Many calderas show remarkable unrest, which often does not culminate in eruptions (non-eruptive unrest). In this context the interpretation of the geophysical data collected by the monitoring networks is difficult. When the unrest is eruptive, a vent opening process occurs, which leads to an eruption. In calderas, vent locations typically are scattered over a large area and monogenic cones form. The resulting pattern is characterized by a wide dispersion of eruptive vents, therefore, the location of the future vent is not easily predictable. We propose an interpretation of the deformation associated to unrest and vent pattern commonly observed at calderas, based on a physical model that simulates the intrusion and the expansion of a sill. The model can explain both the uplift and any subsequent subsidence through a single process. Considering that the stress mainly controls the vent opening process, we try to gain insight on the vent opening in calderas through the study of the stress field produced by the intrusion of an expanding sill. We find that the tensile stress in the rock above the sill is concentrated at the sill edge in a ring-shaped area with radius depending on the physical properties of magma and rock, the feeding rate and the magma cooling rate. This stress field is consistent with widely dispersed eruptive vents and monogenic cone formation, which are often observed in the calderas. However, considering the mechanical properties of the elastic plate and the rheology of magma, we show that remarkable deformations may be associated with low values of stress in the rock at the top of the intrusion, thereby resulting in non-eruptive unrest. Moreover, we have found that, under the assumption of isothermal conditions, the stress values decrease over time during the intrusion process. This result may explain why the long-term unrest, in general, do not culminate in an eruption. The proposed approach concerns a general process and is applicable to many calderas. In particular, we simulate the vertical displacement as occurred in the centre of Campi Flegrei caldera during the last decades, and we obtain good agreement with the data of a leveling benchmark near the center of the caldera.