I discovered that temporalis is giving wrong results and the impacts for some yrs are incorrectly omitted. I tried to play around but could not discover the reason for this. I tried also on other machines, changing the method and increasing cutoff and calculations number but nothing.
I reproduced the error in the file attached. Jjust extract the folder and run the notebook in a project with ecoinvent 2.2 installed (changing the porj name in the notebook) and will do everytng, register the db and run calculations) where the calculations are first done with temporalis and then year by year. They should return the same results but some years are just neglected from temporalis (e.g. 2011,12,13 with GWP and 2013 with FDP). I checked also the raw and these yrs are not there but really do not understand where put the hands...it is so weird