CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   2d Heat transfer external flow (

Richard belcher August 4, 2008 20:47

2d Heat transfer external flow
Im doing a study on a heated cylinder in a uniform cross flow situation ,and my wish is to calculate the avg nusselt number over the surface of the cylinder for various reynolds numbers.

at the moment my results are out by by 40%ish percent from known correlations and my mesh is very refined around the cylinder (yplus 0.1-0.5)and has been checked by others.

Cylinder is defined as fixed 40C temperature, SST and thermal energy models. 100 cycles per timestep

What i dont understand is i believe all my expressions for calculating it are correct but i would appreciate if someone could cross check them for me as it is been a few days now troubleshooting this problem.

here is my list of expressions , my approach is to find heat flux using CFX and use the convection eqn to find h. and then using that to calculate Nu (my diam = 19mm ,0.019m)

It also requires i define Cp (or so it tells me) but i am unsure where i can input this , couldnt turn up anythign in help.

Ts = areaAve(T)@cylinder

q = areaAve(Wall Heat Flux)@cylinder

h = q/(Ts-Tinf) Tinf = 298K

(h*0.019[m])/(areaAve(Thermal Conductivity)@cylinder)

and for defning outlet temperature(m.cp.deltaT)

(((area()@inlet)*q)/(mdot*Specific Heat Capacity at Constant Pressure )) + Tinf

Thanks alot,

Glenn Horrocks August 4, 2008 21:42

Re: 2d Heat transfer external flow

I assume you have a fully validated model and things like timestep, mesh, physics are all OK.

If you are in the laminar flow regime then hopefully you can get the heat transfer better than that. If you are turbulent the accuracy might be inherent in the correlations and no amount of model tweaking will get it right. Also, if turbulent then consider the turbulence transition model. It will make a big difference to the heat transfer is some regimes.

Other comments: * Consider using the in-built h variable rather than calculating it yourself. * Consider using variable properties for the air (eg viscosity, k, Cp versus temperature)

Glenn Horrocks

All times are GMT -4. The time now is 05:37.