Using IPR in gaslift optimization #6653

totto82 · 2025-12-03T11:21:03Z

Start using IPR in gaslift optimization.
Gives a significant speed-up compared to the old code

totto82 · 2025-12-03T11:21:13Z

jenkins build this please

totto82 · 2025-12-04T08:03:36Z

jenkins build this failure_report please

totto82 · 2025-12-04T15:03:26Z

jenkins build this failure_report please

hakonhagland · 2025-12-04T15:54:37Z

Do we have any profiling results demonstrating the speedup?

totto82 · 2025-12-05T08:23:08Z

jenkins build this failure_report please

totto82 · 2025-12-05T08:32:39Z

jenkins build this failure_report please

totto82 · 2025-12-05T11:44:59Z

jenkins build this failure_report please

hakonhagland

I profiled this PR with tracy and compared with master for test case 4_GLIFT_MODEL5.DATA and 4 MPI ranks.

The results indicate a 15% overall speedup by using Inflow Performance Relationship (IPR) for gas lift optimization BHP calculations:

Metric	Master	PR #6653	Improvement
Total runtime	5.81 s	4.93 s	-15.1%
Gas lift BHP solve	562 ms	33 ms	-94%
MPI synchronization	725 ms	394 ms	-46%

It looks very good to me. I will approve this once the comments have been addressed.

hakonhagland · 2025-12-07T18:09:13Z

opm/simulators/wells/GasLiftSingleWell_impl.hpp

    return {-potentials[this->oil_pos_],
            -potentials[this->gas_pos_],
            -potentials[this->water_pos_],
+            bhp,


This now returns BasicRates which seems to indicate to the reader that well rates are returned, but BasicRates now includes bhp which is a pressure.

Should we rename to something like BasicWellState or LimitedWellState? However, that may be confused with the well state? Any other suggestions?

What about RatesAndBhp and LimitedRatesAndBhp ?

hakonhagland · 2025-12-07T18:10:33Z

opm/simulators/wells/GasLiftSingleWell_impl.hpp

 std::optional<typename GasLiftSingleWell<TypeTag>::Scalar>
 GasLiftSingleWell<TypeTag>::
-computeBhpAtThpLimit_(Scalar alq, bool debug_output) const
+computeBhpAtThpLimit_(Scalar alq, Scalar bhp, bool debug_output) const


It might be more readable if bhp is renamed current_bhp or similar? And add a comment above the method describing input parameters and purpose of method.

hakonhagland · 2025-12-07T18:11:54Z

opm/simulators/wells/GasLiftSingleWell_impl.hpp

    OPM_TIMEFUNCTION();
-    const auto& groupStateHelper = this->simulator_.problem().wellModel().groupStateHelper();
-    auto bhp_at_thp_limit = this->well_.computeBhpAtThpLimitProdWithAlq(
+    auto bhp_at_thp_limit = this->well_.computeBhpAtThpLimitProdWithAlqUsingIPR(


Maybe add a comment why we now are using computeBhpAtThpLimitProdWithAlqUsingIPR() instead of computeBhpAtThpLimitProdWithAlq() ?

hakonhagland · 2025-12-07T18:17:09Z

opm/simulators/wells/GasLiftSingleWellGeneric.cpp

 std::optional<typename GasLiftSingleWellGeneric<Scalar, IndexTraits>::LimitedRates>
 GasLiftSingleWellGeneric<Scalar, IndexTraits>::
-computeLimitedWellRatesWithALQ_(Scalar alq) const
+computeLimitedWellRatesWithALQ_(Scalar alq, Scalar bhp) const


Maybe change bhp to current_bhp? And add a comment above the method explaining input parameters and purpose of method.

hakonhagland · 2025-12-07T18:21:20Z

opm/simulators/wells/GasLiftSingleWellGeneric.cpp

                         rates.gas,
                         water_rate,
                         rates.water,
+                         rates.bhp,


The name LimitedRates seems to indicate to the reader that it contains limited rates. But now it also contains bhp which is a pressure.

hakonhagland · 2025-12-07T18:23:41Z

opm/simulators/wells/GasLiftSingleWellGeneric.hpp

            , water_is_limited{water_is_limited_}
            , alq{alq_}
            , alq_is_limited{alq_is_limited_}
+            , bhp(bhp_)


Use curly braces to be consistent with the above initializations

hakonhagland · 2025-12-07T18:26:02Z

opm/simulators/wells/GasLiftSingleWellGeneric.hpp

        }

-        Scalar oil, gas, water;
+        Scalar oil, gas, water, bhp;


Refer to earlier comment about the name BasicRates and including bhp as a member.

hakonhagland · 2025-12-07T18:29:20Z

opm/simulators/wells/WellInterface_impl.hpp

+    template<typename TypeTag>
+    std::optional<typename WellInterface<TypeTag>::Scalar>
+    WellInterface<TypeTag>::
+    computeBhpAtThpLimitProdWithAlqUsingIPR(const Simulator& simulator,


Maybe add a comment above this method about its purpose

hakonhagland · 2025-12-07T18:34:13Z

opm/simulators/wells/WellInterface_impl.hpp

+    {
+        OPM_TIMEFUNCTION();
+        WellStateType well_state_copy = well_state;
+        const auto& groupStateHelper = simulator.problem().wellModel().groupStateHelper();


In general, we should avoid passing around simulator and instead pass groupStateHelper if possible to avoid this kind of hidden dual state pattern where the groupStateHelper is not the same as simulator.problem().wellModel().groupStateHelper(), see: #6631. But I think this requires a larger refactorization, and is better done in a follow-up PR.

totto82 · 2025-12-08T22:40:25Z

jenkins build this failure_report please

akva2 · 2025-12-09T10:16:52Z

This was merged with warnings, https://ci.opm-project.org/job/opm-simulators/3044/gcc/folder.1225362760/

totto82 · 2025-12-09T10:54:15Z

This was merged with warnings

Sorry. I will fix asap.

totto82 force-pushed the fix_glift2 branch from 3f55a2c to 3b0e452 Compare December 4, 2025 08:03

totto82 mentioned this pull request Dec 4, 2025

WIP use IPR in gaslift optimization #6648

Closed

totto82 force-pushed the fix_glift2 branch from 3b0e452 to a60171f Compare December 4, 2025 15:02

totto82 changed the title ~~FOR TESTING: Fix glift2~~ Using IPR in gaslift optimization Dec 4, 2025

totto82 added the manual:irrelevant This PR is a minor fix and should not appear in the manual label Dec 4, 2025

totto82 requested a review from hakonhagland December 4, 2025 15:04

totto82 marked this pull request as ready for review December 4, 2025 15:04

Start using IPR for bhp calculations during gaslift optimization

c98f74a

totto82 force-pushed the fix_glift2 branch from a60171f to c98f74a Compare December 5, 2025 08:22

Fallback on old approach for trivial rates

da6a47e

totto82 mentioned this pull request Dec 5, 2025

Dont use BHP/THP from previous shut wells #6664

Closed

hakonhagland requested changes Dec 7, 2025

View reviewed changes

Clean-up based on review

f9d254a

hakonhagland approved these changes Dec 9, 2025

View reviewed changes

totto82 merged commit 630e4d3 into OPM:master Dec 9, 2025
2 checks passed

totto82 mentioned this pull request Dec 9, 2025

Use old bhp from thp computation for unconverged bhp solves #6669

Merged

Using IPR in gaslift optimization #6653

Using IPR in gaslift optimization #6653

Uh oh!

Conversation

totto82 commented Dec 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

totto82 commented Dec 3, 2025

Uh oh!

totto82 commented Dec 4, 2025

Uh oh!

totto82 commented Dec 4, 2025

Uh oh!

hakonhagland commented Dec 4, 2025

Uh oh!

totto82 commented Dec 5, 2025

Uh oh!

totto82 commented Dec 5, 2025

Uh oh!

totto82 commented Dec 5, 2025

Uh oh!

hakonhagland left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

totto82 commented Dec 8, 2025

Uh oh!

Uh oh!

akva2 commented Dec 9, 2025

Uh oh!

totto82 commented Dec 9, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

totto82 commented Dec 3, 2025 •

edited

Loading