Usually it is a combination of several aspects which influence the difference of LCIA results of a certain product. While each dataset is unique, the most common reasons for the differences between the LCIA results of a product in ecoinvent v2 and ecoinvent v3 (unrelated to the choice of system model) are the following:
1. Dataset updates For example the coconut husk production unit process did not contain any irrigation in ecoinvent v2. After an update of the dataset with more accurate data, ecoinvent v3 it does contain irrigation in version 3, which affects results in many LCIA categories. In general all electricity markets ecoinvent v2 have been updated; this alone has an influence on the LCIA results of almost all products.
2. Changes in the allocation approach Previous versions of the ecoinvent database used mass allocation for many multi-output processes. A review of these datasets revealed that many of these uses of mass allocation were not appropriate and the allocation approach was switched, usually to economic allocation. This affects the resulting inventories and therefore the LCIA results.
3. Updates in the transport distances For version 3, the default assumptions used for transport inputs in datasets where no primary data were available has been updated and improved. The new modelling is sector-based, relies on product properties, and is using real global transport data as a basis.
4. Global supply chains With the introduction of a more global supply chain modelling, many existing activities on non-European and non-Swiss production are now better linked to the datasets on European and Swiss production. This reflects the realities of global trade. Due to the better integration of global sources, the supply chains of localized products often changed, which results in a change in LCIs and LCIA results.
5. Changing from recycled-content allocation to allocation at the point of substitution. Different types of products produced as by-products were cut-off in ecoinvent v2. This approach has been changed in ecoinvent v3 and many by-products were re-introduced into the datasets adapted from version 2. In addition the allocation at the point of substitution means that the burdens (environmental impacts) related to the waste generated are partially assigned to the activity which produces this waste. The removal of the cut-offs can significantly influence the final LCIA results of recycling materials.