Sensitivity Analysis: Testing the Significance of Your Assumptions

A Life Cycle Assessment is only as good as the data and assumptions built in the LCA. As we described in the previous post on developing the function unit, many informed assumptions needed to be made about use patterns of handkerchiefs and facial tissues in the LCA completed by Ecosystem Analytics Inc. We tried to base as many of our assumptions on previously published studies, but still, a judgment had to be made by the LCA practitioner. Assumptions needed to be made on the time length of use considered, the geographic relevance of data used, and the choice of Life Cycle Impact Assessment (LCIA) model.

Luckily, the LCA field has built in the testing of key assumptions into the life cycle assessment process so that we can understand the robustness of the conclusions of any LCA. This process is called Sensitivity Analysis, and according to the ISO 14040 standard on LCAs, sensitivity analysis is the “systematic procedures for estimating the effects of the choices made regarding methods and data on the outcome of a study.”

For this LCA, we performed sensitivity analysis on 3 areas where key assumptions were made:

1. Use Scenarios
2. Country of Production/Electricity Mix Scenarios
3. Impact Assessment Model Scenarios

Use Scenarios

As we described in the previous post on the design of the functional unit, we needed to make assumptions on the frequency of both facial tissue and handkerchief use over time. Two situations were constructed to model use – during respiratory illnesses (Cold) and during periods of wellness (Base Use). Three scenarios were designed to model use during a maximum illness (Max Cold), a minimum illness (Min Cold), and the absence of a respiratory illness (No Cold), based on published nose blow frequencies in a respiratory illness study. Two scenarios were constructed to cover the length of time that a handkerchief would be used before laundering it during the well, Base Use periods (Max Base Use = 1 day, Min Base Use = 7 days). All in all, six one-year use scenarios were created, with the functional unit (Max Cold & Max Base Use) one of them. The 5 other one-year use scenarios were tested as part of the sensitivity analysis, as well as two scenarios that covered the entire useful life of the handkerchief before disposal – Max Life Max Cold & Max Base Use and Max Life Max Cold & Min Base Use.

Country of Production/Electricity Mix Scenarios

As discussed in the previous post on geographic relevance, LCA practitioners often select data from a similar country or region instead of necessarily only using data from the country of production so to be able to use the most comprehensive and reliable LCA databases available. However, the electricity mix is selected in the unit processes to reflect the country of production since electricity often has a large impact on LCA results. For the sensitivity analysis, we flipped the electricity mix used in the model for the handkerchief and facial tissue functional unit. This resulted in scenarios that roughly modeled a case in which the location of manufacture had flipped – the facial tissues were manufactured in China and the handkerchiefs were manufactured in Canada.

Impact Assessment Model Scenarios

Every Life Cycle Impact Assessment (LCIA) model needs to make many informed assumptions to be able to transform a list of unit processes into numerical estimates of environmental impacts. Every LCIA model is constructed a little differently, reflecting new research in the LCA field. To test the importance of our choice of LCA model, we also ran our results through a newer LCIA model – ReCiPe 2008.

In the next post, we will discuss the role of interpretation in understanding LCAs and putting them in context while considering previous studies and the stated study limitations.