Features - Technical
JANUARY 6, 2001
SEA Motorsports Engineering Conference - 2000
BY PETER WRIGHT
The fourth SAE Motorsports Engineering Conference (MSEC) took place in Dearborn, Michigan at the end of November, attracting 38 technical papers on motor sports related subjects. The number of papers is down on previous years, and I believe that this reflects the increasing level of competition in major racing series and the increased involvement of the motor industry. Engineers involved in these series simply do not have the time to write up their research to the level required for publication, and much of the basic research is carried out in cooperation with engineers from major-manufacturer partners, and they do not want their technical secrets widely disseminated.
Nearly 25% of this year's papers are on safety issues, one of the few research topics which is not competitive, and the majority of the remainder fall into one of three categories: Formula SAE (Formula Student in the UK), University research projects, and research by companies selling products or services into the motor sport industry. In some cases, the multiple authors fall into more than one category, providing an insight into the way in which research is funded and carried out in this field. As a result of the confidentiality requirements of racing, not many of the papers are on the subject of cutting edge technologies, however it is through published research that one can sometimes catch a glimpse of what is going on inside racing teams, and it is usually via this medium that technologies eventually emerge into the public domain.
A full list of the papers is appended to the end of this article, and they are available through www.sae.org. I was not able to attend all the oral presentations, but I have highlighted a number of papers, taken in the order they appear in the list, that particularly caught my eye either because they announced something new or because they provided an insight into a particular direction which motor sport technology and engineering is taking.
Motor racing has grown up as an industry to the extent that universities are now studying how it works. Mark Jenkins' paper (2000-01-3535) seeks to determine how a technology such as ground-effect is developed, and how competing organisations respond to the introduction of new technologies.
The dynamic failure of composite structures is probably one of the more complex issues to analyse in mechanics, as Rolls-Royce discovered when they tried to pass the bird-strike test with CFRP fan blades on the RB211. Dynamic FEA software that can analyse composite structures is now being used in the motor industry and in motor racing. Ari Caliskan's paper (2000-01-3536) provides an in-depth look at the theory and practice of the energy absorption behaviour of carbon and glass fibre composite materials, and examines the effect of material and laminate properties, and geometric effects. He notes that predicting the failure modes within the fibre, matrix, and fibre/matrix interface make FEA extremely difficult.
The HANS Head and Neck Support system is fast becoming a standard piece of safety equipment, and the Daimler-Chrysler paper (2001-01-3541) describes some of the extensive test work carried out on it to determine its sensitivity to a number of test conditions, in order to prove its robustness. The effect of neck length, and shoulder belt slack and geometry were assessed, and the critical injury criteria shown to be sensitive to them. However, HANS provides a substantial reduction in injury potential compared to the baseline tests without HANS. 30¼ angled front impacts and side and rear impacts were run and showed that HANS did not increase injury potential in any of these cases, and may even provide increased neck protection in severe rear impacts.
Another paper from the same source (2000-01-3543) describes the development and testing of an airbag system for open cockpit single-seaters, and the results compared with HANS. Both systems provide similar levels of protection, but it is clear that HANS was by far the simpler, lower-cost and less risky system for on-going development. The airbag system could not be mounted in the centre of the steering wheel, as is usual in road cars and the Mercedes DTM cars, but was stored in the forward rim of the cockpit. A gas-generator deployed the bolster shaped bag across the steering wheel and another gas generator inflated it. The problems of explosives in the cockpit, electronics reliability, and dealing with multiple impacts mitigate against the airbag approach when a passive system with equal effectiveness is available.
The work of Gordon Blair on the theory and synthesis of 2-stroke and 4-stroke engines is well known, and this lifetime of work is now embodied in simulation software Virtual 2-stroke and Virtual 4-stroke, available from www.optimum-power.com . His paper (2000-01-3546) describes the use of Virtual 4-stroke to analyse the potential of engine configurations permitted in World Superbike racing, i.e. 750cc 4-cylinder, 900cc 3-cylinder and 1000cc 2-cylinder.
As stated in the paper: "The requirements for simulation model accuracy are clearly spelled out by Blair, such as the employment of non-isentropic thermodynamics and gas dynamics and particle tracking throughout the engine and ducting, branched pipe models which include relevance to inter-pipe junction angles, multi-zone combustion processes, and the use of maps of discharge coefficients based on both pressure ratio and geometry for every type of boundary encountered within the powerplant envelope. In the absence of some or all of the above requirements, some engine simulation engine models need Ôcalibration with experimental data' to be Ôaccurate'. In the design context being investigated here only the engine concept exists so such Ôcalibration' is not possible. Hence, an engine simulation which is Ôabsolute' in its accuracy is essential."
No 900cc 3-cylinder engine has competed in World Superbike racing, but the analysis shows that this configuration has the potential to be very competitive in terms of power output, frontal area, mass and bulk when compared to the alternatives.
Aiolis Engineering Corp. were contracted by Lola Cars International to update, install and calibrate their new 50% wind tunnel (Paper 2000-01-3547). Lola acquired a former British Aerospace wind tunnel, and Aiolis built a new 2.7m x 2.47m test section, contraction, settling chamber, wide-angle diffuser, and fan blades and motor. They also installed a rolling road, overhead balance, and cooling system to maintain constant test conditions. The result is a cost-effective 60m/s wind speed wind tunnel, ideal for developing the types of cars that Lola produces. Full calibration results are presented, and they indicate that the flow quality is better than the original, very high quality, aerospace tunnel.
At the other end of the $$$-scale of motor racing wind tunnels, we get a glimpse of how the big-spending Formula 1 teams, such as McLaren and Toyota, are spending their money on their new, secret wind tunnels. Sverdrup Technology Inc. presented a paper (2000-01-3549) on two new developments in motor racing wind tunnels: adaptive wall test sections and MTS's Flat-Trac rolling road technology. Current 50% model tunnels are limited to 50-60m/s. The new generation is designed for 60% models, with reasonable quality simulation at full scale. Wind speeds of 100m/s will be the new standard, yielding around twice the Reynold's Number of the smaller/slower facilities.
To quote from the paper: "The idea of an adaptive wall test section is simply to shape the side and top walls of the test section such that they correspond to external streamlines that would be present over the vehicle on the open road. More formally stated, "É.if the flow angularity distribution on a control surface surrounding a body coincides with the distribution that would be present at that location for the body located in an infinite domain, the body will experience no interference effects."" For the adaptive wall wind tunnel, the side and top walls of the test section become the control surface. Shaping these walls to correspond to streamlines on the open road means that there will necessarily be no interference effects to degrade the external aerodynamic simulation. This implies that no correction of force coefficients is needed and that the local aerodynamics on the vehicle are properly simulated.
"With racecar developers looking for aerodynamic advantage with the design of essentially every element on the vehicle, situations will arise where the performance increments between various design options are much smaller than the correction increment needed to translate the wind tunnel measurements to on-road performance. Along a similar line of thought, the need for correction arises fundamentally from improper flow simulation over the model. As advanced diagnostic tools such as Pressure Sensitive Paint, Particle Image Velocimetry, and Planar Doppler Velocimetry- which provide detailed, spatially-distributed mappings of local flow structure - become more commonplace in the wind tunnel, proper flow simulation at localized positions on the vehicle is becoming increasingly important. Although post-test corrections are suitable for force and moment coefficients, they are not capable of improving flow simulation in the wind tunnel."
The paper describes how adaptive wall technology is applied, states that blockages as high as 30% are possible without the need for data correction, whereas 20% is the maximum for contoured walls and around 10% for solid walls. The payoff comes from the capability to test larger models in a given sized test section, and hence for a given desired model size the facility size, cost and power requirements are less.
MTS have developed their Flat-Trac, moving steel belt technology over a number of years, applied to tyre test rigs and vehicle dynamics rigs. Only now is it being applied to rolling roads, to overcome the speed limitations of elastomer belt systems. Flat-Trac can either be applied as a full width belt, with vertical force measurement through the belt, or as a 5-belt system with a central narrow belt and individual wheel mini-belts. Both approaches are suitable for full-size vehicles as well as models.
The FIA has mandated the fitment of Accident Data Recorders (ADR) to Formula 1 cars for the last four seasons and to F3000 cars from 2000. The results of the data gathered, along with a description of Delphi Automotive System's ADR2, which is also fitted to IRL cars, is presented in my paper (2000-01-3551). The deceleration performance of run-off areas compared to wet and dry tracks is summarised, and a method of categorising the severity of an impact is described.
The UK's Transport Research Laboratory is commissioned by the FIA to reconstruct severe accidents, particularly where injuries have occurred. The methods used are described in Andrew Mellor's paper (2000-01-3552), in which the results of a number of impacts, reconstructed to reproduce the damage to the helmet and car structure are reported. The research is part of an ongoing TRL-FIA motorsport safety programme and "ÉÉprovides a better understanding of accident kinematics and head injury mechanisms. This work has been extremely important in providing detailed data on driver tolerance to head injury, for both translational and rotational motion, which both contribute significantly to injuries sustained during head impact. This work has indicated that the threshold for driver head injury may be somewhat higher than current published and accepted criteria."
One of the biggest problems with this sort of safety research is knowing exactly the decelerations experienced by the driver. ADR data supplies decelerations at the car itself, from which dummy decelerations can either be modelled or reproduced by either the TRL techniques or on a Hi Ge sled. However they are still only dummy decelerations. Drivers are generally resistant to being fitted with sensors, especially during races. Asking a driver to attach accelerometers to his head and then go out and crash please, is likely to elicit a negative response. The US Air Force Research Laboratory has developed a number of human mounted accelerometer systems for ejector seat research, and these are being adapted for motor sports applications (Paper 2000-01-3557). The most promising are tiny tri-axial accelerometers that fit inside a driver's earplugs, coupled as closely as possible to the skull. Calibration is carried out against a set of accelerometers held firmly between the teeth - something a driver would not do while driving, whereas he might be prepared to have them in his ears.
There was some criticism that the Formula SAE and Formula Student based papers are inappropriate at this conference, as they do not portray the state of the art. They are perhaps, more student theses than pure research, but their publication does build up a useful knowledge base for this wonderful formula. The projects however, do also stimulate members of mechanical engineering faculties to carry out research on motor sport topics, and one such paper is from the University of Leeds, whose Formula Student cars have led the way for the UK (Paper 2000-01-3563). In this paper, three approaches to lap time simulation (steady state, quasi-static and transient strategies) are compared, to evaluate the trade-offs between simulation accuracy and complexity i.e. the need to gather accurate, dynamic input characteristics, software development and computing speed.
Hacker, Lewis and Kasprzak's paper (2000-01-3564) looks at techniques for computer optimisation of racing cars. Taking just three parameters: aerodynamic downforce distribution, weight distribution and roll stiffness distribution, they model the performance of a car on two skid pads of different radii, and analyse the trade-offs between these parameters to determine the optimum solution using "Pareto analysis". The task is very computer intensive and the authors describe how it can best be carried out with a parallel computing architecture. 32 Sun Ultra 5 workstations are employed, configured as a supercomputer. These techniques, though in their infancy, show how "expert systems" may be employed by race engineers in the future to set-up a car.
It seems that the mechanism by which a cycle or motorcycle is steered is still not fully understood. Pierre Ethier's paper (2000-01-3565) describes the limitations of the established "countersteering" theory, and develops a servomechanism steering theory. It is pretty complex stuff, but relevant to the development of motorcycle based machines in which the rider is restrained within an enclosed vehicle.
Two papers from TAG Electronics provide insight into racing car sensors (2000-01-3566) and multi-burst telemetry (2000-01-3567). Sensor reliability is as important as performance, and the techniques used to achieve it in the hostile environment of a Formula 1 car are described. Multi-burst telemetry provides engineers with the means to monitor a running car in almost real time, feeding the banks of computers seen in the pit garages at all the major racing series.
Chuck Hallum has presented papers on his unique tyre analysis approach at the last three MSECs. In the first he put forward an explanation for the extraordinary performance of dragster tyres, and proposed a mechanism by which they are able to generate high levels of grip in the absence of downforce. In the last two he has developed a theory that describes what is happening in the contact patch of a racing tyre, including a thermal model of the tread. His paper this year (2000-01-3571) focuses on the performance of NASCAR tyres, comparing the changes due to a 1psi tyre pressure change with those due to other suspension parameters.
The Lotus Active Suspension system pioneered an approach to suspension dynamics in which each mode i.e. heave, pitch, roll and warp, was given an independent stiffness and damping. Since active suspensions were banned in 1993, race suspension engineers have been seeking to obtain some of the active features in passive suspensions. Erik Zapetal's paper (2000-01-3572) describes a "balanced suspension" that, albeit with considerable complexity, also de-couples the 4 modes passively. He describes the advantages that a soft warp mode achieves in single-wheel bump performance whilst maintaining a high roll stiffness. Lotus eventually abandoned the modal approach for active suspension in Formula 1, returning to an independent, single-wheel strategy with lateral load transfer distribution control.
The FIA has developed a piece of software called Circuit and Safety Analysis System (CSAS) that calculates the speed of a car on a circuit, and then the size of the run-off areas and barriers required to meet the FIA Circuit Guidelines (Paper 2000-01-3573). Deceleration rates on- and off-track are based on statistical data derived from Accident Data Recorders, and plots of the trajectory of an errant car are superimposed on circuit plans in AutoCAD. Barrier characteristics are used to calculate the residual velocity after impact.
Metz and Metz propose a system for slowing cars that deploys a plough-like device beneath the car to engage with the gravel bed (Paper 2000-01-3574). Tests with a 1/10th scale model and a kart are described, but not what would happen if the device is deployed inadvertently.
In the final safety-related paper (2000-01-3575), Edward Kasprzak develops a theoretical approach to analysing barrier placement relative to a corner, showing that there is a "worst" distance for impact severity and examining the sensitivity to placement closer and further than this critical position.
Two engineers at MTS Systems Corp. have been able to utilise the company's facilities to assist them in the design of a Grand Prix racing motorcycle (Paper 2000-01-3576). They have not only been able to use CAD, FEA and fatigue analysis software to design the structure, but they were also able to model the bike's suspension and structural dynamics, and validate the model on a 2-post rig. Pretty scientific stuff for an amateur project!
The final paper (2000-01-3577) describes a large-scale 3-D coordinate digitising system, employing 2 rotating laser heads and an operator-carried wand. Touching the wand on the point to be digitised stores the coordinates in a CPU worn by the operator. Accuracy is claimed to be almost 1/32 inch (0.8mm), which makes this system interesting for the measurement of racing car suspension kinematics, structural alignment and set-up.
The quantity of papers may be down, and the SAE needs to reverse this trend by casting the net wider, but there was plenty of interesting papers to dip into, many of which provide a rare view of some of the things going on inside a secretive industry.
Proceedings of the 2000 SAE Motorsports Engineering Conference.
2000-01-3535 Codifying Engineering Knowledge in Motorsport
Mark Jenkins, Cranfield School of Management
2O00-01-3536 Crashworthiness of Composite Materials & Structures for Vehicle Applications
Ari G. Caliskan, Safety R&D Dept., Ford Motor Co.
2000-01-3537 Fundamental Parameter Design Issues Which Determine Race Car Performance
Andrew Deakin and David Crolla, School of Mech. Eng., The University of Leeds