hi i am trying to simulate incompressible flow of air in the air inatake system of a siemens gas turbine. the above system is for filteration and cotrol of noise of compressor stage of the turbine, i have a problem in finding suitable k and epsilon values when i am trying to solve the overning viscous model by use of fluent. this is not a qustion for fluent techniques and trikcs and tips. i feel that the question may have an answer here. anyway the question is: by chosing the vaues of and epsion is it possible to have control over the values of trubulent viscosity ratio in the domain and help the fluent to reach to a residual for continuity arounf 10^-4 ... i will explain the situation if someone is interested to help me. thanks in advance reza

(1). You really need to get in touch with your support engineer to find out why the code you use is having some problems. (2). Convergence is a complex issue, and it can be linked to the formulation, the mesh, the algorithm, the boundary conditions, the turbulence models, the wall treatment, the geometry, etc... (3). Without the indepth knowledge in these areas, it is hard to know what is causing the problem. And fooling around with a commercial code without the background in these areas, is a waste of time, I think. (even with my background, I don't think, there is any hope for me to solve the convergence problem of a commercial code. In nearly 30 years, I don't have any problem with the convergence of my own codes. But if you are using low Reynolds number models, it will be slower than the wall function model) (4). Like what I said before, trying to emulate the Western society by Russian or Chinese will lead to failure and trying to use commercial cfd codes blindly can only destroy the company using it. It is like sky diving from a tall building by non-professional.(and even for the professional, it is a high risk sport)

hi dear Dr. Chein i do appreciate your very sharp description about the severity and importance of analysis and study of the very fundamental issues of turbulent incompressible flow of air. it seems to me that i had the luck to have the chance of getting to this point in my analysis. i have no access to any academic proffessionals with sufficien knowledge and also free time to help me have the problem of convergence fixed. please explanaine to me that what is your judgment on this kind of problem. air flows in from five different equall vertical openings and then enters the duct of around 10 meters width and height and then enters silencer panell stage after having passed 10 meters and then bend downward into the vertical duct of width 2 meters and finally swirl around a cone placed horizontally in front of compressor inlet area with about 30 m/s speed. the mesh is about 100000 size and dimensions are around 15 meters roughly speaking and i think this would not be challenging problem and it is reqauired may be to have the required basic and critical understanding of the subject. i honestly confess that i plan to pursue my academic studies in the western world but for know i am challenging with a commercial software to understand both the problem solution techniques and also theoretical basis of the case under consideration. with all my hope for your success. reza

hi i finally realised that the problem was initiated by bad three dimmensional modeling in the preprocessing. i corrected the very small surface that was the cause of all of my troubles and the convergence begun afterwards. hope this might be helpfll for others. reza

(1). It is nice to know that you are making progress. (2). The interesting thing is your unknown solution depends on your geometry and mesh. So, your solution is mesh dependent. (3). Getting a solution to converge is just the first step. As somebody has already discovered that mesh refinement could lead to less accurate results. (4). Your goal is still: to find the mesh independent solution, if possible.