big difference (3.6x) between ecoinvent and openLCA results

[gebbissimo1]

Hello openLCA team,

I am not 100% sure, but believe to have found a systematic error in the openLCA LCI and LCIA calculation for some ecoinvent processes (electricity processes):

Process: "electricity, high voltage, production GR, at grid" (, whose output is 3.6kWh electricity... amongst others)
Product system with only this 1 process
Reference target amount: 1kWh

=> CED fossil in ecoinvent web database v2.2: 15.543 MJ-Eq
=> CED fossil with openLCA: 4.31745 MJ-Eq (=1/3.6 x the result of openLCA)

I double checked that the ecoinvent database also has 1kWh and not 3.6kWh as target amount, but it has 1kWh. Also the report "06_XVI_Strommix.pdf" names a similar number (p121) and the payback time of PV modules noted in another ecoinvent report makes more sense assuming that thís openLCA result is wrong. Thus, I can currently only conclude that there's some error in the calculation for this particular process.
I couldn't nail down the error though. If I create a test process with an output of 3.6kWh electricity and some arbitrary inputs, the LCI results are correct. Maybe some confusion with the ecoinvent form?

Help would be highly appreciated, since I regard this as a serious problem. Maybe it also explains other small deviations between openLCA LCIA results and ecoinvent database? (http://openlca.org/forum/viewtopic.php?f=11&t=16179)
If I create a process with just 1kWh of electricity as input and calculate the results, the LCI and LCIA are also 3.6x times less than what they should be according to the ecoinvent web database. This fact makes me wonder, whether every ecoinvent process including electricity has this problem.

Best
Christian


//edit
P.s.: As a more general idea / suggestion: I am not sure in how far calculation results of openLCA are validated for each new version. Would it be possible to implement a validation process for each new version, in which a series of product systems are being evaluated? By evaluating I mean checking the LCI (and LCIA) calculation results for correctness. I would gladly contribute as a (free of charge) tester for some hours per month to make sure the calculations are reliable. Moreover I hope/believe that I'm not the only thinking this way.

//edit2
P.P.s.: If it is of any help, the process "electricity, at Mini CHP plant, allocation exergy (CH)" is affected, too

[aciroth]

Thanks a lot Christian for this offer to contribute as a tester - this is very much appreciated. I will contact you. Regarding the bug with the unit conversion, this should be fixed now in the 1.3 beta versions.
Best wishes,
Andreas

- we of course fully support that correct and reliable results are more important than features!

[gebbissimo1]

Dear Andreas,

I found another small error. I guess it's not a bug in the calculation algorithm, rather in parsing the ecoinvent database:

In the process "operation, passenger car, diesel, EURO5 (CH)" the output in openLCA is 1000km, while according to the ecoinvent v22 database it should be 1km. This leads to the result, that the LCIA factors of this process are diminshed by a factor of 1000.
Fortunately (?) in the process "transport, passenger car, diesel, EURO5 (CH)", whose input is "operation, passenger car, diesel, EURO5 (CH)" the error cancels out. The amount of "operation, passenger car, diesel, EURO5 (CH)" is 628.93km where it should be 0.62893km.

Stil, it makes me a bit nervous having to rely on the fact that this error is canceled out each time the flow is used...

Best wishes
Christian

//edit
I found a similar error in the process "operation, passenger car, electric, LiMn2O4". The amount of electricity input in openLCA is 0.72kWh, while in the ecoinvent database it is 0.2kWh. Hope that helps!