[by_dundar]

Hello everyone,

I am doing retro propulsion (is a rocket engine providing thrust opposing the motion of a vehicle, thereby causing it to decelerate) analysis for my master thesis. I have two problems about that and I want to ask you.


First of all, I would like to share my setup with you.


Problem 1: I am doing the analysis as 1st order. Flow layers coming out from the jet are not perpendicular to the axis. I figured it out when I ran the 2nd order. But I couldn't bring it up. but when i keep running with the 2nd order i get a lot of errors and the residuals are diverged at the end of the day.I tried it with other viscous models but I always got the same result. How can I make the flow layers perpendicular to the axis for 1st order?


Problem 2: The pressure at the port on the forebody closest to the nozzle exit is not captured all by CFD solver. The expansion of the jet flow at the nozzle exit affects the pressure distribution in the region near the nozzle exit. CFD solution cannot be capturing the proper expansion angle at the nozzle exit. I increased the number of cells, I tried different viscous models, but I still did not get any results. Do you have any suggestions for this?

Thank you.

[RaiderDoctor]

Hi by_dundar,

I'm afraid you really haven't given us much to work with. Let's start with your mesh; what does it look like? Give us some pictures and statistics and let's go from there.

On to your first problem. What do you mean by "flow layers"? I typically work with subsonic applications, but I've never heard this term before. Also, given the picture you posted, it appears as though your solution is semi-correct, and qualitatively matches with published results (https://media.springernature.com/lw7..._Fig3_HTML.gif). I agree that the trailing edge looks odd, but I can’t really comment until I see the mesh. Actually, would it be too difficult to post a pic of your entire fluid domain so we could see where you apply your boundary conditions?

Next, let’s talk about your setup. Which residuals were diverging? When? Have you tried adjusting the time step, etc. It's always helpful to go back to the basics and double-check all of your assumptions. Lastly, why do you want to use 1st order? It's inherently inaccurate, and you'll likely get called out on it during your defense.

Now, on to your second problem. Please elaborate a little more on what you mean by the pressure not being captured by the CFD solver. It looks like you graphed some values, so the CFD solver is capturing values. Are they just not what you'd expect? Further, you aren't capturing pressure, you're capturing the coefficient of pressure for various conditions. Why?

Finally, as to the "why isn't the CFD solver showing 'accurate' results" question. This is the fundamental flaw with all numerical solutions; they are all wrong. You can make the best mesh in the world, set the perfect boundary conditions, and acquire excellent convergence. The solution is still wrong, until it is proven to be right by experimental data. So, it looks like you are trying to match with what appears to be said data; how far off is it? What's the R^2 value? Is it good enough? It looks like it matches for increasing values of x/r, so the question you need to answer is why.

I know I’ve given you more questions than answers, but hopefully answering them will help us determine a solution.

[by_dundar]

Originally Posted by RaiderDoctor

Hi by_dundar,

I'm afraid you really haven't given us much to work with. Let's start with your mesh; what does it look like? Give us some pictures and statistics and let's go from there.

On to your first problem. What do you mean by "flow layers"? I typically work with subsonic applications, but I've never heard this term before. Also, given the picture you posted, it appears as though your solution is semi-correct, and qualitatively matches with published results (https://media.springernature.com/lw7..._Fig3_HTML.gif). I agree that the trailing edge looks odd, but I can’t really comment until I see the mesh. Actually, would it be too difficult to post a pic of your entire fluid domain so we could see where you apply your boundary conditions?

Next, let’s talk about your setup. Which residuals were diverging? When? Have you tried adjusting the time step, etc. It's always helpful to go back to the basics and double-check all of your assumptions. Lastly, why do you want to use 1st order? It's inherently inaccurate, and you'll likely get called out on it during your defense.

Now, on to your second problem. Please elaborate a little more on what you mean by the pressure not being captured by the CFD solver. It looks like you graphed some values, so the CFD solver is capturing values. Are they just not what you'd expect? Further, you aren't capturing pressure, you're capturing the coefficient of pressure for various conditions. Why?

Finally, as to the "why isn't the CFD solver showing 'accurate' results" question. This is the fundamental flaw with all numerical solutions; they are all wrong. You can make the best mesh in the world, set the perfect boundary conditions, and acquire excellent convergence. The solution is still wrong, until it is proven to be right by experimental data. So, it looks like you are trying to match with what appears to be said data; how far off is it? What's the R^2 value? Is it good enough? It looks like it matches for increasing values of x/r, so the question you need to answer is why.

I know I’ve given you more questions than answers, but hopefully answering them will help us determine a solution.

Hi RaiderDoctor,

Thank you for your answer and questions.
I will try to answer each question as clearly as I can.

I share mesh photos. You can see in the attachment.
When I say 'flow layers' I think I used a wrong expression. I was talking about flow spots that express each color of mach number. I can explain this in the 'flow layer' picture.

You can find a picture of the boundary conditions in the attached.

When I switch from first to second order, I don't get an error at first, then I get the ''minimum static temperature limit'' error in a few cells in the area where the mach disk is created. (the picture is attached). Then, the number of cells that have the minimum static temperature error starts to increase and I am starting to get a pressure limit error. Then, the errors increase and the maximum temperature limit error is occured and all residuals start to go up. Time step was generally 0.01, and I kept the same time step and I played with Courant number, but again I got the error. I know that the second order will give better results, but I have always these kind of errors in my analysis with the second order and cound not find a way. And I read something on the internet, that is 'it is difficult to run the analysis in second order when flows has high Mach numbers '. In my analysis, Mach number goes up to 16.

I said the pressure is not captured by the CFD solver because I get a different result from the experimental result near the nozzle at x / R = 0.3 as you can see in the chart. This is not only my problem, but also the people who did these kind of analysis faced this difference on the pressure. As our discussion with my advisor, we found that circulation was not sufficiently developed around the nozzle. Pressure and Cp values ​​matched with previous CFD analysis, but in the part close to the nozzle, it does not exactly match the experimental result. Now I'm trying to solve this.

I hope I was clear enough. Thank you again...