Thanks for the clarification Pawel, but should the load be reversed when the Z-axis of the panel is reversed? even if it is not planer? or the program does not understand that?
Also how to make sure that the panel is planer or how to fix that or check that?
After changing local Z of the panel the load will be reversed but for your original file the effect of opposite load direction with/without description will still be observed - because of no planarity of panel resulting in opposite Z for finite elements than for panel.
The sensitivity to such small planarity inaccuracies was reported to developement team to be analyzed and eliminated.
Planarity can be corrected:
1/ or correcting to plane from Edit>Detailed correct
2/ or extending the number of decimals and correcting coordinates of some points of contour
kudos for adding this info
Using "rounding add-in" of Rafal results in bigger "curving" of panel instead of "flattening" it but it helps too.
The effect noticed by mgaafar can be observed for some range of panel unplanarity.
Since almost no one works with warped panels, I think there should be an option that checks all panels if they are planer or not and then fix the warped ones automatically instead of checking and fixing panel by panel.
consider that I had this problem (you can see the first post) more than one time and apparently in a random manner applying hydrostatic load on curved shell.......sometimes is not possible to flatten a cylindrical tank!!!!
I agree that it should not be necessary to eliminate any planarity inaccuracies - especially that sometimes (slightly twisted roofs, intersections of cylinders, etc) they cannot be avoided.
As I have mentioned in my previous answer:
"The sensitivity to such small planarity inaccuracies was reported to developement team to be analyzed and eliminated."