Can someone give a clue as to how to conduct a grid sensitivity test?

I've searched the forum and this is what one of the user recommends:

http://www.cfd-online.com/Forum/main...cgi/read/10797

However, he didn't about the size of the domain, in the case of an external flow.

So in all (I think) there are 3 parameters to vary:

1. domain size 2. no. of pts 3. grid spacing

How and which order should these parameters be varied?

I believe the experts here should be able to give a rough guide.

Thank you.

Varying the domain size will give you the influence of the farfield boundaries on your problem. That is, are they sufficiently far away or are they somehow reflecting and affecting your near field solution.

Varying the # of points and grid spacing is really the same thing if you fix the domain size. Systematic reduction in grid spacing (increase in # of points) will give you a measure of your truncation error.

You can examine each effect separately and the order doesn't matter. In each case you should create at least three meshes to determine a trend.