| Against a backdrop of increasing demand for products, the packaging industry is continually searching for ways to meet the needs of consumers while simultaneously improving the environmentally sustainability of packaging. This requires making informed decisions based at least in part on knowledge of the environmental impacts attributable to the different packaging options being considered. Many companies have turned to using Life Cycle Assessment (LCA) to obtain the necessary environmental impact information. To support this effort, several software programs have been developed to aid in doing an LCA. These programs vary considerably in complexity and focus, leading to questions of how similar are the results they provide. To study the consistency of results across LCA software programs, nine common packaging systems were modeled in each of five different programs: GaBi, SimaPro, openLCA, COMPASS, and Package Modeling. Comparison of the LCA information provided by these programs for the packaging systems showed several significant differences. To better understand why the differences occurred, four simplified systems were created with the intent of minimizing the number of variables involved in the system models. Each of these systems consisted of obtaining and disposing of a single basic packaging material: aluminum, glass, corrugated board, and polyethylene terephthalate (PET). Once again a comparison of results showed significant differences. Using information from these basic material comparisons, and making use of the transparency in LCA calculations provided by GaBi and SimaPro, a study was done that traced the causes of the inconsistencies in results back to how implementations of assessment methods differed between the two programs. A comparison of 14 combinations of basic materials and impact categories involving 3 different assessment methods found 98 instances of characterization factors differing between GaBi and SimaPro. These differences in characterization factors accounted for all the discrepancies in results between the two programs. No errors in calculations were found with either program, and there were no significant differences in the type or amount of inputs and emissions in the life cycle inventories from the two programs. |
