Hi there,

this is a serious problem with fluent 5.5. I am calculating a combustion chamber with several inlets. 3 additional scalar transport equations are defined, and for one of those scalars zero boundary conditions are defined for most of the inlets. In spite of this a negative value stream of this scalar enters the domain through one of the inlets where bc is zero, getting smaller with preceding iteration steps, up to -14, even though this scalar is per definition between 0 and 1. Can anyone help? Is a conflict with user defined variables possible?

Thank you,

Elmar

Is it zero at the boundary or close to it????

If the later, it may be something else. Check diffusion co-efficients (Fluent had default of 1) and how you defined the mass flux that convects the scalar.

You mention that by defition this scalar is between 0 and 1. I gather then this is a mass fraction or soemthing. But Fluent will be unaware of any constrainsts that need to be imposed physically on the user scalar solution (boundedness etc.), in which case you will almost certainly have to ensure this using udfs during the solution process. The easiest way is probably to use Patankar's (see his book) method with a source term linearization to ensure the solution remains within specified limits.

Greg Perkins

Thank you, Greg!

Information you requested: boundary condition at this particular error-producing inlet is zero. The scalar corresponds to a mixture fraction as included in FLUENT's Pdf model. Therefore you don't have any source term in the transport equation for this scalar - no evaporation of liquid fuel assumed.

Diffusion coefficient at inlet bc is already set to 0, and convective flux is set to mass flow rate. I am still wondering why this only happens with second order discretization. Can anyone commend on that?

Perhaps it is a bug. Have you tried asking Fluent????

Greg