Hi,

I've got two questions here. It's quite urgent. I would be very pleased if someone could help me (you can answer by e-mail or on the forum). Thanks.

Q1--If during your analysis, you find that that Reynolds number changes the flow from laminar to turbulent as the water flows around the cooling jacket circumference, say half way round; Then would you change the calculation so that the rest was performed with a turbulent flow model.? Or would you leave it to calculate the rest still using a laminar flow model.

Q2--Can CFX cope well with flow regimes that vary in nature. "What does it do" so to speak.?

i.e. A flow that changes from laminar to transition and then turbulent flow during the flow path.?

A laminar flow does not give any provision for the turbulent flow characteristics. The turbulent flow does. Assume that we are talking about the RANS based turbulence models. Look at the low Reynolds number k-epsilon for example, both turbulent and molecular diffusion terms are accounted for in the k and e equations.

If and when you get to the turbulent regimne, the molecular diffusion/viscosity term will disappers and the you will remain with the turbulence model(dominated by turbulent diffusion/viscosity). Conversely, the Reynolds stress term (u'u'_bar) that we need to model in the turbulent flow will fall of if the flow is laminar and you will remain with the mean velocities to deal with. I guess you are aware that the turbulent viscosity is a property of the flow.

I am not sure i understood your second question. But I believe that most of the flows we deal with have got regions of laminar and tubulent flows in the same domain. And CFX deals with all this.

regards Anne

Thanks Anne,

I asked on the Ansys helpdesk. Here is their answer:

"I think that the answer to 1 depends on what the purpose of the simulation is. But potentially, yes, it can be important to model the transition from laminar to turbulent flow (you will for example have different heat transfer behaviour on a wall depending on whether you have a laminar or turbulent case).

2. To model transition, we have a correlation based transition model in CFX-5 that essentially predicts a quantity called 'intermittency'. This intermittency will affect the turbulent production and destruction terms in the (SST, 2 equation) turbulence model. So you essentially end up with a laminar model where the intermittency is very small, and with a turbulent model where the intermittency is large."

Hi,

What you are talking about is modelling laminar to turbulent transition. As CFX support says, CFX has a transition model which may be able to model this. I have no experience with it. Be aware that modelling transition is a difficult thing to do accurately and is not recommended for beginners.

Note that the omega based turbulence models can reduce to laminar flow as omega gets very small. This means flow which contain laminar and turbulent sections can sometimes be modelled using an omega turbulence model. These models probably will not model transition properly, but if transition is caused by flow over a bluff body or some other obvious triggering feature this approach might work OK. If this approach is applicable to your flow then it is much simpler than using a transition model.

Just some ideas.

Glenn Horrocks

Hi,

the CFX-5 transition model is actually pretty accurate, it was validated against a large number of test cases, see for example the IGTI paper GT2004-53454. The most important thing is to have a good estimate for your inlet turbulence level. If you have no idea what it is for your device than you can always do a sensitivity study by varing the inlet turbulence intensity from a low value (say 0.2 percent) to a high value (say 6 percent). That way you can judge the effect of transition on your device.

hope this helps,

Robin Langtry