CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   STAR-CCM+ (http://www.cfd-online.com/Forums/star-ccm/)
-   -   Exhaust Manifold Transient Heat Transfer Simulation (http://www.cfd-online.com/Forums/star-ccm/86258-exhaust-manifold-transient-heat-transfer-simulation.html)

 Jonny6001 March 17, 2011 20:27

Exhaust Manifold Transient Heat Transfer Simulation

Hello,

I am looking at the conjugate heat transfer characteristics of an exhaust manifold, how different regions have varying heat transfer properties with time. With the aim of predicting manifold metal temperatures.

My boundary conditions are pressure inlets at each of the four manifold runners, specified with a table of pressure and time. My outlet is also a pressure outlet with a time table. The temperature at each inlet and outlet is also defined using this method.
I have initially run the gas only region of the manifold to investigate the dynamics of the gas. The inlet and outlet mass flow against time looks good but the model had to run for 3 complete engine cycles for the behaviour to become repetitive (using time steps equal to 3 crank degrees).

I have not included the solid region in to the model and included the heat transfer from the gas in to the solid.
My problem is, the heat transfer from the gas in to the solid is not really though of on a crank angle time scale. So when I give the metal region an initial temperature, it hardly changes because of the very short time scales.
The gas only model took approx 3 days to run 3 complete engine cycles (3x720degrees). Three complete engine cycles is nowhere near enough time required for the manifold metal temperatures to equalise from the initial values to its equilibrium state.
For example, if you cooled a region of the manifold briefly, it would take several seconds for the manifold to reach equilibrium again. 1 revolution in my time scale takes 0.01seconds and half a day to solve.

I have already tried to solve the manifold metal temperatures initially for a static case hoping the temperatures would be closer to equilibrium than an universal distribution. But the problem remains that it takes a long time.

Has anyone else worked on this type of case before and can offer any methods to try and overcome the issue? Any advice is greatly appreciated.

Thanks

I suggest you simulate the flow and heat transfer separately. First simulate the flow to find the boundary condition for the heat transfer.

 Jonny6001 March 20, 2011 20:05

Unfortunately it's not strictly possible to simulate both separately as they depend on each other.

Do you know if it's possible to run part of the model unsteady and other regions steady?
For example, if the gas region boundary conditions are transient, can the other regions be run steady? What would the effects of this be?

Thanks a lot.

 Pauli March 21, 2011 12:47

1) Run the fluid transient until the cyclic result repeats.
2) Cycle average the results.
3) Run a steady state heat transfer analysis.
4) Iterate through steps 1-3 three times.

 Jonny6001 March 21, 2011 15:27

Hi Pauli, your method makes sense.

So you say run the gas region transiently until it becomes periodic then use the cycle averaged values for the whole of the gas domain in a steady state heat transfer simulation? So my whole gas domain will have values for velocity, pressure, temperature etc based on the cycle averaged values?

How is it possible to cycle average the value of each cell and then use them as input for the heat transfer steady case?

Thanks a lot for your time.

 Pauli March 21, 2011 16:22

You only need to cycle average the exhaust manifold wall values (temperature & heat transfer coefficient). They are the boundary condition for the steady state heat transfer calculation.

Output from the steady state heat transfer calculation is the wall temperature boundary condition for the next CFD run.

It's a weakly coupled system. So it should converge after 2-3 iterations.

 JBeilke March 23, 2011 03:09

But running a steady state heat transfer calculation ist not what you want if you need a transient heat transfer simulation.

 Pauli March 23, 2011 12:00

If you are interested in transient manifold temperatures:
1) Compute the cycle average flow parameters for two discreet operating points.
2) Run a transient heat transfer analysis between those points.

 Jonny6001 March 23, 2011 12:39

The temperature fluctuations in the manifold transiently will be very small so the actual transient temperatures are not important.
It eventually reaches what can be considered a steady state.

The transient importace is in the gas region to evaulate the cycle averaged heat transfer properties through the contact boundary.

That is how I feel about the situation.

 All times are GMT -4. The time now is 03:15.