THE PREDICTION OF TYRE FRICTION ON WET ROADS UNDER BRAKING AND CORNERING by M.Sjahdanulirwan, CivE ITB (Indonesia), MSc B' ham (UK) A thesis submitted in fulfilment of the requirements for the Degree of Doctor of Philosophy School of Civil Engineering The University of New South Wales Australia February 1993
326
Embed
the prediction of tyre friction on wet roads under braking and ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
THE PREDICTION OF TYRE FRICTION
ON WET ROADS
UNDER BRAKING AND CORNERING
by
M.Sjahdanulirwan, CivE ITB (Indonesia), MSc B' ham (UK)
A thesis submitted in fulfilment of the requirements for the Degree of
Doctor of Philosophy
School of Civil Engineering The University of New South Wales
Australia
February 1993
11
CERTIFICATE or ORIGINALITY
I hereby declare that this submission is my own work and that, to the best of my knowledge and belief, it contains no material previously published or written by another person nor material which to a substantial extent has been accepted for the award of any other degree or diploma of a university or other institute of higher learning, except where due acknowledgement is made in the text.
SUPERVISOR'S CERTIFICATE
(Signed)
(Name)
(Date)
M.SJAHDANULIRWAN
r certify that this thesis is, in my opinion, in a form suitable for examination for the degree of Doctor of Philosophy.
(Signed)
(Name) Assoc.Professor W.O.Yandell
(Date)
iii
Abstract
This thesis extends the prediction of tyre-road friction from surface texture and
tread rubber properties developed by Yandell et.al [1983], to cover all ranges of
percent slip (Sx) and slip angles (a) by means of developing an analytical tyre-road
friction model. The proposed model consists of two main components: the locked
wheel Braking Force Coefficient (BFC) or maximum Sideway Force Coefficient
(SFC), and the tyre stiffnesses.
It has been found from early works, that the coefficients of locked-wheel braking
(i.e. at Sx = 100%) and the maximum sideway force (i.e. at optimum a) can be
predicted with reasonably well agreement, in which the factors of surface texture,
water film thickness, tread rubber properties, and certain operating conditions have
been taken into account. The mode of operation for any conditions, such as braking
or cornering, would be the important factor to be considered when dealing with all
ranges of Sx and a.
From previous information it is known that at below the optimum value of Sx (or
a), the relationship between BFC and Sx, or SFC and a, is controlled primarily by
the elastic tyre properties. The effects of "road-contributed" friction (µ) appear
dominantly at and above the optimum value of Sx (or a). It is assumed that the
elastic tyre properties can be represented by the tyre stiffnesses (longitudinal and
lateral), whereas the available friction (µ) will be represented by any value of
coefficient of braking (or cornering) from optimum to maximum value of Sx (or a).
The choice of locked-wheel BFC or maximum SFC as input parameter for the
candidate's tyre model is primarily based on two reasons. First, that the methods for
predicting the locked-wheel BFC or maximum SFC from surface texture
measurements already exist [Yandell et.al (1983)]; hence, the ultimate goal to
develop a theoretical technique for predicting tyre-road friction, from surface texture
data, still be extended. Second, that the equipment for measuring the locked-wheel
BFC or maximum SFC are commonly available (such as the trailer used on
ASTM-E274, and SCRIM); hence, the data from these types of equipment can be
processed directly to obtain the friction for all ranges of Sx and a.
IV
At the present time, the tyre stiffnesses C,. and Cy can be obtained
experimentally by measuring forces and slips (or deformations), either on a road
surface or in a laboratory. Especially for stiffness measurements in a road surface,
the candidate has developed an empirical model for the variation of tyre stiffness
due to vertical load, inflation pressure, and tyre ( or vehicle) speed.
Attempts to predict tyre stiffness directly from other tyre properties or related
parameters are left for future refinement. For stiffness measurements in a laboratory,
the need for applying an empirical model of stiffness variation is probably less
beneficial than the accuracy gained with direct measurement on different variations
(vertical load, inflation pressure). An empirical factor may still be required, however,
if the change of vehicle speed has a significant effect on tyre stiffness.
Field measurements of the coefficients of frictional force are carried out by using
the Multi Mode Friction Test Truck (MMFTT). This vehicle was capable for
measuring pure braking, pure sideway (up to 15 degrees), and braking-in-turning
modes. The use of test wheel motor and gear box allows the accelerating and steady
state braking to be performed.
The analysed data from MMFTT are compared with theoretically predicted
coefficients of friction. The candidate's model is also verified with the laboratory
data obtained from another worker [Sakai (1982)]. The concept of the independence
of tyre stiffness from road conditions, and the difference between the slip stiffness
and the deformation stiffness, have been further examined by additional experiments.
It appeared that the theoretical method of obtaining the coefficient of tyre-road
friction under all conditions, from tyre stiffness and locked-wheel BFC (or maximum
SFC), is feasible. As only one friction parameter is required in the proposed model,
the time and cost associated with the measurements (and calculation) of the friction
parameters can be reduced. In addition, the use of unsymmetric trapezoidal forms of
distribution provides the user with an option and flexibility in representing many
types of actual tyre pressure distribution.
V
Acknowledgements
The candidate wishes to express his thanks to all those who assisted in this
research project and in the preparation of the thesis, including:
Dr.W.O.Yandell, Associate Professor at School of Civil Engineering UNSW, for
his encouragement, guidance, advice, and particularly for the constructive criticism
and helpful supervision throughout the work;
Dr.W.H.Cogill, Researcher, for his critical comments and helpful discussion,
especially on mathematical terms;
Dr.Q.Yang, Research Fellow, for reading certain chapters of the thesis and
p~oviding suggestions for improvements;
Mr.M.J.Kerkhoff, Professional Officer, for his assistance on the electrical matters,
for providing the guidelines on operating the Strain Gauge Amplifier & Acrolog 400,
and for supplying the Program to Read Strain Gauge and Speed Output;
Mr.R.B.McKinnon, Laboratory Officer, for his expert workmanship in the
construction of calibration equipment, and for his help in carrying out the
preliminary test;
Mr.P.J.Gwynne, Technical Officer, for his help during calibration and field
experiment, and especially for operating the Multi Mode Friction Test Truck
(MMFTT);
And to all the other members of the staff in the School and Laboratory Workshop
who have willingly given advice and assistance whenever it was requested.
The experimentation carried out during this research would not have been
possible without the financial support of the Roads and Traffic Authority (RTA) of
New South Wales, and this is gratefully acknowledged.
Finally, the candidate acknowledges his wife, Atu, for her support and patience in
many ways and always standing by him.
Abstract Acknowledgements Table of Contents List of Figures List of Tables
Introduction
Chapter 1. Review of Literature
vi
Table of Contents
1. 1. Nature and Components of Tyre-Road Friction 1.1.1. Hydrodynamic component of friction 1.1.2. Adhesive friction 1.1.3. Hysteretic friction 1.1.4. Abrading friction
1.3. Contact and Slip Between Tyre and Roadway 1.3.1. Classifications of tyre motions 1.3.2. Types of slip 1.3.3. Sliding and skidding 1.3.4. Contact area 1.3.5. Stress distribution
1.4. Surface Texture Description and Measurement 1.4.1. Description of surface texture 1.4.2. The measurement of surface texture
1.5. Hydrodynamic Effects and Tread Rubber 1.5.1. Reynolds theory of lubrication 1.5.2. Fluid film thickness 1.5.3. Hydroplaning 1.5.4. Models for calculation of water film thickness 1.5.5. Measurement of tread rubber properties
1.6. Theory of Hysteretic Sliding Friction 1.6.1. The mechano-lattice analogy 1.6.2. Mechano-lattice unit 1.6.3. The friction of rubber sliding on an asperity 1.6.4. The hysteretic friction of road surfaces
1.7. Problem Definition and Method of Investigation 1. 7 .1. Problem definition 1.7.2. Method of investigation
Page No. iii
V vi xi
XX
1
3
3 5 5 6 7
8 8
11 15
22 22 23 25 26 27
33 35 39
44 44 45 46 47 50
52 53 54 55 58
59 59 64
vii
Chapter 2. Measurement and Prediction of Tyre-Road Friction 65
2.1. Methods of Measuring Tyre-Road Friction 65 2.1.1. Sideway force measuring method 66 2.1.2. Braking force measuring method 67 2.1.3. Stopping distance method 67 2.1.4. Portable skid resistance testers 68
2.2. Some Theoretical Methods of Predicting Tyre-Road Friction 68 2.2.1. Schonfelds's photo-interpretation of pavement 69 2.2.2. Leu and Henry's model of skid resistance 71 2.2.3. Texture friction meter 72
2.3. Theoretical Prediction of the Locked-Wheel Braking and Sideway Force Coefficients 73
2.3.1. Modification of the theory of hysteretic friction 73 2.3.2. Some assumptions of rubber/water temperature 73 2.3.3. Prediction of the locked wheel braking and sideway force coefficients 75
2.4. The Longitudinal and Lateral Tyre Stiffness 76 2.4.1. Effects of normal (vertical) load 77 2.4.2. Effects of inflation pressure 79 2.4.3. Effects of speed 80 2.4.4. Effects of tyre wear (bald & new) 80 2.4.5. Measurement of tyre stiffness and its related parameters 81 2.4.6. Clarification of tyre stiffness formula and its units 84 2.4.7. The empirical model for tyre stiffness variation 86
2.5. The Existing Tyre Models Under Braking and Cornering 92 2.5.1. Analytical and semi-empirical models 93 2.5.2. Empirical models 97 2.5.3. Miscellaneous models 98
2.6. The Candidate's Model for the Prediction of Tyre-Road Friction Under Braking and Cornering 99
2.6.1. General features of the model 100 2.6.2. Geometry of tyre-road contact 102 2.6.3. Pressure distribution and coefficients of friction 103 2.6.4. Location of sliding boundary 105 2.6.5. The calculation of shear forces 107 2.6.6. Brief procedure using locked-wheel BFC 109 2.6.7. Illustrative example 112 2.6.8. Alternative model using maximum SFC 120
Chapter 3. The Parametric Study of the Model 121
3.1. General 121
3.2. Results of the Main Parameters 122 3.2.1. The effects of longitudinal stiffness 123 3.2.2. The effects of lateral stiffness 126 3.2.3. The effects of locked-wheel BFC 129
viii
3.3. The Significance of All Variables 3.3.1. Dependent and independent variables 3.3.2. Model specification 3.3.3. Regression analysis 3.3.4. Significance of longitudinal force coefficient 3.3.5. Significance of lateral force coefficient 3.3.6. Significance of resultant force coefficient
3.4. Comparison With Other Tyre Models 3.4.1. General description of tyre models 3.4.2. Summary of formulas 3.4.3. Model input data 3.4.4. Model response comparisons
Chapter 4. Field Measurement of Friction
4.1. Objectives of the Friction Measurement
4.2. Design of Measurement 4.2.1. Sample size 4.2.2. Randomization 4.2.3. Choice of levels used
4.3. The Multi Mode Friction Test Truck 4.3.1. The test wheel 4.3.2. The recording device 4.3.3. The road wetting device
4.4. Calibration 4.4.1. Wheel load calibration 4.4.2. Braking force calibration 4.4.3. Sideway force calibration
4.5. Experimentation 4.5.1. Sequence of experiment 4.5.2. Test procedure 4.5.3. Data logging program and corrections
4.6. Possible Variability on Test Results 4.6.1. Fluctuation of voltage supplied 4.6.2. Setting up of slip angle 4.6.3. Test location 4.6.4. Speed 4.6.5. Inflation pressure 4.6.6. Temperature 4.6.7. Rubber properties
133 133 134 135 137 139 142
145 145 146 152 153
158
158
158 158 159 159
163 163 164 164
169 169 169 172
172 172 173 174
175 175 176 176 177 177 177 178
ix
Chapter 5. Analysis of Experimental Results 179
5 .1. General 179
5.2. Presentation of Data 180 5.2.1. Friction coefficient versus time 180 5.2.2. Correction for test wheel speed 184 5.2.3. The actual percent slip on entering the measured section 190
5.3. Transformation into Percent Slip and Slip Angle 191 5.3.1. Correction for wheel speed delay 191 5.3.2. The smoothened technique for wheel speed 193 5.3.3. Transformation of time into percent slip 194 5.3.4. Extrapolations and analysis of variance 203
5.4. Determination of Parameters for Theoretical Prediction 208 5.4.1. Longitudinal stiffness 208 5.4.2. Lateral stiffness 210
5.5. Discussion of Results 211 5.5.1. General results 211 5.5.2. Effects of tyre pressure 215 5.5.3. Effects of the rate of braking 216
Chapter 6. Theoretical Prediction of Tyre-Road Friction Under Braking and Cornering 217
6.1. General 217
6.2. Verification of Tyre Stiffness Variation Model 218
6.3. Input Data for Prediction and Comparison of Tyre-Road Friction 220 6.3.1. From candidate's experimental result 220 6.3.2. From Sakai's laboratory data 221
6.4. Theoretical Prediction and Comparison with Experimental Result 223
6.5. Theoretical Prediction and Comparison with the Sakai's Laboratory Data 232
6.6. Theoretical Prediction of Slippery Road 239
6.7. Theoretical Prediction Using Tyre Stiffness from Load-Deflection Measurements 241
6.6. Viability and Reliability of the Theoretical Prediction of Tyre-Road Friction as a Substitute for Actual Skid Tests 248
X
Chapter 7. Conclusions 251
7 .1. Theoretical Findings 251 7 .1.1. Types of tyre stiffness 251 7 .1.2. Description of friction coefficient 252 7 .1.3. The role of tyre stiffness 252 7 .1.4. Model for tyre stiffness variation 253 7 .1.5. The shape of tyre pressure distribution 253 7 .1.6. Calculation of frictional forces 254 7.1.7. Theoretical effect of some input parameters on the tyre-road friction 255 7 .1.8. Theoretical model response comparisons 256
7.2. Experimental Finding5 257 7.2.1. Rectification of experimental problems 257 7.2.2. The significance of the variation of tyre pressure and rate of braking 259 7 .2. 3. Verification of tyre slip stiffness variation model 260 7.2.4. Comparison between measured and theoretically predicted friction 261 7.2.5. Comparison of Sakai's laboratory data
with the candidate's theoretical prediction of friction 261 7.2.6. The effect of modifying the quantification of lateral stiffness
in the candidate's friction prediction 262 7.2. 7. Independence of tyre stiffness from road conditions 262 7.2.8. The relationship between slip stiffness and deformation stiffness 263
7.3. Benefit of the Model 264
7.4. Summary of Conclusions 266
7.5. Future Work 269
References 270
Appendix A: Basic Computer Program for Calculation of Frictional Forces Using Locked-Wheel BFC A-1
Appendix B: Basic Computer Program for Calculation of Frictional Forces Using Maximum SFC B-1
Appendix C: Procedure to Operate Strain Gauge Amplifier C-1
Appendix D: Procedure to Operate Acrolog 400 D-1
Appendix E: Program to Read Strain Gauge and Speed Output E-1
Appendix F: Results of investigation using MMFfT, with higher speed (20 mph & 30 mph) F-1
Appendix G: Mathematical Derivations of Tyre Stiffness Formula G-1
xi
List of Figures
1.1. Schematic representation of 3 zone concept.
1.2. Components of rubber friction.
1.3. Basic tyre structures.
1.4. Effect of speed and road texture to skid resistance.
1.5. Friction mechanism due to adhesion and hysteresis.
1.6. Lateral force coefficient vs slip angle of 178-15 tyre.
1.7. Braking force coefficient vs slip angle of 178-15 tyre.
1.8. Peak and slide traction envelopes of 178-15 tyre in braking-in-a-tum.
1.9. Model experiments for: (a) free rolling, (b) side slip, (c) braking, (d) driving. Travelling direction from right to left.
1.10. Typical contact area for an automobile tyre.
1.11. Pressure distribution down the length of a tyre with no tread pattern, at constant speed 15 kph, inflation 6.5 atm, load: (a) 1680 kp, (b) 3740 kp.
1.12. Pressure distribution across the width of a tyre with no tread pattern, constant speed 15 kph, load: (a,b,c) 1680 kp, (d) 3740 kp; inflation: (a) 3.5 atm, (b) 5 atm, (c,d) 6.5 atm.
1.13. Pressure distribution down the length of a smooth tyre, load 1680 kp, infl. 6.5 atm, speed: (a) 10.2 kph, (b) 8.4 kph, (c) 8.9 kph; acceleration: (a) 0.6 m/s2,
(b) '1.6 m/s2, (c) 4.1 m/s2•
1.14. Pressure distribution down the length of a smooth tyre, load 1680 kp, inflation 6.5 atm, speed: (a) 10 kph, (b) 35.7 kph, (c) 14.5 kph; deceleration: (a) 4.1 m/s2, (b) 2.7 m/s2, (c) 2 m/s2•
1.15. Pressure distribution of bias-belted H size tyre, 26 psi, 0° slip angle, 100% T and R load = 1580 lb.
1.16. Longitudinal shear stress distribution in the running direction of a tyre under free rolling.
1.17. Longitudinal shear stress distribution in the running direction of a bias-belted H size tyre under traction, 26 psi, 0° slip angle, 100% T and R load= 1580 lb.
1.18. Longitudinal shear stress distribution in the running direction of a smooth tyre under traction, load 1590 kp, inflation 6.5 atm.
1.19. Longitudinal shear stress distribution in the running direction of a smooth tyre under braking force, load 2610 kp, inflation 6.5 atm.
Xll
List of Figures (continued)
1.20. Lateral force intensity along the contact spot, at slip angles up to 12°.
1.21. Lateral force intensity: (a) due to side slip (lg), and (b) due to simultaneous side slip and circumferential slip (tc).
1.22. Mathematical definition for numerical evaluation of surf ace texture.
1.23. Profile ratio and "filtered profile ratio".
1.24. Moore's bearing area method.
1.25. Definition of surface characteristics.
1.26. Influence of the number of inflections.
1.27. Illustration of terms of the road surface texture.
1.28. The texture effect on film thickness as a function of time.
1.29. Measured relationship between the damping factor of filled tread rubber, speed, and rubber temperature.
1.30. Simplified of measured hysteresis loop.
1.31. A unit of the mechano-lattice analogy.
1.32. The assembled units simulating a long section of rubber sliding on an asperity.
1.33. Flow chart of the analogy computer programme.
1.34. Mechano-lattice-determined relationships between hysteretic friction, damping factor of sliding rubber, and average slope of contact of triangular prisms and cylinders.
1.35. Analysis of measured texture profiles into scales.
1.36. Schematically presentation of the coefficient of hysteretic friction of the road surface.
1.37. Measured BFC vs predicted BFC, on many types of pavement (concrete, bituminous) and macro-texture (closed, mild, smooth).
1.38. Measured SFC vs predicted SFC, on many types of pavement (concrete, bituminous) and macro-texture (closed, mild, smooth).
2.3. Friction weights of texture parameter E or F.
xiii
List of Figures (continued)
2.4. Texture Friction Meter (TFM) vs British Pendulum Number (BPN) and Multi Mode Friction Test Truck (MMFTT).
2.5. Schematic representation of masking of texture.
2.6. Longitudinal stiffness coefficient (CxC) vs tyre load of radial-ply (R), bias-belted (B), and bias-ply (D) tyres.
2.7. Lateral stiffness coefficient (CyC) vs tyre load of radial-ply (R), bias-belted (B), and bias-ply (D) tyres.
2.8. Longitudinal force (Fx) versus travel of flat bed table supporting a radial tyre at the indicated vertical loads.
2.9. Lateral force (Fy) versus travel of flat bed table supporting a radial tyre at the indicated vertical loads.
2.10. The kinematics of tyre motion.
2.11. The distorted equatorial line.
2.12. The equatorial line with adhesion & sliding region.
2.13. Root line and deflection of the profile lugs.
2.14. Simplified diagram for calculation (or prediction) of frictional forces.
2.15. Tyre-road contact geometry, and deformation in nonsliding portion of contact patch.
2.16. Idealized pressure distribution within the contact patch.
2.17. Longitudinal and lateral components of (available) friction coefficient.
2.18. Longitudinal and lateral components of frictional forces coefficient.
2.19. The contribution of tyre stiffness (C) and the available friction (µ) to the frictional forces (F). At slips beyond the critical value, then Jls = µx ( or µa = µy).
2.20. Determination of the available longitudinal friction at zero slip (µ50).
2.21. Determination of the available lateral friction at zero slip (µa0)
2.22. Calculation of shear (frictional) forces.
2.23. -Tyre model (--) and measured data (o o o) of H78-14 tyre (28 psi) with a/21 = 0.06 and b/21 = 0.20. Left: Fx & FY vs Sx at a= 4°. Right: FY vs a at Sx = 0.
xiv
List of Figures (continued)
2.24. Tyre model of H78-14 (28 psi) with b/21 = 0.20, using a/21 = 0.11 (--) and a/21 = 0.01 (------). Left: Fx & Fy vs Sx at a= 4°. Right: Fy vs a at Sx = 0.
2.25. Tyre model of H78-14 (28 psi) with a/21 = 0.06, using b/21 = 0.01 (--) and b/21 = 0.40 (------). Left: Fx & Fy vs Sx at a= 4°. Right: FY vs a at Sx = 0.
2.26. The computed relationship between µx and µY at various a (0 ), with a/21 = 0.06 and b/21 = 0.20.
2.27. The computed relationship between µx and Uy at various Sx (% ), with a/21 = 0.06 and b/21 = 0.20.
2.28. Determination of the available lateral friction at zero slip (µa0)
3.2. ex vs µscr• for various µd.
3.3. ex vs µx, a = 0°, µd = 0.6.
3.4. ex vs µx, a = 4°, µd = 0.6.
3.5. ex vs µx, a = 8°, µd = 0.6.
3.6. ex vs µx, a = 32°, µd = 0.6.
3.7. ex vs µy, sx = 0.1, µd = 0.6.
3.8. ex vs µy, sx = 0.2, µd = 0.6.
3.9. ex vs µy, sx = 0.5, µd = 0.6.
3.10. ex vs µx, sx = 0.1, µd = 0.6.
3.11. ex vs µx, sx = 0.2, µd = 0.6.
3.12. ex vs µx, sx = 0.5, µd = 0.6.
3. 13. ey vs a.er> for various µd.
3.14. eY vs µa.er, for various µd.
3.15. ey vs µy, Sx = 0, µd = 0.6.
3.16. ey vs µy, Sx = 0.1, µd = 0.6.
3.17. ey vs µy, Sx = 0.2, µd = 0.6.
3.18. ey vs µy, Sx = 0.5, µd = 0.6.
List of Figures (continued)
3.19. Cy vs µx, ex= 4°, µd = 0.6.
3.20. Cy vs µx, ex= go, µd = 0.6.
3.21. Cy vs µx, ex = 32°, µd = 0.6.
3.22. Cy vs µy, ex = 4°, µd = 0.6.
3.23. Cy vs µy, ex = go, µd = 0.6.
3.24. Cy vs µy, ex = 32°, µd = 0.6.
3.25. µd vs SXCI" for various ex.
3.26. µd vs excr, for various CY'
3.27. µd vs µsa• for various Cx-
3.2g. µd vs µacr, for various Cy-
3.29. µd vs µx, ex = (}°, ex = 35.
3.30. µd vs µx, ex = 4°, ex = 35.
3.31. µd vs µx, ex = go, ex = 35.
3.32. µd vs µx, ex = 32°, ex = 35.
3.33. µd vs µy, Sx = 0, CY = 25.
3.34. µd vs µy, Sx = 0.1, Cy = 25.
3.35. µd vs µy, Sx = 0.2, Cy = 25.
3.36. µd vs µy, Sx = 0.5, CY = 25.
3.37. µd vs µx, sx = 0.1, ex= 35.
3.3g_ µd vs µx, sx = 0.2, ex = 35.
3.39. µd vs µx, sx = 0.5, ex= 35.
3.40. µd vs µy, ex = 4°, Cy = 25.
3.41. µd vs µy, ex = go, Cy = 25.
3.42. µd vs µy, ex = 32°, Cy = 25.
xv
3.43. Comparison of longitudinal slip (Sx) versus longitudinal force coefficient (µx = FxfFz) of six tyre models operated at the indicated slip angle (degrees).
xvi
List of Figures (continued)
3.44. Comparison of longitudinal slip (Sx) versus lateral force coefficient (µy = F/Fz) of six tyre models operated at the indicated slip angle (degrees).
3.45. The interaction of longitudinal force coefficient (µx = FxfFz) with lateral force coefficient (µy = F/Fz) at the indicated slip angle (degrees).
3.46. Comparison of the straight-ahead longitudinal force response of six tyre models.
3.47. Comparison of the free-rolling lateral force response of the six tyre models.
4.1. Medium-textured bituminous surface.
4.2. Catadioptric targets behind the MMFTT.
4.3. The recording device on cabin.
4.4. Hydraulic and braking controls on cabin.
4.5. Water discharging apparatus & wind shield (in front of test wheel), at near the right side of MMFTT rear axle.
4.6. Infra-red photo detector ( ..1.) and the disc brake for test wheel (i), at near the left side of MMFTT rear axle.
4. 7. Calibration of vertical, longitudinal and lateral forces.
4.8. The MMFTT and the arrangement for calibration.
4.9. Wheel load calibration curve.
4.10. Braking force calibration curves for angles 0°, 10°, and 15°.
4.11. Sideway force calibration curves for angles 0°, 10°, and 15°.
4.12. Testing sequence.
5.1. Experiment result vs Time of treatment A (3 replicates). Legend: -c-(BFC), -<>-(SFC), L-(Wheel Speed), -(Yeh.Speed).
5.2. Results of IP 30 psi & fast braking, for treatments A (0°), I (10°), Q (15°). Legend: -c-(BFC), -<>-(SFC), L-(Wheel Speed), -(Yeh.Speed).
5.3. Results of IP 30 psi & slow braking, for treatments C {0°), K (10°), S (15°). Legend: -c-(BFC), -<>-(SFC), L-(Wheel Speed), -(Yeh.Speed).
5.4. Results of IP 20 psi & fast braking, for treatments E (0°), M (10°), U (15°). Legend: -c-(BFC), -<>-(SFC), L-(Wheel Speed), -(Yeh.Speed).
xvii
List of Figures (continued)
5.5. Results of IP 20 psi & slow braking, for treatments G (0°), 0 (100), W (15°). Legend: ---o-(BFC), -<>-(SFC), L-(Wheel Speed), -(Yeh.Speed).
5.6. The comparison between the corrected Wheel Speed (WS) on entering the measured section, and the actual MMFIT speed (VS) averaged over the measured section.
5.7. BFC (---o-) & SFC (-<>-) vs percent slip (S,J; IP 30 psi & fast braking, for treatments A (0°), I (10°), Q (15°).
5.8. BFC (---o-) & SFC (-<>-) vs percent slip (Sx); IP 30 psi & slow braking, for treatments C (0°), K (100), S (15°).
5.9. BFC (---o-) & SFC (-<>-) vs percent slip (Sx); IP 20 psi & fast braking, for treatments E (0°), M (10°), U (15°).
5.10. BFC (---o-) & SFC (-<>-) vs percent slip (Sx); IP 20 psi & slow braking, for treatments G (0°), 0 (10°), W (15°).
5.11. SFC vs slip angle; IP 30 psi & fast braking, from treatments AIQ. Legend: -<>- (SFC at Sx = 0%), -1:,.- (SFC at Sx = 100%).
5.12. SFC vs slip angle; IP 30 psi & slow braking, from treatments CKS. Legend: -<>- (SFC at Sx = 0%), -1:,.- (SFC at Sx = 100%).
5.13. SFC vs slip angle; IP 20 psi & fast braking, from treatments EMU. Legend: -<>- (SFC at Sx = 0%), -1:,.- (SFC at Sx = 100%).
5.14. SFC vs slip angle; IP 20 psi & slow braking, from treatments GOW. Legend: -<>- (SFC at Sx = 0%), -1:,.- (SFC at Sx = 100%).
5.15. Determination of longitudinal stiffness (Cx) from the relationship of BFC and Sx (% ), for treatment A (0°).
5.16. Determination of lateral .stiffness (Cy) from the relationship of SFC and a. (deg), for treatments AIQ.
5.17. The locked-wheel BFC & SFC, measured at the first and the second half time of full braking.
5.18. The locked-wheel RFC, measured at the first and the second half time of full braking.
5.19. The uniformity of measured section (25 m), assessed by steady-state BFC at constant slip (Sx) of 20.8%.
5.20. The wheel load at brake off (average of 1 second before activation and 1 second after release), and at brake on.
xviii
List of Figures (continued)
6.1. Input data ( · · c · ·) of Fx versus Sx (at angle 0°), various F2 , from laboratory data [Sakai (1982)].
6.2. Input data ( · · <> · ·) of F Y versus a ( at Sx 0% ), various F 2 , from laboratory data [Sakai (1982)].
6.3. Field experiment (c), and theoretical values of BFe (at angle 0°) by candidate's model: field ex, a/21 = 0.1 (--) & 0.06 (------).
6.4. Field experiment (a), and theoretical values of BFe (at angle 10°) by candidate's model: field ex & ey, a/21 = 0.1 (--) & 0.06 (------).
6.5. Field experiment (<>), and theoretical values of SFe (at angle 10°) by candidate's model: field ex & ey, a/21 = 0.1 (--) & 0.06 (------).
6.6. Field experiment (a), and theoretical values of BFe (at angle 15°) by candidate's model: field ex & eY' a/21 = 0.1 (--) & 0.06 (------).
6.7. Field experiment (<>), and theoretical values of SFe (at angle 15°) by candidate's model: field ex & ey, a/21 = 0.1 (--) & 0.06 (------).
6.8. Field experiment (<>), and theoretical values of SFe (at Sx 0%) by candidate's model: field ey, a/21 = 0.10 (--) & 0.06 (------).
6.9. Field experiment (~). and theoretical values of SFe (at Sx 100%) by candidate's model: field ex & ey, a/21 = 0.10 (--) & 0.06 (------).
6.10. Field experiment (a), and theoretical values of BFe by candidate's model: field ex & ey, a/21 = 0.10, a = 10° (--) & 8° (------).
6.11. Field experiment ( <> ), and theoretical values of SFe by candidate's model: field ex & ey, a/21 = 0.10, a = 10° (--) & 8° (------).
6.12. Field experiment (c), and theoretical values of BFe by candidate's model: field ex & ey, a/21 = 0.10, a = 15° (--) & 12° (------).
6.13. Field experiment (<>), and theoretical values of SFe by candidate's model: field ex & c;,, a/21 = 0.10, a = 15° (-) & 12° (------).
6.14. Sakai's data (c), and theoretical values of Fx (at angle 0°) by candidate's model: parameters from Table 6.5, a/21 = 0.06, various Fz.
6.15. Sakai's data(<>), and theoretical values of FY (at Sx 0%) by candidate's model: parameters from Table 6.5, a/21 = 0.06, various F2 •
6.16. Sakai's data (0 ), and theoretical values of Fx (at angle 0°) by candidate's model: parameters from Table 6.5, a/21 = 0.08, various Fz-
6.17. Sakai's data(<>), and theoretical values of FY (at Sx 0%) by candidate's model: parameters from Table 6.5, a/21 = 0.08, various Fz.
xix
List of Figures (continued)
6.18. Sakai's data (c), and theoretical values of Fx (at angle 0°) by candidate's model: parameters from Table 6.5, a/21 = 0.10, various F2 •
6.19. Sakai's data (<>), and theoretical values of FY (at Sx 0%) by candidate's model: parameters from Table 6.5, a/21 = 0.10, various F2 •
6.20. Sakai's data (<>), and theoretical values of FY (at Sx 0%) by candidate's model: parameters from Table 6.5 (modified ey), various a/21 & F2 •
6.21. Field BFe (c) & SFe (<>) on slippery road at angles 0° (X), 100 (Y), and 15° (Z); and theoretical values by candidate's model: from Table 6.9, a/21 = 0.10.
6.22. Arrangement for measuring the laboratory longitudinal tyre stiffness (ex).
6.23. Arrangement for measuring the laboratory lateral tyre stiffness (ey).
6.24. Field BFe (c) on main experiment road (at angle 0°), and theoretical values by candidate's model: a/21 = 0.10, ex from Field(--) & Laboratory(------).
6.25. Field BFe (c) on slippery road (at angle 0°), and theoretical values by candidate's model: a/21 = 0.10, ex from Field(--) and Laboratory (------).
6.26. Field SFe (<>) on main experiment road (at Sx 0%), and theoretical values by candidate's model: a/21 = 0.10, eY from Field (--) and Laboratory (------).
6.27. Field transient BFe (c) on main experiment road (at angle 0°), and theoretical values by candidate's model: a/21 =0.10, ex from Field (--) & Laboratory (------). Field steady-state BFe (*) are also shown in this figure.
7.1. Inputs and outputs of the candidate's model under either braking or cornering. Left~ Input: ex, locked-wheel BFe. Output: Fx vs Sx (at angle 0°). Right~ Input: CY' locked-wheel BFe (or max SFC). Output: FY vs a (at Sx 0%).
7.2. Inputs and outputs of the candidate's model under both braking and cornering. Left~ Input: ex, ey, locked-wheel BFe (or max SFe). Output: FY vs Fx (various Sx). Right~ Input: ex, ey, locked-wheel BFe (or max SFe). Output: Fy vs Fx (various a).
XX
List of Tables
2.1. Assumed rubber/water temperature.
2.2. The decrease of e"e with increasing F2 for 10.00-20/F tyres, at speed 8-88 kph.
2.3. The decrease of eye with increasing F2 of 078-15 tyres, at speed 6 mph, 28 psi.
2.4. Lateral spring rate (Ky) vs inflation pressure (IP) for truck tyres.
2.5. The relationship between eye and F2 at different speed, for 078-15 tyre, 28 psi.
2.6. The e" and eY of truck tyres (10.00-20/F) in 3 states of wear.
2.7. Results of Fx and FY' at a. = 4°, showing contribution from tyre stiffness (ex, ey) and friction coefficient (µ).
2.8. Results of ~. at Sx = 0, showing contribution from tyre stiffness (Cy) and friction coefttcient (µ).
3.1. Detail of Independent Variables.
3.2. a-Coefficient (t-statistics)
3.3. Model Input Data.
4.1. Detailed treatments, showing the sequence of experiment (in bracket).
5.1. Raw data of treatment A (3 replicates).
5.2. Raw data of treatment A (mean).
5.3. The Recorded and Actual speeds (kph), on entering the measured section, for treatment AeEO (0°).
5.4. The speeds of Test Wheel (kph) on entering the measured section, before and after correction, compared with the speeds of MMFTT (kph) over the measured section.
5.5. The wheel speed (WS), vehicle speed (VS), and percent slip (Sx), for each treatment (Tr), on entering the measured section.
5.6. The wheel speed delay, for all treatments.
5.7. The parameters of normal equations of WS vs T, for all treatments.
5.8. Transformation into Sx scale of treatment A.
5.9. Summary of friction coefficients versus slip, for treatments ACEO (0°).
5.10. Summary of friction coefficients versus slip, for treatments IKMO (100).
xxi
List of Tables (contimied)
5.11. Summary of friction coefficients versus slip, for treatments QSUW (15°).
5.12. Representative friction coefficients versus slip, for analysis of variance.
5.13. Analysis of variance of Braking Force Coefficient (BFC).
5.14. Analysis of variance of Sideway Force Coefficient (SFC).
5.15. Longitudinal stiffness, from zero slip angle treatments (A, C, E, G).
5.16. Lateral stiffness, from zero percent slip values of all treatments.
6.1. Measured and predicted values of longitudinal stiffness coefficient (CxC).
6.2. Measured and predicted values of lateral stiffness coefficient (CyC).
6.3. Parameters for theoretical prediction, extracted from field experiment.
6.4. Input data of Fx and FY at various FZ' from laboratory data.
6.5. Parameters for theoretical prediction, extracted from laboratory data.
6.6. Theoretical prediction of BFC & SFC, versus percent slip (Sx), at various a/21, from the candidate's experimental result.
6.7. Theoretical prediction of SFC, versus slip angle (a.), at various a/21, from the candidate's experimental result.
6.8. Theoretical prediction of Fx vs Sx (a. = 0°), Fy vs a. (Sx = 0%), at various a/21, from Sakai's laboratory data.
6. 9. Parameters for theoretical prediction on slippery road.
6.10. Tyre stiffness from the field (main exp.) and laboratory (load-def.).
Introduction
When a road surface becomes wet, its ability to provide friction is greatly
reduced. For safe driving under wet conditions, a minimum level of friction between
the tyre and the road surface must exist. A knowledge of the availability of this
friction is therefore necessary.
It is a common way to express the skid resistance property of a pavement either
by Braking Force Coefficient (BFC) or Sideway Force Coefficient (SFC). Because of
the skid resistance force results from the interaction between tyre and road, the
friction coefficients are not solely the indicator of pavement properties. In addition
to the contribution of the tyre, some other factors are also involved in influencing
the friction: speed, load, water film thickness, temperature etc.
Many pieces of equipment such as ASTM Skid Trailer (for braking force) and
SCRIM (for sideway force), have been widely accepted as standards for measuring
skid resistance. Whereas the braking friction will determine the braking ability of a
vehicle (such as its stopping distance), the sideway friction will determine the
steering ability of a vehicle (such as its cornering speed).
In general, however, and especially under emergency condition braking and
cornering are present together, and the ability of either braking or steering is
modified by the presence of the other. It is then important to know how the
tyre-road friction will behave under all ranges of percent slip (Sx) and slip angles
(a).
Two ranges of S" (or a), in regard to the magnitude of friction, are generally
considered to have the different controlled factors. These are at below the optimum
value of S" (or a) which are controlled primarily by the tyre properties; and at or
above the optimum value of S" (or a) which are controlled by both the tyre and road
properties. It is important to realise this difference, when dealing with the prediction
of the tyre-road friction under braking and cornering.
This study forms part of the continuing research into tyre-road friction with the
ultimate aim of predicting the skid resistance of a road surface. The work therefore
1
includes the prediction and the validation of the coefficient of friction by field
experiment. The fundamental concepts for the prediction of friction had been
developed by Yandell [1970]. Their application to practical problems requires
modification such as works done by Taneerananon [1981]. However, to cover all
ranges of percent slip (Sx) and slip angles (a) it is then necessary to develop further
the method of prediction.
Two main works are done for this thesis: the development of the method of
prediction, and the field measurement of friction. Included in the prediction of
friction are the calculations of tyre stiffness, locked-wheel BFC or maximum SFC.
The field measurement of friction, on the other hand, is intended to serve
simultaneously two purposes: to supply the parameters needed for the prediction, and
to provide some comparisons for the results of prediction.
The literatures relevant to this research are reviewed in Chapter 1, which are
ended by the problem definition and the method of investigation. In Chapter 2 the
measurements and predictions of tyre-road friction, including the proposed
candidate's tyre model, are described. The effects of parameters influencing the tyre
road friction by using the proposed model, and the comparison of response between
this model and other tyre models, are presented in Chapter 3.
The field measurements of friction, including the calibration prior to the
experiment and the possible variability on test results, are explained in Chapter 4. In
Chapter 5 the analysis of experimental results and the determination of parameters
for theoretical prediction are described. Chapter 6 presents the comparison of the
experimental results with the predictive coefficients of friction, the verification of the
candidate's model with Sakai's laboratory data, the examination of the independence
of tyre stiffness from road conditions, and the investigation into the difference
between the slip stiffness and the deformation stiffness.
The theoretical and experimental findings of the research, including the benefit of
the candidate's tyre model, are presented in Chapter 7. The summary of conclusions
and the future work are also included in this chapter.
2
CHAPTER 1. REVIEW OF LITERATURE
1.1. Nature and Components of Tyre-Road Friction
When a vehicle executes maneuvers such as cornering, braking/accelerating, or
combination of two, the handling of this vehicle depends upon the available frictional
force between its tyres and the road surf ace. Frictional forces are highest on dry roads
where coefficient of friction of 0.7 to 1.0 are usual [Allbert and Walker (1965-66)]. If
the water or any lubricants, which act as medium of low tangential shear, wetted the
road surface, the friction available is drastically reduced. The reduction in friction
level depends on many factors, including: pavement/lubricant conditions (e.g. surface
texture, amount of water, temperature), tyre (e.g. tread pattern), and operating
conditions (e.g. speed, mode of operation).
The mechanism of friction development of a tyre rolling, sliding or combined
rolling/sliding on a wet surface, is commonly described by the concept of 3 zones of
contact area [Allbert and Walker (1965-66), Moore (1966), Harris (1968), Keen
(1968)]. The term "contact" in this concept is not necessarily associated with the direct
contact between tyre and road surf ace, since part ( or even all) of the water covering
the road may not be expelled during the time available. In fact, it just indicates the
area of road surface receiving pressure from the tyre, and thus generate resistance to
the tyre movements (see Figure 1.1).
In the forward region or sinkage zone, the tyre contacts the bulk of water covering
the road. The major component of frictional force then is due to the hydrodynamic
drag. Gough [ 1959] also included the hysteresis loss, as a component of frictional
force, although there is no direct contact between the tyre and road asperities. In other
words, tyre deformation can occur either with dry or lubricated contact. It is believed,
however, that the hysteresis friction does not occur if an excess of water exist, in
which the rolling or sliding tyre may plane on the water and prevent the rubber
flowing over the asperities. Nevertheless, the frictional force developed in this zone is
very small in comparison to other zones. When the vehicle velocity increases, the time
for expelling the water diminishes, the boundaries of the zones move backwards.
3
WATER FILM
ROAD SURFACE
DIRECTION OF MOTION
I T·YRE TO GROUND CONTACT I I I • 1
;
Zone I Region of unbroken water film, 'bullt' displacement zone. Zone 2 Region of partial breakdown of water film, 'thin film' zone. (Road surface
asperities penetrating.) Zone 3 Region of dry tyre to road contact, 'dry' zone.
Fig.1.1. Schematic representation of 3 zone concept. [Allbert and Walker (1965-66)].
Along with increasing velocity, the dynamic pressure generated in the water (p = 0.5pV2/g, where p is water density, V is vehicle velocity, and g is gravitation) also
increases. If the water pressure exceeds the tyre vertical pressure then the tyre is
supported by a layer of water film and the forward zone occupies the whole contact
area. This condition is often referred to as hydroplaning.
In the middle region or transition zone, the tyre proceeds to drape over the larger
asperities and make contact with the lesser asperities. The frictional force developed is
the drag of partial breakdown of lubricant film, and the hysteresis loss. Due to the
squeeze effect, the drag in this zone is greater than in the forward region (although
both of them are still very small). The hysteresis loss is built up as more tyre tread
envelops the projecting aggregate of the pavement surface. Additionally, it is possible
however, that the adhesive friction is also being developed in this zone, as there is dry
contact between tyre and road asperities. Clearly, by intimate contact, the molecular
interactions between rubber and pavements will lead to adhesive bonds, having high
shear strength.
In the rear region, the tread rubber and road surface are in dry contact. The drag of
water film is diminished, and replaced by the adhesion, where the surface now is in
molecular contact. The hysteresis loss will continue to occur due to the deformation of
tyre tread, as it is caused by the indentation of surface texture. Actually, even under
dry conditions, there is no such thing as a perfectly dry contact region between tyre
and pavement; fine water films contaminate all surfaces when the humidity is non
zero. Therefore, when reference is made to a dry contact, this implies a region in
which the fluid film thickness has been reduced to that thickness present on what is
casually and technically incorrectly termed dry pavement [Browne et.al (1982)].
4
If the road asperities are sharp, the tearing of rubber is possible, and the friction is
increased [Tabor (1958)]. This additional component is commonly referred to as
abrading friction. This component, together with other components mentioned above,
will be described in details in the following section.
1.1.1. Hydrodynamic Component of Friction
The hydrodynamic drag, buildup from forward region, can be divided into two
components: water drag and viscous drag [Trollope and Wallace (1962)]. The water
drag is derived primarily from the impact of the water on the leading edge of tyre.
This force is a function of film thickness, water viscosity, and impact speed. Although
the pressure developed by impact speed can exceed the tyre vertical pressure (i.e.
results in hydroplaning), the contribution (in tangential direction) to resist the tyre
longitudinal movements is likely to be of negligible magnitude.
The viscous drag is derived from the viscous resistance of the water which is
continuously sheared between tyre and pavement. This component can be calculated
from Newton's law of viscous shear. Roberts [1971] found that the viscosity of water
increased dramatically when the water film thickness was decreased below 500 A (5x10-6 cm). Hence, the thinner the film thickness, the greater the contribution of
viscous drag. Some theoretical calculations, however, showed this friction due to
viscous shear to be insignificant [Moore (1967), Ludema (1975)].
1.1.2. Adhesive Friction
The adhesion component of friction is due to the making and breaking of atomic
junctions between the rubber and pavement surfaces [Meyer (1968)]. There is general
agreement that adhesion has distinct viscoelastic properties. The coefficient of
adhesional friction is known to exhibit a peak value at creep speeds in about 1 cm/sec
[Ludema and Tabor (1966)]. This peak will shift to a lower speed with decreasing
ambient temperature, and to a higher speed with increasing ambient temperature. In
practice, therefore, this friction most often decreases as sliding speed increases.
5
In ideal case, the true area of contact of two atomically smooth surfaces would be
equal to the nominal · area of contact, and adhesive friction would be very high.
However, for the usual surface which is very rough on the atomic scale, contact would
be limited to the highest proturbances on the surfaces.
Even though the mechanism of adhesive friction is so far not fully understood, it is
well known that on well-lubricated surface the adhesion component may be greatly
reduced due to weakening of the shear strength of bond by the water film. As an
approximation, a film thickness of just exceeding the molecular dimension (i.e.
Angstroms), is probably enough to significantly reduce the direct bonding of tyre
rubber to the road material. An experiment by Roberts and Tabor [1968], using the
glass surface (moving at a uniform speed on the rubber hemisphere), showed that the
friction coefficient decrease sharply from 0. 7 to 0.07 by increasing film thickness from
80 A to 110 A, and then decrease gradually to 0.048 at the film thickness of 180 A.
1.1.3. Hysteretic Friction
On a textured surface, when a tyre is sliding or rolling, the tread rubber is subjected
to continous deformation by the asperities. The hysteresis force is a function of
damping energy loss in deformation (so called deformation loss). The rubber in front
of an asperity is compressed when it approaches the asperity and expands when the
rubber flows over the asperity. A certain amount of energy is required to compress the
rubber element in front the asperity but owing to damping losses only a part of this
energy which is stored in the rubber is recovered. The lost energy is converted into
heat. The loss caused by damping results in an incomplete recovery of rubber upon
expansion, thus creating unsymmetrical pressure distribution shown in Figure 1.2. The
net horizontal component of the pressure force is the hysteresis component which acts
in the opposite direction to sliding.
When the textured surface is perfectly lubricated the adhesion component
disappears, but the hysteresis component always exist as the asperities can deform the
rubber surface. The texture of the surface and the damping property of the tread
rubber, therefore, will determine the magnitude of hysteresis friction.
6
AOHESION DU'ORMATION TEARING WEAR
Fig.1.2. Components of rubber friction. [Kummer (1968)].
In general, the road surface is composed of a number of scales of texture ranging
from microscopic sizes to about 3/4 inch. According to Yandell [1974], it is the
scale's absolute slope (rather than the scale's size) which has a direct effect to the
hysteresis friction. The coefficient of hysteresis friction is the sum of the coefficients
generated on each scale. The size of scale, however, will influence the masking effect
of surface lubricant films, and the sliding speed of peak friction.
The damping properties of the rubber is the most important one for the hysteresis
friction, beside other characteristics: resilience, hardness, and elasticity. The increase
in hysteretic friction due to the higher damping properties of the rubber, is well
established by the fact that tyres made of synthetic rubber provide higher skid
resistance than that given by their natural counterparts [Maycock (1965-66), Savkoor
(1966)].
1.1.4. Abrading Friction
This component of friction is contributed by the fine scale texture with sharp
angular projections on the hard (stone) surfaces. The sharp asperities tear or wear
particles from the rubber. According to Schallamach [1954], this process is due to
tensile failure, induced by stress concentration in the rubber surface behind the track
asperities. It seems, however, that the heating of rubber (such as on braking or
cornering) · also participateS it) develop this process. The force exerted on the tyre
tread rubber to abrade it is the measure of this friction.
7
It was found experimentally by Yandell and Gopalan [1976] that the finer scales of
texture are responsible for greater amount of abrasion but participate less in generation
of hysteresis friction as the speed of sliding increases. Abrasion also increases as the
damping factor, the load and the sliding speed of the rubber increase [Yandell
(1971a)]. On well-lubricated surfaces, however, the abrading friction, similar to the
adhesive friction, is believed to be insignificant. Likewise, it could be argued that the
abrading component of friction might have already been accounted for under one of
the other components of friction [Ludema (1975)].
1.2. Factors Affecting Tyre-Road Friction
There are many of such factors, 47 have been listed by Holmes et.al [1972], but
some of these factors have been to have insignificant effect on the skid resistance.
Moyer [1959] identified 15 variables as the major factors contributing to the large
variation in the friction coefficient measured on various road surfaces. In general, the
variation in skid resistance can be attributed to the following groups, as mentioned
before: pavement/lubricant conditions, tyre, and operating conditions.
1.2.1. Pavement/Lubricant Conditions
(1) Road Surface Texture
Road surface texture refers to the distribution and the geometrical configuration of
the individual aggregates on the road surface. The texture is generally divided into two
components, namely: (a) the macro-texture, to refer to the large scale texture of the
pavement which represent the easily visible asperities, and (b) the micro-texture, to
refer to the fine scale texture on the surface of individual pieces of aggregate. In
quantitative study, however, it had been shown that road surface texture can be
divided into more than two scales [Yandell and Gopalan (1976)]. The function of the
texture is to provide a close contact with the tyre surf ace, to secure sufficient
deformation of the tread rubber, and to facilitate the removal of surface water
especially when the tyre surface is smooth.
8
The role of macro-texture in removing bulk water was demonstrated by Wallace
and Trollope [1969]. They found that as the texture depth increased the normal water
force underneath the tyre decreased resulting in the increase of tyre friction. An
experiment on wet surfaces by Schlosser [ 1977] showed that the influence of macro
texture is mostly prominent at high speeds. His results showed, even with worn tread,
that the locked wheel and sideway force coefficients remained almost constant up to
a speed of 100 km/h on the coarse macro-texture surfaces. It is well known that at high
speed, shorter time will be available for the water to drain away, thus the remaining
water film would be thicker if adequate drainage is not provided as with the cases of
smooth macro-texture surfaces.
The effect of micro-texture in reducing the thin water film (hence establishing dry
contact with the tyre surface) was quantitatively studied by Rohde [1976]. He found
that the time of descent of a tread element under constant load decreased significantly
when the micro roughness amplitude was increased. The shape of the micro-texture
also influenced the sinkage time, with the triangular texture pattern taking shorter time
than the square pattern to reach a given minimum film thickness. The increases in the
level of harshness of micro-texture, on the other hand, can cause an increased tyre
wear without a proportional increase in the tyre friction [Yandell and Gopalan (1976),
Lees et.al (1977)].
(2) Aggregate Characteristics
The characteristics of aggregate which influence the skid resistance are the shape,
grading, type, state of wear or polishing. The pressure distribution of the surface of
asperities pressed into tread rubber depends on the shape of asperities rather than their
size, provided that the macro-texture is coarse enough to drain off the surface water
[Tabor (1959)], its spacing is such that there is no texture saturation by the rubber
[Yandell ( 1970, 1971 b)], and visco elastic frequency effects do not intrude.
Laboratory study by Stephens and Goetz [ 1960] found that the mixture made from
finer gradation gave higher values of skid resistance. It is generally agreed that all
types of limestones are highly polishable and result in extremely slippery roads under
wet condition although they give high skid resistance when newly laid [Finney and
Brown (1959), Gray and Renninger (1965)]. Sandstone and quartzite aggregates were
found to posses the highest permanent skid resistance properties [Havens (1959)].
9
According to Lees [1984-85], aggregates which have a distinct difference in hardness
between primary minerals (such as quartz, augite, hornblende) and weathered
secondary minerals (such as kaolinite, chlorite, terpentine, sericite) will also produce
high skid resistance.
With all types of aggregates and methods of construction of roads it is possible to
get high skid resistance when the road surface is new. However, with age, the
aggregates under the action of heavy traffic tend to polish, resulting in lower skid
resistance. The polishing of the road stones is principally due to the continual attrition
of fine abrasive mineral particles found on the road surface caused by the traffic
movements, and it is seen that the finer the detritus, the greater will be the degree of
polishing [Maclean and Shergold (1960)]. Stiffler [1969] concluded that the abrasives
of size less than 10 micron diameter will cause polishing.
(3) Seasonal Effects on Surface Texture
Giles and Sabey [ 1959] reported a marked difference in test results on wet surf aces
for winter and summer. The coefficient of friction is higher in winter than summer.
The above report reveals that the road surf ace was found to be covered with fine dust
during summer. The dust polished the roadstone thus resulting in a lower friction
coefficient. In winter (when the road was wet for 60% of the time), the dust particles
were quickly washed off the tyre and the abrasion patterns tended to disappear. The
polished stones got roughened by weathering in the presence of water.
( 4) Amount of Water
The effect of water film on friction has been found to be the predominant factor
compared to the speed and temperature dependent viscoelastic effect [Clamroth and
Heidemann (1968), Lupton and Williams (1972)]. Giles [1959] and Besse [1972]
found that the friction continue to decrease as the water film increase, until the film
thickness of about 0.02 in (0.508 mm) where the friction tends to level off.
(5) Temperature
It is well known that with increasing temperature (air, surface, tyre tread), the skid
resistance tends to decrease [Giles and Sabey (1959), Grosh and Maycock (1968),
Meyer and Kummer (1969), Meyer et.al (1974)]. Both Giles et.al and Grosh et.al
10
measured an increase in rubber resilience with increased temperature of rubber which
indicates a decrease in hysteresis losses of rubber.
The effect of water temperature was studied by Meyer et.al [1974]. They measured
skid resistance on several pavements with water temperatures of 140° F (6C.f C) and
60° F (15.5° C) and found that the difference was about 1 skid number (0.01 locked
wheel braking force coefficient), the higher temperature gave the lower values. Hence,
it is seen that the temperature dependence of friction is through its effect on rubber
properties rather than the rate of change of water removal caused by viscosity of the
water (in which case the friction would increase and not decrease at higher
temperature).
It is found that the temperature dependence of wet friction is dependent on the
texture of the road surface. On wet coarse textured road surfaces the decrease in
friction is more than that on wet fine textured road surfaces [Giles and Sabey (1959),
Maclean and Shergold (1960)]. The possible explanation is that on the coarse texture,
the temperature will be higher than for the smooth surface due to the greater contact
stresses (as caused by lesser actual contact area), and due to the break down of the
carbon filler [Yandell et.al (1983)].
1.2.2. Tyre Factor
The choice of tyre type (cross ply, radial) for use in a vehicle can have a large
influence on skidding characteristics. The variability in wet skid friction has been
investigated by Allbert and Walker [1965-66], for example up to 4:1 for tread pattern
design, and up to 1.8: 1 for changes in tyre compound. To fulfil its function the tyre
(tread) compound, as being reflected by its physical properties (such as: resilience,
hardness, elasticity, and damping), must meet a number of different tyre requirements.
A compromise in many properties sometimes must be sought, so that a tyre with high
skid resistance as well as high abrasive resistance, can be obtain simultaneously
[Peterson et.al (1974)]. In the following, those most important factors: tyre type, tread
pattern, damping property, and hardness, will be described in details.
11
A .IIAS ,l Y • . IADIAl IElTED C - IIAS HlTED
Fig.1.3. Basic tyre structures. [Davisson (1969)].
(1) Tyre Type
The three basic types of tyres in use nowday are the bias-ply (sometimes called the
cross-ply or conventional), the radial-ply (belted) tyre, and the bias-belted tyre. These
three types are shown in Figure 1.3. Within each of the three basic tyre types, many
variations are possible such as: low or high aspect ratio (i.e. ratio of the section high
to the section width), and tube or tubeless type.
Radial-belted tyres although have lateral spring rates considerably lower than
bias-ply tyres, but usually have higher cornering stiffness properties [Davisson
(1969)]. In other words, for a given slip angle, the lateral force developed in
radial-belted is greater than bias-ply. According to Davisson [1969], belted tyres have
less tread movement in the contact area, and exhibit a slower rate of wear than bias
tyres.
Results from DeVinney [1967] on hydroplaning tests show that radial-belted tyres
have greater coefficient of friction. He explain that the "belt" gives a rigidness to the
tread which serves to keep the grooves open as the tyre rolls through the contact
patch, providing greater water "drainage".
Tests on cornering coefficient by Sakai et.al [1978] found that when the water film
is thin (1 mm) radial-ply tyres are considerably superior to cross-ply tyres. In contrast,
when the water film becomes thick (5 mm) the radial-ply tyres are superior in the low
speed region but inferior in the high speed region.
12
(2) Tread Pattern
From the point of view of removal of water from the ground tyre contact, the tread
pattern and the road surface macro-texture function in a somewhat similar manner.
The tread pattern provides outlet channels for the surf ace water to escape from the
contact area, under the squeezing action of the tyre. By this reasoning, the tread
pattern should have a greater effect on smooth textured surfaces where drainage is
generally poor. Conversely, tread patterns are least effective on rough open textured
surface. There is also a contrast under dry and wet conditions, where best dry skid
resistance is obtained when a tyre has no tread design at all. However, tread design
plays a vital role in wet friction where accidents due to friction mostly occur on wet
surfaces in~tead of on dry surfaces.
The simplest and the most common tread pattern is the provision of longitudinal
grooves on the tread surface which, has shown to increase the friction on wet surfaces
by 20 to 100 percent depending upon the initial value of the friction [Marick (1959)].
The skid resistance was also found to increase with the number of grooves and groove
widths [Allbert and Walker (1965-66), Maycock (1965-66)]. This was attributed to the
greater groove volume in relation to the total drainage area, to the slightly higher
pressure acting in the contact zone as the groove area increases and probably to the
shorter travel paths of water to reach the drainage channels [Maycock (1965-66)].
On the other hand, Kienle [1974] found that the effectiveness of widening grooves
is asymptotic, and there is an optimum tread width for a given number of ribs. A
similar result is obtained by Veith [1977], using the "fractional groove volume" 0
which he defined as the ratio of groove volume to the total tread volume, as a measure
of a tread pattern effect. He found an exponential relation between the locked wheel
coefficient of friction and 0, where the tread pattern effect is seen to reach a limit at 0
of about 0.4 (beyond which the coefficient remain unchanged with increased 0).
In addition to longitudinal grooves, the lateral edges on the tread surface have been
found to provide wiping action over the wet surface. On some rough surfaces, it was
seen that a suction of 10 psi (7 g/mm2) developed behind the transverse grooves
[Wallace and Trollope (1969)]. This agreed with Gough's observation that sharp
transverse edges of the tyre tread increase the skid friction by wiping water from the
13
road surface [Gough (1954-55)]. Slits in the tyre function as pockets to absorb local
lubricant pressure and promote dry contact [Allbert and Walker (1965-66)].
Since the fall of friction coefficient with speed is associated with the greater
difficulty the tyre has in displacing the water beneath it, it is evident that tread pattern
proves more effective at high speeds. Maycock [ 1968] found that the tread pattern,
even of the simplest design such as straight ribs, gave on the smooth surface a large
improvement in both peak and locked wheel braking force coefficients with the speed
range of 30 mph to 60 mph. At low speeds the effects of tread pattern are not
significant especially on coarse textured road surfaces.
(3) Damping Property
The damping factor is usually measured by allowing a weighted pendulum to strike
a rubber sample from a given height. The elasticity is the percentage of potential
energy regained at the first rebound. Therefore a rebound to the original height would
indicate 100 percent elasticity, that is no damping loss. The energy lost in the process
is called damping energy and is dissipated in the rubber as heat. Yandell [1970]
defined the damping factor as the energy lost divided by the energy applied during one
complete loading-unloading cycle. Rubber with high damping losses (low resilience)
will give high values of friction coefficient [Sabey et.al (1970)].
As rubber is partly viscoelastic material, its damping losses are influenced by
temperature and frequency of loading. The damping decreases as temperature
increases, and increases with increasing frequency. With increasing temperature, peak
damping for a given rubber compound decreases and shifts towards higher frequencies
[Kummer and Meyer (1962)]. Furthermore, it was found that the damping for synthetic
rubber is always considerably greater than for natural rubber. The superiority of
natural rubber, however, is for the low heat buildup characteristic [Davisson (1969)].
( 4) Hardness
The hardness is usually measured by Shore Durometer [Bashore (1937)]. Rubber
hardness has been less definitely correlated with friction. There are reports in the
literature that increasing rubber hardness increases friction [Marwick and Starks
(1941), Grime and Giles (1954-55), Giles et.al (1962), Goodwin and Whitehurst
(1962), Sarbach et.al (1965)], or has no effect [Sabey and Lupton (1964), Bassi
14
(1965)]. The general explanation for the increased friction is that the harder rubber
gives a smaller contact area, resulting in higher contact pressure, better drainage and
hence increased friction [Csathy et.al (1968)].
Carr [1967] reported that the relative influence of rubber hardness depends on the
texture on which it slides. Report from Percarpio and Bevilacqua [ 1968] found that on
slippery road surfaces, the increased hardness conversely leads to decreasing friction.
Their report then suggest that friction increases with hardness only on highly abrasive
surfaces.
1.2.3. Operating Conditions
(1) Speed
Numerous tests have been carried out to study the effect of speed on friction
Fig. 1.8. Peak and slide traction envelopes of 178-15 tyre in braking-in-a-tum. [Bergman (1977a)].
21
1.3. Contact and Slip Between Tyre and Roadway
For a wheel to produce friction forces, either braking or sideway, there must be
some reduction in wheel speed. A wheel is said to be slipping if it circumferential
velocity is different from its travelling velocity. The friction and normal forces needed
for vehicle support, guidance and maneuvers arise in the tyre contact area. The normal
pressure distribution, for example, does not suddenly rise at the edge of the contact
area from zero to a finite value, but rather increases, gradually or suddenly at a finite
rate. The low normal pressure at the edge of contact patch (hence small adhesion
limits) perhaps leads to a conclusion derived by Novopol'skii and Tret'yakov [1963],
that even when the tyre is rolling, there are slip areas where contact begins and ends.
In the following, the characteristics of tyre motions, their contact area, and the stress
distribution will be described in detail.
1.3.1. Classifications of Tyre Motions
According to Pacejka [ 1966] the motions of a tyre can be distinguished into two
main groups, namely steady-state (stationary) and non steady-state motions. The first
group is associated with constant slip velocity in the rolling direction and
perpendicular to it. For example: a. Tyre rolling with constant braking slip or at
constant slip angle. The second group have a character varying with time for lateral
slip and angular motions. For example [Hegmon (1982)]:
bl. Tyre rolling with fast sweep of slip angle (- 10°/second) at constant braking slip.
b2. Fast braking to wheel lock (- 0.5 second) at constant slip angle.
Another classification is made by including a third group between those two
groups, representing a small time interval of the steady-state behaviour [Uffelmann
(1983)], so called quasi steady-state motions. For example [Hegmon (1982)]:
c 1. Tyre rolling with slow sweep of slip angle (- 3° /second) at constant braking slip.
c2. Slow braking to wheel lock (- 5 second) at constant slip angle.
Most operating conditions can be regarded as the steady-state and quasi steady-state
motions when the automobile is cruising in a normal manner. Under emergency or
panic stops, the non steady-state motion often occurs.
22
1.3.2. Types of Slip
Slip has been defined as the ratio of the difference between the angular wheel
velocity when the tyre is rolling or slipping to the angular wheel velocity when the
tyre is rolling [Kummer and Meyer (1967)]. Slip is commonly expressed as a
percentage by multiplying by 100. It is convenient to define slip as a vectorial quantity
S thus:
v-V S° = (-) X 100 V
where v is the travelling velocity, V is the circumferential velocity, and v is the
magnitude of v.
For a pure circumferential slip, the slip in percentage is given by: v-V S = (-) X 100
V
When braking (v > V), S is positive with the maximum value of 100% for a locked
wheel (V = 0). The maximum braking force is obtained at a slip between 7%-30%
[Giles (1964), Holmes and Stone (1969), Schallamach and Grosch (1982)]. During
acceleration (v < V) is negative and becomes negative infinity when the stationary
wheel spins [Schallamach and Grosch (1982)].
When the wheel rolls in a direction making the slip angle a with its plane, the
situation is one of pure side slip, and the slip is given by:
S = sina x 100
There is a critical value of a where the sideway force developed is a maximum, and
beyond this angle the force begins to decrease. This value is around 5° - 20" [Bradley
and Allen (1930), Holmes and Stone (1969), Veith (1971), Shah and Henry (1978)].
The mechanism of slip produced by braking/driving and cornering was illustrated
by model experiments by Schallamach and Turner [1960], and Schallamach and
Grosch [1982]. A wheel of 2.5 inch diameter and 0.5 inch thick was made to travel
over a Perspex track under different conditions of slip. The wheel surface was marked
with lateral equidistant markings, the orientation of these markings on the slipping
tyre indicated the tangential stress distribution in the contact area.
23
(a)
(b)
(c}
(d)
Fig.1.9. Model experiments for: (a) free rolling, (b) side slip, (c) braking, (d) driving. Travelling direction from right to left. [Schallamach and Grosch (1982)].
The strains set up in the contact area of a slipping wheel are showed in Figure 1.9.
All cases have in common that a circumferential element of the wheel on entering the
contact area adheres to the track at first. As the element moves further into the contact
area, the imposed slip produces a deflection which increases linearly with increasing
distance from the front edge. This is clearly seen in Figure 1.9b, where the deflection
is at right angles to the wheel plane. The corresponding surface stress increases in the
same sense until a limiting frictional stress is reached and the element begins to slide
back towards its undeformed position.
24
A braking force (Figure 1.9c), lengthens an element in the circumferential direction
before entering the contact area, and the element adheres at first to the track in this
state of strain. The deflection of the wheel increases linearly with increasing distance
from the front edge, until sliding commences towards the front of the contact when the
tangential stress exceeds the limiting frictional stress. Under a driving torque,
contraction of an element takes place before entering~ontact region (Figure 1.9d).
1.3.3. Sliding and Skidding
As mentioned above, when appreciable (or even quite small) sideway or braking
forces are applied, sliding begins at the rear of the contact surface in which the
available friction is exceeded. According to Kamm [1938], there is appreciable sliding
at the rear of the contact surface even in straight unbraked rolling, at high speeds. In
the general case, the contact surface can thus be divided into adhesion and sliding
regions, an adhesion region being defined as one in which there is no relative motion
between coincident points of the tyre and ground surfaces.
In extreme cases of course the relative motion is one of pure sliding, for which the
term "skidding" may be reserved. In vehicle scale, skidding as defined by Stonex
[ 1959] is the motion of a vehicle under partial or complete loss of control caused by
(pure) sliding of one or more wheels. Transition between partial sliding and skidding
of a tyre is of particular importance since it corresponds to total loss of tyre directional
stability. Schuster and Weichsler [1935] point out that as long as there is even partial
adhesion, the tyre may be said to have a "virtual guiding rail" as a result of frictional
forces (this "rail" is an elastic one).
The sliding or skidding potential of a road surface is commonly measured by
locked wheel or side slip modes of operation of skid testing machines. The locked
wheel test simulates the longitudinal sliding conditions of a braked vehicle, whereas
the sideway test simulates sliding condition when a vehicle is cornering.
25
8:00 x 14 Automotive Tire Four ply Royon Bloa Ply Construction 24 psi inflation 1.25" Deflection 1350 lbs lood - ~---~ -- - - - - - - -
Fig.1.10. Typical contact area for an automobile tyre. [Browne et.al (1982)].
1.3.4. Contact Area
The shape of the contact area between tyre and roadway depends on the tyre cross
section shape and structure. For example, the contact area between an aircraft tyre and
a flat surface usually appears to be nearly elliptical in shape. For an automotive tyre a
somewhat different set of relationships exists due to the fact that the usual
construction involves the use of a relatively heavy tread, particularly in the shoulder
region. In this case any significant contact spread over the entire width of the tyre
between shoulders so that the contact area tends to have essentially straight parallel
sides, and the width of this contact area is nominally independent of tyre deflection. A
typical contact area is shown in Figure 1.10.
Experimental evidence indicates that tyre deflection is the most important variable
governing the area of contact. Results from Hadekel [1952] show that the relationship
between tyre deflection and gross contact area is nearly linear. If inflation pressure
and load are simultaneously varied so as to maintain constant tyre deflection, the tyre
contact area will remain effectively constant [Browne et.al (1982)]. Using a single
truck tyre,anexperiment by the U.S. Army Engineers Waterway Experiment Station
[1964] showed that in general the contact area of a slowly rolling tyre is slightly lower
than of a standing tyre.
When fluid contaminants such as water, oil, slush, or mud are present on the
pavement surface, (dry) contact area between tyre and pavement is reduced due to the
persistence of a fluid film in portions of the formerly dry contact area [Horne and
26
Dreher (1963), Yeager (1974)]. Just how much the dry contact area will remain
depends on tyre, pavement, fluid, and vehicle factors. According to Browne et.al
[1982], the fractional dry contact area is reduced, with increasing of: speed, fluid
(viscosity, density, depth), size of tread elements; and with decreasing of: inflation
pressure, carcass rigidity, width of grooves (number held constant), number of grooves
(width held constant), and groove depth (width held constant).
1.3.5. Stress Distribution
At the interface between the tyre and the roadway an element of tyre surface area is
acted upon by a force vector which can be expressed as two components, one
perpendicular to the contact surface, called the normal component, and one tangential
to the contact surface. This latter component may be further decomposed into two
components, each lying in the contact plane, but one parallel to the central plane of the
tyre and the other perpendicular to it. These components in the contact plane are
commonly called the shear components. With equal validity they could be expressed
as components parallel and perpendicular to the direction of travel of the wheel. Either
decomposition would be useful for describing the shear effects.
(1) Normal Contact Stresses
Contact pressures between tyre and ground are not necessarily uniform, mainly
because of tyre detail design and partly because of road surface irregularities.
Measurements of contact pressure confirm that since the contact pressure are about
50% greater than the inflation pressure, the casing stiffness plays a part in the
load-carrying, otherwise an infinitely flexible casing would result in contact pressures
equal to inflation pressure [Gough (1958-59)].
Modem instrumentation techniques have allowed measurement of normal pressures
under a variety of conditions, using many different tyre types. Some of this data has
been reviewed briefly by Hadekel, with particular emphasis on the early work by
Martin [1936] and Markwick and Starks [1941] for road vehicle tyres, and by Teller
and Buchanan [1937] and Kraft [1941] for aircraft tyres. The following facts emerge
from the above data [Hadekel (1952)].
27
N E ~ 0 r----.--,----,--"T'-"T""T---.r, .i' 2 t\----t------lf-------'-------l- ............_ ' w 4 a . b...-! a: I ;-}! ::, 6 _________________ ..
a 8t-----+----+--+--+--'---,........! flOO~~-,-~~,........,~_.__....,._,
50 100 150 . 200 250 300 mm LEADING EDGE LENGTH nm. TRAIUNG EDGE
Fig.1.11. Pressure distribution down the length of a tyre with no tread pattern, at constant speed 15 kph, inflation 6.5 atm, load: (a) 1680 kp, (b) 3740 kp. [Bode (1962)].
60 40 20 20 40 60 80 100 mm POSITION ACROSS WIDTH
Fig.1.12. Pressure distribution across the width of a tyre with no tread pattern, constant speed 15 kph, load: (a,b,c) 1680 kp, (d) 3740 kp; inflation: (a) 3.5 atm, (b) 5 atm, (c,d) 6.5 atm. [Bode (1962)].
- 4 .., ~6i-::~~~-ri-~-~+=~~~~-~ en ~8t--+--+---+---+--+------l f .
100 50 100 150 200 250 300mm POSITION IN LENGTH OF CONTACT PATCH
Fig.1.13. Pressure distribution down the length of a smooth tyre, load 1680 kp, inflation 6.5 atm, speed: (a) 10.2 kph, (b) 8.4 kph, (c) 8.9 kph; acceleration: (a) 0.6 m/s2, (b) 1.6 m/s2, (c) 4.1 m/s2• [Bode (1962)].
LEADING EDGE N 0....--..--..--..--........,,.,......,---, e ~ 2 A .. 4 i,j
~ 6 en :a1--+-"---------~-----1 a: ~,00..__...,_-------,,-,--300mm
Fig.1.14. Pressure distribution down the length of a smooth tyre, load 1680 kp, inflation 6.5 atm, speed: (a) 10 kph, (b) 35.7 kph, (c) 14.5 kph; deceleration: (a) 4.1 m/s2, (b) 2.7 m/s2, (c) 2 m/s2• [Bode (1962)].
28
Fig.1.15. Pressure distribution of bias-belted H size tyre, 26 psi, 0° slip angle, 100% T and R load = 1580 lb. [Lippmann and Oblizajek (1974)].
a. There is (as might be expected) a tendency for the pressure to rise from the centre
outwards, up to some point at some distance from the edges, but more particularly
so in transverse direction, the zones of highest pressure being near the sides of the
contact surface. According to Martin, near the centre the stress may be actually less
than the tyre air pres$ure.
b. There is a fairly gradual drop to zero at the edges.
Detailed measurements on truck tyres have been reported by Bode (1962], who
reports data such as shown in Figures 1.11-1.14. Additional data has been reported by
Lippmann and Oblizajek [1974] on car tyres. One set of pressure distribution from
their work is illustrated in Figure 1.15. They found (as expected from Figure 1.9b) that
by introducing steering, on one side of the tyre the length increases, and decreases by
a similar amount on the other side of the tyre.
A number of investigations have examined the role of velocity in modifying the
distributions, including the work of Bode as well as the work of 2.akaharov and
Novopol'skii (1957]. In general, the results seem to show that increasing speed causes
an increasing vertical contact pressure at the forward end of the patch and a decreasing
value at the rear portion of the contact patch.
29
Pos.111on .along contilct length.
Fig.1.16. Longitudinal shear stress distribution in the running direction of a tyre under free rolling. [Novopol'skii and Nepomnyashchii (1967)].
RIB NUMBER
~ : o~,---~ DISTANCE INTO ~ -20 PATCH, inches
Fig.1.17. Longitudinal shear stress distribution in the running direction of a biasbelted H sire tyre under traction, 26 psi, O" slip angle, 100% T and R load = 1580 lb. [Lippmann and Oblizajek (1974)].
(2) Tangential Stresses
According to Browne et.al [1982], the direction and magnitude of the longitudinal
tangential stresses of a driving or braked wheel are determined by the sum of the
stresses created in the rolling of a free wheel and the additional stresses created by the
application of a torque. In driving, immediately before the contact the tread elements
are contracted, and they are stretched at the point at which contact is released, as a
result the shear is opposing travel direction. In braking, the situation is reserved, and
the shear is in the direction of travel
30
Acceleration a: 0.96 m/s1
b: l.28m/s1
c: 1.53 m/s1
d: 2.02 m/sl e: 3.22 m/s1
f: 5.66 m/s1
Velocity a: V:4l.7km/h b: V= 17.0 km/h c V•25.8km/h d: V = 17. 7 km/h e: V = 7.4 km/h f: V= 8.9 km/h
Fig.1.18. Longitudinal shear stress distribution in the running direction of a smooth tyre under traction, load 1590 kp, inflation 6.5 atm. [Bode (1962)].
t - .
deceleration a: -0.4 m/sZ b: - 1.23 m/s2
c: - 1.5 m/s2
d. - 2. 45 m/s2
e. - 2. 78 m/s2
f: -4.3 m/s2
Velocity a: V = 17.6 km/h b: V = 17. 7 km I h c: V= 16.9 km/h d: V = 15.4 km/h e: V • 16.9 km/h f: V = 17.2 km/h
Fig.1.19. Longitudinal shear stress distribution in the running direction of a smooth tyre under braking force, load 2610 kp, inflation 6.5 atm. [Bode (1962)].
The form of longitudinal shear stress in a free rolling tyre without steer angle has
been studied, such as by Novopol'skii and Nepomnyaschii [1967] and Lippmann and
Oblizajek [1974]. Their results are shown in Figures 1.16-1.17. Similar results are
obtained by Markwick and Starks [1941], who state that the distribution on a
longitudinal section is somewhat similar for a standing tyre, as it is caused by the
bending of tyre (carcass) with consequent symmetric strain distribution.
Extensive experimental work on the longitudinal component of shear stress has
been published by Bode [1962], such as shown in Figures 1.18-1.19. It can be seen
that the influence of braking or tractive forces is to throw the major part of this
longitudinal component to the rear of the contact patch. Even moderate values of such
forces are sufficient to completely change the stress distribution.
31
0.1 0.2 0.3 0.4 0.5 0.6 0.7 08 0.9 ID I.I 1.2 SIDE SLIP (inchn)
Fig.1.20. Lateral force intensity along the contact spot, at slip angles up to 12°. [Browne et.al. (1982)].
If a tyre is caused to move at a slip angle, that is in a direction different from the
straight ahead, it will develop a ground frictional force resisting that slip. A pneumatic
tyre adheres to the road in the front part of the contact, where it accommodates the
slip angle by lateral deflection. Sliding is confined tQ the rear part of the contact,
where the elastic forces overcome friction. For this reason, the contact area can then
be separated into two regions. As stated by Browne et.al [1982], the first region is one
of essentially static contact between tyre and the road swface, but nevertheless a
region in which secondary slip may exist; whereas the second region, adjoining the
first, is the region of primary slip where the tangential swface stresses exceed the
local frictional stresses available. In the first (adhesion) region, since the stresses in
the contact are simply related to the deflection, it is expected that the lateral stress is
gradually increased until it meets the second (sliding) region, where after that the lateral
stress is decreased following the decrease in the available friction stresses. A typical
of lateral (cornering) force intensity is given in Figure 1.20
Tyres are often required to transmit lateral and longitudinal forces to the ground at
the same time, a common instance being driving wheels in a curve. Simultaneous
lateral and longitudinal slips alter the relation of the lateral and longitudinal forces
with their own slip. This mutual interference is pronounced at large slips when a
substantial part of the contact is sliding because the total available force p must be
shared by the two forces. A simplified example of stress distribution due to
simultaneous slips is shown in Figure 1.21 [Schallamach and Grosch (1982)].
32
X X
a
Totol Traction Lateral Traction
X X (bl
Fig.1.21. Lateral force intensity: (a) due to slip (t.), and (b) due to simultaneous side slip and circumferential slip (fe). [Schallamach and Grosch (1982)].
1.4. Surface Texture Description and Measurement
1.4.1. Description of Surface Texture
The skid resistance which pavement material can provide is strongly dependent on
the textural characteristics of the surface. The term "surface texture" in this context
does not mean the composition of the surface (so much binder, sand, etc), but the
geometrical form of the road surface. The complex and random nature of a road
surface makes it difficult to represent the surface characteristics by a general single
parameter. Posey [1946], for example, suggested that 3 parameters for a representative
length of profile (i.e. histograms of the profile itself and of its slope and curvature),
give sufficient information to permit a complex characterization of the texture;
whereas Moore [1965] attempted to quantify the "feel" of a texture by expressing its
geometrical features as size, spacing and shape factors. In addition, Myers [1962]
listed a new series of single elementary parameters to define texture, but carefully
indicated that depending on the particular application each of the new parameters
might be considered more useful than the others. For example, the RMS (root mean
33
square) of the second derivative of the profile (i.e. its degree of curvature at peaks, or
sharpness) would be most appropriate to determine the degree of wear which a surface
has undergone. In the following, the parameters which are commonly used for
characterization of surf ace texture, will be described in details.
(1) Texture Depth
There is a general trend toward increasing friction coefficient and skid numbers
when deeper surface textures are encountered. However, the typical scatter of the data
about the line of best fit would be of questionable value. Similar results are also
obtained when the texture depth is correlated with speed gradients.
The lack of a definitive relationship between texture depth and both skid number
and speed gradient was attributed, in part, to the possibility of poor test repeatability
[Doty (197 4)]. On the other hand, Lees and Katekhda [197 4] stated that there is no
justification in relating the average texture depth to the drop of friction with speed,
since unconnected voids in the surface are included in the measurement while they
play no part in dissipating the water between the tyre and the road surface. They
argued that disconnected voids act on surface as reservoirs retaining water which aids
the lubrications of nearby particles and prevents full deformation of the tyre into the
road texture.
The average depth, however, is useful for broad classification of surface texture. In
France, for example, texture depth (ID) is classified as: very fine (ID :s; 0.2 mm), fine
(0.2 mm < ID :s; 0.4 mm), medium (0.4 mm < ID :s; 0.8 mm), coarse (0.8 mm < ID :s;
1.2 mm), and very coarse (ID > 1.2 mm). Pavements with very fine-textured are to be
prohibited, while very coarse-textured are used in special cases: danger zones
following a straight line or frequent frost zones [Elsenaar et.al (1977)]. The other
versions of average depth, described in the next section, are shown in Figure 1.22.
(2) Centre Line Average Height (CLA)
The CLA is the measure used in BS 1134: 1972 to specify the fineness of the
surface finish of machined pieces. It is defined as the average value of departure of
the profile from its centre line, whether above or below it. The centre line is defined as
a line conforming to the prescribed geometric shape of the profile and parallel to the
general direction of the profile throughout the sampling length such that:
34
Fig.1.22. Mathematical definition for numerical evaluation of surface texture. [Marian (1962)].
r ydx (y~O) + r ydx (y<O) = o
The Cl.A then is given as:
hCLA = ~f ydx
where y = f(x) is the equation of the profile (see Figure 1.22b).
(3) The Average Depth
The average depth is used (in Europe) as an equivalent of Cl.A in Britain.
Referring to Figure 1.22c the average depth from the crest line is given by:
hAVR = ~f ydx
where y = f(x) is the equation of the profile.
( 4) Root Mean Square
The root mean square height of a profile is given by (see Figure 1.22b):
h_ = j ~fy'dx
This measure accentuates the effect of sharpness and distinguishes between rounded
and sawtooth type textures.
35
Line DI peaks i----.,o~-....... ---#'~--.... -.-= ....... -_:--""-<-~....,_-----,_,...~..----+1s,. ot points -- llnel lf'--___,;iiif,--~1--f--;;.._~ _ _,_ _ _;_.......;:~-,..~~+,-~-+Line C11 depth D bel- peaks
b
Profile ratio • total 11119111 DI prDlile o ..,.glll of base lilw b
·Filtered profile ratoo • w dotllh D n 9o = ...,. of letlpths of profile - deplll D • nf••
,._ of baseline ,_,,..
It,,. 1
Fig.1.23. Profile ratio and "filtered profile ratio". [Sabey (1968)).
8~ s2 s, Reference plane
Fig.1.24. Moore's bearing area method [Moore (1975)).
(5) The Swedish Standard
The Swedish standard is numerically greater than the CLA height, average depth or
root mean square height. The two lines x and y are positioned so that lJ> = K1L and
I.V = KiL (Figure 1.22a). The Swedish standard specifies that K1 = 0.05 and K2 = 0.10. The distance between the parallel line x and y is the measure of surface
roughness.
( 6) Profile Ratio
The measure is the ratio of the profile length to the projected length. This bears
some relation to texture depth and takes into account the shape of the profile [Sabey
(1968)), but does not give any account of the three-dimensional aspect of the surface
texture [Lees and Katekhda (1974)). In addition, the "filtered profile ratio" which is
the profile ratio for the tops of the asperities only, over a different depth D, below the
line of the peaks can be evaluated (see Figure 1.23).
36
(7) Moore 's Bearing Area Method
Moore [1975] used a mathematical expression to represent a single profile. A series
of fiction parallel planes are drawn successively below the reference plane at
distance 61, 6z and so on, so that they intersect different number of asperities N1 (see
Figure 1.24). Two equations may be written:
Ni= Co 6m
Ai= C1 + ½ 6D
where Ni is the number of asperities intersected by the ith plane located at distance
from the reference plane,
Ai is the total contoured area of asperity intersected by this plane,
Ci is the plateau area,
C0, m are constants specifying the height spacing or statistical distribution of the
asperities,
<;, n are constants specifying the mean shape of asperities.
He found, for example, that on concrete road surface: m = 2, n = 3.
(8) Mean Width of Surface Voids
The parameter can be obtained from the asperity density prints. It was found to
correlate with the steepness of the friction speed curve, the closer the surface the
steeper the negative sloped curve [Schulze and Bechmann (1962)].
(9) Mean Hydraulic Radius (MHR)
Moore [1966] measured the drainage capacity of a surface in terms of the mean
hydraulic radius:
MHR = M Vv /tN°·5
where M is the instrument constant,
vis viscosity of water,
t is the recorded time for a fixed volume of water to drain,
N is the number of asperities per square inch of surface texture.
He then used the MHR to predict the wet sliding coefficient of road surf ace.
37
b
m
h
e, f, g & h lie on plane of best fit rtrJ MAXlMUM SLOPE : Angle which triangular planes
a b d , bed ctc make with plane efgh
ABSOWTE SLOPE in DIRECTION of SLIDING :
ldhehael
AREA of SURFACE over· PROJECTED AREA:
abd+ bed efgh
a.
n 0 p
n,o,p & q lie on plane of best fit
CURVATURE: ko-~
no
SIZE of INFLECTION:
ko--~ _Ip-~ no op
b.
Fig.1.25. Definition of surface characteristics. [Yandell (1969)].
Both profiles have similar average absolute Slopes and hysteretic sliding resistances
The lengll'lof drainage paths d & D vary inversely with the r.imb•rs of inflections pH unit length
a.
-- --
Both profiles have similar average absolute slopes
Th• lower profile has fewer inflections per unit length and a lower theoretical hysteret,c lnct,on
b.
Fig.1.26. Influence of the number of inflections. [Yandell (1969)].
( 10). Parameter used by Yandell
Yandell [1969] represented the pertinent characteristics of the
road surface texture by the following parameters:
a. Texture depth.
b. Maximum slope.
38
q
c. Average absolute slope in the direction of sliding.
d. Surface area per unit projected area.
e. Distribution of surf ace.
f. Curvature.
g. Number of inflection.
These parameters can be used for many purposes, such as on the calculation of the
hysteresis friction, the polishing of road stones, and the abrasion of rubber. Some of
them are illustrated in Figures 1.25-1.26. One important parameter, the average
absolute slope, is found to have good correlation with coefficient of hysteresis friction.
It is used with the damping factor of rubber in the mechano-lattice analysis to predict
the coefficient of hysteretic friction.
1.4.2. The Measurement of Surface Texture
The road surface texture is normally categorized by two features, the large scale or
macro-texture which represents the easily visible asperities in the surface and the fine
scale or micro-texture which describes the harshness or state of polish of the stone
surface. (See Figure 1.27).
The method of measuring the surface texture largely depends on the application in
question. Over 20 methods have been used for measurement of surf ace texture, and it
is interesting to note that tests using these methods, have not, on the whole, produced
consistent or reproducible quantitative results [Rose et.al (1973)]. Some of the
important and commonly used methods will be described below; which can be broadly
classified into four categories [Taneerananon (1981)].
a. Volumetric: Sand patch, Grease smear, and Silicone putty.
b. Profile: Texturemeter, Row of needles, Stylus, and Profilograph.
c. Photography: Stereo-photogrammetry, and Stereo-photo interpretation.
d. Miscellaneous: Outflow meter, Surface prints, Laser beam, and Texture friction
meter.
( 1) Sand Patch
It involves spreading a known volume of sand over a circular area until flush with
the tips of the asperities. Average texture depth, the ratio of volume to area, is the
39
road surface macro micro
rough harsh
rough polished
smooth harsh
smooth polished
Fig.1.27. Illustration of terms of the road surface texture. [Sabey et.al (1970), Schlosser (1977)].
measure of surface texture. It is rather simple test, but it evaluates macrotexture only
(the size of the sand used prevents the very fine channels from being measured).
(2) Grease Smear
The principle is same with the sand patch but the grease is applied instead of sand.
The method is used usually for obtaining the texture depth of fine textured road
surfaces, as it has better ability than the sand to fill the narrow channel of the surf ace
[Smith and Fuller (1969)].
(3) Silicone Putty
It is similar in principle to the sand patch and grease smear method. A known
volume of silicone putty is formed into an approximate sphere and placed on the road
surf ace. A recess in a plate is centered over the putty, and the plate is pressed down in
firm contact with the surface. Average diameter of the deformed putty is recorded.
When tested on a smooth flat surface with no texture, the silicone putty will
completely fill the recess. Therefore, a decrease in the measured diameter indicates an
increase in texture depth. This procedure was used by researchers (at Texas
Transportation Institute) to evaluate macrotexture [Rose et.al (1973)].
40
( 4) Texturemeter
The instrument, developed at Texas Transportation Institute, consists mainly of a
series of evenly spaced, vertical, parallel rods mounted in a frame [Rose et.al (1973)].
All except two rods can be moved vertically against spring pressure and independently
of one another. One rod at each end of the device is fixed to the frame for support.
Each movable rod has a hole through which a taut string is passed. One end of the
string is fixed to the frame, and the other is tied to the spring loaded stem of a 0.001
inch dial gauge extensiometer mounted on the frame. When the frame is pressed onto
the road surface, any irregularities in the surface will cause the string to form a
zig-zag line which will give a dial reading. On smooth surface the reading is zero,
therefore the coarser the surf ace the higher will be the dial reading.
(5) Row of Needles
A row, 15 cm long of closely spaced needles guided by a frame is dropped
vertically onto the road surface. The profile measured as depicted by the tips of the
needles is then photographed. The accuracy is limited by the thickness and spacing of
the needles [Astrov (1962)].
(6) Stylus
The method is well known and widely used in mechanical engineering. It gives
more information about a surface than other methods for most surface types [Richards
(1967-68)]. Moore [1966] developed a stylus device to measure the coarse texture of
the road surface. Vertical movement of the stylus are sensed by an electric linear
variable differential transformer and fed into an oscillograph. A similar apparatus was
designed by Yandell to measure the fine texture of the road asperities [Taneerananon
(1981)].
There are, however, some difficulties associated with the stylus method. The
conical shape of the stylus prevents reentrant angles from being detected. If the radius
is large fine crevices will be missed, measurement of troughs and peaks will be
inaccurate. Too fine radius may cause damage to the surface by ploughing. An
optimum angle and sharpness then must be selected, to obtain a compromise between
the conflicting requirements of not ploughing (or sticking) and true reproduction of the
surface profile.
41
(7) Profilograph
This instrument is designed to scribe a magnified profile of road texture as a feeler
probe is drawn across the surface [Rose et.al (1973)]. A mechanical linkage system
magnifies vertical movement of the probe, and the resulting profile is recorded on a
chart. In addition, upward vertical deflection of the probe are recorded on a counter as
the cumulative vertical peak heights of the surface texture through the length
transversed by the probe. Average peak height is obtained by dividing cumulative
peak heights by the number of peaks.
(8) Stereo-Photogrammetry
Two photographs of the macro-texture taken vertically from two distinct points give
sufficient information to produce stereo-photograph which are measured with a
comparator and parallel bar. The height readings are accurate to 0.01 inch [Sabey and
Lupton (1967)]. Yandell and Gopalan [1976] used stereo-pairs from a scanning
electron microscope to measure very fine texture.
(9) Stereo-Photo Interpretation
This method was developed by Schonfeld [1970,1974]. Colour stereo-photographic
transparencies or prints of approximately 6 inch square sections of road surfaces are
obtained and viewed through a micro-stereoscope or mirror-stereoscope. Texture
elements of the surface are classified visually and are rated subjectively according to
an established severity rating for each of several parameters.
( 10) Outflow Meter
Moore [ 1965] described a simple apparatus by which could be measured the
drainage capacity of a surface in terms of MHR (mean hydraulic radius). The
instrument is a transparent cylinder, about 5 inch diameter and 12 inch height with a
rubber ring glued to the bottom face. The cylinder is loaded onto the road surface so
the rubber ring will drape over the aggregate pieces in a way that simulates the
draping of tyre tread. There is no pressure applied on the water except its weight. The
time taken for a known volume of water to drain away is recorded. The short duration
of time or high rate of flow is associated with high macro-texture or high permeability
of the pavement or both.
42
A theoretical relation between MHR and the slope of friction/speed lines was
obtained using stated assumption [Moore (1966)]. The method shows good
discriminating ability [Orchard et.al (1970)]. However, the disadvantages of this
apparatus are: (a) It uses a thin rubber ring which might be affected by an odd particle
or cavity in a manner not representative of drainage in a typical tyre contact patch
[Lees and Katekhda (1974)]; (b) The variability of the readings on very smooth
surfaces is high [Moore (1968)].
A number of modifications have been made to the outflow meter. The high pressure
outflow meter, differs from the original device in that width of the rubber ring has
been increased, time measurement has been automated and the water is pressurized as
opposed to gravity flow [Henry and Hegmon (1975)]. A modification incorporating an
elliptical rubber plate in place of the circular rubber annulus, is aimed to study the
drainage paths lengths for different directions [Lees and Katekhda (1974)].
( 11) Surf ace Prints
Meyer [1964] obtained the asperity density prints by placing an aluminium foil on
the surface and providing a controlled impact onto a rubber disc placed on top of the
foil. The sharp asperities pierce the foil and the number of piercings per unit area is
the measure of texture.
(12) Laser Beam
This method was described by Gee et.al [1975]. The principal elements are laser
source and receiver. Both are off-the-shelf items. Light is emitted from the laser and is
incident on the road surface. The light reflected from the surface is generally scattered
in all directions. The polarization (alignment of the electric field vector) is also
changed after scattering. That is, linearly polarized light will experience
"depolarization", where reflected light is no longer linearly polarized, but is elliptically
polarized. The degree of ellipticity is a function of road surface characteristics, and is
represented by the ratio of the minor and major axes of the polarization ellipse. The
higher ratio indicates coarser texture.
The method appears simple and apparently suited for operation from a moving car.
The correlation coefficient between laser depolarization ratio and skid number,
however, is only about 0.5-0.6 [Gee et.al (1975)]. Another apparatus, the Swedish
43
Laser Road Surface Tester (RST), can measure both fine and rough macro-texture,
where the RST root mean square fine macro-texture is correlated far better with sand
patch texture depth than rough macro-texture. When the result of RST is compared
with the measurement by the Sideways Coefficient Routine Investigation Machine
(SCRIM), no significant correlation was achieved [Jameson et.al (1988)].
(13) Texture Friction Meter
Yandell and Mee have developed a device, that is a small computer controlled
instrument that can accurately sample pavement surface texture to give in seconds the
sideways and locked-wheel friction for several travelling speeds. It analyses a
television image of a sharp knife shaped laser beam which shines on the road surface
as the test vehicle moves along the road. The computer simulates a pneumatic tyre
travelling on that measured wet texture ['Uniken' (1989)].
1.5. Hydrodynamic Effects and Tread Rubber
To study quantitatively the masking effect of the fluid film on tyre-road friction, a
knowledge of the fluid film thickness is necessary. The effective film thickness which
exists between the tyre and road surface during sliding or rolling is different to the
applied film thickness. The Reynolds equation is commonly used as the basis for the
calculation of effective film thickness. According to Yandell et.al [1983], this film
thickness is dependent on the mode of sliding (sideway or locked wheel), the speed of
travel, and the temperature of the water.
1.5.1. Reynolds Theory of Lubrication
Reynolds [1886] applying the Navier-Stokes equations derived what is known as
the Reynolds Equation. In its original form this equation is written as:
!(h':) + !0':) = 6v[ ~0 +U,): + 2v1 ]
where h : fluid film thickness
44
p : fluid film pressure
v : absolute viscosity of fluid
V 1 : velocity of vertical separation
U0, U1 : horizontal velocities of lower and upper surfaces
x, z : coordinate axes on the same plane.
The equation can either be solved by analytical or numerical methods. The
analytical or closed form solution requires the film thickness to be expressed as a
function of x and y. However, as the two dimensional equation can not normally be
integrated directly, one dimensional equation is usually solved by assuming an
infinitely long or short bearing. The pressure p is then obtained as a function of x. The
numerical methods, such as the finite difference and the finite element technique are
common with the aid of digital computers.
1.5.2. Fluid Film Thickness
Various investigators have used different water film thicknesses in their
experimental and theoretical studies. Daughaday and Balmer [1970] used an effective
film thickness of 0.02 inch (0.508 mm) in the thick film region and a thickness of
0.005 inch (0.127 mm) in the semi-dry region. Film thicknesses between 0.03 inch
(0.762 mm) and 0.08 inch (2.03 mm) were used by Allbert and Walker [1968] in their
laboratory test. The commonly used water film thickness in field measurement is 0.02
inch since beyond this thickness the friction tends to level off as was found by Giles
[1959] and Besse [1972]. The internal watering system as specified by the
ASTM-E274 also gives water film thickness of approximately 0.02 inch.
In field tests, however, film thicknesses are difficult to measure with a high degree
of accuracy. This is more difficult on a coarse textured surfaces since the film is non
uniform and the depth varies from top to bottom of asperities and from one asperity to
another. Thus for field tests the water film thickness refers to the applied or nominal
film thickness. The minimum film thickness of 0.0002 inch (0.0051 mm) is assumed
by Taneerananon [1981] in the theoretical calculation of the effective remaining film
thickness between the tyre and road surf ace.
45
1.5.3. Hydroplaning
Hydroplaning can be described as a phenomenon when the tyres of a vehicle
travelling at some critical speed are detached from the wet pavement surface. As a
result, the ability of the tyres to develop braking or cornering forces is drastically
reduced. On a flooded surface if the vehicle speed exceeds the tyre hydroplaning
speed, hydroplaning will occur and control of the vehicle is completely lost. The
hydroplaning speed is a function of the pavement surf ace, the water depth and various
physical tyre parameters. The depth of water on the pavement required for
hydroplaning can vary considerably depending on the particular combination of the
type of road surface texture and the tyre tread design. Patterned tyres operated on
coarse textured surfaces require the greatest depth, while smooth tyres operated on
smooth road surfaces require the least water depth. Hydroplaning can be further
classified into two groups namely dynamic hydroplaning and viscous hydroplaning.
(1) Dynamic Hydroplaning
This situation generally occurs on a flooded surface. The fluid inertia force is the
important factor in this type of hydroplaning. As the moving tyre contacts the
stationary fluid, it causes a sudden change in the momentum of the fluid which in turn
creates hydrodynamic pressure which acts on the surface of the tyre. The
hydrodynamic pressure tends to increase with the square of the tyre speed. A critical
speed is reached when the hydrodynamic pressure force equals the tyre load. This
speed is called "tyre hydroplaning speed" and can be computed by the semi empirical
formula put forwards by Home and Dreher [1963].
This formula ignored the effects of tyre tread design, fluid viscosity and road
surface texture. It was given as:
VP = 10.35./p
where VP = Tyre hydroplaning speed (mph),
p = Tyre inflation pressure (psi).
46
(2) Viscous Hydroplaning
Viscous hydroplaning occurs when there is a thin film of fluid present on the road
surf ace. The internal friction or shear force between the fluid layers prevents the fluid
from escaping from under the tyre footprint. Viscous forces arising from the viscosity
of the thin fluid is dominant in this type of hydroplaning. The pressure force which
develops while the tyre is sliding over the contact patch is calculated by means of the
Reynolds Equation. This pressure force tends to separate the tyre from the road
surface and reduce the amount of friction being generated. The eliminating of the
intervening film is rather difficult. In normal rolling or sliding, it is eliminated by the
squeezing action of the tyre onto the surf ace asperities.
1.5.4. Models for Calculation of Water Film Thickness
Saal [1936] studied the slipperiness of the road by using the sinkage equation
originated by Reynolds. Assuming the contact area of the tyre to be of a flat elliptical
shape, he obtained a reduced water film thickness after a rolling tyre had passed
through the wetted road surface. The duration of time of sinkage was taken as the time
required for the contact segments to travel past the footprint region. The amount of
friction that could be developed was then dependent on whether the height of the
asperities was greater than the residual film thickness. The model, despite its
simplicity gave an insight to the role of the remaining water film in the development
of friction on wet surf ace.
In his study of viscous hydroplaning, Moore [ 1967] used asperities of sinusoidal
shape to represent a pavement surf ace. Individual asperities generate pressure forces
which provide a net uplift tending to separate the sliding rubber from the asperity
surfaces. He divided the film into three region: inlet, central (or foil bearing), and
outlet. Using Reynolds Equation in two dimensional form, the pressure forces were
calculated for the inlet and outlet regions, the central region had constant pressure. He
attributed the main source of friction to the adhesion between microtexture at the
asperity tips and the tyre rubber providing that the micro-texture depths exceed the
thickness of the foil bearing film. The model was applied to a rolling tyre which is
accelerating on a wet surface.
47
Daughaday and Balmer [1970] used a mathematical model to study the rolling and
skidding behaviour of a tyre. The flow of water was divided into: an exterior flow
region, an inlet region, and a footprint region (consists of thick film region and semi
dry region). The amount of thick film penetrate into the footprint region determined
condition of the hydroplaning. In the semi dry region each individual load carrying
asperity was either in dry contact with the tyre surface or the fluid film thickness was
small compared to the depth that the asperity indented the tread surface. For partial
hydroplaning case, the main source of friction was derived from the contact of road
asperity tips with the tyre surf ace in the semi dry region. The other source of friction
was from the viscous forces acting in the thick film region. The model assumed dry
contact exists and thus considered adhesion as the major source of friction. It would
appear that the theoretical friction coefficient depends significantly on the estimation
of the coefficient of friction of the semi dry region.
Taneerananon [ 1981] presented models which using Yandell' s Theory of Hysteretic
Sliding Friction [Yandell (1971 b)] attribute all the friction generated to the hysteresis
process. Adhesion is not considered because of the ever presence of the water film or
contaminants on the texture surf ace. Using Reynolds Equation he developed models
for the sliding tyre (simulation of locked-wheel friction), and for rolling tyre
(simulation of sideway friction). Together with the surface texture measurement and
tread rubber properties, he used these models to predict the locked-wheel BFC, and
(maximum) SFC. Each of these models will be described below.
( 1) Model for Sliding Tyre
The main purpose of this model is to obtain the fluid film thicknesses which are . important in determining the masking of the surf ace texture. The model assumes a one
dimensional flow of fluid in the direction of sliding motion of a tyre. No flow is
assumed to take place in the direction at right angles to the tyre motion. The sinkage
of the tyre takes place simultaneously with the sliding, and the solution has to be
obtained by trial and error method.
The following assumptions were made in the calculation.
a. The tyre surface is parallel to the asperity surface when calculating the film
thickness.
b. The resultant pressure force acts through the centre line of the asperity.
48
c. The flow is laminar and the inertia forces are negligible compared to the viscous
forces.
d. The asperity swface is either smooth or rough.
The first assumption appears reasonable considering the film thickness relative to the
dimension of the asperity. In the second assumption, the actual resultant would act
slightly off the centre line towards the thin end of the film, however owing the relative
smallness of the asperity, the assumption would seem to be justifiable. Taneerananon
examined the third assumption using formula from Pai [1956], and he found that the
inertia force can be ignored. To fulfil the fourth assumption, the effect of micro
texture on the final film thickness then is computed for a range of values of sinkage
time [Taneerananon and Yandell (1981)].
(2) Model for Rolling Tyre
A model of a tyre sinking onto the asperities surface is used to compute the film
thickness. When a wheel is rolling at an angle to its direction of motion, a sideway or
lateral force is developed (due to lateral slip), which acts perpendicular to the wheel
plane. Since there is little or no relative motion between the rolling tyre and the road
swface at the interface, only the sinkage of the upper tyre surface will be considered
in this model. Taneerananon assumed that the hydrodynamic uplift which may be
generated as a result of lateral sliding is small and can be safely ignored. Hence the
model of sinkage is similar to the sliding tyre, except that the tyre motion is sinking
instead of sinking+ shearing. It is clear, that as the speed of travel increases, the time
available for sinkage decreases so that the water films will remain at greater thickness.
One example of relationship between time of sinkage and water film thickness,
derived from Reynolds equations, is shown in Figure 1.28. The effect of tyre tread . (smooth, patterned) is taken into account through the magnitude of fluid film pressure.
During locked-wheel friction, sinkage and shearing are taking place simultaneously.
As the shearing rate increases, an existing water film will increase in thickness for a
given normal contact pressure [Yandell et.al (1983)]. By just referring to this
consideration, it can be expected that the braking coefficient is lower than the sideway
coefficient.
49
mm in
·04
·OJ
·02
01
0
10·' inches
,::;
.. .. • C ,. u
15
i 10 ~
E
E :,
E C
:J: 5 cylindr,coL shape roughness ., depth .ex,, ~ l
Fig.1.28. The texture effect on film thickness as a function of time. [Taneerananon and Yandell (1981)].
1.5.5. Measurement of Tread Rubber Properties
The main tread rubber property to be measured is the damping factor, which is
defined as the area inside the stress-strain hysteresis loop divided by the area under the
loading curve in an unconfined triaxial load-unload test on the tread rubber with
homogeneous stress conditions .
. A number of methods have been used to measure this property, including the
rebound pendulum, which measures the energy returned divided by the energy applied
to the rubber from a hemispherical anvil. Another method involves measuring the
bounce height of a steel ball on the tread rubber. According to Yandell et.al [1983],
most of the methods used are not suitable since (1) they do not load the rubber over
the frequency range encountered by the skidding tyre, and (2) the pulsing stress
pattern is not homogeneous.
50
Temperature
07
! .Q
0·6 2 0
~ u 2
"' C ·o. 05 E 0 0
IO 20 40 50mph
Equivalent slidi119 speed (otter Zonkinl
Fig.1.29. Measured relationship between the damping factor of filled tread rubber, speed, and rubber temperature. [Yandell et.al (1983)].
To overcome those disadvantages above, Zankin [1981] constructed a special
apparatus for measuring the damping factor behaviour of tread rubber for speeds up to
80 kph and temperature up to 15C>°C. Figure 1.29 shows the relationship between
sliding speed, rubber temperature, and the damping factor of a filled 60:40 styrene
butadiene rubber (SBR) - polybutadiene rubber (BR) of tread rubber, with about 33%
high structure carbon black and a Shore A durometer hardness of 66. This figure was
used in road friction prediction.
Another rubber property to be measured, that is when calculating the effect of
rubber mass inertia to friction coefficient, is its density. This effect is significant over
about 40 kph. At low sliding speed the increase of friction coefficient with increased
density is minimal. However, a 3% increase is obtained for 80 kph, when the density
increases from 0.033 lb/cu in to 0.038 lb/cu in [Taneerananon (1981)].
51
1.6. Theory of Hysteretic Sliding Friction
Controversy exists between those attributing the greater part of skidding resistance
on wet surfaces to inter-molecular adhesion and those attributing it to stress-strain
hysteresis in the rubber. Tabor [1952] demonstrated the connection between the
rolling resistance of rigid bodies on rubber and the damping properties of the rubber.
Greenwood et.al [1961] maintained that the increased coefficient of friction observed
as the apex angles of water lubricated cones decreased, was due to the damping
energy to the sliding rubber.
Moore [1967], using the elasto-hydrodynamic theory of lubrication and Tabor's
hysteretic prediction, attempted to demonstrate that hysteresis played only a minor roll
in the skidding resistance of an automobile tyre on a wet road. Kummer [1966]
presented one semi empirical formula for predicting the hysteretic friction and another
for predicting adhesive friction. The adhesion was presented as a hysteresis
phenomenon.
Yandell [1968] presented a mathematical mechano-lattice analogy for predicting the
coefficient of hysteretic friction of rubber sliding on simple geometric asperities. Next
Yandell [1969] proposed and proved a method of applying the results of his
mechano-lattice analogy to the prediction of the coefficients of hysteretic friction of
rubber sliding on lubricated roadstones.
Taneerananon [ 1981] found that in most cases the prediction of hysteretic friction
agrees reasonably well with the measured locked-wheel coefficient when the effects of . the temperature are taken into account. For the sideway force coefficient, the
prediction under estimated the measured coefficient. He explained that the damping
factor of rubber may have increased as a result of the mastication but has not been
accounted for in the prediction.
In the following, brief descriptions of the theory of hysteretic sliding friction will
be presented [Yandell (1971b), Yandell and Holla (1974), Yandell et.al (1983)].
52
1.6.1. The Mechano-Lattice Analogy
Yandell used his mechano-lattice analogy to simulate and analyse the stresses in
and determine the friction of rubber as it slid over triangular prisms or cylinders. The
analogy consisted of an array of connected units designed to simulate the behaviour of
a long section of the sliding rubber in plane stress. In order to give a rigorous
simulation of the rubber the analogy was arranged to allow for large deformations,
possess an appropriate Poisson's ratio and Young's modulus and a damping factor of
a magnitude commensurate with the type of rubber, temperature and speed of sliding.
The analogy was shown to be reasonably accurate when coarse grid predicted
coefficients of friction were compared with those measured by a British Pendulum
friction tester on smooth brass triangular asperities with various apex angles [Yandell
(1969)].
The extent of load-deflection hysteresis, during the single point cyclic loading of a
rubber, is specified by a damping factor. This factor is defined as the energy dissipated
divided by the energy applied in one complete cycle of load, or, referring to Figure
1.30, the dotted area of the hysteresis loop divided by the hatched area under the
loading path of the loop. An applied or measured load-deflection hysteresis loop is
made more mathematically useful if simplified, that is replacing it with a
parallelogram-shaped loop of similar area as shown in Figure 1.30. A model,
consisting of an elastic and a frictional element in parallel, which yields the simplified
hysteresis loop is also shown in Figure 1.30.
The relation between damping factor to the Young's moduli and maximum internal . friction is given by
where p = internal friction coefficient= F'/(t'emax - t\min)
S = stiffness coefficient of the model.
It can thus be seen that for a given rigidity the damping energy per cycle depends on
the internal friction, be it viscous or Coulomb, and on the maximum load range. The
model described above is the basis for the behaviour of the individual elements
comprising the rubber simulating units.
53
Deflection a. Simplified hysteresis loop
b. Model exhibiting behaviour of simplified loop ~ = damping factor= dotted area/hatched area E0 =elastic Young's modulus a::.a/c £,=frictional Young's modulus a::.b/c F' = maximum internal friction.
Fig.1.30. Simplified of measured hysteresis loop. [Yandell (1971b)].
HU,J) -3-h (I,J)
~ _L ---:,
A = horizontal and vertical elements V =volume elements; S=shear elements E~ = elastic stiffness factor E, = c · E~ frictional stiffness factor; c = a constant.
a. Analogue of mathematical model
VCl,J)~ y- f 0
I,J l,J +1
I 1r~vU,Jl
--0'1-
I -- (;
0 I +1, J I +1,J+1
b. Movement of analogue joints
Fig.1.31. A unit of the mechano-lattice analogy. [Yandell (197lb)].
1.6.2. Mechano-Lattice Unit
I I I
Figure 1.31a shows the mechanical analogue of one of 264 mathematical units
forming the model used for simulating the plane stress behaviour of a long section of
rubber. Each unit is made up of eight elasto frictional elements similar to that shown
in Figure 1.30. There are two horizontal, two vertical, two shear and two volume
elements. The volume and shear behaviour are separated by two crosses to rotate. This
is necessary to allow any Poisson's ratio (a).
54
When volume changes are occurring, the shear cross rotates with no additional
loading on its elements. When shear changes are occurring, the volume cross can
rotate and no additional load is put upon its elements. Shear strains in a unit are
functions of differences in diagonal lengths and volume strains are functions of
changes in the sum of diagonal lengths.
The stiffness factors of the elastic components of the elements are calculated
assuming that no friction is present. McHenry [1943] was one of the original
proposers of the simple lattice analogy for simulating continua with a Poisson's ratio
restricted to 1/3 and obeying the theory of linearised elasticity. He calculated the
stiffness factors of a simple cross braced lattice simulating the behaviour of a
non-buckling plate of unit thickness. The (elastic) stiffness factors are:
(E)))/2(1 +cr) horizontal and vertical elements
{E))V2)/(1-cf) volume elements
(E)))/(l+cr).../2 shear elements
1.6.3. The Friction of Rubber Sliding on an Asperity
A total number of 264 units shown in Figure 1.31 are connected at their joints to
simulate the long section of rubber sliding with plane stress as shown in Figure 1.32.
All boundary joints except those on the lower edge are fixed. About 14 of the central
bottom joints are deflected to the shape of the single fixed smooth asperity. The apex
joint is fixed to but is free to move with the asperity. The joints contacting the asperity
are free to move along its surface. The body joints are all free to move in response to
unbalanced forces acting on them.
As can be seen the units to the left of the asperity have been loaded while those to
the right have not. The extreme right-hand column of units is assumed to be unloaded.
As they move across the asperity corresponding elements go through cycles of load
and deflection to generate hysteresis loops similar to that shown in Figure 1.31. The
area of the hysteresis loop is governed by the damping factor of the rubber and the
maximum elastic stress range experienced by the element. A computer program
obeying the properties of the units ensures this. A simplified flow chart of the
computer program is shown in Figure 1.33.
55
X)C)<)(
)( )c )(
)c )( X )<
)< )< )< )c X )c
X )( ><
I Joints on free surface
Fig.1.32. The assembled units simulating a long section of rubber sliding on an asperity. [Yandell (197lb)].
READ: RUBBER:-!.elastkYoung's modulus E.,. frictional Young's modulus £1; Poisson's ratio u; inter-nal frictional coelT. p. ASPERITY :-slope or a/Rad.; penetration a, surface friction; joints at interface.
½
SET UP: simulating grid (9 x 34); boundary conditions and initial deflections of joints.
i COMPUTE: elastic stiffness factors of horizontal, vertical, volumetric and shear elements.
i ...... COMPUTE: lengths, rotations, elastic and frictional forces in elements taking hysteretic behaviour into ,..
account. The maximum force range experienced by each element.
+ RESOLVE AND SUM: forces in elements horizontally and vertically at joints.
! I MOVE: joints horizontally and vertically in response to any unbalanced forces acting on them. Joints in i contact with the asperity are moved along its surface taking surface friction into account.
NO ! DO forces on free joints approach zero?
YES fvEs
HA VE joints at interface been adjusted.
lNo
WRITE : displacements, boundary reactions, stresses and the coelT. of hyst. friction, /h.
Fig.1.33. Flow chart of the analogy computer program. [Yandell (1971b)].
The program calculates and stores the lengths and rotations of all elements. The
elastic and frictional forces are calculated for each element in each stage of loading.
This is done incrementally working from right to left so that the stress history retains
its continuity. The element has gained a frictional as well as an elastic increment of
force by the move and has commenced to follow the loading path of its hysteresis
loop. After a number of additional sequential moves to the left, the frictional force in
the element reaches its maximum F', the less steep elastic line of the hysteresis loop is
56
0-3
COEFFICIENT
CF HYSTERETIC
FRICTION
0·2
0·1
--- CYLINDERS
-- TRIANGULAR PRISMS
OF of 0·5 / Of BO km/hr // with inertia--,,
,,, /
/
,
DAMPING FACTORS~--"(without inertial effect l
AVERAGE SLOPE OF CONTACT
0 O·I 0·2 0·3 0-4 0·5 0·6
0·5
a 07 R
Fig.1.34. Mechano-lattice-detennined relationships between hysteretic friction, damping factors of sliding rubber, and average slope of contact of triangular prisms and cylinders. [Yandell et.al (1983)].
followed until unloading commences and by a reverse process the loop is completed
and possibly recommenced after the element has travelled across the array.
Subroutines in the computer programme ensure this. As their distance from the free
surface increases the size of the hysteresis loops, predictably, decreases.
. When all the forces in the elements have been calculated, each joint is moved in
response to any unbalanced force transmitted to it from the connecting elements.
However, the joints are moved in one operation and in only small increments so that
stability will be maintained. A damping function which is controlled by the largest
unbalanced force in the array is used for moving the joints. Once the joints have been
moved the forces in the elements are recalculated. The procedure is continued until all
the free joints have moved to stable positions and sustain no significant unbalanced
force. The programme checks periodically to ensure that no joint at the asperity
interface is in tension or penetrates the surface of the asperity.
57
scale 3
Fig.1.35. Analysis of measured texture profiles into scales. [Yandell et.al (1983)].
The coefficient of hysteretic friction fh is determined by dividing the vectorial sum
of the horizontal forces acting on the joints which contact the asperity by the vertical
reactions on these joints. The horizontal, vertical and shear direct stresses at the centre
of each unit are calculated together with the major and minor principal and the
maximum shear stresses. The coefficients of hysteretic friction generated by rubbers,
with a number of damping factors, as they slid over isosceles triangular prismatic and
cylindrical asperities with various average slopes of contact, were calculated by the
mechano-lattice analogy, such as shown in Figure 1.34.
L6.4. The Hysteretic Friction of Road Surfaces
Yandell's theory assumes that the total texture of a road surface can be analyzed
into components or scales ranging from microscopic size up to half or one inch.
Figure 1.35 shows how the texture is analyzed into a number of scales.
Rubber is assumed to "flow" over each scale and generate hysteretic friction. The
coefficient of hysteretic friction at a given speed, was the sum of the coefficients on
each scale of texture, as shown in Figure 1.36.
58
_.c. 2.0
z 0 ~ So! a:: u.
!:! I-l&I a:: 1.0 l&I I-II)
?i: u. 0 o.s ~ l&I
i::i it 0 l&I 0 u
depressed envelope due to masking of some scoles by surface water
Peak fridion on scolel
E:::::::::;:::~:::;:=:.......----,,----::,~----::r=-..:.lc~oa.rses tl 10 20 30 40 so 60 70
mile/h
0 10 20 30 40 so 60 70 80 90 100 110 SLIDING SPEED km/h
Fig.1.36. Schematically presentation of the coefficient of hysteretic friction of the road surface. [Yandell and Holla (197 4)].
1.7. Problem Definition and Method of Investigation
1. 7 .1. Problem Definition
As it will be described in the next section. the field measurement of skid resistance
will require large capital investment, while the test results may not be easy to repeat.
Furthermore. the testing done at high speeds or at difficult locations such as sharp
curves and steep gradients may be dangerous. Therefore. any reliable method of
p~edicting the coefficient of friction from surface texture will be useful and desirable.
Yandell [1970] devised a means for predicting the coefficient of friction between a
rubber slider on road surface. He attributed all the friction to the hysteretic losses in
the rubber. He used the mechano-lattice analogy to calculate coefficient of hysteretic
friction for a range of surface texture and rubber characteristics. From the measured
surface characteristics he obtained the coefficient of hysteretic friction for the divided
scales of texture, the sum of which is the coefficient of friction of the surface. Yandell
was able to verify his theoretical coefficients by comparing with the values measured
with a British Pendulum tester.
59
Holla [1974] used Yandell's method for calculating coefficient of friction and his
empirically constructed damping factor versus frequency curves to predict the sliding
coefficient of friction on a road surface. Assuming the viscoelastic properties of
rubber to be the dominating factor, he computed different maximum coefficients of
friction and different frequencies of loading from the average slope and pitch of the
surface texture. A single wheel trailer was used to measured the locked wheel
coefficient of friction under two arbitrary levels of wetness obtained by external
application of water. Therefore the effect of water film can only be quantitatively
estimated. Holla attributed the difference between the prediction and measurement of
friction coefficient to the abrasion friction, the viscous friction of the water film and
the masking of texture scales which can not be assessed quantitatively without
knowing the water film thickness.
Gopalan [1976] carried out a laboratory study of the abrasion of rubber and
attempted to predict the coefficient of friction of a sliding rubber block. Unlike
Yandell's work in which glycerine was used as lubricant and the amount masking of
fine scales of texture was deduced experimentally, Gopalan used water as lubricant
and therefore required a knowledge of water film thickness in order to determine the
masking effect. He estimated the film thickness by the theoretical method given by
Moore [1967]. He also deduced the same film thickness from the experimental data
produced by Wallace and Trollope [1969]. Also, because much finer scales were left
unmasked by water, Gopalan used electron microscope stereo pairs for fine texture
measurement.
Taneerananon [1981] rigorously established the thickness of water film present
between the tyre and road surfaces. The inertia effect of rubber on the coefficient of
hysteretic friction has been quantitatively studied using the mechano-lattice analogy
evolved by Yandell. He then was able to predict the locked-wheel braking and
sideway force coefficients, by applying his two models for calculating water film
thickness. Taneerananon attributed the discrepancies between the prediction and
measurement of friction coefficient to the under estimation of the temperature of the
tread rubber; and then suggested to introducing an empirical factor to account for
temperature effect.
60
At the same time, Zankin [1981] constructed a special apparatus for measuring the
damping factor behaviour of tread rubber for speed up to 80 kph and temperatures
ranging up to 150°C. He found that the hysteresis behaviour of filled rubber was only
partly viscoelastic. High hysteresis losses at room temperature in carbon-black-filled
rubbers are greatly reduced at high temperatures owing to the temporary breakdown of
filler particle agglomerates. His result seems to agree with Payne's work [1974] who
founded that the tread rubber sliding over the coarse texture will break down the
carbon filler and generate higher temperature simultaneously. It was then suggested
that the introduction of a separate set of rubber temperatures for each road surface
reduces the average absolute error (between prediction and measured value of friction)
to 5 percent [Yandell et.al (1983)]. The SFC and BFC for a simulated pneumatic
passenger car travelling at a number of different speeds were predicted with
reasonable accuracy from surf ace texture measurements and tread rubber properties.
In this work which is a continuation of the effort to predict the coefficient of
friction on road surfaces, it was proposed to investigate the tyre-road friction under
braking and cornering. In other words, the prediction of locked-wheel BFC (at 100%
slip) and maximum SFC (at optimum slip angle) which have already been done
[Yandell et.al (1983)], will be extended to cover all ranges of percent slip and slip
angle.
In developing the model for the theoretical prediction, it was proposed to involve
either the locked-wheel BFC or maximum SFC, as input parameter. It has been found
from Taneerananon's works [1981], that the locked-wheel BFC (i.e. at 100% slip) and
the maximum SFC (i.e. at optimum slip angle) can be predicted with reasonable . agreement, in which the following factors have been taken into account: surface
texture, water film thickness, tread rubber properties and certain operating conditions.
The mode of operation, however, was only represented by the models for sliding and
rolling. Therefore, to extend the prediction, the mode of operation should also be
extended covering all percent slips and slip angles.
It should be noted that although the locked-wheel BFC or maximum SFC are
included in the model, in practical application they can be predicted from surface
texture data, instead of being directly measured. In other words, the prediction of
locked-wheel BFC or maximum SFC could be treated as sub-models of the main
61
model. Hence, the ultimate goal to develop a theoretical technique for predicting
tyre-road friction from surface texture data is still continued.
To assess its validity, it was proposed to conduct experimental tests on road
surfaces, so that the theoretical prediction can be compared with the field results. The
braking and sideway force tests were carried out.
Extensive verification programs for Yandell's theory of hysteretic sliding friction
were carried out in 1981-82 by Yandell and Taneerananon, involving the prediction of
BFC and SFC operating separately of a simulated passenger car pneumatic tyre on
many different types of pavement for a number of travelling speeds. The predictions
were based on the measured pavement surface texture and the measured damping and
stiffness properties of the tread rubber. The University of New South Wales multi
mode friction test truck developed by Yandell and Ferguson was used to directly
measure the friction values. Figures 1.37-1.38 show some of the results.
It is proposed to extend this work further by introducing the concept of tyre
stiffness into the relationship between braking and sideways friction.
This study will involve field friction testing which will require the following
considerations:
a. The measurement of friction on the road to the desired accuracy requires repetition
of measurements, taking into consideration the time required in taking
measurements and doing analyses.
b. The precision of the estimate will increase with increasing number of replications.
However, the number of replicate measurements are limited by the time required
for testing, which may influence the test results through uncontrolled factors such
as change in pavement surface characteristics, temperature (given the long period
of measurement). Thus, the number of replications is determined from a number of
considerations rather than precision alone.
c. The behaviour of a pneumatic tyre is more complex than that simulated by Yandell,
Taneerananon and Zankin. This is partly due to the non uniformity of the normal
stress distribution in the contact area as a result of bending of the carcass. Factors
like the one just mentioned are difficult to include in the prediction, and therefore
some discrepancies in the prediction are to be expected.
Fig.1.37. Measured BFC vs predicted BFC, on many types of pavement (concrete, bituminous) and macro-texture (closed, mild, smooth). [Taneerananon (1982)].
1 Measured SFC ( N maximum)
0.9 Veh le le Speed 90%
I::.. 10 mph Confidence 0.8 0 30 mph
D 50 mph 0.7
0.6
0.5
0.4
0.3
0.2 Regression Line
0.1 Slope = 0.938
0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Predicted SFC
Fig.1.38. Measured SFC vs predicted SFC, on many types of pavement (concrete, bituminous) and macro-texture (closed, mild, smooth). [Taneerananon (1982)].
63
In spite of all the difficulties associated with the prediction of the tyre-road
frictional coefficient, it was proposed to investigate the problem in the following
manner.
1. 7 .2. Method of Investigation
1. Study the longitudinal and lateral stiffnesses of tyres in relation to operating
conditions and tyre properties.
2. Develop the model for tyre stiffness variation.
3. Study the theoretical prediction and the existing models for tyre-road friction
under braking and cornering.
4. Develop the model for the prediction of tyre-road friction under braking and
cornering.
5. Study the effects of several factors affecting tyre-road friction by using the
proposed model, and compare the model response with other tyre models.
6. Measure the braking and sideway forces under different operating conditions and
find the corresponding coefficients of friction.
7. Obtain from the measurement in (6) the parameters needed for the model.
8. Predict the tyre-road friction under braking and cornering using the proposed
model.
9. Study the variations of the experimental coefficients of friction under various
operating conditions.
10. Compare the theoretical coefficients of friction in (8) with the experimental
coefficients of friction in (6) . . 11. Modify, where appropriate, the proposed model for prediction by using the
comparison in (10).
12. Study the application and measurements of tyre stiffness.
64
CHAPTER 2. MEASUREMENT AND PREDICTION OF TYRE-ROAD
FRICTION
Notation (Used frequently in Chapter 2) a, b = length of increasing & decreasing pressure zone (unit length) BFC = braking force coefficient
Cx = longitudinal slip stiffness (force/slip, or force/fraction slip), or longitudinal deformation stiffness (force)
Cy = lateral slip stiffness (force/degree, or force/radian), or lateral deformation stiffness (force)
½C = longitudinal slip stiffness coefficient (per slip, or per fraction slip), or longitudinal deformation stiffness coefficient
C,C = lateral slip stiffness coefficient (per degree, or per radian), or lateral deformation stiffness coefficient
IP = tyre inflation pressure (psi) 21, w = tyre contact length & width (unit length)
Kx, Ky = longitudinal & lateral carcass spring rate (force/length per unit area) kx, ky = longitudinal & lateral tread element spring rate (force/length per unit area)
P = tyre normal pressure (force/unit area) SFC = sideway force coefficient
Sx = longitudinal slip (%, or per fraction slip) Sy = tancx V = velocity (unit length/unit time) ex = slip angle (degree, or radian)
All reported the skid resistance in terms of straight forward braking and sideway
force coefficients. Numerous methods have since been developed to measure skid
resistance. The common examples are sideway force, braking force, stopping
distance (or decelerometer), and portable skid resistance testers.
2.1.1. Sideway Force Measuring Method
This method is based on contention that the critical maneuver with regard to
skidding is cornering. The test wheel which is free to rotate is set at a predetermined
angle to the direction of motion, and the sideway force acting normal to the plane of
the wheel is measured. The frictional resistance is then expressed in terms of the
sideway force coefficient which is the ratio of the sideway force to the vertical load.
The maximum sideway force is obtained at a critical angle of inclination of the
test wheel. This critical angle is a function of pavement surface parameters, and is
usually slightly below 20°. The instrumentation of the test vehicle usually involves
electrical strain gauges for measuring the force and an automatic system to obtain a
continuous graphical representation of the test results. SCRIM (Sideways Coefficient
Routine Investigation Machine) is an example of this measuring method (Figure 2.1).
The sideway force method is accurate (reproducibility ± 1 to 2 percent), yields
continous measurements along the road, the wear of the test tyre is uniform, and
there are no accident hazards involved. Depending on the degree of sophistication
aimed at, the initial equipment cost may be relatively high.
66
2.1.2. Braking Force Measuring Method
A test wheel of a trailer positioned in the normal running mode of a vehicle is
braked and dragged over the surf ace to be tested at a constant velocity, thus
generating a frictional tyre-road force. The wheel may either be locked, giving
sliding friction, or allowed to run at a controlled percentage slip, to find the peak
friction at 5 to 20 percent slip.
Trailer tests can be carried out at normal traffic speeds with minimum
interference to traffic, and the procedure is fast and accurate (reproducibility ± 1 to 2
percent). While the initial cost of the equipment is relatively high, the unit testing
cost (cost per test site) is usually very reasonable [Csathy et.al (1968)].
2.1.3. Stopping Distance Method
The test vehicle is brought to the predetermined initial speed and its brakes are
applied hard instantly to lock the wheels. The stopping distance from the point of
application of the brakes to the point at which the vehicle stopped is measured,
either by means of a stopmeter with a fifth wheel or with a tape from a chalk-mark
made by a brake-activated detonator [Csathy et.al (1968)]. The results of the tests
are reported either as stopping distance at a given speed, or as the average (braking)
coefficient of friction, using the simple formula f = V2/30S, where V is the initial
speed in mph and S is the stopping distance in feet.
This method is relatively inexpensive, closely simulates the critical conditions of
emergency braking. On the other hand, it is normally limited to speeds below 30 to
40 mph, because there is an element of accident hazard involved and elaborate
traffic control measures are required. The long stretch of road is required for testing.
The reproducibility of the measurements varies from 2 to 15 percent.
As an alternative to stopping distance, a decelerometer is used in a similar testing
technique. As the brakes are applied, the deceleration of the vehicle is measured
instead of the stopping distance. The coefficient of sliding friction is numerically
equivalent to the deceleration expressed in g's. The problems at stopping distance
67
method may be minimized by using only very short skids rather than skids to a stop.
Although it is a satisfactorily accurate (typical reproducibility ± 2 to 3 percent), the
testing procedure is rather difficult to standardize.
2.1.4. Portable Skid Resistance Testers
The most widely used among these testers is the British Portable Skid Resistance
Tester. The apparatus measures the frictional resistance between a rubber slider and
the wetted road surface. The slider consists of a block of tread rubber 1/4 inch thick
and 3 inch wide mounted on the end of a pendulum arm. The pendulum arm is
released from the horizontal position so that the slider traverses along the test
surf ace and swings upwards after leaving the test surface. The work done against
friction during sliding is equated to the loss in energy of the pendulum arm thus
enabling the direct calibration of the scale. The reading is given as 100 times the
friction coefficient and known as Skid Resistance Value (SRV) or British Pendulum
Number (BPN).
Good correlation exists between the readings of the pendulum tester and the skid
test results of patterned tyres at speeds of 30 mph. This instrument is relatively
inexpensive, however, there are also some disadvantages, such as: results obtained
on coarse textured surfacings can be misleading, only covered small area of the road
surface, difficulties of measurement in busy traffic, and unable to measure skid
resistance at high speed [Croney (1977), Salt (1977)].
2.2. Some Theoretical Methods of Predicting Tyre-Road Friction
The difficulties associated with direct methods of road friction measurement have
lead many researchers to look for other means of obtaining skid resistance. The high
cost of setting up skid testing vehicles is probably a main consideration. In addition,
particularly with the locked wheel skid testers, the repeatability of measurements by
a tester, according to Meyer et.al [1974], is generally not adequate. The variability of
test results is of course due to a number of factors as discussed in Section 1.2.
68
Surface Texture Qllllllification
Pavement Surface Texture aaBRification
Macro-Texture Of Projections Micro-Texture Of
Density Of
Approximate Distribution
Projections Background Approximate Angulari~y or Projections, Parameter
Height Width Hanhne88 Harshness Number C** As A Percentage E** F A* B Of Total Area
D
Cavity in 0 0mm 16mm 0-12 Percent Surface in
Matrix
Round 13-37 Percent POLISHED
I l/4mm 8mm (No Texture Visible)
SMOOTH (Texture Visible But
2 1/2 mm 4mm Subang. 38-62 Percent Micro-Projections Too Small For Visual Estimate OfHeight)
FINE GRAINED 3 I mm 2mm Angular 63-87 Percent (Micro-Projections Appr.
1/4 mm High) t
COARSE GRAINED-
88-100 Percent SUBANGULAR
4 2mm (Micro-Projections Appr.
Particles leS& 1/2 mm High Or More) t
Than 2mm Wide Are COARSE GRAINED-.
Regarded As ANGULAR 5 4mm
Background (Micro-Projections Appr. 1/2 mm High Or More) t
6 8mm
•A texture element ha, a heiBht dimeruion only if the ,urroundi~ area below it• peak u drained. ••For the u•e of the Photo-Interpretation Chart for A•phalt Pavement• the foUow~ adju.tment u made: if the C-Parameter Number u 2, then the E-Parameter Number u raued by 1; if the C-Parameter Number u 3, then the E-Parameter N.umber u raued by 2. fMicro-projection• mmt protrude by an amount not leu than lu,lf theu width.
2.2.1. Schonfelds's Photo-Interoretation of Pavement
The six parameters used to classify the pavement surface are:
1. Parameter A denotes the height above the matrix of projections on the pavement
surface.
2. Parameter B denotes the width of the surface projections at the top of the matrix.
3. Parameter C denotes the shape of the projections.
4. Parameter D denotes density of projection distribution as the proportion of the
69
ic Ill ... c IIC .. Q
it .,, ... Ill
"' :,: ii: Q
.,, ... "' :c .. Cl u 0 iii ... ~ Q
ic z "' 0 .. ;:: z C, u IIC ii 0 ... ... .,, .. z "' :E "' IIC u ! IIC
"' m :E :, z Q
it f!?
70
ASPHALT PAVEMENT ( RANDOM TEXTURE J
60
TEST SPEED 60 mph --TEST SPEED 30 mph ---
50
E or F •o
30
20
10
O -==~10-~20~. -c'30',---A0'---5,'-0-60..._____.,70----l....180--90'---l00.L.....P-ER_CJ...EN_T_O...JF
I I i I SURFACE D-0 D-1 D-2 D-3 D-•
NOTE, IF THE DENSITY PARAMETER IS 0-1. 0-2 0A 0-3. ADD TO THE FRICTION WEIGHT FOR PARAMETER E OBTAINED FROM THIS GRAPH THE FRICTION WEIGHTS FOR PARAMETERS A. B ANO F OBTAINED FROM CHART NO. 2, CHART NO. 3 OR CHART NO. •. AE$PECTIVEL V _
Fig.2.3. Friction weights of texture parameter E or F. [Schonfeld (1974)].
whole surface area occupied by projections.
5. Parameter E denotes the sharpness, angularity or roundness of the microtexture
on the projections' surface.
6. Parameter F denotes the sharpness, angularity or roundness of the micro
projections in the background matrix.
The number given to each parameter together with its description is shown in
Figure 2.2. A pavement surface is thus classified by a set of 6 texture parameter
numbers. Using correlation charts of texture code number and skid test results on
asphalt pavements obtained by an ASTM skid trailer or others, the photo-interpreted
skid number of a pavement is obtained as the sum of the texture parameters' friction
weights which, are given at two speeds of 30 and 60 mph.
70
One disadvantage of the method is that it is rather subjective. For example, with
parameter C which specifies the shape of projections, it is required to be given as a
whole number. Suppose the shape of projections is between subangular (C2) and
angular (C3), and the harshness of the projections is polished (El), then (for asphalt
pavements) it can affect the parameter E by 1 scale depending on whether the
parameter C is rounded to 2 (corresponding with E2), or to 3 (corresponding with
E3). A difference of 1 in E parameter number can give a variation of friction
weights up to 25 at 30 mph. (or up to 20 at 60 mph) which is a large variation,
considering the maximum friction weight for E is only 70 at 30 mph (or 50 at 60
mph). See Figure 2.3.
2.2.2. Leu and Henry's Model of Skid Resistance
Leu and Henry [1978] developed a model to predict the skid number (SN) as a
function of speed (V). The model contains two constants which are functions of
macro-texture and micro-texture. The percentage skid number-speed gradient (PSNG)
which has been shown to be a function of macro-texture is assumed to be
independent of speed in the derivation of the model which is given as:
SN= SN0 exp {(-PSNG/lOO)V}
where SN0 = zero speed intercept and is assumed to be a function of micro-texture.
Using the sand patch method to evaluate macro-texture in terms of mean texture
depth (MD), and British Pendulum Number (BPN) to describe micro-texture, they
found that:
PSNG = 4.l(MDr0·47
SN0 = -31 + 1.38 BPN
Hence the skid number is expressed as:
SN = (-31 + 1.38 BPN) exp {-0.041 V (MD)"0·47 }
Good comparison with measured skid number (ASTM E274-77) was obtained at
speeds of 40 and 60 mph. They also found that skid number at any speed can be
related to a measured skid number at one speed and a macro-texture measurement.
The model's advantage as claimed by the authors is that it clearly distinguishes the
roles of macro-texture and micro-texture. It is interesting to note that the model
71
gives good predicted skid resistance, despite the relatively crude method of obtaining
the macro-texture depth. The effects of water film thickness and other test conditions
are not considered here.
2.2.3. Texture Friction Meter
As mentioned in Section l.4.2 (13), Yandell and Mee (with improvement from
Sawyer) have developed a device that can accurately sample pavement surface
texture. Using Yandell's theory of hysteretic friction, the information from the
surface texture is analysed to give the sideways and locked-wheel friction of a
pneumatic tyre for several travelling speeds. This device is still under trial, however,
some meaningful results have been obtained as part of a pavement management
system [Yeaman (1989)]. Figure 2.4 shows relationships between it and BPN and
It is often experienced that the stability and controllability of a vehicle varies
during the progress of tyre wear. Considering the area of contact between tyre and
road surface, it can be expected that generally the tyre stiffness is higher for bald
(smooth) tyres than for new (patterned) tyres. Sutantra and Harrison [1985]
80
developed the expressed lateral force (Fy) in pounds in terms of the slip angle (a) in
degrees, as follows.
For new bias-ply: a= 0.0036S(Fy)1·19408
For bald bias-ply: a = 0.0030852S(Fy)1.199158
For new radial-ply: a= 0.00301003(Fy}1·207861
For bald radial-ply: a= 0.0023636(Fy)1·222203
(2.6)
(2.7)
(2.8)
(2.9)
From equations above, it can be calculated that the increase of c;. with tyre wear
from new to bald is about 12% for bias-ply, and about 13% for radial-ply. It is
expected that for longitudinal stiffness (ex), the similar results will occur since e. is
also associated with the area of contact.
Table 2.6. The ex and eY of truck tyres (10.00-20/F) in 3 states of wear. [Tielking et.al (1973)].
Tyre Progress ex (lb/slip) ey (lb/deg)
New 42,000 523.4
Half Worn 52,000 690.S
Fully Worn 60,000 771.S
For truck tyres, an increase of ex and eY with the wear progress is shown in
Table 2.6. The results show that the increase of tyre stiffness is about 45%, from
new to fully worn. Hence, the effects of tyre wear is more pronounced for truck
tyres than car tyres.
2.4.S. Measurement of Tyre Stiffness and Its Related Parameters
The longitudinal slip stiffness (ex) has been defined as the initial slope of the
pure longitudinal force (F x) with respect to the slip ratio (Sx), at zero slip ratio.
Similarly, the lateral slip stiffness (ey) is defined as the initial slope of the pure
lateral force (Fy), with respect to the slip angle (a), at zero slip angle. Therefore, the
measurement of tyre stiffness can be related with the measurement of Fx or FY on
road surface, during braking or cornering. As the tyre forces, even at near zero slip
81
or slip angle, are different at any set of operating conditions (e.g. load, inflation
pressure, speed, and tyre wear), the tyre stiffness may also be measured at any set of
operating conditions.
The slip tyre stiffness has also been related with the tread element spring rate,
and the length and width of tyre contact [Bernard et.al ( 1977), Sakai ( 1981 ), Gim
and Nikravesh (1990)], as follows.
ex = dFxfdSX = 0.5 kx(21)2w (2.10)
eY = dF /da = 0.5 ky(21)2w (2.11)
where: ex = longitudinal slip stiffness (force/slip, or force/fraction slip).
eY = lateral slip stiffness (force/degree, or force/radian).
kx = longitudinal tread element spring rate, or rate of change of longitudinal
force with respect to deformation (force/length per unit area).
Icy = lateral tread element spring rate, or rate of change of lateral force with
respect to deformation (force/length per unit area).
21, w = tyre contact length & width (unit length)
The measurements of carcass spring rates (Kx, Ky), which will be equal to the
tread element spring rates (kx, Icy) if the carcass is rigid, can be done in the
laboratory either at complete sliding [Thieme et.al (1982)] or without/near sliding
[Tielking and Mital (1974)]. If the deformation without sliding is higher than at
complete sliding for a given horizontal force, the spring rate without sliding will be
lower than at complete sliding. If the contact length & width are not changed, then
the deformation stiffness without sliding will also be lower than at complete sliding.
Examples of the measurement of carcass spring rates are presented below [Tielking
and Mital (1974)].
The measurements of Kx are made with the wheel plane in the direction of flat
bed travel. The deflected tyre is. locked against the flat bed table which is manually
moved in increments of 0.1 inch. Fx readings are taken at each displacement until
the tyre begins to slide. The onset of tyre sliding is evident in the representative data
for a radial tyre shown in Figure 2.8. The Kx is the slope of the linear portion of the
longitudinal load-deflection plot. As seen in Figure 2.8, the slope is essentially
constant for the 900, 1100, and 1600 lb vertical loads. The approximate value of Kx = 1000 lb/inch for that tyre's contact area is obtained at these loads.
82
400
300
200
100
Longitudinal Force Fx (lb)
.1
p • 24 psi
! <~\ Ta.le
.2 .3 .4 Table Travel (in)
-·-·- 700 lb. -x-x-900 lb. -A-4-1100 lb.
------- 1600 lb.
.s
Fig.2.8. Longitudinal force <Fx) versus travel of flat bed table supporting a radial tyre at the indicated vertical loads. [Tielking and Mital (1974)].
SOO
400
300
200
100
Lateral Force F y (lb)
24 psi
• 2 .4
-•-•-700 lb.
-X-X-900 lb.
-A-A-1100 lb.
-0-0-1600 lb.
.6 .8 1.0 1.2
Table Travel (in)
Fig.2.9. Lateral force (F1) versus travel of flat bed table supporting a radial tyre at the indicated vertical loads. [Tielking and Mital (1974)].
83
The Ky is obtained from load-deflection measurements made with the wheel plane
perpendicular to the direction of flat bed travel. Lateral force measurements are
made as the flat bed table is moved in small increments until the tyre begins to
slide. The Ky is the slope of the linear portion of the resulting lateral load-deflection
plot. As seen in Figure 2.9, tyre vertical load has little influence on the lateral spring
rate of a radial tyre. The value of Ky = 500 lb/inch for that tyre's contact area is
obtained from the data shown in Figure 2.9.
2.4.6. Clarification of tyre stiffness formula and its units
It can be seen that the left-hand sides of Equations (2.10-2.11) have the unit of
force/unit slip or force/unit angle, whereas the right-hand sides have the unit of
force. The difference on this unit raises some confusions about the origin of the
stiffness formula. On most tyre models, the unit of tyre stiffness must be equal to
the unitsof frictional forces (see Section 3.4).
The mathematical verification of tyre stiffness formula has been shown elsewhere
(e.g. Schallamach and Turner [1960], Livingston and Brown [1969], and Tielking
and Mital [1974]). Basically, the stiffness formulae are derived from the phenomena
of either elastic deformation (in longitudinal direction) in the forward part of the
contact region, or elastic deformation (in lateral direction) following the path of an
element parallel to the direction of motion. Some of their mathematical derivations
are presented in Appendix G.
Graphically, the parameter Cx is the slope at the origin of the Fx versus Sx data
measured in straight-ahead rolling. It is seen, that the angle between this slope and
the horizontal axis is not really a definite angle, since the vertical and horizontal axis
have different units. Analogous to this, the slope of the Fx versus deflection, on
obtaining the parameter kx, is also not a definite angle due to their different units.
Hence, it seems to be meaningless for correlating these angles.
The problem now arises with the practical application of Equation (2.10), such as
the application on the model for predicting/calculating the frictional forces: what
value of slip should be assigned so that the dimension of Cx will be in force similar
84
to the right-hand side of Equation (2.10).
As a first solution, it is useful to rearranged Equation (2.10) in the following
forms.
C1 Cx = 0.5 kx(21)2w (2.10a) (force) (force)
where c1 is a coefficient (from O to 1) representing the value (or fraction) of slip
assigned; and the dimension on the left-hand side of Equation (2.10a) now is force.
On practical application, the selection of c1 will influence the magnitude of frictional
forces at below the critical percent slip, since this region is controlled primarily by
the elastic tyre properties (see Table 2.7). In most tyre models, c1 = 1 was used.
The second problem, is whether the value of ex is quantitatively equal with the
value of 0.5 kx(21)2w. It is then useful to rearranged Equation (2.10a) in the
following forms.
C1 ex = m1 0.5 kx(21)2w (2.10b) (force) (force)
where m1 is a constant to equate the longitudinal tyre stiffness measured from Fx
versus Sx data (i.e. measurement of ex) and from load-deflection data (i.e.
measurement of kx). The value of m1 will be dependent on many factors, such as the
difference in tyre motions (steady-state, quasi steady-state, non steady-state) or
methods of deformation measurement (without/near sliding, complete sliding). This
will be discussed further in Section 6. 7.
Graphically, the parameter eY is the slope at the origin of the FY versus a. data
measured at zero longitudinal slip. Again, it is seen, that the angle between this
slope and the horizontal axis is not really a definite angle, since the vertical and
horizontal axis have different units. Analogous to this, the slope of the F Y versus
deflection, on obtaining the parameter ky, is also not a definite angle due to their
different units. Hence, again it seems to be meaningless for correlating these
angles.
Analogous to the case in longitudinal stiffness, on practical application the value
of angle should be assigned so that the dimension of eY will be in force similar to
the right-hand side of Equation (2.11).
85
As a first solution, the Equation (2.11) then will be rearranged in the following
forms.
C2 Cy = 0.5 k,(21)2w (2.lla) (force) (force)
where c2 is a coefficient (from O to 1) representing the fraction (from a = x/2) of
the angle assigned; and the dimension on the left-hand side of Equation (2.1 la) now
is force. On practical application, the selection of c2 will influence the magnitude of
frictional forces below the critical slip angle, since this region is controlled primarily
by the elastic tyre properties (see Tables 2.7-2.8).
Unlike the selection of coefficient c1 for longitudinal stiffness, the selection of
coefficient ei for lateral stiffness has been different between one investigator and the
others.Most of them used ei = 0.637 (or slip angle assigned a = 57 deg= 1 radian),
whereas some of them used c2 = 0.011 (or slip angle assigned a = 1 deg). Perhaps,
the ones who select a = 1 radian had followed the previous assumption of small slip
angles, that tana = a can be accepted only if a is in radians. (For example, tan4° =
0.0699 = a = 4° = 0.0698 radian). This will be discussed further in Section 6.5.
The second problem, is whether the value of Cy is quantitatively equal with the
value of 0.5 k,(21)2w. Analogous with the case of Cx, the Equation (2.lla) then can
be rearranged in the following form.
c2 CY = m2 0.5 ky(21)2w (2.1 lb) (force) (force)
where m2 is a constant to equate the lateral tyre stiffness measured from FY vs a data
(i.e. measurement of C,) and from load-deflection data (i.e. measurement of le,). Again, the value of m2 will be dependent on many factors, such as the difference m
tyre motions or methods of deformation measurement.
2.4.7. The Empirical Model for Tyre Stiffness Variation
It is desirable (from the point of view of cost and practicalities), if the tyre
stiffness can be predicted instead of being measured. However, by considering the lack
of data and available information, it is not possible in the present study to predict
(from the tyre properties and operating conditions) the stiffness of the tyres. Instead,
86
it is proposed to predict the variation of tyre stiffness. In other words, it is attempted
to construct a model which will be able to predict the change of tyre stiffness at
any set of operating conditions, once the tyre stiffness at one set of operating
conditions has been measured (known).
From previous information, it is found that several factors will influence the
magnitude of tyre stiffness. In regard to the contribution of each factor, it can be
said that the significant factors are vertical load (F2), inflation pressure (IP), tyre
speed (V), and wear progress. Further studies of the effect of wear rate leads to
assumption that the modelling of this factor is considered less important. One reason
is that the time duration taken from one state of wear to another significant state of
wear is considered long enough. Moreover, the change of wear from one state to
another state is less frequent (in comparison with other factors). Hence, each
significant state of wear (e.g. new, half worn, smooth) can be regarded as a separate
type of tyre.
Following information from: Figures 2.6-2.7 [Schuring et.al (1976)], Table 2.2
[Ervin (1976)], Table 2.3 [Bergman (1977b)], Equations 2.4-2.5 [Sutantra and
Harrison (1985)], Table 2.4 [Tielking et.al (1973)], and Table 2.5 [Bergman
(1977b)], the general form of the mathematical model then can be expressed as
follows.
exe1 = exeo + f(Cxeo, F20, IP0, V 0, Fzi• IP 1, V 1)
where exe = longitudinal tyre stiffness coefficient
eye = lateral tyre stiffness coefficient
F2 = normal (vertical) load
IP = inflation pressure
V = tyre speed
subscript 1 = predicted (for left-hand side)
subscript 0,1 = measured (for right-hand side)
(2.12)
(2.13)
Three forms of the relation between dependent and independent variables are
considered. Eventually, one is selected on the basis of how well it fits the
experimental data.
y =a+ bx+ cx2 (2.14)
87
Y = axb (2.15)
y = abx (2.16)
In general, the curves obtaining from equations above are not so steep, hence if
necessary, a small extension (extrapolation) outside the range of the model's data,
still can be reasonably applied. The characteristics of each curve are as follows
[Yeomans (197 6)]:
1. Equation (2.14) is most frequently used to fit data having a curve shape.
Depending on the sign of coefficients, the shape can be a concave or convex. A
finite value (a constant) of dependent variable is obtained when the independent
variable equals to zero.
2. Equation (2.15) is simpler than Equation (2.14). The deficiency lies in its near
zero value of independent variable. Depending on the sign of b-coefficient, the
curve can have an assymptote to y positive-infinity (concave downward, b<0) or
to y negative-infinity (convex upward, 0<b<l). Accordingly, any extrapolation, if
required, must be restricted not to cover the near zero values of independent
variable. For b> 1 a finite value (a constant) of dependent variable is obtained
when the independent variable equals to zero.
3. Equation (2.16) is similar to Equation (2.15). However, the sign of b-coefficient
must be positive, otherwise there is no real results for non-integer independent
variable, or there is alternately a positive and negative results for integer variable.
For 0<b<l the curve becomes concave downward, and for b>l the curve
becomes convex upward. This model also has a finite value (a constant) when
the independent variable is zero.
From Figure 2.6 [Schuring et.al (1976)], it is seen that the CxC decreases almost
linearly with increasing load (F2). The Equation (2.16) is selected as best suited for
that particular data. The coefficients are established by statistical method.
For bias belted (B) passenger car tyre,
Cx C = 22 (0.9997l2 (2.17)
Assuming that b.CxCICxCo = (CxC1-CxC0)/CxCo is same for bias belted (B), bias ply
(D), and radial ply (R), then the mathematical model for the effect of vertical load
(for passenger car tyres) is given in the following form.
CC = ((0.9997l21 -(0.9997f"'Jc C = ((0.9997f21 - l} C
b, X F X O F x 0 (0,9997) ZO (0,9997) ZD
(2.18)
88
where Fz = vertical load (lb), in the range of extended model's data 225-1350 lb.
Similarly, for lateral tyre stiffness, from Figure 2.7 [Schuring et.al (1976)]:
eye = 0.31 (O.999fz (2.19)
AC C = ( (0.999f'-(O.999f .. )c C = ((0.999f1 -l !r C (2.20) y l (O.999f zo y O (O.999f zo )'---y O
For truck tyres, see Tables 2.2 [Ervin (1976)] and 2.3 [Bergman (1977)], it is found
that the Equation (2.15) is best suited for longitudinal stiffness, whereas Equation
(2.16) is more suitable for lateral stiffness. Accordingly, using coefficients
established by statistical method, the model for the effect of vertical load (for truck
tyres) is given in the following forms. CxC = 288(Fz>-0.39S
AC C = zl zO C = zl - 1 C ((F t0.39S_(F t0.39S} ((F t0.39S }
x (Fzot0.39S x o (Fzo>-0.39S x o
where Fz = vertical load (lb), within the range of model's data 2250-9110 lb.
eye = O.28(O.9994fz
AC C = ( (O.9994fz,_ (0.9994fzo)c C = ((0.9994fz, _ 1 !r C y l (O.9994fzo y 0 (O.9994fzo ry 0
(2.21)
(2.22)
(2.23)
(2.24)
The model for the inflation pressure (IP) effect is best suited with Equation
(2.14). Using coefficients in Equations (2.4) and (2.5) from Sutantra and Harrison
[1985], the ratio of (Cy1-Cy0)/Cyo is first established. Then by assuming ACyC/CyCo = (CyCcCyC0)/CyCo = (Cy1-Cy0)/Cyo• the mathematical model for the effect of inflation
pressure (for passenger car tyres) is given in the following forms.
For bias-ply passenger car tyre,
AC C = ------------ C (4.32(IP1-IP0)- O.O674{(IP1 ) 2 -(IP0)2}f
Y 26.4 + 4.32(1P0 )- 0.0674(IP0 ) 2 Y 0 (2.25)
For radial-ply passenger car tyre,
(5.3(1Pl -IPO) - 0.0916{(IP.}2-(IP0)2})
ACyc = 2 cyc0 33.5 +5.3(1P0)-O.0916(1P0)
(2.26)
89
where IP = inflation pressure (psi), within the range of model's data 10 psi - 40 psi. Similarly, for truck tyres data in Table 2.4 [Tielking et.al (1973)] the ratio of
(Ky1-l(,o)/Kyo is first established. Then by assuming ACyC/CyCo = (CyCi-C,Co)/CyCo =
(Ky1-Ky0)/Kyo, the mathematical model is given in the following forms.
For normal truck tyre (e.g. 11.00-22/G or smaller),
AC C = ---------- C (-14(IP1-IPO) + 0.1 {(IP/-(IP0)2}}
Y 238O-14(IP0) +O.l(IP0)2 Y 0 (2.27)
For wider truck tyre (e.g. 12.00-2O/G or greater),
AC C = ----------- C (-56.5(IP1 -IPO) +0.45{(IP/-(IP0)2}}
Y 354O-56.5(IP0) +O.45(IP0)2 Y 0 (2.28)
where IP= inflation pressure (psi), within the range of model's data 80-100 psi.
From Table 2.5 [Bergman (1977b)], it can be deduced that CyC increases
(non-linearly) with increasing speed (V). By plotting the data of CyC and V, it is
found that the increase is best fitted with a second-order relation (Equation 2.14).
Accordingly, the equations in Table 2.5 can be formulated as follows.
CyC =a+ bF2 + cV + dV2 (2.29)
The coefficients (a,b,c,d) then can be obtained by least square method.
Equation (2.91) may be rewritten, and further simplifies to:
Xs' µF/1-Sx) --------------21 {2-(a/21)-(b/2l)H(C S )2+(C S )2 }0·5
X X y y
108
(2.84)
(2.85)
(2.86)
(2.87)
(2.88)
(2.89)
(2.90)
(2.91)
(2.92)
f.
1,2 Case 3 Case 3 ._ _ __._ _____ ___._s~ 0 Sxcr 1
._ _ ____, _____ ___..._ o(
00 ~,r
Fig.2.19. The contribution of tyre stiffness (C) and the available friction (µ) to the frictional forces (F). At slips beyond the critical value, then J.1s = Jlx (or µa = µy).
Using Equations (2.80) and (2.81), then:
_ CxSx(X1'J F----- + x 1-Sx 21
(µF z cos0) { 1-0.5 (b/21 )-(X1'/2l)}
h-0.5(a/21)-0.5(b/21)}
_ CYSY ( X1' J (µFzsin0){1-0.5(b/21)-(X 5'/21)}
FY - 1-Sx l 21 + h-0.5(a/21)-0.5(b/21)}
Case (3): a ;;:: X5'
(2.93)
(2.94)
From Equations (2.55) and (2.56), it is found that since both pressure and stress are
linear functions of~·. sliding will occur at all points OS~· s a. Hence, Equations
(2.80) and (2.81) reduce to:
Fx = cos0 r wcrmu(X')dX' (2.95)
FY = sin0 f1wcrmu(X')dX' (2.96)
On carrying out these integrations, the following results are obtained:
Fx = µFzcos0 (2.97)
F y = µFzsin0 (2.98)
It is seen that for Case 3 there is no direct contribution of Cx and Cy-
2.6.6. Brief Procedure Using Locked-Wheel BFC
From Equations (2.57) and (2.58), the available longitudinal & lateral friction
coefficients at zero slip (J.1s0 and ~) must be obtained. This can be solved by using
the equations available for the frictional force coefficients.
109
Data: a/21, b/21, ex, Cy, µd, Fz Variable: Sx, a
Setting: a= O"-+ 8 = 0° then from Eq.(2.57): µ = µ, = f1(JI.o, SJ (la)
Using Eq.(2.93), Xa'/21 = fz{JI.o, SJ, and Eq.(la) then Fx = f3(JI.o, SJ (lb)
Differentiate Eq.(lb) and equalize to zero then dFxfdSx = fiJI.o, SJ = 0 (le)
Using Eq.(2.92) and Xa'/21 = a/21 then µ50 = fs(SJ
Substitute Eq.(ld) to Eq.(lc) then dF xfdSx = fiSx) = 0
obtain sx from f6, named sx = sxcr
Substitute Eq.(le) to Eq.(ld) then µ.o = fs(SJ = fs(Sxcr)
(le)
(lt)
Fig.2.20. Determination of the available longitudinal friction at zero slip (Ps0).
Data: a/21, b/21, ex, Cy, µd, Fz Variable: Sx, a
Setting: Sx = 0 -+ 8 = 1t/2 then from Eq.(2.58): µ = µa = g1(Jlao, a) (2a)
Using Eq.(2.94), Xa'/21 = gz<µa0, a), and Eq.(2a) then F = g3(µa0, a) (2b)
Differentiate Eq.(2b) and equalize to zero then dF da = &(µa0, a)= 0 (2c)
Using Eq.(2.92) and Xa'/21 = a/21 then µa0 = gs( a) (2d)
Substitute Eq.(2d) to Eq.(2c) then dF/da = gia) = 0
obtain a from g6, named a = <la (2e)
Substitute Eq.(2e) to Eq.(2d) then µa0 = gs(a) = g5(<Xcr) (2t)
Fig.2.21. Determination of the available lateral friction at zero slip (µao).
110
Data: a/21, bill, Cx, Cy, Pd• Fz Variable: Sx, a
Using procedure (1) and (2), detennine: Ps0, Pa0
J,
Calculate: Ps = Ps0-(µs0-Pd)Sx Pu = Pa0-Wa0-Pd)2<X/1t p = p.+(µ«-pJ28/1t
Calculate X.'/11 using Eq.(2.87) l
Yes J,
Case (1): Calculate Fx, Fy Eqs.(2.88, 2.89)
X,' ~ 21-b
No J,
Calculate X.'/ll using Eq.(2.92)
Yes J,
Case (2): Calculate F x• F y
Eqs.(2.93, 2.94)
Fig.2.22. Calculation of shear (frictional) forces._
1
No J,
Case (3): Calculate F x• F Y
Eqs.(2.97, 2.98)
The boundary conditions to be considered in this model are (Figs. 2.17 and 2.18):
a. The relationship between the available friction coefficients (Jls, µJ and slips (Sx,
a.) is assumed in this study to be linear (Equations (2.57) and (2.58)).
b. The frictional force coefficients (µx, µy) are assumed to equal the available
friction coefficients (Jls, µJ from the locations of the maximum value (at d.FxfdSx
= 0, or d.F/da. = 0) up to the locked-wheel position (at Sx = 100%, or a.= 1t/2).
As mentioned before the locked-wheel BFC is assumed to equal the frictional
force coefficient at a. = 1t/2, whereas the maximum BFC (at d.FxfdSx = 0) is not
necessarily equal to the maximum SFC (at dF /da. = 0). For a given locked-wheel
BFC (µd) the determination of the available friction coefficients at zero slip (µ50, Pa0)
are presented in Figures 2.20 and 2.21. The frictional forces then are calculated
using procedure shown in Figure 2.22. The derivation of mathematical equations will
be given in an illustrative example.
111
2.6.7. Illustrative Example
The following data was taken from Bernard et.al [1977]: Cx = 10000 lbs/slip, Cy
= 220 lbs/deg = 12605 lbs/rad, Fz = 1100 lbs. By using the relationship of
longitudinal force <Fx) as a function of percent slip (Sx), the locked-wheel braking
coefficient (µd) of H78-14 tyre (at 28 psi) is obtained from their curve as 0.87.
A good match between measured data and tyre model will be influenced by an
appropriate choice of pressure distribution (a/21, b/21). From previous information
[Bode (1962), Lippmann and Oblizajek (1974)], it is found that for braking/cornering
mode the ranges of a/21 and b/21 are normally about 0.05-0.20 and 0.10-0.30,
respectively. These values are dependent also on tyre type and operational factors
(e.g. speed, inflation pressure, and load). In this example, the a/21 and b/21 are
selected as 0.06 and 0.20 which are found to give closer agreement between
measured data and the tyre model.
1. Determination of µso• µscr• and Sxcr-
Since Sxcr is located at the boundary between Case 2 and Case 3, equations in Case 2
can be used for the determination of the required parameters.
a. From Equation (2.57):
µ = µs = µso-(µs0-µd)Sx
b. From Equations (2.93), (2.92) and (la):
F = -(CxSx>-1(1-Sx)(µsO-µsOsx +µdSx)2F;
X {2-(a/21)-(b/21)}2
F z { 4-2(a/21) + (a/21 )(b/21) + (b/21 )2- 4(b/21 )} +---------------(µs0-µsOsx +µdSx>- 1 {2- (a/21)- (b/21)}2
c. Differentiate Equation (1 b) to Sx and equalize to zero:
Fig.2.23. Tyre model(--) and measured data (o o o) of H78-14 tyre (28 psi) with a/21 = 0.06 and b/21 = 0.20. Left: Fx & FY vs Sx at a.= 4°. Right: Fy vs a. at Sx = 0.
As can be seen from Tables 2.7 and 2.8, the contribution from tyre stiffness (Cx,
Cy) decreases with increasing Sx (or a.), whereas the contribution from friction
coefficient (µ) continues to increase. At Sxcr (or a.er) there is no direct effect of tyre
stiffness. However, the frictional forces are still affected indirectly, through the
influence on the steepness of the friction coefficient, from Pso ( or Pao) to µd. At Sx = 100% (or a. = 900) the magnitude of frictional forces coefficient then equals to µd
which is independent of tyre stiffness.
A BASIC computer program (see Appendix A) has been written for the routine
calculations of frictional forces. The computing time for obtaining a.er has been
minimised without loosing accuracy of the result. This can be done by providing two
loops for achieving the convergence. The first loop with large increments is intended
to quickly bring the position into the region of convergency. The second loop then
allows smooth increments to be executed within this region. As a result, about 20
seconds CPU time ( using PC A T-286) are required to run the program.
Figure 2.23 shows the comparison of tyre model and measured data of H78-14
tyre (28 psi). This figure demonstrates the extent to which the tyre model agrees
with experimental data.
116
Force (x1000 lbs) 1.2
1
o.a
0.6
0.4
0.2
fx
D +--........ --r-----r--.,--,...........----r-----..-----r--.
0 10 20 30 40 50 60 70 80 90 100
Sx (%)
1.2 Force (x1000 lbs)
1
/-------- F, o.e
: I I
0.6 I I ' ' I
0.4 '
0.2
D +-----r-----r-......... ""'T""----r-r---r--,----,
D 10 20 30 '10 50 60 70 BO 90
0< (deg)
Fig.2.24. Tyre model of H78-14 (28 psi) with b/21 = 0.20, using a/21 = 0.11 (--) and a/21 = 0.01 (------). Left: Fx & FY vs Sx at ex= 4°. Right: FY vs ex at Sx = 0.
Force (x 10 0 0 lbs) 1.2
1
D.B
0,6
0.4
0.2
D +--........ --r-----r--.,--..--....----r-----..-----r----.
D 10 20 30 40 50 60 70 80 90 100
sx (%)
Force (x1000 lbs) 1.2
1
D.B
0.6
' ' 0.4 ' ' I ' ' I
0.2 ' '
D +-----r-----r-......... ""'T""----r-r---r--,----.
D 10 20 30 'ID 50 60 70 BO 90 0< (deg)
Fig.2.25. Tyre model of H78-14 (28 psi) with a/21 = 0.06, using b/21 = 0.01 (--) and b/21 = 0.40 (------). Left: Fx & Fy vs Sx at ex= 4°. Right: Fy vs ex at Sx = 0.
Figure 2.24 shows the sensitivity of using other values of a/21, where the tyre
model using a/21 = 0.01 is compared with a/21 = 0.11. As can be seen, a big
difference is found near optimum value of Sx (for Fx vs Sx at ex = 4°), and near
optimum value of ex (for FY vs ex at Sx = 0). For the curve of FY versus Sx, a small
difference is found near the low and middle values of Sx.
117
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Uy
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
U X
Fig.2.26. The computed relationship between Px and Py at various a. (0 ), with a/21 = 0.06 and b/21 = 0.20.
Unlike a/21, the change of b/21 is found to be nearly small, such as shown in
Figure 2.25. However, a significant difference still occurs for the curve of FY vs Sx,
near the low values of Sx. These results suggest an optional model with unsymmetric
pressure distribution is justified.
Figure 2.26 shows the computed relationship between braking force coefficient
(Px) and sideway force coefficient (µy) at various slip angles (a.). This figure
represents the braking-in-a-tum condition, obtained from braking the tyre up to
wheel lock up. A curve (dashed line) fitted to the ends of. the curves plotted for the
individual slip angles, describes the slide values of the resultant braking-cornering
coefficient. Since the friction coefficient at a. = 900 is assumed equal to the
locked-wheel coefficient, this curve is approximately in the form of a circle through
points Py= µd and µx = µd.
Figure 2.27 shows the computed relationship between braking force coefficient
(Px) and sideway force coefficient (py) at various percent slips (Sx). This figure
represents the turning-in-braking condition, obtained from turning the braked tyre.
As can be seen from this figure, when the tyre is steered until 90° (in theoretical
situation), the Py will reach the value equal to Pd and will be independent from Sx-
where Y is the frictional force coefficient at any chosen combination of variable
levels x1, x2, x3, x4, x5, x6, and x7•
Variables x1, x2, x3, x4, x5, x6, and x7 are the increasing zone of pressure
distribution (a/21), longitudinal stiffness (Cx), lateral stiffness (Cy), locked-wheel BFC
(µd), wheel-load (Fz), percent slip (Sx), and slip angle (a), respectively in
experimental units (e.u.). The 78 a-coefficients consist of: 1 constant, 9 main effects
(7 linear effects, 2 quadratic effects), 33 two-factor interactions (21 linear x linear
interactions, 12 linear x quadratic interactions), and 35 three-factor interactions (35
linear x linear x linear interactiom). See Table 3.l for 288 data u5,ed (2x2"2"2i<2><3x3).
3.3.3. Regression Analysis
The stepwise regression procedure was used when selecting variables in
regression. In principle, any variable which provides a non significant contribution
was removed from the model.
Table 3.2 shows the calculated a-coefficients for the response equation. The R2
and the S (estimated standard deviation about the regression line) give an indication ~ of how well fitted equation explains the variation in the data.
1. Main effect ~ Py will increase with increasing Cr
2. Two factor interaction ~ Py will increase with increasing CY at high values of a/21,
139
at low values of Fz, and at low values of Sx. This increase of Py with increasing
Cy will be reduced at extreme values of a, particularly at low a. 3. Three factor interaction ---+ Py will increase with increasing CY at high values of
a/21 & low values of F z ( or vice versa), at high values of a/21 & low values of Sx
(or vice versa), at high values of a/21 & a (or low values of a/21 & a), at high
values of Fz & Sx (or low values of Fz & Sx), at high values of Fz & low values
of a (or vice versa), and at high values of Sx & low values of a (or vice versa).
d. Locked-wheel BFC (pd)
From Table 3.2:
b.Py = 0.154(pd)
From equation above, it can be concluded that: Py will increase with increasing pd
(main effect). There is no significant interaction effect between Pd and other
1. Main effect ---+ Py will increase with decreasing Fz.
2. Two factor interaction ---+ Py will increase with decreasing Fz at high values of a/21,
at high values of Cy, and at low values of Sx. This increase of Py with decreasing
Fz will be reduced at extreme values of a, particularly at low a. 3. Three factor interaction ---+ Py will increase with decreasing Fz at high values of
a/21 & Cy ( or low values of a/21 & Cy), at high values of a/21 & low values of Sx
(or vice versa), at high values of a/21 & a (or low values of a/21 & a), at high
values of Cy & low values of Sx (or vice versa), at high values of Cy & a (or low
values of Cy & a), and at high values of Sx & low values of a (or vice versa).
1. Main effect --+ µY will increase with decreasing Sx.
2. Two factor interaction --+ µY will increase with decreasing Sx at high values of a/21,
at high values of Cy, and at low values of Fz. This increase of µY with decreasing
Sx will be reduced at extreme values of a, particularly at low a. 3. Three factor interaction --+ Py will increase with decreasing Sx at high values of
a/21 & Cy (or low values of a/21 & Cy), at high values of a/21 & low values of Fz
(or vice versa), at high values of a/21 & a (or low values of a/21 & a), at high
values of Cy & low values of Fz (or vice versa), and at high values of Fz & low
1. Main effect ~ Jlrs1 will increase with increasing Cy.
2. Two factor interaction ~ Pni will increase with increasing Cy at high values of
a/21, at low values of Fz, and at low values of Sx. This increase of Jlrs1 with
increasing Cy will be reduced at extreme values of a, particularly at low a. 3. Three factor interaction ~ Jlrs1 will increase with increasing Cy at high values of
a/21 & low values of Fz (or vice versa), at high values a/21 & low values of Sx (or
vice versa), at high values of a/21 & a (or low values of a/21 & a), at high values
of Fz & Sx (or low values of Fz & Sx), at high values of Fz & low values of a (or
vice versa), and at high values of Sx & low values of a (or vice versa).
d. Locked-wheel BFC (µd)
From Table 3.2:
AJlrs1 = 0.279(µJ
From equation above, it can be concluded that: Pni will increase with increasing µd
(main effect). There is no significant interaction effect between µd and other
Notes: (a), (b), (c), (d), (e), and (f) refers to, respectively, tyre models of: HSRI-I, HSRI-11, HSRI-III, Good Year, Sakai, and the Candidate's. § estimated from: µ = µ0(1-~ V).
152
3.4.4. Model Response Comparisons
The tyre models were compared over a range of S" and Sy values covering braking
at slip angles varying from zero to 16 degrees.
Figure 3.43 presents the longitudinal friction force response of six models
operated at five slip angles. The candidate's model shows a significant difference
with HSRI-1 and Good Year models, and a small difference with HSRI-11, HSRI-111
and Sakai models. In Figure 3.44 the lateral friction force response of six models
operated at four slip angles are presented. The candidate's model shows nearly ihe
same result a$ the Sakai model.
Figure 3.45 compares the interaction of longitudinal force with lateral force. The
candidate's model shows a similar result to the Sakai model in terms of the slight
increase of Py with increasing Px at low slip angles. However, the candidate's model
differs significantly in terms of the separation of the endpoints of the friction curve,
with the Sakai and Good Year models which assume a constant coefficient of sliding
friction.
Figure 3.46 compares the straight-ahead longitudinal friction force response of the
six models. The candidate's model which assumes a friction coefficient decreasing
with slip (or slip speed), shows a decreasing Px as S" increases after the peak, while
the Sakai and &ood Year models show flat responsesafter the peak. Probably, these flat
response are suitable for very rough surf aces, which show a slight increase of Px
with increasing speed, as found in literature review [Sabey et.al (1970)].
Figure 3.47 compares the free-rolling lateral force response of six models. The
candidate's model produces nearly the same curve as the Sakai model, and nearly
the same curve at smaller slip angles as the HSRI-111 and Good Year models.
Fig.3.43. Comparison of longitudinal slip (Sx) versus longitudinal force coefficient (Jlx = FxfFz) of six tyre models operated at the indicated slip angle (degrees).
154
u, 1
O.B
0.6
0.4
0.2
(a) HSRl-1
D .____....,__......____.....__.....____.____. _ _.___.___,____, 0 10 20 30 40 50 60 70 BO 90 l00
s. (%)
u, 1
O.B (c) HSRl-111
0.6
0.4
0.2
D ..__.......___.__.__...._......____.....__....____.____.___. 0 ID 20 30 40 5D 60 70 BO 90 100
Fig.3.44. Comparison of longitudinal slip (Sx) versus lateral force coefficient (µy = F/Fz) of six tyre models operated at the indicated slip angle (degrees).
155
1 u, 1 u,
(a) HSRl-1 (b) HSRl-11
0.8 0.8
0.6 0.6
4 4
M 0.4
0.2 0.2
0 0 0 0.2 0.4 0.6 0.8 0 0.2 D.4 0.6 0.8
u. u.
1 u, 16
1 u, 16
(c) HSRl-111 (d) Good Year 1
0.8 O.B 8
0.6 0.6
0.4 0,4
0.2 0.2
0 0 0 D.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
u. u.
1 u, 1 u, 1& (e) Sakai 16 (f) Candidate 12 12 O.B 0.8
0.6 0.6
4 0.4 0.4
0.2 0.2
0 0 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
u. u.
Fig.3.45. The interaction of longitudinal force coefficient (Px = FxfFz) with lateral force coefficient (}ly = F/Fz) at the indicated slip angle (degrees).
156
1 Ux
0.9
o.e
0.7
0.6
0.5
0.4
0,3
0.2
0.1
0 0 10
C
e
(ab) HSRl-1/11 (c) HSRl-111 (d) Good Year (e) Sakai (f) Candidate
20 30 40 50 60
Sx (%) 70 BO 90 100
Fig.3.46. Comparison of the straight-ahead longitudinal force response of six tyre models.
1
0.9
O.B
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Uy
[ a b) H SR 1-1/ I I (cd) HSRl-111/Gd Year
[e) Sakai [f) Candidate
0 _______ __._ _____ ......._ ____ __,_ _____ _.
0 B 12 16
Slip Angle (deg)
Fig.3.47. Comparison of the free-rolling lateral force response of six tyre models.
157
CHAYfER 4. FIELD MEASUREMENT OF FRICTION
Notation (Used frequently in Chapter 4) BF = braking force (force) SF = sideway force (force) WL = wheel load (force) VBr = voltage of braking force (voltage) VCr = voltage of sideway force (voltage)
VWL = voltage of wheel load (voltage)
4.1. Objectives of The Friction Measurement
The Multi Mode Friction Test Truck (MMFTT) was used to measure the braking
and sideway forces of the tyres on road surfaces. See also Section 4.3.
The main objectives for carrying the friction measurement are:
1. To obtain the parameters needed (e.g. tyre stiffness) for the prediction of
tyre-road friction using the proposed model.
2. To validate the theoretically obtained values of coefficient of friction, by
comparing with the measured data.
In addition, some investigations into the effect of certain operational factors (i.e.
tyre pressure, and the rate of braking) will also be carried out.
4.2. Design of Measurement
4.2.1. Sample Size
Repeatability in skid testing is always difficult to achieve. This is due to, among
the other things, the intrinsic variability of the experimental material (such as road
surface and tread rubber) and the accuracy of the experimental work.
158
From Taneerananon's works [1981], it was found that by using a sample size of
two replicates, the 90% confidence limits averaged for all conditions of friction
coefficient tests is ± 0.035, which represents ± 8.4% of the average Braking Force
Coefficient (BFC) and± 4.9% of the average Sideway Force Coefficient (SFC).
In this present experiment, it is decided that an error of ± 5% is to be tolerated.
Hence, the sample size must be increased. By using the sample size of three
replicates, the 90% confidence limits averaged for all conditions of friction
coefficient becomes ± 0.016 (assuming the same standard deviation), which will
represent ± 3.9% of the average BFC and ± 2.3% of the average SFC.
4.2.2. Randomization
One purpose of randomization is to take into account the possible systematic
variations in the experiment, and thus transform them into independent and random
variations. If randomization is not employed, the quoted measure of reliability may
be biased. Further, any inference would be unsupported by a meaningful probability
statement.
There are, of course, situations in which complete randomization is either
impossible or uneconomical. Whereas, on the other hand, a completely systematic
design will have no statistical back up for proper analysis. Therefore, some
intermediate position between the two extremes of complete randomization and a
strictly systematic design is more realistic. In other words, some degree of
randomization is required for the valid application of statistical analysis.
4.2.3. Choice of Levels Used
In general, when choosing levels for testing, consideration should be given to the
practical side of tests, that is whether the results of the investigation can be extended
or applied to practical cases. On the other hand, the implementation of the chosen
levels may be limited by the available resources or the capacity of the test
equipment, in such cases then some compromise has to be made.
159
For a factorial experiment, such as the investigation into the effect of certain
operational factors, the number of levels to be used should also be considered. A
minimum two levels for each factor is possible, where this will give an estimate of
the average change of the response parameter over the range investigated. If it is
desired to have some idea of the shape of response curve, then of course, at least
three levels would be required.
By considering that the factorial experiment is only the secondary objective in
this friction measurement, it was decided to use only two levels for such an
investigation, in which the quantitative conclusions are considered sufficient.
a. Speed
The normal speed of vehicles travelling on wet road surf aces is about 10 mph -
60 mph. The choice of high speed was fixed by the limitation of test vehicle as well
as safety considerations.
The length of test road section is another consideration in choosing the level of
speed. For example, if a 30 mph (50 kph) is chosen, and a 4 second observation is
needed, then the length required for the measured section is 55 m, which should be
in homogenous condition. By adding at least a 30 m length for the starting section
(to reach the desired speed) and another 30 m length for the stopping section, then
the minimum length of the test road section would be 115 m, which ideally should
be in a straight ahead condition and free from disturbing access.
In this work, for the purpose of reducing the effect of vibrations, a test speed of
10 mph wasused.
b. Type of pavement
The common type of road pavement will be used in this work. That is a
medium-textured bituminous road surface, with maximum aggregate size of 0.5 inch
(see Figure 4.1). The length of the measured section was 25 m. This length enables
around 5 secondsobservation with 10 mph of test speed.
160
c. Rate of braking
The MMFIT has a capability to continously measure the frictional forces (about
10 measurements per second), from zero percent slip until 100 percent slip
(locked-wheel). The effect of braking rate (time interval from brake application to
full braking) will be investigated by using two levels of the braking rate, that is a
fast braking (= 0.5 second) and a slow braking (= 2 second).
d. Slip angle
For the purpose of obtaining the lateral tyre stiffness (Cy), the shape of the
response curve, particularly at low slip angles,
levels of slip angle is a necessity.
is required. A minimum three
It wa~ found from the literature review, that the lateral forces (for mode without
braking) increased drastically from a slip angle of 0° to lQ°, while from 10° to 25°
the lateral forces increased gradually, reaching its peak, and then decreasing
gradually. Therefore, it is expected that between 0° and lQ° the high variability of
lateral forces will be encountered. Hence, if the variability will be reduced into an
acceptable level, the setting up of test wheel into the designated angle must be
carried out with a high degree of accuracy for this range of slip angles.
Considering also the maximum slip angle that can be set up using this
MMFIT, slip angles of Q°, 10°, and 15° were used in this experiment.
e. Wheel load
The normal wheel loads of vehicles are about 500 lb - 1000 lb for cars, and 800
lb - 6000 lb for trucks. Unfortunately, the MMFIT (with the car tyre size of test
wheel) was unable to provide such normal range of wheel loads.
The total weight of engine, on which the test wheel is attached, was about 400 lb
(181.4 kg). However, at the time of experiment, the maximum load that can be
applied to the test wheel, using the available hydraulic force, was only about 200 lb.
Probably, the only advantage of using this low wheel load (200 lb) is its
lower influence on the operation of the test vehicle, such as the influence on the
vehicle speed during the application of the test wheel brake.
161
f. Tyre type
As found in the literature review, many types of car tyre are available for normal
use. In this present study, the 5.20 x 10 inch cross-ply tyres with smooth tread (no
tread pattern) were used.
The choice of smooth tyre was much more influenced by its high contribution to
the traffic accidents on wet roads. Furthermore, it is considered that the use of
smooth tyre will eliminate the effect of tyre wear (due to testing) on the frictional
force results.
The tread rubber of these smooth tyres is made of TR 457 compound which is a
mixture of Styrene Butadine Copolymer and Poly Butadine in the proportion (60/40
SBR/BR).
g. Tyre pressure
The common inflation pressure range for passenger car tyres is 20 psi to 30 psi.
In this present work, the effect of inflation pressure will be investigated using two
levels of pressure, that is 30 psi (207 kN/m2) and 20 psi (138 kN/m2).
h. Water film
The range of water film occurring in practice varies from just wet (shower/light
rain) to flooded (long period of rain). Test results given by Meyer et.al [1974]
showed that there was little or no change of the skid number for water films of 0.02
inch (0.508 mm) or thicker (flooded), for test speed up to 50 mph.
The effect of inaccurate setting up of pump pressure (due to the changing
temperature of pump engine, fluctuation of power voltage, or variation in the test
truck speed) on producing the water film, hence on the results of friction, will be
less significant by using the flooded water level.
The use of this flooded level, will not cause dynamic hydroplaning for 10 mph
test speed, where according to Home and Dreher [1963] this phenomena generally
occurs, on a flooded surface, above the following speeds:
For 30 psi: V = (10.35)~30 = 56.7 mph
For 20 psi: V = (10.35)~20 = 46.3 mph.
162
4.3. The Multi Mode Friction Test Truck
The vehicle was designed by Yandell primarily for use in measuring tyre-road
friction to validate the Texture Friction Meter. The water discharging system was
designed by Taneerananon [1981]. For this present work, a new computerised data
logging system has been installed, to replace the old ultra-violet chart recorder.
The test truck consists of a six tonne Dodge chassis with the retractable test
wheel mounted on the back .. Water tanks for road wetting are mounted behind the
driver's cabin. A 120 cubic ft (3.4 cubic meter) per minute high pressure pump
delivers the water to the road surface. An instrument and operator's cabin is
mounted on the rear of the truck behind the water tanks (see Figures 4.3-4.4). A 2
kilowatt 240 volt generator provides electric power for instruments and for the
hydraulic power unit.
As indicated earlier, despite the capability to measure braking and sideway
forces, there are 3 main limitations of the MM F T T :
1. Inability to turn (continously) the slip angle during the measurement. Hence, it is
not possible to simulate the turning-in-braking condition. Instead, it can only
simulate the braking-in-turning condition.
2. The maximum slip angle is limited up to 15°-20°.
3. The maximum wheel load is lower than that usually used on the passenger car.
4.3.1. The Test Wheel
The transverse engine, gear box, locked differential and right hand front driving
wheel of a Morris Mini Minor is mounted on the rear of the truck on a hydraulic
lowering device. The steerable driving wheel acts as the friction test wheel. The left
hand wheel has been removed and replaced by a disc brake for braking the test
wheel through the locked differential.
The brake is actuated by a solenoid through an adjustable needle valve so that
braking can, if necessary, be applied slowly enough to allow peak braking to be
detected.
163
The Morris Mini suspension was redesigned with no castor angle to allow the
accurate measurement of the forces. The longitudinal, lateral and vertical forces on
the test wheel are sensed by electric transducers and their amplified values are
recorded by a laptop computer.
4.3.2. The Recording Device
An ACRO-400 Data Acquisition and Control Unit system is used for continously
( ~ 10 measurements per second) recording the loads (vertical, longitudinal and
lateral) on the test wheel. Amplified bridge outputs from a six channel amplifier are
fed into the recorder. The test wheel speed sensed as pulses per second, is converted
to voltage and read directly into the recorder. The procedure to operate the Strain
Gauge Amplifier and ACROLOG-400 are attached (see Appendix C and D). The
configuration of the recording device is shown in Figure 4.3.
A device is used to tum the brake on at the commencement of the test strip and
to tum it off at the end of the test strip. Catadioptric targets (see Figure 4.2) are
placed at each end of the test strip facing the passing test truck. A modulated
infra-red photo detector (see Figure 4.6) turns a relay on the first target and off at
the second. The relay operates the on and off position of the solenoid.
4.3.3. The Road Wetting Device
The MMFTT has the capability to produce a belt of water about 6 inch (15 cm)
wide, laid immediately in front of the test tyre. The water is deposited by means of
one of a number of different sized water jets. The water pressure and the jet size are
adjusted so that the water is emerging from the jet at the test truck speed, so there is
no relative velocity between the water and the road, and consequently no splashing.
The flow rate is also adjusted so that the desired water film thickness is laid for a
particular speed. Hence, the flow rate has to be increased as the truck speed is
increased for a given film thickness. A wind shield is fitted around the water
discharging apparatus (see Figure 4.5) to ensure that all water reaches the road in the
desired strip.
164
Fig.4.1. Medium-textured bituminous surface.
Fig.4.2. Catadioptric targets behind the MMFIT.
165
Fig.4.3. The recording device on cabin.
Fig.4.4. Hydraulic and braking controls on cabin.
166
Fig.4.5. Water discharging apparatus & wind shield (in front of test wheel), at near the right side of MMFIT rear axle.
Fig.4.6. Infra-red photo detector ( ..L) and disc brake for test wheel (i), at near the left side of MMFIT rear axle.
167
Fig.4.7. Calibration of vertical, longitudinal and lateral forces .
Fig.4.8. The MMFIT and arrangement for calibration.
168
4.4. Calibration
4.4.1. Wheel Load Calibration
The wheel load was calibrated by using a special designed rig and winch as
shown in Figures 4.7-4.8. The deflections were converted into voltage and recorded
as the load being applied to the test wheel through a spring balance. A wheel load
calibration curve is given in Figure 4.9.
W L (lbs) 400
350
300
250
200
150
1 DO
50
WL = 375VWL
D i.:::;_ _ __._ __ --'---.&......----'-----'---.&......-----1.-----L---_.__ _ __.
0 D .1 0.2 0.3 0.4 0.5 0.6 VWL (voltage]
Fig.4.9. Wheel load calibration curve.
4.4.2. Braking Force Calibration
0.7 0.8 0.9
For calibration of braking force, the test wheel was locked from rotating by the
application of the brake. A belt was wound around the wheel, the loose end of
which was connected to a spring balance (see Figures 4.7-4.8). The horizontal load
(in longitudinal direction) applied to the wheel was noted from the spring balance
reading and the corresponding voltage were recorded on the computer.
Notes: Mean of Recorded speed= 18.27 kph. Mean of Actual speed= 16.07 kph. Correction factor= (16.07)/(18.27) = 0.8796.
Table 5.4 presents the test wheel speed (WS), before and after correction for all
treatments, and the comparison with the actual vehicle speed (VS). This comparison
between the corrected WS and the actual VS, are also displayed in Figure 5.6.
184
Table 5.4. The speeds of Test Wheel (kph) on entering the measured section, before and after correction, compared with the speeds of MMFIT (kph) over the measured section.
0 .9 .-----~~~~~~~~~~~~~~~~~~~~;;;;~~~;;;;:;~:;--------, 18 0.8 vs = 17.26 16
0.7 14
0.6 12
0.5 10
0.4 8
0.3 6 0.2 4 0.1
O 2
-0.1 0 o 10 20 30 40 50 60 10 so
0.9 .--------------------------, 18
0 8 vs = 16.17 Q · 16
0.7 14
0.6 12
0.5 10 0.4
8 0.3
6 0.2
0.1 4
0 2
-0.1 0 0 10 20 30 40 50 60 70 80
Time (x 0.107 sec)
Fig.5.2. Results of IP 30 psi & fast braking, for treatments A (D°), I (10°), Q (15°). Legend: --0- (BFC), -<>- (SFC), 1-- (Wheel Speed), - (Veh. Speed).
186
BFC & SFC Speed (Km/h) 0.8 C 18 0. 7 VS = 15.91 16
0.9 ,----;=========;;;:;:;:;;;;;;;;;;;;;;,;;;;;;;;-------;:;--, 18 VS = 17.57 K
0.8 16
0.7 14
0.6 12
0.5 10 0.4
8 0.3
6 0.2 4 0.1
O 2
-0.1 0 0 10 20 30 40 so 60 70 80
0.9 ,----------------------------, 18 VS = 16.56 s
0.8 16
0.7 14
0.6 12
o.s 10 0.4
8 0.3
6 0.2 0.1 4
0 2
-0.1 0 0 10 20 30 40 50 60 70 80
Time (x 0.107 sec)
Fig.5.3. Results of IP 30 psi & slow braking, for treatments C (0°), K (100), S (15°). Legend: -a- (BFC), -<>- (SFC), L- (Wheel Speed), - (Yeh. Speed).
Fig.5.4. Results of IP 20 psi & fast braking, for treatments E (0°), M (HY'), U (15°). Legend: -o- (BFC), -<>- (SFC), L- (Wheel Speed), - (Veh. Speed).
188
BFC & SFC Speed (Km/ h) 0.8 G 18 0. 7 VS = 15.95 16
Fig.5.6. Comparison between the corrected Wheel Speed (WS) on entering the measured section, and the actual MMFIT speed (VS) averaged over the measured section.
5.2.3. The Actual Percent Slip on Entering the Measured Section
Due to the lesser values of test wheel speed in comparison with the vehicle speed
(especially at slip angles HY' and 15°, as shown in Figure 5.6), it is assumed that
some slips have occurred, even though the brake was not activated. The exact causes
of this slip are not known. One possible explanation is that the lateral forces due to
slip angle create a drag force through the wheel axle, and then increase the frictions
at the joints with the axle shaft. In addition, it is possible that with a low wheel
load, such as used in this experiment, some local sliding can occur on a freely
rolling test tyre drawn by a vehicle.
Furthermore, the changes of assumed constant vehicle speed may also have an
effect on creating a slip, as the test wheel was not rotated by its own power. For
example (see Figure 5.6), at zero slip treatments (A, C, E, G), it can be expected
that the lower values of treatments A, E and G are probably due to an increase
(accelerating) of the test vehicle speed on entering/across the measured section;
190
whereas the higher value of treatment C may be due to a decrease of the test vehicle
speed (decelerating). Ideally, this could be confirmed if the history of the vehicle
speed (particularly when approaching the catadioptric targets) had be recorded. With
only the test wheel speed being recorded, it is rather difficult to reach a conclusion.
Also the test wheel speed itself is based only on 1 second duration, instead of an
instant speed or a shorter duration.
The actual percent slip on entering the measured section can be found by
assuming that the decrease in speed is linearly proportional to the increase in
percent slip. As an example, for treatment A (see Table 5.4), the actual percent slip
was = (16.35-16.30)/(16.35) x 100% = 0.3%. For the test wheel speed greater than
the vehicle speed (i.e. treatment C) it is assumed that no slip occurred, and the
percent slip on entering the measured section is assumed to be zero. The actual
percent slip for all treatments are presented in Table 5.5.
Table 5.5. The wheel speed (WS), vehicle speed (VS), and percent slip (Sx), for each treatment (Tr), on entering the measured section.
Tr ws vs sx Tr ws vs sx Tr ws vs sx A 16.30 16.35 0.3% I 15.20 17 .26 11.9% Q 14.72 16.17 9.0%
C 16.54 15.91 0.0% K 15.87 17.57 9.7% s 15.75 16.56 4.9%
E 15.89 16.06 1.1% M 14.94 16.42 9.0% u 14.71 16.39 10.2%
G 15.55 15.95 2.5% 0 15.43 16.86 8.5% w 14.18 16.06 11.7%
5.3. Transformation Into Percent Slip and Slip Angle
5.3.1. Correction for Wheel Speed Delay
From Figures 5.2-5.5, it can be seen that generally some time has elapse before
the wheel speed begins to decrease, following the significant changes of friction
coefficients. One possible source of this delay is
wheel speed.
191
inaccuracy of the recording
The magnitude of the wheel speed delay (for shifting purposes) will be estimated
as follows.
a. Drawing a line of the average rate of decreasing Wheel Speed (WS) with Time
(T). The values of this average rate, however, are dependent upon the range of
the Time chosen. This uncertainty will be eliminated, by establishing the
following conditions:
- The range of the Time should cover the dominant range of WS. In this case the
range of T which covers ~ 85% of the range of WS is considered appropriate.
- The representative line will be determined by the least squares method. In this
case the coefficient of determination (R2) ~ 0. 70 is regarded as sufficient.
Several lines which satisfied those conditions can be drawn. The line which gives
the minimum range of Time is selected. This line is T = f(WS).
b. Obtaining the last Time at which the percent slip (Sx) is still zero. This can be
done by intercepting the line of (a) above with the line of the Vehicle Speed
(VS). In other words, by substituting WS = VS into T = f(WS), the Time where
the wheel speed is assumed begin to decrease can be obtained.
c. The wheel speed delay then is the difference between the Time in (b) above and
the Time= 10 (when brake is activated).
Table 5.6. The wheel speed delay, for all treatments.
Ranges of data Tr
T ws (kph)
A 16-27 16.60 - 1.63 C 20-32 16.19 - 0.08 E 15-28 15.60 - 1.01 G 18-29 14.97 - 1.14 I 18-22 14.98 - 1.55 K 20-31 14.53 - 0.79 M 19-22 14.91 - 1.81 0 16-30 14.96 - 0.74 Q 15-27 14.43 - 0.10 s 19-29 15.47 - 1.38 u 17-28 13.76 - 1.07 w 15-25 13.99 - 1.64
T = 27.916 - 0.817WS 0.82 4.56 T = 32.017 - 0.769WS 0.94 9.78 T = 28.224 - 0.878WS 0.91 4.12 T = 29.749 - 0.900WS 0.74 5.39 T = 22.675 - 0.266WS 0.71 8.08 T = 30.338 - 0.796WS 0.89 6.35 T = 22.218 - 0.224WS 0.88 8.54 T = 29.659 - 0.869WS 0.93 5.36 T = 27.834 - 0.862WS 0.94 3.90 T = 28.512 - 0.712WS 0.84 6.72 T = 26.702 - 0.812WS 0.80 3.39 T = 27.940 - 0.790WS 0.78 5.25
Average delay (x 0.107 sec)= 5.95 (== 6)
192
Table 5.6 presents the wheel speed delay for each treatment, with average value
of 6 (x 0.107 sec). Generally, it is also found that on average the delay is lower (T = 5.5) for fast braking treatments (AEIMQU), and is higher (T = 6.5) for slow braking
treatments (CGKOSW). While the inaccuracy of the wheel speed recording is the
only possible source known for this delay, it is reasonable to assume that a shift of
less than the speed accuracy (one second duration) can be justified. Hence, a shift of
T = 6 (x 0.107 sec) is used for all treatments.
5.3.2. The Smoothened Technique for Wheel Speed
As can be seen from Figures 5.2-5.5, the decrease of Wheel Speed (WS) with
increasing Time (T) does not follow a smooth relationship. This is due to the fact, as
mentioned in Section 5.2.3, that the Wheel Speed in this experiment is based only
on 1 second duration, instead of an instant speed or a shorter duration.
Many techniques are available for smoothing the data, ranging from the free hand
curves, moving averages, or by using mathematical curves. To minimize an error
due to subjectivity, and to obtain a smooth curve, a mathematical curve then will be
used for the smoothing technique of Wheel Speed vs Time curve.
The form of the mathematical equation will be selected in accordance with the
general trends of the Wheel Speed (WS) vs Time (T) curve. From past experience,
and from the results of WS vs T curve in Figures 5.2-5.5, it can be said that the WS
vs T curve may be approximated by a bell shaped curve. Accordingly, the normal
curve (Gaussian) equation will be used for the smoothing process:
y = ae -o.s(X)2 .... WS = ae --O.S{(T-u)/sl2 (5.1)
The magnitude of (a) in Equation (5.1) is the maximum value of WS. In this case
(a) is selected as the average value of WS from T=lO to T=15. The position of (u)
is the commencement of Time of the WS vs T curve after shifted (hence, u = 16).
The magnitude of standard deviation (s) will be determined from the available WS
data, starting from T = 16, with a number of T data (n) about 14-16 is considered
enough to represent the WS. The equation for obtaining (s) is as follows:
193
l:ln(WS) = n.ln(a) - 0.5(s)"2l:(T-16)2 (5.2)
An example of this smoothing technique is given here for treatment A. From
Table 5.2, it can be found that the average WS from T=lO to T=15 is 16.55 (km/h).
From Figure 5.2, it is seen that the WS has approached the zero value at T=30. The
magnitude of (s) then will be obtained from the WS data, ranging from WS at T=16
until T=30. Hence, the value of (n) is 15. The l:ln(WS) can be obtained as 21.460.
Then ln(a) = ln(16.55) = 2.806. Using Equation (5.2), the magnitude of (s) is found
as 4.959. The normal (Gaussian) equation for treatment A then is as follows: WS = 16.55--0.S{(f-16)/4.959>2 (5. la)
Using the same procedure as above, the normal equations for other treatments
can be found. Table 5.7 presents the parameters of normal equations of Wheel Speed
(WS) versus Time (T) curves for all treatments, in which: (n) is the number of T
data, (a) is the maximum value of WS, and (s) is the standard deviation of WS data.
Table 5.7. The parameters of normal equations of WS vs T, for all treatments.
Tr T (n) (a) (s) Tr T (n) (a) (s)
A 16-30 (15) 16.55 4.959 M 16-30 (15) 14.92 4.494 C 16-31 (16) 16.26 7.949 0 16-31 (16) 15.28 5.948 E 16-30 (15) 15.65 4.841 Q 16-29 (14) 14.53 4.249 G 16-30 (15) 15.16 6.089 s 16-30 (15) 15.70 5.484 I 16-30 (15) 15.09 4.480 u 16-29 (14) 14.23 4.260 K 16-31 (16) 15.82 6.414 w 16-30 (15) 14.15 4.924
5.3.3. Transformation of Time into Percent Slip
As it found in the literature review, percent slip (S,i:) will increase proportionally
as Wheel Speed (WS) decreases. Hence the transformation of Time into Sx can be
done using a simple equation: (VS-WS)NS = (1-Sx)100%, where VS is the vehicle
speed (in which Sx is assumed to be zero), and Sx = 0% is given for WS ~ VS.
Table 5.8 presents the transformation of time into longitudinal percent slip (Sx)
scale for treatment A. The summary of friction coefficients versus Sx, for all
treatments, are presented in Tables 5.9-5.11, and Figures 5.7-5.10.
194
Table 5.8. Transformation into Sx scale of treatment A
Fig.5.15. Determination of longitudinal stiffness (Cx) from the relationship of BFC and Sx (% ), for treatment A (0°).
The longitudinal stiffness coefficient (½C) is equal to the coefficient of (b)
multiplied by 100. Hence, for treatment A the CxC = 0.0388 (per % slip) x 100 = 3.88 (per 100% slip). Using the WL data in Table 5.9 the longitudinal stiffness (½)
is equal to (3.88 x 204.4) = 793 lb/100% slip.
Using the similar approach as described above, the longitudinal stiffness for other
treatments can be obtained. Table 5.15 presents the results of longitudinal stiffness,
from zero slip angle treatments (A, C, E, G).
Table 5.15. Longitudinal stiffness, from zero slip angle treatments (A, C, E, G).
Fz = vertical load (force) 21, w = tyre contact length & width (unit length) ~. Ky = longitudinal & lateral carcass spring rate (force/length per unit area) ~. ky = longitudinal & lateral tread element spring rate (force/length per unit area) SFC = sideway force coefficient
Sx = longitudinal slip (%, or per fraction slip) a = slip angle (degree, or radian)
µd = locked-wheel friction coefficient subscript 0,1 = initial (measured), final (measured or predicted)
6.1. General
In this Chapter, the results from the candidate's experiment will be used for two
purposes. The first is to supply the parameters needed for the theoretical prediction
of tyre-road friction under braking and cornering. The second is to provide field
measurements for comparison with the candidate's theoretical prediction.
By considering some limitations of the experiment performed, such as the low
values of wheel load and the limited ranges of slip angle, another comparison will
also been done. That is to compare the laboratory data, obtained from other worker's
data [Sakai (1982)], with the candidate's theoretical prediction, where the parameters
needed are also supplied by this laboratory data.
217
The verification of tyre stiffness variation model will be shown in Section 6.2.
The input data for theoretical prediction, from both the candidate's and Sakai's data,
will be presented in Section 6.3. The theoretical prediction and its comparison, will
be presented in Section 6.4 for the candidate's experimental results, and in Section
6.5 for Sakai's laboratory data. Section 6.6 presents the theoretical prediction of a
slippery road, whereas the theoretical prediction using the tyre stiffness from load
deflection measurements will be shown in Section 6.7. Finally, the viability and
reliability of the theoretical prediction of tyre-road friction, as a subtitute for actual
skid tests, will be described in Section 6.8.
6.2. Verification of Tyre Stiffness Variation Model
As presented in Section 2.4.5, the empirical model for tyre stiffn~ss variation
consists of several equations (Equations 2.32-2.37), which can be selected according
to the tyre type. In this Section, the Equations (2.32) and (2.34) will be verified with
the data from experiment.
According to Equation (2.32), the predicted longitudinal tyre stiffness coefficient
(CxC1) will depend on: the initial longitudinal tyre stiffness coefficient (CxC0), the
initial wheel load (Fz0), and the present wheel load (Fz1). Referring to Table 5.15, the
average values from field measurements are assumed to be the initial (actual) values,
and the predicted values of CxC1 for treatments A, C, E, and G are calculated based
on their present values of wheel load.
Table 6.1 presents the comparison between the longitudinal tyre stiffness
coefficient (CxC) from field measurements and its predicted values using the
candidate's empirical model. The measured values in columns 2 and 3 are obtained
by using MMFIT, and their average values in the first row are assumed to be the
initial value. The predicted longitudinal stiffness coefficients (CxC1) in column 4 are
obtained based on the initial value, and the corresponding present values of wheel
load (Fz1).
218
Table 6.1. Measured and predicted values of longitudinal stiffness coefficient (CxC).
,a·a ·a ·a Fz (kg) ·a .. a 600 a .. ·D .. D 600 a .a·a ·O •D aa ' ·a .. a a SOO . ' ·a .. D 500 Ca aa .a·a ·O ·D ·a .. a a .. ·a .. D 400 400 Ca ao ·D·C ·a ·a ·D .. a a .. ·a .. D 300 300 a 0 a aD
,D•O . ·C ·D ·a .. a 200 Do a .. ·D .. a 200
a a aD aD ·a·a ·a ·D ·D .. a . . a . . ·D . . D 100 100 aaa a
0 0 10 20 30 40 50 60 70 80 90 100
Sx (%)
Fig.6.1. Input data ( · · a · ·) of Fx versus Sx (at angle 0°), various Fv from laboratory data [Sakai (1982)].
From Table 6.10, it is found that there are some discrepancies between the tyre
stiffness from the laboratory and from the field, which can be attributed into the
following factors.
1. The accuracy of measurement.
The stiffness from the field rely on a few points in the relationship between BFC
(or SFC) and Sx (or a), while the relationship itself was derived from the
transformation of time into percent slip. The accuracy of transformation will
depend on the smoothened technique for wheel speed, and on the correction for
wheel speed delay. The measurement of stiffness from the laboratory, however,
more straightforward, can exhibit scatter results if there are some inconsistensies
in the experiment methods used to obtain such data [Loeb et.al (1990)].
2. The difference on tyre motions.
It is clear that the data for obtaining field Cx (Figures 6.24-6.25) were derived
from the tyre motion which can be regarded as quasi steady-state or non steady
state motions (i.e. slow or fast braking to wheel lock), whereas the data for
243
obtaining field Cy (Figure 6.26) were derived from the tyre motion which can be
regarded as steady-state motion (i.e. tyre rolling with constant braking slip and
constant slip angle). On the other hand, the load-deflection stiffness (both ½ and
Cy) were derived from the static measurements which will be, presumably, close
to steady-state motion and higher than transient motion. A possible explanation is
the effect of non-homogeneous shear resistance between the tyre and road
surface during transient sliding.
3. The different methodsof deformation measurement.
As mentioned in Section 2.4.5, the laboratory measurement of tyre stiffness
without sliding will be lower than the measurement with complete sliding.
Referring to Equations 2.10-2.11, it is more likely that the slip measurement will
be close to the deformation measurement at the condition of complete sliding
[Thieme et.al (1982)] instead of without sliding [Tielking and Mital (1974)]. In
other words, the stiffness derived from the slip measurements (of steady-state
motion) will be close to load-deflection stiffness with complete sliding, but will
be higher than load-deflection stiffness without sliding.
4. The assumption of carcass rigidity.
The tread element spring rates (kx, ky) for calculating the deformation stiffness
(Cx, Cy) are obtained from measuring the carcass spring rates (Kx, Ky), which are
only equal under the condition of rigid carcass. In actual case the carcass was
not perfectly rigid, hence a higher value of tread element spring rates (kx, ky)
should be used. The degree of carcass rigidity in lateral direction is expected to
be lower than in longitudinal direction, hence a greater correction (i.e. addition)
will be required for lateral deformation stiffness (Cy) than for longitudinal
deformation stiffness (Cx).
244
0.8 BFC (at Angle 0)
0.7 D D a
a a 0.6 D ------cr-- -------------y·-o 0.5 a D D
,l 0.4
, a D I
I
0.3 : I . .
0.2 . a a .
I . 0.1 .
: alP 0
-0.1
-0.2 0 10 20 30 40 50 60 70 80 90 100
Sx (%)
Fig.6.24. Field BFe (a) on main experiment road (at angle 0°), and theoretical values by candidate's model: a/21 = 0.10, ex from Field(--) & Laboratory(------).
Fig.6.25. Field BFe (a) on slippery road (at angle 0°), and theoretical values by candidate's model: a/21 = 0.10, ex from Field(--) & Laboratory(------).
Fig.6.26. Field SFe ( <>) on main experiment road (at Sx 0% ), and theoretical values by candidate's model: a/21 = 0.10, <; from Field (--) & Laboratory (------).
From Figures 6.24-6.25 it is seen, that the laboratory-predicted BFe generally
overestimates the candidate's measured field data. Results from Bergman and
Beauregard [1974] show that under steady-state conditions, the lateral slip stiffness
(at a = 0) is greater than that for the transient conditions. It may be expected, that
this phenomenon is also valid for longitudinal slip stiffness (ex). In other words,
under steady-state conditions the slip ex (at Sx = 0) will be greater than that for
transient conditions. Hence, when the deformation ex is close to steady-state
conditions, the laboratory-predicted values will also be higher than that for transient
conditions. Presumably, in this case factor (2) has a dominant effect over factor (3),
From Figure 6.26 it is seen, that the laboratory-predicted SFe generally
underestimates the candidate's measured field data. Since the field eY was derived
from steady-state motion (which is closer to static laboratory measurement), so the
effect of factor (2) is less relevant, and the discrepancy is mainly due to factor (3), i.e
the '5tiffnes5 without 5\iding (in laborator_y) will be lower than the stiffness with complete
Fig.6.27. Field transient BFC (c) on main experiment road (at angle 0°), and theoretical values by candidate model: a/21 = 0.10, Cx from Field (--) & Laboratory (------). Field steady-state BFC (*) are also shown in this figure.
Figure 6.27 presents the comparison between theoretical BFC and transient field
BFC. The steady-state field BFC, which were obtained by using the test wheel
motor, are also shown in this figure. The rotating test wheel (against the direction of
motion) created slips when the MMFTT was running at constant speed without
applying the brake.
It can be seen from Figure 6.27 that the steady-state BFC is generally higher than
the transient BFC. It should be noted that these steady-state values are on the
conservative side, since they are based on overall averages along the measured
section (25 m), while the skid resistance property at the area of below the locked
wheel braking (- 11 m) is actually higher than at the area of full braking (- 14 m).
(See Figure 5.19).
Another possible effect due to the difference on tyre motions, i.e. factor (2)
above, is through the shapes of the pressure distribution (a/21, b/21). Probably at
constant slip (steady-state) the a/21 will be higher (i.e. flatten pressure distribution)
than at slow braking (quasi steady-state) or fast braking (non steady-state). In other
247
words, at steady-state conditions the frictional forces become higher due to the
flatten pressure distribution.
Results from Bode [1962] confirm that the pressure distribution is rather flatter at
low deceleration (see Figure 1.14). On the contrary, results from the candidate's
main experiment show that the slip stiffness at slow braking is rather lower than at
fast braking. So, even if the slip stiffness is slightly decreased in the case of
candidate's results, the frictional forces may be still high if the effect of the flatten
pressure distribution is more dominant than the effect of decreasing slip stiffness
(see Figure 2.24).
Providing that the tread element spring rates (kx, ky) can be measured (obtained),
it can be concluded that for steady-state conditions both the tyre stiffness from slips
and load-deflection measurements with complete sliding can be applied for
predicting the frictional forces. For quasi steady-state or non steady-state motions,
however, a lower value of tyre stiffness or a lower value of a/21 may be required if
the tyre stiffness is derived from load-deflection measurements. For load-deflection
measurements without sliding, a higher value of tyre stiffness may be required.
The introduction of an empirical factor (see Section 2.4.6) for using load
deflection stiffness, to account for: i) other than steady-state motions, ii) other than
measurements with complete sliding, iii) other than measurements of tread element
spring rates, appears to be a practical solution.
6.8. Viability and Reliability of the Theoretical Prediction of Tyre-Road Friction
As a Substitute for Actual Skid Tests
Yandell and Gopalan [1976] used the mechano lattice analogy and the theory of
hysteretic friction to predict the coefficient of friction of a sliding rubber blocks at
one constant speed. A modified portable skid resistance tester was used to measure
the actual coefficients.
248
Taneerananon [1981] and Yandell et.al [1983] extended the precision of the
method used by Holla [1974] for estimating the coefficient of locked wheel friction
of road surfaces at various speeds. In addition he predicted the (maximum) sideways
force coefficients as well as the locked wheel coefficients from surface texture
measurements. The actual coefficients were also mea~ured by the Multi Mode Friction
Test Truck (MMFTT).
The candidate has extended the prediction of frictional force coefficients to cover all
ranges of percent slip and slip angles encountered by motor vehicles. Even though
some discrepancies (due to less controlled factors) were found to exist between the
predictive and the experiment values (measured by MMFTT), the method gave a
generally good agreement and correct trend of the coefficients. Furthermore, using
his modified model, an even c.\01,er agreement was found when the predictive values are
compared with the laboratory data obtained b.Y 5akai [1<182 ], in which all factors are
presumably assumed to be in fully controlled conditions.
The present experiment involved a common type of road pavement (i.e.
medium-textured bituminous surface). Referring to the literature survey, it is most
likely that the friction of the other types of road pavement (with an exceptional of
very rough surface) can also be predicted with reasonably good agreement, as their
behaviour (in respect to percent slip and slip angle) will also have a similar trend
with this present pavement. Recent investigations (see Appendix F), using higher
speeds (20 mph & 30 mph), al,o showed a similar trend to those with the 10 mph
speed used.
One possible difficulty in obtaining a reliable estimation of tyre-road friction, is
the determination of the shape of the tyre pressure distribution, particularly the input
value of a/21. An empirical model for determining this distribution, or a list of
expected pressure distributions under a variety of conditions, may be the practical
solution. This can be achieved by gathering more information about the effects of
the relevant factors (such as: tyre types, wheel load, inflation pressure, and wheel
speed), into the shape of the distribution.
At the present time, the tyre stiffnesses Cx, and Cy can be obtained
experimentally by measuring forces and slips (or deformations), either in a road
249
surface or in a laboratory [Tielking and Mital (1974), Thieme et.al (1982)].
Especially for stiffness measurements on road surfaces, the candidate has developed
an empirical model for the variation of tyre stiffness due to: vertical load, inflation
pressure, and tyre speed.
Attempts to predict tyre stiffness directly from other tyre properties or related
parameters are left for future refinement. For stiffness measurements in the
laboratory, the need for applying empirical models is probably less beneficial than
the accuracy gained with direct measurement on different variation (vertical load,
inflation pressure). An empirical factor may still be required, however, if there is a
sufficient change of vehicle speed so that it has a significant effect on tyre stiffness.
The theoretical method of obtaining the coefficient has the advantages of low
capital cost and relatively more safety when compared with measurements made
by a skid testing vehicle. More advantages will be gained if the method of
obtaining the locked-wheel braking coefficient (or maximum sideway coefficient),
from surface texture data [Taneerananon (1981) and Yandell et.al (1983)], has been
further improved and established. The Yandell-Mee Texture Friction Meter Mark II
[Yandell and Sawyer (1992)] is at least as reliable as direct measuring methods.
Above all, the theoretical method would yield values of coefficient of friction which
have less degree of difficulty and potential errors associated with the field
measurement by actual skid tests.
250
CHAPTER 7. CONCLUSIONS
7 .1. Theoretical Findings
7 .1.1. Types of Tyre Stiffness
For both longitudinal and lateral directions, there are two types of tyre stiffness:
force versus slip stiffness (i.e. Cx = dFxfdSx at Sx = 0, and Cy = dF/da at a = 0),
and force versus deformation stiffness (i.e. Cx = 0.5 kx(21)2w, and Cy = 0.5 ky(21)2w).
The equality between the slip stiffness and the deformation stiffness has been
mathematically proved elsewhere (see Appendix G).
Two problems were found, however, with the practical application of tyre
stiffness. The first problem is the value of slip or slip angle to be used in the
calculation of slip stiffness, so that the dimension of slip stiffness will be in force,
similar to the dimension of deformation stiffness and the dimension of frictional
forces. The second problem is that the quantitative relation between the slip stiffness
and the deformation stiffness may show the two to be unequal.
On the first problem, it was found that the assignment of 100% longitudinal slip
and 9()° slip angle into the slip stiffness calculation, generally gives satisfactory
results in terms of the agreement between the measured and the predicted values of
frictional forces (see also Section 7 .2.10). Other workers used empirical and lesser
values of slip stiffness (see Section 2.4.6).
On the second problem, it was found that the difference in tyre motions (steady
state, quasi steady-state, non steady-state) will influence the magnitude of slip
stiffness (see Figure 6.27); while it was expected that the methods of deformation
measurement (without/near sliding, complete sliding) will influence the magnitude of
deformation stiffness. The candidate suggested the introduction of an empirical
factor to account for different types of tyre stiffness when comparing either slip
stiffnesses or deformation stiffnesses (see also Section 7.2.12). For example it can be
seen in Figure 6.27 that steady-state slip stiffness is greater than transient slip
stiffness.
251
7 .1.2. Description of Friction Coefficient
The candidate deduced from the literature two types of friction coefficient: the
available (or nominal) friction coefficient (µ) which can be resolved into the
longitudinal (µJ and lateral (µJ components (see Equation 2.59), and the frictional
force coefficients (µx, µy).
The available friction coefficients (Jls, µJ has been assumed to be linear functions
of sliding speed or slips (see Equations 2.57-2.58), while the frictional force
coefficient is zero at zero slip (see Figures 2.17-2.18). At slips beyond the critical
value, i.e. complete sliding without any elastic deformation, the frictional force
coefficient equals the available friction coefficient. The available friction coefficient
when a = rc/2 is assumed to equal the locked-wheel coefficient (µd).
7 .1.3. The Role of Tyre Stiffness
Simulation using the candidate's tyre model suggests that the magnitude of the
tyre-road frictional force is significantly influenced by the stiffness of the tyre, both
in longitudinal (½;) and lateral (Cy) direction. In general, it means that the skid
resistance is not only the property of the road pavement. The tyre stiffness can also
be regarded as independent of pavement/lubricant conditions. This applies to both
types of stiffness: slip stiffness and deformation stiffness. (Available friction
influences frictional force for all percentages of slip; see Figure 2.19).
It was found ( see Figure 2.19, Tables 2. 7-2.8), that the direct contribution of tyre
stiffness to the frictional forces is in the area of below the critical percent slip (Sxcr)
and critical slip angle (Cler). The contribution from tyre stiffness decreases with
increasing percent slip or slip angle, whereas the contribution from friction
coefficient (µ) continues to increase.
The candidate has shown algebraically (see Section 2.6. 7) that beyond the critical
percent slip or slip angle, the tyre stiffness still contributes, indirectly, to the
frictional forces. The effect is through the steepness of decreasing frictional forces
beyond the critical percent slip or slip angle area. He also showed that the
252
magnitude and the position of critical percent slip and slip angle are also determined
by the tyre stiffness. The role of tyre stiffness then is diminished when the condition
of 100% slip (locked-wheel) or 90 degree slip angle was achieved.
7.1.4. Model for Tyre Stiffness Variation
Unlike most vehicle components whose mechanical properties do not vary with
operating conditions, it is found that the tyre stiffness is strongly influenced by basic
operating conditions such as: wheel load, inflation pressure, speed, and tyre type.
From the point of view of cost and practicability, it is desirable that the tyre stiffness
be predicted using empirical formula, instead of being measured at each condition.
Due to the lack of data and available information, however, it is not possible in
the present study to predict tyre stiffness (from tyre properties and operating
conditions). Instead, the candidate has predicted the variation of tyre stiffness. In
other words, the candidate attempted to construct a model (see Equations 2.32-2.37)
which will be able to predict the change of tyre stiffness for any set of operating
conditions, once the tyre stiffness for one set of operating conditions has been
measured (known).
The proposed empirical model (see Section 2.4.7), incorporates the significant
factors influencing the tyre stiffness for each tyre type: wheel load (Fz), inflation
pressure (IP), and tyre speed (V). Data for this model were gathering from the
literature survey. The factor of wear-rate is not included in the model, due to the fact
that the time duration, taken from one state of wear to another significant state of
wear, is considered long. Hence, each significant state of wear (e.g. new, half worn,
smooth) is regarded as a separate type of tyre.
7.1.5. The Shape of Tyre Pressure Distribution
Contact pressures between tyre and ground are not necessarily uniform, mainly
because of the tyre detail design and partly because of the road surface irregularities.
Based on the measured data from the literature, however, only the pressure
253
distribution across the width of a tyre is normally treated as uniform. To represent
the pressure distribution down the length of a tyre, some forms are available, such
as: elliptical, hyperbolic, and symmetric trapezoidal.
Certain factors are found to significantly influence the shape of pressure
distribution. Among the other things are: tyre type, wheel load, inflation pressure,
tyre speed, and mode of operation (braking, driving, cornering). It is also found that
the shape of pressure distribution is not necessarily symmetrical, due to the
influences of tyre speed and mode of operation.
The candidate has proposed a model for the shape of tyre pressure distribution in
a form of a trapezium, with contact length of 21, and slopes length of a and b. This
modelled pressure distribution is intended to offer flexibility in representing many
types of actual pressure distribution (as found in the literature review), by selecting
the appropriate value of its slope (a/21, b/21). In comparison to symmetric trapezoidal
[Bernard et.al (1977)], this model provides the user with one additional option of an
unsymmetric trapezoidal distribution. The maximum normal pressure over the
contact patch then is given by: P max = Fj{21-0.5(a+b) }w, where Fz is the vertical
(wheel) load, and w is the width of the contact patch. By taking a = b, the
symmetric trapezoidal distribution is obtained; whereas by taking a = b = 0, there
will be a uniform pressure distribution.
7 .1.6. Calculation of Frictional Forces
The frictional forces (Fx and Fy) of an automobile-tyre travelling over a road
surface, are believed to be dependent upon many factors [Holmes et.al (1972)] which
have, on each, various degrees of significance. These factors can be grouped into: (a)
operational, (b) tyre, (c) lubricant, and (d) pavement. By incorporating those factors,
an analytical model has been developed by the candidate, which enables the
calculation of frictional forces over all ranges of percent slip and slip angles.
The proposed model is based mostly on the classical analysis of the rolling tyre
by Fiala [1954]. Some of the general mathematical equations are derived in a similar
manner lo other investigators such as Dugoff et.al [1969, 1970], and Bernard et.al
254
[1977]. By using the differential approach, however, the candidate was able to
reduce the friction parameters required for predicting the frictional forces. The
benefit of the candidate's model will be shown in Section 7.3.
7.1.7. Theoretical Effect of Some Input Parameters on the Tyre-Road Friction
The effect of some input parameters (½, ey, a/21, b/21, µd, and Fz) are studied
theoretically using the candidate's model explained in Section 2.6. The ranges of
parameter were selected to represent the normal values which commonly occur in
practice. The variables which significantly affect the tyre-road friction were identified
by means of regression analysis.
(1) Effect of longitudinal stiffness (ex)
The main effect is the increase of BFe with increasing ex- The rate of increasing
BFe will be greater at low values of Sx and a. In addition, the magnitude of critical
longitudinal friction coefficient (Pscr) will also increase with increasing ex, and the
position of critical percent slip (Sxcr) will move into the lower value. The increase of
SFe with increasing ex is found to be insignificant. At a = D° there is no effect of
Cx on SFe.
(2) Effect of lateral stiffness (ey)
Similar to ex, the main effect is the increase of SFe with increasing eY. The rate
of increasing SFe will be greater at low values of a and Sx. Additionally, the
magnitude of critical lateral friction coefficient (µacr) will also increase with
increasing Cy, and the position of critical slip angle ( acr) will move to a lower
value. The increase of BFe with increasing eY is found to be insignificant. At Sx =
0% there is no effect of eY on BFe.
(3) Effect of increasing-zone of pressure distribution (a/21)
The main effect is the increase of BFe and SFe with increasing a/21 (decreasing
leading steepness, as shown in Figure 2.16). The rate of increasing BFe will be less
at extreme values of Sx, particularly at high values of Sx. Similarly, the rate of
increasing SFe will be less at extreme values of a, particularly at high values of a.
255
(4) Effect of decreasing-zone of pressure distribution (b/21)
The effect of varying b/21 on the BFC and SFC has been found to be nearly
small. The significant effect is obtained for SFC at low values of Sx and a., where
the SFC decreases with increasing b/21 (decreasing trailing steepness, Figure 2.16).
(5) Effect of locked-wheel BFC (µd)
The main effect is that the BFC and SFC, as well as the magnitude of critical
longitudinal friction coefficient (PsCI') and critical lateral friction coefficient (µa.a-),
will increase with increasing µd. The rate of increasing BFC will be less at low
values of Sx (and high values of a.). Similarly, the rate of increasing SFC will be less
at low values of a. (and high values of Sx). In addition, the position of critical
percent slip (Sxcr) and critical slip angle (a.er) will move into the higher value with
increasing Pd·
(6) Effect of wheel load (Fz)
The main effect is that the BFC and SFC will decrease with increasing Fz. The
rate of decreasing BFC will be less at extreme values of Sx. Similarly, the rate of
decreasing SFC will be less at extreme values of a..
7 .1.8. Theoretical Model Response Comparisons
In order to identify the effects of various assumptions made during derivation, the
candidate's model has been compared with other tyre models: HSRI-I, HSRl-11,
HSRI-III, Good Year, and Sakai. All tyre models were compared over a range of Sx
and Sy values covering braking at slip angles varying from zero to 16 degrees; and
by using the same input data for tyre stiffness and wheel load, approximately the
same friction coefficient, and different forms of pressure distribution (see Table 3.3).
In general, the candidate's model was found to be give median results relative to the
other model's results.
For the percent slip (Sx) versus braking force coefficient (µx) at five slip angles
(see Figure 3.43), the candidate's model shows a significant difference with HSRI-1
and Good Year models, and a small difference with HSRl-11, HSRI-ID, and Sakai
models. For the Sx versus Px at angle O (see Figure 3.46), the candidate's model
256
which recognises the possibility that a friction coefficient decreases with slip or slip
speed, shows a decreasing Px as Sx increases after the peak, while Sakai and Good
Year models show flat response after the peak. Probably these flat responses are
suitable for very rough surfaces, which show a slight increase of Px with increasing
speed, as found in literature review [Sabey et.al (1970)].
For the percent slip (Sx) versus sideway force coefficient (µy) at five slip angles
(see Figure 3.44), the candidate's model shows nearly the same result as the Sakai
model. For slip angle (a) versus µY at Sx = 0% (see Figure 3.47), the candidate's
model produces nearly the same curve as the Sakai model, and nearly the same
curve at smaller slip angles as the HSRI-111 and Good Year models.
For interaction between braking force coefficient (Px) and sideway force
coefficient (µy) at five slip angles (see Figure 3.45), the candidate's model shows a
similar result to that of Sakai's model in terms of the slight increase of µY with
increasing Px at low slip angles. However, the candidate's model differs significantly
in terms of the separation of the endpoints of the friction cur,we, wit.tit.he Sakai and
Good Year models which assume a constant coefficient of sliding friction.
7.2. Experimental Findings
7.2.1. Rectification of Experimental Problems
(1) Correction for Test Wheel Speed
It is known that the rolling radius, on which the recorded wheel speed of the
MMFIT (Multi Mode Friction Test Truck) was based, can be affected by many
factors, such as tyre size. Therefore, the recorded test wheel speed -..as considered as
an approximation only, and need of correction. A procedure has been set up for the
correction. The basic principle in obtaining the correction factor was that the speed of
the MMFIT over the measured section, was assumed to be e1ual with the wheel speed at
approaching the first catadioptric target (at zero slip angle, free rolling, and start of
brake activation).
257
For slip angles other than zero degree, another correction was also carried out.
First, the correction factor at zero slip angle (as obtained above) was applied to the
recorded test wheel speed, to obtain the corrected test wheel speed in the direction
ofthewheelplane. Second, this corrected test wheel speed then is divided by cosa
(where a slip angle), to obtain the corrected test wheel speed in the direction of
travel.
(2) Correction for Wheel Speed Delay
It was found from experiment that generally some time had elapse before the
wheel speed began to decrease, following a significant change of friction
coefficient. One possible source of this delay was the inaccuracy of the recording
wheel speed (the speed is based on 1 second duration, instead of an instant time or a
shorter duration). A procedure then was set up to estimate the correction needed
(Section 5.3.1). The result of the correction factor, however, was found to be too
small. A slightly greater correction then is seen to be a practical solution.
(3) The Locked Wheel Braking Force Coefficient of Friction
It was found that the locked-wheel BFC tended to decrease with increasing
duration of application (see Figures 5.2-5.5). Also a small increase of the
locked-wheel SFC was found with increasing duration of brake application. It has
been concluded, that the decrease of slip angle due to the looseness of wheel angle
wa&only a minor cause for decreasing the locked-wheel BFC during the full braking,
whereas the decreasing slip angle was seen to be the primary source for increasing the
locked-wheel SFC. A lower skid resistance in that particular section of road
contributed slightly to the decrease of locked-wheel BFC.
(4) Local Sliding During Free Rolling
It was found for slip angles other than zero, that the wheel speed (after
correction) on entering the measured section was lower than the average speed of the
258
vehicle over the measured section. One possible cause was the occurrence of local
sliding during free rolling (before brake activation). This may be due to the fact that
the wheel load during brake off (before brake activation and after brake release) was
significantly lower than the wheel load during brake on (see Figure 5.20).
With the difference in speed mentioned above, some slip by definition had
occurred, even though the brake was not yet activated. One possible explanation for
this slip was that the lateral forces due to slip angle creates a drag force through the
wheel axle, and then increases the frictions at the joints in the axle shaft.
7 .2.2. The Significance of the Variation of Tyre Pressure and Rate of Braking
The significance of the following effects was determined by analysis of variance
as described in Section 5.3.4. All the data was gathered by the candidate using the
Multi Mode Friction Test Truck (MMFTT) on a medium-textured bituminous road
surface (see Figure 4.1), with the conditions of: 30/20 psi tyre pressure, smooth tyre,
flooded water, and 10 mph average speed.
(1) Effect of Tyre Pressure
a. Braking Force Coefficient (BFC)
The main effect is the decrease of BFC with increasing tyre pressure. This effect
is significant for angles 0° and 10°, but insignificant for an angle of 15°. The
interaction effect of tyre pressure and percent slip was found to be insignificant. The
interaction effect of tyre pressure and the rate of braking is insignificant for angles
0° and 10°, but significant for angle 15°. The percentage of decrease of BFC with
increasing tyre pressure is greater at a slow rate of braking.
b. Sideway Force Coefficient (SFC)
Theoretically, there will be no effect of tyre pressure on SFC for angle Q°. The
result of the analysis variance for angle Q° therefore should be read as a mechanism
of deviation from ideal conditions (such as the symmetrical or vertical position of
the tyre to the pavement). For angles lQ° and 15° the effect of tyre pressurewasfound
to be insignificant. The interaction effect of tyre pressure and percent slip, and the
interaction effect of tyre pressure and the rate of braking were also found to be
insignificant.
259
(2) Effect of the Rate of Braking
a. Braking Force Coefficient (BFC)
The main effect was the increase of BFC with increasing rate of braking. This
effect was significant for angles 0°, 10°, and 15°. The interaction effect of the rate of
braking and percent slip was also significant for angles 00, 10°, and 15°. The
percentage of increase of BFC with increasing rate of braking is greater at the
middle percent slip. As mentioned before, the interaction effect of the rate of braking
and tyre pressure is insignificant for angles 0° and 100, but significant for an angle
of 15°. The percentage of increase of BFC with increasing rate of braking is
greater at high tyre pressure.
b. Sideway Force Coefficient (SFC)
The main effect was the decrease of SFC with increasing rate of braking. This
effect was significant for angles 100 and 15°, but insignificant (as theoretically
expected) for angle 00. The interaction effect of the rate of braking and percent slip
was significant for angles 00 (should be negligible) and 15°. The percentage of
decrease of SFC with increasing rate of braking is greater at the middle percent
slip. The interaction effect of the rate of braking and tyre pressure is significant for
angle 0° (should be negligible), and insignificant for angles 100 and 15°.
Similar results regarding the effect of the rate of braking could not be found in
the literature. It may be concluded that increasing the rate of brake application will
increase braking and decrease sideway force for the above conditions.
7.2.3. Verification of Tyre Slip Stiffness Variation Model
The resultsu5ingthecandidate's measured field data (Tables 6.2-6.3) show a good
agreement between the candidate's predicted and the measured values, either for
longitudinal or lateral tyre stiffness, with discrepancies less than 10%. In addition, it
seems that for even closer agreement, a factor involving of the rate of braking could
be incorporated in the longitudinal tyre stiffness variation model.
260
7.2.4. Comparison Between Measured and Theoretically Predicted Friction
It was found that for BFC the predicted values w-ere generally lower than the
experimental values at the middle-lower percents slip, but show a good agreement at
the middle and high percents slip. For SFC, on the other hand, the predicted values
show a good agreement only for high percents slip, but the predicted values are
higher than the experiment values for middle percents slip, and lower than the
experimental values for low percents slip.
Some of the above discrepancies may be due to some poorly controlled
experimental conditions such as: the looseness of the test wheel position, and the
delay of wheel speed measurement response. It was thought that closer agreement
could be achieved by using a smaller slip angles after a certain percent slip (say,
after Sx ~ 30%) on theoretical prediction, and by using a greater correction of the
wheel speed delay (say, 7 x 0.107 second) on the measured data.
The above improvements, however did not affect the discrepancies of the SFC
versus slip angle at zero percent slip. In other words, the lateral stiffness was only
slightly affected by this improvement, and so the theoretical values of SFC versus
slip angle at zero percent slip remain lower than the measured values. A
modification of the prediction model is thus required for the input values of lateral
stiffness. The effect of this will be discussed later.
7.2.5. Comparison of Sakai's Laboratory Data With the Candidate's Theoretical
Prediction of Friction
It was found that the predicted values, especially for data of Fx versus Sx, are
generally in a good agreement with the laboratory data. In most cases, the predictive
values with a/21 = 0.06-0.10 can be used for low wheel loads (Fz), whereas a/21 =
0.06-0.08 are suitable for higher wheel load. The selection of a/21 was seen to be
more sensitive with increasing the tyre stiffness (both longitudinal and lateral). For
the data of FY versus slip angle (ex), particularly for lower values of slip angle, the
predictive values under estimated the laboratory data. This is seen in Figures
261
6.14-6.19. The zero and 100% slip values of friction in the prediction were made equal to
Sakai's laboratory values.
7.2.6. The Effect of Modifying the Quantification of Lateral Stiffness in the
Candidate's Friction Prediction
The candidate has used the laboratory data of Sakai [1982.], in which all factors
are assumed to be in fully controlled conditions, to examine his model.
The predicted values of lateral forces (Fy) are less than both the candidate's
experimental results and Sakai's laboratory data. A modification of the candidate's
model seemed to be a necessity. This has been done by using a higher lateral tyre
stiffness (Cy) in the prediction.
In the modified model, the input of lateral stiffness (Cy) is force/1.57 radians (or
force/ 90°), instead of force/radian (or force/57.3°) as in original model. This
modification did not influence the predicted values of longitudinal force (Fx) versus
Sx at angle 0°. However, by using the modified model, a closer agreement between
the predicted and the measured values was achieved.
This modified model also produced more consistent or similar results between FY
versus Cl at Sx = 0 (see Figure 6.20) and Fx versus Sx at ex = 0° (see Figures 6.14,
6.16, and 6.18). In addition, the full range of ex (00 S ex S 90°), as a result of
modification, has an analogy with the full range of Sx (0% S Sx S 100%) within the
model.
7.2. 7. Independence of Tyre Stiffness from Road Conditions
It was found, that the slip type tyre stiffness obtained from the main experiment
can be used to predict the BFC and SFC on other pavements (i.e. slippery road) with
reasonably good agreement to the experimental data provided that the locked-wheel
BFC or maximum SFC are measured on the other road. The experimental data on
this slippery road were gathered by Multi Mode Friction Test Truck (MMFTT) with
262
the same tyre and operational conditions used in the previous pavement experiments
(main experiment).
7 .2. 8. The Relationship Between Slip Stiffness and Deformation Stiffness
It was found that there are some discrepancies between the values of tyre
stiffness obtained from the field (i.e. slip measurement) and from the laboratory (i.e.
deformation measurement). The discrepancies have been attributed to the factors of:
accuracy of measurement (number of points, corrections, and inconsistencies),
difference on tyre motions (steady-state, quasi steady-state, and non steady-state),
different methods of deformation measurement (without sliding, and with complete
sliding), and the inclusion of the carcass spring rates instead of the tread element
spring rates. It should be noted that the two types of stiffness have been proved
theoretically equal at zero friction force (see Appendix G).
Providing that the tread element spring rates (kx, ky) can be measured (obtained),
it has been suggested that for steady-state conditions both the tyre stiffness measured
from slip and deformation (with complete sliding) are nearly equal, and can be
applied for predicting the frictional forces. For quasi steady-state or non steady-state
behaviour, however, while slip stiffness gives an accurate prediction of friction
measurement the use of deformational stiffness over predicts measured friction. A
possible explanation is the effect of non-homogeneous shear resistance between the
tyre and road surface during transient sliding. Hence, for other than steady-state
motions, a lower value of deformation stiffness is required to predict the frictional
forces. For deformation measurement without sliding, a higher value of deformation
stiffness is required to predict the frictional forces accurately. (See also Section
2.4.5).
The introduction of an empirical factor for deformation stiffness, to account for:
i) other than steady-state motions, ii) other than measurements with complete sliding,
iii) other than measurements of tread element spring rates, appears to be a practical
solution.
263
7 .3. Benefit of the Model
The candidate's tyre model for calculating the tyre-road frictional forces has the
following main benefits.
1. The model requires one friction parameter less than that required by the semi
empirical model developed by Bernard et.al [1977] or the analytical model
developed by Gim and Nikravesh [1990, 1991]. Hence, the time or cost
associated with the measurements (and calculation) of the friction parameters can
be reduced.
2. The model is able to calculate frictional forces from tyre stiffness (as
representing the factors of operation and of tyres), and either locked-wheel BFC
or maximum SFC (as representing the factors of operation, of tyres, of lubricant,
and of the pavement). (See Figures 7.1-7.2). For all calculations of frictional
forces, the secondary inputs needed are: the assumed shape of pressure
distribution (a/21, b/21), and the applied wheel load (Fz).
3. The candidate's model, with a higher lateral stiffness, is more precise than others
since the true stiffness at zero frictional force is used rather than an empirical
and lesser value. (See Section 2.4.6).
4. Since the locked-wheel BFC (or maximum SFC) can be predicted from surface
texture measurements [Yandell et.al (1983)], this model can be used to predict
the tyre-road frictional forces for various degrees of slip and slip angle (including
critical or peak friction), merely from tyre stiffness and surface texture data.
5. Since the equipment for measuring the locked-wheel BFC or maximum SFC is
commonly available (such as the trailer used on ASTM-E274, and SCRIM), the
data from this equipment can be processed directly to obtain the tyre-road
frictional forces for all ranges of percent slip (Sx) and slip angle (a). That is the
behaviour of combined braking and cornering friction can beeasily predicted, as
shown in Figure 7 .2.
264
D
I predicted from
the model from c., locked-w hee I BF C
locked-wheel BFC [measured)
~ Slip (5.c) 100% 0
c, (measured) predicted from
the model /
max SFC (alternative input)
SFC at 90° = locked-wheel BFC
[measured)
Slip angle (oc) 90°
Fig. 7 .1. Inputs & outputs of the candidate's model under either braking or cornering. Left~ Input: ex, locked-wheel BFe. Output: Fx vs Sx (at angle 00). Right~ Input: ey, locked-wheel BFe (or max SFe). Output: FY vs a (at Sx 0%).
Input: c., c,, locted-wheel BFC [or max SFC)
predicted from / the model
Input: Cx, c,, locked-wheel BFC (or m11x SFC)
predicted from / the model
~ ~ Fig.7.2. Inputs & outputs of the candidate's model under both braking and cornering. Left~ Input: Cx, Cy, locked-wheel BFe (or max SFe). Output: FY vs Fx (various Sx). Right~ Input: ex, ey, locked-wheel BFe (or max SFe). Output: FY vs Fx (various a).
265
7.4. Summary of Conclusions
1. It was found that the application of slip stiffness (i.e. ex = dF xfdSx at Sx = 0, and
Cy = d.F/da at a= 0) using 100% longitudinal slip and 90° slip angle, generally
gives satisfactory results in terms of the agreement between the measured and the
predicted values of frictional forces. The slip stiffness used is the tangent of the
slope of the friction versus slip curve at zero friction and not some empirical
lesser value used by others.
It was expected that the values of deformation stiffness (i.e. ex = 0.5 kx(21)2w,
and Cy = 0.5 ky(21)2w) will be influenced by the methods of deformation
measurement (without/near sliding, complete sliding). It was determined
experimentally that the equality between the deformation stiffness and the slip
stiffness was also influenced by the tyre motions (steady-state, quasi steady-state,
non steady-state).
2. The available friction coefficient (µ), which can be resolved into the longitudinal
and lateral components (p5, µJ, has been expressed as a function of slip. At slips
beyond the critical value, the frictional force coefficients (µx, µy) are equal to the
available friction coefficients. At a = rt/2 the friction coefficient is assumed to
equal the locked-wheel coefficient (µd).
3. The candidate deduced mathematically from hi~ predic:tion model, with inconclusive
evidence from field measurement, that the direct contribution of tyre stiffness to
the frictional forces is in the region of below the critical percent slip (Sxcr) and
critical slip angle (acJ Beyond this area, the tyre stiffness still contributes
indirectly, and diminishes when the condition of 100% slip (locked-wheel) or 90°
slip angle was achieved. See al&o Figure 2.19 and Tables 2.7-2.8.
Using the data from the literature survey, the candidate proposed an empirical
model which will be able to predict the change of tyre stiffness for any set of
operating conditions, once the tyre stiffness for one set of operating conditions
has been measured (known). The equation~ (Eq~.2.~2-2.37) are ~et out in Section 2.4.7. The
results using the candidate's measured field data confirm a good agreement between
the predicted and the measured values of tyre stiffness.
266
4. The candidate has proposed a modification to an existing model for the shape of
tyre pressure distribution in a form of trapezium, with contact length of 21, and
slope length of a and b. This model offers flexibility in representing many types
of actual pressure distribution, by selecting the appropriate value of its slope
(a/21, b/21). It was found using Sakai's laboratory data that the selection of a/21 is
more sensitive with increasing tyre stiffness.
5. An analytical model has been developed by the candidate, which enables the
calculation of frictional forces over all ranges of percent slip and slip angle. The
ability to calculate frictional forces from tyre stiffness and either locked-wheel
BFC or maximum SFC, is a feature of the candidate's model. In general, the
candidate's model is found to be in the median position when the model
response was compared with 5 tyre models: HSRI-I, HSRI-11, HSRI-III, Good
Year, and Sakai. The way of u~ing the candidate's model and its example are presented in Sections 2.6.6-2.6.8.
Since the locked-wheel BFC (or maximum SFC) can be predicted from surface
texture measurement [Yandell et.al (1983)], this model can be used to predict the
frictional forces, from tyre stiffness and surface texture data. With the availability
of equipment for measuring the locked-wheel BFC or maximum SFC, the data
from this equipment can be processed directly to obtain the frictional forces for
all ranges of percent slip (Sx) and slip angles ( ex). See al~o Figures 7.1 - Z Z.
6. The effect of some input parameters (Cx, CY' a/21, b/21, µd, and Fz) has been
studied using the candidate's model. Almost all parameters have significant
effects on the BFC and SFC, except for b/21 which is found to be small.
7. A procedure has been set up for correcting the test wheel speed of Multi Mode
Friction Test Truck (MMFTT) by using: the calculated vehicle speed (over the
measured section), the recorded test wheel speed (at approaching the first
catadioptric target), and the slip angle used.
Another procedure has also been set up to estimate the correction needed for the
wheel speed delay, by using the data of the decreasing wheel speed with time. A
slightly greater correction than when using this procedure was required to
improve the estimation.
267
8. It was found that during the fully braking period the magnitude of locked-wheel
BFC tends to decrease with time.
9. The effect of tyre pressure and the rate of braking have been studied by means
of analysis of variance. The BFC decreased with increasing tyre pressure,
whereas the effect of tyre pressure on SFC was found to be insignificant. The
BFC increased with increasing the rate of braking, whereas the SFC decreased
with increasing the rate of braking.
10. In general, the results from the candidate's experiment and Sakai's laboratory
data, show a good agreement between the measured and predicted values.
While two lesser-controlled factors were involved in the candidate's
experiment, a closer agreement can be achieved by using a smaller slip angle
(after certain percent slip) on the theoretical prediction, and by using a greater
correction (for wheel speed delay) on the candidate's field measured data.
For both candidate's experimental results and Sakai's laboratory data, the
predicted values of SFC (versus slip angle) at zero percent slip, was found to
under estimate the measured values. A closer agreement can be achieved by
using the modification in the input of the candidate's tyre model, so the input
values of Cy is force/1.57 radians (or force/90°) instead of force/radian (or
force/57.3°) as in the original input of the candidate's tyre model.
11. It was found that the magnitude of field tyre stiffness (i.e. slips measurement) is
practically independent of road conditions, and gives an accurate prediction of
friction measurement.
268
7.5. Future Work
1. Gathering more information on the factors influencing the tyre stiffness. The
results then can be used for further refinements of the model or empirical factor
for tyre stiffness variation. It is equally important, to develop the model for the
prediction of tyre stiffness (from other tyre properties or related parameters).
2. Gathering more information on the factors influencing the shape of tyre pressure
distribution. The results then can be used for selecting the appropriate shape of
pressure distribution (a/21, b/21) for each condition.
3. Carry out extensive field tests on different types of pavement at different speeds,
to further examine the model of frictional forces calculation. The results then can
be used for establishing the model, or (where appropriate) to derive a set of
empirical factors in theoretical prediction.
4. Extend the research on predicting the tyre-road friction under braking and
cornering, from tyre stiffness and locked-wheel BFC (or maximum SFC), in
which the latter parameter is obtained from the surface texture data.
269
References
1. Agg, T.R. (1924), "Tractive Resistance and Related Characteristics of Roadway Surfaces", Eng. Exp., Iowa State College, Vol. 22, No. 36.
2. Allbert, B.J. and Walker, J.C. (1965-66), "Tyre to Wet Road Friction at High Speeds", Proc. Inst. Mech. Engrs. (London) Automobile Div., Vol. 180 (Pt 2A), No. 4, pp 105-121.
3. Archard, J.F. (1953), "Contact and Rubbing of Flat Surfaces", Journal of Applied Physics, Vol. 24, No. 8, pp 981-988.
4. Archard, J.F. (1957-58), "Elastic Deformation and Loss of Friction", Proc. Royal Soc., A-243, pp 190-205.
5. Astrov, V .A. (1962), "Influence of the Roughness of a Road Pavement on the Tyre-road Adhesion", The Roads, No. 11.
6. Balcker, E., Nyborg, L. and Pacejka, H.B. (1987), "Tyre Modelling for Use in Vehicle Dynamics Studies", SAE Paper, No. 870421, pp 1-15.
7. Bashore, H.H. (1937), "A New Revised Rebound Resiliometer", Rubber Chemistry and Technology, Vol. 10, pp 820-826.
8. Bassi, A.C. (1965), "Measurements of Friction of Elastomers by the Skid Resistance Tester", Rubber Chemistry and Technology, Vol. 38, pp 112-122.
9. Bergman, W. (1961), "Theoretical Prediction of the Effect of Traction on Cornering Force", SAE Trans., Vol. 69, Paper 186-A, pp 614-640.
10. Bergman, W. (1977a), "Skid Resistance Properties of Tires and Their Influence on Vehicle Control", TRR, No. 621, pp 8-18.
11. Bergman, W. (1977b), "Critical Review of the State-of-the-Art in the Tire Force and Moment Measurements", SAE Paper, No. 770331, pp 1-15.
12. Bergman, W. and Beauregard, C. (1974), "Transient Tire Properties", SAE Paper, No. 740068, pp 1-15.
13. Bergman, W. and Clemett, H.R. (1975), "Tire Cornering Properties", Tire Science and Technology, Vol. 3 (3), pp 135-163.
14. Bernard, J.E., Segel, L. and Wild, R.E. (1977), "Tire Shear Force Generation During Combined Steering and Braking Maneuvers", SAE Paper, No. 770852, pp 1-17.
15. Besse, J.P. (1972), "Water Film Thickness Effects on the Friction Between Tires and Pavements", Auto. Saf. Res. Program Rep. S51, Pennsylvania State Univ. 76 pp.
16. Bethune, J.D. and Read, K. (1974), "Road Surface and Skid Resistance Testing Vehicle", Proc. ARRB, Vol. 7, Pt. 7, pp 380-393.
270
17. Bode, G. (1962), "Krafte und Bewegungen unter rollenden Lastwagenreifen", A.T.Z. 64 (10), pp 300-306.
18. Box, G.E.P., Hunter, W.G., and Hunter, J.S. (1978), "Statistics for Experimenters", John Wiley, New York.
19. Bradley, J. and Allen, R.F. (1930), "Factors Affecting the Behaviour of Rubber-Tyred Wheels on Road Surfaces", Proc. Inst. of Automobile Engineers, Vol. 25, pp 63-92.
20. Browne, A., Ludema, K.C. and Clark, S.K. (1982), "Mechanics of Pneumatic Tires - Chapter 5: Contact Between the Tire and Roadway", NBS Monograph 122, pp 303-349.
21. Carr, C. I (1967), "Contribution of the Rubber Compound to the Wet Skid Resistance of Tires", HRR, No. 214, pp 1-6.
22. Carroll, P.J. (1965), "The Coulomb Contribution to Rubber Hysteresis", M.Sc Thesis, The Penn. State Univ.
23. Clamroth, E. and Heidemann, W. (1968), "Measurement of Rolling and Sliding Friction Coefficients of Tread Compounds", Rubber Chemistry and Technology, Vol. 41, pp 908-942.
24. Croney, D. (1977), "The Design and Performance of Road Pavement", London.
25. Csathy, T.I., Burnett, W.C. and Armstrong, M.D. (1968), "State of the Art of Skid Resistance Research", HRB, Spee. Report 95, pp 34-48.
26. Daughaday, H. and Balmer, G.G. (1970), "A Theoretical Analysis of Hydroplaning Phenomena", HRR, No. 311, pp 1-10.
27. Davisson, J.A. (1969), "Design and Application of Commercial Type Tyres", SAE Paper, No. 690001, pp 1-32.
28. Denny, D.F. (1953), "The Influence of Load and Surface Roughness on the Friction of Rubber-like materials", Proc. Phys. Soc., 66B, pp 721-727.
29. Dermoredjo, S.P.S. (1992), "Effects of the Rate of Braking Application on the Frictional Force Coefficient by Using MMFTT", M.Eng.Sc Project Report, The University of New South Wales.
31. Doty, R.N. (1975), "Study of the Sand Patch and Outflow Meter Methods of Pavement Surface Texture Measurement", ASTM, STP 583, pp 42-61.
32. Draper, N.R. (1981), "Applied Regression Analysis", John Wiley, New York.
271
33. Dugoff, H., Fancher, P.S. and Segel, L. (1969), "Tire Performance Characteristics Affecting Vehicle Response to Steering and Braking Control Inputs", Final Report, National Bureau of Standards Contract No. CST-460, HSRI, Univ. of Michigan, pp 1-105.
34. Dugoff, H., Fancher, P.S. and Segel, L. (1970), "An Analysis of Tire Traction Properties and Their Influence on Vehicle Dynamic Performance", SAE Paper, No. 700377, pp 1219-1243.
35. Elsenaar, P.M.W., Reichert, J. and Sauterey, R. (1977), "Pavement Characteristics and Skid Resistance, TRR, No. 622, pp 1-25.
36. Ervin, R.D. (1976), "Measurements of the Longitudinal and Lateral Traction Properties of Truck Tires", Braking of Road Vehicles, Inst. Mech. Engrs. Conf. Publications, pp 93-100.
37. Fancher, P.S. (1972), "Tire Traction Grading Procedures as Derived from the Maneuvering Characteristics of a Tire-Vehicle System", Final Report, Volume II, NBS Contract No. 1-35715, HSRI, Univ. of Michigan.
38. Fiala, E. (1954), "Seitenkrafte am Rollenden Luftreifen", VDI-Zeitschrift 96, pp 973-979.
39. Finney, E.A. and Brown, M.G. (1959), "Relative Skid Resistance of Pavement Surfaces Based on Michigan Experience", First Int. Skid Prevention Conf., Pt. II, pp 435-459.
40. Foster, R. (1961), "Friction of Rubber at Low Sliding Velocities", M.Sc Thesis, Dept. of Mech. Engng., The Penn.State Univ.
41. Fromm, H. (1954), "Sideslip and Guiding Characteristics of the Rolling Wheel", NACA, TM 1365, pp 191-216.
42. Fuller, D.L., Hall, G.L. and Conant, F.S. (1984), "Effect of Testing Conditions on Rolling Resistance of Automobile Tires", SAE Paper, No. 840068, pp 1-11.
43. Gee, S., King, W.L. and Hegmon, R.R. (1975), "Pavement Texture Measurement by Laser: A Feasibility Study", ASTM, STP 583, pp 29-41.
44. Giles, G.C. (1959), "Some European Methods for the Measurement of Skidding Resistance", Proc. 1st Int. Skid Prevention Conf., Pt.I, pp 267-296.
45. Giles, C.G. (1964), "Some Recent Developments in Work on Skidding Problems at the Road Research Laboratory", HRR, No. 46, pp 43-59.
46. Giles, G.C. and Sabey, B.E. (1959), "A Note on the Problem of Seasonal Variation in Skidding Resistance", First Int. Skid Prevention Conf., Pt. II, pp 563-568.
47. Giles, G.C., Sabey, B.E. and Cardew, K.H.F. (1962), "Development and Performance of the Portable Skid-Resistance Tester", ASTM, STP No. 326, pp 50-74.
272
48. Gim, G. and Nikravesh, P.E. (1990), "An Analytical Model of Pneumatic Tyres for Vehicle Dynamic Simulations - Part 1: Pure Slips", Int. J. of Vehicle Design, Vol. 11, No. 6, pp 589-618.
49. Gim, G. and Nikravesh, P.E. (1991), "An Analytical Model of Pneumatic Tyres for Vehicle Dynamic Simulations - Part 2: Comprehensive Slips", Int. J. of Vehicle Design, Vol. 12, No. 1, pp 19-39.
50. Golden, J.M. (1981), "A Theory of Wet Road-Tyre Friction", Wear, Vol. 71, pp 307-331.
51. Goodwin, W.A. and Whitehurst, E.A. (1962), "The Minimization of Variables in Equipment and Techniques for Performing Skid Trailer Tests", ASTM, STP No. 236, pp 29-41.
52. Gopalan, M.K. (1976), "Influence of Surface Texture on Friction and Abrasion of Rubber", Ph.D Thesis, The University of New South Wales.
53. Gough, V.E. (1954-55), Discussion on Paper, "The Skid Resisting Properties of Roads and Tyres", Proc. Inst. Mech. Engrs. Auto. Div., p 46.
55. Gough, V.E. (1959), "Friction of Rubber on Lubricated Surfaces", Discussion of a paper by D.Tabor, Rev. Gen. du Caoutchouc, Vol. 36, No. 10, p 1409.
56. Gough, V.E. (1974), "A Tyre Engineer Looks Critically at Current Traction Physics", The Physics of Tire Traction - Theory and Experiment, Eds. D.F. Hays and A.L. Browne, pp 281-297.
57. Gray, J.E. and Renninger, F.A. (1965), "The Skid Resistance Properties of Carbonate Aggregates", HRR, Vol. 120, pp 18-34.
58. Greenwood, J.A., Minshall, H. and Tabor, D. (1961), "Hysteresis Losses in Rolling and Sliding Friction", Proc. Roy. Soc., A259, pp 480-507.
59. Greenwood, J.A. and Tabor, D. (1958), "The Friction of Hard Sliders on Lubricated Rubber: the Importance of Deformation Losses", Proc. Phys. Soc., 71, pp 989-1001.
60. Grime, G. and Giles, C.G. (1954-55), "The Skid-Resisting Properties of Road and Tyres", Proc. Inst. Mech. Engrs. Auto. Div., Vol. 7, pp 19-30.
61. Grosh, K.A. and Maycock, G. (1968), "Influence of Test Conditions on Wet Skid Resistance of Tire Tread Compounds", Rubber Chemistry and Technology, Vol. 41, pp 479-493.
62. Hadekel, R. (1952), "The Mechanical Characteristics of Pneumatic Tyres - A Digest of Present Knowledge", S & T Memo, TPA 3/fIB, British Ministry of Supply.
273
J
63. Harris, A.J. (1968), "Road Surface Texture and the Slipperiness of Wet Roads", HRR, No. 214, pp 18-23.
64. Havens, J.H. (1959), "Skid Prevention Studies in Kentucky", First Int. Skid Prevention Conf., Pt. II, pp 333-340.
65. Hegmon, R.R. (1965), "Development of Friction Analyser and Study on the Friction of Lubricant Rubber", Auto. Safety Res. Prog., Rep. S20, Dept. of Mech.1Engng., The Penn. State Univ.
J 66. Hegmon, R.R. (1968), Ph.D. Thesis, Penn. State Univ.
67. Hegmon, R.R. (1975), "Contact Temperature of a Skidding Tyre", Wear, Vol. 33, pp 17-23.
68. Hegmon, R.R. (1982), "Reliability of Locked-Wheel Skid Resistance Tester Confirmed", Public Roads, Vol. 46, No. 3, pp 92-101.
69. Hegmon, R.R. and Henry, J.J. (1973), "Thermal Analysis of a Skiding Tire", Wear, Vol. 24, pp 361-380.
70. Henry, J.J. and Hegmon, R.R. (1975), "Pavement Texture Measurement and Evaluation", ASTM, STP 583, pp 3-17.
71. Hofelt, C. (1959), "Factors in Tires that Influence Skid Resistance - Part V: Effect of Speed, Load Distribution and Inflation", Proc. 1st Int. Skid Prevention Conf., pp 173-187.
72. Holla, L. (1974), "Influence of Surface Texture on the Coefficient of Friction of Road Surface", Ph.D Thesis, The University of New South Wales.
73. Holmes, K.E. and Stone, R.D. (1969), "Tyre Forces as Functions of Cornering and Braking Slip on Wet Surfaces", RRL Report, LR 254, pp 1-26.
74. Holmes, T., Lees, G. and Williams, A.R. (1972), "A Combined Approach to the Optimation of Tyre and Pavement Interaction", Wear, Vol. 20, pp 241-276.
75. Home, W.B. and Dreher, R.C. (1963), "Phenomena of Pneumatic Tire Hydroplaning", NASA, TN D-2056, pp 1-52.
76. Jameson, G.W., Baran, E. and Sheldon, G.N. (1988), "Australian Experience With the Swedish Laser Road Surface Tester", Proc. 14th ARRB Conf., Part 8, pp 244-258.
77. Kamm, W. (1938), "Gesamtfahrwiderstandsgleichung fur die Fahrwiderstande von Personenkraftfahrzeugen", DKF, ZB 24.
78. Kant, S., Rao, D.L.P. and Munjal, M.L. (1975), "Prediction of the Coefficient of Friction for Pneumatic Tyres on Hard Pavement", Proc. Inst. Mech. Engrs., Vol 189, pp 259-266.
80. Kem, W.F. (1967), "Coefficient of Wet Friction of Tire Treads", Rubber Chemistry and Technology, Vol. 40, pp 984-1013.
81. Kienle, R.N. (197 4 ), "The Role of the Tread Pattern - A Blend of the Simple and Complex", The Physics of Tire Traction - Theory and Experiment, Eds. D.F. Hays and A.L. Browne, pp 241-255.
82. Kraft, P. (1941), "Die Kraftverteilung in der Beruhrungsflache zwischen Reifen und Fahrbahn", Bericht 140 der Lilienthal-Gesellschaft, pp 11-14. English version (1954): "Force Distribution in the Contact Surface Between Tire and Runway", NACA, TM 1365, pp 31-40.
83. Kummer, H.W. (1966), "Unified Theory of Rubber and Tire Friction", Eng. Research Bull. B-94, Penn. State University.
84. Kummer, H.W. (1968), "Lubricated Friction of Rubber - Discussion", Rubber Chemistry and Technology, Vol. 41, pp 895-907.
85. Kummer, H.W. and Meyer, W.E. (1962), "Measurement of Skid Resistance", ASTM, STP 326, pp 3-28.
86. Kummer, H.W. and Meyer, W.E. (1966), "New Theory Permits Better Frictional Coupling Between Tyre and Road", FISITA Conf. Munich.
87. Kummer, H.W. and Meyer, W.E. (1967), "Tentative Skid Resistance Requirements for Main Rural Highways", NCHRP Report 37, pp 1-80.
88. Lander, F.T.W. and Williams, T. (1968), "The Skidding Resistance of Wet Runway Surfaces with Reference to Surface Texture and Tyre Condition", RRL Report, LR 184.
89. Lees, G. (1984-85), Lecture Notes on Material Engineering, Univ. of Birmingham, UK. (unpublished).
90. Lees, G. and Katekhda, I.E.D. (1974), "Prediction of Medium and High Speed Skid Resistance Values by Means of a Newly Developed Outflow Meter", Asphalt Paving Technology, Vol. 43, pp 436-464.
91. Lees, G., Katekhda, I.E.D., Bond, R. and Williams, A.R. (1977), "The Design and Performance of High Friction Dense Asphalts", Skidding Accidents, TRR, No. 624, pp 40-51.
92. Leu, M.C. and Henry, J.J. (1978), "Prediction of Skid Resistance as a Function of Speed from Pavement Texture Measurements", TRR, No. 666, pp 7-18.
93. Lippmann, S.A. and Oblizajek, K.L. (1974), "The Distribution of Stress Between the Tread and the Road for Freely Rolling Tires", SAE Paper, No. 740072, pp 1-31.
275
94. Livingston, D.I. and Brown, J.E. (1969), "Physics of the Slipping Wheel - I. Force and Torque Calculations for Various Pressure Distributions", Rubber Chemistry and Technology, Vol. 42, No. 4, pp 1014-1027.
95. Livingston, D.I. and Brown, J.E. (1970), "Physics of the Slipping Wheel - II. Slip Under Both Tractive and Lateral Forces", Rubber Chemistry and Technology, Vol. 43, No. 2, pp 244-261.
96. Loeb, J.S., Guenther, D.A., Chen, H.F. and Ellis, J.R. (1990), "Lateral Stiffness, Cornering Stiffness and Relaxation Length of the Pneumatic Tire", SAE Paper, No. 900129, pp 1-9.
97. Ludema, K.C. (1975), "Physical Factors in Tyre Traction", Physics in Tech., Jan., pp 11-17.
98. Ludema, K.C. and Tabor, D. (1966), "The Friction and Viscoelastic Properties of Polymeric Solids", Wear, Vol. 9, pp 329-348.
99. Lupton, G.N. and Williams, T. (1973), "Study of the Skid Resistance of Difference Tire Polymers on Wet Pavements With a Range of Surface Textures", ASTM, STP 530, pp 101-116.
100. Maalej, A.Y., Guenther, D.A. and Ellis, J.R. (1989), "Experimental Development of Tyre Force and Moment Models", Int. J. of Vehicle Design, Vol. 10, No. 1, pp 34-51.
101. Maclean, D.J. and Shergold, F.A. (1960), "The Polishing of Road Stone in Relation to the Resistance to Skidding of Bituminous Road Surfacings", Dept. of Sci. and Indust. Res., R.R. Tech. Paper No. 43.
102. Marian, J.E. (1962), "Surface Texture in Relation to Adhesive Bonding", ASTM, STP 340, pp 122-149.
103. Marick, L. (1959), "Factors in Tires that Influence Skid Resistance - Part I: The Effect of Tread Design", Proc. 1st Int. Skid Prevention Conf., pp 155-162.
104. Markwick, A.H.D. and Starks, H.J.H. (1941), "Stresses Between Tire and Road", J. Inst. Civ. Engrs. (London), pp 309-325.
105. Martin, H. (1936), "Druckverteilung in der Beruhrungsflache zwischen Rad und Fahrbahn", Krafttfahrtechnische Forschungsarbeiten, No. 2.
106. Maycock, G. (1965-66), "Studies on the Skidding Resistance of Passenger-Car Tyres on Wet Surfaces", Proc. Inst. Mech. Engrs. (London), Automobile Div., Vol. 180 (Pt 2A), No. 4, pp 122-151.
107. Maycock, G. (1968), "Skid Resistance of Passenger Car Tires on Wet Surfaces", Rubber Chemistry and Technology, Vol. 41, pp 780-806.
108. McHenry, D. (1943), "A Lattice Analogy for the Solution of Stress Problems", J. Inst. Civil Engrs., No. 2, pp 59-82.
276
109. Meyer, W.E. (1964), "Some Results of Research on Skid Control", 10th FISITA Congress, Paper No. B-5, Tokyo.
111. Meyer, W.E. and Kummer, H.W. (1969), "Pavement Friction and Temperature Effects", HRB, Sp. Report No. 101, pp 47-55.
112. Meyer, W.E., Hegmon, R.R. and Gillespie, T.D. (1974), "Locked Wheel Pavement Skid Tester Correlation and Calibration Techniques", NCHRP Report 151, pp 1-100.
113. Moore, D.F. (1965), "Drainage Criteria for Runway Surface Roughness", I. Roy. Aeronautical Soc., Vol. 69, pp 337-342.
114. Moore, D.F. (1966), "Prediction of Skid-resistance Gradient and Drainage Characteristics for Pavements", HRR, No. 131, pp 181-203.
115. Moore, D.F. (1967), "A Theory of Viscous Hydroplaning", Int. I.Mech. Sci., Vol. 9, pp 797-810.
116. Moore, D.F. (1968), "An Elastohydrodynamic Theory of Tire Skidding:, Int. FISITA Congress, Paper No. 2-02, Barcelona.
117. Moore, D.F. (1975), "Principles and Applications of Tribology", Pergamon Press, pp 1-388.
118. Moyer, R.A. (1934), "Skidding Characteristics of Automobile Tyres on Roadway Surfaces and Their Relation to Highway Safety", Bulletin 120, Iowa Eng. Exp. St., Iowa State College.
119. Moyer, R.A. (1959), "A Review of the Variables Affecting Pavement Slipperiness", First Int. Skid Prevention Conf., Pt. II, pp 411-433.
121. Nordeen, D.L. (1968), "Application of Tire Characterizing Functions to Tire Development", SAE Paper, No. 680409, pp 1421-1435.
122. Nordeen, D.L. and Cortese, A.D. (1963), "Force and Moment Characteristics of Rolling Tires", SAE Paper No. 713A, pp 325-347.
123. Novopol'skii, VJ. and Nepomnyashchii, E.F. (1967), "The Interaction of a motor vehicle tyre tread with the road surface", Abrasion of Rubber, Ed. DJ. James.
124. Novopol'skii, VJ. and Tret'yakov, O.B. (1963), "Slip of the Elements of the Tread Pattern in the Contact Area of Tires", Soviet Rubber Tech., Vol. 22, p 25.
277
125. Orchard, D.F. (1947), "Investigation of Skidding on Roads", J. Inst. Civil Engrs., Vol. 27, p 179.
126. Orchard, D.F., Yandell, W.O. and Lye, B.R.X. (1970), "A Quick Method of Measuring the Surface Texture of Aggregate", Proc. 5th ARRB Conf., Vol. 5, Part 5, pp 325-341.
127. Pacejka, H.B. (1966), "The Wheel Shimmy Phenomenon", Diss. Tech. University of Delft.
128. Pacejka, H.B. and Fancher, P.S. (1972), "Hybrid Simulation of Shear Force Development of a Tire Experiencing Longitudinal and Lateral Slip", Proc. XIV lnt. Automobile Technical Congress, FISITA, London, pp 3n8-3/85.
129. Papenhuyzen, P.J. (1938), "Friction Experiments in Connection With Slipping of Automobile Tyres", Der Ingenieur, Utrecht 53 (45): V, 75.
130. Payne, A.R. (1974), "Hysteresis in Rubber Vulcanizates", J. Polym. Sci. Polym. Symposia, No. 48, pp 169-196.
131. Percarpio, E.P. and Bevilacqua, E.M. (1968), "Lubricated Friction of Rubber -Part V : Influence of Rubber Resilience and Hardness on Friction", Rubber Chemistry and Technology, Vol. 41, pp 870-880.
132. Peterson, R.F., Eckert, C.F. and Carr, C.I. (1974), "Tread Compound Effects in Tire Traction", The Physics of Tire Traction - Theory and Experiment, Eds. D.F. Hays and A.L. Browne, pp 223-239.
133. Phillips, B.D.A. (1973), "The Static, Steady State and Dynamic Characteristics of Pneumatic Tires", Ph.D. Diss., Lanchester Polytechnic.
134. Pike, E.C. (1949), "Coefficient of Friction", Journal Royal Aero. Soc., Vol. 53, pp 1085-1094.
136. Reynolds, 0. (1886), "On the Theory of Lubrication and its Application to Mr. Beauchamp Tower's Experiments, including Experimental Determination of the Viscosity of Olive Oil", Proc. Roy. Soc., Vol. XL, pp 191-203.
137. Richards, P.J. (1967-68), "Review of Methods of Measurement and Assesment of Surface form and Texture", Proc. Inst. Mech. Eng., Vol. 182, Pt. 3K, Paper 41, pp 453-465.
138. Roberts, A.O. (1971), "The Shear of Thin Liquid Films", J. Physics D: Appl. Phys. Vol. 4, pp 433-440.
139. Roberts, A.O. and Tabor, D. (1968), "Surface Forces: Direct Measurement of Repulsive Forces due to Electrical Double Layers on Solids", Nature, Vol. 219, pp 1122-1124.
278
140. Rogers, L.C. (1972), "Theoretical Tire Equations for Shimmy and Other Dynamic Studies", J. of Aircraft, Vol. 9, No. 8, pp 585-589.
141. Rohde, S.M. (1976), "On the Effect of Pavement Microtexture on Thin Film Traction", Int. J. Mech. Sci., Vol. 18, pp 95-101.
142. Rose, J.G., Hutchinson, J.W. and Gallaway, B.M. (1973), "Summary and Analysis of the Attributes of Methods of Surface Texture Measurement", ASTM, STP 530, pp 60-77.
143. Saal, R.N.J. (1936), "Laboratory Investigation into the Slipperiness of Roads", J. Soc. Chemistry and Industry, Vol. 55, pp 3-7.
144. Sabey, B.E. (1958), "Pressure Distribution Beneath Spherical and Conical Shapes Pressed into a Rubber Plane and Their Bearing on Coefficients of Friction under Wet Conditions", Proc. Phys. Soc., Vol. 71, pp 979-988.
145. Sabey, B.E. (1966), "Road Surface Texture and the Change in Skidding Resistance with Speed", RRL Report, LR 20, pp 1-19.
146. Sabey, B.E. (1968), "Wet Road Skidding Resistance at High Speeds on a Variety of Surfaces on Al", Proc. ARRB, Vol. 4, Pt. 2, pp 1512-1529.
147. Sabey, B.E. and Lupton, G.N. (1964), "Friction on Wet Surfaces of Tire-Tread-Type Vulcanizates", Rubber Chemistry and Technology, Vol. 37, pp 878-893.
148. Sabey, B.E. and Lupton, G.N. (1967), "Measurement of Road Surface Texture", RRL Report, LR 57.
149. Sabey, B.E., Williams, T.E. and Lupton, G.N. (1970), "Factors Affecting the Friction of Tires on Wet Roads" Int. Auto. Saf. Conf. Comp., SAE Paper, No. 700376, pp 324-340.
150. Sakai, H. (1969), "Theoretical Study of the Effect of Tractive and Braking Forces on Cornering Characteristics of Tire", Journal of the Society of Automotive Engineers of Japan, Vol. 23, No. 10.
151. Sakai, H. (1981), "Theoretical and Experimental Studies on the Dynamic Properties of Tyres - Part 1: Review of Theories of Rubber Friction", lnt. J. of Vehicle Design, Vol. 2, No. 1, pp 78-110.
152. Sakai, H. (1982), "Theoretical and Experimental Studies on the Dynamic Properties of Tyres - Part 4: Investigations of the Influences of Running Conditions by Calculation and Experiment", Int. J. of Vehicle Design, Vol. 3, No. 3, pp 333-375.
153. Sakai, H., Kanaya, 0. and Okayama, T. (1978), "The Effect of Hydroplaning on the Dynamic Characteristics of Car, Truck and Bus Tires", SAE Paper, No. 780195, pp 1-22.
279
154. Salt, G.F. (1977), "Research on Skid Resistance at the TRRL (1927-1977)", RRL Report, SR 340.
156. Savkoor, A.R. (1965), "On the Friction of Rubber", Wear, Vol. 8, pp 222-237.
157. Savkoor, A.R. (1966), "Some Aspects of Friction and Wear of Tyres Arising from Deformations, Slip and Stresses at the Ground Contact", Wear, 9, pp 66-78.
158. Schallamach, A. (1954), "On the Abrasion of Rubber", Proc. Phys. Soc., B67, pp 883-891.
159. Schallamach, A. (1958), "Friction and Abrasion of Rubber", Wear, Vol. 1, pp 384-417.
160. Schallamach, A. and Grosch, K. (1982), "Tire Traction and Wear", Mechanics of Pneumatic Tires, Ed. S.K. Clark, NBS Monograph 122, pp 415-454.
161. Schallamach, A. and Turner, D. (1960), "The Wear of Slipping Wheels", Wear, Vol. 3, pp 1-25.
163. Schonfeld, R. (1970), "Photo Interpretation of Skid Resistance", HRR, No. 311, pp 11-25.
164. Schonfeld, R. (1974), "Pavement Surface Texture Classification and Skid Resistance Photo-Interpretation", The Physics of Tire Traction - Theory and Experiment", Eds. D.F. Hays and A.L. Browne, pp 325-338.
165. Schulze, K.H. and Beckmann, L. (1962), "Friction Properties of Pavements at Different Speeds", ASTM, STP 326, pp 42-49.
166. Schuring, D.J., Tapia, G.A. and Gusakov, I. (1976), "Influence of Tire Design Parameters on Tire Force and Moment Characteristics", SAE Paper, No. 760732, pp 1-17.
167. Schuster, R. and Weichsler, P. (1935), "Der Kraftschluss Zwischen Rad und Fahrbahn ", Automobiletechnische Zeitschrift.
168. Segel, L. ( 1966), "Force and Moment Response of Pneumatic Tires to Lateral Motion Inputs", Trans. ASME, J. Engr. for Ind., Vol. 88B(l), pp 37-44.
169. Shah, V.R. and Henry, J.J. (1978), "Determination of Skid Resistance - Speed Behaviour and Side Force Coefficient of Pavements", TRR, No. 666, pp 1-44.
170. Sharp, R.S. (1989), "On the Accurate Representation of Tyre Shear Forces by a Multi-Radial-Spoke Model", Proc. 11th IAVSD Symposium, pp 528-541.
280
171. Sharp, R.S. and El-Nashar, M.A. (1986), "A Generally Applicable Digital Computer Based Mathematical Model for the Generation of Shear Forces by Pneumatic Tyres", Vehicle System Dynamics, Vol. 15, pp 187-209.
172. Smith, L.L. and Fuller, S.L. (1960), "Florida Skid Correlation Study of Skid Testing with Trailers", ASTM, STP 456, pp 4-101.
173. Staughton, G.C. and Williams, T. (1970), "Tyre Performance in Wet Surface Conditions", RRL Report, LR 355, pp 1-44.
174. Stephens, J.E. and Goetz, W.H.(1960), "Designing Fine Bituminous Mixtures for High Skid Resistance", Proc. HRB, Vol. 39, pp 173-190.
175. Stephens, J.E. and Goetz, W.H. (1961), "Effects of Aggregate Factors on Pavement Friction", HRB Bull. No. 302, pp 1-17.
176. Stiffler, A.K. (1969), "Relation Between Wear and Physical Properties of Roadstones", HRB, Sp. Report 101, pp 56-68.
177. Stonex, K.A. (1959), "Elements of Skidding", Proc. 1st Int. Skid Prevention Conf., pp 1-4.
178. Stutzenberger, W.J. and Havens, J.H. (1958), "A Study of the Polishing Characteristics of Limestone and Sandstone Aggregates in Regard to Pavement Slipperiness", HRB Bull. No. 186, pp 58-81.
179. Sutantra, I.N. and Harrison, H.L. (1985), "The Effects of Operational Parameters on Vehicle Directional Responses", SAE Paper, No. 852187, pp 93-105.
180. Tabor, D. (1952), "The Mechanism of Rolling Friction", Philosophical Magazine, Serial 7, Vol. 43, No. 345, pp 1055-1059.
181. Tabor, D. (1959), "The Importance of Hysteresis Losses in the Friction of Lubricated Rubber", Proc. 1st Int. Skid Prevention Conf., pp 211-218.
182. _Taneerananon, P. (1981), "The Analysis of the Mechanism of Tyre Friction on Wet Roads", Ph.D Thesis, The University of New South Wales.
183. Taneerananon, P. (1982), Confidential Report. (unpublished).
184. Taneerananon, P. and Yandell, W.O. (1981), "Microtexture Roughness Effect on Predicted Road-Tyre Friction in Wet Conditions", Wear, Vol. 69, pp 321-337.
185. Teller, L.W. and Buchanan, J.A. (1937), "Determination of Variation of Unit Pressure over the Contact Area of Tires", Public Roads, Dec., pp 195-198.
186. (van Eldik) Thieme, H.C.A., Dijks, A.J. and Bobo, S. (1982), "Measurement of Tire Properties", Mechanics of Pneumatic Tires, pp 542-613.
281
187. Thirion, P. (1946), "Coefficients of Adhesion of Rubber", Rev. Gen. Caoutch, Vol. 23, p 101.
188. Tielking, J.T. and Mital, N.K. (1974), "A Comparative Evaluation of Five Tire Traction Models", Report No. UM-HSRI-PF-74-2, Univ. of Michigan, pp 1-162.
189. Tielking, J.T. and Schapery, R.A. (1980), "Calculation of Tire-Pavement Shear Force", ASME Symp. Proc., The General Problem of Rolling Contact, ASME AMO, Vol. 40, pp 19-39.
190. Tielking, J.T., Fancher, P.S. and Wild, R.E. (1973), "Mechanical Properties of Truck Tires", SAE Paper, No. 730183, pp 1-11.
191. Trollope, D.H. and Wallace, K.B. (1962), "The Nature of Skid Resistance", Proc. ARRB, Vol. I, pp 779-791.
192. Uffelmann, F. (1983), "Automobile Stability and Handling Dynamics in Cornering and Braking Maneuvers", Vehicle System Dynamics, Vol. 12, pp 203-223.
193. 'Uniken' (1989), "Friction Meter Device: A Boon for Road Safety", UNSW News, No. 274, pp 1-2.
194. U.S. Army Engineer Waterways Experiment Station, (1964), "Stresses Under Moving Vehicles - Report 4", Technical Report No. 3-545.
195. Veith, A.G. (1971), "Measurement of Wet Cornering Traction of Tires", Rubber Chemistry and Technology, Vol. 44, pp 962-995.
196. Veith, A.G. (1977), "Tire Wet Traction Performance: The Influence of Tread Pattern", Skidding Accidents, TRR, No. 621, pp 113-125.
197. Wallace, K.B. and Trollope, D.H. (1969), "Water Pressure Beneath a Skidding Tyre", Wear, Vol. 13, pp 109-118.
198. Yandell, W.O. (1969), "The Effect of Surface Geometry on the Lubricated Sliding Friction and Polishing of Roadstones", Aust. Rd. Res., Vol. 3, No. 10, pp 50-68.
199. Yandell, W.O. (1970), "Measurement of Surface Texture of Stones with Particular Regard to the Effect on the Frictional Properties of Road Surfaces", Ph.D Thesis, The University of New South Wales.
200. Yandell, W.O. (1971a), "The Use of Mechano Lattice Analogy for Determining the Abrading Stresses in Sliding Rubber", Rubber Chemistry and Technology, Vol. 44, pp 758-770.
201. Yandell, W.O. (1971b), "A New Theory of Hysteretic Sliding Friction", Wear, Vol. 17, pp 229-244.
282
202. Yandell, W.O. (1974), "The Relation Between the Stress Saturation of Sliding Rubber and the Load Dependence of Road Tyre Friction", The Physics of Tire Traction - Theory and Experiment, Eds. D.F.Hays and AL.Browne, pp 311-323.
203. Yandell, W.O. and Gopalan, M.K. (1976), "The Relation Between Surface Texture of Roads and Friction and Abrasion of Tyre Tread Rubber", ARRB Proceedings, Vol. 8, pp 8-14.
204. Yandell, W.O. and Holla, L; (1974), "The Prediction of the Coefficient of Friction from Surface Texture Measurement", Proc. ARRB, Vol. 7, Pt. 7, pp 354-363.
205. Yandell, W.O. and Sawyer, S. (1992), Report to Roads and Traffic Authority of NSW. (unpublished).
206. Yandell, W.O., Taneerananon, P. and Zankin, V. (1983), "Prediction of Tire-Road Friction from Surface Texture and Tread Rubber Properties", Frictional Interaction of Tire and Pavement, ASTM, STP 793, pp 304-322.
207. Yeager, R.W. (1974), "Tire Hydroplaning: Testing, Analysis, and Design", The Physics of Tire Traction - Theory and Experiment, Eds. D.F. Hays and A.L. Browne, pp 25-64.
208. Yeaman, J. (1989), "Pavement Maintenance Management for Local Government", Pavement Management Service Pty. Ltd., Dynatest PMS.
209. Yeomans, K.A. (1976), "Statistics for the Social Scientist: 2 Applied Statistics", Penguin Education, Great Britain.
210. Yeow, S.H., El-Sherbiny, M. and Newcomb, T.P. (1978), "Thermal Analysis of a Tyre During Rolling or Sliding", Wear, Vol. 48, pp 157-171.
211. Zakaharov, S.P. and Novopol'skii (1957), "Distribution of Specific Pressure of a Tire on the Road at High Velocities", Trudy Tauchno-Issledovatel'skoga Instituta Shinnoi Pronyschlennosti, Sbomik 3, Methody Rascheta i Ispaytania Automobil'nykh Shin: U.S.S.R. Nauchno, pp 139-153.
212. Zankin, V.G. (1981), "Determination of Damping Behaviour of Viscoelastic Substrates from High Speed Rolling Friction Tests", Ph.D Thesis, The University of New South Wales.
283
A-1
Appendix A
100 REM BASIC COMPUTER PROGRAM FOR CALCULATION OF 110 REM FRICTIONAL FORCES USING LOCKED-WHEEL BFC 120 REM CONSTANT: A a/21 B = b/21 C = Cx (force/slip) 130 REM D Cy (force/rad), or D (mod.) = Cy (force/1.57rad) 135 REM E µd F = F z (force) 140 REM VARIABLE: G Sx H = a (rad) 150 REM**************************************************************
160 READ A, B, C, D, E, F 170 LET P A*(2-A-B) 180 LET N = 2-A+A*B*0.5+B*B*0.5-2*B 190 LET G = (E*F*(N-P)/C+P*P*0.5)/(E*F*(N-P)/C+P*N-P*P) 195 IF A= 0 THEN 213 200 LET IA= P*C*G/(F*(l-G)) 210 LET L = (IA-E*G)/(1-G) 211 GOTO 216 213 LET IA= E 214 LET L = E 216 PRINT "a/21 = ";A, "b/21 = ";B, "F.= ";F 218 PRINT "Cx = ";C, "Cy = ";D, "µd = ";E 22 0 PRINT "Sxcr = "; G, "µscr = "; IA, "µ. 0 = "; L 225 IF A= 0 THEN 304 230 FOR R OTO 1.57 STEP 0.001 240 LETS= 1.5708-2*SIN(R)*COS(R)*(N/P-1)
+2*COS(R)*COS(R)*E*F*(N/P-1)/(D*P)-R 250 IF ABS(S) < 0.1 THEN 270 260 NEXT R 270 FOR H R-0.0005 TO 1.57 STEP 0.00001 280 LETO= l.5708-2*SIN(H)*COS(H)*(N/P-l)
+2*COS(H)*COS(H)*E*F*(N/P-l)/(D*P)-H 290 IF ABS(O) < 0.001 THEN 310 300 NEXT H 302 GOTO 310 304 LETH= 1.5708 305 LET IB = E 306 LET M = E 308 GOTO 330 310 LET IB = P*D*TAN(H)/F 320 LET M = (IB-E*H*.636619772)/(1-H*.636619772) 330 PRINT "<lcr 340 PRINT"
= ";H,
F x'', " "~r = ";IB, "~0 = ";M
Fy", " µx", " ~", " 350 FOR H = .06981317 TO 1.57 STEP .06981317 360 LET K = M-(M-E)*H*.636619772 370 LET YA= l/(1+(2*B-A*B-B*B)*D*TAN(H)/(K*F)) 380 IF YA>= (1-B) THEN 520 390 LET YB = K*F/((2-A-B)*D*TAN(H)) 400 IF YB >= (1-B) THEN 420 410 IF YB >= A THEN 470 420 LET RC 0 425 LET CR RC/F 430 LET SC K*F
435 LET CS= SC/F 440 LET RSC= SQR(CR*CR+CS*CS) 450 PRINT RC, SC, CR, CS, RSC 460 GOTO 560
LET RB 0 LET BR RB/F
A-2
470 475 480 485 490
LET SB D*TAN(H)*YB*YB+K*F*(l-.5*B-YB)/(1-.5*A-.5*B) LET BS SB/F LET RSB = SQR(BR*BR+BS*BS)
500 PRINT RB, SB, BR, BS, RSB 510 GOTO 560 520 LET RA 0 525 LET AR RA/F 527 IF B = 0 THEN 532 530 LET SA D*TAN(H)*YA*YA+K*F*(l-2*YA+YA*YA)/(2*B-A*B-B*B) 531 GOTO 535 532 LET SA= D*TAN(H) 535 LET AS= SA/F 540 LET RSA= SQR(AR*AR+AS*AS) 550 PRINT RA, SA, AR, AS, RSA 560 NEXT H 570 FOR G 580 590 600 610 620
FOR LET LET LET LET
H J K
I XA =
.1 TO .99 STEP .1 0 TO 1.57 STEP 0.06981317 L-(L-E)*G M-(M-E)*H*.636619772 J+(K-J)*ATN(TAN(H)/G)*.636619772
630 IF XA >= (1-B) THEN 735 640 LET XB = I*F*(l-G)/((2-A-B)*(SQR(C*C*G*G+D*D*TAN(H)*TAN(H)))) 650 IF XB >= (1-B) THEN 660 655 IF XB >= A THEN 700 660 LET PC I*F*COS(ATN(TAN(H)/G)) 665 LET CP PC/F 670 LET QC I*F*SIN(ATN(TAN(H)/G)) 675 LET CQ QC/F 680 LET PQC = SQR(CP*CP+CQ*CQ) 685 PRINT PC, QC, CP, CQ, PQC 690 GOTO 770 700 LET PB
720 LET PQB = SQR(BP*BP+BQ*BQ) 725 PRINT PB, QB, BP, BQ, PQB 730 GOTO 770 735 IF B = 0 THEN 743 740 LET PA= C*G*XA*XA/(1-G)+I*F*COS(ATN(TAN(H)/G))
*(1-2*XA+XA*XA)/(2*B-A*B-B*B)
741 743 745 747
GOTO 745 LET PA LET AP IF B = 0
C*G/ (1-G) PA/F THEN 753
A-3
750 LET QA D*TAN(H)*XA*XA/(1-G)+I*F*SIN(ATN(TAN(H)/G)) *(1-2*XA+XA*XA)/(2*B-A*B-B*B)751 GOTO 755
753 LET QA D*TAN(H)/(1-G) 755 LET AQ QA/F 760 LET PQA = SQR(AP*AP+AQ*AQ) 765 PRINT PA, QA, AP, AQ, PQA 770 NEXT H 780 NEXT G 785 FOR H 0 TO 1.57 STEP 0.06981317 787 LET K = M-(M-E)*H*.636619772 788 LET I= E+(K-E)*ATN(TAN(H)/G)*.636619772 790 LET PD= I*F*COS(H) 795 LET DP PD/F 800 LET QD I*F*SIN(H) 805 LET DQ QD/F 810 LET PQD = SQR(DP*DP+DQ*DQ) 815 PRINT PD, QD, DP, DQ, PQD 820 NEXT H 830 STOP 840 DATA 0.06, 0.2, 10000, 12605, 0.87, 1100 850 END
B-1
Appendix B
100 REM BASIC COMPUTER PROGRAM FOR CALCULATION OF 110 REM FRICTIONAL FORCES USING MAXIMUM SFC 120 REM CONSTANT: A= a/21 B = b/21 C = Cx(force/slip) 130 REM D = Cy (force/rad), or D (mod.) = Cy (force/1.57rad) 135 REM IB = µ._,,r F = Fz (force) 140 REM VARIABLE: G = Sx H = a (rad) 150 REM************************************************************** 160 READ A, B, C, D, IB, F 170 LET P = A*(2-A-B) 180 LET N = 2-A+A*B*0.5+B*B*0.5-2*B 185 IF A= 0 THEN 216
ATN (F*IB/ (P*D)) 190 LET R 200 LET E = (R-1.5708+2*SIN(R)*COS(R)*(N/P-1))
*0.5*D*P/(COS(R)*COS(R)*F*(N/P-1)) 210 LET M (IB-E*R*.636619772)/(1-R*.636619772) 211 GOTO 216 212 LET R 1.5708 213 LET E = IB 214 LET M = IB 216 PRINT "a/21 = ";A, "b/21 = ";B, Fz = ";F 218 PRINT "Cx = ";C, "Cy = ";D, "µacr = "; IB 220 PRINT "acr = ";R, "µd = ";E, "µ..0 = ";M
230 LET G = (E*F*(N-P)/C+P*P*0.5)/(E*F*(N-P)/C+P*N-P*P) 235 IF A= 0 THEN 253 240 LET IA= P*C*G/(F*(l-G)) 250 LET L = (IA-E*G)/(1-G) 251 GOTO 260 253 LET IA= IB 254 LET L = IB
260 PRINT "Sxcr = ";G, "µscr = ";IA, "µso= ";L 340 PRINT" F " " X I F " " y , µ " " X I
350 FOR H = .06981317 TO 1.57 STEP .06981317 360 LET K = M-(M-E)*H*.636619772
~", "
370 LET YA= 1/(1+(2*B-A*B-B*B)*D*TAN(H)/(K*F)) 380 IF YA>= (1-B) THEN 520 390 LET YB = K*F/((2-A-B)*D*TAN(H)) 400 IF YB >= (1-B) THEN 420 410 IF YB >= A THEN 470 420 0 LET RC 425 430 435 440
LET LET LET LET
CR SC CS RSC=
RC/F K*F SC/F
SQR(CR*CR+CS*CS) 450 PRINT RC, SC, CR, CS, RSC 460 GOTO 560 470 LET RB 475 LET BR 480 LET SB 485 LET BS
500 PRINT RB, SB, BR, BS, RSB 510 GOTO 560 520 LET RA= 0525 LET AR= RA/F 527 IF B = 0 THEN 753
B-2
530 LET SA= D*TAN(H)*YA*YA+K*F*(l-2*YA+YA*YA)/(2*B-A*B-B*B) 531 GOTO 535 532 LET SA= D*TAN(H) 535 LET AS= SA/F 540 LET RSA= SQR(AR*AR+AS*AS) 550 PRINT RA, SA, AR, AS, RSA 560 NEXT H 570 FOR G 580 FOR H 590 LET J 600 LET K 610 LET I
.1 TO .99 STEP .1 0 TO 1.57 STEP 0.06981317 L-(L-E)*G M-(M-E)*H*.636619772 J+(K-J)*ATN(TAN(H)/G)*.636619772
620 LET XA = l/(1+(2*B-A*B-B*B) *(SQR(C*C*G*G+D*D*TAN(H)*TAN(H)))/(I*F*(l-G)))
630 IF XA >= (1-B) THEN 735 640 LET XB = I*F*(l-G)/((2-A-B)*(SQR(C*C*G*G+D*D*TAN(H)*TAN(H)))) 650 IF XB >= (1-B) THEN 660 655 IF XB >= A THEN 700 660 LET PC= I*F*COS(ATN(TAN(H)/G)) 665 LET CP PC/F 670 LET QC I*F*SIN(ATN(TAN(H)/G)) 675 LET CQ QC/F 680 LET PQC = SQR(CP*CP+CQ*CQ) 685 PRINT PC, QC, CP, CQ, PQC 690 GOTO 770 700 LET PB
715 LET BQ QB/F 720 LET PQB = SQR(BP*BP+BQ*BQ) 725 PRINT PB, QB, BP, BQ, PQB 730 GOTO 770 735 IF B = 0 THEN 743 740 LET PA= C*G*XA*XA/(1-G)+I*F*COS(ATN(TAN(H)/G))
*(1-2*XA+XA*XA)/(2*B-A*B-B*B) 741 GOTO 745 743 LET PA 745 LET AP 747 IF B = 0 750 LET QA
C*G/(1-G) PA/F THEN 753 D*TAN(H)*XA*XA/(1-G)+I*F*SIN(ATN(TAN(H)/G)) *(l-2*XA+XA*XA)/(2*B-A*B-B*B)
751 GOTO 755 753 LET QA = D*TAN(H)/(1-G) 755 LET AQ = QA/F 760 LET PQA = SQR(AP*AP+AQ*AQ)
765 PRINT PA, QA, AP, AQ, PQA 770 NEXT H 780 NEXT G
B-3
785 FOR H = 0 TO 1.57 STEP 0.06981317 787 LET K = M-(M-E)*H*.636619772 788 LET I= E+(K-E)*ATN(TAN(H)/G)*.636619772 790 LET PD 795 LET DP 800 LET QD 805 LET DQ
I*F*COS(H) PD/F I*F*SIN(H) QD/F
810 LET PQD = SQR(DP*DP+DQ*DQ) 815 PRINT PD, QD, DP, DQ, PQD 820 NEXT H 830 STOP 840 DATA 0.06, 0.2, 10000, 12605, 0.9324, 1100 850 END
C-1
Appendix C
PROCEDURE TO OPERATE STRAIN GAUGE AMPLIFIER
1 . Power Supply
a. When operating from 24V power supply, set small power switch to down position (large power switch can be in any position).
b. When operating from 240V main supply, set small power switch in upward position and use large power switch. The red power indicator should come in either case.
2. Balancing Strain Gauge Bridges
a. The strain gauge bridges must be balanced before use. To do this, first select the channel to be balanced (1 through 6).
b. On the appropriate channel select the required attenuation (topmost control) before going on to the next step. Notes: channel 1 [WHEEL LOAD] should be set to 21 (i.e. position between 18 and 24), channel 2 [SIDEWAY FORCE] set to 3 (i.e. position between O and 6), and channel 3 [BRAKE FORCE] set to 12.
c. Select R on the function switch (rotary switch at bottom of panel), then return control marked R so that a zero reading on the meter is read.
d. Set the function switch to Rx and turn the control marked X so that a "null" reading is obtained, that is, the point at which the needle swings closest to the zero position, note that it doesn't have to actually read zero in this case.
e. Switch back to R position on the function switch and rebalance as in step (c).
f. This completes the balancing procedure for one channel, leave the function switch in the R position for operation. Each channel being used should be balanced in this way.
3 . Operation
a. To start reading, ensure all channel function switches are set to the R position.
b. Make sure that the channel select switch (far right hand side) is set to an unused, or to the SET TO MARK position.
D-1
Appendix D
PROCEDURE TO OPERATE ACROLOG 400
1. Starting logging program
a. Switch computer on (red button on rear), and note that there is no disk in the A drive (floppy disk).
b. The C:\> prompt should now appear. c. Type in 'cd acrolab\term' <return>, note the space. d. C:\ACROLAB\TERM> should now appear. e. Type in 'term' <return>. f. Hit <return> twice. g. The Ready should now appear. h. Hit Fl. i. Type in 'truck.bas' <return>. j. Hit F3. k. At the prompt type in 'a:\filename.dat' <return>.
This tells the logger that data is to be recorded in this file on the specified drive (in this case drive A and file filename.dat). Ignore the above two steps (i.e. F3 etc.) if a disk log is not required.
Note: Wherever possible, avoid using drive C (i.e. the hard disk drive) as writing to this drive whilst in motion may cause damage to the drive.
1. Type in 'run' <return>, the logging program now starts.
2. Logging and writing data
a. Read the screen for instructions. b. Enter in a test run name at the prompt and <return>.
The data logger is now logging data and writing data to the floppy disk.
c. To stop logging hit 'CTRL C' the Ready prompt will now appear. At this point the data file on the floppy disk is still open, that is, further data samples taken on subsequent runs will be appended to this file.
3. Temporary interrupting of logging activity
a. If it is required to increase the pressure in the hydraulic system, then hit 'CTRL C' to stop the program. Now increase the hydraulic pressure.
b. The logging program now has to be reloaded into the ACROLOG, that is, hit Fl, then type in 'truck.bas' as described above. The data file is still open and should still record data, that is, F3 does not have to be hit again.
D-2
4. Returning to DOS and restarting
a. To return to DOS hit 'ALT X'. On return to DOS the data file on disk will be closed, that is, if any subsequent test runs are made logging into the same file name, the existing file will be overwritten.
b. Before travelling to another location, hit 'CTRL C' then 'ALT X' then switch the computer, the instruments and the brake controller off.
c. On restarting make sure that the data file is a different name (e.g. truckl.dat, truck2.dat) otherwise existing data will be overwritten; this must be done whenever 'ALT X' has been hit (i.e. to return to DOS) as this actually closes the data file.
E-1
Appendix E
10 REM 20 REM 30 REM 40 REM 50 REM 60 REM 70 REM 80 REM 90 REM 100 REM 110 CLS
+------------------------------------------------------------+ U.N.S.W./R.T.A. ROAD FRICTION TEST VEHICLE
PROGRAM TO READ STRAIN GAUGE AND SPEED OUTPUT VERS 2.2 MAY 1991 M.J.KERKHOF
CONVERTS, PRINTS AND LOGS ANALOG AND DIGITAL INPUTS ACRO-400 MUST BE IN SIMULATION MODE.
SET SWITCH 1-1 ON. +------------------------------------------------------------+
130 RESOLVE=13:REM SET RESOLUTION OF A-TO-D CONVERSION TO 13 BITS 140 RANGE(l)=l0:REM SET RANGE CHANNEL 1 TO+/- 10v 150 RANGE(2)=10:REM SET RANGE CHANNEL 2 TO+/- 10v 160 RANGE(3)=10:REM SET RANGE CHANNEL 3 TO+/- 10v 170 DCONFIG(3) = 0:REM SET DIGITAL LINES 17 TO 24 AS INPUT 180 COUNTER= 0:REM DISABLE COUNTER 190 COUNT= 0:REM RESET COUNTER 200 S = VAL(TIME$(7,8)) :REM CONVERT SECONDS TO A NUMBER 210 LOCATE 2,25:REM LOCATE CURSOR 220 PRINT "ACROLOG 400 DATA LOGGING PROGRAM" 230 LOCATE 5,1 240 PRINT "CTRL-C TO STOP LOGGING" 250 PRINT "CTRL-C THEN ALT-XTO RETURN TO DOS" 260 PRINT "F3 TO LOG DATA TO DISK"; 270 LOCATE 9,2 280 INPUT "ENTER TEST RUN NAME: ",N$ 290 LOCATE 9,2 300 PRINT" 310 LOCATE 11, 2
"· REM REMOVE LINE
320 PRINT "FRICTION TEST RECORDED ON ";DATE$;" AT ";N$ 330 LOCATE 13,2 340 PRINT "TIME"; TAB ( 15), "LOAD"; TAB (25), "S. W. F."; TAB (35), "BRAKE";
TAB(47),"km/h" 360 G = VAL(TIME$(7,8)) :REM READ SECONDS FROM R.T. CLOCK, CONVERT
370 380 390 400 410 420 430
SECONDS TO A NUMBER IF G = S THEN GOTO 450 ELSE GOTO 380 S = G COUNTER= 0:REM STOP COUNTER R = COUNT:REM READ VALUE IN COUNTER COUNT= 0:REM RESET COUNTER COUNTER= 1:REM START COUNTER Rl R/200:REM CALCULATE NUMBER OF WHEEL REVS/SEC
440 R2 ((Rl*l.612)*3600)/1000:REM CALCULATE SPEED IN KM/H 450 Vl VIN(l) :REM READ INPUT FROM CHANNEL 1 460 V2 VIN(2) :REM READ INPUT FROM CHANNEL 2 470 V3 VIN(3) :REM READ INPUT FROM CHANNEL 3 480 IF DIN(l7) THEN B$ = "OFF" ELSE B$ = "ON":REM READ BRAKE STATE 490 LOCATE 16,1:REM LOCATE CURSOR 500 REM NOW CONVERT AND PRINT INPUTS 510 PRINT TIME$;%8F2,TAB(12),Vl;TAB(23),V2;TAB(33),V3;TAB(43),R2;
TAB(63),B$ 520 LOCATE 20,1:REM SEND CURSOR NEAR END OF PAGE 530 GOTO 360 540 END
130 RESOLVE=13:REM SET RESOLUTION OF A-TO-D CONVERSION TO 13 BITS 140 RANGE(l)=l0:REM SET RANGE CHANNEL 1 TO+/- 10v 150 RANGE(2)=10:REM SET RANGE CHANNEL 2 TO+/- 10v 160 RANGE(3)=10:REM SET RANGE CHANNEL 3 TO+/- 10v 170 DCONFIG(3) = 0:REM SET DIGITAL LINES 17 TO 24 AS INPUT 180 COUNTER= 0:REM DISABLE COUNTER 190 COUNT= 0:REM RESET COUNTER 200 REM T$ = TIME$(7,8) :REM READ SECONDS FROM R.T. CLOCK ON STARTUP 210 S = VAL(TIME$(7,8)) :REM CONVERT SECONDS TO A NUMBER 220 LOCATE 2,25:REM LOCATE CURSOR 230 PRINT "ACROLOG 400 DATA LOGGING PROGRAM" 240 LOCATE 4,1 250 PRINT "CTRL-C TO STOP LOGGING" 260 PRINT "CTRL-C THEN ALT-XTO RETURN TO DOS" 270 PRINT "F3 TO LOG DATA TO DISK"; 280 LOCATE 9,2 290 INPUT "ENTER TEST RUN NAME: ",N$ 292 LOCATE 9,2 295 PRINT:REM REMOVE LINE 300 LOCATE 11,2 310 PRINT "FRICTION TEST RECORDED ON ";DATE$;" AT ";N$ 320 LOCATE 13,2 330 PRINT "TIME";TAB(15),"LOAD";TAB(25),"S.W.F.";TAB(35),"BRAKE";
TAB(45),"SPEED";TAB(58),"BRAKE ON/OFF" 340 G = VAL(TIME$(7,8)) :REM READ SECONDS FROM R.T. CLOCK, CONVERT
SECONDS TO A NUMBER IF G = S THEN GOTO 440 ELSE GOTO 360 S = G COUNTER= 0:REM STOP COUNTER R = COUNT:REM READ VALUE IN COUNTER REM COUNT= 0:REM RESET COUNTER COUNTER= 1:REM START COUNTER Rl R/200:REM CALCULATE NUMBER OF WHEEL REVS/SEC
350 360 370 380 390 400 410 420 430 R2 ((Rl*l.595)*3600)/1000:REM CALCULATE SPEED IN KM/H 440 Vl VIN(l) :REM READ INPUT FROM CHANNEL 1 450 V2 VIN(2) :REM READ INPUT FROM CHANNEL 2 460 V3 = VIN(3) :REM READ INPUT FROM CHANNEL 3 470 IF DIN(17) THEN B$ = "OFF" ELSE B$ = "ON": REM READ BRAKE STATE 480 LOCATE 15,1: REM LOCATE CURSOR 490 PRINT TIME$;%8F2,TAB(12),Vl;TAB(23),V2;TAB(33),V3;TAB(43),R2;
TAB(63),B$:REM CONVERT AND PRINT INPUT 500 LOCATE 18,1:REM SEND CURSOR NEAR END OF PAGE 510 GOTO 340 520 END
F-1
Appendix F
RESULTS OF INVESTIGATION USING MMFTT, WITH HIGHER SPEED (20 MPH & 30 MPH). Extracted from Dermoredjo (1992].
1. Example for 20 mph (± 32 kph) vehicle speed, 10° slip angle, 0.5 second rate of braking, 20 & 30 psi tyre pressure.
Coefl. of Frlotlon 1
0.8
o.e
OA
0.1
0
-0.2 0
-
\
\ ./
-
+ ......
{ \ .
I I - - - c.. -
.,_ -. -,,.
.
--
,.
-
- -- -
T
-
- .. - . .
I _.I!
-.
~
21
20
11
10
I
lpNd(ka/la) as
ao
21
20
11
10
a -0.2 ..__...__..___..___..1...-_ _.___.J__.J__..J__..J...._..J 0
Rate or 1wa1 .. _o.r ,.... ............. I I I i ! I I
i i I I I
~ ! l __.
I r .
I !
-I
- -
+ ,__ ~""'--k....
I I
-+- ....... ,__ Ooetf.
-
½ .
.....
I I . - -I-· -· ! I I i I ,A!' r-:+"
eo
ao
20
10
eo
80
20
10
Alllle10
Data 002
G-1
Appendix G
MATHEMATICAL DERIVATIONS OF TYRE STIFFNESS FORMULA
After Schallamach and Turner [1960], Dugoff et.al [1969], Livingston and Brown [19691, Tielking and Mital [1974].
__;,...__.!..,j,l~L.......L--...:;:'vll;.;..;H.:.:£ E:.:L;_P;..;;L;.;..;A.;.;.;NE;;.____.._ X
FRIC.TIOWI.L'"'------'I r-,, F'ORCE
Vx
Fig.G.1. The kinematics of tyre motion.
-contact leng¼h-
p' CA/tCASS ~
Fig.G.2. Left: tyre-road contact geometry. Right: deformation in adhesion region of contact patch. [Dugoff et.al (1969)].
EI
Fig.G.3. Rectangular tread elements attached to an elastically supported deformable ring of bending stiffness EI. [Tielking and Mital (197 4)].
G-2
Notation
Cx = longitudinal slip stiffness (force/slip, or force/fraction slip), or longitudinal deformation stiffness (force)
~ = lateral slip stiffness (force/degree, or force/radian), or lateral deformation stiffness (force)
= longitudinal & lateral force (force) = tyre contact length & width (unit length) = longitudinal & lateral carcass spring rate (force/length3)
= longitudinal & lateral tread element spring rate (force/length3)
= longitudinal slip (%, or per fraction slip) = tana = resultant of tyre velocity (unit length/unit time) = longitudinal & lateral tyre velocity (unit length/unit time) = free-rolling velocity (unit length/unit time) = resultant of slip velocity (unit length/unit time) = longitudinal & lateral slip velocity (unit length/unit time) = slip angle (degree, or radian) = longitudinal & lateral stress (force/unit area) = arc tan S/Sx (degree, or radian)
1. Pure longitudinal slip under braking
Consider the case of pure longitudinal slip under braking (see Figure G.1). Since the slip angle (a) is equal to zero, then the direction of motion is parallel to the wheel plane.
When the tyre is free-rolling, the rolling velocity (Vr) is equal to the longitudinal velocity (Vx) of the wheel center.
Vr = Vx (free-rolling) (G.1)
When braking is applied, the longitudinal velocity (Vx) and the rolling velocity (Vr) are reduced by a certain amount each instant in time. Elastic deformation and sliding in the contact region affect the reduction in longitudinal velocity and rolling velocity. The longitudinal velocity decreases until equal to the longitudinal slip velocity at the condition of a lockedwheel, whereas the rolling velocity decreases until equal to zero when the the wheel is locked. The difference is the longitudinal slip velocity (V sx> which is given by:
VSX = vx - vr (G.2)
It is convenient to define a parameter Sx as a fraction of longitudinal velocity which indicates the amount of braking.
sx = VsxfVx (G.3) This parameter may exhibit values from O in the case of a free-rolling wheel to 1 in the case of a locked-wheel. By Equation (G.2), the definition of Sx may be written as
G-3
Sx = 1 - VJVx Eliminating Vx from Equations (G.2) and (G.3), one obtains
VsxfVr = SJ(l-Sx)
(G.4)
(G.5)
As the angle of slip is equal to zero, the magnitude of the slip velocity Vs = I Vs I is the component of Vs in the direction of wheel plane, ~=~ ~~
Consider now the deformation in the contact patch (see Figure G.2). Since the slip angle (a) is equal to zero, the typical point P (X, Y) lies on the wheel plane.
In time at, the base point P' will move into the contact region a longitudinal distance X' determined by the rolling velocity Vr.
X' = Vrat (G.7)
If there is braking, the contacting point P, which moves relative to Y with the wheel center velocity V x, will cover an additional distance (X - X') as indicated in Figure G.2. The displacement of point P from the axis Y is
X = Vxat (G.8) Eliminating time from Equations (G.7) and (G.8) gives an expression for the longitudinal displacement (X - X').
X - X' = (Vx/Vr - l)X' (G.9) or, in terms of the longitudinal slip speed given by Equation (G.2),
X - X' = (VsxfVr)X' (G.10)
In adhesion region, the deformation (X - X') is produced by static friction with a limiting coefficient of friction µ. The static friction force per unit area of element (ax) required to produce this displacement depends upon the longitudinal stiffness (~) of the element (see Figure G.3),
ax = ~(X - X') (G.11) and, using Equation (G.10),
(Jx = ~(VsxfVr)X' (G.12)
It is convenient to define the non-slip contact length 21 on the equator of the tyre carcass to which the X,Y coordinates are fixed. (Note: 21 is the contact length of the free-rolling tyre at zero slip angle). If it is assumed that the behaviour is uniform in the Y direction, the distributed · contact shear stress CJx is defined as function of X only. The resultant force Fx, transmitted to the tyre mounting rim, is now given by the following integral.
F = w P1a dx (G.13) X Jo X
where w is the contact width.
If the rolling and slip velocities are such that X' ~ 2t there is complete adhesion over the entire contact region and from Equations (G.12) and (G.13):
G-4
Fx = 0.5 w I<x<VsxfVr)(21)2 (G.14) The use of Equation (G.5) permits the adhesive forces, given by Equation (G.14), to be written in the following way:
Fx = 0.5 w kx{Sx/(1-Sx)}(21)2 (G.15)
The longitudinal stiffness for complete adhesion, Cx is given by ex= dFJdSX (at sx = 0) (G.16)
It should be noted from Equation (G.17) that the dimension of½ is force per unit slip, whereas the dimension of 0.5 kx(21)2w is force. The right hand side of Equation (G.16) is the slip stiffness, whereas the right hand side of Equation (G.17) is the deformation stiffness.
It can be concluded that at zero slip the longitudinal slip stiffness is mathematically equal to the longitudinal deformation stiffness (but not apply for other values of slip).
2. Pure lateral slip under cornering
For the case of pure lateral slip, the longitudinal slip Sx is equal to zero, and the direction of motion make an angle a with the plane of the wheel (see Figure G.1).
Analogous to the longitudinal slip parameter, the lateral slip parameter SY is defined to be:
Sy = Vs/Vx = tana (G.18) This parameter may exhibit values from 0 (free rolling, at a = D°) to oo
("locked-wheel", at a. = 1t/2). With Sx = 0, then Vx = Vr (G.19)
and hence, Sy = Vs/Vr (G.20)
As the longitudinal slip is equal to zero, the magnitude of the slip velocity V s = I V s I is the component of V s perpendicular to the plane of the wheel,
Vs = Vsy (G.21)
Consider now the deformation in the contact patch (see Figure G.2). The typical contact point P (X, Y) lies in the direction of motion relative to the origin (point 0).
In time at, the contact point P is displaced laterally a distance Xtana Xtana = V5yat (G.22)
In adhesion region, the deformation Xtana is produced by static friction with a limiting coefficient of friction µ. The static friction force per unit area
G-5
of element (cry) required to produce this displacement depends upon the lateral stiffness (ky) of the element (see Figure G.3).
cry = kyXtana (G.23) and, using Equations (G.18) and (G.19),
cry = ky(Vs/Vr)X (G.24)
The resultant force Fy, transmitted to the tyre mounting rim, is now given by the following integral.
F =W~~ ~~ Y Jo Y
where w is the contact width, and 21 is the non-slip contact length.
The integration of Equation (G.25) under the condition of complete adhesion yields
Fy = 0.5 w ky(Vs/Vr)(21)2 (G.26) The use of Equation (G.20) permits the adhesive force, given by Equation (G.26), to be written in the following way:
FY = 0.5 w kytana(21)2 (G.27) And for small slip angle, tana "" a, hence
Fy = 0.5 w kya(21)2 (G.28)
The lateral stiffness for complete adhesion, Cy is given by CY = dF/da (at a= 0) (G.29)
Differentiating Equation (G.28) into a (at a = 0) yields, CY = 0.5 kyC21)2w (G.30)
It should be noted from Equation (G.30) that the dimension of CY is force per unit angle, whereas the dimension of 0.5 ky(21)2w is force. The right hand side of Equation (G.29) is the slip stiffness, whereas the right hand side of Equation (G.30) is the deformation stiffness.
It can be concluded that at zero slip angle the lateral slip stiffness is mathematically equal to the lateral deformation stiffness (but not apply for other values of slip angle).
EXPLANATORY NOTES
A. Modification of Summary of Conclusions
7.4. Summary of Conclusions
1. Using the data from the literature survey, the candidate proposed an
empirical model which will be able to predict the change of tyre
stiffness (C , C ) for any set of operating conditions (F , IP, V)., X y Z
once the tyre stiffness for one set of operating conditions has been
measured (known). The equations (Eqs. 2.32-2.37) are set out in
Section 2.4.7.
2. The results using the candidate's measured field data confirm a good
agreement between the predicted and the measured values of tyre
stiffness. This can be seen, for example, in Table 6.2 where one set
of operating conditions is presented in the first row, and any set
of operating conditions are presented in columns 2-4 of rows 2-4. The
predicted values in column 6 are calculated using Eguation 2.34. The
measured values in column 5 of rows 2-4 are obtained by using MMFTT.
3. The candidate has proposed a modification to an existing model for
the shape of tyre pressure in a form of trapezium, with contact
length of 21, and slope length of a and b. This model offers
flexibility in representing many types of actual pressure
distribution, by selecting the appropriate value of its slope (a/21,
b/21). The results from the literature survey (Section 1.3.5) can be
used as a guidance for choosing the slope. It was found using Sakai's
laboratory data that the selection of a/21 is more sensitive with
increasing tyre stiffness.
4. An analytical model has been developed by the candidate, which
enables the calculation of frictional forces over all ranges of
1
percent slip and slip angles. The ability to calculate frictional
forces from tyre stiffness and either locked-wheel BFC or maximum
SFC, is a feature of the candidate's model. In general, the
candidate's model is found to be in the median position when the
model response was compared with 5 tyre models: HSRI-I, HSRI-II,
HSRI-III, Good Year, and Sakai.
5. The candidate deduced mathematically from his prediction model, with
inconclusive evidence from field measurement, that the direct
contribution of tyre stiffness to the frictional forces is in the
region of below the critical percent slip (S ) and critical slip xcr
angle (a ). Beyond this area, the tyre stiffness still contributes er
indirectly, and diminishes when the condition of 100% slip (locked-
wheel) or 90° slip angle was achieved. See Figure 2.19 and Tables
2.7-2.8 for contribution of C, C andµ to frictional forces. X y
6. The way of using the candidate's analytical model for predicting the
frictional forces are presented in Sections 2.6.6-2.6.8. By knowing
(or measuring) the input of C , C , locked-wheel BFC, a/21, b/21, and X y
F , the values of F and F at various S and a can be calculated. This Z X y X
can be seen, for example, in Section 2.6.7 where ex= 10000 lbs/slip,
c = 12605 lbs/rad, locked-wheel BFC = 0.87, a/21 = 0.06, b/21 = y
0.20, and Fz = 1100 lbs. By using the candidate's analytical model,
it was found that for Sx = 0.1 and a = 4°, the Fx = 685 lbs and FY =
515 lbs. The values of Fx and FY at other values of Sx and a are
tabulated in Tables 2.7-2.8.
7. Since the locked-wheel BFC (or maximum SFC) can be predicted from
surface texture measurement (Yandell et.al (1983)], the candidate"s
analytical model can be used to predict the frictional forces, from
tyre stiffness and surface texture data. With the availability of
equipment for measuring the locked-wheel BFC or maximum SFC, the data
from this equipment can be processed directly to obtain the
frictional forces for all ranges of percent slip (Sx) and slip angles
(a). See also Figures 7.1-7.2.
2
8. The effect of some input parameters (C , C , a/21, b/21, µd and F ) on X y Z
frictional forces has been studied using the candidate"s analytical
model. Almost all parameters have significant effects on the BFC and
SFC, except for b/21 which is found to be small.
9. In general, the results from the candidate's experiment and Sakai's
laboratory data, show a good agreement between the measured values
and the predicted values (using the candidate"s analytical model).
However, for both candidate's experimental results and Sakai's
laboratory data, the predicted values of SFC (versus slip angle) at
zero percent slip, was found to under estimate the measured values.
A closer agreement can be achieved by using the modification in the
input of the candidate"s tyre model, so the input values of cy is
force/1.57 radians (or force/90°) instead of force/radian (or
force/57.3°) as in the original input of the candidate's tyre model.
B. Explanation of Section 3.3
a. There are 288 data available for the development of regression
model. These data are created from the independent variables in
Table 3.1 (2x2x2x2x2x3x3 factorial). For example:
Data 1: a/21 = 0.05, C = 10000 lbs/slip, C = 10000 lbs/rad, X y