RESEARCH REPORT340 A STUDY OF VARIOUS CAR ANTI-LOCK BRAKING SYSTEMS. TRANSPORT AND ROAD RESEARCH LABORATORY Department of Transport

Transcription

1 TRANSPORT AND ROAD RESEARCH LABORATORY Department of Transport RESEARCH REPORT340 A STUDY OF VARIOUS CAR ANTI-LOCK BRAKING SYSTEMS by B J Robinson and B S Riley Crown Copyright The views expressed in this Report are not necessarily those of the Department of Transport. Extracts from the text may be reproduced, except for commercial purposes, provided the source is acknowledged. Vehicle Safety Division Vehicles Group Transport and Road Research Laboratory Crowthome, Berkshire, RG11 6AU 1991 ISSN

2 CONTENTS Page Page Abstract Braking on a curve Introduction Ford Escort Background Fiat Uno Theory Honda Prelude Systems One channel - Rear axle control Mazda 626, BMW 320i, 17 Ford Granada 4.3 Split surface braking Two channel - Front wheel control Two channel - Front and rear axle control Ford Escort Fiat Uno Honda Prelude Three channel - Front wheel and rear axle control 2. The Test Vehicles 3. The Test Procedures 3.1 Description of surfaces 3.2 Measuring equipment used 3.3 Disabling of the anti-lock systems Mazda BMW 320i Ford Granada 18 Discussion Other test results Effect of anti-lock braking 19 on accidents The test manoeuvres Straight line braking Braking on a curve Split surface braking The Test Results 4.1 Straight line braking The costs and benefits 19 of anti-lock 6 Conclusions Acknowledgements References 20 Appendix: Locked Wheel Stopping 21 Distances Ford Escort Fiat Uno Honda Prelude Mazda BMW 320i Ford Granada 11

3 Ownership of the Transport Research Laboratory was transferred from the Department of Transport to a subsidiary of the Transport Research Foundation on I st April This report has been reproduced by permission of the Controller of HMSO. Extracts from the text may be reproduced, except for commercial purposes, provided the source is acknowledged.

4 A STUDY OF VARIOUS CAR ANTI-LOCK BRAKING SYSTEMS ABSTRACT The braking performance of cars equipped with six different anti-lock braking systems has been assessed. The performance with the anti-lock systems functioning has been compared to that with the systems disabled. Straight line braking tests have been conducted on a range of single (that is uniform) surfaces. Similar surfaces have also been used for braking on a curve tests. Tests have also been performed on split high/low grip boundary surfaces. In all tests heavy, emergency type, braking was applied and the initial speed and stopping distance measured. A note was also taken of the rotational stability of the vehicle under such conditions of heavy braking. The vehicles were tested both unladen and fully laden, and over the initial speed range of km/h. All the anti-lock systems provided a high degree of vehicle stability and controllability under all test conditions. The deceleration characteristics of each system were found to be closely related to their degree of complexity and cost. A brief analysis of real life accident data indicates that all the anti-lock systems tested have a potential to reduce the number and severity of road accidents. 1. INTRODUCTION 1.1 BACKGROUND Anti-lock braking systems were first developed by the aircraft industry in the early 1950s. By the beginning of the 1970s several car manufacturers were developing their own systems. Recent technical advances, particularly with microprocessors, have led to anti-lock systems now being widely available across the whole car range. This report describes work carried out at the Transport and Road Research Laboratory to investigate the braking performance of a range of cars fitted with various modern anti-lock systems. In all tests the brakes were applied as hard and as fast as possible, thus simulating an emergency stop. This allowed the possible real life benefits of anti-lock, such as improved stability and shorter stopping distances, to be properly evaluated. 1.2 THEORY An anti-lock braking system is one that modulates brake pressure to prevent over-rapid deceleration of the sensed wheel or wheels. If a wheel is allowed to decelerate too rapidly under braking then it may lock. The potential benefits of anti-lock control in terms of reduced stopping distances and increased controllability can be demonstrated with the aid of Figure 1. Side force ~ Brake force Force coefficient Fp ~L I I Braking slip Fig.1 A typical relationship between brake force coefficient and side force coefficient as wheel slip increases from 0 Ifree rolling) to 1 (locked wheel)

5 Figure 1 shows how the braking force on a wheel varies with wheel slip for a braked wheel on a typical road surface. The brake force coefficient when the wheel is locked (i.e. wheel slip = 1), I~,, is about 70 per cent of the peak brake force coefficient, I~p. This peak brake force usually occurs somewhere in the slip range 0.1 to If an anti-lock system can modulate the brake pressure to a wheel such that the slip of that wheel is kept within a range that gives an average brake force higher than i~ L, then the retardation afforded by that wheel will be higher than the locked case and hence the stopping distance will be shorter. The controllability and stability of a vehicle are determined by the amount of available side forces on each tyre. Figure 1 shows that in the locked wheel condition the side force coefficient is reduced to zero. Relating this to real world situations, if the front wheels of a vehicle lock, while the rears continue to rotate, then no sideforces will be available to steer the car. Those available at the rear will keep the car travelling in a straight line regardless of steering input by the driver. This is an uncontrollable, but stable, situation. If the rear wheels lock, while the fronts continue to rotate, then there will be no side forces available at the rear to balance the side forces generated at the front and thus prevent the car spinning violently about its vertical axis. This is an uncontrollable unstable situation. If all four wheels lock then no side forces are generated and the vehicle will drift, again in an uncontrollable, though mainly stable, way. Maximum side forces occur at zero braking slip. As Holmes and Stone (1969) noted, this means controllability and stability can only be maximised with zero brake force. The anti-lock system designer therefore has to weigh up the relative importance of improved stopping distances as against better controllability. He or she may decide to minimise stopping distances by designing the system to operate at slips giving the highest brake forces, but accept a lower level of stability and controllability, although still better than the locked wheel case with no anti-lock. Alternatively the designer may decide that controllability is more important and hence the system will operate at lower slip levels. The stopping distances may be no better, or even slightly worse, in this case than those achievable with locked wheels. 1.3 SYSTEMS The designer has to decide what levels of slip to keep the sensed wheels at, what wheels on the vehicle are to be sensed, and how the brake pressures to all the wheels are to be controlled. The following are some common anti-lock configurations One channel - Rear axle control This simple system, described by Madison and Riordan (1969) was first used in the late 1960's. The speeds of the two rear wheels are sensed. The brake pressure to both of them, through one brake line, is controlled by whichever wheel begins to lock first. This logic, known as 'Select Low', ensures that neither rear wheel can lock. Straight line stability should be better with this system than with locked wheels, but as the front wheels are allowed to lock there are no steerability benefits. Although the original one channel system has been superseded, Yoneda et al (1983) described a version of it used by one Japanese manufacturer until quite recently. FRONT Wheelspeed sensor Control channel REAR Select low Fig.2 One channel system - Rear axle controlled Two channel - Front wheel control This is a relatively low cost system currently fitted to some front wheel drive cars with diagonally split braking systems. The front wheels are sensed and controlled independently, while regulating brake pressure to their diagonally opposite rear wheel via an apportioning valve. This system ensures that under conditions of panic braking on a split-mu surface it may be possible to lock one of the rear wheels, but never both. FRONT Individual control 4' Wheel speed sensor ==~== Control channel Apportioning valve REAR Fig.3 Two channel system - Front wheels controlled Two channel - Front and rear axle control This slightly more complex system was used until recently and is described by Satoh and Shiraishi (1983). All four wheels are sensed and each axle is controlled. The rear axle is controlled by Select Low, the front axle by 'Select High'. This allows one front wheel to lock, but

6 starts to modulate the brake pressure to the front wheels when the second wheel is about to lock. By preventing one of the front wheels from locking the system should ensure that a degree of steerability is maintained. 2. THE TEST VEHICLES Table 1 describes each of the six cars tested, including the make and type of anti-lock system (see Figures 3,4 and 5), tyres fitted and vehicle weights. ~1' Wheel speed sensor Control channel The laden weights were obtained by ballasting the vehicles with steel weights or gravel filled containers. The resulting vehicle weights were as close to the manufacturer's gross vehicle weight as possible without overloading the axles. m FRONT Select high Fig.4 Two channel system - Front and rear axles controlled REAR Select low Three channel - Front wheel and rear axle control This is the most sophisticated of the common anti-lock systems. First used on a production car in 1971, as described by Douglas and Schafer (1971), it is now found on many upper range passenger cars. Usually all four wheels are sensed, the front two are controlled independently, whilst those on the rear axle are controlled by Select Low. On some equivalent systems the rear wheel speeds are sensed within the differential, with the select low control logic still being used. The Ford Escort has the mechanical two channel system described fully by Newton and Riddy (1984). This senses and controls the front wheels independently, but has no wheel lock detection capability on the rear axle. Instead the rear wheels are linked to their diagonally opposite front wheels via load sensing (apportioning) valves. These valves govern the brake fluid pressure (and thus the braking force) at the rear wheels to ensure that under heavy braking both rear wheels cannot lock at the same time. The Fiat Uno has an electronic variant of the two channel system which operates on the Escort (see Figure 3). The Honda also has a two channel electronic anti-lock system, but it is a front/rear axle split system as described earlier (see Figure 4). The front wheels are select high controlled, allowing one, but not both, to lock. The rear wheels are controlled by select low, which allows neither to lock. The Mazda, BMW and Ford Granada all have versions of the three channel electronic system described earlier in the paper (see Figure 5). All have independent control of the front wheels and select low control of the rear wheels. The Mazda system has been described by Shinomiya et al (1988), that of the BMW by Leiber and Czinczel (1979) and the Granada system by Bleckmann et al (1983). Wheel speed sensor ==~== Control channel 3. THE TEST PROCEDURES 3.1 DESCRIPTION OF SURFACES Tests were carried out on a variety of surfaces with the vehicles both laden and unladen. The test surfaces are described fully in Table 2. FRONT Individual control REAR Select low When required the test surfaces were wetted by a hose and water cannon or sprinkler system. This ensured that there was a uniform layer of water on the surface. Fig.5 Three channel system - Front wheels and rear axle controlled 3.2 MEASURING EQUIPMENT USED A fifth wheel was mounted on the rear of the vehicle and connected to an electronic unit positioned inside the car. This unit had two digital displays, one for speed and one for distance travelled and was also connected to the brake pedal. When the brake pedal was pressed, this froze the speed display and activated the distance read-

7 TABLE 1 The Test Vehicles Vehicle Anti-lock system Tyre type Weights (kg) make and type Front Rear Front Rear (Total) (Total) Ford Escort 1.4L Fiat Uno Turbo ie Honda Prelude EX-M Mazda 626 GT BMW 320i Ford Granada Scorpio Lucas Girling SCS, Fig SR (1075) (1325) AP Lockheed, Fig 3 175/60 R (975) (1310) Honda ALB, Fig 4 185/70 HR (1180) (1490) Girling-Sumitomo, Fig 5 195/60 R (1375) (1850) Bosch ABS, Fig 5 195/60 HR (1225) (1500) Teves Mk II, Fig 5 195/65 R (1475) (1830) TABLE 2 The Test Surfaces Name Texture Tested Wet/Dry Skid Resistance Bridport rounded gravel Macadam carpet Smooth Mastic asphalt Fine Textured Asphalt (FTA) Motorway surface Rough, polished Wet only Low Smooth, polished Wet only Low Smooth, harsh Wet and dry High Rough, fairly harsh Wet and dry High out. When the vehicle had come to a halt the displays thus showed the speed at which the brakes were applied and the overall braking distance. From these figures the average deceleration of the vehicle could be calculated using the formula: a = v 2/2d where 'a' is the average deceleration in ms -2, 'v' is the initial velocity in ms -1 and 'd' is the braking distance in metres. Throughout this paper the units of deceleration are given as fractions of the gravitational acceleration, g (9.8 ms-2). 3.3 DISABLING OF THE ANTI-LOCK SYSTEMS Most of the testing was done to compare the performance of a vehicle with its anti-lock system functioning to that with it disabled. The electronic systems could be disabled simply by altering one part of their electrical systems, e.g. by removing a fuse or disconnecting a wire. In all cases this resulted in complete loss of anti-lock capability but retention of the full conventional braking system. The Ford Escort has a mechanical system. Each front wheel is sensed and modulated completely independently of the other. This means it is impossible to fully

8 disable the anti-lock system unless mechanical alterations are made to both modulators. The chances of faults developing in both modulators at the same time are likely to be remote and tests were therefore carried out on this vehicle to simulate "half" failure of the system, that is with just one modulator disabled. Tests were also conducted with the complete system disabled. With the anti-lock modulator disabled, the relevant half of the braking system corresponded to that of a conventionally braked vehicle. The electronic systems may also fail partially, e.g. failure of a solenoid operated valve, or a wheel speed sensor, but these failure modes were not the target of this investigation. When a vehicles' anti-lock system was disabled, the resulting braking system usually represented that of a standard, non anti-lock, version of that vehicle. This was not always the case, however, because the anti-lock system is not, necessarily, simply an "add on" device to a conventional braking system. Other parts of the system can be changed to suit the braking characteristics of the anti-lock equipped vehicle. The Ford Granada and Ford Escort fall into this category. All anti-lock systems are designed to be fail-safe, that is they revert to full conventional braking if a failure occurs. It is therefore reasonable to assume that the braking, stability and controllability performance of the anti-lock disabled vehicles all lie within the usual range of performances from standard, conventionally braked vehicles. 3.4 THE TEST MANOEUVRES The tests conducted on each vehicle are detailed in Table 3. Not all the tests could be conducted on the Honda due to inclement weather and other factors Straight line braking This test involved driving the vehicle with all four wheels on the same test surface. The test speeds varied from 20 km/h up to 100 km/h and were increased in 10 km/h increments. At the required test speed the footbrake was applied as quickly and as hard as possible to simulate an emergency stop. At the end of each run when the vehicle had come to a halt a note was made of the initial velocity and braking distance. From this the average deceleration was calculated. This test was performed with the anti-lock system fully, and in some tests with the Ford Escort partially, operational and then with the anti-lock disconnected. The vehicles were tested both unladen and laden. The results are shown in the tables in section to Braking on a curve This test involved the use of a curve, of approximately 80 degrees of arc, which was marked using cones on the appropriate test surface. The inner radius of the curve was 60m and the outer radius was 63m. The purpose of this test was to evaluate the effectiveness of the anti-lock systems under combined conditions of braking and cornering. Initially the vehicle was driven, unbraked, through the curve. The speed was increased each run in small increments until control was lost, thus determining the Unbraked Breakaway Speed (UBS). This gives a guide to the upper limits of adhesion, and hence controllability, that the tyres can provide. Any reduction in side-forces will inevitably lead to a loss of control when the vehicle is at this limit. If an anti-lock system can provide high sideforces under heavy braking, then the maximum speed from which the vehicle can be brought safely to a rapid halt, whilst remaining within the marker cones, will be near to the UBS. TABLE 3 The Test Programme Surface Test No A/L Full A/L Half A/L No A/L Full A/L Half A/L Wet Bridport Straight EFHMBG EFHMBG E Curve EFHMBG E Wet Mastic Straight EFHMBG EFHMBG E Curve EFHMBG E Wet FTA Straight EFHMBG EFHMBG E Curve EFHMBG E Dry FTA Straight EFHMBG EFHMBG E Curve EFHMBG E Wet Motorway Straight EF MBG EF MBG E Dry Motorway Straight EF MBG EF MBG E Bridport/FTA Split EFHMBG EFHMBG E Mastic/FTA Split EF MBG EFHMBG E E - Escort, F - Fiat, H - Honda, M - Mazda, B - BMW, G - Granada EFHMBG EFHMBG EFHMBG EF MBG EF MBG EF MBG EF MBG EF MBG EFHMBG EFHMBG EFHMBG EFHMBG EFHMBG EFHMBG EF MBG EFHMBG EF MBG EF MBG EFHMBG EFHMBG 5

9 The vehicle was driven into the curve at a constant speed and the footbrake was applied as hard and as fast as possible. Calculations of average deceleration were then made from the initial speed and braking distance displays. The test speeds were gradually increased each run until a loss of control was experienced. During all the runs a vehicle was deemed to have failed the test if it could not be steered to stay within the lane of marker cones. These tests were only conducted with the anti-lock systems in use. No locked wheel tests were conducted because if both the front or rear wheels locked there would be a definite loss of control and test failure. Because the Ford Escort could be run with just half of its anti-lock system functioning, it was tested with only the inner front wheel sensing, and then the outer wheel sensing, as well as with full anti-lock. Section 4.2 gives the results of the tests in the form of maximum successful braked speed as a percentage of the Unbraked Breakaway Speed. A result of 100 per cent indicates that the vehicle could be driven at the maximum possible speed into the curve, and the anti-lock system would still provide sufficient side forces to enable full use of the brakes without serious loss of control Split surface braking This test involved braking with the near-side wheels on a different surface to the off-side wheels. The layout of the track enabled two combinations of surfaces to be tested upon, namely Wet Bridport/Wet FTA and Wet Mastic/Wet FTA, giving in each case a low/high friction combination. The purpose of this test was to determine the controllability of an anti-lock equipped vehicle when it is braked heavily on a split high-low grip surface. This type of condition can be found, for example, when snow, ice or water lies to one side of an otherwise dry road. The test involved driving the vehicle at various constant speeds increasing from 20 km/h in 10 km/h increments onto the test area. The foot brake was then applied as hard and as quickly as possible. Once the vehicle had come to a halt the initial speed and braking distance were noted, and the average deceleration was calculated. With the anti-lock system disconnected a vehicle is likely to spin violently during this manoeuvre if all four wheels lock. This is due to the wheels on the high grip side of the vehicle slowing down faster than those on the low grip surface. The Ford Escort was tested with full anti-lock, as well as its two "half" anti-lock modes. The system was tested with anti-lock control of the front wheel on the high grip surface only, and then on the low grip surface only. Section 4.3 gives the maximum speed at which the manoeuvre could be performed for the various anti-lock conditions. For the test to be completed successfully the vehicle had to be brought to a halt, with some steering correction if necessary, in a straight line. The definition of a "straight line" in this instance is when either side of the vehicle remains on the surface that it was on when braking commenced. An anti-lock system which enhances controllability of a vehicle will give higher maximum speeds in this test. Speeds over 100 km/h were not investigated, so for an anti-lock system with good controllability we would expect speeds of 100 km/h to be achieved. Lower maximum speeds than this may indicate that a system has been designed to minimise stopping distances at the expense of controllability. 4. THE TEST RESULTS 4.1 STRAIGHT LINE BRAKING The following sections (4.1.1 to 4.1.6) present a summary of the results from the straight line braking tests for each vehicle tested. Average decelerations are given for the lower speed range (20-60 km/h), the upper range ( km/h) and over all tested speeds. The table in section includes the decelerations obtained from the Escort with half its anti-lock system operational, as explained in the Test Vehicle section. All the remaining tables refer to results either with full anti-lock or no antilock. The difference between the average decelerations obtained with anti-lock operational and that with it disabled is shown in percentage form. All percentages are relative to the corresponding non anti-lock case. A positive percentage indicates a higher deceleration with anti-lock than without it. A negative value demonstrates that the anti-lock system produced a lower deceleration, and hence longer stopping distance, than was achievable by allowing the wheels to lock. Figure 6 converts deceleration to stopping distance over a range of initial velocities Ford Escort A summary of the results from this vehicle is given in Table 4. From the Table two general trends emerge. First, with no anti-lock, the average decelerations obtained from the laden vehicle are broadly similar to those with it unladen. With full anti-lock the laden values are consistently slightly lower than the unladen. Second, and more importantly, the average decelerations tend to be higher with no anti-lock than with partial or complete anti-lock. The fact that, when the anti-lock system was disabled, the unladen and laden results are similar, indicates that all four wheels of the vehicle locked under both loading conditions. In this situation the deceleration of the vehicle is virtually independent of mass.

10 1.0 I= I! /! ~ i ~ ~,! km/h O km/h 100 km/h 60 km/h km/h 20 km/h ~-!... t" 0 L Stopping distance (m) Fig.6 Chart to convert average deceleration to stopping distance at various initial velocities The second observed trend, that on many surfaces decelerations are not improved by the anti-lock system, implies that the system does not generally maintain the brake effort at each wheel at a sufficiently high level to ensure better adhesion than can be obtained by allowing the wheels to lock. On all but two of the surfaces at low speeds (60 km/h and below) decelerations without antilock are higher than the half anti-lock values. The only exceptions are the Wet Mastic and Dry Motorway surfaces. The half anti-lock values are themselves higher than the full anti-lock values on all surfaces and under all conditions except Wet Mastic, unladen. Overall, the full anti-lock decelerations are on average about 13 per cent lower than those with locked wheels. At speeds of 60 km/h and above the average deceleration without anti-lock is still slightly better than the two anti-lock cases on the Wet FTA, Dry FTA, Wet Bridport and Wet Motorway (laden only) surfaces. The full antilock decelerations are again lower than the half anti-lock values. The two anti-lock operational cases give better decelerations than with locked wheels when the vehicle is braked from these higher speeds on the Dry Motorway and Wet Mastic surfaces. This is also the case, though to a lesser extent, on the Wet Motorway, unladen. Averaging the results over the whole tested speed range, it can be seen that there is an improvement in decelerations obtained with the anti-lock system partly or wholly operational only on the Wet Mastic and Dry Motorway. The half anti-lock decelerations are between 2 and 8 per cent higher than the locked wheel values, while the improvements with full anti-lock lie in the range 1 to 12 per cent. On the other surfaces the half anti-lock decelerations are lower than those with no anti-lock by between 2 and 6 per cent. The full anti-lock decelerations are lower by between 10 and 19 per cent. When operating, the two channel mechanical anti-lock system fitted to the Ford Escort provides better decelerations on only three of the surfaces tested and then only under certain conditions of loading and initial speed. The best performance was undoubtedly on the Wet Mastic with the vehicle unladen. Under these conditions the decelerations achieved with the anti-lock system fully operational were higher than those obtained by allowing the wheels to lock at all speeds over about 30 km/h. With the vehicle laden on this surface there was an improvement in the performance at speeds over about 50 km/h. An improvement was found at speeds in excess of 40 km/h when unladen and 50 km/h when laden on the Dry Motorway. On the Wet Motorway, with the vehicle unladen, a very slight improvement was found at speeds over 60 km/h. At all other speeds, on all other surfaces and under all other loading conditions tested, the antilock system did not increase the decelerations. The reason for the enhanced performance on the Mastic may be that the ratio of peak to locked wheel tyre/road friction coefficient is quite high, allowing the anti-lock system to operate within a larger range of brake pressures that produce brake force levels in excess of the locked wheel value. This locked wheel value can be further diminished by the fact that the surface is smooth and so it is easier for a layer of water to get between the tyre and the road. If the wheel is rotating the tyre tread will be able to disperse most of this water, but this process will not be as effective if the wheel is locked. 7

11 TABLE 4 The Test Results - Escort - Straight Line Braking Low Speed High Speed (<60 km/h) (_>60 km/h) AIISpeeds ( km/h) Testsufface Loading Anti- Avrge Per Avrge Per Lock decel cent decel cent (g) change (g) change Avrge Per decel cent (g) change Wet Bridport None Half.35-7,31-3 Full.31-18, None Half Full Wet Mastic None Half Full None Half Full Wet FTA None Half Full None Half Full Wet Motorway None Half Full None Half Full Dry FTA None Half Full None Half Full Dry Motorway None Half Full None Half Full Per cent changes calculated before rounding Low Speed and High Speed figures include result at 60 km/h.

12 As is explained in the Introduction any anti-lock system has to achieve a compromise between good deceleration and good controllability. This test shows that the Escort system often does not manage to reduce stopping distances, particularly at lower speeds. At higher speeds (approaching 100 km/h) there is generally very little difference between the stopping distances of the anti-lock and non anti-lock vehicle. It should be stressed, however, that the stopping distances achieved with the anti-lock system are perfectly acceptable for a modern passenger car. The split surface and curve tests are needed to show the level of controllability that the system can provide Fiat Uno A summary of the results from this vehicle is given in Table 5. This vehicle was tested with its anti-lock system fully operational or with no anti-lock at all. It is apparent from the Table that, with the anti-lock system disconnected, the decelerations achieved with the vehicle laden were broadly similar to those with it unladen on all the wet surfaces. This is similar to the results experienced with the Escort and could again indicate that all four wheels TABLE 5 The Test Results - Fiat Uno - Straight Line Braking Low Speed High Speed All Speeds (<60 km/h) (_>60 km/h) ( km/h) Testsufface Loading Anti- Avrge Per Avrge Per Lock decel cent decel cent (g) change (g) change Avrge Per decel cent (g) change Wet Bridport None Full None Full Wet Mastic None Full None Full Wet FTA None Full None Full Wet Motorway None Full None Full Dry FTA None Full None Full Dry Motorway None Full None Full Per cent changes calculated before rounding Low Speed and High Speed figures include results at 60 km/h 9

13 locked under both loading conditions. With the system disconnected there is a slight loss in performance when the vehicle is laden on the two dry surfaces. It is possible that the high grip of these surfaces prevented the rear wheels locking in this condition and the brake force produced was not as high as that with locked wheels. It is clear from the Table that at low speeds (60 km/h and below) the anti-lock system reduces the decelerations achievable on all of the surfaces tested. The average decelerations vary from 3 to 30 per cent lower than the locked wheel values over the range of surfaces. The average reduction is about 14 per cent. The performance of the system is better at higher speeds. The unladen vehicle gave higher average decelerations with the anti-lock operational at speeds of 60 km/h and above on all but one of the test surfaces (Dry FTA). The improvements were about 10 per cent on average. The laden vehicle, however, produced improvements on just the Wet and Dry Motorway surfaces (16 and 3 per cent respectively). On the other surfaces the laden anti-lock decelerations were usually about 5 per cent lower than with locked wheels. Averaging over the whole of the tested speed range, there was a very slight increase in decelerations achieved by the anti-lock system with the unladen vehicle on the Wet Mastic and Dry Motorway. When unladen on the remaining surfaces and when laden on all the TABLE 6 surfaces, there was a decrease in decelerations of between 2 and 20 per cent, 8 per cent being the average value. The two channel electronic anti-lock system fitted to the Fiat Uno generally gives lower decelerations, and hence longer stopping distances, than can be achieved by disconnecting the system and allowing the wheels to lock. This is true on all the surfaces tested, both unladen and laden, at speeds below about 40 km/h. Above this speed the with anti-lock decelerations are the same or better than the non anti-lock results on the Dry FTA, Dry Motorway and Wet Mastic (all unladen). Above 60 km/h the same is true on the Bridport (unladen), Wet Motorway (unladen and laden), Dry FTA and Dry Motorway (both laden). The anti-lock system does not equal or better the non anti-lock decelerations until speeds of 80 km/h are reached under the remaining conditions-of Wet FTA, Bridport and Mastic (all laden). As was the case with the Ford Escort, which had a mechanical variant of this front wheel controlling, two channel anti-lock system, the decelerations achieved with the system operational were often lower than those with it disconnected. They were, however, acceptable for a modern passenger car Honda Prelude The Test Results - Honda - Straight Line Braking A summary of the results from this vehicle is given in Table 6. Low Speed High Speed All Speeds (<60 km/h) (>60 km/h) ( km/h) Test Surface Loading Anti- Avrge Per Avrge Per Lock decel cent decel cent (g) change (g) change (g) Avrge decel change Per cent (g) Wet Bridport None Full None Full Wet Mastic None Full None Full Wet FTA None Full None Full Dry FTA None Full Per cent changes calculated before rounding Low Speed and High Speed figures include results at 60 km/h

14 The limited availability of the vehicle and bad weather prevented the complete set of tests from being conducted. Nevertheless, sufficient testing was carried out to allow the observation of some important trends. There is almost no difference between the decelerations achieved with the vehicle unladen to the corresponding values with it laden on the Bridport and Mastic surfaces. The Wet FTA, no anti-lock condition produces laden decelerations about 10 per cent lower than those with the vehicle unladen. This could be due to a degree of underbrakingto the rear axle which meant the rear wheels could not be made to lock when the vehicle was laden on this surface. The two channel, front and rear axle controlling anti-lock system gives higher decelerations than those with it disabled on all the surfaces and over almost the whole tested speed range. The improvements are greatest at higher speeds, up to 60 per cent on the Wet Mastic. The smallest improvements in deceleration were found on the Wet FTA surface, just 1 per cent at low speed, unladen, rising to 12 per cent at high speed, laden. Over the whole speed range the improvements ranged from 4 to 39 per cent, averaging about 17 per cent. It is therefore clear that this electronic anti-lock system is able to reduce stopping distances on a wide variety of surfaces, regardless of initial speed or loading condition. The stability and controllability aspects of the system, which allows one front wheel to lock, need to be investigated using the braking on a curve and split surface tests Mazda 626 A summary of the results from this vehicle is given in Table 7. The laden decelerations are generally lower than those obtained from the unladen vehicle. It was found that, even with the anti-lock system disabled, the rear wheels could not be made to lock. This indicates a degree of under-braking to the rear axle which is probably sufficient to produce the deterioration in braking experienced when the vehicle was loaded to the maximum levels specified by the manufacturer (Table 1). On all but one of the surfaces at low speed and on all surfaces at high speed, the anti-lock equipped vehicle produced higher decelerations than those achieved in the equivalent non anti-lock test. The Wet FTA surface was the only exception, producing a small reduction in deceleration at low speed when the vehicle was unladen. The improvements are more marked at high speeds than at lower speeds BMW320i A summary of the results from this vehicle is given in Table 8. There was found to be little consistent variation between the laden and unladen results. The decelerations achieved with the anti-lock system functioning, however, were found to be consistently higher than those without anti-lock. Improvements ranged from just 2 per cent at low speed on the Wet FTA to an impressive 125 per cent at high speed on the Wet Mastic. Averaging over the whole speed range, the improvements varied between 10 and 69 per cent, with an average of about 25 per cent. These figures suggest the three channel electronic antileck system fitted to the BMW has excellent deceleration capabilities. Only on the Wet FTA is there a reduction in decelerations achievable with anti-lock at speeds up to about 50 km/h. On all the remaining surfaces, and under both loading conditions, the anti-lock system gives higher decelerations than with locked wheels at all speeds above about 20 km/h Ford Granada A summary of the results from this vehicle is given in Table 9. There is little variation between the two loading cases on most surfaces. The laden decelerations obtained on the Bridport are rather lower than those with the unladen vehicle, particularly when the anti-lock system is operational. At low speeds there is very little difference between the decelerations achieved with locked wheels and those with the anti-lock system operating on the Wet Motorway, Wet FTA and Dry FTA surfaces. On the remaining surfaces large improvements over the locked wheel performance were experienced. At higher speeds the anti-lock system produces better decelerations than locked wheels on all the surfaces tested. Averaging over the whole tested speed range, the anti-lock decelerations vary from 6 to 49 per cent higher than those obtained by allowing the wheels to lock. The average improvement is about 22 per cent. This three channel, electronic anti-lock system is generally able to improve on the locked wheel stopping distances on all the test surfaces, regardless of loading. Only at speeds below about 40 km/h is there a performance reduction on some surfaces. Averaging over the whole tested speed range, the antilock system produced decelerations between 7 and 34 per cent higher than the equivalent non anti-lock cases. The average improvement was 16 per cent. It is clear that this electronic system is able to reduce stopping distances under most conditions. Only at very low speeds is there sometimes little or no improvement. 11

15 TABLE 7 The Test Results - Mazda - Straight Line Braking Low Speed High Speed All Speeds (<60 km/h) (_>60 km/h) ( km/h) Test Surface Loading Anti Avrge Per Avrge Per Lock decel cent decel cent (g) change (g) change Avrge Per decel cent (g) change Wet Bridpo~ None Full None Full Wet Mastic None Full None Full Wet FTA None Full None Full Wet Motorway None Full None Full Dry FTA None Full None Full Dry Motorway None Full None Full Per cent changes calculated before rounding Low Speed and High Speed figures include results at 60 km/h

16 TABLE 8 The Test Results - BMW - Straight Line Braking Low Speed High Speed All Speeds (<60 km/h) (>60 km/h) ( km/h) Test Surface Loading Anti Lock Average Per Average Per Average Per decel cent decel cent decel cent (g) change (g) change (g) change Wet Bridport None Full None Full Wet Mastic None Full None Full Wet FTA None Full None Full Wet Motorway None Full None Full Dry FTA None Full None Full Dry Motorway None Full None Full Per cent changes calculated before rounding. Low speed and high speed figures include results at 60km/h. 13

17 Test Surface Loading TABLE 9 The Test Results - Granada - Straight Line Braking Low Speed High Speed All Speeds (<60 km/h) (>60 km/h) ( km/h) Anti Average Per Average Per Average Per Lock decel cent decel cent decel cent (g) change (g) change (g) change Wet Bridport None Full None Full Wet Mastic None Full None Full Wet FTA None Full None Full Wet Motorway None Full None Full Dry FTA None Full None Full Dry Motorway None Full None Full Per cent changes calculated before rounding. Low speed and high speed figures include results at 60km/h. 14

18 TABLE 10 The Test Results - Braking on a Curve Maximum possible speed (%UBS) Vehicle Loading Anti-Lock Wet Bridport Wet Wet Dry Mastic FTA FTA Escort inner outer full inner outer full Fiat full ~full Honda full full Mazda full full BMW full full Granada full 100 full BRAKING ON A CURVE Table 10 shows the results from the braking on a curve tests. The figures represent a percentage of the unbraked breakaway speed (UBS) that each vehicle could attain under full braking whilst remaining controllable. In the case of the Escort three figures are shown. One represents anti-lock on just the inside front wheel on the curve, one represents anti-lock on just the outside front wheel on the curve, and the final one shows anti-lock on both wheels Ford Escort From Table 10 it can be seen that there is very little difference between the results obtained with the vehicle laden and unladen. There is also very little difference between the results obtained with the two half anti-lock conditions. This implies that either condition gives substantially the same degree of controllability. Using full anti-lock gave a significantly better performance than either of the two half anti-lock cases. Braking from speeds of between 93 and 100 per cent of the UBS was achieved with full anti-lock. The two half anti-lock cases gave speeds of between 75 and 90 per cent of the UBS. Therefore it can be summarised that using the full anti-lock system on the Escort gives more controllability when panic braking on a curve than is possible with half the system disabled. However, even in the half anti-lock mode there is a degree of controllability that would be impossible if both wheels on any one axle were allowed to lock. In the straight braking tests the half anti-lock condition generally produced higher average decelerations than was possible using the full system. However, from the braking on a curve tests it is clear that the full system is preferable because the amount of side force generated by the tyres is of greater importance in this condition. The deceleration curves in Figures 7 and 8 show that on the slippery Bridport the decelerations achieved by the two half anti-lock modes are slightly higher than those of the full anti-lock case. On the Mastic the full anti-lock and front outer sensing modes generally produce similar decelerations and slightly higher than those obtained from the front inner sensing mode. On the Wet and Dry FTA surfaces, however, the front outer sensing mode achieves slightly higher decelerations than either full antilock or front inner sensing. The outer sensing mode allows the inside front wheel to lock. When braking in a curve it is the inside wheels which are less heavily laden and hence which are most likely to lock. It is therefore likely that the front inside wheel did lock during this manoeuvre and, as was the case during the straight line tests, the vehicle produced slightly higher average decelerations. 15

19 0.6 Deceleration (g) Full A/L..,,i,- Front inner sensed -0- Front outer sensed Deceleration (g) I I I I I I I I I I I I I I I I I I Velocity (km/h) t I f f I I 1 I I f Velocity (km/h) 0.4 Wet Bridport Wet Mastic Wet FTA Dry FTA Fig.7 Curve tests - Escort - Deceleration (g) Full A/L ~ Front inner sensed,-~ Front outer sensed Deceleration (g) iiiiiiiiiiiiiiiiiii iiiiiiiiiiiii Iiiii iiiiiiiiii i i I r i I [ I I = t L i I I L I I J J I = I [ I I I I I , , , Velocity (km/h) Velocity (km/h) Wet Bridport Wet Mastic Wet FTA Dry FTA 0.4 Fig.8 Curve tests - Escort Fiat Uno It can be seen from Table 10 that there is little or no difference between the results achieved with the vehicle laden and unladen. The largest difference was only 6 per cent of UBS and was obtained on the Wet Bridport surface. The maximum possible speeds whilst braking on the curve ranged from 94 to 100 per cent of the UBS. This indicates that the two channel electronic anti-lock system fitted to this vehicle provides a very high degree of controllability during this manoeuvre Honda Prelude The maximum achievable speeds whilst braking on a curve were at least 80 per cent of the UBS on all surfaces and under both loading conditions. This lowest level of performance was obtained on Wet Mastic, the other surfaces produced results ranging from 89 to 99 per cent. Shortage of test time with this vehicle meant that only one attempt could be made at each speed increment. Therefore the results may slightly underestimate the true performance of the braking system. The speed during braking was increased in 5 km/h increments. This would mean that if a test run was unsuccessful at 65 km/h, but successful at 60 km/h, then 60 km/h would be recorded as the maximum possible speed, whereas speeds up to 64 km/h may have been possible. The UBS was, however, found by increasing the speed in 1 or 2 km/h steps, as were the maximum achievable speeds of all the other vehicles. The "select high" operation of the anti-lock system meant that one wheel on the front axle was able to lock. The inside wheel did occasionally lock during this manoeuvre, 16

20 particularly on the wet Mastic surface. With this wheel locked the amount of side force enabling the vehicle to be steered will be reduced. This is a possible explanation for the slightly lower maximum speeds that were achievable on the Wet Mastic surface. Despite the possibility of one front wheel locking, the controllability of this vehicle with its two channel electronic anti-lock system remains very high under conditions of emergency braking on a curve Mazda 626, BMW 320i, Ford Granada The three channel anti-lock braking systems fitted to these vehicles were all found to give a high degree of controllability at all speeds up to (or very near) the UBS, and under both loading conditions. The maximum speeds achieved ranged from 89 to 100 per cent of the UBS. 4.3 SPLIT SURFACE BRAKING Table 11 shows the results from the split surface tests, in the form of maximum speeds at which each of the vehicles tested could successfully complete the manoeuvre. Speeds over 100km/h were not attempted Ford Escort The Ford Escort was tested in its two "half" anti-lock modes. Table 11 shows these results with the sensed front wheel on either the high grip surface (High p.t), or on the low grip surface (Low ~). Speeds of between 50 and 70 km/h were achievable when the anti-lock system was completely disabled. This was the case under both loading conditions and on both pairs of surfaces. With only one wheel detecting wheel TABLE 11 The Test Results - Split Surface Braking Vehicle Loading Anti-Lock Bridport/FTA Maximum speed (km/h) Mastic/FTA Escort None Low # High ~ Full None Low ~ High ~ Full Fiat None Full None Full Honda Mazda None 50 (Nottested) Full Full None Full None Full BMW Granada None Full None Full None Full None Full

21 lock on the FTA surface (High I~), the results were very similar to the no anti-lock condition. At speeds greater than those stated the vehicle rotated violently through an angle of more than 90 degrees. When the anti-lock was only acting on the front wheel on the low grip surface (Low ~t), the results were identical to the full anti-lock mode. The vehicle could be braked hard from speeds of 100 km/h and brought to a halt in a straight line and with very little steering correction. This applied to the laden and unladen conditions, and to both pairs of surfaces. It can be concluded that the fully operational anti-lock system on this vehicle will provide good controllability under split 11 braking conditions. If half the system is operational then the wheel being sensed must be on the low # surface to give the same amount of stability. If the sensed wheel is on the high surface then the controllability of the vehicle is no more than it is under the no anti-lock condition Fiat Uno With the two channel anti-lock system operational the vehicle could be braked heavily at 100 km/h under full control. There was very little steering correction necessary to maintain a straight course. However, with the anti-lock system disabled it was found that the maximum speeds that the vehicle could be braked from were 50 to 60 km/h. Beyond these speeds the vehicle would veer from the split # surface and run onto the FTA. There were also some instances of the vehicle rotating through as much as 90 degrees, particularly at higher speeds. This applied to both the laden and unladen conditions, and both pairs of test surfaces. The vehicle appeared to be fairly stable during most of these test runs with the anti-lock system disabled. The reason for this is that although the front wheels were locking up, the rear wheels continued to rotate. This will give the wheels sufficient side grip to prevent violent rotation, although the vehicle will still veer off course and become unsteerable. The conclusions are that under split tt braking conditions the anti-lock system fitted to this vehicle gives very good controllability Honda Prelude It can be seen from Table 11 that on both combinations of test surface and under both loading conditions the antilock equipped vehicle remained fully controllable when heavily braked from a speed of 100 km/h. With the anti-lock system disabled the unladen vehicle would rotate violently through almost 90 degrees at 50 km/h on the Bridport/FTA surface. As a result of this and shortage of test time, no further testing was conducted without the anti-lock system operational on the split grip surfaces. When use was made of the anti-lock system the front wheel on the low p. surface could lock. This would mean that the available side-forces are reduced and that the controllability may be adversely affected. During the tests it was found that although a large amount of steering correction was needed, control of the vehicle could be maintained. The degree of correction was found to be greater in the unladen condition. This is probably due to less force acting down on the rear wheels or less braking force being needed to cause the rear wheels to slip or momentarily lock. This will generate a reduction in the available side forces and therefore a reduction in stability Mazda 626 From Table 11 it can be seen that the results for this vehicle were identical for both the laden and unladen conditions. Without the anti-lock system the vehicle could not be braked to a halt in a straight line at speeds over 70 km/h on the Bridport/FTA, and 50 km/h on the Mastic/FTA. However, the vehicle exhibited similar stability characteristics to the Fiat when the anti-lock system was disabled, that is the rear wheels would not lock and so the car simply veered off course. When the anti-lock system was used the vehicle could be braked hard from 100 km/h, remaining under the full control of the driver. Similar deceleration results were achieved from all of the test conditions. The decelerations obtained with and without anti-lock are similar at speeds below about 40 km/h. At higher speeds the anti-lock system produces better deceleration values. Overall this anti-lock system gives very good controllability. At speeds over about 40 km/h it will also improve the stopping performance of the vehicle BMW 320i It can be seen from Table 11 that the vehicle was capable of maximum braking from 100 km/h on the split I.t surfaces. Minimal steering correction was needed to achieve this. Without the anti-lock system the vehicle could not be braked to a halt without rotating violently at speeds above 40 or 50 km/h Ford Granada When the anti-lock system was used the vehicle was able to stop in a straight line from 100 km/h using full braking. Without the anti-lock system the maximum speed at which full braking could be accomplished without the vehicle rotating violently was between 30 and 50 km/h. This applied to both loading conditions and to both pairs of test surfaces (Table 11). 18

22 The anti-lock system fitted to this vehicle produced very good controllability coupled with very good deceleration performance in the split # test. 5. DISCUSSION The results outlined in the previous sections indicate that all the anti-lock systems are able to offer substantial improvements in the stability and controllability of the vehicles to which they were fitted. The more complicated and expensive electronic three channel systems were also able to bring the vehicles to a more rapid halt, that is give higher average decelerations, than would normally be possible without anti-lock. The costs and benefits of the various systems need to be considered carefully. For example although the mechanical, two channel, SCS system fitted to the Escort was generally unable to better the decelerations achieved with locked wheels, it is likely that any failure of the anti-lock system could only affect half of the braking system. The other half remains fully functional, offering good controllability on most surface conditions and decelerations that are often better than with full anti-lock control of both braking circuits. The simpler systems may have better reliability and fail-safe characteristics than their more complicated and more expensive rivals. It has not been the intention of this programme of work to investigate this aspect of anti-lock braking systems. 5.1 OTHER TEST RESULTS Wallrich and Schindler, of TUV Rheinland, have described work to assess the ability of average drivers to use the potential benefits of anti-lock in simulated emergency situations. The system tested was the mechanical SCS system fitted to a Ford Escort. The results suggested that drivers of the anti-lock equipped vehicles were likely to be involved in about 20 per cent fewer accident situations, on a uniform surface, when having to brake and steer around an obstacle. This figure rose to 30 per cent when simply stopping in a straight line, and 60 per cent or more when having to brake while driving round a bend or on a split grip surface combination. The TRRL tests did not find that any significant reduction in stopping distances was generally possible with this particular anti-lock system, and hence the 30 per cent figure may not seem to be supported. It should be noted, however, that the TRRL results were obtained with one vehicle, either with its anti-lock system operational or with it disabled. The T0V tests used two separate vehicles, one anti-lock equipped, the other not. The base braking systems on the two versions are different. This may account for at least some of the supposed discrepancy. Differences in driver reactions to a vehicle that has locked wheels to one that has not may also cause some lengthening., of the without anti-lock stopping distances found by TUV. The driver of a vehicle that has locked its wheels may be tempted to release the brakes slightly in order to allow control of the vehicle to be retained. 5.2 EFFECT OF ANTI-LOCK BRAKING ON ACCIDENTS Given a knowledge of what the various anti-lock systems can achieve on a test track, under test conditions, it would be desirable to translate this information into the possible benefits of anti-lock in real world accident situations. Unfortunately very little work has been done by the international vehicle safety community in this area. Treat (1980) gives the results of an in-depth study of 215 accidents in Indiana, USA, occurring during the mid 1970s. Various improvements to vehicle braking systems were considered in this study. It was found that four wheel anti-lock brakes would have certainly prevented 3 per cent of accidents, probably prevented or reduced the severity of a further 5 per cent and possibly a further 5 per cent. In total 13 per cent of car accidents might have been prevented or reduced in severity by anti-lock. Severity reduction was defined as a reduction in impact speed of 10 mph (16 km/h) or 25 per cent, whichever was the greater. The anti-lock system was assumed to not affect stopping distances on dry roads and reduce them by 10 per cent on wet roads. The TRRL research described in this paper into some more modern anti-lock systems shows that many are capable of stopping distance improvements of 10 or 20 per cent on both wet and dry surfaces. Even the cheaper and simpler systems can give slight improvements at higher speeds (above about 60 km/h). These results, combined with the TUV results discussed in the previous section, mean that it may be reasonable to hope that rather more than 13 per cent of accidents might now be affected by anti-lock. 5.3 THE COSTS AND BENEFITS OF ANTI-LOCK A cost benefit analysis may be undertaken by assuming that a percentage of all accidents involving cars may be prevented. If, for example, a 15 per cent reduction is used, just over 33,400 accidents would be prevented each year (1989 figures, Department of Transport (1990)). Assuming further a car population of 19 million, a car life of ten years, an average accident cost of 19,320 (Department of Transport (1990)) and a discount factor of six per cent, yields a Net Present Value of benefits over the ten years of about 250 per car. Therefore, provided no more than this amount of money is spent in equipping each car with anti-lock braking, the benefits in accident savings will exceed the costs. The SCS system used on the Escort cost around this amount; some proposed simple electronic two channel systems might cost less; the more complex three channel electronic versions at present cost considerably more. The more complex three channel systems can on some surfaces increase average decelerations by 20 or 30 per cent or more and should give greater benefits than their cheaper counterparts. Larger scales of production and advances in electronics could well reduce the cost of 19

23 manufacturing these more complicated systems in the future, and hence improve their viability. The method of comparing benefits against costs described here relies critically on a satisfactory estimate of accident savings and, as presented so far, the assumed accident reduction of 15 per cent can only be used in the calculation as a guide as to whether or not car anti-lock braking is worth having. As more and more cars are fitted with anti-lock, either as standard or as an optional extra, the national STATS 19 data will become more useful as an indicator of benefits since make and model of car can now be obtained. However, more detailed accident data also needs to be studied to give a better idea of how the accident avoidance potential of anti-lock can be used. Although the evidence so far suggests that anti-lock could have worthwhile benefits there are still arguments raised against its use. For example there are worries about reliability and fail-safe characteristics, which need to be explored. Some people believe there will be an increase in the number of rear end impacts while only part of the car population is fitted with anti-lock. Also frequently raised are the 'risk compensation' arguments suggesting that drivers of anti-lock vehicles will brake later and harder because the risk of wheel lock is eliminated, or will corner faster because they know they can brake on corners. Similar arguments have often in the past been brought against new safety features, such as better tyres, disc brakes or seat belts and the extent to which they apply can only be estimated after substantial amounts of real life experience. It may be worth insurance companies considering some sort of additional financial incentive to encourage their customers to buy cars fitted with anti-lock, if they prove safer on the road. 6. The deceleration characteristics were found to be closely related to the degree of complexity and cost of the systems. 7. The more expensive three channel electronic systems significantly increased the deceleration levels attainable on all the surfaces and at all but quite low speeds. An appreciable increase at all but low speeds was also provided by the cheaper two channel, dual axle control, electronic system. There was a slight increase provided by the two channel, front wheel only, systems on some of the test surfaces but on most, particularly at lower speeds, the anti-lock actually reduced the braking deceleration slightly, although never to an unacceptable level. 8. Detailed analysis of accident data is required before any definite conclusions can be drawn regarding the relative importance of greater controllability compared with reduced stopping distances. 9. All the anti-lock systems tested are considered to have a potential to greatly reduce the number and severity of road accidents. A basic cost benefit analysis indicates that all the systems tested, including the most sophisticated when allowance is made for future cost reductions, might be financially viable if this potential is realised. 7. ACKNOWLEDGEMENTS The work described in this report was carried out in the Vehicle Safety Division of the Vehicles Group of the TRRL. Thanks are due to the TRRL garage staff and vehicle suppliers for their help and assistance. 6. CONCLUSIONS 1. Six modern anti-lock braking systems have been tested on a variety of road surfaces, with the vehicles both unladen and fully laden, and at speeds ranging from 20 to 100 km/h. 2. The performance of each system has been compared to that of the same vehicle with the system partially or wholly inoperative. 3. The deceleration characteristics have been assessed by heavily braking the vehicles from a known speed to rest whilst driving in a straight line on several surfaces having single (that is uniform) coefficients of friction. 4. The controllability of the systems has been assessed by heavily braking whilst driving on a curved path on single coefficient surfaces and whilst driving in a straight line over split coefficient surfaces. 5. The vehicle controllability provided by all the anti-lock braking systems was found to be excellent. 8. REFERENCES BLECKMANN, H.W, J. BURGDORF, H. VON GRUNBERG, K. TIMTNER and L. WEISE (1983). The First Compact 4-Wheel Anti-Skid System with Integral Hydraulic Booster. SAE Paper No Detroit, DEPARTMENT OF TRANSPORT (1990). Road Accidents Great Britain The Casualty Report. HMSO. DOUGLAS, J.W. and T.C. SCHAFER (1971). The Chrysler "Sure-Brake"- The First Production Four-Wheel Anti-Skid System. SAE Paper No Detroit, HOLMES, K.E. and R.D. STONE (1969). Tyre Forces as Functions of Cornering and Braking Slip on Wet Road Surfaces. TRRL Report No. LR 254. Transport and Road Research Laboratory, Crowthorne. LEIBER, H. and A. CZINCZEL (1979). Antiskid System for Passenger Cars with a Digital Electronic Control Unit. SAE Paper No Detroit,