Found this post on the net which I think is not properly solved.
It is based on NAFEMS LE3: HEMISPHERICAL SHELL linear elastic benchmark.
Prepomax result is 185.1mm compared to expected result 185mm.
Benchmark or compares between one to each other are not easy task, someone need carefully to be sure the element type, formulation and boundary conditions are exactly the same.
Even i do a lot of studying problems, still prefer using “test and example” names instead and left question mark when some discrepancies exist. Sometimes i do contact directly to the developer and got replied about clarity and advices.
*edited
it seems latest version of CalculiX solver treat node restraint in local coordinate system of shell element, not as previous version which needed to define locally by transform. Probably, link of benchmark above ignoring them.
below an example were previously a problem in symmetry condition without transform assigned for each of vertical, horizontal and curve edges.
