RIGOROUS MODELING IN DYNAMIC REGIME AND PREDICTIVE CONTROL OF COMPRESSIBLE FLUIDS TRANSPORT SYSTEMS INTEGRATED TO CENTRIFUGAL COMPRESSORS.
natural gas network, centrifugal compressors, surge avoidance control,
model predictive control.
With the worldwide increase in energy demand, natural gas has emerged among other fossil fuels. Once it has a lower environmental impact and a better energetic efficiency, pipeline networks can distribute it. Thus, many authors have developed techniques to predict network gas behavior to increase process performance and competitiveness. Despite that, the usage of more complex centrifugal compressors for network modeling is lacking in the literature. Such equipment plays a pivotal role in network optimization and control applications once they provide degrees of freedom and limits the compressor map process. Of such limits, the most significant one is the surge line, which coincides with the maximum isentropic efficiency and limits the process to stable conditions. From these assumptions, this work contributes by describing the compressor map using a phenomenological model able to predict the gas properties with an equation of state. Thus, the pivotal contributions in this thesis can be summarized as follows: (i) to evaluate the influence of the phenomenological compressor map approach in natural gas network transient simulation and compression systems control, (ii) to analyse the influence on the equation of states in simulations, as well, in the transient states of compressor map, and (iii) to propose surge avoidance control strategies for the compression systems in both lumped parameters and distributed parameters models. The open-loop simulations showed a significant impact of the modeling hypotheses (compression system model and equations of state) on the mass flow and the discharge pressure of the network. In addition, a significant impact of these hypotheses on the surge avoidance constraints was observed, which later proved to be relevant for the efficiency of the process in closed-loop simulations. In addition, an IHMPC with the extended model strategy was proposed with feasibility assurance via slack variables to control compression systems. This strategy proved to be promising compared to a conventional NMPC by having a similar response to it but a significantly lower computational time. In addition, the proposed control law is efficient in simulations of the control of systems while keeping the process in safe conditions in both modelings: lumped and distributed parameters, thus presenting itself as a good option for in loco implementation.