Abstract
Purpose
Improving the quality and quantity of unit process datasets in Life Cycle Inventory (LCI) databases affects every LCA they are used in. However, improvements in data quality and quantity are so far rather directed by the external supply of data and situation-driven requirements instead of systematic choices guided by structural dependencies in the data. Overall, the impact of current data updates on the quality of the LCI database remains unclear and maintenance efforts might be ineffective. This article analyzes how a contribution-based prioritization approach can direct LCI update efforts to datasets of key importance.
Methods
A contribution-based prioritization method has been applied to version 3 of the ecoinvent database. We identified the relevance of unit processes on the basis of their relative contributions throughout each product system with respect to a broad range of Life Cycle Impact Assessment (LCIA) indicators. A novel ranking algorithm enabled the ranking of unit processes according to their impact on the LCIA results. Finally, we identified the most relevant unit processes for different sectors and geographies.
Results and discussion
The study shows that a relatively large proportion of the overall database quality is dependent on a small set of key processes. Processes related to electricity generation, waste treatment activities, and energy carrier provision (petroleum and hard coal) consistently cause large environmental impacts on all product systems. Overall, 300 datasets are causing 60% of the environmental impacts across all LCIA indicators, while only 3 datasets are causing 11% of all climate change impacts. In addition, our analysis highlights the presence and importance of central hubs, i.e., sensitive intersections in the database network, whose modification can affect a large proportion of database quality.
Conclusions
Our study suggests that contribution-based prioritization offers important insights into the systematic and effective improvement of LCI databases. The presented list of key processes in ecoinvent version 3.1 adds a new perspective to database improvements as it allows the allocation of available resources according to the structural dependencies in the data.
Similar content being viewed by others
Notes
The term product includes both goods and services.
Our definition of LCI database follows the definition of the Shonan Guidance Principles (Sonnemann and Vigon 2011).
The workload for updating a dataset varies greatly. Depending on the given level of completeness of a dataset, it might take weeks (e.g., when a global dataset is disaggregated into many country specific datasets) or only 1 h (e.g., when the quantity of one emission flow in one dataset is updated). Updating a dataset typically involves data collection, data entry/manipulation, and data submission to peer review (using a software tool called EcoEditor). Any change in an existing datasets or the submission of new datasets requires a peer review to be accepted into the database (Weidema et al. 2013). Therefore, one person-day should be considered as a rough but realistic estimate of the average effort associated with updating a dataset.
In principle, CA can also be applied on the inventory level (Heijungs and Kleijn 2001). We focus on environmental impacts, since we believe that a CA on the inventory level is of little practical interest in the context of prioritization of improvement efforts.
Table S1 in SI1 (Electronic Supplementary Material) provides detailed information on all matrices and equations involved and SI2 (tab. “LCI database example”) shows the example in excel.
The key adaptations concern the diagonalization (expressed by the hat, ˄) of vector e in Eq. 3, which facilitates an in-depth analysis of the contribution per process and the use of a demand matrix I (instead of a demand vector f) (Reinhard et al. 2016).
In this context, the identity matrix explicitly introduces a unitary product demand (reference flow) for all product systems (Heijungs and Suh 2002, p. 85). Algebraically, this multiplication has no effect, and for numerical efficiency, it can be omitted.
Note that e expresses environmental impacts (in terms of CO2 equivalents) and not the elementary flow CO2.
The operation in Eq. 7 ensures that the total environmental impacts refer to exactly one unit of a process and not one unit of product. It has no influence on our simple example but is important to achieve correct results in the presence of loops.
We use the modulus for the calculation of β. This reduces distortions due to negative contributions which can result from negative characterization factors and/or negative elementary flows (e.g., an uptake of heavy metals by plants in a cultivation dataset) in the biosphere matrix.
To date, a value lower than one was only observed for very simplistic database example (e.g., the sum of βcon is lower than one (0.36) in our very simplistic database example) but not for actual LCI databases.
Note that, once computed, S can be reused without further adaptation and consequently one matrix inversion is sufficient for the computation of any set of LCIA indicators.
As shown by Reinhard et al. (2016), the importance of one and the same unit process can vary a lot across different LCIA indicators.
As the zeros in the MLC tables do not relate to any process index, they are not required for the ranking algorithm.
In the presence of more than one successor of equal size—as with 1.00 (Y) and 1.00 (X) in IPCC 2007 for a ccv of 0.77—the successor with the lower h index is selected.
Only the index/indices which is/are not already recorded will be added to the LCIA indicator specific row of the prioritization table.
One possibility is to vertically concatenate the rows (LCIA indicators) of the prioritization table (Fig. 2, step 3) into one matrix and then count the occurrence of each process ID separately for each ccv (column). SI3 (Electronic Supplementary Material) shows our implementation in MATLAB (see function “determineProcessrelevance”).
The system model is based on the cut-off approach where primary (first) production of materials is always allocated to the primary user of a material. Furthermore, a primary producer of a recyclable material does not receive any credit for its provision. Therefore, recyclable materials are available burden-free to recycling processes, and secondary (recycled) materials bear only the impacts of the recycling processes (Wernet et al. 2016).
In this system model, processes are only allowed to have more than one output if the non-unitary output represents a recyclable product.
Scientific quality is determined according to the recommendations in the ILCD-Handbook (EC-JRC 2011).
With inventory support, we refer to the total amount of unit processes in the database which have a contribution according to a particular LCIA indicator, i.e., which include an elementary flow addressing one of the characterization factors contained in a particular LCIA indicator (see Table S4 in SI1 (Electronic Supplementary Material) for some examples).
International Standard Industrial Classification of All Economic Activities, Revision 4 (see http://unstats.un.org/unsd/cr/registry/isic-4.asp)
Note, however, that the method offers some indirect support for the identification of blank spots, namely via the identification of processes which are modeled at a too generic level. Such process models should be replaced by more specific ones in a spatial or in technological scale.
References
Bourgault G, Lesage P, Samson R (2012) Systematic disaggregation: a hybrid LCI computation algorithm enhancing interpretation phase in LCA. Int J Life Cycle Assess 17:774–786
Ciroth A, Muller S, Weidema B, Lesage P (2016) Empirically based uncertainty factors for the pedigree matrix in ecoinvent. Int J Life Cycle Assess 21:1338–1348
Duclos J-Y, Araar A (2006) Poverty and equity: measurement, policy, and estimation with DAD. Springer
Ebner F (2013) Results, interpretation and reporting: GaBi Software. In: Gabi-Softw. http://www.gabi-software.com/software/gabi-software/gabi-5/functionalities/results-interpretation-and-reporting/. Accessed 16 Aug 2013
EC-JRC (2011) International Reference Life Cycle Data System (ILCD) Handbook- Recommendations for Life Cycle Impact Assessment in the European context. First edition November 2011. EUR 24571 EN. Luxemburg. Publications Office of the European Union
Finnveden G, Hauschild MZ, Ekvall T, Guinée J, Heijungs R, Hellweg S, Koehler A, Pennington D, Suh S (2009) Recent developments in Life Cycle Assessment. J Environ Manag 91:1–21
Goedkoop M, Oele M (2004) SimaPro 6 - Introduction to LCA with SimaPro
Heijungs R, Huijbregts M (2004) A review of approaches to treat uncertainty in LCA. Complex Integr Resour Manag Proc 2nd Bienn Meet Int Env Model Softw Soc
Heijungs R, Kleijn R (2001) Numerical approaches towards life cycle interpretation five examples. Int J Life Cycle Assess 6:141–148
Heijungs R, Suh S (2002) The computational structure of life cycle assessment. Kluwer Academic Publishers, Dordrecht
Hellweg S, Milà i Canals L (2014) Emerging approaches, challenges and opportunities in life cycle assessment. Science 344:1109–1113
Huijbregts MJ (1998) Application of uncertainty and variability in LCA. Int J Life Cycle Assess 3:273–280
Majeau-Bettez G, Strømman AH, Hertwich EG (2011) Evaluation of process- and input–output-based life cycle inventory data with regard to truncation and aggregation issues. Environ Sci Technol 45:10170–10177
Mutel C (2012) Framework and tools for regionlization in life cycle assessment. ETH
Reinhard J, Mutel C, Wernet G et al (2016) Contribution-based prioritization of LCI database improvements: method design, demonstration, and evaluation. Environ Model Softw 86:204–218
Sonnemann G, Vigon B (eds) (2011) Global guidance principles for life cycle assessment databases: a basis for greener processes and products (Shonan Guidance Principles)
Steubing B, Wernet G, Reinhard J, Bauer C, Moreno-Ruiz E (2016) The ecoinvent database version 3 (part II): analyzing LCA results and comparison to version 2. Int J Life Cycle Assess 21:1269–1281
Tillman A-M (2000) Significance of decision-making for LCA methodology. Environ Impact Assess Rev 20:113–123
Weidema B, Bauer C, Hischier R et al (2013) Overview and methodology. Data quality guideline for the ecoinvent database version 3
Wernet G, Bauer C, Steubing B, Reinhard J, Moreno-Ruiz E, Weidema B (2016) The ecoinvent database version 3 (part I): overview and methodology. Int J Life Cycle Assess 21:1218–1230
Acknowledgements
We gratefully acknowledge Christopher L. Mutel for the valuable comments and suggestions given to improve the manuscript. We are thankful to the four anonymous reviewers for their insightful comments and inputs.
Author information
Authors and Affiliations
Corresponding author
Additional information
Responsible editor: Yi Yang
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
This article contains supplementary material in two file formats: Supplementary Information 1 (SI1) is in Adobe pdf format, a file which complements the article with additional figures and discussions, and SI2 (Electronic Supplementary Material) in Microsoft Excel (.xlsx) format, a file provided for dynamic exploration of the data processed and discussed in the article.
ESM 1
(DOCX 3.77 mb)
ESM 2
(XLSX 28.6 mb)
ESM 3
(M 1 kb)
ESM 4
(M 1 kb)
ESM 5
(M 3 kb)
ESM 6
(M 1 kb)
ESM 7
(M 485 bytes)
Rights and permissions
About this article
Cite this article
Reinhard, J., Wernet, G., Zah, R. et al. Contribution-based prioritization of LCI database improvements: the most important unit processes in ecoinvent. Int J Life Cycle Assess 24, 1778–1792 (2019). https://doi.org/10.1007/s11367-019-01602-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11367-019-01602-0