Test Cases: 2D or 3D?
 

Discoganya recently added a new validation case (Flow across a square cylinder), and that set me to thinking. Should we be classifying these cases on the geometry or the flow field? I personally think it should be the geometry, but was wondering what others thought.

jason

Re: Test Cases: 2D or 3D?
 

There are many criteria by which we could classify. This is merely a matter of rearranging the links on the main test cases page. The internal structure of the wiki pages does not change. As more test cases get added, we can think of creating a better table of contents. We could also create multiple classifications. For example one page could list classification based on 1d, 2d, 3d. Another could list it based on inviscid, laminar, turbulent. And so on.

Re: Test Cases: 2D or 3D?
 

Good point. It is easy to forget that a Wiki does not have any inherent tree-structure and each page/article can be linked to in many different ways.

But to answer the original question concerning classification into 2d or 3d. I'd vote for that we categorize based on the flow-field and not the geometry. My motivation is that when you are looking for a 2D test-case you are most likely interested in testing your code in 2D and you do not want a test-case which has a 3D flow-field.

Re: Test Cases: 2D or 3D?
 

My concern with going with the flow field as the classification criteria is that might lead to two articles on the old "flow over the cylinder" case. One for laminar or RANS-type validation in 2D (that I have done and was going to add), and one for LES/DES/DNS-type validation in 3D (as is already there for the square case). To me, it would be best if there was one article addressing both. But that is just my opinion.

Re: Test Cases: 2D or 3D?
 

I think the classification shall be based on the models, that is if we see any comm solver and if we see its tutorials, they are always divided into categories related to models they use, so, i would personally divide them into ,

laminar

turbulence

--rans

--les

--other etc

multiphase

--vof

--etc

combustion

--finite rate

--flamelet

---etc

heat transfer

-- etc

the main benefit of this way of classification is when someone is looking for something, he knows what area of cfd is want the problem in.

further if you are not satisfied by this, you can classify them into 1d 2d and 3d

turbulence

-- 1d

---rans

---les

---other

-- 2d

---rans

---les

--- others

-- 3d

---rans

---les

--others

the user can now know exactly where to go.

Re: Test Cases: 2D or 3D?
 

I think the classification shall be based on the models, that is if we see any comm solver and if we see its tutorials, they are always divided into categories related to models they use, so, i would personally divide them into ,

1. <LI> laminar <LI> turbulence
   1. <LI>rans <LI>les <LI>other etc
   <LI>multiphase
   1. <LI>vof <LI>etc
   <LI>combustion
   1. <LI>finite rate <LI>flamelet <LI>etc
   
   <LI>heat transfer
   1. <LI>etc

the main benefit of this way of classification is when someone is looking for something, he knows what area of cfd is want the problem in.

further if you are not satisfied by this, you can classify them into 1d 2d and 3d

1. <LI> turbulence
   1. <LI> 1D
      1. <LI>rans <LI>les <LI>other etc
      
      <LI> 2D
      1. <LI>rans <LI>les <LI>other etc
      
      <LI> 3D
      1. <LI>rans <LI>les <LI>other etc

the user can now know exactly where to go.

Re: Test Cases: 2D or 3D?
 

how can 2d cordinate to 3d coordinate convert & which formula use to easy convert 2d to 3d coordinate ?