LCA of Chemicals
Life Cycle Assessment of fine chemical production: A case study of pharmaceutical synthesis
Gregor Wernet • Sarah Conradt • Hans Peter Isenring • Concepción Jiménez-González • Konrad Hungerbühler
Supporting information: description of the energy separation algorithm
The method applied in this work was developed for another work (Wernet et al. submitted) but could be applied to the results for Substance A. The basic principle is the process-based approach followed by the ecoinvent database and our inventory work. In this approach, all processes exist on a unit process level. Ecoinvent also supplies aggregated data, the so-called system processes. These have only inputs from the biosphere and emissions, no inputs from the technosphere – all requirements of upstream processes are calculated and included in the system process. When calculating the results of an LCA, the aggregated LCI of the process under analysis is calculated and then assessed. Unfortunately, once the aggregated inventories are compiled, it is not possible to separate specific flows by their uses. This is problematic specifically for the analysis of chemical products, as the main material feedstocks in this case, oil and natural gas, are also the main energy sources. A high need of oil and gas for a specific product therefore is not a good indicator for the energy needs of a production, as the production may be mass-intensive. Therefore, as we desired to assess specifically the importance of energy use over the production cycle, we created a different approach.
The Brightway LCA software was modified for this purpose to assess the data in several ways. First, a setup was created that calculated regular LCIs and LCA results. The results here matched results from other software such as SimaPro. The second setup was a modified calculation where all processes in the system (meaning the ecoinvent database and the process data generated in this work) were classified as to whether or not they were energy-producing processes. That included all processes whose output was heat, steam or electricity. The modified calculation treated these processes as having no inputs and no outputs except for the desired form of energy. In layman’s terms, the calculation described a hypothetical scenario in which these forms of energy are freely available without any environmental burdens. By comparing the two scenarios, the exact impacts of energy use including all upstream processes can be determined.
In a further approach, another calculation setup was created to determine the impacts of transport processes on the overall results. As the impacts of transport proved to be minor, both categories were combined into the ”energy-related” category described in the main article. Note that one cannot simply combine the results of the two individual categories, as energy production requires transport and transport requires energy production. A simple addition would therefore lead to double counting and to an exaggeration of the energy-related impacts. Instead, a new calculation setup was created where both categories were included.
During the preparation of the calculation, the ecoinvent database was analyzed for processes where the unit process approach was not consistently employed. The most important problem were the data on about a dozen basic chemicals in ecoinvent based on APME (Plastics Europe) data. These were available only as aggregated results, which would render our approach unfeasible. Fortunately, the original source data for the ecoinvent datasets included information on the fractions of impacts due to energy production and use as well as transport. These processes and others with similar issues were manually corrected to ensure that each calculation setup used the correct data.
1