Refinements of methods for life cycle impact assessment are directed at reducing errors and quantifying and reducing uncertainties in results. The uncertainty reduction benefits of such refinements depend upon the structure of the inventory model. The structure of inventory models in general are investigated using an economic input/output life cycle assessment model of the US economy. Percentiles for the share of total upstream emissions contributed by the set of processes in each supply tier are presented for US commodities and several important pollutants. Capturing at least 90% of the total direct plus upstream emissions for criteria air pollutants and toxic releases for at least 75% of the commodities in the US economy requires full modeling of direct emissions plus the first five supply tiers.
A method is developed and applied for streamlining Input/Output models. The method yields several conclusions relevant to risk-based life cycle impact assessment. The depth and breadth requirements for capturing a high percentage (e.g., > 80%) of total emissions vary widely across products or commodities. Models focusing on just direct plus the 15-20 most important processes in tier 1 can capture a median of just over 40% of emissions. To capture more than 60% of total emissions for more than half of all commodities requires models requires models with more than 4000 process instances. To well characterize the total impacts of products, life cycle impact assessment methods must characterize foreground process impacts in a site-informed way and mean impacts of far-removed processes in an unbiased way.
Keywords: Life Cycle Assessment, Life Cycle Inventory Analysis, uncertainty analysis, boundary truncation, Input/Output Analysis, Impact Assessment, streamlining