CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Problems in CFX 5.5.1 (

Sulviyanti February 6, 2003 00:14

Problems in CFX 5.5.1
Hi, I got a problem in running my analysis. My analysis is on simulation of temperature and airflow pattern of air conditioning unit in a car. My objective of my analysis is to simulate the temperature distribution and airflow pattern in the car compartment(cooling the compartment). The car compartment has 2 inlets situated infront of the dashbord and an outlet at the bottom dashboard situated at left side of the car compartment.

Below are the settings in CFX Build:

1. Geometry:

model is imported from Pro/Engineer200i^2 using IGES file.The model is then create as solid B-rep.

2. Fluid domain:

domain is defined as Air at STP,steady state,heat transfer as thermal energy and turbulence model used k-epsilon.


involve 2 inlets and 1 outlet.

inlet 1: velocity=4.4 ms^-1

Temperature=12 C

inlet 2: velocity=3.6ms^-1

Temperature= 19 C

outlet : Relative Pressure=0 Pa

4.Initial condition set domain as individual. For temperature inital value=50 C...(this is the temperature of the compartment before cooling)

5.Mesh parameter: set as max edge length=5.0

6.Solver parameter set max no of 30 time steps to be converge at 0.0001

With these setting, the Solver has run the analysis with total cpu time of 2814 sec.

My problem was, the results from the CFX post didnt came out rigth as I expected.The temperature distribution for the compartment was not reliable e.g.the difference between front area and rear area temperature. Front compartment has temperature of 323 K and rear compartment of 285 k.The pattern of the temperature distribution showed that the temperature didn't distributed randomly.

My refference of my settings were based on Tutorial:Static Mixer.Did I have the right setting?What must I do to get a reliable results?

jh February 6, 2003 01:40

Re: Problems in CFX 5.5.1
I think you have to check two points.

1, the timestep you used. When you use a timestep of 0.0001 second and run the simulation for 30 iterations, that means you final solution is at its age of 0.003 second counting from the initial values. Will you expect a great change of solution?

2, mesh size. length=5? m, cm or mm? Is that fine enough to resolve the boundary layer close to the solid walls?

Good luck!

Bob February 6, 2003 06:27

Re: Problems in CFX 5.5.1
A quick way to check if your solution is converged either: a) create monitor points around the inside of your model and output variables such as temperature and velocity. b) restart your model (using a larger timestep say o.1sec ) then after 50 timesteps stop the run and view the solution in post. If there is a difference then its more than likely that you have not run the solution for long enough. Why did you choose such a small timestep. There are some very good tips in the online help (although it needs a little searching on your part) for setting up models. In the tips they cover time step selection for different types of problem. Good luck Bob

sulviyanti February 6, 2003 06:28

Re: Problems in CFX 5.5.1
max edge length used is 5cm,and default value given is 5.9 cm.CAn you explain more on relation of iterations and timesteps? Thank you...

Mike February 7, 2003 00:55

Re: Problems in CFX 5.5.1
A couple of points.. 1. set your number of timesteps to a high value (say 200) and the convergence criteria to RMS 1e-5. The solver will then finish when the solution has converged to a reasonable level - hopefully much less then 200 iterations! The timestep size should be a fraction of the residence time of the air in the domain, my guess would be at least a couple of seconds.

2. Make sure your advection scheme is NOT Upwind. High Resolution should be better.

3. Use a smaller mesh length scale if you can afford the CPU time.


Mike February 7, 2003 01:41

Re: Problems in CFX 5.5.1
... the previous comments are assuming you're running a steady state calculation. Or is it transient? Mike

sulviyanti February 7, 2003 17:46

Re: Problems in CFX 5.5.1
I'm running the steady state and now trying to run transient state.

Now I have manage to run and get the reliable results for my previous analysis. I have increased my time step from 30 to 150. The result do came out right.

I tried to run the transient state problem is in setting the timestep.Do you have any idea how to set up the time step for the transient state?

All times are GMT -4. The time now is 05:40.