Apart from what you have done and assuming that you have done manual and near perfect PA/Flow calibrations :
- try to rotate on angle, not much space in your print but … .
- Try to bump the bed to the max 110C
- Reduce the fan speed to the minimum which does not affect your print it seems that you can do it bellow 20 or even completely off for the speeds and model you are using ( assuming aux and chamber are off )
- Check your bed if flat and not curved after you preheat it to 100C and something straight you can use sheets of paper under to measure i had 0.6mm(fixed)
Check your flow in the sliced model and try to make it the same(unified). not sure how to do it perfectly in OrcaSlicer/Bambu . But with speeds and line widths you can be make it close - Try using the same layer heights and same line widths and same speeds all around. This should be able to make the flow unified can be checked in the sliced model , e.g. a bit more unified cooling
Bringing down the speed to 70mm/S helped for some of the models - as suggested reducing the Print temp helps , but for me that caused more problems as getting to a place with bad layer adhesion and in my tests was not a great improvement compare to the other things i tried. Best outcome for me was with unified flow, reduced fans, reduced speed, blanket and slowly reducing the bed temperature after the print
- increasing or decreasing infill percentage is worth trying as well , medium success on some prints in my case with increasing
The first layer speed is not good enough as the problem happens from the top layers when not unified shrinkage(cooling or different flow) is happening. For some prints i was able to overcome the problem by reducing the print speed bellow 70mm/S and minimum or no fan, Also the chamber is releasing the temp unevenly so a i did a thermo blanket around
While it prints check if the build plate is not lifting , magnet is not strong enough sometimes if too much warp , some binder clips helps
Unfortunately no silver bullet here , similar problems on some prints for months some was able to resolve some changed the print
There are some threads to improve increase the wait time with heat Bed-on after the print is finnished you leave the bed on and slowly bringing it down ,as well as actively heated chamber . found that M191 with 110C and 5C decrease every 5-10 minutes helps on some prints.
Please advise how if you solved it and what worked for you i solved it for most models but time to time is still a problem , but have not tried the active chamber heating and the AUX( part fan) modifications
Thermo camera on the heat bed and on the model can help you to identify variations , i am borrowing one from time to time to see how bad it is . Also on the build plate i see more than 10C initially variations especially around the edges and even after soak time 4-5C variations and more
EDIT: changing the infill Gyroid and grid did not make for me much difference but changing the infill orientation did in some cases , which was similar result to rotate the object for some prints