Sorry my net browser somehow was pasting the whole data in the body of the post and the formatting was all screwed up. Anyway I was able to download the file and took a quick look.
The structure was having very high displacements even for PERFROM ANALYSIS. So things were falling apart when PDELTA 4 ANALYSIS was carried out. This is due to the fact that the structure had low lateral stiffness with the columns bases being pinned and the moment releases applied to the connecting members. As a quick fix, I changed the MY MZ releases to partial releases.
A modified file is attached for your reference.
(Please visit the site to view this file)