Assuming the expansion joint is infinitely stretchable, the expansion joint itself would expand out nearly to F=kx where F=PAaff, if it's just an expansion joint and two blinds.

If we push both ends back together to initial state, we need to exceed and then meet F=kx.

If we want to push it further - such as taking on thermal displacements from pipe, then we have additional x to include into our analysis.

Of course, our metal isn't unobtainium, and it will also be inclined to flex. So the expansion joint will absorb some fraction, the pipe will absorb some fraction, the vessel wall will absorb some fraction, and the bending from nozzle to vessel base will absorb some fraction.

They're all flexing together in series, so 1/keq = 1/k1 + 1/k2 + 1/k3... thus resulting in a total load somewhat less than P*A+k(expansion joint)*x(thermal).

In the case of tie joints, you have a non-linear scenario where F=kx if x is compression and F=(k1+k2)x when in tension, where k1 = rubber and k2 = tie rod.