Learning Forum

[aram.alizadegan]

Hi everyone

I am working on a tall building CFD simulation and I need wind forces on building but I have a problem about residual convergence. I have a tall building (30*45*180 m) inside a domain (1800*3600*1440 m) with these boundary conditions:

[Nikhil Narale]

Hello,

Can you try monitoring the parameter of your interest (for example, pressure or force) and not only the residuals and see if that converges (Reaches to a constant value after some iterations)?

Nikhil

[aram.alizadegan]

Thank you very much for your kind reply Nikhil
I have created this comparison to give better understanding about the problem and also yesterday I understood from this tutorial (/courses/index.php/courses/basics-of-turbulent-flows/lessons/simulation-exampleshomework-quizzes/topic/wind-flow-between-two-buildings-simulation-example/), that if I increase the Maximum Turb. Viscosity Ratio from Limits to 1e+10, the equations will be converged.
In all three models (simple without scaling, scaled and increased Maximum Turb. Viscosity Ratio) pressure and force amounts Reach a constant value. but only in simple model, the building's pressure distribution is similar to the results of wind tunnel experiments approximately.
In this situation, Which way should I use? And if the residuals don't converge but the amount of pressure and force reach a constant value, are they correct?

[Rob]

The viscosity ratio limiter is designed to stop solutions running away with silly values. Unfortunately, built environment models tend to cover length scales that require a higher value on the limiter: I typically add 3-4 0s to the limit before starting a run. The reason it's not converging when you hit the limiter is (I think) covered in the Theory Guide.
Coming back to monitors, residuals etc. The residual is a function of the error in the model. If you have any transient effects in a steady solver (eddy shedding or more commonly wake movement) residuals will stick at around 1e-2 or so. The bulk of the flow is stable, but bits aren't. We then use monitors of useful numbers (temperature, pressure, velocity etc) to see if the values we're interested in are changing, and if so, by how much. It's then for you to decide if the residual is low enough, the monitor is constant enough and the flux balance near enough to use those results.

[aram.alizadegan]

Hi rob
Now I understand what happens
I appreciate the information and advice you have shared.
thanck you

[Rob]

You're welcome.