Experimental Testing, Subsystem Model Validation, and Design of a Variable Displacement Hydraulic Motor

Abstract

The design, testing and model validation of a novel variable displacement hydraulic motor is presented in this Thesis. This novel, radial piston, low-speed high-torque motor achieves variable displacement using a 7-bar 2-degree of freedom mechanism that infinitely varies the piston stroke. This motor is called the Variable Displacement Linkage Motor (VDLM). One potential application of the VDLM is in the propulsion drive of off highway equipment such as a compact track loader. Efficiency is improved at a component level by designing a motor that is inherently efficient and at the system level by adding the additional degree of freedom of variable displacement. Varying motor displacement allows reduced flow demand in the hydrostatic loop at high motor rotational speeds while maintaining high torque output at low rotational speeds. This novel architecture is beneficial because currently available solutions for traction drives either do not feature continuous displacement variation or require a gearbox. Nearly all off-highway equipment is currently fueled with diesel or gasoline engines. With the push to decarbonize these energy sources, fluid power systems that can more seamlessly integrate into hybrid electric or fully electric powertrains are in demand. The VDLM technology has significant potential to fill this market gap. A series of models were created to thoroughly predict the performance of the VDLM. A Multi-Objective Genetic Algorithm (MOGA) was utilized to create a Pareto front of potential mechanism geometries and a comprehensive Matlab code analyzed each solution for validity. From the Pareto front, a prototype was selected from which to build a single cylinder prototype. Displacement of this prototype was 42cc/rev and featured 6 discrete displacement settings from 50% to 100% fractional displacement. The purpose of this single cylinder prototype was to collect experimental data to validate the models. Modeling is a closed-loop process, and experimental data was required to validate the models. In the case of the single cylinder VDLM, mechanical losses were the primary focus. Losses were modeled and grouped into five primary sources: main taper roller bearings, case windage, shaft seal, valve actuation, and linkage. A staged-assembly method was developed that allowed individual loss components to be isolated and each model was validated independently. Two test stands were designed and assembled for the purpose of collecting the energy loss data. Instrumentation, such as pressure sensors, a torque transducer, and a piston position sensor, collected data at a sample rate of 10kHz and was recorded on a DAQ system for future analysis. Low magnitude torque tests were conducted on the test stand at the University of Minnesota and high magnitude torque tests were conducted on the Milwaukee School of Engineering test stand. Model predictions and experimental results were compared side-to-side. Correlation was improved by altering physics-based model parameters such as the coefficient of friction, viscosity, and radial clearance in the valve. The final validated model prediction achieved correlation to the experimental data to within 10% error when averaged through an entire piston cycle. Data analysis of the pressure dynamics inside the cylinder was conducted to qualify the valve design. Design revisions were required during the testing phase to make the single-cylinder prototype operational. The valve return spring force was increased, a linkage over-travel stop was designed and installed, and the valve timing was corrected. A failure of the VDLM occurred at 19.3 MPa pressure differential testing. A thorough teardown of the VDLM was conducted following the failure to determine the root cause. The valve was seized in its bore which was attributed to deflection or debris in the valve bore. Valve timing was interrupted due to the valve sticking and a hydro lock condition occurred. This pressure spike exceeded the clamping capacity of threaded fasteners in the cylinder block and caused an O-ring to fail. The failed O-ring allowed high pressure fluid to enter the VDLM housing which overwhelmed the case drain port and caused the main shaft seal to blow. Linkage components also failed due to the over pressurization. Wear throughout the motor was analyzed during the teardown. From the testing and tear down, design guidelines for future VDLM’s were drawn. Utilizing the validated models from generation one, a second generation VDLM was designed. The goals for the generation two VDLM were to be smaller in overall diameter, include live, continuous displacement adjustment control, and feature a full array of pistons and cylinders. The purpose of this motor was to conduct hardware in the loop testing. Constraints such as constant top dead center through all displacements and the requirement for the roller follower and rocker pivot axis of rotation to be coincident were relaxed to achieve smaller motor diameters. A Pareto front of potential motor geometries was developed, and a 5-piston 7-lobe individual was selected. This motor featured continuously adjustable displacement from 50% to 100% fractional displacement and displaced 300 cc/rev. The detailed design of the mechanism is described in this thesis from an early conceptual phase to a final phase where part prints are developed. The preliminary VDLM solution was reviewed for validity; layouts were determined, bearing sizes were validated and a CAD model was developed. Fits and tolerances for the linkage were determined and materials and surface finishes selected. Linkage design was validated using Ansys Finite Element Analysis (FEA). Because the VDLM is a complex assembly, geometric simplifications were made to create a more basic CAD model of the linkage for the purpose of the FEA study. Simplifications included that a single cylinder block and linkage assembly was created, a symmetric boundary condition was applied, and the position where the linkage experienced the highest internal loads was determined. 35 MPa of fluid pressure was applied to the piston and an FEA solution was determined. Two metrics were used to determine design validity, deflection, and stress. Finally, detailed 2D part drawings were developed, and custom manufactured parts were ordered and assembled. Part assembly and fit-up was critiqued to close the solution development to implementation loop.