I am also doing MPA analysis on a steel frame with b31element for beams and columns and S4R shell elements for steel shear wall (3D). Well, in my point of view the right forward procedure is to first apply deal a live loads through subsequent general static steps, and then use a general Riks step to perform a load-displacement control analysis! however I am experiencing convergence issues in Riks step after applying dead and live loads! the structure does not deform toward the target displacement anymore! it will go into irrelevant deformed shapes! however I need to apply dead and live loads too! if I perform the Riks analysis alone, everything goes well, somehow it seems the Riks step cant find the correct stability path in case of an initial imperfection generated by applying dead and live loads! or maybe the deformed S4R shell elements trigger this convergence issue in Riks analysis?

Any suggestion on this issue is a great help and is highly appreciated! thank you for your time!

Your problem seems interesting, please share the file if you can.