CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Need help with a rocket nozzle simulation (

badboyz31 June 13, 2013 07:01

Need help with a rocket nozzle simulation
Hello members of CFD Online. I've got a problem with a rocket nozzle simulation.

In this simulation, I have a supersonically moving rocket nozzle with internal combustion. You may see how I doid the meshing with ICEM CFD (unstructured mesh) in this picture :
The rocket nozzle is encased inside a tube structure (there is no volume mesh in the space between the nozzle and the case. The rocket also ejects high speed air through the nozzle which I modelled as a inlet in front of the nozzle.

Then, I ran it in CFX. The inlet inside the rocket were meant to be releasing air at 80 bar and 3000 K. The rocket itself is moving at 640 m/s at sea level condition.

CFX gave me an immediate overflow error in the process. However, by setting the inlet inside the rocket to release air at 60 bar and 1000 K. The calculation went smoothly until the 91st iteration.

I would like to know what could have made the error and if you need more information, I'll be glad to give it to you.

And I'd want to know how to set the heat capacity ratio (gamma) of the air inside CFX ? (in the simulation, I used Air Ideal Gas, but I want to change some properties of it)

ghorrocks June 13, 2013 18:51

gamma is the ratio of specific heats, so change the specific heats and you will change gamma.

Your convergence problems sound like typical failures to converge - have a look at this FAQ:

diamondx June 13, 2013 19:10

go with a structured mesh, it easy to do it in icem, we can help you in the geometry and meshing section

ghorrocks June 13, 2013 20:50

I agree, improved mesh quality is a really good place to start with simulations which are tricky to converge.

badboyz31 June 20, 2013 06:00


Originally Posted by ghorrocks (Post 433929)
I agree, improved mesh quality is a really good place to start with simulations which are tricky to converge.

Alright then, remeshing.

However one more thing, I noticed that the maximum mach number increases to a thousand mach with each iteration. I think that it's the signal for bad iteration. Is there a way to avoid it ?

ghorrocks June 20, 2013 06:32

That means your simulation is diverging. You need to improve the numerical stability of your simulation. Assuming the physics is correctly set up that would mean double precision numerics, better mesh quality, smaller timestep or a better initial condition.

badboyz31 June 24, 2013 08:38

Alright, got it. Although i got to wait for several hours to get the simulation done, but finally it converges.

Thanks everyone

badboyz31 June 24, 2013 08:47

Oh, one more thing.

Once I'm done with the simulation, I need to know the presssure coefficient along a X-wise polyline (which is on the inner surface of the nozzle) and plot it into a X-Y graph in CFD Post. I can't find the function Cp in the graph properties. Any trick that does it ?

And is there a way to plot 3D graph on CFD Post ? I need to plot total temperature at the cross section of the nozzle's throat (that means, X-Y-Temperature graph).

ghorrocks June 24, 2013 20:05

Your first question sounds like simple chart processing. You might need to define a variable for your Cp, but it is then easy to make a chart of it along a line.

But CFD-Post has limited capabilities to do the 2D surface/3D chart you describe. I would export it and do that in python, matlab or tecplot.

badboyz31 June 25, 2013 19:02

So i just get to the table viewer and define Cp as a new variable with its own equation ?
Am I going to type something like this in the table viewer : Cp=(2/1.4*Mach_number@X)*(@X/pressure@freestream-1)

And then, how to export the data on a cross section ? I only know how to export data from a chart (since there is an export button).

Finally, is it possible to pause an iteration process inside the solver manager while it's running and then resume it later ? (Let's say that I had to shut it down first to prevent overheating from long-duration calculation)
Sorry for asking so much, but I'm really a noob in CFX.

ghorrocks June 26, 2013 06:04

The CFD-Post tutorials describe how to add a variable and do stuff with it. You want to define a general variable (not just a spreadsheet cell) so you can use it as a variable to plot on objects.


Pause - you can stop a simulation with stop and restart using an initial conditions where you use the results from the last simulation. But thsi can take several minutes to actually stop by the time it wraps up all processes and write the output file. If your CPU has overheating problems then turn your overclocking down or get a better cooling fan. That is going to cause you no end of grief if you run CFD on a computer which overheats anytime you give the CPU something to think about. Also note current CPU have heat sensors built into the CPU and throttle their own speed shen they run hot - unless you are an overclocker and have disabled that too :)

badboyz31 June 26, 2013 06:59

I'm running the simulation on a laptop which has limited overclocking ability but i guess i'll try to disable intel turbo boost to see if it cools down. Sometimes, I'll have to pause the simulation to rest the laptop sometimes.

Thanks so much for everyone's help

Ranjan April 5, 2014 03:55

Rocket Nozzle flow simulation in OpenFOAM
Dear Foamers,

Can any of you recommend how to simulate supersonic flow through nozzle? Which solver can I use?

Thanks in advance

ghorrocks April 6, 2014 08:36

I have no idea. Try the open foam forum. This is the CFX forum.

Ivanrips September 22, 2014 00:06

Nozzle tutorial
Here tutorials de Laval nozzle in CFX

Part 1

Part 2

Part 3

Part 4


All times are GMT -4. The time now is 18:56.