Rapidly decreasing deltaT for interDyMFoam
 

Hey,

So I'm currently running a parallel (although this also happens in serial) case using interDyMFoam to dynamically refine my mesh. Ignoring the likely issues with restriction on where I can decompose my mesh to avoid horrible load balancing, the case happily runs for a few iterations, including mesh refinement.

After a little while however (around 170s clocktime), the deltaT value starts dropping alarmingly (before this point it is around 10^-6, which is expected, but drops approximately an order of magnitude each iteration beyond this point), and then timePrecision is increased to compensate for presumably the data being otherwise labelled at the same time.

I am using the adjustableRunTime setup with Courant values limiting the time steps, but the Courant values being output are not equal (or even particularly close to) the limits I have set (0.75 for maxCo, 3 for maxAlphaCo, with maxDeltaT at 1).

If anyone can suggest why my case suddenly goes a little crazy, I would appreciate the help. I have attached the output of the data run and have listed below a smaller section of the output. The first file has the relevant sections (I think) in, but I also attached the second half in case the exit messages are important.

Attachment 24827
Attachment 24828

Yours,

Christian Butcher


Edit: The case is decomposed for 4 processors, in case that is relevant for the fact that some of the Courant numbers are in fact becoming higher, whilst others sit around 0.2. I assume this might be different processor regions. They are split using simple (2 2 1) and the mesh is cylindrical with the z axis as the axis of symmetry. The refinement is occurring at one end of the cylinder, for the higher z values (and should be at least approximately axisymmetric)

Hi Christian,

I am currently experiencing similar problems with interPhaseChangeFoam in 2.2.2. Have you found the solution to your problem and want to share?

Peace,

Artur

Dear Artur,

I'm afraid I don't remember now - I stopped working on that simulation at the end of August last year, and since then have been trying to set up and run other simulations/set-ups.

Some of my current work is using interDyMFoam, or similar, and these do not suffer from the same problem (OF-2.2.2, 2.3.0) so I can't really guess what might be the problem.

Sorry for not having a useful response but thought this might be better than leaving you waiting here indefinitely.

Good luck and best wishes,

Christian

Dear All,

I am using OpenFOAM version 2.2.0.

I was able to run the tutorial "floating Object" which is in 3D in interDyMFoam without any issues.

I am trying to run the same case in 2D.

The following changes have been made, to run the case in 2D

1.In the blockMeshDict file the y coordinates range is changed.
In 3D y axis range was from 0 to 1 and for 2D y axis range is changed as 0 to 0.1.
For the front and back faces,created patch frontandback as empty

2.In the topoSetDict file:
In 3D, box (0.35 0.35 0.44) (0.65 0.65 0.56) was changed in 2D as box (0.35 0.0 0.44) (0.65 0.1 0.56)

3.In the pointDisplacement file:

In 3D

floatingObject
{
type sixDoFRigidBodyDisplacement;
centreOfMass (0.5 0.5 0.5);
momentOfInertia (0.08622222 0.08622222 0.144);
mass 9.6;
rhoInf 1; // needed only for solvers solving for kinematic pressure
report on;
value uniform (0 0 0);
}

In 2D was changed as:

floatingObject
{
type sixDoFRigidBodyDisplacement;
centreOfMass (0.5 0.05 0.5);
momentOfInertia (0.01952 0.08352 0.08);
mass 9.6;
rhoInf 1; // needed only for solvers solving for kinematic pressure
report on;
value uniform (0 0 0);
}
NOTE: Mass not changed

With a change in mass,the changes are as follows:

floatingObject
{
type sixDoFRigidBodyDisplacement;
centreOfMass (0.5 0.05 0.5);
momentOfInertia (0.08133 0.0348 0.033);
mass 4.0;
rhoInf 1; // needed only for solvers solving for kinematic pressure
report on;
value uniform (0 0 0);
}


4.For all the boundary condition files in the folder 0,the patch frontandback is specified as empty.

I run the case and and the run blows with delta T decreasing rapidly.

--> FOAM Warning :
From function Time::operator++()
in file db/Time/Time.C at line 1029
Increased the timePrecision from 12 to 13 to distinguish between timeNames at time 0.0444347581734
GAMG: Solving for p_rgh, Initial residual = 0.00606079623689, Final residual = 5.1801156616e-05, No Iterations 2
time step continuity errors : sum local = 4.69383557933e-05, global = 4.4792886392e-07, cumulative = -0.000506180874993
GAMGPCG: Solving for p_rgh, Initial residual = 1.52101895333e-05, Final residual = 6.35747944249e-09, No Iterations 3
time step continuity errors : sum local = 0.000245342044169, global = -1.53144705592e-06, cumulative = -0.000507712322048
smoothSolver: Solving for epsilon, Initial residual = 0.00132531232957, Final residual = 9.99802871216e-07, No Iterations 41
bounding epsilon, min: -1.04792539095e+26 max: 8.15921076625e+28 average: 3.4517817453e+25
smoothSolver: Solving for k, Initial residual = 0.910004972962, Final residual = 0.0190942329344, No Iterations 1000
ExecutionTime = 120.48 s ClockTime = 121 s

Interface Courant Number mean: 0.00473017730727 max: 14.3965768911
Courant Number mean: 2.42817054949 max: 10937.5681779
deltaT = 6.09445041823e-18
Time = 0.04443475817338


Kindly can somebody please let me know where I am making a mistake and what I could do, to run the case in 2D successfully.Thank you.
Regards,
Gautami.