On-Line Self-Tuning for Centrifugal Pumps Driven in Parallel Mode Using Dynamic Optimization

This contribution presents a self-tuning method for centrifugal pumps driven in parallel mode. The target is to minimize the energy demand by adjusting the rotational speed of each pump during operation and by detecting switch on/off points. One requirement to the algorithm is a save tuning process, thus the given setpoints have to be satisfied - which is solved by communicating control loops. A further issue is to manage varying setpoints - which is handled by a developed dynamic optimization algorithm. For minimizing the target value, the Nelder-Mead algorithm is used. The program is tested on the basis of flow rate control using a modeled pump system. This Matlab/Simulink based model is presented and verified. After that, the tuning program is also implemented in a PLC to optimize two pumps in a test rig.