Search

Pump failure is a general concerned problem in the hydraulic field. Once happening, it will cause a huge property loss and even the life loss. The common methods to prevent the occurrence of pump failure is by preventative maintenance and breakdown maintenance, however, both of them have significant drawbacks. This research focuses on the axial piston pump and provides a new solution by the prognostic of pump failure using the classification of machine learning. Different kinds of sensors (temperature, acceleration and etc.) were installed into a good condition pump and three different kinds of damaged pumps to measure 10 of their parameters. Then, several classification models were trained by different machine learning algorithms including Decision Tree, K-Nearest Neighbor, and Neural Network. These classification models can detect the “unhealthy” state of the pump and locate the damaged parts from real-time data. Finally, the performances of these models were evaluated by different metrics including Training Time, Accuracy and etc. According to the results, for the prognostic of axial piston pump failure, the algorithm with the highest accuracy is Fine KNN; while the algorithm that has a good balance of training time and accuracy is Fine Tree; finally, the Deep Neural Network has a good accuracy and can be designed with different hidden layer structures, in addition, the trained net of the Deep Neural Network can also be reserved and retrained with new data.

Virtual prototyping of positive pumps and motors is incredibly useful, increasing speed of development and lowering cost. The calculations of fluids discharge pressure, displaced by a swashplate-type-axial piston pump/motor is dependent on the load impedance driven by the unit. The load impedance is estimated by a variable orifice found downstream of the unit. Similarly, the suction line impedance is also estimated with an orifice. The current project aims to evaluate the various loading architectures in their ability to quickly and accurately create the desired impedance. The robustness of the loading architectures was tested on a wide range of pump sizes, speeds, pressures, and displacements. The estimated initial conditions of the loading architectures, such as the orifice areas and line volumes and pressures are very efficient means of converging to the correct discharge pressure. To develop a model, based on past simulation, a full factorial design of experiments (DOE) was performed to generate the required training data set. The DOE revealed the models sensitivity to the normally unknown variables which contributed to large variations in accuracy. However, after analyzing the impact of other variables, an alternate approach was taken. By creating a variant of the simulation, which sets port pressures equal to line pressures, orifice areas can be effectively removed from the equation. In order to compare the performance, ideal valve plates specific to each code variant should be used. This is because arbitrary valve plates may favor one over the other, yielding extraneous conclusions. An existing genetic algorithm found the accuracy was unaffected while greatly improving speed.

Noise control is an important factor in evaluating the design of positive displacement machines. This research project aims to develop new tools in MATLAB, with emphasis on visual approaches, to comprehensively characterize the noise generated by positive displacement machines in spatial, temporal and frequency domains. Sound pressure level (SPL), sound intensity level (SIL) and loudness were calculated and plotted on a measurement surface surrounding the pump, which illustrates the spatial distribution of the sound field. In order to highlight the phenomenon within specific frequency bands, Butterworth filters were used to isolate desired frequencies, such that specific harmonic content or 1/3 Octave bands content can be analyzed. In addition to static visualization methods, videos were created by compressing a series of sphere plots in time to illustrate the dynamic characteristics of the measured sound. Corresponding phase plots were generated at the same time to statically illustrate these dynamic characteristics shown on videos. The successful generation of the various methods of characterizing and visualizing pumps generated noise was evaluated on a large sample set of more than 150 measure grids with nearly 30000 individual microphone measurements. The creation of these analytics has already changed the conversation and challenged the state of the art in hydraulic Noise Vibration and Harshness (NVH).

Hydraulic hybrids transmissions have the potentially to substantially improve the fuel efficiency of on road vehicles. In fact recent studies have demonstrated that this technology can improve fuel economy by upwards of 30% over competing electric hybrids. To further improve the fuel economy and performance of this technology a novel blended hydraulic hybrid transmission has been constructed at the Maha Fluid Power Research Center. While this novel hybrid architecture created by the Maha lab has many benefits over conventional systems, there are a number of control challenges present due to several discrete modes of operation. And though improving fuel economy is quite important, in order for a vehicle to gain wide spread acceptance in a global market positive driver perception and comfort must be assured. Thus it is important to develop control schemes which enhance and optimize drivability. Controller development began by improving an existing hydraulic hybrid powertrain model constructed in MATLAB Simulink. Control schemes for individual modes of operation were then proposed and investigated in the improved simulation model. Finally once sufficiently well-developed these control scheme were applied, tuned, and evaluated on the Maha lab’s hydraulic hybrid demonstration vehicle. The controllers developed showed an overall improvement in drivability as well as a great improvement in the transitions between discrete modes of operations. This held true for both the simulation model and more importantly when implemented in the demonstration vehicle. The new controls proposed in this work help to enhance the driver’s perception of a novel and highly efficient hydraulic hybrid transmission.

In this paper a feasibility study has been conducted to determine a technique for the measurement of lubricant fluid film thickness in the piston-cylinder interface of an axial piston pump. The thickness of the lubricant film has a significant impact on the efficiency of the piston pump yet it still remains an uncertainty. If the lubricant film is too thin then friction will cause excessive damage resulting in deformation of interacting surfaces resulting in further uncertainty in thickness measurement. If the lubricant film is too thick then there is excessive leakage which will compromise the efficiency of the system. Acoustic, inductive, capacitive and optical techniques were investigated through literature review and it was decided to investigate further into acoustic techniques. Experimental procedures using ultrasound to measure the lubricant fluid film thickness are investigated and an experimental procedure is explored further into in this paper. The governing factors of measuring fluid film thickness were determined. The effect of surface deformation on lubricant thickness is also investigated and a technique of imaging using ultrasonic sensors is presented. A CAD model displaying sensor locations and mounting apparatus in piston-cylinder interface is also presented. A feasible method for the measurement of lubricant fluid film thickness is proposed. These measurements will be used to validate existing models which predict the behavior of lubricant fluid film.

Variable displacement vane pumps (VDVP) are one type of positive displacement pumps used in automatic transmission vehicles for lubricating the gears, cooling the transmission and actuating the clutches. Though fixed displacement pumps are widely used, they output a constant effective flow at a given speed. Depending on pump sizing considerations, the pump can be oversized at high speeds because flow demand of the transmission is independent of engine speed. The excess flow returns to the tank through an orifice that causes the oil to heat up, increasing the energy required for cooling and reducing the efficiency of the transmission. A VDVP is used because the effective flow is adjustable to meet transmission demands any speed. The VDVP studied in this paper is unstable because of a feedback loop between pressure fluctuations at the pump’s outlet and the control system. To design a better VDVP with increased stability and controllability, a simulation model of the current pump design needs to be validated against measurements. Therefore a test rig design with required pump modifications is proposed in this study. Six pressure sensors are installed such that they measure the pressure in each of the pump’s displacement chambers. A LVDT is used to measure the pump’s eccentricity changes as the displacement is varied. Part of this study was to develop a custom MATLAB script that combines all the sensor data to show a pressure profile for each displacement chamber. The pump modifications for sensor installation and test rig implementation was designed.

Axial piston pumps are integral to a variety of hydraulic systems due to their versatility and reliability, but most attempts to model them for design purposes have failed. However, software which fully models the fluid structures within these pumps, such as the FSTI gap design program, has improved significantly. Unfortunately, current discrepancies between simulated and measured friction forces on the cylinder due to the fluid drag and piston movement need to be eliminated so that more accurate simulation of these pumps in operation is possible. Friction measurements and FSTI simulation outputs were analyzed with MATLAB filters and graphs, resulting in the conclusion that mixed friction, a form of partially lubricated contact friction, is the best explanation for the discrepancy. To model and predict the mixed friction, the Stribeck curve, a relationship between piston velocity, fluid viscosity, contact pressure, and the coefficient of friction between the piston and cylinder, was used, allowing a computational model to be developed to post-process the FSTI results and determine the mixed friction. The greatly improved model resulted in simulated total friction increases of nearly 300 percent in cases involving rough surfaces and high pressures. Though this module still needs to be fully integrated into the FSTI gap design program, the developed mixed friction model greatly increases the accuracy of the simulation by bringing the magnitude of the simulated friction much closer to realistic measured values.

Displacement Controlled (DC) hydraulics system is booming recently due to its high efficiency and accurate controllability, while the surplus numbers of pumps restrains its further development. DC pump switch technique is a promising application to reduce the number of pumps as well as the cost products to promote DC hydraulics technique. It is necessary to support an innovative DC hydraulic pump switching system with a controller which would allow rapid and accurate estimation and adaptation for all the parameters, like pressures and velocities, in the system. Dynamics and Hydraulics characteristics have initially been investigated for a simplified DC pump switching system testing rig, which is the Joint Integrated Rotary Actuator. Based upon the results from the test rig, three different kinds of models - Simmechanics, Nonlinear and linear - have been established for validating the state space model, which will be used for linear controller establishment. Matlab and Simulink will be implemented for validating predicting data. After fully validating data, labview will be utilized to build up a simple controller and generally build up its complexity. The ideas and experience came up with in this project would be modified and implemented for other DC hydraulics system controller development.

Hydraulic pumps and motors are vital components used in many applications today. Specifically, the axial piston pump is important because it is reliable, relatively compact, and has a high horsepower-to-weight ratio. These features make this type of pump very advantageous in hydraulic systems. Maintaining proper lubrication between surfaces in an axial pump, such as the slipper and swashplate, is imperative in order to have smooth operation of the system and prevent metal-to-metal contact. The aim of this research is to find the optimal slipper design and fluid film thickness to simultaneously maintain a balanced pressure and decrease power loss in the pump. Standard slipper designs are compared to multi-land designs through simulations. The simulations are run using the latest models developed by Schenk (2012). The geometry of the slipper along with input operating conditions calculate the gap height of the fluid film, as well as the pressures related to the interface. From this gap height, the efficiency of lubrication between the slipper and swashplate are determined. The results from the simulations are used to compare a design currently in use to both a standard and multi-land design. The simulation studies showed that a careful balance between pocket and orifice diameter is required to reduce leakage while minimizing friction.