from the slope deflection curve it appears, the linear elastic part is very short lived (in terms of strain). ofcourse you'll observe variation in results with increment size until, you get more than 2-3 data points in the elastic zone. There's a term in FEA convergence which is time increment convergence. Keep on reducing increment size until you don't see much variation initial ealstic modulus.

P.s. if you notice, it appears the response goes to plastic in just one increment (very sharp transition from elastic to inelastic).

Hello Samuel,

Try increasing number of data points in the field output. Lets say 50 or 100 points in the given time period at equally spaced intervals. Try doing this and hope you can solve your issue.

Not sure I am answering your question, but in Abaqus , you can go in "Field Output Request", then you change "Frequency" to "From time points", then you can write a table where the time points at which you want the output are specified

In a purely elastic simulation, you can reach the maximum displacement in a single time step.

However, this is not possible in an elasto-plastic model.

If your simulation reach the yield at the first time step, you will have to reduce the timestep.

If the simulation reaches the yield point in the first time step, the solver can converge anyway, but the result is inaccurate because it becomes dependent on the time step chosen.

Hope this helps.

Nicolo