search

equinor / fmu-ensemble

Python objectification of reservoir model ensembles left on disk by ERT.
GNU General Public License v3.0
12 stars 19 forks source link

Mismatch computation of rate vectors incorrect #257

Open berland opened 2 weeks ago

berland commented 2 weeks ago

Rate vectors in Eclipse are valid in time intervals, not at points in time. Eclipse writes the values of these rates at the report step at the end of the interval.

When computing mismatches for a rate vector at a specific date, users usually assume the opposite, that a rate that is set using WCONHIST for a specific date can be matched at that date (thus assuming that Eclipe writes the value of the interval to the start of the interval).

A test illustrating this is in https://github.com/equinor/fmu-ensemble/pull/256/files

asnyv commented 2 weeks ago

Guess we are then back to the everlasting question of "what is a rate vector". E.g. matching on a flowing pressure like BHP would have the same issue as what you describe above, even though it is not a rate of course, while a reservoir pressure would in many cases be more natural to interpolate between the time steps.

But utilizing the "is_rate" metadata is probably better than nothing at least.