Dear users,

I would like to know your opinion about a case if this should be a trasient or a steady state simultion.

I have to simulate a offshore platform with 4 decks and a fire in a second floor. The wind is blowing from East with a constant velocity. What i am intersted in the smoke modelling.

All the guys from here says that clearly it is a transient simulation, but i still think that this is a steady state.

Can anyone help me with this,

Joana

Hi,

I have no experience with fire or smoke modelling but maybe I can give you some hints on the problem.

I assume that you are neither going to model the spread of the fire nor the fire itself. You seem to be interested in modelling something like a fixed source of smoke. Is this true?

In this case I would agree that you could use steady state modelling to obtain the wind field. After the wind field is converged to steady state you may fix it in a second, time dependent, simulation that does only model the time dependent spread of the smoke in the fixed wind field. I cannot say in how far this would be realistic.

One would probabely have to model quite a number of different ambient wind directions and speeds, according to the wind conditions on site.

Hope I could help you a bit

regards

Yeah, although the wind field upstream might be steady, the wind flow around the structure will definately be unsteady (transient), so therefore the smoke path will also be unsteady.

In real life there are infact very few flows which are truely steady in nature, perhaps creeping flows or very low reynolds number flows.

Regards James

It all depends on what you want to get out of the simulations! What are you looking for and how will the data be used? Physically, the problem is undoubtedly unsteady. But are you interested in, say, the puffing frequency of smoke, the large scale dynamics of the flow, or just the time averaged envelope of the flow. If you were to simulate this problem as unsteady for long enough of a time, and were to then time average the results, you'd see how different the two solutions are (they are _very_ different even in the "simple" case of a plume with no wall interaction, etc.)

On the other hand, the shear size of the problem you claim to want to solve makes me wonder how serious a computation this will eventually become, irrespective of whether you assume steady flow or not!!

Adrin Gharakhani

Wind flow is always unsteady, this is of course true, as it is highly turbulent. I thought, however, that we are referring to the Reynolds-averaged flow with use of some turbulence model like k-e or so. Clearly, some kind of large-eddy modelling would be perfect in this case, but is not feasible in an engineering context so far, as I know.

It is not really clear to me actually when one must expect a reynolds-averaged flow to be unsteady (with fixed geometry and boundary conditions). Probabely in cases of strong backflow behind abstacles combined with some unstable numerical scheme. But this is something I never actually got my head around.

regards

Tim

It's not the "unsteadiness" of wind flow that should be of major concern here (though the flow is unsteady); it's the flow instabilities due to density variation (fire smoke, etc.) that is more of an issue here ... Note that in this case, even a laminar flow can/does become unsteady; turbulence only adds further complication ...

Adrin Gharakhani

i just thought of a simple case which refers to a close a domain with a fixed heat sourse in it and adiabatic boundaries. it is obviously a unsteady case, for there is no outlet for the heat.But if one of its boundary conditions is set to a fixed temperature, it could change to a steady problem. So if the domain to compute in the smoke modeling problem is rather big that all the smoke generated is restricted in it, the problem is unsteady with no doubt.But if the domain is set with some outlet for the smoke and heat generated by the fire to escape away, it is possible to achieve a steady state if the effect of turbulence is not taken into account.

I am talking about physics not numerics (and choice of computational domain size).

Let me be a bit more specific. If you solve the plume rise as an unsteady problem, you'll see unsteady large eddy motion - think of it as a train of (interacting) vortex rings, etc. So, in an unsteady simulation you'll see puffing of lumps of eddies at certain frequencies. Now when you average out this unsteady flow, it turns out that you see no lumps of large eddies - you just see a "straight cylinder" of plume rising. The envelopes of the two flows look very different. BTW, this is true even in the laminar flow case.

But, again, it all depends on what it is one wants to get out of a simulation.

Adrin Gharakhani