In many instances, pressure causes no structural response in CAESAR II analyses - exceptions are with long pipelines and FRP pipe where axial stiffness is low enough to develop significant system deformation. You can consider this structural response due to pressure in other systems by turning on the (axial) "Bourdon effect". On the individual component level, untied expansion joints can also develop axial distortion due to pressure. In the typical CAESAR II model, the force of the pressure times the entire, effective bellows area is applied on either end of the joint rather than on the internal surfaces projected up- and down-stream of the joint. I believe there are several Forum posts on this application.
I suggest you run a load case with pressure alone. The results will show the CAESAR II nozzle loads. My guess is there will be a compressive load on your nozzle if you entered an effective ID (and pressure) for that XJ. But, as you say, this load includes the calculated force carried by the back of the pump housing. To get a more accurate estimate of the nozzle load, I suggest you break the XJ's axial pressure load into two groups: 1) the load based on the ID of the pipe, located on the back wall of the pump housing (or on the impeller), there is a similar load upstream of the XJ, probably on a bend, and 2) the differential of the effective XJ area and the inside area of the pipe applied on either end of the XJ. If you call these loads "F1", the pressure-alone case would now be "P1+F1". If this works for you, then a good load case for your operating loads on the pump would be something like W+P1+T1+F1.
That force due to pressure may either increase or decrease the total load on your nozzle.