Learning Forum

[Jason Sum]

I am conducting a drag analysis on a Ahmed Body.
I learnt that in general, one should not merely rely on convergence plot in the residual graph to conclude the accuracy of the simulation, some desired parameters need to be monitored as well.

Here is my current result on residual and cd (desired parameter monitored):

I would like to ask:
1. Is it numerically reasonable for some types of simulations, allowing residuals of some parameters (e.g. k, continuity, veloctiy components) not truly converging but staying flat consistently at 1E-1/1E-2. How can I determine (in what types of simulation) if further improvement or refinement need or I should tolerate and allow the flat residual and focus on the desired parameter monitoring? How to judge overall?

2. Regarding the y+ value capturing boundary layer fluid viscous behavior, I heard that y+ value should be <1 for the viscous sublayer, some says 30 < y+ < 300 (correct me if I mess sth up plz). By some calculations in my case on ahmed body, y (first layer height) is so small (like 1e-5 meter) and aspect ratio will be high which causing poor mesh quality near inflation layers. How can I generally define a proper layers for capture turbulence while maintaining good mesh quality? 

Thank you for answering and addressing my struggles!!

[Sunil Lilhare]

Hello, in general if the mesh is good and simulation setup is correct, then the residuals fall below prescribed value after some iterations. However, for steady state simulations, sometimes the residuals drop initially and then fluctuates about mean value. In such cases, additional report definition plots should also be monitored.

Additionally, if it is inferred that the convergence has been reached, then proper mass or energy balance should also be checked.

For such cases where residuals and report definition plots fluctuates about the mean value and do not fall down for any number of further iterations performed, the simulation should be continued in the transient mode for better accuracy.

In the shared images it looks like the residuals are falling, so the iterations should be continued further.

For detailed information on how to select proper y plus values, please have a look at the Lesson 2 of the AIS course on Turbulence Modelling in Ansys Fluent whose link is given below.

Turbulence Modeling in Ansys Fluent | Ansys Innovation Courses