SimScale CAE Forum

Time dependant momentum source for non inertial reference frame


#1

Hello,

Here is my situation:
I am currently trying to find numerically the added mass coefficients on a completely immersed ellispoid, as a verification test for future studies with more complex geometries. Thoeretical results can be obtained considering a potential flow, hence I have made the hypothsis of incompressible laminar flow, with a slip wall condition on the solid as there I could not find any potential flow solver.
As a first step, I am trying to solve the case for a constant linear acceleration, so I went for a steady case study, with a time dependant velocity inlet. However, such a referential can not be considered as inertial, and a source term has to be added in the conservation of momentum equation ( more details in this paper https://www.diva-portal.org/smash/get/diva2:916575/FULLTEXT01.pdf).

The problem is:
It seems I can not define a time dependant speed accounting for momentum source. I have also tried to consider a transient study with solid linear body motion but I am facing the same problem.

Is my approach valid, and is there any way to solve this problem ?

here is a link to the project : https://www.simscale.com/projects/semilien/added_mass/

Thanks


#2

Hi @semilien!

We do support momentum sources but without any time-dependancy! I can have a look at the paper later this evening and tell you if there might be a workaround for this. In the meantime I am tagging my colleagues, the @PowerUsers_CFD, to jump in in the meantime.

Cheers,

Jousef


#3

Hi everyone,

Not exactly familiar with this type of simulation, but if you want to implement a momentum source in space that is time dependent, don’t think that is possible yet. OpenFOAM offline may be possible, but not sure if you want to delve into that.

Regarding your BCs, is the flow not time dependent? Did your time dependent inlet work?

Cheers.

Regards,
Barry


#5

Hello,

Thanks @Get_Barried for the suggestion, even though the resulting forces are not supposed to be time dependent, the flow is, and a transient approach improved the results in this way. The time dependant inlet works pretty well in these conditions.

There is a large difference, in the pressure resulting forces between steady and unsteady case, suggesting that unsteady pressure is not negligible, which is what I am expecting. In order to be the closest as possible to potential flows, I separated the pressure in steady and unsteady parts, and tried to consider only the unsteady part for the forces calculation. Steady pressure law is proportionnal to the square of velocity according to Bernoulli and has been verified through computation. However the forces given by the unsteady parts are about 35% lower than the theoretical estimations. Do I make any mistakes by separating pressure ?

I investigated further on the acceleration proportionnal source term and it seems it can be neglected when the ratio between accelration and square velocity is high enough. Nevertheless, I will probably need it in the future for specific cases, such as rotationnal motions, thus I might consider using offline openFOAM.

I still get high pressure residuals, so I am going to dwell on solver configuration for the moment, as I think the mesh is rather correct.
edit : I noticed the pressure gradient is close to constant, would it be effective to assign pressure initial conditions giving such results (or is it done by the potential foam initialisation), and if so, is there a way I can do it ?

Cheers

Emilien


#6

Hi @semilien,

I am unsure unfortunately on this area. Lets see if the other @PowerUsers_CFD have some input or are more familiar with this type of simulation.

Setting up openFOAM offline is quite tedious and rather unfriendly even if it works so if you need some help on setup I can guide you there, assuming its not already setup.

Agree, your mesh is fine. No issues there so I wouldn’t consider it a significant contributing factor to the result deviation you are experiencing. Assigning an initial pressure condition should not affect your results greatly but I suggest you can try. Potentialfoam does initialize the solution but thats about it and sometimes it even causes the sim to be unstable. I believe you can assign this under initial conditions in the simulation tree.

Apologies for the lack of insight into your issue.

Cheers.

Regards,
Barry


#7

Hi @Get_Barried,

I reviewed my calculations and it seems I made a mistake on the theoretical estimations. I also improved the pressure residual by setting relative tolerance on all pressure solvers to 0, and divided the absolute one by 10.
I couldn’t find where to apply specific intial pressure conditions, however potential foam optimisation (which I was already using) seems to be effective. I am still getting a 0.08 pressure residual , which is better than the previous results, nonetheless not satisfying.

Results are in fact only arround 6.5% lower than theory, which is way more acceptable. Do you think the poor residual might explain this difference, and if so, would setting tighter convergence criteria help to reduce this gap ?

I am going to investigate the momentum source term contribution, I will keep this topic updated if I get some significant improvements. (edit : the addimensionnal coefficient corresponding to the source term with these conditions is 0.05, suggesting that it might explain a most of the differencies).

Thank you for the help

Regards,
Emilien