Models for predicting gear pump efficiency

The research article 'On the assessment of lumped parameter models for gear pump performance prediction' has been published in Elsevier journal Simulation Modelling Practice and Theory (Volume 99, February 2020, 102008).


The present work describes a statistical approach for the assessment of discrete models for gear pump efficiency prediction. A critical discussion is performed on the input data assumptions that are commonly adopted to carry out the analysis, with particular regards to the actual bearing clearances and casing radial clearances. The proposed model adopts well-established techniques for simulating the pump fluid-dynamics in association with a novel approach, which allows us to take into account the effects produced by the gearpair micromotions. Moreover, the possibility to study both spur and helical gears, as well as non-unitary transmission ratio gearpairs, has been included, in order to ensure the wide applicability of the model in modern design solutions. Measured data obtained from an extended experimental campaign, involving 20 nominally identical samples of the same pump design, are used to establish the assessment procedure. Each sample is geometrically characterized by measuring the actual clearances at the end of the production process and then tested at 14 different working conditions, leading to 280 tests. The entire set of test conditions is then adopted to carry out a trace-driven simulation analysis, showing that the lumped parameter approach may reach different levels of accuracy depending on both the analyzed working conditions and the simulated pump samples.

The results underline that reliability and accuracy of this kind of model should be evaluated with respect to a population of pumps, defined on the basis of a statistical approach, since referring to a single pump sample may easily lead to an over/under-estimate of the quality of the proposed model. In addition, they also demonstrate that real clearance values need to be included in the model to obtain high fidelity estimations.

Access the complete article on ScienceDirect.