CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Phoenics (
-   -   %Error on earth (

Sukun Kurusathian January 14, 2001 23:40

%Error on earth
Dear, I've just use Phoenics 3.3 to simulate hot air flow phenomena in an oven. The result looks good, but during earth calculation, % error was very high. Why % error on earth solver is so high (~1.00e+4)? What does it mean? Thank you very much.

Javier Larrondo January 15, 2001 11:58

Re: %Error on earth
Clearly I'm not an expert. But I have noticed the same thing, but fortunately the cases that I have done, even that they have high errors (~10e+05) are good enough to understand the fenomena and the values are in range.

I found an answer in a lecture that is in POLIS, but I don't understand the RESREF value, maybe someone can explain us this.

you can find the lecture in:

if you read this lecture the residuals presented in EARTH, are normalized by RESREF.

So, if you have an error or residual (ep) around 0.1 in every cell and you have a case with 10000 cells (~21x21x23) which it's not a big case, so the sum(ep)=10e+03 and normalized by RESREF is the value that appears in EARTH so maybe RESREF is not big enough or of the order of sum(ep), but I don't know the value of RESREF, and that is what I understood from the lecture.

Well, that's all I can say, I hope it will be useful.

Javier L.

PattiMichelle March 2, 2001 13:29

Re: %Error on earth
Hello: I have been similarly confused by combustion problems in PHOENICS. When I view the %Error and RES/RESREF plots in Earth, they decrease up to a point, then simply don't converge any more no matter how long I let iterations continue. I have tried increasing and decreasing relaxation but to no avail. Many or most of the library cases WILL converge completely but mine do not. I might mention I am doing combustion simulation with variables H2, O2, OH, HO2, H2O, etc... and these may present ill-formed or stiff problems that simply don't converge completely by numerical methods.


All times are GMT -4. The time now is 21:00.