3
$\begingroup$

I am a beginner in CFD having written few codes for laminar flow cases using SIMPLE and some other explicit solvers. Now, I want to use my solvers and some other models to solve for the turbulent flow cases. In this regard, problem of decaying turbulence got my attention.

Since I absolutely do not know about how to make the transition from laminar to turbulent, how to write a code for it. I need some basic guidelines, papers and literature that can help me to write a code for this particular problem using the Navier-Stokes equations. And also, understand the basic concepts of the same. Thanks!

$\endgroup$
  • $\begingroup$ It seems likely that you're interested in turbulence models rather than full direct numerical simulation (DNS). That might be worth adding to your question since it really affects the answer. $\endgroup$ – Doug Lipinski Apr 13 '16 at 16:32
  • $\begingroup$ I completely agree, you need to distinguish whether you're interested in looking at turbulence modeling or DNS. Papers by Kim and Moin are a good starting point for DNS. $\endgroup$ – Charles Apr 15 '16 at 17:48
3
$\begingroup$

The first thing to keep in mind is that there is no one turbulence model that works well in all situations. You need to choose the right model for the right situation. See this link for reference: https://engineering.stackexchange.com/questions/336/which-turbulence-models-are-suitable-for-cfd-analysis-on-a-streamlined-vehicle-b/344#344

There are many types of models, all with varying degrees of accuracy and computational cost. Generally speaking: the more accurate the model is, the more computational cost it incurs. So you need to consider very carefully how much computational expense you can afford versus how much accuracy you really need.

With that said, it is not trivial to convert laminar code to one that handles turbulence. Depending on the model, you may need up to 7 additional variables and equations, and heuristics to specify these new values at the boundaries.

Depending on whether you're considering compressible or incompressible flow, you may need to modify your momentum and energy equations with additional terms which include these new variables, along with varying expressions for the turbulent viscosity and turbulent Prandtl number.

My recommendation: have a look at OpenFoam, an open source cfd solver with lots of turbulence models already implemented. You can get a good idea of how to modify your code from this (assuming your solver uses finite volume method).

Admittedly, OpenFoam has a steep learning curve and may take a while to master navigating it. But I think this is your best bet when it comes to learning implementation of generalized cfd code.

$\endgroup$
  • $\begingroup$ Thank you so much for the detailed answer. My laminar code uses a Finite difference framework for that matter. I am looking for an unsteady turbulent flow case, which is not very complicated and could be considered as a starting step for someone who wants to work with turbulence. Based on that, as you suggested, I could go on for some model which could give me reasonable accuracy in reasonable time. For unsteady laminar flow, my test cases were Taylor-Green decaying vortex and flow over suddenly moving plate. Now, what could be the potential test cases for turbulent flow that have benchmarks? $\endgroup$ – Tanmay Agrawal Apr 13 '16 at 2:29
  • $\begingroup$ Turbulence is well understood and there are known profiles for shear layers and jets. Also, these setups are not affected by walls which come with their own difficulties. $\endgroup$ – Jan Apr 13 '16 at 18:48
  • $\begingroup$ @jan Do you have any references to recent work that claim turbulence is now "well-understood"? $\endgroup$ – Lysistrata Apr 27 '16 at 2:47
  • $\begingroup$ I meant well understood for the jet and shear layer. $\endgroup$ – Jan Apr 27 '16 at 13:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.