It's probably your expansion joint thrust load. If you enter an effective ID for your XJ and you have pressure on your XJ spreadsheet, CAESAR II will apply a pressure thrust load on either end of the XJ (pressure times effective bellows area). If you have tie rids in your model, that thrust load would put your rod(s) in tension and little pressure thrust load will bleed into your nozzle. But without any tie rod in the model, that thrust load simply pushes on your suction nozzle. This load, in the real world, does not find itself on the nozzle, it is on the impeller and casing behind it.
If your XJ is intended to compensate for axial growth, then tie rods are not part of your design. If, however, the XJ is installed for its lateral compensation, then tie rods can be used to carry the pressure thrust load.
If you want pressure thrust in your untied model, place a force on the first elbow upstream of the pump and pointing away from the pump and assume the counteracting pressure thrust (pointing towards the pump) is carried by the pump base. More sophisticated modelling would have you break the thrust load into two components - 1) based on the inside pipe area and located at the upstream bend 2) based on the effective area minus the inside pipe area and located on the upstream end of the XJ.
_________________________
Dave Diehl