I am analyzing a built up thin walled steel I-Beam formed by connecting two channel sections back to back. So there is an obvious large surface in contact. The model has geometric as well as material non-linearity. Upon analyzing the problem using the General, Static Step, the analysis aborts in the first increment itself. I have tried damping as well as reducing the initial increment size, but nothing works. However, referring to a research paper, the author has solved this issue by decomposing the Step into two- In the first short Step, an initial small displacement is used to bring the Channels into contact and in the next step a displacement controlled loading is applied. How do I decompose the Step into two? Is there any other possible solution?

Any help will be greatly appreciated.