I would consider attempting to place XS’s for each structure using the XS 1 through XS 4 convention, and perhaps use XS 4 of the pedestrian bridge as XS 1 of the roadway bridge. It’s usually just a difficult thing to model, with such limited space.
Take a look at the bridge modeling methods, if you don’t like the way the results look through the structure (e.g. severe drawdowns in a steady flow analysis). I would recommend not worrying about the skew of the bridge(s) until you’re okay with the results of a straight run. Then, constrict the opening based on skew as deemed appropriate.
Let me know if this isn’t helpful,
David