From your analysis, linear behaviour seems relevant to experimental. From the load-deflection curve, it's always hard to find what's wrong as non-linear behaviour depends on several factors. I'd encourage you to check the boundary condition first, then by adjusting non-linear material property. After that, I'd go for checking the analysis convergence norm (load, displacement, energy norm).
As can be seen from the curve, up to the yield point your analysis is correct. Most of the times numerical analyses have a little bit more stiffness than experimental ones. For instance, the boundary condition is being taken into account ideally in numerical analyses, while it is not ideal in experiments.
For the second branch, there are lots of parameters playing an effective role. If your concrete is normal concrete without using any steel fibers, the mistake may stem from nonlinear material properties of concrete or steel rebars.
1-Be sure you have defined concrete damage plasticity (CDP) correctly.
2-If you did not take the strain hardening of the steels, it may affect somehow the second branch.
3- Change the size of mesh in Abaqus. (i.e decrease their size)
4- Be sure of the correctness of your material property.
If you did not reach a satisfactory result after checking these steps, I suggest changing your stress-strain concrete model.
hi..it is seen from your load - deflection curve that experimental and numerical results are considerably good up to peak load..after the peak load the numerical results shows with increase in strain there is a decrease in load..this is mainly due to many parameter but the main parameter is constitutive model used to describe the behaviour of beam..so which model is used to simulate the response of beam to the applied load..you might have used work softening model or since the beam is stiffer material it response to the load itself is nonlinear in nature..ok all the best..
This is could be due to buckling. In your experiment, your loading direction and the deflection could happen concurrently. But for CAE, you would have chosen an element that has spatial degrees of freedom. Even in case of slight error in geometry or lateral effects coming into picture because of degrees of freedom, which are not exactly same as in test, the structure is clearly buckling...
I think there are two or three issues in need to address in your FE model.
a. Your initial stiffness resulting from FE result is higher than the experimental outcome. Address this issue by giving appropriate input such as material modeling of concrete. May be modulus of elasticity of composite section (FE model) is higher than that of the experimental samples.
b. check your stress strain curve of the rebar. Sometime if rebar dia is small (say 10 or 12mm) it is observed that strength is much higher than anticipated value.
c. Strain hardening parameter steel need to be adjusted.