So I have a structure with a pile foundation and a gravity load acting on the structure. The demand curve is drawn by running an elastic analysis. Pile-Soil is modelled as p-y, t-z, and q-z curves. When i run a pushover analysis that starts after the gravity nonlinear case ends, i find a high loss of initial stiffness. However, if pushover starts with zero initial conditions, the curve almost follows the elastic demand curve. Can anyone provide any reasoning behind this phenomenon? Note that all layers of soil mobilize in skin friction under gravity non-linear case.