Learning Forum

[vkt1990]

I am doing my project on odor source localization. I am doing the simulation using ansys fluent 19.1 software to understand the behavior of CO2 gas.
1st I performed simulation in steady state and then in the transient state.
but I am not able to understand the reason why flow behavior in both cases is different. I want to understand it mathematically. I am attaching the images of both. 1st image of steady and second of transient 

[Rob]

Looking at those images, how long was the transient model run for? 

Steady state means we don't account for time, so in simple terms it's the result when time reaches infinity. The transient solver solves each time step, so your result will have reached a finite time: in your case possibly not long enough to reach the equilibrium state. 

[vkt1990]

thanks for your reply... actually I am using ANSYS software first time and currently, my aim is to trace the trajectory of the plume of different gases in the different environment as well as in different boundary conditions.
so I want to know am I going in the right direction?

[Rob]

As ANSYS staff I'm limited in scope when it comes to offering advice. 

In your case you need to determine what goal you wish to achieve: do you want to know how far the plume goes or how long it takes to develop? The former is steady state, the latter is transient. Now consider that a steady solution takes a few hundred to a few thousand iterations to converge and has one answer: a transient solution takes thousands of timesteps each of 15-20 iterations so will take significantly longer to run. 

The successful application of simulation is as much down to planning the work as it is to driving the software. Once you know the project goals, you can plan the modelling strategy and then do it. 

[Raef.Kobeissi]

I would also add that certain flows cannot be resolves as steady state problems. Some problems are transient in nature, you need to consider that. You can obviously perform a steady state simulation on them but sometimes that would give you in an inaccurate solution. For your case, I would go for a transient simulation.

[vkt1990]

thanks for your valuable reply...exactly I want to know how far the plume goes.
I want to trace the path of the plume till velocity of the plume reduces to zero. so that I can understand the behavior of the plume.

[Rob]

OK, I'd go with steady state then. However, depending on the plume behaviour you could find convergence isn't ideal as the flow may have transient features as Raef mentions. Note, I'm using the software to get a "good enough" answer: you need to decide to what level you need to defend your work, and (possibly more importantly) how long you've got.  A velocity of zero probably isn't easily found, plot the axial velocity along the plume centre and see if literature has any half-life type correlations. 

You are going to need to be very careful with the mesh and choice of turbulence model, so there's quite a lot of the manual etc to read up on. 

[DrAmine]

All suggestions provided here are very valuable and help you understand the differences between steady-state and transient behavior.

But I am just curious to know a thing: Which numerical method have you used in your steady-state solution? Coupled with Pseudo-Transient or just SIMPLE based approach?

[vkt1990]

SIMPLE based approach

[DrAmine]

Okay. Thanks!

[Benoit85]

Hello,

I also have general question regarding steady vs transient flows.

If one would try to simulate a physically non-steady problem with a steady-state solver, should it then always be possible to find a (mathematically) converged solution? Even though this solution does not correspond to the physical reality? -OR- Is it possible that the solution would never really converge (regardless mesh quality and solver parameters etc.), meaning there IS NO steady mathematical solution either?

For example: I'm simulating the flow field in a hybrid rocket engine. I started with a non-reactive flow using the species transport model. Sovler: steady-state. Turbulence: just k-epsilon to start with. The solution however does not really converge: a shock wave in the nozzle (due to over-expansion of nozzle) keeps moving a few mm up and down the nozzle. The residuals are fluctuating 2 orders of magnitude. PS: However, as soon as I activated the chemcial reactions, there was enough pressure and the shock wave left the nozzle, and the solution converges properly.

Thanks for any advice

[Rob]

Depends on your definition of converged. The solution may fluctuate slightly between many steady state solutions but the change may be small enough to ignore. You then need to decide if the result is good enough for your purpose, or whether you need to switch to transient.