I assume that the problem can occur because of the following reasons:

1) both splicing bars are located within same "FE layer" of concrete. In other words, adjacent nodes of rebars are connected to the same concrete FE with master slave conditions (and bond as well). Try to place them in different concrete layers
2) define bond properties to overall length of bar instead of short part only
3) double check bond properties. make a small pull out test with the same bond to check whether this behaviour is what you expect
4) i would advice you to use ATENA Science. I believe it doesn't affect to result in this simple case but it is better for your future tasks and the user interface is much better

can you give detailed description of the experiment with material properties?

First of all, there is no sense in using Arc Length method in your problem. Please check Theory manual for further details. Just in a few words: define prescribed displacement to steel frame and change solution method to Newton Raphson.

Secondly, try to define another bond law. Theoretically you should get closer curve if bond law would be less stiff. So I would define something like bilinear curve with three points: 0mm-0MPa, 0.02mm-2MPa, 1mm-2MPa
In this case your ultimate force should be much less than with previous bond. If it is less than the experiment, try to adjust bond law to get closer. Moreover, you don't have an exact bond law and the curve given in MC is not an exact relationship for all cases. You have to modify it to see how it affects to result.