Jouku's reasoning is sound. The Pressure Thrust (P*A) exists on either side of the fabric joint. The inside "diameter" area force goes to the equipment or pipe/ducting on each side and must be restrained by those items. For example on a HRU (large Heat Recovery Unit) the vendor has physical stops (round not slotted holes) at one of the foundation column lines on either side of the fabric joint location.

If you put a fabric joint on the inlet to the HRU, then the pipe/duct must consider this thrust force in the pipe/duct system. The Flange will see the net bellows area thrust ("convolution" area) distributed around the flange face.

If the stress engineer misses or does not understand this, the system may have survived without much blow apart deflections if the items on either side of the fabric joint were heavy and the frictional forces limited the perceived "perpetual motion machine" deflections.