Optimize cross section buckling

I am making a comparison between different beams and looking at the lateral buckling. I have a beam on 2 supports, 3 supports, 4 supports and 5 supports. the length of the fields between the beams is always 8.95 meters and the loads are always 77 kN/m. When optimizing the cross-section, it always displays a larger profile because the Mcr is getting smaller:

2 supports: HEB500
3 support : HEB800
4 support :HEB900
5 support :HEB1000

I think this is not correct because a beam at several supports gets a small buckling length, which makes the Mcr bigger. How is this possible?

ligger-2 steunpunten verdiepingsligger lijnlast optimalisatie goed.gh (71.0 KB)
ligger-4 steunpunten verdiepingsligger lijnlast.gh (81.7 KB)
ligger-5 steunpunten verdiepingsligger lijnlast.gh (83.8 KB)
ligger-3 steunpunten verdiepingsligger lijnlast.gh (77.3 KB)

Dear @b.hunen,
the problem is related to the way how Karamba3D estimates a beam’s buckling length (). Up to Karamba3D 1.3.2. build 190918 the buckling length of a beam was calculated as the distance between neighboring nodes which connect to more than two elements - or double that value in case one end is free. This explains why Mcr gets smaller when the total length of the beam increases.
In the new build of Karamba3D (see www.karamba3d.com/nightly-builds) supports are included as endpoints in the calculation of buckling lengths (see here).
The automatically calculated values of the buckling length may be checked by disassembling the beam-information. They show up as negative values. Positive values signal a user supplied buckling length:

– Clemens