A modified theoretical model coupled with the nondominated sorting genetic algorithm II (NSGA-II) for performance prediction of centrifugal pumps

The accuracy of theoretical models for performance curve prediction has been a persistent challenge, particularly when operating conditions deviate from the rated values. Furthermore, improving the prediction accuracy for centrifugal pumps remains a focal point in this field of research. This research proposes a novel theoretical model that employs the energy-flow variation theory to derive the characteristic curve equation for centrifugal pumps. By utilizing the nondominated sorting genetic algorithm II for multi-objective optimization, the model optimizes the coefficients in the theoretical model, enabling precise calculation of the performance curves for head, power, and efficiency. Notably, the proposed model offers enhanced generality and effectiveness as it is not restricted by the geometric parameters of the pump's overcurrent components. Experimental validation was conducted on two centrifugal pumps with specific speeds of 64.1 and 128.2, yielding mean absolute relative errors of 1.44% and 0.59% for head, 2.9% and 4.13% for power, and 6.41% and 3.91% for efficiency, respectively. This research significantly contributes to the field of pump performance prediction models, aligning well with the requirements of engineering practice.