Background qualitative analysis of the European reference life cycle database (ELCD) energy datasets – part II: electricity datasets

The aim of this paper is to identify areas of potential improvement of the European Reference Life Cycle Database (ELCD) electricity datasets. The revision is based on the data quality indicators described by the International Life Cycle Data system (ILCD) Handbook, applied on sectorial basis. These indicators evaluate the technological, geographical and time-related representativeness of the dataset and the appropriateness in terms of completeness, precision and methodology. Results show that ELCD electricity datasets have a very good quality in general terms, nevertheless some findings and recommendations in order to improve the quality of Life-Cycle Inventories have been derived. Moreover, these results ensure the quality of the electricity-related datasets to any LCA practitioner, and provide insights related to the limitations and assumptions underlying in the datasets modelling. Giving this information, the LCA practitioner will be able to decide whether the use of the ELCD electricity datasets is appropriate based on the goal and scope of the analysis to be conducted. The methodological approach would be also useful for dataset developers and reviewers, in order to improve the overall Data Quality Requirements of databases.


Introduction
The European Platform of Life Cycle Assessment (EPLCA), a project initiated by the Institute for Environment and Sustainability (IES), has the objective of promoting Life Cycle Thinking (LCT) and providing appropriate support to business and public administrations within the European Union (EU), as well as in close coordination with international activities. This support is essential, and is being achieved through the development of a number of different deliverables, being the European Reference Life Cycle Database (ELCD) one of them. The ELCD provides core Life Cycle Inventory (LCI) data from front-running EU-level business associations and, where not available, other sources. Several energy-related data are provided within the ELCD, since energy is a key input to most environmental analyses of products or processes. The ELCD latest version can be consulted on the JRC webpage: http:// eplca.jrc.ec.europa.eu/ELCD3/.
Data quality in LCA still represents a major bottleneck to a broader use of LCA and environmental footprint methods in business and in policy (Fazio et al. Method applied to the background analysis of energy data to be considered for the European Reference Life Cycle Database (ELCD). Springer Plus -Submitted in 2014). Under the framework of ISO standards (ISO 14044:2006)  The objective of this analysis is to identify areas of potential improvement of the ELCD electricity datasets quality, considering data available in third party life cycle databases and from authoritative bodies and/or business associations. The work has consisted in analysing and comparing electricity datasets from different databases, considering the ELCD database as the basis for this analysis. This effort has been carried out in two stages: i) Selection of datasets, databases and quality standards, in order to assure the methodology, ii) Analysis and qualitative comparison of the datasets, each selected electricity dataset was analysed according to previously defined quality standards. Then, findings and recommendations were derived in order to identify the potential improvements of ELCD datasets.

Selection of datasets and databases
The energy datasets to be analysed should be representative of a significant share (such as 40 to 60%) of the European electricity market and associated technology mixes/geographic origins, therefore a deep review of the most updated data in terms of electricity for EU-27 has been conducted.
According to European statistics (EUROSTAT, 2012;EC, 2011), the energy sources that contribute the most to electricity generation in 2011 were the following: Nuclear (27%), coal (26%), gas (23%), hydro (13%) and wind (4%). Other renewable energy sources have lower contribution to electricity generation in EU-27, such as biomass and waste, and solar energy (3% and 0.68%, respectively). However, due to their foreseen potentials, their contribution is expected to increase in the future. So, electricity from these sources was considered for the analysis. An electricity mix for EU-27 was also taken into account.
The latest ELCD version includes one dataset of European average electricity mix as well as electricity mix datasets from each EU-27 country. However, the unit processes used to build the datasets cannot be broken down into technologies. This limitation had to be solved, since the final objective of the analysis is to analyse the quality of the different datasets, focusing on the underlying models and data used. These ELCD electricity mix datasets by country have been originated from PE International (GaBi 2012). Taking into account the above mentioned limitation, the use of specific datasets from GaBi for conducting the analysis seemed to be essential. Whenever ELCD database did not provide the required datasets, GaBi datasets from the last updated version were analysed. It must be noticed that GaBi provides these datasets for each EU-27 country, but does not include datasets for each technology referring to the European context (which are available in the developer's internal database -PE International-, but so far not in the commercially available databases, i.e. electricity production from hard coal, European Mix). As a first approximation, in order to take into account the European energy market, the datasets by country were chosen from GaBi database considering only those countries that sum up 60% of the electricity produced in Europe for each technology (this value has been decided by the leaders of this evaluation as a first approach, and considering that it will be representative enough for the European energy market). Hereinafter, the nomenclature of ELCD energy datasets will refer to GaBi datasets.
In order to identify those countries that sum up more than 60% of the electricity produced in Europe by technology, data of electricity production by sources from Eurostat (data from 2010) were collected and analysed. Germany (23%), United Kingdom (21%) and Poland (20%) were the main producers of electricity from hard coal; Germany (41%), Czech Republic (14%), Poland (14%) and Greece (9%) were the main contributors to lignite electricity production; the main producers of electricity from natural gas were United Kingdom (20%), Italy (20%), Germany (13%) and Spain (10%); and the main producers of electricity from nuclear power were France (47%) and Germany (15%). Then, Table 1 shows the eighteen chosen datasets as the base for the comparison with other datasets. These datasets have been compared to their counterparts from three other databases: Ecoinvent v2.2 (Ecoinvent 2012), GEMIS 4.7 (GEMIS 2012), and E3 database (E3 2012). Considering theses databases and the availability of datasets, Table 2 presents the list of datasets to be finally analysed. The database selection have been made irrespective of the methodological compliance of the database/datasets with the ILCD quality criteria: it was indeed assumed that although other databases might have lower data quality rating (DQR) according to ILCD rules (because they were not specifically developed using these rules), datasets would represent interesting benchmarks and

Quality criteria for analysis
The evaluation has been based on the quality indicators developed within the ILCD handbook (EC-JRC, 2010a, 2010b, 2011: Technological representativeness (TeR), Geographical representativeness (GR), Time-related representativeness (TiR), Completeness (C), Precision/Uncertainty (P) and Methodological appropriateness and consistency (M). Each of those has been evaluated according to the degree of accomplishment of the criterion (from 1 to 5), and an overall DQR of the datasets has been calculated by summing up the achieved quality rating for each of the quality criteria indicator, divided by the total number of considered indicators, as described in Garraín et al. Background qualitative analysis of the European Reference Life Cycle Database (ELCD) energy datasets -Part I: Fuel datasets. Springer Plus -Submitted in 2014. The quality indicators described in the ILCD Handbook (EC-JRC, 2011) provide a general framework to evaluate datasets. When applying these indicators to specific sectorial datasets, it is necessary to redefine them based on the specific characteristics of the processes/technologies in order to identify key aspects. For this purpose, a deep pre-analysis of the technology situation was conducted, considering the European market context. The main features for assessing each criterion are similar to those described in  Table 3 shows both quality criteria definitions and values considered.       Table 4 shows the rates of the quality criteria assessment of the selected ELCD electricity datasets. Information contained in each dataset and additional confidential documents provided by the dataset developer (GaBi, 2012) were considered to define a final single value for each criterion. It should be noticed that only one dataset of each technology has been included in this article in order to show the full application of the evaluation method.

Discussion
The comparison of the selected datasets from different databases, referred to the same technology, can lead to the identification of potential improvements in each quality criteria. Moreover, relevant Authoritative Sources and Business Associations, which could provide additional information to improve the quality of the ELCD results, can be also identified in order to enhance the overall quality of data. It must be remarked that many recommendations are related to future updated versions of ELCD electricity datasets. Table 5 shows a summary of the findings and recommendations that arose from such cross assessment.

Conclusions and recommendations
This extended analysis of the ELCD electricity datasets aimed at providing better founded information related to its data quality, following the indicators developed and described within the ILCD handbook (EC-JRC, 2011). This analysis, together with the study on ELCD fuel datasets (Garraín et al. Background qualitative analysis of the European Reference Life Cycle Database (ELCD) energy datasets -Part I: Fuel datasets. Springer Plus -Submitted in 2014), have meant an opportunity to implement these quality indicators to different datasets for the first time. It has had two main consequences. Firstly, the implementation of the quality indicators to the energy-related datasets from the ELCD has been used to understand the room for improvement in future ELCD versions. Additionally, it has also served to identify whether these data quality indicators are applicable and useful for the database developers in general, as well as for the LCA practitioners. It should be stated that results obtained from this analysis ensure the quality of the energy-related datasets to any LCA practitioner, and provide insights related to the limitations and assumptions underlying in the datasets modelling. Giving this information, the LCA practitioner will be able to decide whether the use of the ELCD datasets is appropriate based on the goal and scope of the analysis to be conducted. Along the current analysis, several assumptions have been made in order to facilitate the analysis, such as the selection of databases and datasets or the definition of data quality indicators (DQIs). The results have to be understood under this context. Taking those considerations into account, the data quality assessment conducted in here should not be extrapolated to datasets under different contexts. Furthermore, the analysis has been performed only in a selection of the most representative electricity datasets from the ELCD as well as from the other selected databases. The conclusions obtained in this analysis cannot be extrapolated to other type of datasets, nor can be used to compare databases among them.
From the deep analysis conducted, it must be highlighted that the ELCD datasets have been modelled based on an extensive review of the most relevant literature and statistics. The documentation used to model the ELCD

C
• To fulfill the criterion in a 100% share, the following flows should be considered: Halon 1211 for ozone depletion, and indium for resource depletion impact category. P • Statistical information used to construct the electricity mixes of each country has been retrieved from the IEA (authoritative source). However, and due to the ELCD database has been developed by the EC in a European context, it seems adequate to use the data reported by each country to Eurostat.

M
• Inclusion of the EoL modelling of PV facilities.

General
• In order to have a more useful database in which users can update the EU27 electricity mix, datasets not only by country but also by technology should be available.
• Some analysed databases make use of energy models to derive future European electricity mixes, although this is not the scope of the ELCD.
Electricity from fossil fuels (hard coal, lignite and natural gas)

TeR
• CCS technologies can be included due to the importance in future environmental scenarios, as stated in several studies (Koornneef et al., 2008;Stanley and Dávila-Serrano., 2012).
• Several future clean coal, lignite and natural gas electricity scenarios can be developed and included in the ELCD datasets, as another database (GEMIS) includes.

C
• To fulfill the criterion in a 100% share, the following flows should be considered: Halon 1211 for ozone depletion, and indium for resource depletion impact category.  ( ) or UNSCEAR (1993( , 2000. In case of the French dataset, conversion data in facilities are available in the ExternE study of the French nuclear fuel cycle (EC, 1995).

TiR
• In both German and French datasets, TiR is the worst scored category in the ELCD database. The reason lies on the use of several old references. However, no better references could be found in the other databases analysed in this study. Other datasets (e.g. Ecoinvent) perform better since the validity period of the dataset is closer to the oldest references. P • Concerning radioactive emissions data, uncertainty can be decreased by using data published by UNSCEAR (2000), considered as an authoritative source.

M
• It could be improved with the consideration of a final repository for spent fuel and high activity waste. Data source can be those included in NAGRA (2002aNAGRA ( , 2002b. Electricity from hydro power TeR • In a future scenario, Small Hydropower Plants (SHPP) should be included due to the potential importance in the mix. According to statistical data from Arcadis (2011), a considerably reduction of electricity from hydropower mix is expected and the large facilities might be the main affected. Then, the share of SHPP in electricity from hydropower mix might increase; although a reduction of their potential is foreseen.
• To get additional inventory data, ESHA (European Small Hydropower Association, www.esha.be) publishes EU data facts and statistics of hydropower generation.

P
• The inclusion of documentation related to the data collection process and additional references to identify the origin of the data values could be useful to achieve a better rating. On the other hand, the IHA (International Hydropower Association, www.hydropower.org/), might be a relevant information source for double checking (annual reports and GHG Risk Assessment Tool that provides estimation of the level of gross GHG emissions from freshwater reservoir).

C
• It can be fulfilled completely with the consideration of Halon 1211 for ozone depletion, and cadmium and indium for resource depletion impact category. It must be highlighted that ELCD includes the emissions due to biomass degradation, while other datasets do not consider them.
Electricity from wind power TeR • Capacity factors and average sizes described are in line with the statistics provided by authoritative sources, such as the European Wind Energy Association (EWEA) and the International Energy Agency (IEA). It would be recommended to include additional energy related datasets can be found in the Life Cycle Thinking Platform web-site (http://eplca.jrc.ec.europa. eu/ELCD3/). In terms of the quality criteria, the analysed ELCD datasets showed a very good performance in the majority of the criteria, especially in those criteria related to TeR, C and M. Nevertheless, several recommendations for improving have been detailed above.
Concerning the different technologies analysed, ELCD datasets have the best quality rating in the majority of the technologies, with the exception of electricity from nuclear datasets in which TiR and M criteria score worse than other databases and PV dataset where M criterion also performs worse than in other databases. Several recommendations have been also made to overcome these limitations.
One of the most relevant weaknesses of the ELCD is the lack of datasets that model electricity produced by each technology in each European country. Currently, the ELCD includes electricity mix datasets for each country, modelled considering an established share of sources that might be different to the needs of the user. The inclusion in future versions will improve the flexibility and usefulness of the database. Moreover, in some renewable technologies (PV or biomass) regional specificities are not always well considered in terms of capacity factors, forest management, etc. In these cases, it would be desirable to split the dataset in different country specific or bioclimatic regions datasets. Although the optimal solution to this limitation would be to model new datasets for electricity production by technology and for each country, this might not be feasible in the short term. An alternative solution would be to model Table 5 Recommendations for improving ELCD electricity datasets by DQI (Continued) documentation, providing more detail concerning the different shares of onshore and offshore power as well as the contribution of each country to the total mix (e.g. the British Wind Energy Association). Additionally, it is recommended to review for future versions other wind options, such as the small and medium scale wind, which might increase in the future, and the re-powering, which substitutes old turbines, increasing the capacity.

GR
• ELCD dataset models a non-defined region in Europe. It must be highlighted that this energy source has a very site-specific resource and therefore, this technology applied in each European country and their contribution to the total electricity generation by wind in Europe might vary. However, ELCD takes into account this particularity by considering the full load hours for the actual region using statistical information. • It should be noted that no additional authoritative source has been found that could improve the ELCD dataset.
Electricity from photovoltaic TeR • It should be noted that this dataset has been modelled in a way that the European current technology is included. Among the other databases, the ELCD dataset contains the most updated information and provides deep details concerning the precision of the data used. To model this technology at least two relevant Authoritative Bodies have been used: the European Photovoltaic Technology Platform (EPTP) and the EurObserv'ER Barometer (www.eurobserv-er. org). The European Photovoltaic Industry Association (www.epia.org) provides detailed information related to the evolution of this sector yearly, and should be considered a relevant source for future versions.

C
• In order to improve the criterion, CFC-14 for climate change, Halon 1211 for ozone depletion, and indium for resource depletion impact category, should be also considered.

M
• ELCD should include also an EoL scenario in future versions (e.g from Lozanovski & Held, 2010). Moreover, it can be improved with the inclusion of a basic scenario of dismantling and waste treatment, considering main materials, such as steel or plastics, as Ecoinvent (2012).
datasets for each technology under a European context, and to introduce parameters in the electricity mix datasets to vary the shares of each technology. In order to give response to any change or advance in technologies, and to be able to model new datasets and/or to modify the current ones if necessary, it is highly recommended to constantly review the evolution of advanced technologies and their share in the European market. According to the approach endorsed by several authors for the modelling of consequential-LCA of energy systems (see e.g. Igos et al. 2014, Vázquez-Rowe et al. 2014, Earles et al. 2013, and Dandres et al. 2012, also future electricity scenarios can be developed using data from energy projective models such as PRIMES or TIMES (The Integrated MARKAL-EFOM System). This is an important improvement of the database that could be very useful for prospective and consequential LCA studies.
Modelling the End of Life (EoL) of the systems appears to be a difficult task due to the novelty of some technologies and the lack of data from other technologies (solar PV, final repository for spent nuclear fuel and natural gas plant dismantling). Efforts on this challenge should be kept in the future.
Regarding the use of authoritative sources, the ELCD database makes extensive use of the statistical information provided by the IEA. This is of course an authoritative source. However, for the European context it seems appropriates the use of data reported by each country to Eurostat. In order to improve precision, it would be advisable to make a more extensive use of Business Associations and Authoritative sources data that have been proposed throughout the analysis.
Since its first release, the ELCD database has been updated two times. The needs of reviewing and updating the ELCD database depend on the different sectors and the technologies. It would be useful to define periods to revise the electricity related datasets. For this purpose, a deep analysis of the learning curves would identify the level of maturity for each technology. Then, special periods for reviewing could be identified by technology.
Finally, it should be noted that the selected databases are in a constant process of updating and improvement, e.g. Ecoinvent v3.0 or GEMIS v4.93, so a detailed analysis of these, together with a deeper analysis of related reference documents (Treyer andBauer, 2013, 2014) can offer further potential improvements to future ELCD versions.