Dear  Fatima Khalid  ,  I am not sure about the load conditions. Is it tensile / compression? Assuming the compressive Forces are below and the tensile Forces are upwards, the Explanation is rather simple. You are pulling in which the tensile Forces of the steel reinforcement determine the resulting Forces for a given elastic strain. When compressing your RC part mostly the concrete takes over the load. You defined different material properties for tensile and compressive loads for the same elastic strain. The Response by the steel and concrete on compression for a given strain leads to higher Forces than for the same tensile strain in which mostly / only the (softer or smaller) steel reinforcement cross-section is taking over the load. Most likely it's all o.k.  

Dear Ms Fatima Khalid and Mr. Rudiger Rentsch

Hello,

It seems to me, in general, computational methods are approximations of the true mathematical models that coincide in solution on the exact solutions of the mathematical models only in the limit of convergence, e.g. very small values of the algorithmic parameters. Accordingly, such as you can see in optimization of symmetric problems and structural systems, the approximately computed solutions need not to result in symmetric solutions. They should be symmetric in the limit of convergence and they would be symmetric provided special consideration in the computations. However, the considerations may sacrifice the efficiency considerably. In brief, convergence to the exact solution is the main point in approximate computations. I have not checked your problem (sorry, the question seemed interesting and I need to leave soon) however in many cases symmetric behavior can be forced to the mathematical model in advance before implementation of a computational method. Have a nice day.

Check strain softening???