Hi CFDEM Developers,
I am working on the coal combustion in OF, I am running a case with around 1500 coal parcels using four way coupling present in OF, But it seems to me it is taking too much. Using 8 processors, it took me a couple of days to simulate the case for 0.5 seconds and it is still running .
I once tried to couple the coal chemistry foam solver with your interface (CFDEM), i come to know some of the complications. I have few questions regarding this.
Is it possible to couple this LIGGHTS/CFDEM with some reacting solver especially with compressible one? As I see in CFDEM, you are using incompressible solver?
I will be happy to implement it and would like to share the comparative results in OF (4-way coupling) and CFDEM approach.
So far I have seen your 4-way coupling results are also for incompressible solver am I right?
Looking forward for your response on this issue.
Regards,
Naveed
cgoniva | Thu, 08/04/2011 - 12:21
Hi Naveed, You are right,
Hi Naveed,
You are right, that the coupling is currently implemented for incompressible flows.
From technical point of view it should easily be possible to plug it into an incompressible solver, but you might violate some of the assumptions inside the models.
It might be best to approach the problem step by step, i.e. first adopt the coupled momentum and mass balance for incompressible flow. Do the validation. In a second step, reactions might be implemented.
Do you know any good literature references and validation examples on CFD-DEM for compressible flows? We could join forces on that topic if you like.
Your described performance seems very bad - did you include reactions there as well (those usually need extremely fine timesteps).
Cheers,
Chris
niqbal | Thu, 08/04/2011 - 13:58
Hi Chris,
Hi Chris,
Thanks for your prompt reply, surely I would like to collaborate on this topic.
Do you know any good literature references and validation examples on CFD-DEM for compressible flows?
Regarding the literature references, I can refer some studies done on the coal gasification where I am working right now.
An experimental study on gassification of Colombian coal in fluidised bed (A. Ocampo et all 2003) is the base study for my work. Which is then numerically simulated by Yu et all (2007), But he is using bubbling approach by the kinetic theory of granular flow (KTGF).
Your described performance seems very bad - did you include reactions there as well (those usually need extremely fine timesteps).....
You are right the performance is very bad that is why I am thinking to go for CFDEM as I think the particle collision and tracking is much efficient in LIGGGHTS, rest of the routine will take the same time on both frames of work environment. Reactions are also inculded and the timestep is small enough to capture the chemistry details (order of 10^-4 to 10^-5).
Regards,
Naveed