CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   PhD dilemma (

shuo May 24, 2008 02:49

PhD dilemma

I have written a code that is very slow to run. I am considering whether or not to start all over again using other source codes (ie numerica by Toro) or contact the authors of the paper whose method I am trying to implement and asks them nicely for the source code. The paper was published back in 1990 so I am hoping they'll be generous enough to share it.

Your advice is greatly appreciated. Shuo

jinwon park May 24, 2008 03:30

Re: PhD dilemma
You can do well. Do not give up. I also suffered from such failure but recovered soon. Check the line-by-line. You can find bugs you has not realized. If not, you can ask someone else around you. I hope you feel better.

su May 24, 2008 15:34

Re: PhD dilemma
can you tell us what is that paper please ?

shuo May 24, 2008 19:26

Re: PhD dilemma
The paper is "Numerical analysis of gas-particle two-phase flows", R.Ishii, Y.Umeda and M.Yuhi, JFM 203:475-515, 1989. I tried to find the author in the web but hasn't been successful in tracking them down.

rieman May 25, 2008 02:53

Re: PhD dilemma
I strongly suggest you to think of numerica by Toro and to read his book. Also you need to think of having well established mathematical analysis for your case i.e. as in his book (chap 4 - 14). I used the same methodology for compressible two-phase flows.

Please let us know how it goes!

shuo May 25, 2008 07:25

Re: PhD dilemma
Hi rieman

Can you recommend some papers on extending Riemann and upwind solver to multi-phase flows? Also how easy it to extend the methods outlined in Toro's book to the Navier-Stokes equations? I am simulating a supersonic impinging jet and viscosity would play a part for the wall jets.

Cheers shuo

Chin Fook May 29, 2008 22:44

Re: PhD dilemma
Writing a code sometimes may give you a real hassle. Debug your code smartly; meaning check by running them section by section (part by part) in order to figure our which routine actually causes the bottleneck. Anologuous to experimental research, we try to reduce the number of variables at a time just to find out what is the effect of a single variable.

Nishu July 7, 2009 18:13


Are there any possibilities to get the NUMERICA code from somewhere?

oopscangel July 8, 2009 09:55

Hey, just send an E-mail to Prof. Toro or send an E-mail to me, and I will give you the NUMERICA codes.

Nishu July 8, 2009 13:35


My email address is
Please send me the codes along with the documentation


All times are GMT -4. The time now is 00:43.