Dear Qas, without seeing your model and at least a sketch of the experiment, I can only guess - see also Troubleshooting, 2.1.1. Please also send a graph comparing the measured and calculated Load-Displacement curve.
Moreover, it is not clear what information you have about the material?
Ad 1.: Your observation suggests compression (or shear) have significant influence. This might be happening in the experimental setup, however, speaking about direct tension, I suppose a problem in representing the boundary conditions (or maybe a somewhat problematic or unusual tensile test setup).
Ad 2.: As you speak about plain = unreinforced concrete, I see no reason for modelling shrinkage as there are no bars acting against the volume change.
Ad 3.: If unsymmetric localization is responsible for the difference, you should model the whole specimen.
Sorry, without seeing your model and having some information about the real setup (see above), I can not recommend any parameter changes. I assume you have already read Troubleshooting, 2.2.18 My analysis results do not match the experiment/expectations. How can I improve my model?