use analytic weights for cprw by default - except when solvent is enabled#6925
use analytic weights for cprw by default - except when solvent is enabled#6925steink wants to merge 2 commits intoOPM:masterfrom
Conversation
|
jenkins build this failure_report please |
|
After first look:
|
atgeirr
left a comment
There was a problem hiding this comment.
The code looks good to me, one minor change requested.
I have not looked at test failures, since you are already taking a shot at that.
| FlowLinearSolverParameters() { reset(); } | ||
|
|
||
| void init(bool cprRequestedInDataFile); | ||
| void init(bool cprRequestedInDataFile, bool cprwUseAnalyticWeights = true); |
There was a problem hiding this comment.
Suggest not adding the default argument, to avoid mistakes.
There was a problem hiding this comment.
This init-function is also called from ISTLSolverGPUISTL.hpp and BlackoilModelNldd.hpp. Would you rather prefer that both those also checks for the solvent? I kind of assumed it was not relevant for them, but that's maybe not a valid assumption indefinitely.
There was a problem hiding this comment.
This
init-function is also called fromISTLSolverGPUISTL.hppandBlackoilModelNldd.hpp. Would you rather prefer that both those also checks for the solvent? I kind of assumed it was not relevant for them, but that's maybe not a valid assumption indefinitely.
For nldd it is relevant, and longer term also the gpu code. I think it is a good example of why the default argument is dangerous, because even if assumptions about it are correct today they might not be in the future.
8bb1283 to
606d0cd
Compare
…L.hpp/BlackoilModelNldd.hpp
Redo #6840, but don't allow using analytic weights when solvent is enabled
Analytic weights are computed for oil/water/gas (and no other) and in a case where a well only contains solvent (e.g., injector with solvent-fraction 1), this leads to singular well system