|
Fluid Power Research Group
Department of Mechanical Engineering
Saskatoon, Saskatchewan
Canada
|
|
| Location |
Saskatoon, Canada |
| Responsible Leader |
Prof. Richard Burton (Assistant Dean, Undergraduate Programs) |
| Laboratory facilities |
About 330 m2 |
| Address |
College of Engineering, University of Saskatchewan
57 Campus Drive, Saskatoon
Canada S7N 5A9 |
| Telephone number |
+306 966 5287 |
| Telefax number |
+306 966 5202 |
| Email |
Richard_Burton@engr.usask.ca |
| Internet Site |
http://www.engr.usask.ca/dept/mee/faculty/burton.html |
From Editor
International Journal of Fluid Power would like to introduce the fluid power research and education centres with their expertise and particular interests in this column. Jumping from continent to continent we like to offer every research centre the opportunity to present itself.
FLUID POWER RESEARCH CENTRES WORLD-WIDE
The University of Saskatchewan is located in the city of Saskatoon, in the heart of the prairie provinces. It is approximately 1200 km east of Vancouver and 2000 km west of Toronto. The College of Engineering is one of 12 Colleges on campus and has about 1300 undergraduate and 250 graduate students. The Department of Mechanical Engineering is one of the larger Departments in the College with about 240 undergraduate and 80 graduate students.
One of the strengths of the Mechanical Engineering Department is the undergraduate program. Our students have placed first in many national and international competitions. Perhaps one of the most noteworthy achievements was in the SAE (Society of Automotive Engineering) super mileage competition in which our students set a world record of 49.6 ml/100km with their vehicle, a record which still holds today for that category.
Group Objectives:
-
To provide a suitable research environment for the postgraduate education of engineers in the fluid power area.
-
To apply state of the art control and monitoring techniques to hydraulic and fluid power systems.
Research Group (History)
Mechanical Engineering has had a long history of research. Dr P.N. Nikiforuk, Department head from 1966 to 1973 and Dean from 1973 to 1996, started the Fluid Power Laboratory in the department in 1973. Along with Drs. Paul Ukrainetz (department head from 1974 to 1982), Jim Wilson ( Head of Division of Controls Engineering, 1969 to 1997) and Bob Besant (department head from 1982 to 1993), research into hydraulic controls systems, valves and fluidic systems was initiated. In the 1970's, Drs. Burton , Schoenau (Department head from 1993 to 1999) and Sargent joined the research group with a focus on expert systems, neural networks and component design. In the 1990's, engineers Doug Bitner and Dan Mourre (part time) became integral members of the research team. In 1998, Dr. S. Habibi (head of the Mechatronics Research Group) has also now become active in the fluid power research area. The strength of this research group lies in the team approach to supervising graduate students and projects and the sharing of financial resources.
The Fluid Power group has graduated about 100 M.Sc. and Ph.D. students over the last thirty years. A review of our publication lists clearly show that our students are recognized as principal authors in work in which they are involved. Our work over the decades has appeared in almost every international fluid power journal and conference proceedings in the world, a fact of which we are very proud.
At present we have 14 graduate students in our program, along with one visiting professor (Dr. J. Huh from the Department of Mechanical Engineering, University of Technology and Education, Korea). Figure 1 shows most of the "gang" as we now exist.
Undergraduate and Graduate Education
The U of S is the only Canadian University presently offering a course in basic hydraulics. This course is available as a fourth year elective and is always over subscribed. The course emphasizes a basic understanding of how components work, their design and assembly into a circuit configuration. Emphasis on the design of circuits using sound engineering practices are integral to the material. The laboratory involves a hands-on fabrication of hydraulic circuits to do a specified job. The students are provided with a pump and load for which they must design and build a circuit to adequately meet the given constraints.
Ms. students usually take two of the following courses and three (or four) other courses to supplement their research areas (Neural networks, Data acquisition, and System Identification are just some examples). Ph.D. students commonly take all four classes in the fluid power area. M.Sc. students normally take 2 years to complete their M.Sc. with Ph.D. students taking about 3.5 to 5 years.
Courses Taught in Fluid Power
-
ME 462.3
Introduction to Hydraulic Circuit Design
This class in the only course offered in Canada at a University level.
- ME 860.3
Fluid Power Systems Introduction to Servo-System Technology (Merritt)
Class project: a complete model of a servo-valve and load simulated on Matab-Simulink©.
- ME 898.3
Advanced Fluid Power
This is a reading class which concentrates on modeling and analysis of Fluid Power Systems.
- ME 898.3
Selected Topics in Fluid Power
This is a reading class in which students are assigned topics which have been the subject of Ph.D. studies at other Universities.
Class Project: Simulation and experimental verification (when possible) of systems used in the aforementioned topics.
Research Facilities
Our research facility, which is comprised of a Fluid Power Lab and Controls Lab, is small (approximately, 330 m2), but well planned and highly utilized. However, it is equipped with state-of-the-art instrumentation and data collection equipment. Among the equipment is a network of computers and software for data acquisition, monitoring, simulation, real time control and graphics. Some of the software packages available include Matlab©, Real Time Workshop©, Autocad, Solid Works and LabWindows. The Controls Lab compliments the Fluid Power Lab as a resource center for computer interfacing equipment, sensors, amplifiers, monitoring instrumentation and two multi-channel signal analyzers.
The Fluid Power Lab contains three large hydraulic power supply units and test benches capable of supplying a range of flow rates up to 130 l/min and pressures up to 300 bar. One of the units, shown in Fig. 2, is designed to provide precise control over temperature, pressure and flow. The lab also houses two hydraulic manipulators with appropriate electrohydraulic valving. In addition, ferrography equipment and a particle analyzer are available for fluid particle wear analysis.
The group benefits from experimental equipment provided by funds from industry. Two of these units are the multiple pump unit and the load sensing pump test rig shown in Fig. 3. A wide variety of research projects was made possible with the aid of this equipment.
The laboratory also houses several smaller portable units in the range up to 20 l/min and 210 bar along with a large variety of pumps, manual and electro-hydraulic valves, motors, actuators, manipulators, pipes and fittings for hydraulic component testing and analysis. For the undergraduate course, the portable hydraulic power supply units, along with a diverse selection of basic hydraulic components and sensors, are available for students to design, build and test their hydraulic circuits as dictated by the particular lab experiment (Fig. 4). Students enjoy this class because of its hands-on and practical nature.
Research Projects
Application of Neural Network Technology to Fluid Power Systems
The U of S fluid power group was one of the first research institutions to apply expert systems and neural network technology to the fluid power area. Projects have been oriented towards the practical implementation of these technologies. Studies have included expert systems for the design of hydraulic circuits (which involved introducing a methodology for design), the use of neural networks for open loop control (an inverse problem) and closed loop control (a multiple input, multiple output problem) and using recursive neural networks for mimicking complex dynamic controllers. Recursive neural networks have also been used for "black box" modeling of real components. These studies are continuing.
Currently, some of the projects which are underway are:
- Modeling of a load sensing pump using recursive neural networks.
(NSERC grant), Lamontagne, Burton, and Ukrainetz.
-
A neural network based PID controller. (NSERC grant), Qian, Scheonau, and Burton.
- A neural network based co-controller to reduce coulomb friction in an actuator.(NSERC grant), Schoenau and Burton.
- A neural network based variable rate fertilizer application system. (NSREC grant), Mourre, Burton and Ukrainetz.
Condition Monitoring of Hydraulic Components
Several studies have been initiated into the use of estimation techniques to predict difficult to measure parameter values of pumps and solenoid valves. A change in these estimated parameter values over time can be used as a measure of the wear or deterioration of a component or system. Critical to this kind of approach is the ability to produce a set of operating conditions that are reproducible over the life of the component or system. This allows a norm signature to be set from which future measurement can be compared to. A simple add-on valving system has been implemented to provide a practical mechanism for creating the necessary operating condition for monitoring a proportional solenoid valve.
Three estimation techniques are being explored: neural nets, extended Kalman filtering and least squared estimation. Each technique is being explored with respect to reliability, sensitivity and computer implementation efficiency.
Currently, some of the projects which are underway are:
-
Estimation and monitoring of parameters in a hydraulic solenoid proportional valve using Least Squares approach. (IRIS 3 - Computer Assisted Machine Operation grant), Ansarian, Burton, Schoenau, and Bitner.
- A technique to facilitate the measurement of signatures in a proportional solenoid valve. (IRIS 3 - Computer Assisted Machine Operation grant), Mourre, Bitner, Burton.
- Estimation and monitoring of parameters in a hydraulic proportional solenoid valve using a Neural Network approach. (IRIS 3 - Computer Assisted Machine Operation grant), Rosa, Burton, Mourre, and Bitner.
- Estimation and monitoring of parameters in a hydraulic proportional solenoid valve using Extended Kalman Filtering Approaches. (IRIS 3 - Computer Assisted Machine Operation grant), Wright, Burton, and Bitner.
- Estimation and monitoring of parameters in a hydraulic pump using Extended Kalman Filtering Approaches. (NSERC), Cao, Burton, Schoenau, and Huh.
- The use of a neural network to map parameter characteristics as a function of spool displacement., Mourre, Burton and Schoenau.
Component Modeling and Design of Hydraulic Components and Circuits
Over the years, many studies on the design of new valves have been initiated including digital valves, high precision flow divider valves, spring valves and fast acting on -off valves and rotary valves, just to name a few. This research has continued with the co-operation of Dr. Jian Ruan from the Zhejiang University of Technology, China. A combined rotary and linear motion spool valve has been investigated and several novel control techniques have been applied. This study is an example of an international co-operative effort between two Universities in different countries.
Modeling of components and circuits is considered to be an integral part of all projects; indeed it is considered to be a part of the education process for all graduate students. We have not developed a software fluid power simulation package (there are many excellent packages available as it is). We do use both the power bond graph and the traditional transfer function in conjunction with commercial simulation packages (Matlab/Simulink©) for any simulation studies. Modeling studies have included such components as pumps, valves, and actuators.
A special study on the design of an Electro-Hydrostatic Actuation System (EHA) is under way. Techniques to control and optimize the performance of this unique system are being investigated. This system has many applications in the robotics and aircraft industry, but a means of reducing the weight of the complete system (including the electric drive) are necessary. Studies are focusing on methods to realistically achieve this.
Currently, some of the projects which are underway are:
-
Design, control, optimization and application of a combined rotary/linear spool servovalve. (NSERC), Ruan, Burton and Ukrainetz.
- Simulation of a hydraulic pump. (NSERC), Dobchuk, Nikiforuk, Ukrainetz, and Burton.
- Neural Fuzzy control of a pump. (NSERC), Dobchuk, Nikiforuk, Ukrainetz, and Burton.
- Neural Fuzzy control of a hydrostatic hydraulic system. (NSERC), Wu, Schoenau and Burton.
High performance hydrostatic actuation systems
An area of research in the fluid power group is the design and prototyping of high performance hydrostatic actuation systems. The objective of this research is to produce self-contained units that combine the benefits of conventional servo-valve controlled hydraulic systems and direct drive electrical actuators, namely high torque/mass ratio and modularity. Emphasis is placed on theoretical analysis backed by experimental results. In this relation, a prototype of a system referred to as the ElectroHydraulic Actuator (EHA) has been produced and used for experimental purposes. EHA has proved to be a unique device with its own characteristics and has required hydraulic components such a novel symmetrical actuator that are specifically tailored to its needs.
The current thrust of research in this area are:
-
Force control of EHA by using pressure feedback and neural network based feedforward compensation. (NSERC), Gnanam and S. Habibi.
- Health monitoring and diagnostics using Kalman filtering. (Commonwealth) Chinniah, Burton and Habibi.
- Design optimization through mathematical modeling and evolutionary programs (NSERC), Singh, Zhang and Habibi.
BACK
©2000 Fluid
Power Net All rights reserved. Disclaimer