In the present work, we determine how three well documented models of the magnetospheric electric field, and two different mechanisms proposed for the formation of the plasmapause influence the radial distance, the shape and the evolution of the plasmapause during the geomagnetic storms of 28 October 2001 and of 17 April 2002. The convection electric field models considered are: McIlwain's E5D electric field model, Volland-Stern's model, and Weimer's statistical model compiled from low-Earth orbit satellite data. The mechanisms for the formation of the plasmapause to be tested are: (1) the MHD theory where the plasmapause should correspond to the last-closed-equipotential (LCE) or last-closed-streamline (LCS), if the E-field distribution is stationary or time-dependent respectively; (2) the interchange mechanism where the plasmapause corresponds to streamlines tangent to a Zero-Parallel-Force surface where the field-aligned plasma distribution becomes convectively unstable during enhancements of the E-field intensity in the nightside local time sector. The results of the different time dependent simulations are compared with concomitant EUV/IMAGE observations when available. The plasmatails or plumes observed after both selected geomagnetic storms are predicted in all simulations and for all E-field models. However, their shapes are quite different depending on the E-field models and the mechanisms that are used. Despite the partial success of the simulations to reproduce plumes during magnetic storms and substorms, there remains a long way to go before the detailed structures observed in the EUV observations during periods of geomagnetic activity can be accounted for very precisely by the existing E-field models. Furthermore, it cannot be excluded that the mechanisms currently identified to explain the formation of "Carpenter's knee" during substorm events, will have to be revised or complemented in the cases of geomagnetic storms.