Different results from resumed run of propeller in AMI

Summary

Hi, I am running an unsteady simulation using pimpleFoam of a propeller rotatibng at high speed. Obj is to get the thrust for validation. I use AMI so I have a domain. In the domain, there's a cylinder and a propeller within it. Similar to the propeller tutorial. I ran it for 24hr, which is the max time allowed on our cluster. Time simulated is 0.05s. The thrust oscillates slightly around 3.9N. I then try to resume the job to run for another 24 hrs. When I look at the new results, I found that the thrust oscillates much more, going a bit heywire. To test if the error is due to resuming, I resumed my job from 0.025 and then compared the results. The resumed case is very different from the non-resumed case.

Steps to reproduce

Run the simulation for certain duration. stop. Resumed run. The results obtained is different from the non-resumed case.

Example case

Not possible to attach case due to special propeller configuration. But if it's realy a bug, I can create the case with a new propeller.

What is the current bug behaviour?

Different thrust results when resumed. I have attached the cd output

What is the expected correct behavior?

Same thrust results whether resumed or not

Relevant logs and/or images

Environment information

• OpenFOAM version :2206 and 2312 tested
• Operating system :redhat
• Hardware info :
• Compiler :gcc

Possible fixes