In my simulation of analyzing the effect of vane deflection of air pushed by a propeller in a shroud, I set up some variables (like cd, cl, force, moment) to monitor. After running up to 10000 iterations, I have the cd and force converge but cl and moment did not converge.
Can anyone please give me an insight to the reason why these two parameters refuse to converge but keep moving back and forth
Without insight into the flow solution, it is impossible to tell. Several reasons can explain. Either your mesh is not good, your time step is too large, your BC are wrong, etc.
However, let me speculate. Assuming that your flow solution is mostly ok, it could also be the results of vortex shedding in a separated area of the flow. If this is the case, it means that you do not have a stationary problem.
Dear Uchenna
With the iterative methods, convergence is not a guaranteed outcome. There can be several reasons. For example, if at any place any of your variable touches zero and in next step it is being used in denominator, you wont get a solution however it should create some errors or warning etc. Look at another situation where one of your variable has many roots like square root etc, and in that case the coding may converge to a nearest solution which may not be the desired one. Also note that, many CFD models and equations have some attached conditions for validity like an equation may only be true for a certain range of Reynolds number and your parameters do not match that. I would suggest a literature review to check the model first. Then go over the convergence methodologies like relaxation parameters or helping computer by manually putting the conditions to avoid possible errors. For example, we may write a condition that if a parameter is exactly zero, consider it a very very small value equivalent to zero but not exactly zero etc. Also you can direct the convergence to a certain solution if you know exactly the range where the solution should go. For example, if you expect a solution as positive, put a condition to ignore negative roots etc.
You can always get help from the adviser or an expert to guide you in the trouble shooting but its always a tedious work.
Be patient and keep trying.
Good luck
There may be some possible reasons.
1) Mesh is not proper. You may need to resize the mesh as per the design reqires. You may better use unstructured grid system.
2) Boundery condition may not be compatible with your flow domain.
3) Most importantly you need to change the convergence criteria for all the parameters properly and set to a required value for your domain.
Thank you all for the response.
I think I was able to get a mesh though not of high quality but it is within an acceptable range.
As for the BC: For the fact that my flow is compressible, I am using mass flow inlet and pressure outlet now where I defined the mass flow rate. I tried mass flow inlet and mass flow outlet but the system gave me reverse flow then I changed to pressure outlet.
I have my zones named as inlet, outlet, symmetry, fluid domain, shroud and vanes.
Please, is there any other thing to set under BCs.
I am using density based solver for the simulation.
The area I am still battling with since am still new in CFD is setting convergence criteria.
I have tried adjusting under relaxation factor but that only lowered residual but did not help with convergence.
@ Rahid Naseer: I read through the later part of your response where you a talking about convergence
Then go over the convergence methodologies like relaxation parameters or helping computer by manually putting the conditions to avoid possible errors. For example, we may write a condition that if a parameter is exactly zero, consider it a very very small value equivalent to zero but not exactly zero etc. Also you can direct the convergence to a certain solution if you know exactly the range where the solution should go. For example, if you expect a solution as positive, put a condition to ignore negative roots etc.
Please, how do I go settings manual convergence criteria for the system to treat zero as small value or ignoring negative root?
- Badges
- Science topic