I am not sure how it is possible to get a softening response under load control (Pressure) in ABAQUS. I have applied surface pressure load on 3D shell element in Explicit analysis. What can be a possible numerical backgroud?
you should pay attention when choosing your integration method for shell element. Can you give more details about your model ; (type of element, integration method ?
More information is needed. I assume by a softening response, you're referring to strain increasing non-linearly with pressure e.g. if you have a linear ramp of pressure the resulting strain curve is non-linear and convex. Without any knowledge of your constitutive model, load-time curves, masses, etc. it is impossible to say. Plasticity would be an obvious source if that is included in your constitutive relation. Likewise if there are viscoelastic or poroelastic materials you may be simply observing creep. Finally, since this is an explicit dynamic analysis inertial effects will play a role, as will the temporal properties of how the loads are applied (rate, smoothness, etc.).
To troubleshoot, you may try switching to displacement control to see if you again see the softening response (this time in the stress being non-convex). Also, if possible an implicit, static analysis would eliminate inertia and time dependence (assuming materials are elastic).
That's likely your answer. Is the total deformed area of your shells 20% less than the total undeformed area? In that case, the total reaction force must also be 20% lower to maintain the initial surface pressure calculated at reference.