Solar thermal plants are commonly constituted of different subsystems, such as the solar collector field, accumulation tanks, and gas heaters, to enhance plant performance. Nevertheless, from a process control perspective, the changes in the subsystem's configuration can cause interactions and compromise the controller performance. Therefore, this work proposes a hybrid practical nonlinear predictive control for a solar thermal plant facility, aiming to improve the plant operation performance by considering mixed-logical dynamic models and including the process constraints as linear mixed-integer inequalities in a single control layer. The proposed strategy is compared with the conventional practical nonlinear control with an external decision-maker. Both frameworks are simulated under actual process circumstances using real meteorological data and validated hybrid models of the CIESOL facility (Spain). This paper makes relevant contributions in some aspects. From the control framework perspective, a new formulation of the hybrid control algorithm simplifies the control structure, excluding the decision-maker structure from its task. Moreover, the present study considers variations in load demand, a novel and significant contribution. The results demonstrate that the hybrid nonlinear controller performs better than the conventional approach, with a reference tracking error index approximately 35% lower in a sunny day scenario.