I am using transient simulation and it takes hours to simulate 9 frames. I am thinking of running the steady state 9 times at different end time to see if it takes less time. Are the results the same?
If the process is random for each of these steady state simulation, is there a way to set the seed?
Hi, thanks for using the forum!
Please always share the link to your project so we can provide the best possible help.
Is this a dynamic simulation or a non-linear static? What non-linearities are present? I am gonna guess that it is not equivalent to split the simulation into multiple times, and if it would be, I think the single run would be faster.
What random process so you mean? Can you please elaborate on this?
Most CFD simulations have a deterministic nature. This means that you will always obtain the same results regardless of the number of times you run the same case. Some models coupled to the base CFD model have a stochastic nature. For example, for RANS simulations, particle dispersion relies on a stochastic mechanism to produce a relative velocity vector considering the modelled turbulent kinetic energy. Of course, there are newer stochastic or semi-deterministic approaches to CFD but those are beyond my knowledge.
What you call frame from a transient simulation represents the numerical solution after a certain period of time (in seconds). Therefore, each frame represents the temporal evolution of your numerical solution. On the other hand, for a steady state solution, each frame represents the solution obtained so far after a certain number of iterations, considering a steady-state approach to your numerical problem. So frame after frame, what you are observing is a better “solution” to a steady (hopefully) state fluid flow problem .
Sorry about that. I mean the k-omega SST simulation. @jairogut is what you say still hold true for that simulation?
Link. It is about simulating gas through a straight pipeline.
Basically, if I run a k-omega simulation twice, would I get the same result? And are states at time t+k*epsilon (k = [1,2,3,...]) of a transient simulation equal to k solid state simulation at time t and t+k*epsilon?
The k-omega model as far as I remember does not have any stochastic terms. Therefore, if managed by a PIMPLE pressure velocity coupling will always produce the same results. Regarding your questions on “states”, I do not understand what you mean with “k solid state simualtion”. I’ve already answered what the saved frames mean for both transient and steady approaches.
I mean k solid state simulations, each has end time of k*epsilon, k in [1,2,3,...]