Data Acquisition for the Study of Racecar Vehicle Dynamics
Abstract The aim of the project was to install, calibrate and test two data acquisition systems on a Formula Renault and a Formula Ford car respectively. The cars were then to be tested on track at Silverstone and the data subsequently analysed. The idea was to use the data acquisition systems to study the vehicle dynamics of a racecar using real data. The project was carried out in collaboration with National College for Motorsport (NC4M) at Silverstone Innovation Centre where the two cars are based. A 2D system was installed on the Formula Renault and tested at Silverstone. System compatibility issues were highlighted when installing the system on the car. Sensor Calibration errors and logger problems were found and rectified in the time leading to the test. Various lessons were learnt about the installation and packaging problems posed by data acquisition system. As only a partial system was tested, limited data was available for analysis and further development of the system. Having learnt from the test, fewer mistakes were made with the Pi Delta system. All sensors were calibrated and installed on the car. It was learnt that the Formula Ford would not be tested on track. Hence, focus was shifted to making a workbook in Pi Toolbox using sample data to facilitate data analysis. Various math channels looking into load transfer, ride height changes, roll, dynamic crossweight changes have been set – up to assist in the study of vehicle dynamics. The initial project brief had suggested a comparison between real brake temperature data and a simulated model. Provisions have also been made for the same. The aim was ultimately achieved in three stages:
•
2D system installed and tested on track
•
Pi Delta system installed on the Formula Ford
•
Workbook and math channels set – up in Pi Toolbox
Proposals for the expansion of the systems and advancing the project further have been made towards the end of the report.
1
Data Acquisition for the Study of Racecar Vehicle Dynamics
Statement of Originality
I hereby declare that this thesis and the research documented in this thesis, except where indicated in the text, has been composed entirely by me. The design and construction of the proposed project along with the analytical techniques used, was devised by me. The workbook set – up and conclusions derived have been a result of the undertaken project and have not been influenced by the work carried out by other individuals.
2
Data Acquisition for the Study of Racecar Vehicle Dynamics
Acknowledgements First of all, I would like to thank my supervisor Prof. John Nixon for his continued support, guidance and advice throughout the project. I would also like to thank Dr. Marko Tirovic for his support in setting the project up with National College for Motorsport. I would also like to thank John Grist from Datron Technology for providing support with the 2D system. Thanks also to Robin Cull at Pi Research for carrying out the logger checks and providing the university with a new version of Pi software and help files. Many thanks to Jim Harrod and the rest of the staff at NC4M for their exceptional support and guidance throughout the project.
3
Data Acquisition for the Study of Racecar Vehicle Dynamics
ABSTRACT
1
STATEMENT OF ORIGINALITY
2
ACKNOWLEDGEMENTS
3
1
7
INTRODUCTION
1.1
Project Overview
7
1.2
Background Information
7
1.2.1
National College for Motorsport
7
1.2.2
The Racecars
8
1.2.3
Data Acquisition Systems
8
1.2.4
System Manufacturer’s
9
1.3
2
Project Aims and objectives:
10
LITERATURE REVIEW
11
2.1
History:
11
2.2
Driver associated sensors
13
2.2.1
RPM Analysis
13
2.2.2
Speed Analysis
15
2.2.3
Throttle Analysis
16
2.2.4
Steering Angle Analysis
17
2.2.5
Lateral Acceleration G
18
2.3
Engine Systems
21
2.4
Chassis Sensors
22
2.4.1
Damper Position Potentiometers
22
2.4.2
Load Cells
22
2.4.3
Ride height sensors
23
2.5
Sensors and Instrumentation
2.5.1
RPM sensors
24 24
4
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.5.2
Wheelspeed sensors
25
2.5.3
Temperature Sensors
26
2.5.4
Pressure sensors
27
2.5.5
Linear Potentiometers
28
2.5.6
Rotary Potentiometers
28
2.5.7
Ride Height Sensor
29
2.6
Summary of the Reviewed Text
30
2.7
Conclusion of Literature Review
31
3. EXPERIMENTAL PROCEDURE
32
3.1 Formula Renault and 2D
32
3.1.1
Pre – Installation Test Runs
32
3.1.2
Installation and ergonomics
33
3.1.3
Calibration and Dash set – up Process
34
3.2
Pi System and Formula Ford
36
3.2.1
Pre – installation test runs
36
3.2.2
Sensor Calibration
36
3.2.3
Installation and Ergonomics
37
4
RESULTS
4.1
2D and Formula Renault:
4.1.1 4.2
Pi Delta System
4.2.1
5
Results Obtained from the test session:
Workbook Set – up
DISCUSSION
39 39 41 43 44
52
5.1
2D system and Formula Renault
52
5.2
Pi and Formula Ford
54
5.2.1
Pi Delta installation
54
5.2.2
Pi Toolbox
55
6
CONCLUSION
57
7
FUTURE WORK AND POSSIBILITIES
58 5
Data Acquisition for the Study of Racecar Vehicle Dynamics
8
REFERENCES
61
9
REFERENCES: TABLE OF FIGURES
63
10 APPENDIX A: MATH CHANNELS
65
6
Data Acquisition for the Study of Racecar Vehicle Dynamics
1 Introduction This chapter gives the reader an introduction to the project. The initial project brief has been mentioned. More importantly, the scope of the project, the background information on the cars involved and the systems manufacturer’s has also been mentioned here.
1.1
Project Overview
The project involves the installation, calibration and testing of two different data acquisition systems on a Formula Renault and Formula Ford single seater racecars. The two cars are based at National College for Motorsport with whose co – operation the project has been carried out. The two data acquisition systems being used for this project are 2D and Pi Delta. With the data available, the vehicle dynamics module can be better understood and explained. Load transfers, ride height changes, ride height and load changes due to aero loading, etc. will then be demonstrated using the data. Also, students at NC4M can then be showed the effects of set – up changes carried out by them. The Data acquisitions module can be taken further into a more advanced stage.
1.2
Background Information
Before discussing the aims and objectives of the project, it would be beneficial to have some background information of National College for Motorsport (NC4M ), the two cars, the systems and their manufacturers.
1.2.1 National College for Motorsport National College is an educational institution that specialises in training technicians for the Motorsport Industry. It is based at Silverstone Innovation Centre near Silverstone Circuit. It benefits from its experienced staff, proximity to the track, many F3 teams based at Silverstone and its various industrial links.
7
Data Acquisition for the Study of Racecar Vehicle Dynamics
Their programme revolves around the numerous mechanical aspects of a racecar. Students taking the course perform various activities like stripping and rebuilding various cars, gearboxes and engines. During the course, students get opportunities to work with race teams during test sessions. NC4M has been able to provide its students with various apprenticeships in different race teams ranging from Formula Renault to F3000 to A1GP. NC4M does not however get involved in engineering race cars.
1.2.2 The Racecars Formula Renault: The Formula Renault racecar acquired by the National College for Motorsport (NC4M) used to compete in the Formula Renault BARC championship. The car has since been used for the students to work on as part of their coursework. This constant use is an advantage since all the components on the car are in good working order and do not need an overhaul. This car would be fitted with a 2D system over the course of this project. Formula Ford: The car was initially based at Cranfield University where students have used it as part of their various projects and coursework. The car has competed in the Walter Hayes trophy that takes place every year as part of the Formula Ford festival. In terms of data acquisition, it does not benefit from any electronic aids. The RPM can be read using a tachometer. Inability to store data on the tachometer removes any possibility of basic data analysis. The Pi Delta system would be fitted onto as part of the project.
1.2.3 Data Acquisition Systems 2D systems – WinaRace and Winlt: these are the data analyser software and the logger communication software respectively. The system, hardware included, is quite dated by current standards. It had been acquired from Datron Technologies in 2004 and had been left untouched since. It was therefore quite important to determine the status of the system i.e. license expiry, any hardware failures, inspect connectors.
8
Data Acquisition for the Study of Racecar Vehicle Dynamics
Although these systems are designed to last for long periods of time in harsh environments, it was still essential to carry out a complete system check. Prior to installation, the majority of the effort was concentrated on the analyser software to gain a better understanding of the system flow. Specifics of the analyser software and comparison to other data acquisition software will be discussed later. Although the analyser plays an important role in processing the data, Winlt cannot be ignored because of the simple fact that it is the software that communicates with the logger through a COM port. Any changes to the sensors or the logger and the calibration cannot be carried out without Winlt. Pi Delta and Pi Toolbox: These are the software for Logger communications and logger management respectively. As with the 2D system, the system and the software were acquired in 2004 and had not been put to use since. It was therefore necessary to check the status of the equipment and the loom as well. Various methods were devised for the same and have been mentioned later in the report. Pi Toolbox has been powerful data analysis software and benefits from great flexibility in terms of displaying data. Data filtering and programming math channels has been the highlight of the software and has really been used during the project. Pi Delta is again quite flexible in terms of allowing the configurations and calibration of sensors to be changed quite easily. The specifics have been detailed in various sections of the report.
1.2.4 System Manufacturer’s Datron Technologies: They are a German data acquisition systems company with establishments all over the globe. They make data systems for varied industries ranging from industrial sensor systems to railways to motor racing. In motorsports, they are particularly known for their bike systems. In terms of 4-wheeled motorsport, their systems are used extensively in the BTCC championship.
9
Data Acquisition for the Study of Racecar Vehicle Dynamics
Pi Research: They are a data systems manufacturer with its main focus on motorsport. Their European division is dedicated solely to motorsport. Their products include various data loggers that tend to different levels of the sport starting from the amateur club racers to as high as A1GP and Formula 1. In North America, they are heavily involved with Champcars.
1.3
Project Aims and objectives:
With this knowledge, it is possible to discuss the aims and objectives of the project. The aim is to make two cars available so that they can be used as an educational tool for the study of vehicle dynamics and motorsport data acquisition. Having completed the installation and testing, the system can be expanded towards a more dedicated programme – recommendations for which will be made over the course of this project. The initial brief also included a comparison between brake temperatures as seen on track and a simulated model. This interchange was to take place across two departments within the university.
10
Data Acquisition for the Study of Racecar Vehicle Dynamics
2 Literature Review The history of data logging and the various channels that are monitored as part of data analysis has been mentioned here. References have been made to contributions of other writers on the subject like Simon McBeath and M. Albayati ( MSc Thesis 2006 – Driver Data Analysis and Comparison ) wherever applicable. The various sensors used to obtain data and their operating principles and manufacturer details have also been discussed following the channel description. The chapter ends with a critical review of the text.
2.1
History:
Data analysis has today become the most important factor in improving car performance. Drivers have become more receptive towards data and have learnt to adapt their driving styles to suit different circuits. Modern day legend such as Michael Schumacher had invested huge amounts of time and brainpower to understand the implications of his driving style on the car and tyres. Being able to respond to the suggested changes and then keep the pace up has made him truly great. But this is referring to a modern day Formula 1 car where a few degrees change in temperature and pressure could mean a difference of a few tenths. But where did it all begin? First instances of data logging and analysis can be attributed to the driver’s answers to the engineer’s questions. RPM values through each corner and handling characteristics were the most prominently asked questions. Answers to these were at best, speculative and subject to the driver’s memory. The first instance of data logging hardware can be traced to the installation of tachometers that were able to record data. RPM and Speed traces put together provide enough information for driver analysis and car performance. It is indeed still the practice at most semi – professional and club events.
11
Data Acquisition for the Study of Racecar Vehicle Dynamics
At professional racing categories such as GT, Touring cars, F3, GP2 and of course, F1 data can be broadly classified as: Driver Associated Channels: How the driver uses the car given to him. Channels like RPM, speed, throttle, steering angle, tyre wear, lateral and longitudinal acceleration provide a good account of the driver’s style. Engine: would contain channels relating to engine oil temperatures and pressures, water temperatures, air pressure in the airbox. Most of the gearbox channels are generally analysed along with the engine. Chassis and set – up: Entire suspension set – up, springs, dampers, packers, ride heights etc are considered here. Load cells and strain gauges are also considered here if they have been installed on the car. Math channels i.e. derivatives of the physical channels are also integrated into these main categories in the same way. Data logging is still considered a black art as most teams adapt the hardware available to them in a direction they want to improve their respective cars in. A good example today would be aero data in F1 being guarded on a level parallel to a country’s defence programme. Car set – ups are loosely tailored to a driver’s style and hence drawing up standards and general patterns for study is a near impossible task. Use of math channels to study interactions of other parameters adds a new level of complexity to the study of data logging. As a result, sources to study the role of data loggers are very limited. It would be logical to discuss individual channels within the groups they are generally associated with. This would provide a better understanding of the interactions between them.
12
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.2
Driver associated sensors
This list consists of sensors on which the driver has a direct influence on. The analysis of these allows for a better understanding of a driver’s style and technique. Points on the track where time is being lost or can be gained can then be identified.
2.2.1 RPM Analysis As mentioned earlier, the oldest form of data recording used tachometers with memory. The earliest ones simply lodged the maximum rpm during the run following the last reset. This however provided little information. “Intelligent Tachometers” which were later available had electronic memory cards that recorded engine rpm data. Further advances followed with the ability to download the rpm memory by linking it to a PC. Initially, the amount of data that could be stored was limited by the memory of the recording tachometer. Typical storage times being in the region of 8 – 10 minutes. This increased greatly with advances in computer technology. Another limitation being the looped or cyclic recording procedure i.e. when the recording memory was full, it would start overwrite from the beginning. In cases where such tachometers were used, data for longer runs would be lost and only the final minutes would be available for analysis. As always is the case, most of the driver channels are expressed in a graphical format against time or distance. The interaction of various inputs has been discussed after considering individual channels.
13
Data Acquisition for the Study of Racecar Vehicle Dynamics
Figure 1: RPM plot with details
Information that can be deduced from a simple RPM trace has been mentioned below: 1. Maximum RPM in each gear. The max speed of the vehicle in each gear can be deduced from this. 2. “General occasions where the driver backs off the throttle to slow the car but doesn’t brake”. 1 3. Occurrence of wheel spin can also be verified from the RPM trace. Matching this to the track map can give clues as to where on the track the car is loosing traction or whether the driver is inducing wheel spin to get off the line, etc. 4. Wheel spin can also assist in the analysis of oversteer / understeer if the rpm data is analysed with other channels like individual wheel speeds or throttle and steering response. 5. Using the values for maximum and minimum RPM values, “the engine’s power band can be matched to the actual demands of the competition, either by better optimising the car’s gearing, or by altering the engine components to produce torque and power at the rpm they are needed.” 2
14
Data Acquisition for the Study of Racecar Vehicle Dynamics
6. Amateur drivers or over zealous drivers can over rev the engine, jeopardising the whole race and resulting in huge expenses for engine rebuild or replacements. Modern data acquisition systems highlight these and the driver can be alerted. This information in itself can allow semi – professional and / or weekend racers to get a good idea of the car and driver’s performance.
2.2.2 Speed Analysis “ Speed is your most useful and most important channel. It tells you what you want to know at every point on the track – in other words how fast are you? Logged speed, in fact, relates to every aspect of car performance: acceleration, braking, chassis performance and set – up, engine performance, aerodynamic performance and driver performance. All these aspects are inter related, but study of the speed trace can point you in a direction of enquiry when analysing your overall performance.” 3
Figure 2: Speed Trace for Thruxton 15
Data Acquisition for the Study of Racecar Vehicle Dynamics
To analyse a speed trace, it must be verified that the maximum and minimum speeds correspond to where they are expected to be. Comparing some of the speed peaks with a track map can help understand the cars gearing. Corner entry and exit speeds can also be looked at to ensure the driver has been consistently driving fast. Differences in speed traces are usually marginal. Any exorbitant changes could be a result of traffic on the track or the driver having gone off. Also while passing someone on track, the driver is likely to take a line different from the racing line resulting in a different speed trace. Such differences are quite obvious and can be confirmed by talking to the driver. Overlays of speed traces from different laps will highlight inconsistencies in the car and driver performance. Better understanding of the performance factors can be obtained by combining speed and rpm together with the track map. Handling characteristics i.e. understeer / oversteer can be judged by looking at these two channels together. For example, if there is a sharp rise in the engine rpm for lesser increase in speed, while exiting a corner, the car might have offloaded a driven wheel or lost grip, causing wheelspin. Such conclusions can be confirmed by checking patterns in the data for subsequent laps.
2.2.3 Throttle Analysis Throttle analysis is closely related to RPM analysis as throttle opening and rpm are closely related. Throttle traces are generally expressed in terms of percentage with 100% being wide-open throttle and 0% being fully closed. In effect, it tells the analyser/engineer what the driver’s right foot is doing at any point on the circuit. However, it does not convey any information about what the optimum point for applying wide-open throttle or for backing off is. These are hugely dependent on track conditions, traffic on track while taking a corner, aerodynamic forces, engine characteristics and the car’s handling. Overlays for throttle traces of two different drivers on the same track can highlight points for comparison thereby giving vital insight into improving performance and driver technique. Analyser software allow for histograms of various channels to be 16
Data Acquisition for the Study of Racecar Vehicle Dynamics
plotted. A histogram of throttle percentage against lap time reveals the driver’s commitment to the use of the throttle.
Figure 3: Throttle Plot at Silverstone International
2.2.4 Steering Angle Analysis Steering angle obviously shows what the driver is doing inside the cockpit. Combined with throttle alone steering angle reveals how the driver tackled a particular corner. Steering traces generally follow the track’s characteristics i.e. “more right hand turns for a clockwise circuit.” 4 Comparison between drivers gives insight into the different lines taken by them. Steering angle combined with other channels like speed, throttle and lateral acceleration can highlight the car’s handling characteristics i.e. understeer or oversteer. It also reveals whether the driver initiated these conditions to balance the car; or whether he was too aggressive while taking a certain corner. Opposite lock is a good indicator of oversteer and/or the driver’s commitment to stay flat out on the throttle while exiting a corner.
17
Data Acquisition for the Study of Racecar Vehicle Dynamics
Combined with the speed trace, the driver’s braking performance while taking corners can be evaluated. However, external factors such as events taking place on the track (overtaking manoeuvres), track topography e.g. adverse camber or gradients on corners will show significant changes in the steering inputs. It is very easy to believe that steering data output was driver error rather than a change in the handling or the environment that induced it.
2.2.5 Lateral Acceleration G “Lateral G is essentially a measure of cornering force, or more strictly, cornering acceleration, and for convenience is usually expressed in units of G where 1G is equal to acceleration due to gravity.”
5
The cornering acceleration generated by a car is
dependant on the grip available from the tyres and the vertical forces generated from it. Softer tyres provide more grip and higher cornering forces can be generated. The cars aerodynamic set up will also contribute significantly to the vertical forces acting on the tyre again increasing cornering G. As mentioned earlier, lateral G when plotted against time or distance with other channels like steering angle, throttle and speed gives a good indication of the car’s handling characteristics. Opposite lock on the steering at the point of highest G value and lowest speed through a corner is an indication of oversteer. Similarly understeer can be seen when the peak lateral G lags the highest steering input.
18
Data Acquisition for the Study of Racecar Vehicle Dynamics
Figure 4: Corner Entry Oversteer
The analysis of longitudinal G has been left out of this discussion as similar deductions for wheelspin and braking can be made from the analysis of other channels like wheel speeds, brake pressures, etc. Lateral G generated gives a good indication of how well the driver is utilising the available grip. Although drivers claim to be on the limit at all times on the track, lateral G traces can show otherwise. It, of course, helps drivers see where more time can be gained. To better understand this concept, lateral G and longitudinal G can be plotted together to give what is known as a circle of friction. Tyre manufacturers provide data for the maximum grip available for a combination of lateral and longitudinal accelerations. Hence a direct comparison can be made between the driver’s ability to use the grip. It must be remembered that the grip available is limited by available torque and power.
19
Longitudinal Acceleration
Data Acquisition for the Study of Racecar Vehicle Dynamics
Lateral Acceleration
Figure 5: G - G Plot
The shape of the G – G plot will not always be circular as the highest acceleration occurs in lower gears where downforce is limited. Also, race cars generally two wheel drive “putting a cap on maximum longitudinal G under acceleration. The lower half will only be circular if the car and driver can actually stay on the limit of available grip when entering, negotiating or exiting a corner. Also, data points close to the circle’s limits are created when either maximum lateral or longitudinal G is being generated, but not when both lateral and long G are being generated. ” 6
20
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.3
Engine Systems
For general purposes, all the pressure and temperature channels are grouped together under engine systems. These include coolant temperature and pressure, oil pressure and temperature, fuel pressure. It is not uncommon to include gear position pots in engine systems. Additional gearbox channels like oil temperatures in the gearbox, if available, are also included here. The pressure and temperature channels give a good indication of the operating conditions. For example, if the coolant is running too hot, then blanking tapes can be taken off the radiator to cool it. Similarly, oil temperatures can also be monitored and changes affected accordingly. The most significant information relayed by monitoring pressures and temperatures are reliability issues. Plotting battery voltage with speed and overlaying with fuel pressure will spot any misfires for corresponding drops in fuel pressure. Oil pressure drops can also be spotted similarly.
Figure 6: Engine Pressures and Temperatures
21
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.4
Chassis Sensors
The most common type of sensors installed for monitoring chassis and suspension behaviour are damper position potentiometers and load cells.
2.4.1 Damper Position Potentiometers They find particular importance to this project as vehicle dynamics can be better understood and explained through the data obtained through them. It is common practice to use damper pots on all four corners of the car. Logging suspension movement depict suspension travel at each corner of the car and the speed of suspension movement. This can assist in fine-tuning suspension springs and dampers for high speed and / or low frequency events as well as diagnosing changes in handling and the aerodynamic balance.
2.4.2 Load Cells Suspension loads can also be monitored by the use of strain gauges bonded directly to suspension members like pushrods to monitor the loads on the suspension. Loads can also be logged using load cells – whose output is based on internal strain gauges mounted co axially with the pushrods. “This enables dynamic pitch and roll changes to be analysed as well as the peak loads developed on bumps and kerbs.” 7 Loads imposed by the aerodynamic downforce can also be analysed.
22
Data Acquisition for the Study of Racecar Vehicle Dynamics
Figure 7: Damper Displacement, Speed and Load
2.4.3 Ride height sensors These are of particular importance whenever significant downforces are in play. Tyre deflections, chassis roll and sideways tilt on the car can be measured using these sensors highlighting their importance in terms of vehicle dynamics. Due to complex nature of factors affecting ride height changes, detailed descriptions have been omitted here. Another reason for the omission being the cars available for this project are unlikely to attain high aerodynamic forces and cause tyre deflection.
23
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.5
Sensors and Instrumentation
The material reviewed thus far has not considered sensors that are installed on the car to record the data nor the problem associated with them. There are considerable challenges in working with sensors and due care must be taken to ensure their smooth operation. Although, two different data acquisition systems were used on two different cars, the sensors are essentially the same – digital or analogue. Sensors such as RPM and wheelspeeds are digital channels where as damper pots, Steering, Throttle, Oil pressure, etc are analogue.
Their installation details and sampling
frequencies have been mentioned here.
2.5.1 RPM sensors RPM forms the most essential and basic analysis channel irrespective of the formulae or car. In case of the Formula Renault, the RPM data is available from the ECU of the car. However, with the Formula Ford, the lack of an ECU has prompted the use of a specified RPM sensor. Of the various types of RPM sensors available, the electrical sensors are the most widely used. The 2D systems “sensor receives pulses from the primary ignition driver.” 8 The Pi sensor called the ‘4 Stroke RPM Box’ “picks up current pulses from the LT side of the ignition coil or voltage pulses from the master HT lead on 4 stroke engines.” 9 As always, care must be taken to place the sensor away from the ignition coil to reduce noise on the signal. The sensor itself is galvanically isolated from the ignition coils of the car.
Figure 8: Pi Express 4 stroke RPM box
24
Data Acquisition for the Study of Racecar Vehicle Dynamics
The engine speed varies significantly over small time periods and a sampling frequency of 100Hz is quite common. Higher frequencies are possible but the recorded data does not reveal much and occupies more space on the logger memory.
2.5.2 Wheelspeed sensors These are generally non-contact inductive devices. The sensor reacts to a metal part being crossed over it. “Inductive signals are then measured and counted”
10
When
suitably mounted i.e. close to brake screws or other parts of the brake itself, the sensor can be triggered and the wheelspeed thus obtained. Manufacturers provide instructions on mounting the sensor and its proximity to its trigger. These instructions must be followed earnestly to ensure proper functionality of the sensor over a long period. The sensor usually fails when it is set too close to the trigger and the trigger actually makes contact with the sensor knocking it off. Sensing distance between 0.5mm and 1.2mm are very common. Another common cause failure is the use of excess force to tighten the lock nuts for Pi wheelspeed sensors. The nuts tend to cut through the sensor causing failure. Maximum operating frequencies are 1 – 2 kHz depending on the manufacturer. Measuring frequencies can then be set according to the number of triggers and the memory available on the logger. Although it is a fundamental channel to the system, speed does not increase as rapidly as damper position changes. Hence, a measuring frequency of 10 – 50 Hz is adequate. Optical wheelspeed sensors are also available for sensing speed but tend to create installation problems. They are also more prone to noise. After having discussed the two fundamental channels for analysis it is now essential to look at sensors that will give a good indication of the operating parameters of the vehicle i.e. pressure and temperature sensors.
25
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.5.3 Temperature Sensors “The measurement of temperature is essential for monitoring and control of a wide range of motorsport systems. The temperature of cooling water and lubricating oil has a direct influence on the performance of power trains, while the temperatures of incoming air, brakes, tyres and even the cockpit can influence vehicle performance.” 11
Due to varied range of temperatures that need to be measured, a selection of temperature sensors is available. The most appropriate temperature sensors for the purpose of the project are thermocouples and infrared temperature sensors. Thermocouples work on the principal that a voltage is generated when a junction is made between two metallic alloys. This voltage is dependent on the temperature difference between the ends of the wires. “The performance of thermocouples is dependent on the composition of the metals used and the range of alloys is available. The most common types are K, J and R. The K type uses chrome – nickel and aluminium nickel alloys.” They operate over a range 200° C to 1100° C. The J types use iron – constantan and have a limited temperature range of -40° C to 750° C. Similarly, the R type thermocouple uses alloys of rhodium and platinum and can operate at temperatures over 1500° C. Their low output voltage makes them sensitive to external noise and interference. Thermocouples can be used to measure all fluid temperatures seen on the two cars. The other type of temperature sensor is the infrared sensor. It is used mostly for brake temperatures. These are of particular importance as the initial spec for the project involved installation of brake temperature sensors to study cooling performance and then comparing them with a simulation model. “As the temperature of the brake disc rises, the level of IR emissions increases. This can be directed onto an IR sensitive photodiode.” 12 Depending on the application, the sampling frequency is set. Water temperature or oil temperature does not change rapidly and a frequency of 1Hz or maximum 10Hz
26
Data Acquisition for the Study of Racecar Vehicle Dynamics
would suffice. For Brake temperatures, the channels are monitored at 25hz to 50Hz. Higher frequencies are used if specific test programmes are being conducted.
2.5.4 Pressure sensors “Pressure sensors operate by measuring the deflection of a diaphragm, which is exposed to pressure on one side.”
13
The sensing element could be a piezoelectric
sensor or a strain gauge connected in a bridge configuration. The sensors can be as small as a few millimetres and operate over a wide range of pressures. It is generally supplied with a housing enabling it to be installed into a tapped hole providing access to the sensing area. Some pressure sensors incorporate sealed vacuum chambers on the inner side of the diaphragm to give an absolute pressure value. Differential pressure ranges can also be measured if required. Sampling frequencies are again dependent on the application. Air pressure is measured at 1Hz where as brake pressures are measured at 25 to 50 Hz. In addition to these, current and voltage sensors can also be used to monitor battery voltages and/or alternator currents. Lambda sensors are used in high-end motorsport applications to monitor exhaust emissions. It is now possible to look at other sensors such as damper pots, ride height sensors and load cells, which will assist improving performance. Throttle and steering position sensors can also be discussed now.
27
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.5.5 Linear Potentiometers These are essentially linear potentiometers “which is a strip of resistive material mounted parallel to the moving component. A wiper attached to the moving component moves along the resistive strip.”
14
This assembly is connected as a
voltage divider that produces a voltage proportional to the position of the wiper. This basic principle is the same irrespective of the manufacturer. Although there are other position sensors such as Linear Variable Differential Transformers available, linear potentiometers are the most widely used due to simple installation and calibration process.
An interesting application of a linear
potentiometer is its use as throttle position sensor. A spring loaded linear potentiometer is installed onto the throttle pedal to give a percentage reading of the applied throttle. When used as a damper position sensor, its sampling rate is set at 100Hz atleast, to monitor movements. As a throttle position sensor, a 50Hz measuring frequency is essential. However, this reads the movement of the pedal and not the throttle itself. Hence it is not accurate and rarely put in practice.
2.5.6 Rotary Potentiometers “Circular resistive tracks similar to those in a standard potentiometer have been used.” 15
These are most widely used as steering and throttle position sensors for engine
management systems. It has also been used as a damper position sensor in some applications. When used as steering position sensor, measuring frequencies range between 50Hz to 100Hz. For throttle position applications, 50Hz is common.
28
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.5.7 Ride Height Sensor A special form of infrared position sensor is used to measure ride height. Mounted to the bottom of the car, the sensor sends out a beam which is reflected off the ground into a detector. By measuring the position of the returning beam, the height can be calculated. Data can only be available once the sensors have been installed on the car and calibrated. It is important to look at the sensors, their operating principles and any concerns they might raise during installations.
29
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.6
Summary of the Reviewed Text
It must be noted that not many publications are available on the subject. Simon McBeath’s Competition Car Data Logging is more comprehensive while the referred thesis deals with only the driver inputs and their relevance. Hence a direct comparison has been avoided here. Simon McBeath’s book is aimed at developing an individual’s interest in data acquisition. It brings novices up to speed with modern data acquisition systems. The importance and relevance of every trace has been discussed in adequate detail. The book also investigates the possibilities of combining different channels. E.g. relation between battery voltage and fuel pressure. Other reliability issues have been mentioned but left to the user. The only criticism for this book being problems and issues with the data being left to be discussed at a later stage – could be confusing for novices. For someone who has developed his thinking based on the book might feel a bit listless after getting to a more advanced stage. The thesis has borrowed much from Simon Mcbeath’s book but is intended for comparison between drivers. Hence only the driver channels could be reviewed using it. Other channels which would have significant effect on performance such as damper pots, load cells, ride height sensors have been mentioned very briefly. Various pressure and temperature channels which are directly affected by driving styles have not found a mention here. The most significant channel to have been ignored would be brake pressures and brake temperatures. Although, this could be due to the nature of the championship that the writer was investigating. It would have been worthwhile to have a good look at Buddy Fey’s Data Power for similar reviews. But this material could not be made available at the time of this report. In terms of manufacturer information sheets, both manufacturers provide the same details about the sensors in different words. The operating fundamentals for the sensors are the same and choice would ultimately depend on the price and the system being used. Pi sensors, however are a better design and have a wider range of operating conditions. The range of measuring frequencies available is also much larger. 30
Data Acquisition for the Study of Racecar Vehicle Dynamics
2.7
Conclusion of Literature Review
The reviewed material is a generalisation for all the data acquisition software and hardware available. It does provide a general idea for analysis but to get meaningful data out of the system it is essential to have the right hardware in the right place and fully functional. The material has completely ignored installation problems for the sensors. The many meters of looms that go into the car can be a great packaging issue. The loggers usually tend to be in the footwell or the cockpit and can be a hindrance for the driver. The download port can be an issue if it is in the wrong place. It is not just the driver that should be considered. Mechanics spend most of the time on the car and the wiring always tends to get in their way – a fact that should not be ignored whatsoever. It is best to talk to the mechanics and then install the looms and the sensors. The idea therefore is to get all the hardware on to the car first in the best possible way and check the functionality of the hardware. Once all the hardware has been assessed, the system and the car can be tested at Silverstone. Data will then be available for analysis. Following the test, basic analysis will be conducted using channel data like RPM, Speed, Steering, Throttle, Lateral G. With damper pots available, it would be logical to assess the set – up using this data. In terms of assessing the vehicle dynamics, Math Channels would be set up to track changes. Body Roll, Loading at the front and rear, etc. can then be tackled. A worksheet/workbook with the math channels will be set – up using one of the systems available so that other channels can then be plotted and studied. Reliability issues can also be looked at but due to the limited amount of data and run time available, it would be difficult to spot trends. Consequently, recommendations and comparisons between the two systems will be made.
31
Data Acquisition for the Study of Racecar Vehicle Dynamics
3. Experimental Procedure The complete process; from installation to calibration and testing of the two systems has been described in this chapter. The associated problems and various solutions obtained have been detailed here. The two systems were studied individually and have thus been presented in the same way. The Formula Renault was to be tested at Silverstone on the 15’th of June, 2007 and hence, got the precedence. The Pi Delta – more comprehensive of the two here was considered later so as to learn from the 2D system and make better use of it.
3.1 Formula Renault and 2D As mentioned earlier, the system is quite dated and a system check was essential. It was only logical to run the system prior to installation on the car. This also made it possible to go though the system thoroughly.
3.1.1 Pre – Installation Test Runs The loom and the sensors were already installed on the car. The logger however was not in place. Hence the logger was installed and the entire system wired up. The first test was to ensure that the logger shows up on the system, which failed to happen. The mini dash powered up when connected. The mini dash allowed the user to go through the various pages and change the dash display settings. However it did register an error – NO CAN ID’s received. The system is partially independent of the computer software and the dash display and the logger communicate with each other. The mini dash was functional but not communicating and the logger had failed to show on any of the COM ports on the computer. This indicated that the problem was at the logger end of the system. It is highly unlikely for a logger to have failed but all fingers pointed in that direction. It was learnt that only one computer had been configured to communicate with the logger, and the system is specific and would not work with any other computer
32
Data Acquisition for the Study of Racecar Vehicle Dynamics
available at the National College. Unfamiliarity with the system led to a belief that the problem had arisen due to the absence of drivers for the logger and the USB lead. Hardware support was acquired from Datron Technologies but did not resolve the problem. After contacting John Grist @ Datron Technologies, it was learnt that only a specific USB lead and a specific mini dash would communicate with the rest of the system. Hence various combinations of USB leads and mini dashes were tried. With the same error recurring, it was deduced that the problem was indeed with the logger. The wiring for the logger did not have clear markings for the negative terminal. This could have resulted in the diode within the logger being burnt out. To verify this, a different logger was tried with a different USB lead, loom and computer. Results confirmed that the problem was indeed with the logger and/or the loom. Following this, the loom and the logger were sent to Datron Technologies in Germany for inspection. As this occurred close to the test session, it was decided to use the second logger and a different loom on the car. Installed sensors would be calibrated and used. No additional sensors were to be tried in the test session on 15’Th June 2007. This would be a test to check the compatibility of the existing hardware and the software during race conditions.
3.1.2 Installation and ergonomics Installation and ergonomics go hand in hand. The data engineer has to make sure that all the wires are out of the way of the driver and the mechanics alike while at the same time being accessible to him. The best way to do this would be to have the loom where the mechanics are less likely to work under normal race conditions. On a racecar there is no such space and the installation was carried out with the chief mechanic present. The change in the logger posed some problems in the installation. The specified loom had two ‘boxes’, which basically are receivers for the connectors from the sensors. Since it was not a loom dedicated to the car, access to some parts of the car was easier than the rest. For example, access to the rear left side of the car was easier. The 33
Data Acquisition for the Study of Racecar Vehicle Dynamics
connectors from the rear suspension potentiometer, the rpm sensor and the water temperature sensor could be received, however the lap trigger, which was on the right side required additional connectors. Similarly, the front end of the car had more than adequate length of the loom posing packaging problems. Box 1 is the primary receiver and is also the point for downloading data from the system. It has a manual switch on it, which completes the system communication loop. For these two reasons, the box had to be placed in the safest possible place on the car i.e. inside the cockpit next to the driver. Care was taken to keep the box out of the driver’s line of sight within the cockpit as the box has a red LED, which blinks continuously and can therefore distract the driver easily. It was placed deep inside the cockpit. This move backfired as it had been carried out in the absence of the driver. With the driver inside the car, the box fitted in snugly next to the driver’s hip. The connectors were not long enough to make any last minute changes in the location of the box. Minute details of the installation process have been left out for obvious reasons. Problems caused by the change in logger will be discussed later in the report.
3.1.3 Calibration and Dash set – up Process This is where Winlt comes to the fore. The fact that analogue and digital channels appear on separate lists makes it convenient to work through the channels. The right channel numbers have to be watched and calibrated as it is easy to get the channel numbers mixed up. It is also worth mentioning ‘drift’. This occurs on analogue channels when they are not connected to a sensor. The displayed values increase continually. When a sensor is connected, the output values are constant and a multiple of the sensor’s range prior to calibration. This acts a quick check to make sure the user is looking at the right channel while calibrating. If a sensor has broken after installation, the corresponding channel will exhibit drift. Although output values due to drift can throw a user off, it is a useful tool.
34
Data Acquisition for the Study of Racecar Vehicle Dynamics
The mini dash set – up is also carried out through here. The required channels are added to a CAN list before they can be allocated to each page on the dash. General practice is to have the lap time and water temperature on page one. The LED’s on top give the RPM so the driver knows when to shift. Another useful function worth mentioning is the ability to load a virtual sensor list for both digital and analogue sensors. If a list was made in a previous outing, the calibration details and the list of sensors is stored within the system. For a later outing, provided there are no problems with the hardware, the sensor details can simply be copied and pasted into the real list. It also updates any tables that might be used for sensors - measuring water temperature for example. Fault identification is quite important when using complex systems with multiple sensors. The best and quickest way to get to a problem is to isolate it. Analogue sensors primarily alter the voltage reading within a range of 0V to 5V usually. A small linear potentiometer or a rotary potentiometer that can function over this specified range can be used as a system-checking device. The sensor itself has to be calibrated by the manufacturer and never used in race conditions. Alternately, a voltage generator can also be used for the same. It however tends to be bulky compared to the sensor. This practice proved useful almost immediately as one of the linear potentiometers on the suspension failed after installation and was identified using the sensor and the drift phenomenon. As the hardware is quite dated, the solder joint on the ground for the potentiometer had gone brittle and had come loose but not broken completely. As a result, the sensor would occasionally behave correctly. It thus escaped initial inspection before installation.
35
Data Acquisition for the Study of Racecar Vehicle Dynamics
3.2
Pi System and Formula Ford
Having learnt from the installation of the 2D system on the Formula Renault, ergonomics were given priority during the installation of the Pi hardware. Hence, no special mention has been made of the same. The logger and the software were dated and were checked before installation on the car. No major issues were faced during the pre – installation test runs or the calibration process as a lot had been learnt from the installation on the Formula Renault. A power supply was provided by Professor John Nixon facilitating the configuration of the entire system set – up away from the car.
3.2.1 Pre – installation test runs To carry out any test runs it was essential to check the logger status. It was also discovered that the software had expired. Consequently, the hardware was taken to Pi Research in Cambridgeshire for checks. The logger was tested to be OK and a newer version of the software obtained. The next stage was to check the loom status by using the sensors. As there was no mini dash available, the computer was the only link to identify faults with the system Following this, the entire set of sensors, the loom and the logger were laid out and connected. The Logger Management Software then showed all the sensors connected on various channels. As the sensors were not calibrated, only the voltage readings were considered to check their status. Spare channels were checked by adding sensors to them and checking their voltage readings.
3.2.2 Sensor Calibration Having found no major issues with the logger, the loom and the sensors, the entire set – up could be configured away from the car. The Logger Management Software provided great flexibility in terms of watching channels with existing calibration on them. The sensor voltage and the corresponding units were set into the logger and the sensors thus calibrated.
36
Data Acquisition for the Study of Racecar Vehicle Dynamics
Only concerns would be to ensure proper functioning of the wheelspeed sensors and changing the calibrations of the Lateral acceleration and the Longitudinal acceleration depending on the final position of the logger inside the car. Similarly, the steering position sensor calibrations would be changed accordingly given its complex location within the car. An RPM sensor was to be bought to complete a basic functional system for the Formula Ford. It was learnt that this car would not be tested during the course of the project, as the car would be ready towards the end and the circuit was not available for then.
3.2.3 Installation and Ergonomics The hardware was taken over to NC4M for installation on to the Formula Ford with the installation being carried out with Professor John Nixon. As is the case, ergonomics were a major concern as there was little space available inside the car for the logger and the loom. The best place for the logger was identified to be on the floor of the car inside the cockpit. The loom was then routed to both sides of the car to get access to the front and the rear end. Packaging the loom away provided the greatest challenge, as the loom is much larger than required for the Formula Ford. The excess loom was tied to the higher ends of the chassis to ensure the footwell has access at all times and the logger itself is not in the way of the driver’s ingress and egress. The brackets for installing sensors like the steering position sensor and the lap beacon were already made from previous attempts of installing the system. Steering position sensor would be mounted under the left side of the chassis tube while the beacon would find a place on the roll hoop. Installation of wheelspeed sensors proved to be a challenge as there are no mounting points on the uprights. Two different uprights on the front did not help matter either. The throttle position sensor could only be mounted behind the throttle pedal, as it is a small spring-loaded linear potentiometer. Again, the lack of space inside the car warranted the need to create new bracket for mounting the sensor. 37
Data Acquisition for the Study of Racecar Vehicle Dynamics
The technicians at NC4M had offered to prepare those brackets and mount them on to the car. With this done, the car can then be readied by installing the remaining sensors and tidying up inside the cockpit of the car. At the time of writing this report, the brackets were being awaited.
38
Data Acquisition for the Study of Racecar Vehicle Dynamics
4
Results
This chapter deals with the results obtained from the Formula Renault during its test at Silverstone Circuit. The Formula Ford and the Pi system have also been mentioned here.
4.1
2D and Formula Renault:
With the Formula Renault being tested at Silverstone, data would be available from the car for analysis. It must be remembered that this was the first test for both the system and the user, in race conditions. As with any system, there was a good chance of Murphy’s Law coming into play for any of the numerous components. Thus far all the system checks, installation and calibration processes had all been carried out in a workshop. The system still had to collect and deliver some useful and meaningful data over a race distance. The system being used for this test comprised of two suspension linear potentiometers, a rpm sensor and the mini dash. Compared to the complete inventory of hardware available, this system was less than basic. Without a wheel speed sensor and a lap trigger, there would not be any differentiation between the laps. These problems will be discussed in the section on general problems. An initial problem on the test day was that the logger failed to identify itself on the system even before the car was sent out for the first session. It would not communicate with the mini dash either, causing it to display the same – Receiving No CAN Id’s – message. Initial fears were that the logger had been damaged during transport. With time in hand, the entire wiring was checked and checked again to make sure everything was connected properly. The fault was the manual switch on Box 1 not being turned on. As mentioned in the previous section, it is the switch that completes the communication loop.
39
Data Acquisition for the Study of Racecar Vehicle Dynamics
The car then went out during its assigned 45 minute session. On his return, the driver confirmed that the LED’s indicating the RPM were working perfectly. Connecting to the computer turned up a surprise, the logger had failed to record anything. A quick check at the starting conditions for the logger revealed that the logger was set – up to start recording after the car had travelled a distance of 11m. The absence of a wheel speed sensor and the lap trigger did not provide the system with any distance data thereby not triggering the logger. This was changed to a time condition of 20 secs. The second session went smoothly. On return, the logger memory was partially full. On downloading, there were traces for both suspension movements and RPM. Again, the lack of distance data and lap trigger meant that all the data was recorded as a single lap of 32minutes. The pattern in the traces did reveal intervals when the lap might have commenced and ended. Closer inspection did reveal a few problems with the data:
•
The suspension travel for the front and rear had values as high as 13mm. This was a glaring mistake. The sensors had been zeroed in the workshop with the car still up on stands. The car was taken to the designated flat spot on the track and the sensors recalibrated and zeroed in. More meaningful data was subsequently obtained in the post lunch session. A few basic operations like changing the range of the axes and defining the mean of the travel had to be carried out in the analyser to make it more legible.
•
The RPM trace was way off the mark with maximum values reaching 45000 – an obvious calibration error. It must be stressed that the RPM sensor had been a last minute addition with the calibration being carried out just before the tests. As RPM is a digital channel, the error was assumed to be with the recorded pulses. It had a default value of 9. Increasing this to 20, the value had climbed even higher to 250000 in the following session. The LED’s had been scrambled as a result, again confirmed by the driver. The pulses were then reduced to as low as 1 with the same result.
40
Data Acquisition for the Study of Racecar Vehicle Dynamics
The channel details were then compared to the virtual sensor list from the system. The default value of 9 pulses was required by the system to function properly. The chief mechanic was consulted on the issue as he had assisted during the calibration of the RPM sensor. He admitted to not running the engine at max speed for fear of overheating. The pulses were then set back to 9 and a partial range defined for the channel. However, a brake master cylinder failure in the final session brought the day to an abrupt end. No data was recorded.
4.1.1 Results Obtained from the test session:
Figure 9: Results from 1'st Test Session 1
2
3
4
41
Data Acquisition for the Study of Racecar Vehicle Dynamics
1 – Trace for front and rear suspension movement Front Suspension – Orange Rear Suspension – yellow 2 – Overview window – showing the entire test session as a single lap 3 – Measuring value – showing values for all the displayed channels It can be seen that the values for RPM are inaccurate. The value of the damper pots confirms that they were not zeroed in on a flat spot. 4 – Circuit Window Error message shows up when a speed channel has not been assigned for any event. Displayed map has been imported from the samples given with the software
Figure 10: Results from 2'nd Test Session 1
2
42
Data Acquisition for the Study of Racecar Vehicle Dynamics
1 – Trace for front and rear suspension pots The sensors were zeroed on a flat spot. As a result, the front and the rear pots have a mean of zero. 2 – Measuring Window The value for the RPM has gone up considerably due to the change in number of recording pulses. The values for the suspension pots are now more sensible due to recalibration. As mentioned earlier, the logger and loom change meant that the test could only be used for a systems check and to check if the system can be further expanded further. A proposal for the same has been made in the Future Work chapter The available data does contribute greatly towards analysis and engineering the car. With more time in hand for the test, it would have been possible to run a complete system with the lap beacon and wheelspeed sensors to get data useful for engineering the car. Lessons learnt from this installation were put to practice with the Pi system. The concepts of signal drifts and calibration procedures were studied in detail and helped in working with the Pi system. Indeed more work was undertaken with the Pi system away from the car to ensure the system would be fully functional and assist in meeting the aims of the project.
4.2
Pi Delta System
On completion of the Formula Renault test session, it was decided to get the Pi Delta system fully functional before it could go on the car. The sensor calibration, loom and logger checks were completed away from the car. The car was only considered during installation of the hardware. It was learnt that the Formula Ford car would not get a test session similar to the Formula Renault, as the circuit would not be available. Hence, entire focus was shifted to the analyser software to get a workbook set – up in Pi Toolbox. Sample F1 data available with the software was used for the same. It was thus a process of 43
Data Acquisition for the Study of Racecar Vehicle Dynamics
reverse engineering since data was available and was to be used for the workbook. It must be remembered that more channels were available for this exercise compared to what would be available with the entire hardware on the Formula Ford. The analysis was left out at that stage so as to not complicate the process. Pi Toolbox allows for various displays that can be used on different worksheets for analysis. Before describing the process of creating the worksheet, it would be appropriate to understand a few basic terms associated with Pi Toolbox: Worksheet – is similar to a single excel spreadsheet that allows for various displays to be inserted. Displays – are various preset pages to which are inserted into the worksheets. Channels can be added to these displays for analysis. Multiple displays with different channels can be added to the same worksheet to analyse various channels that are linked together. e.g. Time/Distance Charts, Bar displays, Excel reports, Split reports, Navigators, X/Y charts, Summary sheets, Maps, etc. Workbook – is a combination of worksheets and displays used to create different pages for analysis. A worksheet can be assigned to some specific channels like dampers, damper speeds and load. All the driver performance related channels were grouped together in different worksheets and studied together. Workbooks are a feature of fully licensed Toolbox versions. Mr. Dave O’Neil at A1GP Team Ireland must be acknowledged for providing the writer with his license for use during the course of the project.
4.2.1 Workbook Set – up From the literature review, the approach to data analysis was set out. Various channels were grouped together as driver related, Engine and Gearbox related or 44
Data Acquisition for the Study of Racecar Vehicle Dynamics
chassis related. The same approach was maintained in setting up the workbook. The most basic channels necessary for analysis would be set – up initially, followed by the engine parameters. The chassis channels were to be looked at last as the resulting analysis would have a significant role in assisting the study of vehicle dynamics and load transfer module. Reliability issues would also be looked at in due course. Driver and Driver compare page Most drivers like to look at their driving in terms of data and where more time can be gained. A simple time distance chart showing speed, rpm, throttle, front brake pressure and gears was set – up for the same. A slightly more advanced version of the same worksheet was set – up for the race engineer to compare driver’s lap times for different outings. Other channels included were the steering angle, lateral acceleration. A G – G plot was also set – up to study the driver’s ability to use the maximum grip available. Every page had been set – up with a navigator for quick selection of lap times. The playback selection in the navigator display can be used to review the whole lap. It finds particular importance where video feeds can be afforded to trace driver movements inside the cockpit.
45
Data Acquisition for the Study of Racecar Vehicle Dynamics
Figure 11: Driver Compare page
A worksheet with individual wheelspeeds plotted has also been set – up as it assists in identifying wheel lock – ups on track. Temperatures and Pressures This worksheet was set – up in two halves using a tabulated outing report and a time/distance chart. The tabulated outing report is a simple column based report, which lists all the laps done in the outing and displays values for the channels selected. It is of particular importance as it allows channels to be selected twice and a different tell tale be selected. For example, oil pressure can be selected twice in the table and the minimum and maximum values be displayed. Other tell tales that can be selected include mean, standard deviation, difference, start and end. Start and end values are of particular importance in monitoring fuel consumption for the outing. This table was set – up for coolant pressure, coolant temp, fuel pressure, oil pressure, oil temperature and air temperature. All the fluids were selected twice to see the minimum and maximum values for the same. 46
Data Acquisition for the Study of Racecar Vehicle Dynamics
The other half of the worksheet consists of a time/distance chart with the same channels. Plotting fluid temperatures and pressures as a waveform assists in spotting trends. For example, if the oil pressure drops out regularly at the same place on track then it can be said that the oil pump is not efficient when the engine is pulling certain revs in that part of the track.
Figure 12: P and T page
A similar display was used for the engine worksheet. The channels monitored here being – Fuel used in Kg, fuel used in ltr, RPM, battery voltage, coolant pressure, coolant temperature, air temperature, oil pressure, fuel pressure and alternator current. Apart from fuel used and alternator current, all the channels were set to record maximum values. The average alternator current was monitored for every outing whereas fuel consumption was set to difference so that lap by lap consumption could be obtained. All channels were also plotted on a time/distance chart to spot trends.
47
Data Acquisition for the Study of Racecar Vehicle Dynamics
Map The software allows for a circuit map to be created using the lateral acceleration and steering angle. The created map can then displayed in every time/distance chart for reference.
Figure 13: Pi Toolbox Create Map Wizard
A separate map report was created where segments were defined. Turns and straights were identified and virtual sectors created by inserting split beacons. This allows creation of a graphical report where channels details for corner entry and exit can be monitored. A split report using the virtual sectors defined has also been set – up. The map report in this instance shows the gear selected, corner entry speed and exit speed along with max. speed through the corner.
48
Data Acquisition for the Study of Racecar Vehicle Dynamics
Figure 14: Map Report
These 5 worksheets set – up would be adequate to carry out any basic analysis of driver performance and engine behaviour. It would also suffice to run a car in amateur or semi – pro racing categories. But it does not consider any of the chassis sensors. A great deal about the set – ups and set – up changes can be learnt through simple use of damper pots and load cells. It is where the importance of math channels comes to the fore. Math Channels Various aspects of set – up changes such as dynamic ride heights, dynamic load transfer and roll can be assessed using Math channels. Damper speed plots allow for low speed and high speed tuning of the dampers. Before setting the displays up for these channels it was vital to write the Math Channels for the channels. Some math channels in the workbook had to be set up, only to be used as channels in the formation of other math channels. Eg. Total brake pressure had to be set – up to create the brake bias channel but the total brake pressure channel itself has never been used anywhere in the workbook.
49
Data Acquisition for the Study of Racecar Vehicle Dynamics
Other channels such as compare time were set up to see where and how much time was lost by a driver while comparing different laps/outings. But going back to the load transfers during cornering, the channels that were set – up were: 1. Brake Bias: ( Front brake pressure /total brake pressure ) * 100 2. Damper velocity: obtained by differentiating channel output from damper pots for all four corners of the car. 3. Load Front: obtained by adding loads from FL and FR load cells. Alternatively, strain gauges could be used and then calibrated to get loads from them. 4. Load Rear: obtained from the rear two load cells or strain gauges. 5. Total Load : addition of front and rear loads. 6. Crossweright Dynamic: to check loading at diagonally opposite corners. ((Load
FL
+
Load
RR)
/
Total
Load
)
*
100
7. Ride Height (FL, FR, RR, RL ) – obtained by defining the static ride height for the outing as a constant and then subtracting the Damper position. outing
constant
[static
ride
height
]
–
Damper
position
8. Roll (front and rear)- obtained by subtracting the FL ride height from the FR ride height. Similarly the rear roll can be worked out. 9. Roll
%:
obtained
from
introducing
conditional
expressions.
for lateral acceleration > 0.95, (( rear roll – front roll ) * 100) / (rear roll + front roll ) &
50
Data Acquisition for the Study of Racecar Vehicle Dynamics
For lateral acceleration < -0.95, ((rear roll – front roll ) * 100 / (rear roll + front roll For a car set – up for any particular circuit with corner weights noted and load cells mounted on the push rods, these channels would allow the user to understand the load transfer through every corner. Damper tuning can also be carried out looking at the speeds. Changes in ride height due to aero loading can also be studied. Consequently, different time/distance charts and XY plots were set – up for chassis channels and the workbook completed. The complete list of math channels along with their syntax has been provided in Appendix A. Only the simple math channels were set – up as part of the workbook.
51
Data Acquisition for the Study of Racecar Vehicle Dynamics
5
Discussion
This chapter reflects on the project and highlights the various issues faced while working on both cars. The work done on the Pi Toolbox software has also been critically analysed. The technique to handle both data acquisition systems was developed through the literature review. Over the course of the project, however, the technique has had to be modified in some cases. With the Pi Delta system, the process has been reversed due to lack of testing time with Formula Ford. These differences have been discussed in the following lines.
5.1
2D system and Formula Renault
The approach to the Renault was to study WinARace and Winlt first, as some of the hardware was already installed on the car. Also, the writer’s inexperience with data acquisition software shifted priorities away from the car. The car also has an ECU leading to the belief that the engine parameters could be monitored. The installation process was undertaken in advance with the test in mind. But problems were discovered with the system. Considerable time was spent in getting support from the manufacturer and isolating the problem. The root cause was identified a burnt out diode in the logger which had to be sent back along with the loom to Germany for checks and repairs. The problems that put the main logger out of action triggered a chain of events that affected the project right until the test day. Two of the main consequences of the logger being out of action were: •
Lack of G sensors: no acceleration data meant there was no way to analyse the car’s progress on track.
•
Lack of a dedicated loom: raised packaging, access problems. Also resulted in running a partial system from the already restricted system.
52
Data Acquisition for the Study of Racecar Vehicle Dynamics
The absence of a dedicated loom was a major set back as the new loom could not reach far enough to receive sensor connectors. This resulted in a loss of a wheel speed sensor and the lap trigger, both of which were installed and operational on the car. The additional link leads required could not be acquired in time for the test. This loss of sensors meant no distance data from the vehicle. As mentioned earlier in the report, no distance data meant all the data was collected as a single lap of 30+ minutes. This then coupled with lack of G sensors meant that no track information could be obtained. The car’s progress could not be monitored either therefore reducing any chance of optimising the set – up through the test. Engine parameters could not be monitored, as the sensors could not be connected to replacement loom. With this handicapped system in place and the RPM sensor being a last minute addition, the test session was reduced to being a glorified system check for a spare logger. The damper data obtained through testing could not be used further due to the absence of distance information thereby eliminating any damper tuning. The data from the first test session had shown exaggerated damper movements. This was identified to the damper pots being zeroed with the suspension at full droop and not on a flat spot. Earlier system checks were restricted to identifying status of the logger, loom and sensors. From the lessons learnt through the test, the process was extended to obtaining data from the system and the car, if possible. Status identifications did just that. System compatibility within the system itself and the system and the car were issues that had not been considered previously. E.g. whenever a wheelspeed sensor is passed over a trigger, pulses will be recorded but during race conditions a faulty sensor drops out showing gaps in the trace – a simple fact not identified by status checks. No further test sessions were available. Financial constraints meant that the logger and loom would not be ready in the immediate future to complete the system. Experience gained by working with touring car teams assisted in compiling a sensor list for further expansion of the system which has been listed in the future works chapter. 53
Data Acquisition for the Study of Racecar Vehicle Dynamics
5.2
Pi and Formula Ford
As mentioned earlier, no further test dates were available. Hence, no race data would be available from the Formula Ford. Focus was shifted to optimising Pi toolbox to represent data whenever it was available. Facilitating the study of vehicle dynamics had been the aim from the outset. Sample data with many more channels was available and thus used for setting up the workbook. The single biggest flaw was the fact that a third party’s data was used. There was no information about the conditions (track, car or weather) for when the data was obtained. Simply put, the biggest error was made during the first step. It wasn’t, however, consequential in the outcome, as the process had been reversed. Instead of testing the car and then acquiring the data, available data was used to learn more about the analyser. The installation and calibration of the system was carried out independently. As a result, the installation and the analyser software have been discussed separately.
5.2.1 Pi Delta installation Most of the brackets required for the sensors like steering position, throttle position, the lap beacon, etc. were available and thus used. Not enough time was spent on improving the sensor mounting positions or packaging. The available loom is longer than necessary for the Formula Ford and has posed problems. The excess wiring has been tied away inside the cockpit and footwell adding weight and reducing available space even further. For normal operation under race conditions, it would be standard practice to get the right loom. For the purpose of the project, the loom offers great flexibility. Excess channels when installed could be accessed easily by simply re routing the loom. Logger position and the loom connectors have been assumed to be optimum. This could prove wrong with a driver inside the car. The logger has also been mounted on the floor of the car providing no protection from vibrations possibly resulting in signal noise.
54
Data Acquisition for the Study of Racecar Vehicle Dynamics
The sensors have been installed and calibrated away from the car. Having learnt from the 2D system, the sensors were physically moved and the results manually logged. But their operation in race condition could not be tested. Wheelspeed drop outs will not register on the system until it is tested in race conditions.
5.2.2 Pi Toolbox The workbook developed has been done so by user preference. For example, the driver page has been set – up with speed, rpm, throttle, brake and gear channel. A particular driver may or may not prefer this set – up. Similarly, driver compare page may not convey what the engineer is looking for. The setting up of math channels is where Toolbox is quite advanced compared to most other data analysis software. The load transfer and ride height channels will provide tons of information on the dynamic state of the car when it is on track. Ride heights and damper position traces can be used together to set the suspension geometry for anti – dive or anti – squat. Geometry changes, ride height changes, aero loading can all be confirmed quickly using these math channels. Some mistakes were made while setting these channels up. The motion ratio and compliance was not taken into account when the damper position page was set – up. As a result, the data when available will show the movement of the damper pot and not the damper itself. Advanced math channels for dynamic ride height changes, pitch, rpm in each gear and understeer / oversteer have been left out as it requires various parameters like compliance in the suspension, etc to be measured. These require investment in equipment and time, which could not be justified for a car not to be tested. Provision has also been made for a separate worksheet for brakes and brake bias. As was stated in the first project brief, simulations were to be run for brake cooling and then compared with data from the racetrack. Provided brake temperature sensors or rubbing thermocouples are added to the system, this comparison can be done at a later
55
Data Acquisition for the Study of Racecar Vehicle Dynamics
stage. Similar comparison could also be made for differential being used on a car using wheelspeed on the rear end. It must be remembered the entire set – up has been carried out using data with many more channels available. The Formula Ford may not benefit from load cells, 4 wheelspeeds, brake pressure sensors etc. and most of the work carried out may not apply. To continue the project further would require additional funding and more time to document the various constants that will be required for finding the final few tenths of a second.
56
Data Acquisition for the Study of Racecar Vehicle Dynamics
6
Conclusion
This chapter provides a conclusion to the undertaken project and summarises the findings. The aim of the project had been to have a Formula Renault and a Formula Ford single seater complete with two different data acquisition systems installed on it. These two cars were then to be used as educational tools for the study of racecar vehicle dynamics and data acquisition. In terms of achieving the aims and objectives, the project has been partially successful. It would have been ideal to have two car sets worth of data, analyser workbooks and math channels set – up. Various problems ranging from installation and packaging to sensor calibration errors and logger burnouts were encountered. Various solutions and improvisations were implemented to overcome these. The approach developed using the literature review had been refined and/or altered over the course of the project. The aim was achieved in three parts: 1
The Formula Renault had a partial 2D system installed and was tested at Silverstone.
2
Lessons learnt from the test were implemented in the installation and calibration of the Pi Delta system on the Formula Ford. The car was not tested on track.
3
Sample data available with Pi Toolbox was used to create a workbook with pages set – up for data analysis. Math channels assisting the study of various aspects of vehicle dynamics were set – up in the workbook.
Further improvements can be made to the process of data acquisition and analysis with time and financial resources to have a complete system installed on a car that will be tested on track more often.
57
Data Acquisition for the Study of Racecar Vehicle Dynamics
7
Future Work and Possibilities
This chapter delves into the work that could be carried out in taking the project further. The possibilities of expanding either system and the emerging possibilities have been discussed here. After running both Pi and 2D systems and having worked with the analyser, recommendations for expanding the system towards a certain objective can now be made. These will obviously be subject to financial restraints. But before getting into financial aspects, it is vital to explore the three possible objectives for the proposed expansion.
•
Driver Oriented System: A system set – up to inspect driver inputs and then compare driving styles and also get feedback on set – up changes from the driver. This is, perhaps, the most inappropriate system to install as neither National College nor Cranfield University specialise in driver training or strategy planning according to individual drivers. A cost justification cannot be provided for the system e.g. a specialised brake pedal sensor from 2D costing £300.
•
Vehicle Oriented Data Acquisition System: This would incorporate a sensor system that allows the students @ NC4M to better understand the effects of the mechanical changes and repairs carried out by them – a representation of the cause and its effect. It must be emphasised here that NC4M is primarily concerned in training technicians for the motorsport industry. The use of a data acquisition system as an engineering tool for racecars will find little exposure here. This would have to be a basic and standard system that goes on to the car with little margin for improvement and/or optimisation.
This
would also be the most cost effective system with immediate results for the students coming in next year @ NC4M.
58
Data Acquisition for the Study of Racecar Vehicle Dynamics
•
Test Oriented System: This could be a very flexible system that allows for more sensors to be installed on to the vehicle at a later date when the car is being tested for some specific performance gains, i.e. brake efficiency, Pressure sensors for aero loading on straights.
This would by far be the most expensive solution to pursue, as it would require an excess of sensors and connectors to be readily available. It would also demand a logger with multiple channels available meaning the loom for the primary logger would have to be repaired. This system, however expensive, would be beneficial to the institution in the longer run. The Formula Renault is also a better car on which a test system can be installed, because it gets out on the track more often than the Formula Ford. It is inspected and constantly repaired hence maintained in prime condition throughout the year. Compared to the Formula Ford car, the Formula Renault it is more current in terms of specifications and design, making a stronger case for it. The 2D system is not as complex as some of the others and could result in more interest from students and teachers alike at the college. The Pi Toolbox however offers more options in terms of data representation. Programming of Math channels, if taken further, could establish a strong link between knowledge gained from books and its applications on a car going around a circuit. The proposed list of sensors for expansion is as follows: No.
Equipment 1
Quantity
Loom
2 specified for the car
2
Wheel Speed Sensor
2 Front and Rear
3
RPM sensor
1
59
Data Acquisition for the Study of Racecar Vehicle Dynamics
4
Gear Position Sensor
1
5
Temperature Sensor
2 Oil and Water
6
Pressure
4
Oil, Coolant, Brake Pressures, Fuel 7
Load Cells
4
8
Damper Pots
3
9
Brake Temperature Sensors
1
It is likely that a vehicle-oriented system will get the nod due to fewer complications and financial considerations. The most viable solution would be to run the Formula Renault with a Pi Delta system along with the workbook set – up in the analyser. This could lead to collaboration between NC4M and Cranfield University in running the cars in a semi - professional racing category, using students, as a joint exercise. Provided the list of sensors is approved, the brake temperature simulation project can be completed. Simple structures could be used as brake ducts for cooling and more information obtained for the same. In terms of sheer possibilities, the list could be endless.
60
Data Acquisition for the Study of Racecar Vehicle Dynamics
8
References 1. Pg 38, Competition Car Data Logging, Simon Mcbeath, Haynes Pubslishing, 2002. 2. Pg 42, Competition Car Data Logging, Simon Mcbeath, Haynes Publishing, 2002. 3. Pg 68, Competition Car Data Logging, Simon Mcbeath, Haynes Publishing, 2002. 4. Pg 31, Driver Data Analysis and Comparison, M. Albayati, MSc. Thesis, 2006, Cranfield University. 5. Pg 93, Competition Car Data Logging, Simon Mcbeath, Haynes Publishing, 2002. 6. Pg 115, Competition Car Data Logging, Simon Mcbeath, Haynes Publishing, 2002. 7. Pg 122, Competition Car Data Logging, Simon Mcbeath, Haynes Publishing, 2002. 8. 4 Stroke RPM sensor Information Sheet, Pi Research, www.pixpress.com Copyright: Pi Research, 2006. 9. 4 Stroke RPM sensor Information Sheet, Pi Research, www.pixpress.com Copyright: Pi Research, 2006
61
Data Acquisition for the Study of Racecar Vehicle Dynamics
10. Standard Wheelspeed Sensor Technical Sheet, Pi Research, www.piresearch.co.uk Copyright: Pi Research, 2006 11. Pg 6. Course Notes Msc. In Motorsport Engineering and Management Motorsport Electronics and Data Acquisition. 12. Pg 7. Course Notes Msc. In Motorsport Engineering and Management Motorsport Electronics and Data Acquisition 13. Pg 10. Course Notes Msc. In Motorsport Engineering and Management Motorsport Electronics and Data Acquisition 14. Pg 3. Course Notes Msc. In Motorsport Engineering and Management Motorsport Electronics and Data Acquisition 15. Pg 2. Course Notes Msc. In Motorsport Engineering and Management Motorsport Electronics and Data Acquisition
62
Data Acquisition for the Study of Racecar Vehicle Dynamics
9
References: Table of Figures
Figure 1: RPM plot with details
14
Pg 38, Competition Car Data Logging, Simon Mcbeath, Haynes Pubslishing, 2002 Figure 2: Speed Trace for Thruxton Pg 68, Competition Car Data Logging, Simon Mcbeath, Haynes Pubslishing, 2002
Figure 3: Throttle Plot at Silverstone International
17
Results obtained as part of the project. Figure 4: Corner Entry Oversteer
19
Pg 98, Competition Car Data Logging, Simon Mcbeath, Haynes Pubslishing, 2002
Figure 5: G - G Plot
20 Pg 113, Competition Car Data Logging, Simon Mcbeath, Haynes Pubslishing, 2002
Figure 6: Engine Pressures and Temperatures
21
Results obtained as part of the project
63
Data Acquisition for the Study of Racecar Vehicle Dynamics
Figure 7: Damper Displacement, Speed and Load
23
Pg 122, Competition Car Data Logging, Simon Mcbeath, Haynes Pubslishing, 2002
Figure 8: Pi Express 4 stroke RPM box
24
4 Stroke RPM sensor Information Sheet, Pi Research, www.pixpress.com Copyright: Pi Research, 2006
Figure 9: Driver Compare page
46
Results obtained as part of the project Figure 10: P and T page
47 Results obtained as part of the project
Figure 11: Pi Toolbox Create Map Wizard
48
Results obtained as part of the project Figure 12: Map Report
49 Results obtained as part of the project
64
Data Acquisition for the Study of Racecar Vehicle Dynamics
10 Appendix A: Math Channels 1
Brake Bias:
(gate([Brake Pres Frt]>60,([Brake Pres Frt] / ([Brake Pres Frt] + [Brake Pres R]))))*100 Reads brake bias only when front brake pressure reads a realistic value. More than 60 Psi in this case 2
Total Brake Pressure
[Brake Pres Frt] + [Brake Pres R] 3
Combined G
sqrt(([Accel Lat]*[Accel Lat])+ ([Accel Long]*[Accel Long])) 4
Compare Distance
compareDist( [Elapsed Lap Time] ) distance based comparion of elapsed lap time 5
Load Transfer %
(([Load FL]+[Load RR])/([Load Total]))*100 6
Damper velocity
derivative( [Damper FL Adjusted], Ignore) 7
Fuel Used in Kg
[Fuel Used] * 0.79 Fuel density for Elf racing fuels available in the UK
65
Data Acquisition for the Study of Racecar Vehicle Dynamics
8
Ride Height
GLOBALCONST(Static Ride Height Front)-([Damper FL Adjusted]*GLOBALCONST(Motion Ratio Front)) 9
Front Roll
[Ride Height FL]-[Ride Height FR] 10 Roll % choose([Accel Lat]>0.95,(([Roll Rear]-[Roll Front])*100)/([Roll Front]+[Roll Rear]),choose(([Accel Lat]<-0.95),(([Roll Rear]-[Roll Front])*100)/([Roll Front]+[Roll Rear]),0)) 11 Roll Rear [Ride Height RL]-[Ride Height RR] 12 Wheelspin [Speed]-(([Speed RL]+[Speed RR])/2)
66
