This work deals with analysis and maintenance of steady-state performance of model predictive controllers. The aim is to use information about input and output constraints of the ideal controller and of the controller actually applied by the operators – some of them are either loosened or tightened based on actual operating conditions. This can have impact on profitability of the process. The procedure is applied to a simplified model of controller implemented at production unit at SLOVNAFT, a.s. refinery. The studied controller processes the total of 47 manipulated, disturbance, and output variables. The analysis provides information on how to move the constraints of the controller to reach the optimal operating point. Python programming language is used to create the application with the graphical user interface that is actually used at the refinery. Acknowledgment: this research is funded by the Slovak Research and Development Agency under the projects APVV-21-0019 and APVV SK-FR-2019-0004, by the Scientific Grant Agency of the Slovak Republic under the grants VEGA 1/0691/21 and VEGA 1/0297/22, and by the European Commission under the grant no. 101079342 (Fostering Opportunities Towards Slovak Excellence in Advanced Control for Smart Industries).