Gridding and geometry
Dear All CFX USER,
Actually I have just found the web site containing a discussion of CFX USER.
I have a problem related to GRIDDING or GEOMETRY. Now, I am simulating a pneumatic conveying drier, a particle of wood is dried by the mixture of hot air and superheated steam. I apply the EULERIAN METHOD to calculate multiphase flow.
Now, I am calculating a vertical straight pipe having the length and diameter of pipe of 13 m and 0.1 m respectively.
In the beginning, I created a shorter pipe i.e. length of 1 m and solved it. I didn't get any difficulties and the calculation could reach a convergent solution. In addition, the maximum number of grid that I have ever used is of 160000 of grid (20x20x400).
And then I tried to used my real problem i.e. length of 13 m and solved it. Unfortunately, CFX-4.3 that I was using didn't work well. It couldn't work after creating a COMMAND FILE, giving information about which solver used to calculate, writing some information about a value of some constants and other information, writing the kind of patches and finally writing the TOTAL FLOW THROUGH INLET AND OUTLET BOUNDARY. And then the calculation was stopped without any information.
For a convergent solution, calculation would be continue with MONITORING THE VALUE OF VARIABLES AT THE MONITORING POINT etc and finally it would be stopped by writing information of residual values.
I had a thinking that it is probably caused of number of grid. I, and then, decreased a number of grid several times and finally I got convergent solution after I applied the number of grid of 2500 (5x5x100). It looked so funny because this grid is quite little and if I compare with a shorter geometry, it has 160000 number of grids and still ok.
Does the shape or length of geometry affect to convergence of calculation? And how big it affect to the convergence of calculation?
To anyone who knows how to encounter my problem, please give me some suggestion.
Thank a lot Muslikhin Hidayat Chemical Engineering Design Chalmers University of Technology Göteborg, Sweden
Re: Gridding and geometry
Dear Hidayat, I think you are facing problem with aspect ratio of element (ratio between the longest and the shortest element side). I don't use CFX 4, but I am sure that aspect ratio and skewness/orthogonallity reduces the accuracy. In tascflow 2.9, the maximum aspect ratio is 100. But for special cases, i.e. recirculation and separation flow,the aspect ratio should be around 1.
You can play with the aspect ratio and hold the grid number, to see its effect on the convergence.
Re: Gridding and geometry
Thank for your suggestion, and I will keep it in my mind.
Actually, I had to apply initial guess for my case. After I set a reasonable initial guess, everything worked.
|All times are GMT -4. The time now is 11:11.|