Manufacturing analysis and design optimization of irrigation pumps for Bangladesh

Abstract

Irrigation pumps are indispensable to the production of major crops in Southeast Asia, especially in Bangladesh where agriculture plays a dominant role in the economy. This thesis first analyzes the manufacturing and performance of prototype irrigation pumps, which are manufactured in Bangladesh using a Thai mixed-flow pump model. Then this thesis optimizes the design of the current model in a cost-effective and energy-efficient manner. This study is based upon several fundamental tenets of fluid mechanics. It begins with the definition of specific speed, which is critical to the pump selection among axial-flow, mixed flow, and centrifugal pumps. The study also discusses friction losses and the Affinity Law concepts, which are incorporated in the analysis of major and minor losses and of scaling at different RPMs. This study proceeds through multiple stages, including design of experiments, full-scale sample testing, prototype testing, computational fluid dynamics (CFD) simulation in SolidWorks Flow Simulation and ANSYS CFX, and results analysis. The research also involves several advanced techniques such as rapid prototyping, reverse engineering, computer-aided design (CAD), CFD, and analysis of variance (ANOVA) during these stages. During this study, the results of prototype and CFD simulation demonstrate good agreement with full-scale testing. From this study, several significant outcomes are generated and presented. The six manufactured samples exhibit geometric differences with an average standard deviation (STD) of 3.6%. These differences have only a small effect on pump performance, with STDs in shut-off head at all RPMs of less than 3.0% in prototype testing, 4.4% in SolidWorks simulation, and 5.6% in CFX simulation. The study implements the friction loss model to the pump system. For full-scale testing rigs, the loss coefficient is found to be 21.75 with a STD of 1.74. For prototype testing rigs, the loss coefficient is found to be 14.2 with a STD of 3.19. The thesis also implements Affinity Law scaling at different RPMs. This is supported by curve fits of the data with R-squared values of greater than 0.9 for flow rate and 0.99 for shut-off head. After that, a method of pump selection is presented for customers to find the most energy efficient pump and its operating condition. Finally, a design optimization of the four major design parameters, along with the significance level of each, is suggested for the current design of Thai mixed flow pump.