I have modelled a Linear Elastic composite beam with dimensions 355x355x4038 mm. The upper layer is a 50 mm thick concrete layer and has a E-value of 40 000 MPa (v=0.17). The lower layer is a 305 mm thick base layer and has a E-value of 207 MPa (v=0.4). A pressure (of 100 kPa) is applied in the middle of the beam over a width of 152 mm. C3D20 elements were used.

When the beam is simply supported there is a small strain (E11) discontinuity at the interface from 117 microstrain (concrete layer) to 123 microstrain (base layer). When the beam is fixed at both ends and prevented from deflecting vertically, the stress (S11) at the top of the base layer is negative (and becomes positive with depth), while the strain at the top of the base layer is positive.

I have modelled the same beam with two parts using a tie constraint. I get the same results.

Why is there this discrepancy? Any help would be appreciated.