Conference Year



PNMPC, Hammerstein model, Compressor test, Nonlinear control, Predictive control


This paper discusses the implementation and presents the results of a suboptimal nonlinear model predictive controller used to control the suction and discharge pressures of compressors under test in a rig. The objective of this rig is to emulate operational conditions to which refrigeration compressors can be subjected when applied in a refrigeration system, such as household refrigerators and freezers, and allow quick measurements of some of the compressor characteristics under those conditions. There is a coupling between suction and discharge pressures and the behavior of such variables is nonlinear with respect to the valve openings, thus the plant to be controlled can be characterized as multivariable and nonlinear. Even though in industry it is common to use linear controllers to control nonlinear plants, the use of nonlinear controllers can bring advantages in terms of performance and robustness. The controller implemented in this paper is the practical nonlinear model predictive control algorithm, which is a general framework that can be used for the implementation of nonlinear model predictive controllers considering almost any class of nonlinear model. Even though model predictive control is harder to be implemented than classical controllers, such as PID, it poses the process control problem in the time domain, so the concepts involved are intuitive and at the same time the tuning is relatively easy, even for the multivariable case. In addition, model predictive control allows constraints, such as valve opening limitations and pressure limits, to be handled during the design phase. This paper considers a specific nonlinear model architecture, the nonlinear Hammerstein model, which is composed of a static nonlinear element in series with a linear dynamic part. Since this model is conceptually simple and presents good results in most of the practical situations, it is widely used in practice when a nonlinear model is desired. The dynamics of the real test rig were identified using this nonlinear model structure and the identification results are discussed. The practical nonlinear model predictive controller was implemented in the real test rig, being tested in a variety of operating conditions. The results of the controller are compared with the ones obtained with a classical PID controller. The modeling approach presented good results and the results obtained in this study show that it is possible to use nonlinear model predictive control algorithms in refrigeration test rigs, and that this use can contribute to increasing the productivity and operational efficiency of compressor tests.