Hello everyone!
I am rather new to the open source community and would look for some help in modelling my problem in the correct way.
My problem is about cooling a particle flow with air (cross flow cooling). Therefore I had a look at the cfdemSolverPisoScalar. But if I understood right, it is considering the temperature field rather in the way of transporting a species than in a thermodynamic manner. According to the documentation this means that the change in temperature is neither infulencing the density of the fluid nor the flow field itself.
But since the fluid in my problem is heated up quite a lot, the density will decrease significantly which will speed up the fluid.
I have worked out the following points on this topic:
(1) Am I misunderstanding the documentation and/or the code and the cfdemSolverPisoScalar is perfectly capable of doing what I am looking for?
(2) Has somebody ever faced a similar problem and tried to implement an equation of state in cfdemSolverPisoScalar to take the change in the flow field into account? In my case I think the incompressible ideal gas law (like in Fluent) would be suitable since the difference between inlet and outlet pressure is small compared to the temperature gradient. But if I would implement such a simple equation of state, this would not lead to the desired influence of the flow field, since the density is not included in the PISO solver (incompressible solver)?! The only influence I could make out is the source term of the momentum equation since it is divied by the density.
(3) Probably the best solution I could think of so far would be to create a new cfdemSolver in combination with a compressible OpenFOAM solver, e.g. rhoPimpleFoam. With this option I could make use of several equation of states implemented in OpenFOAM and also the energy equation would be already included. Is this a reasonable option compared to modifying the cfdemSolverPisoScalar to fit my needs? Or are there maybe some other possibilities I have overlooked. Because one thing I am afraid of is that my simulation would become quite unstable in case of coupling to a compressible solver...
(4) Since I have a continous process in my mind it would make sense to couple to a steady state OpenFOAM solver (e.g. rhoSimpleFoam). Has somebody got a hint how to achieve a steady state in LIGGGHTS? I was thinking about conducting just a simulation of the particulate flow until the transient problem reaches some kind of steady state. Then I could take the field of particles (e.g. one representing particle distribution, like a snapshot, or an average of several representing distributions) and couple just the energy equation to the flow field (I expect the momentum coupling to be very weak anyway so I am not so interested in that). This sounds like a lot of change in the LIGGGHTS code to me or even to
Sorry for the very long text, I hope my explanations didn't get too messy! Since I am rather new to the open source approach I would appreciate any hints, suggestions, experiences or whatsoever (regarding modifications of the code or choice of the model)!!
Best regards,
Harald
*Edit: I have just realized that this is maybe rather a Developer topic, so please feel free to move this post!
Riccardo Maione | Thu, 10/22/2015 - 15:04
Hello Harald,
Hello Harald,
1) No you have to modify cfdemSolverPiso
2) In order to do that you have to change quite a bit the solver. I did that with the reactingFoam solver, and obtained good agreement with some cases in the literature and with the examples in the the cfdem folder, so it can be done.
4) I am not sure, you have to check this, but if you want to disable the momentum coupling I think you can simplu delete these 3 lines in the cfdemSolverPisoScalar.C
Ksl.internalField() = particleCloud.momCoupleM(0).impMomSource();
particleCloud.smoothingM().smoothen(Ksl);
Ksl.correctBoundaryConditions();
To avoid the computation of particles movement, you can avoid to integrate, i.e. delete the command fix nve. however I don't know if the convective heat exchange phase would be computed after.
Hoping this could be helpfull
Riccardo
hunger | Fri, 10/23/2015 - 09:42
Hey Ricardo!
Hey Ricardo!
Thanks for your input!
ad 4)
(a) Haven't thought about this possibility to deactivate the integrator yet! If the fix nve command is necessary to compute the energy equation, another possibility I can think of is just to apply a stopper when a certain particle distribution is present to avoid moving particles.
(b) To disable the momentum coupling the 'forceModel_noDrag' should be enough I guess?! But of course deleting the appropriate lines in the code would do the job ;)
(c) Another idea I had was not really to couple LIGGGHTS and OpenFOAM interactively, but to conduct a LIGGGHTS simulation and use CFDEM to extract a porosity field with the voidfraction function. In a simular manner I would like to extract a field of source terms for the energy equation (maybe with some kind of particle size/mass distribution on the CFD grid in combination with a specific heat capacity of the solids to calculate the Nusselt correlation accordint to [Li, Mason]). I would like to hand this information on to a steady-state OpenFOAM solver (e.g. rhoPorousSimpleFoam which is enhanced with an energy equation). Still have to dig deeper into the source code of both CFDEM and OpenFOAM to find out if this is possible.
mardussa | Wed, 08/29/2018 - 07:42
Hi Harald,
Hi Harald,
This is an old thread but perhaps you will see this.
Did you get any further with implementing temperature dependent properties in cfdemSolverPiso? I am interested in doing something similar because I need the temperature dependent density for use with another code. If you have any suggestions one how to get this working I would be very grateful for some advice.
Kind regards,
Rob