General Approach when Choosing Databases

Question

I have been thinking about how one should approach the choice of databases. As a first time user of openLCA, or any type of LCA software for that matter, I feel lost when contemplating the enormous selection of databases. 

Does anyone here have a standard procedure when selecting databases or data sets for their project? Do you prioritize the organisation behind the sets (GaBI, EcoInvent etc), the geographical cohesion of the data or another parameter?

 Furthermore, it feels a bit awkward to use a huge amount of data that you have had no part in collecting. I know that the organisations providing the data are trustworthy and professional but it still feels strange to use something without proper insight. Essentially, how do you justify the usage of these kinds of databases when talking to people with no experience in LCA?

Finally, how does one prioritize between the usage of personally collected data and data from the databases? Logically, the data I collect on my own would fit my specific scenario well but on the other hand, data from the databases are more likely to reflect a general case/process and are acquired and evaluated by professionals. Are there any guidelines to follow when using these two types of data?    

Sorry if I made a long post. I appreciate any replies or feedback.     

Answer 1

Hi Nils,

sorry for leaving your good comment and question without reply for some time - I feel that the LCA community does not have a clear answer here yet.

Of course, in some restricted, more precisely defined application contexts, it happens that one specific background database is mandatory and only possible to be used (Environmental Footprint, e.g.) which helps to align results of different studies done, but on the other side means that any process and product not available in this "prescribed" database cannot be addressed (if you do not model it yourself) and thus has zero impacts, which is also not perfect if these products and processes are not clearly negligible for the study (which, without knowing them, is always hard to decide). 

For Environmental Product Declarations, there have been also specific databases proposed, with often funny and annoying outcomes (GaBi databases prescribed for EPDs in Germany, ecoinvent in some other countries). This has been softened recently though, so that you can use ecoinvent e.g. also in German EPDs (I am shortcutting a bit the details).

In general, you can consider if you accept databases that only distribute aggregated system processes, where you in the end e.g. need to trust what is done by a German SME (that was recently purchased by a US company), without possibility to really check this work, and where you cannot change anything in the supply chains yourself, or if you don't accept only-system-process databases. Then, in UN GLAD, a data quality concept has been developed which helps to find and assess datasets, and this could in principle be applied for entire databases too but is not available yet. However, in principle, these data quality indicators assess whether the data fits to your modeling approach and to your product and process that you need, and this is of course already now applicable for the database selection.

But overall, as said in the beginning, I am afraid not a too clear answer probably..

Best wishes,

Andreas