hi,

I am trying to simulate turbulent flow over an gas turbine blade.I tried to apply boundary layer at the blade surface to have better control while refining.

when I used a particular boundary mesh I had problems in convergence.but,when I increased the no.of rows in the boundary layer and made it thick,the solution converged beautifully.

Can somebody explain this phenomenon,and also give any advice on boundary layer meshing?

What was your y+ ? If it is too low and you use standard-wall-models, the "solution" often does not converge.

hi,

thanks for the reply,but if the solution does not converge,and I compute the y+ after some iterations,will the y+ indicate the correct trend?

santosh

There is no clear answer to that, it depends on how "not-converged" your solution is and what causes it not to converge.

However, you can quite easily estimate the y+. There is even a small java-script on the net which allows you to do it interactively - you can find a link to it on this site ( http://www.cfd-online.com/Resources/calcs.html#misc, the link is named "viscous grid spacing calculator").

A) u can mesh based on y+ and y* B) what do u mean by correct trend ? do u mean if it is 600 at some stage of calculation, does it mean the correct solution even though it decreases over large number of iterations ?

DC

hi,

by correct trend what I mean is that say,the calculation proceeds to 200 iterations and then diverges,if I estimate y+ value at this point,how correct is this estimate?

santosh

A) At any stage you calculate y+, it will give the correct value B) At that point even though it's within (if) the limit, it doesn't necessarily mean that the solution of the prob is correct