CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (

Carl September 23, 2006 13:45

Dear all,

I am trying to simulate the insulation performance of an particular wall-air-wall arrangement. I have set a 2D basic problem to study the case before trying the 3D model. The experimental set up is basically a heat source of 100 W/m^2 on the bottom of a horizontal wall then comes an arrangement of confined air sections and then another wall where thermocouples read temperature versus time.

The above set up is what I try to model, I have set (going from -y to +y sides): a wall section, then an interphase solid-fluid, then air then another interface solid fluid then an arrangement of 'thick paper' (with properly defined solid-fluid interphases) and finally air another solid-fluid interphase and the ending wall. In the lower face of the bottom wall I have set a boundary condition of 100 W/m^2 (not as source, but as load). Then in the upper face of the top wall I defined natural convection (5W/m^2K @ 20C).

The first problems I had were that as I defined one air domain for all the 'independent' air confined volumes, CFX gave a error asking to define particular air domains for each confined volume. I did it. Then, I tried a transient run and CFX gave errors regarding pressure reference levels for the air domains... I put a reference pressure of zero for all the confined air domains and a pressure level of 1 atm. I tried with timesteps of 0.5 secs, and 0.25 sec, using upwind scheme, euler bcwrd 2nd order, and a timestep of 12 for solid regions (based on Fourier number) and automatic stepping for fluid regions and CFX gave a overflow error....

I better tried then a steady solution, it converged fast but I think the results do not look quite good, although somewhat convincing (I was expecting the steady behavior reached experimentally)... Now, what I try to do is to use this RES file as an input for a transient run... but I am a doing something wrong, since CFX (by default) ask for an 'initialisation' of the domains (I would think that the initial profile is the steady state solution I already have)... and although I use this RES file as an initial solution the error of over flow comes again...(still use upwind scheme)

question: do I have to get rid of the heat flux load in the transient model?... is it something to do with timestepping? (since error is overflow)... or any other ideas you may throw?...

PS. The mesh in the interphase regions is fine respect of the 'away of walls' mesh


Joe September 25, 2006 09:48

Done use the upwind scheme unless absolutely necessary. Try smaller timesteps and use the same timestep for all domains.

If you still dont make progress read this and provide the necessary files:

All times are GMT -4. The time now is 15:42.