Modified well controls to match Intersect results #1202
Conversation
src/coreComponents/physicsSolvers/fluidFlow/wells/CompositionalMultiphaseWell.cpp
Outdated
Show resolved
Hide resolved
src/coreComponents/physicsSolvers/fluidFlow/benchmarks/Egg/dead_oil_egg.xml
Outdated
Show resolved
Hide resolved
src/coreComponents/physicsSolvers/fluidFlow/wells/CompositionalMultiphaseWell.cpp
Outdated
Show resolved
Hide resolved
src/coreComponents/physicsSolvers/fluidFlow/wells/WellSolverBase.cpp
Outdated
Show resolved
Hide resolved
| @@ -303,6 +303,9 @@ class WellSolverBase : public SolverBase | |||
| /// the number of Degrees of Freedom per reservoir element | |||
| localIndex m_numDofPerResElement; | |||
|
|
|||
| /// copy of the time step size saved in this class for residual normalization | |||
| real64 m_currentDt; | |||
There was a problem hiding this comment.
Not a task for this PR, but perhaps we should be passing dt into CalculateResidualNorm()
There was a problem hiding this comment.
Yes I agree that it would be useful to have the dt passed as an argument of calculateResidualNorm here and maybe for others as well. I can take care of this in another PR.
Yes you are right the surface rate option must be used and checked in the integrated tests. In the latest commit it is now used for single-phase flow in |
| // above/below the target pressure depending on well type | ||
| pressureControl = ( wellType == WellControls::Type::PRODUCER ) | ||
| ? 0.5 * pres | ||
| : 2.0 * pres; | ||
| } |
There was a problem hiding this comment.
So isnt' the targetBHP defined even for rate constrained wells as (e.g., as an upper limit for injectors) Is this for the case in which it is not specified by the user? Maybe I am confused by what the reference pressure is. Or is this just like an initial guess?
There was a problem hiding this comment.
Yes you are right rate-constrained wells have a targetBHP as well, and the code will stop if the targetBHP is not specified (for both single-phase and compositional). What is done here is just a not-very-good way to initialize the pressure in rate-constrained wells at the beginning of the simulation, before the first iteration.
A better way would be to set the initial pressure using targetBHP (for instance, slightly below targetBHP for injectors). To be honest, I did not use the targetBHP to initialize the injectors that because I sometimes (often ;) ) use a very large targetBHP to keep the wells at rate constraint during the full simulation, even if it is not realistic (and I don't want this large targetBHP to perturb the initialization...). I am going to add this explanation to the inline comment just before the line you referenced.
Ultimately, to have a really good well initialization and to improve the nonlinear convergence, it would be helpful to guess the well pressure and rate by solving small local well systems (with fixed reservoir values), but there is a little bit of work to do to allow that.
francoishamon
added
ci: run CUDA builds
This PR implements some changes in the well controls to match a simulation result generated with Intersect:
With these changes I was able to match a two-phase DO simulation result with BHP and surface oil rate controls obtained with Intersect. If at some point we need to deal with more complex constraints, more work will be required.
This PR also changes the scaling of the control/pressure difference equations because the previous scaling was creating issues with fgmres + MGR.
Related to https://github.com/GEOSX/integratedTests/pull/114