CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   limit for a variable (https://www.cfd-online.com/Forums/cfx/25211-limit-variable.html)

Luis February 10, 2008 11:56

limit for a variable
 
Hi,

I am using CFX11 to simulate fully-premixed flames with the TFC combustion model. I am not using the combustion model from CFX with multiple components, but instead I am defining a material with properties in function of temperature, solving a transport equation of a new variable for the reaction progress and modeling the reaction source term.

I would like to know if one can adjust a variable between a range. The progress variable should be between 0 and 1, but I got strong overshoots of it (values around 4) at the beginning of the simulation. Then the temperature and enthalpy exceeds the limit and the simulation stops. I am adding also the reaction source term to the energy equation. My reference enthalpy for the mixture is 0. For the ignition, I am using a source point.

If someone knows also another way to ignite the mixture, it would be also helpful.

Thanks

Luis


Glenn Horrocks February 10, 2008 17:27

Re: limit for a variable
 
Hi,

Sounds like a boundedness problem. What differencing scheme are you using for the variable?

I think you can limit the bounds of a variable. I can't remember how to do it but I think it can be done. Talk to CFX support.

Glenn Horrocks

Luis February 10, 2008 17:51

Re: limit for a variable
 
Thanks for your answer. I am using the high definition. I tried also with upwind, but the overshoot appeared too. It would be nice if you would give me some tips too. I am contacting also CFX support.

Thanks,

Luis

Glenn Horrocks February 11, 2008 17:32

Re: limit for a variable
 
Hi,

Second order differencing schemes are prone to overshoots and undershoots, causing boundedness problems. The high res scheme is meant to look at the gradients of the variables and revert back to first order if an under/overshoot is likely so hopefully the high res scheme can minimise the effect. If you have your own model then the high res scheme may not do this prediction correctly and it may still give an over/undershoot. Just my guess as to what might be happening.

I would use upwinding on this variable only for the time being until you have things under control. Then do a study to see if second order differencing makes any difference and only go to it if it helps. Quite often on secondary variables like this second order differencing does not improve accuracy, just makes convergence harder.

Glenn Horrocks

Luis February 11, 2008 18:10

Re: limit for a variable
 
Hi Glenn, thanks for your answer. I tried before also with upwind, but the overshoot appeared too. I will try to change only to upwind the variable. Do you know if one uses a source point, the balance of the transport equation + the source point would lead to an overshoot? I am using a source point for ignition. Is it possible to "patch"(like in Fluent) a variable in determinated domains? I will be grateful if you would send me some information about limiting the variables in case you have it. Also, do you know if probably there is some "adjust function" like in fluent to avoid overshoots?

Thanks again for you help,

Luis

Glenn Horrocks February 12, 2008 06:55

Re: limit for a variable
 
Hi,

Overshoots are common in combustion modelling and similar physics where very rapid changes in fluid state occur.

I don't know what you mean by "patch" a variable, I have never used Fluent.

I don't know of any "adjust function", but I am pretty sure you can specify variables to be bounded. Talk to CFX support about how to do it. I don't know how to do it, I am just pretty sure it can be done.

Glenn Horrocks


All times are GMT -4. The time now is 11:12.