Dear Whoever Can Help Me:

I am fairly new to CFX 4.3 and to CFD for that matter and I am really stuck. I am trying to model simple case of water flowing through two plates. The plates are heated and the water is used as a coolant. I set the Command File to use a Data Base, and I specified the PCP data base that comes with the software package. I specifed water and was able to get decent results until I turned on the Bouancy Option and tried to use weakly compressible flow. I am definately getting the wrong answers now but the funny thing is the results are for air flowing between the plates and not water. The density matches that of air and not water. Now the default fluid under physical properties is air (>>STANDARD FLUID) so I tried to set it to water. Apparently I am not supposed to use a data base and set the standard fluid because solver won't run when I do that.

If anyone could help me out or give me some idea of what I could do differently I would really appriciate it.

Thanks in advance!

Richard Harrison

Utah State University

Hi Richard,

If you are just using water as your fluid the "STANDARD FLUID" option is probably best. In your command file you should have:

>>MODEL DATA
:>PHYSICAL PROPERTIES
:>STANDARD FLUID FLUID 'WATER' STANDARD FLUID REFERENCE TEMPERATURE 300.0

Set the standard fluid reference temperature to a typical temperature in your domain, say the average of the temperature of the two plates - it is just the temperature CFX uses to determine the density, viscosity, thermal expansion coefficient etc. as they vary with temperature.

Alternatively you could directly specify them using >>FLUID PARAMETERS and not use a database at all.

Of course, if your problem is dominated by natural convection, and the temperature difference across your solution is greater than 3 degrees Kelvin the Boussinesq approximation and constant fluid properties is no longer accurate. You then should use variable properties - you will have to specify fluid parameters as a function of temperature and pressure using user fortran, and use a fully compressible simulation then. (E Leonardi, "A numerical study of the effects of fluid properties on natural convection", PhD thesis, University of NSW, Australia 1984) Going into the details of your problem no doubt is a PhD project in itself...

Hope that helps, Glenn Horrocks