Special Session: LCA Data Interoperability – New Solutions to Old Challenges

Despite specifications in LCA like ISO 14048 and internationally-recognized exchange formats like Ecospold and ILCD, LCA practitioners continue to struggle with basic problems of LCA data interchange. Data are simply not easily merged from different LCI and LCIA sources and are not available in forms that enable users to easily bring them into their tools of choice. This can result in lack of availability of data and even more alarming inconsistences in implementation in LCA software that can result in differences in LCA results from the same datasets. New tools have recently emerged that attempt to tackle these issues. To what extent can new tools and data formats help us solve this problem, or do we need to totally rethink how we structure and create LCI and LCIA data? Among the new solutions are new approaches to LCA data development and metadata, new open-source tools providing translation capabilities, and new data formats and technologies that could possibly make smart, machine readable use of LCA data. The purpose of this session is to summarize the outstanding issues and present recently developed solutions, to discuss pros and cons of each, and to suggest a path forward in the community in which one or more of these tools and technologies can become mainstreamed to make use of data more possible and powerful for the average LCA practitioner. Outcomes of this discussion will hopefully inform existing initiatives, such as the recently formed Global Network for LCA Databases, on how best to address data interoperability challenges.

13:30 Pascal Lesage

LCIA implementation in software: Alarming differences

ABSTRACT. It is generally well known that the choice of Life Cycle Inventory (LCI) database or Life Cycle Impact Assessment (LCIA) method can have a significant effect on the results of a Life Cycle Assessment (LCA). What is less well known is that the choice of LCA software can also drastically affect results, even when using the same LCI data and the same LCIA method. Such discrepancies were informally alluded to in some circles of the LCA community, and a paper recently accepted for publication in the Journal of Industrial Ecology (Speck et al., 2015) confirms, for four packaging-related product systems, that there can indeed be differences, and that these differences can affect the conclusions of one’s study. The work presented here compares the results of all ecoinvent 2.2 product systems characterized using both CML and ILCD recommended methods in SimaPro and GaBi, resulting in 3999 comparisons per impact category. While results were very similar for some impact categories (acidification from ILCD recommended methods, global warming from both methods). On /average/, differences were also quite minimal. However, for many impact categories, hundreds of results which should be the same were different by at least one order of magnitude. In some extreme cases, differences were over 5 orders of magnitude. In cases where the datasets associated with these significant differences were significant contributors in a given product system, the conclusions of an LCA could very much be dictated by the choice of software more than the actual environmental performance of the assessed product. The severity of differences for some impact categories can undermine the very credibility of the software tools and, by extension, of LCA itself. The discrepancies are based on differences in interpretation and implementation of elementary flows (LCI) and characterization factors (LCIA). Tools aimed at facilitating cross-platform data transfer and nomenclature harmonization can help. However, a discussion among software providers, LCIA method developers and LCI database providers will be required to root out differences in interpretation of basic data generated by various actors in order to harmonize the implementation of databases and LCIA methods in software tools.

13:45 Chris Mutel

Can’t we all get along? The pain and promise of LCA data interchange

ABSTRACT. Exchanging LCA data with other people or software systems is much more difficult than it needs to be. In this presentation, I examine the current LCA data formats, and discuss the difficulties in linking projects and databases to other projects and databases. Detailed linking algorithms for each data format and software system are provided and illustrated with real world example projects and inventory databases. I also explain and compare the specific problems of each format.

Preliminary results include the following: – Different versions of the biosphere make it difficult or impossible to link even the same database when exported by different software systems – Some inventory databases are not internally consistent – Alteration of process names, units, and categories presents a significant barrier to linking process flows – Failing to support international text in standard encodings (e.g. unicode and utf-8) presents a significant barrier to linking process flows – LCA software as a whole is weakened by a lack of real competition or expectation of interoperability

To address some of these challenges, I developed new software for linking disparate data sources and formats, called brightway2-io [1]. This software recognizes that real-world data sources are messy, and that perfect algorithmic linking is impossible. Given these constraints, sets of linking strategies can be applied in an iterative process by LCA practitioners until a “good enough” result is reached. The software is demonstrated with real data examples.

This presentation was supported by SCCER Supply of Electricity.

[1] https://bitbucket.org/cmutel/brightway2-io

Submitted to special session “LCA Data Interoperability: New Solutions to Old Challenges”

14:00 Wesley IngwersenThomas TransueDavid MeyerMatthew BergmannEzra KahnPeter Arbuckle and Heidi Paulsen

The LCA Harmonization Tool

ABSTRACT. Use of best available data to support life cycle assessments (LCA) is hampered by differences in nomenclature between different datasets. Overcoming these differences is a particularly challenging task because of the size of LCA datasets and the continuous generation of new data. The Life Cycle Assessment Harmonization Tool (LCA-HT) uses advanced technology based on semantic web architecture that will semi-automate the process of harmonizing elementary flows in life cycle inventory (LCI) and life cycle impact assessment (LCIA) datasets. Datasets are imported as lists of elementary flows or as full LCI or LCIA datasets in a newly defined JSON-LD format for LCA. Elementary flows are broken down into components including the material or chemical, compartment, and unit, and harmonized against a user-defined master lists of each of these components. Elementary flows for LCI and LCIA are matched to assure that impact assessment calculations capture all characterized flows. Harmonized lists can be exported as text files or imported into openLCA software. The LCA-HT is a stand-alone OS independent desktop tool that is open-source and will be freely available. An overview of the tool will be given, features explained, and current limitation and next steps discussed.

14:10 Andreas Ciroth and Michael Srocka

JSON-LD: A smarter format for LCA data interchange
SPEAKER: Andreas Ciroth

ABSTRACT. For long, there has been debate about different LCA data formats, about conversion issues between different data formats, and about enabling better data exchange between existing LCA software systems and LCA users. Currently existing LCA data formats are all based on XML, a standard that has emerged at the end of the 1990’s. Overcoming all issues in data conversion seems an enormous effort, partly also because of the limitations of XML, but also because of incompatible concepts between different LCA formats. We are proposing a new data format for LCA data exchange. The format uses JSON-LD, which is a lightweight, modern format for linked data, suited also for large data amounts; it is meanwhile used by major search engines including google for structuring information. A first implementation is available in the open source LCA software openLCA, in a recent project commissioned by US EPA. The new format has several advantages over the existing LCA data formats. It links directly to ontologies for LCA and is lean and human-readable at the same time. It offers further the chance to overcome existing differences in LCA data formats, and thus improve data exchange. The format will be shortly explained and demonstrated, differences and advantages will be discussed. One of the discussion points will be whether it is possible to include and provide the format also for other LCA software systems.

14:20 Brandon KuczenskiWesley IngwersenKrzysztof JanowiczPascal HitzlerGary Berg-CrossCharles Vardeman and Sangwon Suh

Ontology Design Patterns for Semantically Enriched LCA

ABSTRACT. The Semantic web refers to a set of technologies concerned with the assignment of meaning to data. Semantic web tools are designed to facilitate the automated interpretation of data by documenting the relationships among entities and articulating rules of inference about those relationships. An ontology design pattern (ODP) is a description of a set of concepts and their relationships in a particular applicaton domain, put in terms of formal logic. The objective of creating an ODP is to encode a description of some aspect of the world for a specific purpose. Once created, an ODP can serve as a guide for putting data resources in the application domain into semantic terms.

In this talk we present the outcomes of a workshop conducted in March 2015 to develop a set of ODPs for the domain of life cycle assessment in an event called a “Vocabulary Camp.” The workshop collected a group of LCA domain experts (including several special session co-presenters) together with a group of semantic data engineers. The objective of the meeting was to begin to develop a set of ODPs that could be used as a common basis for efforts to semantically enrich LCA data.

The group worked on three distinct and complementary patterns, which correspond to the formative LCA concepts of “Flow”, “Activity”, and “Environmental Impact.” Expressed as collections of logical axioms, the patterns can be applied to existing LCA data sets and serialization formats to relate heterogeneous data sources under a shared conceptual model. Because the relationships are formally specified, data described in their terms can be easily interpreted and reasoned about by automated tools, promoting interoperability efforts without restricting diversity or flexibility in data management.