Journal of Cleaner Production xxx (2016) 1e15

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

Carbon footprint of extra virgin olive oil: a comparative and driver

analysis of different production processes in Centre Italy
Claudio Pattara a, *, Roberta Salomone b, Angelo Cichelli a
a
Department of Economic Studies, University ‘‘G. d’Annunzio’’, viale Pindaro42, 65127 Pescara (Italy)
b
Department of Economics, University of Messina, Piazza S. Pugliatti 1, 98122 Messina (Italy)

a r t i c l e i n f o a b s t r a c t

Article history: In this research, five case studies located in Abruzzo (Italy) were analysed using the Carbon Footprint
Received 2 April 2015 method aiming primarily to quantify the greenhouse gas emissions related to the cultivation of olives
Received in revised form and the production of olive oil (respectively farm phase and mill phase) and to identify the drivers
29 March 2016
behind these emissions (excluding olive tree planting and distribution phase). In respect to the existing
Accepted 29 March 2016
Available online xxx
literature, the research presented here developed more detailed collection of data on farms and selected
the case studies in order to represent the current production situation at the international level (Spain
excluded) as regards technology and size. Furthermore, an uncertainty and sensitivity analysis was
Keywords:
Extra virgin olive oil
performed in order to assess the robustness of results. Five litres of extra virgin olive oil, with primary
Carbon footprint and secondary packaging, were chosen as a functional unit. Results showed that agriculture accounts for
Life cycle assessment emissions of CO2eq ranging from 3.34 to 7.74 kg (mainly due to fertilizer and pesticide treatments),
Greenhouse gases followed by the packaging process in the industrial phase for which CO2eq emissions range from 1.13 kg
Climate change to 3.20 kg (for which glass bottles represent the largest load). The study revealed that a realistic
reduction of greenhouse gas emissions should be based on an efficient use of pesticides and fertilizers.
Drivers are mainly located in the farm factory phase, which is the one with the highest impact, but also
the phase that proves most difficult as regards retrieving detailed data.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction changes concerning cultivation practices, technologies of extraction

The increase of greenhouse gas (GHG) concentrations is
considered the main cause of global climate change (IPCC, 2014). It
is increasingly accepted that of the various manufacturing activities
responsible for GHG emissions, the whole agri-food supply chain is
one of the major contributors to climate change; food production
systems thus contribute a large share of anthropogenic emissions
(Pelletier et al., 2013), estimated to be 19e29% of total GHG emis-
sions (Vermeulen et al., 2012). Between 80 and 86% of these
emissions are caused by agricultural production, while the
remainder comes from pre-production (mainly fertilizer produc-
tion) and post-production activities, such as primary and secondary
processing, packaging, transport, etc (Avraamides and Fatta, 2008).
One of the most important agri-food sectors in the EU is the olive
oil industry. In recent decades, this sector has been affected by major
* Corresponding author. Tel.: þ39(0)8545083222.
E-mail addresses: claudiopattara1@gmail.com (C. Pattara), roberta.salomone@
unime.it (R. Salomone), cichelli@unich.it (A. Cichelli).
and, above all, quality improvement and selection of typical charac-
teristics. In addition to those changes mainly linked to strengthening
the quality of extra virgin olive oil (EVOO) e such as nutritional,
organoleptic, hygiene and typical characteristics ethe scientific
community has also tackled and analysed environmental impacts
related to its production with particular interest. One of the reasons
for this is that the olive oil industry is characterized by great variability
deriving from different cultivation practices, production techniques
and supply chain organization, as well as influences arising from local
characteristics (Rinaldi et al., 2014; Salomone and Ioppolo, 2012).
Among the various environmental impacts, the assessment of
GHGs in a product life cycle perspective is a significant aspect
related to environmental issues that requires full, in-depth analysis.
In fact, numerous international organizations (Environdec, 2014)
are currently working on application rules (Product Category Rules
ePCR) for the various national standards for calculating GHG
emissions. For example, the European Commission has its Product
Environmental Footprint (PEF) project underway, while the Inter-
national Olive Oil Council (IOOC) is developing a protocol for the
quantification and subsequent certification of GHG emissions

http://dx.doi.org/10.1016/j.jclepro.2016.03.152
0959-6526/© 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
2 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15
generated by the production of EVOO. The IOOC and the Joint
Research Centre (JRC) are currently developing the Product Envi-
ronmental Footprint Category Rules (PEFCR) for olive oil, under the
aegis of the European Commission (EC, 2013).
Also on the scientific research level, a number of authors have
carried out Life Cycle Assessment (LCA) studies in the agri-food sector
and on the specific field of olives and olive oil production. The LCA
method was initially applied in the olive sector in the 2000s
(Salomone et al., 2015), and proved to be an effective tool on which to
base operational decisions at individual farm level (Fedele et al., 2014).
A recent publication reports the results of a critical review of the
international literature (Salomone et al., 2015) in which 72 studies
were reviewed, taking into account different aspects, such as the
tool used (LCA, Simplified LCA, Carbon Footprint, Water Footprint,
etc.) and the methodological issues. Four other papers were pub-
lished after the literature review reported in Salomone et al., 2015:
two apply the LCA method (Mohamad et al., 2014; Rajaeifar et al.,
2014) and two refer to the Carbon Footprint (CF) method (Proietti
et al., 2014; Rinaldi et al., 2014).
Therefore, the literature still has few cases where the footprint
label has been applied in the olive oil sector. Indeed, at the time of
writing this research, only five CF (Polo et al., 2010; Intini et al.,
2011; Lucchetti et al., 2012; Proietti et al., 2014; Rinaldi et al.,
2014), one Water Footprint (Salmoral et al., 2011) and one Ecolog-
ical Footprint (Scotti et al., 2009) had been published.
In respect to the existing literature, the research presented here
developed more detailed collection of data on farms by considering
a broader sample, and considered the current trend in technolog-
ical levels for central Italy, Greece and the countries of northern
Africa. Indeed, the main goal of the study was to identify and assess
the phases and inputs with greater environmental impact of each
system in order to identify the potential optimization options of
EVOO production practices and the drivers behind GHG emissions.
In this sense the production system in Abruzzo is similar to that
found in the central regions of Italy (Lazio, Tuscany, Marche, and
Umbria) as regards cultivation, milling facilities, and quality of olive
oil, as well as to that of emerging countries (Morocco, Tunisia,
Argentina, etc.). The higher level of accuracy included the single
processes of the agricultural and industrial phases, excluding the
distribution and end of life that are, as is known, the stages in which
quality and data accuracy are lower (Saner et al., 2012; Accorsi et al.,
2015; Lazarevic et al., 2010).
Further goals of the study were to increase awareness among
practitioners and local stakeholders of the use of environmental
assessment tools in order to increase attention to environmental
concerns and highlight the main methodological issues related to
the application of the CF method in the specific sector of olive oil
production.
This paper is structured as follows:
1. Introduction, summarizing the general aim of the paper, its
collocation in relation to the existing literature and its structure;
2. Materials and Methods, split between the description of Case
studies and of Carbon Footprint analysis (including the
description of the main elements of the CF analysis framework
implemented: functional unit and system boundaries, inventory
analysis and impact assessment);
3. Results, where the Carbon Footprint results are presented;
4. Discussion, divided into:comparison of results with other car-
bon footprint studies; drivers and methodological choices
affecting the impacts and uncertainty and sensitivity analysis;
5. Conclusions, summarizing the main findings of the paper (both
in terms of identification of “hot-spots” and opportunities for
reductions of GHG emissions, and methodological aspect of
applying the CF method in the olive oil production sector).
Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
2. Materials and Methods
In the context of a dearth of application of the CF method in the
olive oil sector, and of growing interest in these kinds of life cycle
perspective studies on climate change and a local/regional imple-
mentation, the present paper focuses on the results obtained by
applying the CF method to the olive oil production sector in
Abruzzo (Centre Italy), through a “cradle to gate” analysis which
includes the phases of cultivation (farm factory phase) and olive oil
extraction and bottling (mill factory phase). The method used for
carrying out the study is the CF according to ISO/TS 14067: 2013
(ISO, 2013) and the PCR “A CPC 21537 Virgin Olive Oil And Its
Fractions” (Environdec, 2014). A Monte Carlo analysis was also
performed to assess the uncertainty of data for which higher
variability was detected (in particular some specific farm processes,
such as fertilization, transport, phytosanitary treatments, and
diesel consumption for tillage processes); based on the outcome of
which, a sensitivity analysis was then carried out (Huijberts et al.,
2001; Weidema and Wesnaes, 1996).
2.1. Case studies
Five case studies are considered and analysed in depth that are
representative of the regional, national and other Mediterranean
production realities in terms of quality, quantity, cultivation, and
extraction technologies. Leading producing Countries that present
similar cultivation techniques and extraction technologies are
Greece, Tunisia, Morocco, Argentina, and Syria; on the contrary,
other leading producing Countries which do not present these main
characteristics are excluded from this relation (e.g. Spain which is
mainly characterized by super intensive cultivation and 2-phase
decanter extraction technology).
Furthermore, to represent fully the structural framework of the
olive oil supply chain in Central Italy (INEA, 2014) the five mills
covered by the study include three cooperatives (Case studies 3, 4,
and 5) and two private mills (Case studies 1 and 2). In fact, about
75% of the farms of NortheCentral Italy that produce olives have an
average size of less than 5 ha and are associated with cooperatives
for the processing of the olive oil (ISTAT, 2013).
Therefore, the choice of case studies (in term of their charac-
teristics such as size, type, etc.) proportionally represents the dis-
tribution of the main characteristics of farms and mills in Abruzzo,
Italy and other leading producing Countries, as highlighted in
Table 1.
Fig. 1 shows the localization of the case studies, while Table 2
presents in detail the characteristics of the case studies, as regards:
 extraction technology (2-phase decanter, 3-phase decanter, and
pressure) e the analysed mills cover all three types of olive oil
processing systems currently used in the majority of Italian
operating plants;
 working capacity expressed in terms of olive milled (quintals/
year) e the pressure system is a discontinuous cycle and
therefore presents working capacities lower than the centrifugal
systems (2- and 3-phase decanter);
 size and organizational structure (large and medium-sized
cooperative mills and medium and small-sized private mills);
 cultivation method (conventional or organic) e three organic
productions and two conventional ones were considered. This
choice was made to evaluate whether organic production,
though representing a best practice in terms of safety of the
inputs used, implies more energy consumption (fuel, electricity,
etc.) due to a lower yield of product per hectare;
 farming system (traditional non-irrigated) e is the same in all
the investigated realities as this is common throughout the
 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15 3

Table 1
Characteristics of the case studies in relation to the olive oil sector in Abruzzo, Italy and other leading producing Countries.

Characteristics Case studiesa Abruzzo Italy Other countriesb References

Average size of olive orchards ~75% under 1.5 ha ~80% under 1.5 ha ~90% under 1.2 ha ~70% under 3 ha IOOC, 2012
ISTAT, 2013
Planting patternc 100% traditional 95% traditional 75% traditional 95% traditional IOOC, 2012
ISTAT, 2013
Size of Mills 100% under 20q/h 70% under 20q/h 70% under 35q/h 50% under 35q/h Cimato, 2008
Pattara et al., 2010
INEA, 2014
IOOC, 2012
Type of mill (Decanter/Pressure) 80/20 60/40 65/35 60/40 Di Giovacchino and Preziuso, 2006
Cimato, 2008
IOOC, 2012
Plain/Hill/Mountain 85% Hill 67% Hill 63.9% Hill 70% Plain IOOC, 2012
12% Plain 33% Plain 35% Plain 30% Hill UNAPROL, 2010
3% Mountain 1.1% Mountain UNAPROL, 2013
UNAPROL, 2014
Package 80% glass 44.7% glass 80% glass 60% glass IOOC, 2012
20% steel can 10.6% Steel can 11% steel can 11% iron steel UNAPROL, 2010
31.7% PET 7% PET 7% PET UNAPROL, 2013
3% other material 2% other material 2% other material UNAPROL, 2014
6% unpacked
Irrigated or dry 100% dry 94% dry 75% dry 65% dry IOOC, 2012
UNAPROL, 2010
UNAPROL, 2013
UNAPROL, 2014
Farms specialized in olive production/ 56% 63% 69% 75% IOOC, 2012
UNAPROL, 2010
UNAPROL, 2013
UNAPROL, 2014
a
Only primary data.
b
Greece, Tunisia, Morocco, Argentina, Syria.
c
Traditional ± 200 trees/ha; Intensive ± 400/500 trees/ha; super-intensive ± 1500 trees/ha.

Fig. 1. Localization of case studies.

Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
4 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15

region and, in general, to almost all farms of Central and
Northern Italy (dry method);
 localization of the case studies (Fig. 1) e mills were also selected
to represent different geographic characteristics, such as the
proximity to coastal areas, and therefore to assess whether the
micro-climate might have a significant impact on the number of
pesticide treatments and on GHG emissions;
 type of primary and secondary packaging e four products
packaged in glass (1 L bottles; cardboard for six) and one can (5 L
cans; four can carton) were included.
In order to ensure the representativeness of the case studies,
data concerning farm surface, surface area of olive groves, cultiva-
tion practices, size and type of mills, and packaging materials were
examined. A survey to gather all the information about the different
aspects was then carried out. For private mills, data was collected
from all the farms, on the contrary the collection of data on farms
relating to the cooperative mills, considering the high number of
farms involved, was carried out through a stratified sampling. Sets
of “farms” reflecting the dimension of the olive groves (values from
0 to 1 ha, 1e2 ha, 2e3 ha, 3e4 ha, 4e5 ha, from 5 to 10 ha and
>10 ha) were created. A proportional number of farms from each
cooperative mill and from each layer (37 from Case 3, 33 from Case
4, and 27 from Case 5) were randomly extracted. In this way, for all
the three cooperative mills data representing 58%e72% of the total
area of olive groves was collected.
2.2. Functional unit and system boundaries
The functional unit (FU) is 5 L of EVOO in its primary and sec-
ondary packaging. This choice was motivated by the fact that for the
examined mills the typical selling unit is a 6-bottlepackage or a 5-L
steel can; in order to allow the comparative study, it was necessary
to have a common FU for the five case studies. Table 2 also shows
the FU with the corresponding reference flows used in the indi-
vidual cases included in the study.
System boundaries, through a “cradle to gate” analysis, include
the phases of cultivation (farm factory phase) and olive oil extrac-
tion and bottling (mill factory phase) as described in Fig. 2: pro-
cesses within dotted line are excluded from the system boundaries.
2.3. Inventory analysis
Table 3 reports inventory data related to the farm factory phase
and a brief description of the processes for which data was
gathered.
The tillage processes (rotary tillage, flail mower, chisel plough, disk
harrowing, harrowing) are usually performed to weed out the sea-
sonal weed infestations and allow inputs of additional nutritional
elements (water and fertilizer) which can be easily absorbed by the
olive trees, thus protecting soil from erosion by meteoric leaching.
Two practices have been commonly used in recent years to offset
high fuel costs and avoid erosion processes (particularly in sloping
ground): cover cropping (flail mower, two or three times a year) or
green manuring. Other tillage processes are harrowing, digging,
ripper, chopping, milling (fuel consumption varies from 4 to 5 L/h of
diesel for ripper or harrowing, to 7e9 L/h for chopping or digging).
The phytosanitary and fertilization treatments include all actions
aimed at fighting major plant pests. The treatments are generally
carried out by using a sprayer connected to the tractor (fuel, water
and phytosanitary products consumption was evaluated). In con-
ventional cultivations (up to 400 plants/ha), the treatment process
can be performed on average 1e1.5 h/ha. As regards fertilization
treatments, typically this process takes place by supplying one or
more nutritional elements directly on the ground or via foliar
fertilization.
Pruning and harvesting may be carried out with jigsaws and
beaters (pneumatic or electric), while the removal of branches is
typically performed manually. Fuel consumption for the operation
of the air compressors (for pneumatic tools) and the energy
required to charge the batteries (for electric tools) was calculated.
Transport of materials and products for all the inputs of each
process, from the place of purchase/production is also included in
the inventory analysis. The fuel consumption for the transport of
olives was considered without including the wear and tear of the
vehicle; this is because the single operation of transport to the oil
mill (typically at a distances of 2e5 km) is negligible compared to
normal use of the vehicle.
Some processes have been omitted from the inventory of the
farm factory phase, for the following reasons:
 servicing of agricultural machinery e even if the replacement of
mechanical or engine oil and other spare parts could have

Table 2
Detailed characteristics of the five case studies.

Characteristics Case 1 Case 2 Case 3 Case 4 Case 5

Extraction Technology Pressure Decanter (2 phases) Decanter (3 phases) Decanter (3 phases) Decanter (3 phases)
Olive milled (quintals/year) 4500 5700 12,000 9000 7000
Olive yield kg per hectare (kg/ha) 6125 7180 7092 6871 5956
Oil yield litre per hectare (L/ha) 1120.91 1345.27 1282.43 1264.92 1025.10
Mill structure and total Private (1 mill; 1 farm) Private (1 mill; 1 farm) Cooperative Cooperative Cooperative (1 mill; 137 farms)
number of farms (1 mill; 329 farms) (1 mill; 258 farms)
Farms sampled 1 1 37 33 27
Cultivation surface (ha) 120 150 270 240 190
Cultivation methods Conventional Conventional Organic Organic Organic
Farming system Traditional non-irrigatedTraditional Traditional Traditional Traditional non-irrigated
non-irrigated non-irrigated non-irrigated
Localization Ortona (CH) Moscufo (PE) Pianella (PE) Pianella (PE) Casoli (CH)
Primary Packaging Glass bottle Steel can Glass bottle Glass bottle Glass bottle
Secondary Packaging Cardboard Cardboard Cardboard Cardboard Cardboard
Reference flow 5 L of conventional EVOO 5 L of EVOO in 5 L of organic EVOO PDO 5 L of organic EVOO 5 L of organic EVOO PDO in
PDO in 5 glass bottles 1 tin and ¼ of in 5 glass bottles and 1 PDO in 5 glass bottles 5 glass bottles and 1
and 1 cardboard box cardboard box cardboard box and 1 cardboard box cardboard box
Functional unit 5 L of EVOO with primary
and secondary package

PDO ¼ protected designation of origin.

EVOO ¼ Extra Virgin Olive Oil.

Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15 5

Fig. 2. System boundaries for the “cradle to gate” Carbon Footprint assessment of EVOO production.

Table 3
Inventory data for the farm factory phase per 1000 kg of olives.

Process Input Unit Case 1 Case 2 Case 3 Case 4 Case 5

Tillage Diesel L 11.791 7.155 9.313 10.281 13.566

Phytosanitary/fertilization treatments Fertilizer Kg NPK 15:5:5 NPK 10:12:7 Cow manure Cow manure NPK 17:5:7
Quantity Kg 13.007 9.058 149.116 139.163 14.461
Pesticide type Bordeaux Dimethoate Bordeaux Bordeaux Bordeaux
mixture mixture mixture mixture
Dimethoate. Pyrethrum Pyrethrum Dimethoate.
Quantity Kg 8.16 5.01 5.08 4.95 9.74
Water L 1017.959 328.134 307.107 434.726 1320.853
Diesel L 7.997 4.292 5.321 6.168 8.915
Pruning and harvesting Diesel L 12.802 6.542 8.780 9.106 14.342
Electricity kWh 0 2.421 0 0 0
Transport of materials and products Inputs transport to farm tkm 2.564 6.722 5.936 5.238 3.786
Olive transport to mill tkm 1.451 2.020 7.345 8.128 4.916

significant impacts, several data limitations are connected to
this process because farmers do not possess accurate related
data (the servicing is usually carried out by external authorized
workshops) and the use of the same machinery for all crops and
land management (therefore allocation problems may occur,
causing a high degree of uncertainty and inaccuracy);
 olive planting phase e in the traditional olive farming system
(density between 200 and 400 trees per hectare) the average life
span of trees is typically more than 100 years, often reaching 150
or even 180 years, and all the olive groves analysed had mature
and ready-for-production plants (over 30e40 years), thus the
production phase of the seedlings and the related planting

Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
6 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15
phase, as well as the time (4e7 years) required for the first olive
harvest, represent a small environmental burden (Salomone and
Ioppolo, 2012; Salomone et al., 2015);
 irrigation e there is no irrigation of olive groves within the
companies analysed (dry method);
 energy consumption related to farm infrastructure e energy
consumption in the farm structures is negligible in a life cycle
perspective of the olive oil product, considering moreover that
the farms are not specialized in olive cultivation alone and farm
structures are often used for other purposes (Salomone et al.,
2015);
 land use change e it was not possible to trace land use prior to
olive cultivation; in addition, considering that in traditional
olive farming systems, land use had not changed for 100e180
years at least, land use change is not a relevant impact category;
 carbon sink e for these items and for those relating to emissions
from biomass combustion (the practice of burning the pruning
on the edge of the fields) the carbon, and related CO2, was
considered as belonging to the short carbon cycle and, as a
result, not to be considered in the general CO2 budget. In the
authors' opinion, this approach is more consistent with the
original purpose of the CF method (BSI, 2011), i.e. to analyse,
quantify and reduce CO2 emissions related to fossil sources. This
decision is further supported by the review of Arzoumanidis
et al., 2014, which highlighted the lack of uniformity in the
application of the accounting method of biogenic CO2, and
found that the standards and calculation methods that assess
biological CO2, include it in a inventory that differs from that of
GHG produced by other biogenic sources.
As regards the mill factory phase (Table 4), a framework for data
collection was created based on inspection of the mills to verify:
types of plants; layout; installed power; presence of mechanisms to
measure the energy used; timing and amount of maintenance
carried out; energy bills and loads; olives processed daily and
during the whole milling season and milling phase; and con-
sumption of natural gas and fuel for forklifts. Furthermore, during
the milling phase, a multimetre (model DMGM380001 Lovato) was
installed on the electrical control of the extraction plant. This
enabled the collection of electrical measurements for each batch of
olives processed (250e500 kg) for all the standard stages of pro-
cessing: washing and defoliation; milling; kneading; extraction (in
centrifugal decanter or traditional system); centrifugal separation
of oil-water emulsion; storage tank; packaging. The average data
related to energy consumption per kg of olives treated and per kg of
oil extracted was considered, based on these figures. The energy
consumption of the system was measured several times on four
separate days of the year and on different batches of olives. Elec-
trical data concerning plant maintenance, energy consumption of
infrastructure and consumption of natural gas, Liquefied Petroleum
Gas(LPG) and diesel used in the company was added. Comparing
the average data (taken from electricity bills) with those collected
in situ, it was possible to differentiate the portion attributable to
activities not related to the production phase (lighting, adminis-
tration, heating, etc.), in order to exclude it. The transport data was
collected, along with all field surveys, through questionnaires filled
out by farmers and olive oil millers. Based on these, the average
data were obtained and used for the calculations.
Some processes of the mill factory phase have been omitted
from this study, for the following reasons:
 maintenance of mill infrastructure e was excluded in accor-
dance with suggestions in the literature (Ecoil, 2007; BSI, 2011;
Salomone et al., 2015). Indeed the frequency of maintenance
(every three or five years) and the type of maintenance
Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
(plastering or painting of walls, maintenance of pipes) is negli-
gible in a life cycle perspective of EVOO;
 all stages following the packaging ethe authors chose not to
consider the steps of distribution and end of life as the main aim
of the research is to investigate in depth the comparison of the
agricultural and industrial phases, which are the ones that
present the greater differences. On the contrary, distribution
and consumption were assumed to present similar impacts and
were therefore excluded from the comparative assessment;
 mill by-product usage e in modern technology mills, effluents
from the extraction of olive oil are no longer considered waste
but by-products: olive mill wastewater, from pressure and 3-
phase decanter, can be used as fertigation through procedures
and limits prescribed by law; pomace, from pressure and from3-
phase centrifugal decanter, can be used for the extraction of the
residual oil by solvent or for energy purposes; wet pomace, from
2-phase centrifugal decanter, can be treated in order to extract
the pits used for energy purposes and then undergo a process of
composting, or can be spread on land as fertilizer. However,
their economic value is very low (olive mill wastewater and wet
pomace V 0/m3, pomace from pressure 3.5e4.5 V/100 kg,
pomace from three-phase decanter 1.5e2.5 V/100 kg) and in the
present study, similarly to some other researches (e.g. Raggi
et al., 2000; Salomone and Ioppolo, 2012) all the impacts were
allocated on EVOO (therefore using an economic allocation
criterion) and by-products were omitted from the inventory.
While it may be useful, or at least desirable, to allocate the
impact of the earlier stages among the various by-products of
the extraction process, it should be highlighted that allocation
problems could result and the issue is still under study by the
Project on the PEFCR for olive oil (EC, 2013).
The data collection process described here aimed to seek the
greatest possible quantity of primary data in order to respond
adequately to the principles required by ISO 14067 (time coverage,
geographical coverage, technology coverage, representativeness,
relevance, completeness, consistency, accuracy and transparency,
and reproducibility). Double counting was avoided during the data
collection phase. Coefficients present in the literature and com-
mercial databases were used (Table 5) for inputs such as electrical
energy, fossil fuels and fertilizer emission.
2.4. Impact assessment
In compliance with the requirements of ISO 14067 (ISO, 2013),
the only impact category considered in this study is Global
Warming. All GHG emissions were calculated in terms of CO2
equivalence over a 100-year time horizon according to the IPCC
method 2007 GWP 100 V1.02 (IPCC, 2007).
GHG emissions per FU were first calculated for each input and
then for each phase (farm factory phase and mill factory phase).
Subsequently, the whole life cycle of each of the five case studies
were compared. Impact assessment results are presented in x 3.
3. Results
Fig. 3 shows the distribution of GHG emissions associated with
the different agricultural practices in the five case studies analysed,
in terms of kg of CO2eq/FU and in percentage terms respectively.
Plant protection treatments are the most significant item in two
farms: Case 1 (conventional cultivation) and Case 5 (organic culti-
vation). In these cases, it is not the cultivation method that acts as a
determinant but the number of treatments carried out (and thus
diesel fuel used) which counts for about 50% of the total value.
Indeed, in Case 5 only a few products are used (bordeaux mixture,
 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15 7

Table 4
Inventory data for the mill factory phase per 1000 kg of olives.

Process Input Unit Case 1 Case 2 Case 3 Case 4 Case 5

Olive oil extraction Water (washing) L 15 17 119 127 100

Electricity kWh 65.892 37.785 40.083 51.148 49.614
Synthetic Rubber Kg 1.212 1.861 2.533 0.587 1.921
Stainless steel Kg 3.667 2.732 0.943 0.482 1.543
Packaging Green glass Kg 108 0 108 108 108
Aluminium cap Kg 0.770 0 0.770 0.770 0.770
non-drip spout Kg 0.157 0 0.157 0.157 0.157
Front Cardboard label Kg 0.218 0 0.218 0.218 0.218
Rear Cardboard label Kg 0.162 0 0.126 0.144 0.144
Shrink cap Kg 1.264 0 1.26 1.26 1.26
Steel can Kg 0 18.722 0 0 0
Cardboard Can label Kg 0 0.0396 0 0 0
Can plastic cap LDPE Kg 0 0.092 0 0 0
Corrugated cardboard box Kg 11.376 24.127 10.951 11.462 11.738
Adhesive tape (LDPE þ Glue) Kg 0.045 0.0648 0.045 0.045 0.045
LLDPE film Kg 0.324 0.185 0.356 0.332 0.345
Pallet Kg 5.987 3.318 6.134 6.211 6.067
Transport of materials Input transport to mill tkm 1.456 2.027 1.323 1.894 2.234

Table 5
Data sources.

Life cycle phase Sub process Process Data sources

Farm factory phase
Mill factory phase
Fertilization Fertilizer production transport, field emission Ecoinvent, 2010;  Consultants, 2014
PRe
Diesel production, transport and combustion Ecoinvent, 2010;  Consultants, 2014
PRe
Fertilizer spreading Measured data
Amount of fertilizer used Measured data
Pest treatment Pesticides production transport Ecoinvent, 2010;  Consultants, 2014
PRe
Pesticides spreading Measured data
Amount of pesticides used Measured data
Diesel production, transport and combustion Ecoinvent, 2010;  Consultants, 2014
PRe
Amount of diesel used Measured data
Harvesting Electricity production, transport and loss Ecoinvent, 2010;  Consultants, 2014
PRe
Amount of diesel used Measured data
Diesel production, transport and combustion Ecoinvent, 2010;  Consultants, 2014
PRe
Tillage (rotary tillage, Diesel production, transport and combustion Ecoinvent, 2010;  Consultants, 2014
PRe
flail mower, chisel Amount of diesel used Measured data
plough, disk harrowing, harrowing)
Pruning Electricity production, transport and loss Ecoinvent, 2010;  Consultants, 2014
PRe
Amount of electricity used Measured data
Diesel production, transport and combustion Ecoinvent, 2010;  Consultants, 2014
PRe
Amount of diesel used Measured data
Transport to olive mill Diesel production, transport and combustion Ecoinvent, 2010;  Consultants, 2014
PRe
Distance Measured data
Amount of diesel used Measured data
Extraction of olive oil Amount of electricity used Measured data
Electricity production, transport and loss Ecoinvent, 2010;  Consultants, 2014
PRe
Amount of water used Measured data
Amount of olive milled Measured data
Amount of olive oil extracted Measured data
Packaging Amount of electricity used Measured data
Packaging materials, production Ecoinvent, 2010;  Consultants, 2014
PRe
Packaging materials, transport Measured data
Amount of packaging materials used Measured data

copper sulphate, etc.), but in larger quantities (10e15 kg/ha) to
those used for plant protection products (1.00e1.5 L/ha) in con-
ventional agriculture. This factor, as well as the higher number of
treatments carried out in organic agriculture (20e30% compared to
the conventional system), causes an increase in the impacts asso-
ciated with the plant protection phase.
On the contrary, fertilization treatments are significant con-
tributors to GHG emissions in all five case studies, followed by
mowing and disk harrowing.
Another factor to consider regards the energy efficiency of the
agricultural machinery used in the farm factory phase, where
tractors and related tools are on average 15/20 years old. This has a
highly significant impact on emissions given the reduced energy
efficiency of such machinery. The operations most affected by low
energy efficiency are, of course, the ones that are most energy
intensive (phytosanitary treatments, chopping, digging), which are
associated with higher emissions per hectare and therefore per
kilogramme of olives and oil produced. New tractors, which are on
average more powerful and show lower consumption for the same
worked surface compared to obsolete models, allow a lower
amount of emissions per kg of olives produced.
The harvesting and transport of the olives to the mill factory
account for between 5% and about 16% of the total emissions of the
farm factory phase, depending mainly on the level of automation
(and thus energy consumption) of collection (viable through
pneumatic or electric equipment) and the quantity of olives

Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
8 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15
Fig. 3. GHG emissions in the farm factory phase (kg of CO2eq/FU.).
transported at each trip to the mill. Case 2 showed lower values for
this aspect, associated with the use of electrical machinery for the
harvest and management of the aerial part of plants and particularly
efficient transport of olives to the mill (maximizing the ratio of olives
transported per trip). On the other hand, the worst cases were the
cooperative mills (Cases 3, 4 and 5) with farms that resisted auto-
mation and technological innovation thus affecting the impacts
related to the consumption of diesel fuel (tractors and machinery
used to produce compressed air for pneumatic saws and harvesters).
As far as green manure is concerned, faba beans were only sowed
in very few cases among the companies considered and therefore do
not appear in Fig. 3. In any case, it should be highlighted that this
agricultural practice plays an important role in term of atmospheric
nitrogen fixation in ground. A continual use of this farming practice
would enable avoidance of nitrogen fertilizers (by the contribution
of about 50e150 kg N/ha, 10-35 P2O5, K2O 30-120) and permit
optimum management of the weed species (which do not grow
when faba beans are present). Furthermore, energy consumption
per hectare for this practice is attributable to operations such as
seeding, mulching and digging, for a total quantity of diesel con-
sumption ranging from 30 to 45 L/ha and related emissions ranging
from 130 to 180 kg CO2eq/ha. These values are remarkably lower
than those of conventional management, which envisages fertil-
ization with synthetic products and mechanical weed management.
Conventional management involves diesel consumption of about
25e35 L/ha for weed control and 3e7 L/ha for spreading fertilizer, to
which the emissions related to the production and use of fertilizer
(which vary according to the composition) must then be added.
Values range from 0.9 to 2.8 kg CO2/kg product for the production of
a generic NPK 15:15:15, and about 2.5 kg N2O per 100 kg of N used on
the ground (Ecoinvent, 2007; IPCC, 2006).
Fig. 4 shows the distribution of GHG emissions associated with
the different mill factory processes in the five case studies analysed,
respectively in kg of CO2eq/FU and in percentage terms.
Only minor differences between the various cases can be
highlighted for the mill factory phase; the highest value is linked
Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
to the traditional extraction system (1.095 kg CO2eq). This is
because this extraction system works with a limited quantity of
olives at a time (200e300 kg) and, even if the power of the ma-
chinery installed is lower (about 30 kW for the pressure mill
system versus 60/100 kW for centrifugal decanter), the energy
consumption per kg of extracted oil is greater for the pressure mill
system.
For the packaging process, the emission values are similar for
glass packaging (common to four of the analysed cases), as the
bottles are analogous in weight and material; the only difference
occurs for Case 2 in which 5-L steel containers are used (0.104 kg/L
of olive oil against 0.605 kg of glass for 1 L of olive oil). This allows
a substantial saving in terms of CO2 emissions (about 0.417 kg
CO2eq for the 5 L steel container and 0.298 kg CO2 for the 1 L glass
bottle).
Table 6 summarizes the total GHG emissions of the five case
studies.
4. Discussion
Results show that the CF varies between 4.48 kg and 10.1 kg
CO2eq for FU. The wide range of these values is related to the fact
that the five case studies refer to different agricultural practices and
different extraction technologies. The highest value (10.1 kg CO2eq)
is that of Case 5, mainly due to the contribution of the farm factory
phase and, within this, to the phytosanitary treatments.
The lowest value (4.4 kg CO2eq) is that recorded by Case 2; this
result depends both on the agronomic management (use of elec-
trical machineries for the harvest and management of aerial part of
plants) and to the mill factory phase (two-phase decanter and use
of the 5-L steel tin as primary packaging).
The other investigated cases (1, 3, and 4) are fairly similar for the
farm factory phase, whereas Case 1 differs appreciably for the mill
factory phase (using the pressure system).
 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15 9

4.1. Comparison of results with other carbon footprint studies Table 6

Total GHG emissions of the five case studies (kg of CO2eq/FU.).

As mentioned above, at the time of writing this research, five
previous CF in the olive oil sector had been published.
Polo et al. (2010) analyse two types of product olive oil (con-
ventional 5-L packaging in PET bottles and “ecological” 1-L glass
bottle). Although data on the production phases was not reported, a
comparison can be attempted: the CF value for the conventional
olive oil (5.495 kg CO2eq) and for the ecological olive oil (1.140 kg
CO2eq) is in line with the results of the present study.
Intini et al. (2011) carried out an assessment of the benefits
arising from the possible use of de-oiled pomace for energy re-
covery. This analysis is thorough and clearly shows what GHG
emissions could be avoided if all the de-oiled pomace was not
destined for residential users (as happens today) but rather to
electricity and heat production plants; unfortunately our work does
not cover the phases of olive oil use and end of life of the by-
products, so a comparison cannot be made.
Lucchetti et al. (2012) analyse only the packaging phase and
quantify emissions as 22 gCO2eq/kg of olive oil worked. It can be
seen that in our study the emissions deriving from the industrial
phase range from 1.14 (Case 2) to 3.20 kg CO2eq for UF (for other
cases studies). The value of Lucchetti, as specified in the work,
derives from an analysis of average data obtained by calculating the
ratio between the total amount of packaged oil and the total con-
sumption of electricity and natural gas. This thus neglects all the
packaging input and the extraction process which, in the present
case study, cannot be considered separately from the entire pack-
aging phase.
Proietti et al. (2014) analyse the first ten years of activity of an
olive grove considering only the agricultural phase; the first two
years are the most impactful (16.578 tCO2e/hectare for the first and
2.892 tCO2eq/ha for the second) and the impact then stabilizes at a
value of about 1.5 tCO2eq/ha. Although they cannot refer to the
production of olives, the results obtained in Proietti et al. (2014) are
consistent with those of the present study (1391.04 kg CO2/ha Case
1; 1473.01 kg CO2/ha Case 3; 1267.44 CO2/ha Case 4; 1586.85 CO2/
ha Case 5), with the only exception of Case 2 (899.98 Kg CO2/a) that,
Farm factory phase Mill factory phase kg CO2eq/FU
CASE 1 65.97% 34.03% 9.41
CASE 2 74.66% 25.34% 4.48
CASE 3 67.29% 32.71% 7.37
CASE 4 65.66% 34.34% 7.63
CASE 5 76.63% 23.37% 10.10
as already mentioned earlier, can be explained by a particularly
efficient management of the olive orchard using the latest gener-
ation machinery.
Rinaldi et al. (2014) report a value of 17.53 kg CO2 eq/L for the
entire life cycle of the product EVOO. This value seems particularly
high and, even if we consider only the steps in common with our
study (7.88 kg CO2 eq/L), the value of GHG emitted appears much
higher than those obtained in our cases (from 0.88 kg CO2 eq/L for
Case 2e2.2 kg CO2 eq/L for Case 5). This extreme difference is to be
found mainly in the yield per hectare (particularly low in the case
studied by Rinaldi et al., :1548 kg/ha), compared to the case studies
analysed in this paper (about 6500 kg/ha).
A summary of the comparison of the results of the five inves-
tigated case studies with the previous CF analyses, is presented in
Table 7. In addition to these few CF studies, it is also interesting to
cite a recent LCA applied in the olive oil sector (Rajaeifar et al., 2014)
which reports a low yield of the crop (slightly over 3000 kg/ha of
olives and 620 kg/ha of olive oil) and a considerable amount of
fertilizers used (about 1300 kg/ha). This significantly affects the
value of the final CF (although the packaging process is not
included) that reaches approximately 2 kgCO2eq/L of olive oil; this
value is similar to the olive oils considered in the present study
(which however include primary and secondary packaging).
The review above highlights the fact that no CF studies have
previously been done to comprehensively analyse the most im-
pactful sub-processes of life cycle stages of olive oil; it is, however,
clear that the data from this study are consistent with those of
previous studies.

Fig. 4. GHG emissions in the mill factory phase (kg of CO2eq/FU.).

Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
10 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15
Table 7
Comparison of results with other Carbon Footprint studies.
Agricultural stage kgCO2eq
CASE 1 6.20
CASE 2 3.34
CASE 3 4.96
CASE 4 5.01
CASE 5 7.74
Polo et al., 2010 n.a.
Lucchetti et al., 2012 n.a.
Proietti et al., 2014 16.57d
2.89d
1.5d
Rinaldi et al., 2014 3.24
Rajaeifar et al., 2014 2.02
n.a. ¼ not available.
a
5 L PET bottle.
b
1 L glass bottle.
c
gCO2eq/kg of olive oil worked.
d
tCO2eq/ha (I -II -III year).
e
1 Litre.
4.2. Drivers and methodological choices affecting the impacts
As specified in the goal and scope, the CF analysis performed
also aims to identify drivers affecting the impacts. The hypothetical
drivers identified were: technology, localization, cultivation
method, protection, farm/mill structure and packaging material.
Results highlighted the following aspects:
 technology e extraction technology is an important driver in the
life cycle of EVOO. Indeed the value of about 1.1 kg CO2eq/FU
related to the traditional extraction system is almost double
compared to other extraction systems (centrifugal extraction
technology). In this sense it is necessary to clarify that in recent
years the pressure systems are ever lower and smaller in size.
This process of replacement has been widespread in Spain since
the ‘70sas well as spreading to southern Italy and northern Af-
rica in the last 20 years, though to a lesser extent. The more
limited diffusion of this technology can be explained by an un-
willingness to innovate in the olive oil extraction sector of Italy,
Greece and North Africa where there is generally a low level of
mechanization, a low agricultural income related to olive
growing and a limited average surface area of farms;
 localization of production areas e case studies located in the
province of Pescara (Cases 2, 3, and 4) have lower emissions in
the farm factory phase due to a lower number of disease
treatments. This is because in the other cases (1 and 5) meteo-
rological conditions favoured the development of pests; indeed,
from the interviews carried out, it emerged that climatic factors
(rainfall, temperature, etc.) greatly affected the operating pa-
rameters by creating suitable conditions for the development of
plant diseases. This clearly highlights the fact that climatic
conditions are one of the main drivers of GHG emissions
because they strongly affect the number of phytosanitary
treatments carried out, as well as the consumption of diesel and
related input. Unfortunately, variables related to climatic con-
ditions remain an element of natural stress that cannot be
effectively managed in the short term for GHG emissions
mitigation;
 cultivation method (organic vs conventional) e organic farming
does not present any advantage in reducing GHG emissions. This
achievement differs basically from the results of other studies
(Romani et al., 2004; Cecchini et al., 2005; Olivieri et al., 2005;
Neri et al., 2012) in which the production under organic
Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
Industrial phase kgCO2eq Total kgCO2eq
3.20 9.40
1.14 4.48
2.41 7.37
2.62 7.63
2.36 10.10
n.a. 5.495a
1.140b
22c n.a.
n.a. n.a.
4.64 7.88e
0.10 2.12e
farming is associated with lower GHG emissions. It is therefore
necessary to further study this issue in future studies, paying
particular attention to other factors such as the localization and
the technological level of the machinery used;
 productivity e olive yield (kg/ha) and oil yield (L/ha) e the method
of production and other agricultural and technological variables
significantly affect the yield of olives per hectare and olive oil
yield. These two variables, in turn, weigh significantly on the
total balance of GHG. Yields of 5000 kg/ha of olives and
extraction yields of 12% imply that the total GHG ought to be
allocated to 600 kg of olive oil. Years when olive production
reaches 9000 kg/ha and yields of extraction are 16% imply that
the total GHG ought to be allocated to 1360 kg of olive oil. From
this it can be noted that the production yield significantly in-
fluences the CF of the end product, even with variations in the
order of 100%;
 organization of the farm/mill structure e the results reveal that
Case 2 (private) is the one that has the lowest GHG balance (for
both phases), while the other private mill (Case 1) has higher
values, mainly in the phases of disease treatment, chopping
weeds and harvest time. On the contrary, the type of organiza-
tion (private or cooperative chain) represents an important
variable for the proper management of the mill factory phase. In
fact, a cooperative chain that often relies on a large number of
farm members and very large volumes of olives, is able to find
funding to keep up to date with technological innovations in the
sector, unlike some smaller private mills that still use pressure
systems that are technologically obsolete because these farms
do not have the funds to upgrade plants. From the point of view
of GHG emissions, the lack of technological upgrading and less
than full volume processing usually means increases in emis-
sions per volume of olive oil produced (the difference between
pressure systems and centrifugal decanter is a notable amount,
while it is smaller for a simple technology upgrade). It should be
stressed, however, that the organization of the supply chain in
the private mill cases provides a vertical integration enabling
management of the farm and mill factory phases and this allows
a more professional management of cropland, which could
make it possible to minimizeGHG emissions, especially in the
farm factory phase;
 packaging e packaging (primary and secondary) has a great
influence on the total amount of GHG within the life cycle of the
olive oil; it can represent up to 3.08 kg CO2eq. In this sense, the
 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15
type of material (glass, tin) is the key driver in terms of related
emissions. For a product such as EVOO, which has low produc-
tion yields per hectare and low yields of extraction of oil from
olives, being able to have primary packaging with a reduced
emission of GHG could provide an important strategic advan-
tage. However, packaging is a key factor in terms of marketing
and sales, and for this reason is often considered as the last part
of the product to be of interest for possible reduction options;
 carbon sink and short term carbon balance e there have been
numerous studies in the olive oil sector (Arzoumanidis et al.,
2014) aimed to quantify a value for carbon storage in the soil,
but so far they have not been exhaustive and have been unable
to consider all possible variables affecting the value of the
accumulation of carbon by olive groves. However, the impor-
tance of CO2 management and its accounting for assessing the
carbon footprint of a product or process is clear. In fact, being
able to account the stocks of carbon in the soil or the classifi-
cation of some stocks of CO2 as belonging to the short cycle, may
make it possible to discount certain stocks (precisely because of
a short cycle) or to have a positive balance (carbon sink) to
calculate the carbon footprint. However, only a direct and
continuous measurement (through coring and subsequent
chemical analysis) would provide significant results. Further-
more, there are also aspects that are not clear regarding the
possible negative changes in soil carbon content, in part
resulting from a change in land use or from incorrect manage-
ment, thus uncertainties still remain about the most suitable
ways to record and account them.
4.3. Uncertainty and sensitivity analysis
The robustness of the results was assessed through an uncer-
tainty analysis (Weidema and Wesnaes, 1996; Huijiberts et al.,
2001); it focused on the agricultural phase as this is the one with
greatest impact and with higher variability of data among the case
studies considered. The uncertainties and variations in the data of
the main farm processes (for which higher variability was detected
during data gathering) have been analysed by a Monte Carlo
Simulation procedure using SimaPro 8.0.1software.The processes
with higher variations were: fertilization, transport (of inputs to
farm and of olives to mill), phytosanitary treatments, and diesel
consumption for tillage processes. In reference to the variables on
distance (transport inputs to the company and transport of olives to
the mill)a uniform distribution was chosen because, from the sur-
veys done, it was seen that within the range of transport distances,
the probability of travelling a short distance was equal to that of
travelling further. A normal distribution was chosen for plant pro-
tection treatments, diesel consumption for tillage and fertilization,
centred on the measurement average because it was considered
impossible for this type of operation to have a random distribution
(uniform) as it must tend to an average value established by the
correct agronomic practices. The simulation was performed with an
iteration of the independent variables for 10,000 cases and with a
confidence interval of 95%. Table 8 reports values for the average,
maximum, minimum, standard deviation and coefficient of varia-
tion derived from Monte Carlo simulations: for each case study, the
coefficient of variation is particularly high for factors relating to
transportation (of the inputs to the farm and of olives to the mill)
and, to a lesser degree, to fertilization processes, plant treatments
and diesel consumption. Although according to the results of
Monte Carlo analysis, important uncertainties are associated
mainly with transport processes, the other parameters were not
neglected (considering their significant role in farming) and a
sensitivity analysis was carried out independently for all of them. In
Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
11
addition, a new hypothesis of packaging (bag in box) was tested
with the sensitivity analysis, because this parameter can be easily
managed by practitioners and local stakeholders.
In particular, the following changes were considered for each
case study for the sensitivity analysis:
a) variation of (±) 20% on diesel consumption-the increased
efficiency has been contemplated assuming the use of trac-
tors having lower fuel consumption;
b) different packaging for all products (bag in box) e the
packaging requires the use of a containment cardboard
(0.218 kg) and a metallic polylaminate bag (0.014 kg) inclu-
sive of the olive oil dispenser. In this case, we forecast that
the packaging phase will be the same in both processes;
c) different transport distance e variations that meet the range
encountered in the data collection phase were assumed;
d) different fertilization strategies -the amount of fertilizer has
been varied in a range of ± 20%;
e) different phytosanitary treatment strategies e the hypothe-
ses involved the change in the amount of pesticides used
by ± 20%.
The sensitivity results showed that with respect to the variable
“diesel consumption”, the variation of emissions due to the one-
field activities is about 10%e25%, while the contribution to total
emissions is about 10%. From this we can see that the contribution
of emissions from fossil fuels account for about 35%e50% of the
total. In this sense, therefore, a hypothetical reduction of 30% on
fuel consumption would theoretically produce a 15% reduction in
total emissions. The change in package would result in a strong
reduction of the emission associated with the packaging phase
(from 35% to 76%), and also reductions of emissions related to
the transport of inputs (due to the lower weight of packaging
transported from the production site to the packaging site). In
relation to different fertilization strategies, changes in total emis-
sions are of modest proportions, ranging from 11% to þ10%. In
fact, although the quantity of fertilizer used varied between 20%
and þ20%, related emissions showed a variation between 77%
and þ60%. This magnification is mainly due to the multiplier effect
that the N2O emission has in the atmosphere; in fact, it has an
emission coefficient nearly three hundred times greater than CO2.
Similarly, there was an increase in emissions from the olive harvest,
caused by the increased productivity of olive groves. This same
effect partly mitigates the total emissions, since a greater GHG
emission (caused by fertilization) was accompanied by a higher
productivity of the olive grove which reduced the value of GHG per
litre of olive oil produced, with variations from þ8% to þ12%. The
same situation occurs when the quantity of fertilizers is reduced, an
action that negatively influences the productivity of olives per
hectare. In this case, there is a decrease in total GHG ranging
from 8% to 15%. As regards the plant protection treatments, the
quantity of products used in the treatment of plant diseases (not
reducing the number of treatments will not interfere with the
defence plan of plant disease) were reduced. The variation of ± 20%
produces a variation of emissions of between þ16% and 13%.There
is even a slight variation linked to the transport of the products, the
range is between ± 7%. All these changes represent only very low
values (±0.05%) on total emissions.
A variation in the distances related to transport (olives and other
inputs) has an almost imperceptible influence, causing variations of
less than 1%on total emissions. Considering that transport distance
was the parameter showing higher uncertainties from Monte Carlo
results, and that its variations cause only minor changes on total
emissions, the effect of this uncertainty on the results and the final
outcome of this study is very limited.
12 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15

Table 8
Monte Carlo analysis on main farm processes.

Fertilizing Input transport Olives transport Phytosanitary treatments Diesel consumption on tillage process

Case 1 Av 13.01 2.56 1.45 8.16 11.79

min 12.81 1.83 1.12 7.56 9.46
Max 13.46 2.73 3.01 9.87 12.87
stdDev 0.46 0.64 0.52 0.35 0.69
CV 3.53% 24.82% 36.14% 4.25% 5.88%
Case 2 Av 9.06 6.72 2.02 5.01 7.16
min 8.66 5.43 1.87 4.52 5.91
Max 9.78 7.86 3.99 6.37 8.63
stdDev 0.79 1.72 0.51 0.52 0.82
CV 8.77% 25.56% 25.13% 10.47% 11.44%
Case 3 Av 149.12 5.94 7.35 5.08 9.31
min 100.56 1.15 0.58 3.95 7.83
Max 168.35 9.36 12.34 5.91 10.87
stdDev 17.86 2.16 3.07 0.49 0.80
CV 11.97% 36.34% 41.81% 9.61% 8.58%
Case 4 Av 139.16 5.24 8.13 4.95 10.28
min 95.86 0.35 0.35 3.97 8.76
Max 178.32 12.36 11.23 5.78 12.96
stdDev 21.21 3.72 2.71 0.49 1.23
CV 15.24% 70.96% 33.34% 9.88% 11.93%
Case 5 Av 14.46 3.79 4.92 8.92 13.57
min 9.87 0.45 1.78 7.29 11.77
Max 17.59 8.45 10.73 10.24 15.39
stdDev 2.03 2.19 2.91 0.90 1.07
CV 14.03% 57.81% 59.13% 10.12% 7.92%

In summary, the above-mentioned sensitivity analysis (Fig. 5)
indicates that a variation in the consumption of diesel fuel produces
a reduction in emissions of around 10%, whereas a change in
packaging (from glass bottle or steel tin to bag in box) resulted in
reductions of overall emissions in all cases (from 5% for Case 2
to 20%for Case 3).
Finally, the analysis of drivers affecting the impacts and the
sensitivity analysis results allowed us to draw up a hypothetical
“best-in-class” case study, which embodies all the processes with
lower GHG emissions. Specifically, this scenario includes:
- for the farm factory phase e reduced use of mechanical working
of land and use of cover cropping; use of the flail mower and
disk harrowing for soil management and weed control; con-
ventional phytosanitary treatments to fight disease in the olive
grove; management of the aerial part and harvesting phase
using electrical equipment; distance between farm and mill
below 3 km;
- for the mill factory phase e use of two-phase decanter to extract
the oil, packaging in 5-L tin can (4 cans per carton).
Fig. 6 compares the “Best-in-class” scenario with the five case
studies. In the hypothetical scenario, the specialization of the
company (which produces only olive oil) and the vertical integra-
tion of the company (agricultural and olive oil extraction phase
under the leadership of a single entity) enable greater efficiency
through better organization the work. Furthermore, it may make it
possible to achieve the economies of scale and scope required to
maintain technologically up-to-date machinery. It is worth recall-
ing that the traditional cultivation system provides low yields of
olives per hectare (from 5000 kg to 7500 kg) and requires a
considerable use of manual labour in the pruning and harvesting
phases. The introduction of automation at some steps (especially
during the harvesting phase) allows a drastic reduction of the la-
bour force and simultaneously an improvement in the quality of
harvested olives (which may be collected at exactly the right period
and sent to the mill as rapidly as possible). The packaging has, as for
many other food products (wine, beer, preserves, etc.), an impor-
tant impact on emissions. The material used (glass, aluminium, tin,
steel, plastic or laminated materials) often affects the final account
of CO2. The use of 5-L tin cans makes it possible to significantly
lower impact compared to glass containers and would permit a
reduction of 2 or 3 kg of CO2 eq. compared to the other case studies.
5. Conclusions
The calculation of CF for the case studies presented here allowed
us to draw several considerations, both regarding the environ-
mental optimization of the olive oil supply chain and the applica-
tion of the CF method in the olive oil production sector.
As regards the environmental improvement of the olive oil
supply chain, the results enabled us to identify “hot-spots” and
opportunities for reductions in GHG emissions. In particular, the
phytosanitary and fertilization treatments are predominantly
responsible for GHG emissions and they often represent greater
loads than mechanical work, ranging from 1.07 to 3.2 kg CO2/FU. On
the other hand, energy efficiency of mechanical equipment repre-
sents a key item that should be carefully controlled because it en-
ables cuts in emissions and an increase in the quality of the olives
produced. Similarly, if olive tree pruning is properly managed and
mechanized, it can lead to significant benefits in both CO2 balance
(a by-product that could be allocated as part of productive inputs)
and economic gain (production of heat at low cost in the farm
buildings). Hot-spots for each case study were identified, mainly
focused on plant protection treatments, fertilizers and packaging.
Furthermore, a number of drivers affecting the impacts were
identified: climatic conditions that determine the number of phy-
tosanitary treatments; technological factors that affect the energy
efficiency of operations in the olive orchard and in the industrial
phase, and material used for packaging. These results allowed us to
draw up a sort of low-carbon guideline that included the identified
opportunities for GHG emissions reduction that are summed up in
a hypothetical best-in-class case study which would permit a
reduction of 2e3 kg of CO2eq compared to the best case study of the
five analysed here and of about 5.76 kg of CO2eq compared to worst
one. The methods for assessing the quality of data allowed us to
make important considerations on the most impactful phases and
inputs and to conclude that a vertical organization of the olive oil

Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15 13

Fig. 5. Sensitivity analysis.

12.000

10.10
kg CO2eq/FU
9.40
10.000 kg CO2eq/FU

2.36
7.76
7.73 kg CO2eq/FU
8.000 3.20 kg CO2eq/FU
Mill factory phase

Farm factory phase

2.41 2.62
6.000
4.48 4.43
kg CO2eq/FU kg CO2eq/FU

4.000 1.13 7.74 1.14

6.20
4.96 5.01
2.000
3.34 3.20

0.000
CASE 1 CASE 2 CASE 3 CASE 4 CASE 5 Best in class

Fig. 6. Best-in-class scenario (kg of CO2eq/FU).

supply chain could bring important benefits in reducing emissions.
This is confirmed by the best-in-class scenario that includes many
existing elements of a private mill.
Moreover, three main issues could be highlighted relating to the
methodological aspect of applying the CF method in the olive oil
production sector:
 lack of context-specific data with a suitable level of detail for
each process and sub-process e there was a certain difficulty in
retrieving data, especially for the farm factory phase. In fact, the
process of interviews and data collection was lengthy as the
analysed sample was quite large and the data had to be
reorganized so as to be consistent with the required format.
There was also a lack of data for single processes and sub-
processes on the farm, and a general lack of knowledge of the
CF method, which made it even more difficult to obtain data. In
respect to the existing literature, the research presented here
had more detailed collection of data on farms, but future
research in this field should improve the availability of this kind
of data;
 difficulties in the quantification of carbon storage within the soil
and tree structures e carbon stored in the aerial structures of
the olive tree belongs to the short carbon cycle (pruning of the
branches is done every 1e2 years), while carbon stored in the

Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
14 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15
roots and that accumulated in the soil can fall within the long-
term carbon cycle (especially where the average life of a tree
reaches and exceeds 100 years) and, in this context, quantifying
and accounting carbon matter in the total GHG emissions still
appears difficult and subject to a high amount of uncertainty.
Further analysis is recommended to obtain robust and exhaus-
tive data on these aspects;
 need to improve the understanding of the role of climatic con-
ditions on the number of phytosanitary treatments e in the CF
method this should be reflected by using GHG estimation
techniques incorporating climate factors and by considering
temporal boundaries including a proper number of years (at
least for multi-annual crops such as olive groves, vineyards and
orchards) in order to take into account the variability in yields
per hectare and the influence of weather conditions on the
number and type of pesticide treatments.
Finally, with reference also to the PEFCR project, it can be said
that the present study has reaffirmed the need for a data collection
phase (especially for the agricultural phase) that is as clear and
extensive as possible. It is the most impactful phase, which presents
the greatest difficulties given its duration (about 10 months), its
heterogeneity (consumption of chemicals, fuel consumption,
alternating operators of agricultural machinery), and the great in-
fluence related to the territorial context (e.g. for climate conditions,
impacts distribution).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.jclepro.2016.03.152.
References
Accorsi, R., Versari, L., Manzini, R., 2015. Glass vs. plastic: life cycle assessment of
extra-virgin olive oil bottles across global supply chains. Sustainability 7,
2818e2840.
Arzoumanidis, I., Fullana-i-Palmer, P., Raggi, A., Gazulla, C., Raugei, M.,
Benveniste, G., Anglada, M., 2014. Unresolved issues in the accounting of
biogenic carbon exchanges in the wine sector. J. Clean. Prod. 82 (1), 16e22.
Avraamides, M., Fatta, D., 2008. Resource consumption and emissions from olive oil
production: a life cycle inventory case study in Cyprus. J. Clean. Prod. 16 (7),
809e821.
BSI-British Standards Institution, 2011. PAS 2050:2011 e Specification for the
Assessment of the Life Cycle Greenhouse Gas Emissions of Goods and Services.
British Standards, London, UK. Available at: http://shop.bsigroup.com/en/forms/
PASs/PAS-2050/ (last access July 2015).
Cecchini, F., Franchini, F., Neri, P., Guerra, E., 2005. Valutazione dell’Impatto ambi-
entale dovuto alla produzione biologica, integrata e convenzionale di 1 l di olio
d’oliva di aziende delSud d’Italia. Technical report PROT-P135e091. ENEA,
Rome.
Cimato, A., 2008. Ricerca e sperimentazione applicativa & trasferimento delle
innovazioni tecnologiche agli operatori della filiera olivicola/olearia. In: La
centralita del frantoio nella filiera olivicolo-olearia. Convegno 20 Aprile 2008,
Pescara.
Di Giovacchino, L., Prezioso, S., 2006. Utilization of olive mill by-products. In:
Proceedings of Olivebioteq, pp. 379e389.
Ecoil, 2007. Life Cycle Assessment (LCA) as a Decision Support Tool (DCT) for the
Ecoproduction of Olive Oil. Ecoil final technical report. Technical University of
Crete. Available at: http://www.ecoil.tuc.gr/technical_final%20report_F.pdf.
Ecoinvent Centre, 2007. Ecoinvent-report No. 15.a. Swiss Centre for Life Cycle In-
ventories. Ecoinvent Centre, Dübendorf.
Ecoinvent Centre, 2010. Ecoinvent Data V3.0. Swiss Centre for Life Cycle Inventories.
Ecoinvent Centre, Dübendorf.
Environdec, 2014. PCR “UN CPC 21537 Virgin Olive Oil and Its FRACTIONS”. http://
www.environdec.com/en/PCR/Detail/?Pcr¼5934#.U2uNAvl_tLo (last accessed
10.05.08.).
European Commission, 2013. 2014. Single Market for Green Products Initiative,
Fiche Olive Oil. http://ec.europa.eu/environment/eussd/smgp/index.htm (last
accessed on 09.02.205.).
Fedele, A., Mazzi, A., Niero, M., Zuliani, F., Scipioni, A., 2014. Can the life cycle
assessment methodology be adopted to support single farm on its environ-
mental impact forecast evaluation between conventional and organic produc-
tion? an Italian case study. J. Clean. Prod. 69, 49e59.
Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152
Huijberts, M.A.J., Norris, G., Bretz, R., Ciroth, A., Maurice, B., Von Bahr, B., et al., 2001.
Framework for modeling data uncertainty in life cycle inventories. Int. J. LCA 6
(3), 127e132.
INEA e Istituto Nazionale di Economia Agraria, 2014. Indicazioni per la valutazione
del rischio di frodi e indirizzo delle attivita’ di controllo nel settore olivicolo-
oleario. Rapporti. INEA 2014.
Intini, F., Kühtz, S., Rospi, G., 2011. Energy recovery of the solid waste of the olive oil
industriese LCA analysis and carbon footprint assessment. J. Sustain. Energy &
Environ. 2, 157e166.
IOOC e International Olive Oil Council, 2012. Overview of the Olive and Olive Oil
Sector Itemised by Country and Based on the Official Replies to IOC Question-
naires, 17 October 2012. http://www.internationaloliveoil.org/estaticos/view/
136-country-profiles.
IPCC e Intergovernmental Panel on Climate Change, 2006. Guidelines for National
Greenhouse Gas Inventories.
IPCC e Intergovernmental Panel on Climate Change, 2007. In: Solomon, S., Qin, D.,
Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.),
Contribution of Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change, 2007. Cambridge University Press,
Cambridge, United Kingdom and New York, NY, USA.
IPCC e Intergovernmental Panel on Climate Change, 2014. In: Pachauri, R.K.,
Meyer, L.A. (Eds.), Climate Change 2014: Synthesis Report. Contribution of
Working Groups I, II and III to the Fifth Assessment Report of the Intergov-
ernmental Panel on Climate Change [Core Writing Team. IPCC, Geneva,
Switzerland, p. 151.
ISO e International Organization for Standardization, 2013. ISO/TS 14067:2013
Greenhouse Gasesdcarbon Footprint of Productsdrequirements and Guide-
lines for Quantification and Communication. The International Organization for
Standardization, Geneva.
ISTAT, Istituto nazionale di statistica, 2013. Atti del 6 Censimento Generale del-
l'Agricoltura. ISTAT, ISBN 978-88-458-1779-3.
Lazarevic, D., Aoustin, E., Buclet, N., Brandt, N., 2010. Plastic waste management in
the context of a European recycling society: comparing results and un-
certainties in a life cycle perspective. Resour. Conservation Recycl. ISSN: 0921-
3449 55 (2), 246e259. December 2010. http://dx.doi.org/10.1016/j.resconrec.
2010.09.014.
Lucchetti, M.C., Romano, I., Arcese, G., 2012. Carbon Footprint: un’analisi empirica
per la produzione di olio. In: Paper presented at the VI Convegno della Rete
Italiana LCA “Dall’Analisi del Ciclo di Vita all’impronta Ambientale” Bari.
Mohamad, R.S., Verrastro, V., Cardone, G., Bteich, M.R., Favia, M., Moretti, M.,
Roma, R., 2014. Optimization of organic and conventional olive agricultural
practices from a life cycle assessment and life cycle costing perspectives.
J. Clean. Prod. 70, 78e89.
Neri, E., Rossetti, F., Rugani, B., Niccolucci, V., Pulselli, F.M., Bastianoni, S.,
Marchettini, N., 2012. Life Cycle Assessment and eMergy application to the
comparison between organic and conventional productions. In: Paper Pre-
sented at the 8th International Conference on LCA in the Agri-food Sector,
Saint- Malo, France. (October).
Olivieri, G., Falconi, F., Pergreffi, R., Neri, P., Romani, A., 2005. Life Cycle Assessment
forenvironmental integrated system in the olive oil tuscan company. In: Paper
Presented at the European Meeting of the International Society for Industrial
Ecology, Bologna, Italy. (January).
Pattara, C., Cappelletti, G.M., Cichelli, A., 2010. Recovery and use of olive stones:
commodity, environmental and economic assessment. Renew. Sustain. Energy
Rev. 14 (5), 1484e1489.
Pelletier, N., Ibarburu, M., Xin, H., 2013. A carbon footprint analysis of egg pro-
duction and processing supply chains in the Midwestern United States. J. Clean.
Prod. 54, 108e114.
Polo, J.A., Salido, J.A., Mourelle, A., 2010. Calculation and verification of carbon
footprint in agricultural products. In: Paper Presented at the LCA Food VII
Conference on Life Cycle Assessment in the Agri-food Sector, Bari, Italy.
PRe Consultants, 2014. SimaPro V.8.0. Available from: http://www.pre-
sustainability.com.
Proietti, S., Sdringola, P., Desideri, U., Zepparelli, F., Brunori, A., Ilarioni, L., Nasini, L.,
Regni, L., Proietti, P., 2014. Carbon footprint of an olive tree grove. Appl. Energy
127, 115e124.
Raggi, A., Cichelli, A., Pagliuca, G., Costantino, A., 2000. A screening LCA of olive husk
combustion for residential heating. In: Paper Presented at the 1st World Con-
ference on Biomass for Energy and Industry, Sevilla. (June).
Rajaeifar, M.A., Akram, A., Ghobadian, B., Rafiee, S., Heidari, M.D., 2014. Energy-
economic life cycle assessment (LCA) and greenhouse gas emissions analysis of
olive oil production in Iran. Energy 66, 139e149.
Rinaldi, S., Barbanera, M., Lascaro, E., 2014. Assessment of carbon footprint and
energy performance of the extra virgin olive oil chain in Umbria, Italy. Sci. Total
Environ. 482, 71e79.
Romani, A., Olivieri, G., Neri, P., Tacconi, D., Pinelli, P., Lapucci, C., Ieri, F., 2004.
Qualita , Certificazione e Life Cycle Assessment (LCA) per oli vergini e acque
reflue di frantoio. In: Paper presented at the XXI Congresso Nazionale di Mer-
ceologia, “Risorse naturali e sviluppo economico- sociale. Contributo delle sci-
enze merceologiche”, Foggia. (September).
Salmoral, G., Aldaya, M.M., Chico, D., Garrido, A., Llamas, M.R., 2011. The water
footprint of olives and olive oil in Spain. Span. J. Agric. Res. 9 (4), 1089e1104.
Salomone, R., Cappelletti, G.M., Malandrino, O., Mistretta, M., Neri, E.,
Nicoletti, G.M., Notarnicola, B., Pattara, C., Russo, C., Saija, G., 2015. Life cycle
assessment in the olive oil sector. In: Notarnicola, B., Salomone, R., Petti, L.,
 C. Pattara et al. / Journal of Cleaner Production xxx (2016) 1e15 15

Renzulli, P.A., Roma, R., Cerutti, A.K. (Eds.), Life Cycle Assessment in the Agri-
food Sector: Case Studies, Methodological Issues and Best Practices. Springer
International Publishing, pp. 57e121.
Salomone, R., Ioppolo, G., 2012. Environmental impacts of olive oil production: a
Life Cycle Assessment case study in the province of Messina (Sicily). J. Clean.
Prod. 28, 88e100.
Saner, D., Walser, T., Vadenbo, C.O., 2012. End-of-life and waste management in life
cycle assessment. Int. J. Life Cycle Assess. 17 (4), 504e510. May 2012.
Scotti, M., Bondovalli, C., Bodini, A., 2009. Ecological Footprint as a tool for local
sustainability: the municipality of Piacenza (Italy) as a case study. J. Environ.
Impact Assess. Rev. 29, 39e50.
UNAPROL e Consorzio Olivicolo Italiano, 2013. Filiera olivicola. Analisi strutturale e
monitoraggio di un campione di imprese. Campagna 2011-2012. http://www.
unaprol.it/images/stories/filiera%20olivicola%202012-2013.pdf.
UNAPROL e Consorzio Olivicolo Italiano, 2014. Filiera olivicola. Analisi di scenario e
studi di settore. II annualita . http://www.unaprol.it/images/stories/Filiera%
20Olivicola%202012-2013_copy.pdf.
UNAPROLe Consorzio Olivicolo Italiano, 2010. Filiera olivicola. Analisi strutturale e
monitoraggio di un campione di imprese. Campagna 2008-2009. http://www.
unaprol.it/images/stories/documenti/Filiera%20Olivicola%202008-2009.pdf.
Vermeulen, S.J., Campbell, B.M., Ingram, J.S., 2012. Climate change and food systems.
Annu. Rev. Environ. Resour. 37 (1), 195.
Weidema, B.P., Wesnaes, M.S., 1996. Data quality management for life cycle in-
ventories e an example of using data quality indicators. J. Clean. Prod. 4 (3e4),
167e174.

Please cite this article in press as: Pattara, C., et al., Carbon footprint of extra virgin olive oil: a comparative and driver analysis of different
production processes in Centre Italy, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.03.152