framework. 1998.

[3] Huijbregts M, Norris G, von Bahr B. General framework to compare and develop methods for
uncertainty assessment in LCI, working paper from SETAC LCA workgroup on data availability
and data quality. 1999.
[4] Rebitzer G, Fullana P, Jolliet O, Klopffer W. State of the art: advances in LCA and LCM; an
update of the liaison of the two LCA-planets. Int J LCA 2001;6(4):187–91.
[5] Hischier R, Baitz M, Bretz R, Frischknecht R, Jungbluth N, Marheineke T, McKeown P, Oele M,
Osset P, Renner I, Skone T, Wessman H, De Beaufort ASH. Guidelines for consistent reporting of
exchanges from/to nature within life cycle inventories (LCI). Int J LCA 2001;6(4):192–8.
[6] Bjorklund AE. Survey of approaches to improve reliability in LCA. Int J LCA 2002;7(2):64–72.
[7] ISO 15686. Buildings and constructed assets—service life planning—part 1: general principles. 2000.
[8] Commission of the European communities. Green Paper on Integrated Product Policy. COM, 2001,
p. 68.
[9] ISO/TR 14062. Environmental management—integrating environmental aspects into product
design and development. 2002.
[10] Lair J. Failure modes and effects analysis and service life prediction. Report of IEA Task 27—
Building envelope components—Performance, durability and sustainability of advanced windows
and solar components for building envelopes, Sub-task C—Project C2, September 2001.
[11] Ardente F, Beccali G, Cellura M, Lo Brano V. Life cycle analysis of solar thermal collector (first
part: life cycle inventory). First report of the International Energy Agency (IEA)—Task 27—Per-
formance of solar facade, Subtask C—Project C1, Environmental Performance, March 2003.
[12] Ardente F, Beccali G, Cellura M, Lo Brano V. The environmental product declaration EPD with a
particular application to a solar thermal collector. Ecosystems and Sustainable Development,
Advanced in Ecological Sciences no. 18, Witpress publisher, 2003, p. 325–35.
[13] Ardente F, Beccali G, Cellura M, Lo Brano V. The emission saving related to the use of renewable
energy sources: the case study of solar thermal collectors. Air Pollution XI—Eleventh International
Conference on Modelling, Monitoring and Management of Air Pollution, Witpress publisher, 2003,
p. 295–304.
[14] ISO14041. Environmental management—life cycle assessment—goal and scope definition and
inventory analysis. 1999.
[15] Italian Standardisation Organisation UNI 10349. Heating and cooling of buildings. Climatic data,
1994.
[16] Duffie JA, Beckman WA. L’energia solare nelle applicazioni termiche, Liguori, 1978 [Italian lan-
guage].
[17] ANPA—Italian National Agency for environment protection. Banca dati italiana a supporto della
valutazione del ciclo di vita, Italian environmental database version 2.0, 2000.
[18] Ardente F, Beccali G, Cellura M, Lo Brano V. Life cycle analysis of solar thermal collector (second
part: sensitivity analysis of results). Second report of International Energy Agency (IEA)—Task 27—
Performance of solar facade, Subtask C—Project C1, Environmental Performance, October 2003.
[19] Öko-Institut (Institut für angewandte Ökologie). Global Emission Model for Integrated Systems
(GEMIS), Version 4.1, 2002, German environmental database.
[20] Boustead Ltd. Boustead Model, environmental database, 2001.
[21] International Iron and Steel Institute. LCI data of steel products—Hot-dip galvanised coil, June
1999.
[22] European Aluminium Association. Environmental profile report for the European aluminium
industry, report of the European Aluminium Association, April 2000.
[23] Hay A, Orlandi MC, Koundakjian P. EUROFER stainless steel LCI study, EUROFER Avestapo-
larit—Arcelor Group. Available from: http://www.nrcan.gc.ca/mms/canmet-mtb/mmsl-lmsm/
enviro/lifecycle/.
[24] Sicily Regional Government. Relazione sullo stato dell’ambiente della Sicilia 2002, regional official
environmental report, 2002 [Italian language].
[25] Gomber A. Results of durability testing of antireflective glazing, Dissemination workshop of IEA
task 27, Freiburg (Germany), 6 October 2003.