Moving beyond point-values in LCAs
Posted on October 10, 2013 by willgriffiths
Life cycle assessment (LCA) is a useful tool in assessing a products total impact across its whole life time from it’s production, right the way through to it’s sale. LCA has been key to SEAT’s investigations but there are limitations when dealing with the diverse range of production methods and facilities in Asian aquaculture. In a new paper published in the International Journal of Life Cycle Assessment, which was presented at the 53rd LCA Discussion Forum in Zurich, experiences from the SEAT project have been called upon to develop a new protocol for these LCAs and making statistical analysis to lead to more confident conclusions. Here Patrik Henriksson explains:
Aquaculture is more diverse than terrestrial farming systems, both in terms of the number of species that are farmed and the variety of different farming systems. Salmon farming is probably the most familiar to Westerners and this is one of the most homogenous aquaculture systems, dominated by single species systems that rely on commercial feeds and formulated inputs. Asian farming systems, however, have evolved to become far more diverse, often capitalising on available opportunities (see table below). In the SEAT project, the focus is on four countries and within each country there are two focus species. Each farmed species is, in turn, often divided into two or more distinctly different farming practices, each with its own set of environmental interactions. These different farming practices may be a result of differences in regional conditions (e.g. shrimp from Surath Thani or Chanthaburi province in Thailand), capital investments (e.g. high or low-level ponds in China), scale of production (e.g. small, medium or large scale pangasius farms in Vietnam), local legislation (e.g. mixed mangrove-shrimp farms in Vietnam) integration with other species (e.g. different combinations of shrimp, prawn, rice and fish in Bangladesh) or simply tradition.
When investigating farming impacts, ideally, each farm should be evaluated individually, with respect to its unique set of inputs and consequent outputs. Naturally, however, this is an impossible task when evaluating such a diverse matrix of possible systems. Averaging becomes a necessity but this has the unavoidable result of simplifications. This is why we generally want to present standard deviations around averaged data as a proxy for the underlying diversity, here referred to as spread.
As mentioned, when conducting life cycle assessments (LCAs), averaging is a norm. Meanwhile, given the large complex systems that the underlying life cycle inventories (LCIs) often present, estimates for spread are almost never incorporated in the field of LCA. Even highly generalised products (eg. biofuel from US corn) are presented using only point-values. This hampers not only statistical testing of conclusions, but also reduces the potential of being able to identify improvement options in value chains.
As part of SEAT’s Life Cycle Assessment work package (WP3), we addressed this issue and produced a protocol for including overall dispersions in LCA. Alongside spread, inherent uncertainty and unrepresentativeness were identified as leading causes for overall dispersions in LCIs. The research outcomes were recently published in the International Journal of Life Cycle Assessment (with supporting material available at the Leiden University website). Interest in this research also resulted in an invited talk at the 53rd LCA Discussion Forum in Zurich, Switzerland, where this methodology was well received.
The protocol will be adapted to LCAs conducted as part of the SEAT project. This will assist in the analysis of the large diverse dataset collected to date, and allow for statistical conclusions to be made. It will also highlight data gaps and promote ranges, rather than point values, allowing for more fair representation of environmental sustainability in qualitative product schemes, such as the EAFI.