 |
  S.I.L.2233 HANSELMAN AVE. SASKATOON,SK S7L 6A7 WHEELWATCHTABLE OF CONTENTS1.0 SCOPE 2.0 BACKGROUND 3.0 OBJECTIVES 4.0 TEST CONFIGURATION 5.0 ROAD TESTS • 5.1 INTRODUCTION • 5.2 TESTS WITH WHEELS SECURE • 5.3 TESTS WITH A LOOSE WHEEL • 5.4 OTHER RESULTS 6.0 ANALYSIS • 6.1 WHEELEX STRAIN • 6.2 TENSION IN ANGLE BRACE (Load Bolt) • 6.3 TEMPERATURE RISE 7.0 CONCLUSIONS APPENDIX A WheelWatch Test Plan #1 1.0 SCOPE This report documents the results of tests conducted to determine the preferred method of detecting a loose wheel on a vehicle equipped with a Wheelex Loose Wheel Management System. 2.0 BACKGROUND The Wheelex Loose Wheel Management System developed by Ibex Resources Corp. has been demonstrated to be an effective means of ensuring that wheels on highway vehicles are safely restrained even if they should become detached while the vehicle is travelling. Tests performed on a test track have been confirmed by operational experience. In no case has a loose wheel escaped even after travelling considerable distances. The most dramatic confirmation resulted when the lug nuts on one wheel of a fully loaded trailer accidentally came off while the vehicle was travelling at highway speed. Although the wheel was completely detached from its hub, and would normally have rolled away from the vehicle, the Wheelex Loose Wheel Management System restrained its motion so that it was limited to rolling in place adjacent to the vehicle. Figure 2.1 shows marks left on the highway indicating that, despite the very considerable forces exerted by the loose wheel, it was forced to stay in place until the driver noticed the condition and safely brought the truck to a stop.  To further increase the effectiveness of Wheelex, it is proposed to develop a sensor which will provide the vehicle operator with an early warning signal indicating that a wheel has become loose and is trying to escape. This sensor has been dubbed "WheelWatch". 3.0 OBJECTIVES From a knowledge of the manner in which Wheelex functions, it was determined that the most promising indicators of a "wheel-off"condition would be one or more of the following: i) force on the Wheelex frame as evidenced by straining of the vertical support member; ii) force on the Wheelex frame as evidenced by straining of the angle brace; iii) vibration of the Wheelex frame; and iv) temperature rise of the Wheelex frame at the point where the tire rubs. Other potential indicators, such as smoke from the wheel or enhanced noise, were considered to be less practical and were not pursued further. Accordingly, a series of tests were planned to quantify the various effects with the final objective being to identify which effect or combination of effects would produce the most effective method of sensing a wheel-off.4.0 TEST CONFIGURATION A highway flatbed trailer, shown in Figure 4.1, was fitted with a large box containing two tanks capable of carrying a combined total of 22,000 lb of water concentrated over the wheels under test. This enabled tests to be conducted with both light and heavy loads simply by adding or removing water. FIGURE 4.1. Trailer fitted for test.  Figure 4.1 also shows an instrumentation case suspended below the trailer chassis directly in front of the wheel to be tested. A close up view of the instrumentation case is shown in Figure 4.2. In this view the Wheelex Loose Wheel Management System can also be seen. It consists essentially of a steel "C-Frame" wrapped around the wheel at hub height.The C- Frame is secured to the trailer chassis. The instrumentation case includes a video camera viewing the region where a loose wheel would come in contact with the C-Frame and a digital data logger and lap top computer to record the outputs from various sensors. When the covers were fitted, the camera viewed through a small clear plastic window.  FIGURE 4.2. Instrumentation case with covers removed. The sensors employed during the test consisted of: i) two accelerometers, each capable of measuring accelerations up to +/-30g; one mounted on the trailer chassis; the other mounted directly on the Wheelex C-Frame; ii) six strain gauges: - one, configured as a quarter bridge, bonded directly to the vertical Wheelex bar where it would sense the strain caused when the wheel rubbed the Wheelex; - one, configured as a quarter bridge, bonded to a substrate which was, in turn, bonded to the Wheelex vertical bar; and - a group of four, configured as a full bridge, one on the vertical bar of each of the four Wheelex assemblies protecting the four wheels of the trailer; iii) two thermocouples, forming a cold-junction/hot-junction pair, one mounted adjacent to the point where a loose wheel would rub against the Wheelex bar and the second mounted on the Wheelex frame away from the heat caused by rubbing; iv) a digital real time clock located within the field of view of the camera so that the events recorded by the camera could be synchronized with the events recorded by the data logger; and v) a speed sensor measuring the rate of rotation of one of the wheels. The photograph in Figure 4.3 shows the manner in which the strain gauges were mounted. 
FIGURE 4.3. Strain Gauges mounted on Wheelex. Data from the video camera was recorded on a VCR located in the cab of the tractor unit. A video monitor in the cab also gave a real time view of what was happening. Prior to the commencement of the first run, all sensors were calibrated and appropriate scale factors were entered into the data logger so that all data would be recorded in the appropriate engineering units. 5.0 ROAD TESTS 5.1 INTRODUCTION Tests were carried out in accordance with Wheel Watch Test Plan #1 (Document 4073TDAA -RO1, dated February 7, 2000), copy attached as APPENDIX A. Prior to each test, the calibrations of the strain gauges were confirmed. After initial "shake-down" tests to ensure that the instrumentation was fully functional, a series of runs were undertaken, some with the water tanks empty, some with the tanks full, some with all wheels secure, and some with one wheel loose. With the data logger sampling and recording every parameter 100 times per second the amount of data collected is voluminous. Key portions of that data are presented in this document. 5.2 TESTS WITH WHEELS SECURE It is logical to expect that the greatest amount of bouncing around would result when a lightly loaded trailer is towed at highway speeds over rough roads. The data confirmed that this is so. Figures 5.1 and 5.2 show some strains measured in Wheelex caused by normal driving over roads of varying quality. Figure 5.1 shows results for "Wheelex Strain"(*1) and "Load Bolt(*2) The peak to peak Wheelex Strain is about 401bf and the peak to peak Load Bolt tension is about 1501bf. As would be expected, the Temperature Rise, i.e., the differential temperature between the two thermocouples, is essentially zero.   The worst case condition, with all wheels intact, occurred when travelling along a roughly graded track across a hay meadow at 50 km/h. Again, the tanks were empty. Figure 5.2 shows the comparable chart. Under these conditions, the peak to peak Wheelex Strain increased to about 125 lbf and the peak to peak Load Bolt tension increased to about 700 lbf. Again, the Temperature Rise is negligible, peaking at about 4°C. The existence of any temperature differential at all is attributable to small local cooling or heating effects related to solar heating and vehicle movement. *1."Wheelex Strain" is the bending strain experienced by the vertical Wheelex bar, expressed in terms of vertical load on the Wheelex restraining bar, i.e., a 50 lb weight hanging on the bar would result in a Wheelex Strain of 50 lbf. *2."Load Bolt" is the tension in the Wheelex Angle Brace as measured with a load bolt incorporating a strain gauge bridge. 5.3 TESTS WITH A LOOSE WHEEL Any sensor developed to sense the existence of a loose wheel, must be able to experience conditions such as those portrayed in Figures 5.1 and 5.2 and be able to identify these as "normal driving conditions". At the same time the sensor must be able to detect the minimum loads or temperature changes created by a wheel rubbing against the Wheelex C Frame. To determine the magnitudes of signals that might be expected when a wheel comes loose, tests were done, first on a track graded across a stubble field, then on a stretch of highway. For both of these tests a full 22,000 lb of water was loaded onto the trailer. The results appear in Figures 5.3 and 5.4.  Figure 5.3 shows the loads created when the vehicle travelled at 60 km/h along a track graded in a stubble field. The peak to peak Wheelex Strain (although hard to read on this graph) is about 50 lbf, comparable to what it was over rough ground when the wheel was secured. Because of this, Wheelex Strain does not appear to be a useful parameter for use as an indicator of loose wheels. The Load Bolt tension is more promising. Its peak to peak amplitude has increased to about 1200 lbf compared to the 700 lbf observed earlier. Although this is a significant increase, it is not dramatic enough to make for an indicator which is capable of reliably detecting loose wheels yet is free from false alarms. Temperature Rise is the dramatic variable. Whereas during normal operation the temperature of the two thermocouples remains nearly the same, under wheel-off conditions the thermocouple near the rub point heats up considerably. In this run, the temperature rose to about 350'C where it leveled off. This is in sharp contrast to the temperature differential of only four degrees observed in the absense of wheel rubbing. The final test was similar to the above except that it was conducted on a smooth paved highway at speeds up to 90 km/h. The graph in Figure 5.4 shows a segment during which the speed was increasing from 25 to 50 km/h. Prior to reaching 25 km/h the wheel was not contacting the Wheelex frame. As shown on the graph (by the sudden increase in Load Bolt tension at 15:23:44), the wheel then attempted to move outwardly and was restrained by the Wheelex C- Frame. Initially there is little effect on Wheelex Strain but it then rises sharply to about 1000 lbf. If this effect had also been seen in the test illustrated by Figure 5.3, Wheelex Strain would be a very good indicator. But it wasn't. The reasons for this apparent anomaly will be discussed later.
The most effective indicator is again the Temperature Rise. Although it only had time to rise 25°C during the short time segment displayed here it continued to rise, stabilizing at about 350°C as the vehicle continued on. It is important to note that the tire does not rub against the Wheelex C-Frame continuously. Wheelex operates by guiding the wheel back into position any time it tries to move away from the vehicle. Even with a loose wheel, there are periods - some short, some longer - when the tire is not rubbing. An example of that is shown near the end of the period covered in Figure 5.4. This does not detract from the effectiveness of Wheelex, nor does it imply that forces or temperature are not good indicators of a wheel-off. What it does mean is that the indicator chosen, and the algorithm used to exploit it, must be capable of reacting to short term contact between the wheel and the Wheelex C-Frame.
The other significant finding is that the environment in which the sensor is expected to operate is a very hostile one. Figure 5.5 shows the vibration levels encountered during a wheel-off at 80 km/h. The Chassis Acceleration shows 10 to 20g peaks with an occasional extreme of 30g or more, briefly exceeding the nominal +/-30g range of the accelerometer. Although these are greater than would normally be found, they are manageable. The Wheelex Acceleration is much greater. In the 10 second window provided by Figure 5.5 the acceleration is off scale about 50% of the time. Because the accelerometer saturated, we do not know what the peak levels were. Suffice to say they are very high. The message for WheelWatch is that if any portion of the WheelWatch sensor is mounted on the Wheelex frame it will have to be extremely robust. 6.0 ANALYSIS 6.1 WHEELEX STRAIN In Section 5.0 it was noted that Wheelex Strain, i.e., the bending strain in the Wheelex Vertical bar, does not promise to be a reliable indicator of loose wheels. To understand why that is so, it is necessary to examine the geometry of Wheelex. Figure 6.1 shows a simplified drawing of Wheelex as seen from the front. When a wheel comes loose and moves outwardly, it encounters the Wheelex C-Frame and applies an outward force, F(out) Due to friction between the tire and the frame, there is also a downward force, F(down). The resultant is a force F(net). If F(net) acts along a line that passes below the point of attachment to the vehicle chassis (shown in blue in Figure 6. 1), the Wheelex C-Frame will tend to move outward and upward, causing the Wheelex vertical bar to flex and hence generating a strain which would be detected by the strain gauge. If, on the other hand, F(net) acts along a line that passes above the point of attachment to the vehicle chassis (shown in green), the Wheelex C-Frame will tend to move downward and inward, generating a strain of the opposite polarity. In the special case where F(net) acts along a line that passes through the point of attachment to the chassis (shown in red), there will be no movement of Wheelex and no signal will be generated by the strain gauge. In the course of the tests reported in this document, all three of these conditions were observed, accounting for the exteme variability in Wheelex Strain measurements. Hence, Wheelex strain, or any other strain measurement that measures flexing in the vertical bar, is not a reliable indicator of wheel-off's. 
6.2 TENSION IN THE ANGLE BRACE (Load Bolt) The measurement of tension in the angle brace is not subject to the geometrical problem that affected Wheelex Strain. In fact, as shown in Figure 6.1, the force F(net) acts along a line roughly parallel to the angle brace so that both F(out) and F(down) add to the tension. The load bolt produced large and predictable signals during normal operation and during wheel-offs. Unfortunately there was not a great margin between the two. Normal operation produced peak to peak signals of about 700 lbf while wheel-offs produced signals of up to 1200 lbf, in one case, and 2200 lbf in another case. Even based on this limited sampling of results, it is apparent that there would be a need for a rather sophisticated algorithm to uniquely distinguish wheel-off from wheel-on conditions. Some work has been done on developing such an algorithm based on the observation that most of the Load Bolt fluctuations during wheel-on are symmetrical about the baseline value whereas the fluctuations during wheel-off are superimposed on a large positive tensile load. Simple low pass filtering would remove most of the fluctuations (which have a frequency of about 10 Hz) leaving only the more slowly varying unidirectional signal resulting from the tire rubbing against the C frame. This approach is promising, but additional work may not be warranted because there is, what appears to be, an even better indicator. 6.3 TEMPERATURE RISE Temperature Rise(*3) provides a very clear indication of tire contact, with little evidence of contaminating effects. Although a simple "decision threshold" set at about 20°C above the baseline(*4) would be adequate, it is useful to also calculate the "Rate of Temperature Rise" (with an alarm setpoint of +2°C/s) and use both as indicators. Since the sensor will incorporate a microprocessor, this added processing is a zero-cost increment to the manufacturing cost. Figure 6.2 indicates how effective these indicators would be. The "Rate of Temperature Rise" provides the quickest indication of trouble since it reacts to the initial heating. Later, as temperature stabilizes, the absolute amount of Temperature Rise is a more useful indicator.Together they are better than either one alone. (*3)"Temperature Rise" is the temperature difference between a point near where the tire rubs against the Wheelex frame and a reference point some distance away, also on the Wheelex frame. (*4) The baseline would be determined by the long term (10 sec) running average.  Use of temperature as the indicator of wheel-off's has the added advantages of being electronically simple and economical to manufacture. Thermocouples also meet the "robustness" requirement to survive in the hostile environment presented by a loose wheel trying to escape. A minor disadvantage of sensing temperature at the rubbing point is that it requires electrical wiring be brought to that point. Given the location and its potential for mechanical damage there is a need to provide adequate protection for the wiring. There will also be a requirement for a connector in the electrical cable so that the Wheelex C-Frame can be removed for tire servicing. 7.0 CONCLUSIONS The Developmental Tests for WheelWatch are considered to be very successful. The key objectives were i) to identify a suitable indicator that would uniquely determine when a wheel had come off while at the same time being immune to normal background conditions; and ii) to identify an algorithm to be used in conjunction with the indicator signal to determine when an alarm condition had occurred. These objectives have been met by identifying "Temperature Rise" and "Rate of Temperature Rise" as joint indicators of a wheel-off. This would be done by using two temperature sensors (thermocouples) configured as a differential pair, one mounted on the Wheelex C-Frame near the point where the tire of a loose wheel would rub, the second mounted a convenient distance away, also on the C Frame. The output signal would be a measure of the difference in temperature between the two mounting locations. By using "temperature differential" rather than just "temperature", the system becomes independent of variations in ambient temperature. The digital processor would compute the temperature rise above its baseline value and would also compute the rate of temperature rise. Either condition, differential temperature exceeding its setpoint (20°C) or rate of temperature rise exceeding its setpoint (20°C/sec), would set off the alarm. TEST PLAN #1 1.0 SCOPE 2.0 BACKGROUND 3.0 TEST CONFIGURATION 4.0 DATA LOGGER CHANNEL ASSIGNMENTS 5.0 TEST PROTOCOL 5.1 LABORATORY TESTS 5.2 ROAD TEST -- WHEEL INTACT 5.3 TRACK TESTS - WHEEL LOOSE 6.0 ANALYSIS 7.0 REPORT 1.0 SCOPE This Plan describes the tests that will be used to specify the operating environment in which WheelWatch will be required to operate and to identify the most promising detection technique. These tests make up Phase 1 of the WheelWatch development program. 2.0 BACKGROUND WheelWatch, when installed on a typical vehicle, will be expected to detect "wheel-off conditions with a high degree of reliability. At the same time it is expected to ignore false inputs caused by normal road conditions including roughness, snow, mud, water on road, high temperatures, low temperatures, etc. These two requirements may sometimes be in conflict. It is the objective of these tests to identify the most promising detection technique and to specify how sensitive it must be to a wheel-off and how insensitive it must be to the various extraneous inputs. The two most probable techniques for detection of wheel-offs are considered to be: i) measurement of strain in the Wheelex structure as the tire rubs the bar, and ii) measurement of differential temperature between the Wheelex bar (near where the tire rubs) and the ambient. The tests described in this document focus on these two techniques. The strain measurement approach appears to be more easily implemented and is currently the preferred one. 3.0 TEST CONFIGURATION The tests shall be conducted using a highway trailer fitted with a Wheelex Loose Wheel Management System on at least one wheel. The instrumentation on the trailer shall include: iii) Accelerometer (vertical) mounted on trailer chassis near axle under test. iv) Accelerometer (vertical) mounted on Wheelex restraining bar near point where tire would touch during wheel-off. v) Thermocouple mounted on outboard face of Wheelex restraining bar at point where tire would rub. vi) Thermocouple mounted on Wheelex bar well away from point where tire would rub. vii) Strain gauge on vertical bar of Wheelex at point of maximum bending moment during a wheel-off. viii) Load bolt to measure tension in Wheelex Angle Brace. ix) Four strain gauges mounted on the four Wheelex vertical bars (Gauges connected and monitored as a single full bridge) x) Vehicle speed sensor (fifth wheel with tachometer or magnetic pickup on one of the trailer wheels). xi) Data logging computer to record data from the above sensors. xii) Alarm horn - to test audibility (preferably remotely controlled from chase vehicle). xiii) Alarm light - to test visibility (preferably remotely controlled from chase vehicle). xiv) Video camera (including time-of-day display synchronized with the computer clock to within one second) mounted directly in front of the wheel under test to give a clear view of the wheel and Wheelex and show a) when the tire comes into contact with restraining bar, and b) the amount of deflection of the Wheelex assembly. xv) Protective housings (with temperature control, if necessary) for the computer and camera. xvi) Power source(s) for all the above. Other equipment shall include: i)A highway tractor unit to pull the trailer. ii)A chase vehicle to allow observers to follow the tractor/trailer. iii)Two way voice radio communication between the chase vehicle and the operator of the tractor iv)Remote control unit for alarm horn and alarm light. 4.0 DATA LOGGER CHANNEL ASSIGNMENTS The data logger shall be configured to record data at the rates shown in Table 4.1. Data Logger Channel Assignments  Note 1: Signal is a differential output of +/- 2 VDC biased at about 3.6 VDC. Requires differential input to data logger. 5.0 TEST PROTOCOL 5.1 LABORATORY TESTS With the equipment installed and operating each sensor shall be calibrated by applying known force, temperature, acceleration, etc. To the greatest extent practical, preliminary tests shall be conducted by exercising each sensor in a manner approximating the full range of conditions expected during the tests. Data shall be logged during these simulations. The logged data shall be played back and examined for readability and reasonableness, 5.2 ROAD TESTS - WHEEL INTACT These tests may be done on public roads and highways or on a test track. For each of the test conditions listed below, at least three runs shall be carried out, one each at a speed of approximately 50 km/hr, 75 km/hr and 100 km/hr. Because some tests are not repeatable, additional runs may be required to enable a range of responses to be recorded. With the wheel under test fully intact and running normally, the instrumented trailer shall be towed over a wide variety of road conditions, including as many of the following conditions as time and availability permit. i) normal smooth highway ii) potholes iii) bumps (e.g., speed bumps, rumble strips, obstacles) iv) washboard road surface v) water pooled in depressions on road vi) heavy snow coverage vii) mud viii) freezing rain or slush 5.3 TRACK TESTS - WHEEL LOOSE After the successful completion of the Road Tests in 4.2, the trailer shall be taken to a suitable section of track or unused highway for wheel-off tests. The tests conducted at the track shall attempt to reproduce a range of wheel-off conditions between the extremes of : •a very gentle momentary contact of the tire on the Wheelex bar, •and full continuous hard contact between the tire and the Wheelex bar. As with the Road Tests in 4.2, each test shall be repeated at each of the three speeds (50, 75, and 100 km/hr). In addition the test route shall include both gentle and sharp turns to both left and right. 6.0 ANALYSIS The computer data from the Road Tests and Track Tests shall be presented in graphical form and examined visually with the intention of identifying the most promising detection technique. The graphical data shall be annotated with data from the video camera. Particular attention shall be paid to identifying:
iii) the minimum signal of the same type that resulted from a wheel-off condition. The band between this maximum and minimum represent the detection window. A WheelWatch Monitor designed with a detection threshold anywhere between the maximum normal condition and the minimum wheel-off condition would reliably detect wheel-off conditions yet not respond to normal disturbances encountered during road travel. 7.0 REPORT The tests shall be recorded in a written report showing the test configuration, the test results and the conclusions that were drawn. |
