There can be several reasons for this. The overall behavior depends on the definition of hinge characteristics and the location and number of hinges. In many cases, if everything else is fine, the analysis might stop after at certain locations the hinge reaches the peak (just before the descending branch initiates) due to instability of analysis. You should try using "Restart using secant stiffness" approach for re-distribution of forces/moments once the hinge reaches its peak. Also try reducing the step size and using more iteration steps.

This problem has often arisen because of the high rigidity of the shear wall which prevents the structure from reaching ruin.

Thank you for reply. Time period seems to be reasonable. So, I don't think high stiffness is the problem in my case. Do you have any other suggestions?