Poor data quality undermines LCA results

Life-cycle inventory (LCI), the data gathering stage of an LCA, is a "very data-intensive exercise which implies an accuracy that is not there", Bea de Smet, (Environmental Quality Manager, Procter & Gamble) told a European Chemical New and Chem Systems conference in Brussels in November.

Even for packaging LCIs, a product group which has been analysed extensively, a 20% difference in the results "is what you need to be sure that you have got significant savings," according to Peter Hindle of P&G. For the Swiss supermarket chain, Migros, this decision point is 30%.

A typical LCI "will incorporate easily up to 10,000 single data entry points," said Bea de Smet. The data will come from many countries and diverse sources, and may represent out-of-date technologies and different energy or waste management infrastructures. Major errors can also stem from the arbitrary setting of system boundaries.

Dishonesty has also been the cause of many data quality problems. "Because so many past LCAs have been defensive," said Claude Fussler (Environment, Health and Safety Vice-President, Dow Europe), "data was often provided on a selective basis to present the product and its application in the best light. This has tainted the credibility of both the data sources and the LCA methodology."

The Society of Environmental Toxicology and Chemistry (SETAC) has been developing LCA methodology through several workshops and publications. It held a workshop in October 1992 to draw up a "conceptual framework" for LCA data quality, with the establishment of minimum acceptable data requirements.

The report on the workshop proceedings includes a survey of 33 LCA practitioners in Europe and North America. It concludes that they "appear to recognise the tenuous nature of much LCI data and appropriately couch their reports in cautionary language." However, it also warns the "opportunities for selective out-of-context use of LCI information are still possible."

The survey found that data quality assessment "is by no means as widespread, uniform or rigorous as it should be." The SETAC report goes on to suggest how such assessments should, in general terms, be performed.

Either a qualitative or a quantitative approach can be used to give data quality indicators. In one quantitative method, scores are given for each data point based on age, representativeness, completeness, whether it is an industry average or site-specific, and other factors. These are then given a weighting and amalgamated through the inventory to give an overall data quality score. The method is useful in forming an opinion on the overall data quality but, like LCA itself, expressing the score numerically may "imply a precision that does not exist," SETAC cautions.

The report recommends that databases should be developed and augmented as a means of improving data quality. A commonly available database could also reduce the costs of performing LCAs and increase their use as an environmental management tool.

The Society for the Promotion of Life-cycle Assessment Development (SPOLD) has already commissioned a feasibility study of an open database of standard inventory data (ENDS Report 226, p 25 ). SPOLD hopes to have a database containing 50 or so commodity data sets ready by December 1995. The quality of each will be peer reviewed.

Similar initiatives are under way in Canada, Japan and the USA. In one programme, Chem Systems is preparing a North American plastic industry life-cycle inventory database.

The need for a database has also been underlined in an interim report by the "Groupe des Sages", which is being funded by the European Commission to draw up guidelines for the application of life-cycle assessment in the EC eco-labelling programme. A public database, the European Library of LCA Data, should be set up for commodity materials, fuel and services, it says.

The Groupe des Sages is also standardising an LCA methodology for use in defining eco-labelling criteria. However, a consistent approach is unlikely to emerge for some time because the methodology is still in its infancy.

In the meantime, the Groupe has identified numerous research needs. The main problem lies at the impact assessment stage. Here the Groupe suggests that the method developed by CML in the Netherlands, and supported by SETAC, should be used (ENDS Report 231, pp 20-22 ). This relies on the aggregation of data from the LCI based on "equivalency factors" for each impact category. For the global warming impact category, for example, the global warming potential of each greenhouse gas in the LCI can be used for this purpose.

For some impact categories, calculating equivalency factors is fairly straightforward. However, the Groupe has not reached a consensus on whether others - particularly those whose impact is on a local or even regional level, such as ecotoxicity, acidification or human toxicity - should even be included. Instead, further research is advocated, suggesting that "regional and local impacts may in due time be brought within the scope of LCA."

Meanwhile, a technical committee of the international standards body, the ISO, has completed the first stage of its development of a life-cycle assessment standard or guideline. This is part of the ISO 14000 family of environmental management standards. The work is not expected to be completed until after 1997, when there may be a consensus on impact assessment.

A decision whether to have a standard or a guideline has not been taken. Some LCAs are for internal use only, in which case a guideline is sufficient. However, LCAs are also used to make comparative claims in the public domain. Here there is a need for a standard with a validation or critical review process.

The ISO paper warns that LCA involves assumptions, value judgements and trade-offs and therefore is not a purely scientific endeavour. As such, the results of LCA should be "cautiously interpreted and applied" and the decision-making process should not rely solely on LCA. The results should not be reduced to a simple overall conclusion, it says.