Back to Table of
ContentsINDUSTRIAL APPLICATIONS OF DYNAMIC DATA ACQUISITION - FIRST SEMESTER EXPERIENCES IN A MULTI-DISCIPLINARY LABORATORY COURSE
ABSTRACT
A junior-level multidisciplinary laboratory course centered around industrial applications of dynamic data acquisition and analysis is described. The course was developed with funding from an NSF Instrumentation and Laboratory Improvement (ILI) grant and is offered as a NSF Foundation Coalition (FC) course. It is also open to traditional engineering students. In the course, teams of four (maximum) students are drawn from more than one discipline and with complementary interests and skills. It is intended that the associations developed within and amongst the teams will enable cooperative, multidisciplinary design projects in the senior year.
The course consists of four weeks of introductory material followed by four laboratory modules, each concerning a specific application of signal acquisition and analysis. Currently, these modules include speech encoding and enhancement, machinery sound power measurement, machine condition monitoring, and motor condition monitoring. Each module is designed to be independent and the modules may be presented in any order. The course concludes with a culminating design project. Three instructors are involved in teaching the course, one from Mechanical Engineering and two from Electrical Engineering.
In this paper, information on the course structure, content, hardware and software is provided. Problems encountered in the first semester are recounted and adjustments to the course structure are suggested to address these problems.
Some of the problems are common to any new laboratory course. Other difficulties are unique to the structure of this FC course, including those associated with multi-disciplinary team teaching, and technology enabled education. Methods of improving the efficiency and effectiveness of instruction are proposed. Despite start up difficulties, the Fall 1996 student reviews were highly enthusiastic. There has students demand for the course to offered again in the Fall 1997 semester.
INTRODUCTION
Students enter the College of Engineering at The University of Alabama (UA) in either the traditional or the FC track. Freshman and sophomore FC students take a coordinated series of courses that integrate calculus, physics and engineering. These students make extensive use of technology tools and software, such as MAPLE, and engage in team based, active learning starting in their first semester. The laboratory course discussed here serves as a junior level FC elective. It is also open to students in the traditional AE, EE, IE and ME curricula. Only mechanical and electrical engineers (both FC and traditional) enrolled in the course when it was first offered in the Fall of 1996.
Funds for additional instrumentation to support this course were provided by a 1995-1997 NSF ILI grant and UA matching funds totaling $90,000. The course was conceived as a mutlidisciplinary, applications based laboratory course that incorporated team based, active learning elements. It is a junior level elective accessible to aerospace, electrical, mechanical, and industrial engineering students. Funding from the FC allowed for development of a technology enhanced curriculum.
In this paper, the structure and content of this multi-disciplinary laboratory course are described. Lessons learned during the first semester are discussed throughout. Information on instrumentation and software is given in discussions of the laboratory modules. Cited model names and numbers do not represent endorsements by the authors and are provided for information purposes only. Prices, where quoted, are approximate.
COURSE STRUCTURE
PC Based Learning Tools
The three-credit-hour, semester long course consists of both classroom activities and laboratory exercises. Many of the early classroom activities are interactive, making use of PC-based demonstrations and exercises. This requires a classroom equipped with networked computers and projection capabilities for the instructor's PC. During the fall semester of 1996, when the course was offered for the first time, MATLAB movies developed for the fundamentals section were available on the network and on the WWW. As the course matures, greater use will be made of data and resources available over the WWW.[1,2]
The advantages of computer based instructional tools are clear. They include active involvement of the students and the ability to rapidly construct and visualize solutions (e.g., successive approximations to periodic waveforms obtained with Fourier series expansions). These are very important advantages, but there are disadvantages, too.
Two of the difficulties that we encountered in a PC based classroom during the first semester were poor classroom acoustics and poor visibility. Room acoustic treatment (to absorb noise generated by the computers) made it very difficult for students in the back of the room to hear the instructor. Student views of the whiteboard were partially obscured by the PCs; instructor views of the students were substantially obscured. As a result, a few of the students busied themselves playing solitaire or surfing the web. These problems may be alleviated in a new computer classroom that will be used the next time the course is taught.
When interactive demonstration tools or computer based exercises are used in the classroom, students require considerably more time than the instructor needs to complete a task. This has a significant impact on the amount of new material that can be covered in a 50 minute lecture hour. It may be that these educational tools are best used in two or three hour instruction periods.
Lastly, the time required to develop computer based instructional materials that truly enhance the learning process is considerable. We made use of an enthusiastic, senior level electrical engineering student to develop many of the MATLAB movies and demonstrations.
Student Teams
Our use of teaming and active learning in the first semester was a mixed success. The fourteen students enrolled in the course consisted of five FC students (junior level EEs and MEs), six traditional ME and 3 EE seniors in their graduating semester. The students were grouped into teams of three or four, consisting of both electrical and mechanical majors.
As a general rule, the groups of three performed better as teams (cooperating, dividing up the workload, meeting regularly) than did the groups of four. However, the most obvious disparity in team performance was between the groups that had two or more FC students and those that did not. Compared to the traditional students, the FC students really worked as a team. To make matters worse, several of the senior level students were vocally adverse to teaming. Two measures supported by the students to improve team performance are: 1) power to fire unproductive team members; and 2) anonymous team performance reviews after each laboratory module.
When the course is taught again next semester, team training exercises will be incorporated into the classroom activities during the first weeks. The meaning and extent of teamwork in the course will also be more strongly emphasized during the first class meeting.
One of our objectives in establishing multidisciplinary teams was to promote interaction among students in different disciplines that would enable cooperative, multidisciplinary senior design projects. It is too soon to tell whether or not this objective will be met, but the interaction between the traditional ME and EE students outside the classroom during the first semester was impressive. The interaction among FC students from different disciplines was already well established.
Active Learning
While active learning most definitely took place in the laboratories, it was not as fully employed in the classroom. Even in the laboratories, there was substantial variation in the extent to which the three different professors employed active learning. This resulted in comments on students' appraisals such as:
"I think every module should follow Dr. X's method which is, explain, show us how to do them and have answers for all the labs."
"Dr. XX makes us dig for things, but, as frustrating as it sometimes was, this was probably good."
Here, as with teaming, it is important that it be made clear to the students on the first day of class what will be expected of them. The course must then be consistent with these expectations.
Team Teaching
Team teaching makes this course a truly multi-disciplinary experience for the students. One of the two instructors for the Fundamentals module and the instructor for the Speech Encoding and Enhancement module is Dr. Stern (EE), who specializes in communication systems. The other instructor for the Fundamentals module and the instructor for the Sound Power and Machine Condition Monitoring modules is Dr. McInerny (ME), who specializes in noise, vibration and fault detection. The instructor for the Motor Monitoring module is Dr. Haskew (EE), whose specializes in power conversion systems and fault detection. Students get the benefit of the instructors' enthusiasm for their specialized fields as well as access to their laboratories.
We learned in our first semester that maintaining consistency in terminology, instructional methods, as well as homework and laboratory report formats, is both very important and not a trivial task. We did a good job of maintaining consistency in terminology and in continuity of subject matter, but were not as consistent in terms of expectations for homework, pre-laboratory assignments, and laboratory reports (despite established laboratory report guidelines). Another issue that requires careful coordination is the return of assignments and tests from previous modules during the next professor's classroom activity periods. With planning and ongoing communication, these problems can be remedied.
COURSE CONTENT
The course begins with an introductory module covering the fundamentals of digital data acquisition and analysis. This is followed by a series of independent laboratory modules centered around specific industrial and commercial applications of dynamic data acquisition and analysis. The course concludes with each team performing a two week culminating project based on one of the application modules. Teams give oral reports on their projects.
The laboratory module topics were selected based on industry applications and the backgrounds of the instructors. Industry representatives continue to provide input on the course content. Each application-specific module is designed to be independent, thus could be presented in any order. The modular structure of the course allows for the addition or deletion of modules in response to changes in industry.
Four industrial and commercial application modules were presented in the Fall of 1996. These modules were Speech Encoding and Enhancement, Machinery Sound Power Measurements, Machine Condition Monitoring and Diagnostics, and Motor Condition Monitoring and Diagnostics. The students also did a small independent project at the end of the semester.
It is likely that over the next few years we will continue to adjust the compromise between technical depth and breadth of the topics covered in this course. At present, it appears that either one laboratory module or the culminating project should be eliminated in order to allow the students more time to absorb the application specific concepts and to become more proficient in the use of the different analyzers.
In selecting instrumentation for the application modules, an effort was made to obtain equipment and software representative of that used in industry. A Stanford Research Systems SRS 780 was chosen for the Fundamentals and Speech Encoding modules based on a compromise between bandwidth and price. Although the basic spectral functions behind the acoustic and machine condition monitoring applications are the same, there is not an analyzer available that performs all the necessary calculations and is compatible with widely used software. Thus, two other types of dual channel analyzers were purchased for the Sound Power and Condition Monitoring modules. The use of mutiple, application specific analyzers has the advantage of exposing the students a variety of analyzers.
The cost of instrumentation for this course could be reduced if a single analyzer were used in all of the laboratory modules. Data Physics recently introduced a PCMCIA card based spectral analyzer. It can be used with any notebook or desktop PC that has a PCMCIA slot. The PCMCIA card has anti-aliasing filters, an analog to digital converter and a 50 Mflop DSP chip. It comes with software that converts recorded data to a variety of formats for importation into programs such as ME Scope (operating deflection shape and modal analysis) or MATLAB. The cost of the Data Physics system, the DP 104, ranges from ~ $4000 to $7000 (exclusive of the PC) depending upon options.
There are a couple of drawbacks associated with the use of such a system for this course. First, application specific software would have to be purchased separately or written by the user and external power supplies would be needed for the microphones and accelerometers. Secondly, the analyzer purchased by the authors for the Sound Power module is an ANSI standard Type I sound level meter. This makes it useable for a wide variety of measurements outside of the classroom. A number of issues would have to be addressed before a PC based data acquisition system could be used for standardized acoustic measurements.
Fundamentals Module
The Fundamentals module consists of four weeks covering the basic concepts of data acquisition and digital signal processing (DSP). Topics include Fourier analysis, sampling and reconstruction, aliasing, linear and nonlinear quantization, dynamic range, Fast Fourier transform relationships, windowing ,and filtering. Emphasis is on the qualitative and applied aspects, details of digital signal processing algorithms, and digital filtering are omitted. Concepts are reinforced through laboratory experiments and MATLAB exercises, using both pre-recorded data files and data gathered by the students. Sensors are covered in the application modules in which they're introduced.
Material for this module is drawn from manufacturers application notes and texts. [3-7] Sections of the application notes are included in the course packet available at the student bookstore. We are considering incorporation of some of the relevant MATLAB based laboratories from Reference 8.
In the Fundamentals laboratories, students learn how to use the Stanford Research Systems SRS780, a 100 kHz, 2 channel spectral analyzer (~$10000). This analyzer is used again in the Speech Communications module. Instructions for the use of SRS780 analyzer are included in the course packet. Several function generators are also used in the Fundamentals modules. Other than the difficulties cited in the earlier section on PC Based Learning Tools, the Fundamentals module was very successful and was well received by the students.
Speech Encoding and Enhancement
The objective of this module is to demonstrate data acquisition principles and certain signal processing techniques used in speech encoding and enhancement. Digital speech encoding and enhancement has many commercial and industrial applications, including new-generation digital cellular telephones, speech encryption devices for privacy and compact disks. This module consists of experiments covering performance evaluation of various speech encoding techniques and the use of digital signal processing to reduce background noise.
Using a PC-based data acquisition system, students begin the module by making digital recordings of their voice, both in a quiet environment and with background noise. Using the "quiet" recordings, students investigate various speech encoding techniques, including Pulse Code Modulation (PCM) at 64 Kbits/sec and Delta Modulation (DM) at 16 Kbit/sec. Students then examine the performance of each encoding technique in terms of speech quality, computational complexity, and encoding rate.
Next, students examine the recordings made with background noise. Various processing techniques such as additional filtering and adaptive noise cancellation are evaluated. As with speech encoding, the noise reduction techniques are rated in terms of computational complexity and improvement to speech quality.
In conjunction with this module, we plan to include coverage (perhaps as a student project) of the use of digital signal processing to for the enhancement of speech. Students will use speech generated by speaking-impaired individuals and employ speech enhancement techniques to improve the intelligibility and quality of the handicapped person's speech. The experiment will be performed in cooperation with faculty and students from the Department of Communicative Disorders.
Equipment used in this module includes the SRS780 analyzer, electronic filters, two 100 MHz PCs with sound cards and speakers, and a couple of inexpensive microphones.
Sound Power Measurements
In this module, students gain an understanding of why machinery sound power measurements are needed (hearing protection and annoyance issues) and how sound power data are utilized by engineers. The basic equations for free field acoustic radiation and room acoustics are introduced. The decibels, 1/3-octave and octave band levels, and A-weighting are covered. The relationship of A-weighting to human hearing sensitivity is established.
Students are introduced to standards (ANSI and ISO) applicable to the measurement of sound power. Various methods of sound power measurement are discussed: free-field (outdoor or hemi-anechoic room) measurements, reverberation room methods, and sound intensity measurements. The relative merits of the three methods is discussed. The use of sound intensity in this laboratory provides the students with their first exposure to the use of two channel measurements and issues associated with phase accuracy.
In the first laboratory exercise, measure noise generated with a white noise source, amplifier and speakers. Students calibrate the measurement system using a pistonphone and then make, for later comparison, 1/3-octave band, A-weighted 1/3-octave band and high resolution FFT sound pressure level measurements. The measured data is recorded to memory in the LD2900 and then transferred to a floppy disk for post processing and plotting in EXCEL.
In the second and last laboratory exercise of this module, students measured the sound power radiated by a dehumidifier using the scanning method of sound intensity. The sound intensity calibrator was not available the first semester, but will be used for system calibration in the laboratory next fall. Measurements are made on five faces of a 1 meter cube constructed with dowels centered around the dehumidifier (or other noise source). Students download the measured 1/3-octave band data and calculate in EXCEL (although the LD2900 has the capability internally) the overall sound power in 1/3-octave bands. In their reports, they are asked to comment on the predominant noise frequencies and radiation directions.
Equipment for this module includes a calibrated Acculab sound source (~$1800), a Larson Davis LD1260 sound intensity probes with phase matched microphones (List ~ $5000), a Larson Davis CAL290 sound intensity calibrator (list ~$4200), and a LD2900 portable two-channel real time analyzer with 1/3 octave, octave, and FFT capabilities(list ~$12000 with selected options). Note that the Larson Davis equipment was purchased as a package which included some used demonstration equipment, which reduced the total cost. Unlike the SRS780, the LD2900 is a portable, battery powered analyzer that is easily hand held. This permits a great deal of flexibility it the selection of laboratories exercises.
Based on experience gained in the first semester, students need more time to absorb the background material and to become familiar with the use of the LD2900 analyzer. Instructions on the use of this analyzer were not included in the course packet the first semester. We have the manufacturer's permission to include relevant sections from the LD2900 Training Manual in the course packet next fall.
Machine Condition Monitoring and Diagnostics
The objective of this module is to introduce students to the use of vibration measurements for machine condition monitoring and diagnostics. Machinery health conditions reflected in order-related spectral components (1x running speed and higher harmonics) including imbalance, misalignment, and loose couplings are examined. Non-order-related vibrations are discussed only briefly. Topics covered include sensors (proximeters, velocimeters, and accelerometers), optimal sensor locations and mounting, normalization of frequency scales in terms of multiples of the base RPM, and the use of bin or band levels.
Coverage of diagnostics is limited to the more basic methods involving the trending of spectral levels, although reference is made to more sophisticated or complex analysis utilized for gear- and bearing-diagnostics (e.g., the so-called "envelope averaging method" and cepstral methods used for bearing analysis). Students are taught techniques to identify imbalance, out-of-alignment, and shaft looseness conditions.
In the first laboratory exercise, students use a tachometer and an accelerometer on a four bladed, direct-drive portable fan. The variability of single average spectra and the trade-off between spectral resolution and measurement time are examined. Acceleration measurements are made with the fan firmly mounted to a concrete base both with and without a mass added at the tip of one of the blades. With the imbalance mass in place, a synchronously averaged acceleration versus time measurement is recorded. Lastly, the imbalance mass is removed and the fan detached from the base and another set of data is recorded. The data sets are downloaded from the Diagnostic Instruments PL302 analyzer to a PC via an RS-232 connection for post processing and plotting in EXCEL.
In the final laboratory, students acquire and analyze data on an induction motor coupled to a DC generator. On one side of the coupler a steel disk with holes at two different radii is mounted. Industrial use accelerometers with plug in type connectors are stud mounted to mounting pads permanently bonded to the motor - generator. The optical probe and tachometer are also used. Students acquire data with the system in "good" operating condition (balanced and aligned); out of balance (using solid 3/4" bolts in holes of the disk); and out of alignment (support legs on the outboard side of the motor are lowered). Using EXCEL and the downloaded data sets, students are asked to show how the imbalance and out of alignment conditions are reflected and differentiated in the data.
Instrumentation for this laboratory module includes a tachometer with optical probe, 6 IMI industrial use accelerometers with cables (~$900), a PCB portable accelerometer calibrator (list ~$1800), and a Diagnostics Instruments PL302 real time dual-channel analyzer / data logger(list ~$14000). Analyzer capabilities include those required for run-up and run-down testing, time synchronous averaging, FFT, time waveforms, frequency response functions, zoom analysis, and machinery balancing software. This analyzer was selected over competing products based on its real time analyzer capabilities, compatibility with machine condition monitoring software widely used in Alabama industry (ENTEK-IRD), and price. Two analyzers, one a demonstration unit, were purchased as a package whose price reflected both university discounts and the used demo unit.
The material covered in this application module was more difficult for the EE than the ME students to grasp. When next taught, the background material on the dynamics of rotating machinery will be covered more slowly. In addition, another basic laboratory exercise could be added so that students become more familiar with the use of the analyzer and with the physics of rotating systems.
In the future, we hope to replace the motor-generator with a motor driven fan, one in which the shaft alignment could be more closely controlled. If well planned, such a test bed could be readily modified to accommodate a gear box or journal bearings. Students would then be able to do more advanced machinery diagnostics or perform fan balancing for their final project.
Motor Condition Monitoring and Diagnosis
As first conceived, this module was to cover the use of current measurements for motor condition monitoring. Motor monitoring software packages similar to machine condition monitoring software exist for this purpose. However, feedback from industry indicated that the software is of limited usefulness. Broken rotor bars and eccentric rotors are readily detected, but impending insulation breakdown is not.
Drs. Haskew and McInerny are involved in research on the detection of insulation breakdown in variable speed motors. This is a timely topic, as the use of adjustable speed motors is growing rapidly in manufacturing and variable flow heating, ventilating and air conditioning systems. Associated with these drives are power system harmonics, noise, motor overheating (due harmonics on the motor side of the drive), and insulation breakdown in retrofitted and rewound motors.
Thus, when this course was offered for the first time, we chose to focus this module on power system harmonics associated variable speed drives. Students were introduced to the basics principles of three phase induction motors and Pulse Width Modulated (PWM) motor control. The first laboratory exercise was performed in the EE department's Power Electronics Laboratory and focused on the operation of power electronic motor drives.
For the second laboratory, we utilized a research facility in which a three phase, 440V, 15 HP motor is controlled with a PWM variable speed drive. The motor load is provided by a DC generator. Hall effect transducers were used for measuring current and resistive voltage dividers were used for potential measurements. In this laboratory, students experienced the operation of a variable speed motor and developed an understanding of the harmonic distortion induced by a PWM controller.
Students had a difficult time with the pace and details of the material covered in this module. This was particularly true of the junior level ME students who had had no previous exposure to motors. In response to student feedback, this module is being modified and will be substantially different when offered next semester.
Culminating Project
For the final two week project in the Fall of 1996, each team chose from a list of projects proposed by the instructors and drawn from the application modules. These included: demonstration and explanation of a database / trending program used for machine condition monitoring, measurements of the directivity and frequency distribution of sound generated by a PWM controlled AC induction motor as a function of drive frequency, design and performance of a ventilated sound reduction enclosure for an air compressor, frequency dependent impedance of an AC induction motor, and design of voice encoding techniques for Internet voice communications. Each team made a final presentation to the class and to the instructors. (More extensive final projects drawn from local industry were planned and may be employed in the future if one of the application modules is omitted.)
CONCLUSION
The multidisciplinary junior level laboratory course described in this paper is a concept that works. Coverage of background material prior to the application specific laboratories must be thorough, yet not too detailed. With team based active learning, both in the classroom and the laboratory, students learn by doing and from one another. This is well summarized by student comment made on one of the course evaluation forms "This class changed my perspective in a lot of ways. I knew nothing about anything we covered prior to this class, but I realized how much I am able to 'self teach' myself and draw from others in their fields." Team teaching provides students with instructors who are enthusiastic about their fields, as well as providing them with access to specialized laboratories. Close coordination between instructors is necessary for consistency and continuity.
REFERENCES