Hello!

I am having difficulties getting Fluent to accurately model the following unsteady heat transfer model:

I have a fluid-solid interface problem (i.e. a steel pressure vessel containing propane), wherein heat is applied to a small part of the exterior of one wall. The idea is that the heat transfer through the wall will cause natural convection (and thus fluid motion) of the propane within the tank.

However, after running the simulation with 100 time steps of approximately 5 seconds the fluid remains stationary, and the temperature remains the same throughout the both the fluid and the walls of the tank.

I am using the following conditions in Fluent 6.0:

Segregated Unsteady 1st Order Implicit Solver Energy equation enabled Laminar viscous flow model with visous heating Piecewise linear density with temperature Constant material properties PISO pressure-velocity coupling 20 iterations per time step

My walls are defined with a finite thickness and have a specified heat flux with a specified shell conduction.

Does anyone know if I am missing something that prevents me from getting a valid solution? Any suggestions or changes that you can think of? Could this be a pre-processing problem (i.e. some form of error in generating the mesh in Gambit?)

Thank-you very much!

-A

Hi there,

This is a wild stab in the dark because i had a similar problem.

Make sure you have scaled your mesh (grid/scale).Im assuming you created your geometry in mm and fluent reads it in meters. If this is the case as in my problem, the heat had to travel through a 2m thick wall instead of a 2mm!!!

Hope this helps.

Good luck

Thanks for response, but the scale in Fluent is o.k. that is not the problem.

Hi

I have never model unsteady heat transfer, however I have model unsteady flow field.

Reading your case setup, I wonder why do you choose 1er order discretization, I really suguest that you choose second order for all the equations and simple c as velocity pressure coupling.

The selection for PISO must to have certain condition that you should read before appling it to your case.

Best regards

Alex Munoz

I would suggest you to perform staedy analysis first. If you have already opened gravity, you should trigger the flow in the direction you think it is going to flow by using the initial conditions panel (initial velocity in that direction -opposite to the gravity- which is very close to zero, like 0.001 m/s, may help it) What makes me think is that even if the natural conv. did not work properly, there should have been changes in the temperature. So the problem may be due to the non-matching mesh at the interface.

regards emre

may i know that the properties of u's fluid is constant or not? you should define the fluid properties, specially density will change with temperature, can use some function provided in fluent.

we know that convection is due to density change, so the density of the fluid cannot be constant.

hope this help.