《高等路面设计理论》课程授课教案（讲义）Prediction of TyreRoad Friction

940030Paper No.PREPRINTDuplication of this preprint for publication or saleis strictly prohibited without prior written permissionof the Transportation Research Board.Title: THE Y-M TEXTURE-FRICTION METERMARK 2(New Title: The Prediction ofTire-Road Friction from TextureMeasurements)Author($): William O.YANDELL & S.SAWYERTransportation Research Board73rdAnnual MeetingJanuary 9-13,1994Washington, D.C

MEETINGANNUALRESEARCHBOARDTRANSPORTATIONTHEWASHINGTONDCUSA9-13JAN1994The Prediction of Tyre-Road FrictionfromTextureMeasurementsbyW.O.Yandell and S. Sawyer4School of Civil EngineeringUniversity of NewSouth WalesP.O. Box 1, Kensington, N.S.W. 2033,Australia

SUMMARYFor the past two decades the author and his research students have been working on apurely theoretical means for predicting tyre-road friction.It is based on a faithfulsimulation of a pneumatic tyre sliding over the wet texture of the road surface. Thisinvolved the stress-gross strain analysis of the tread rubber, the effect of shear rate,heat and lubrication. A device called the Yandell-Mee Texture Friction Meter isdescribed here and is the end product of this research. If when placed on a roadsurface it samples a 60 cm long total texture profile to an accuracy of 0.05 mm andpredicts side force and locked wheel wet friction for three speeds in seconds. Sincethe result only varies with texture changes this is an excellent control tool forpavementengineers.1.INTRODUCTIONThe author and his colleges have been studying the part played by surface texture ontyre-road friction since 1968 (see references 1 to 12). Much of the work wasinfluenced by that of Tabor [13] and Kummer and Meyer [14]. It was assumed thattyre-road friction was due to hysteretic energy loss in the tread rubber as it flowedover the road surface texture and inter molecular adhesion would not occur on wetroads. Yandell [3] summarised some of the basic elements involved in hystereticsliding friction and the change in micro topology of road surfaces in service. Theauthor [3] set out the principles of the mechano-lattice stress-strain analysis for grossdeformations used in the prediction of hysteretic friction from one texture parameterthe average absolute slope. It was also shown how the friction of small stone surfaceslubricated with liquids of various viscosities sliding on tread rubber could be predictedinthelaboratory

Before the mechano-lattice stress-strain analysis can be used the damping and resilientproperties of the tread rubber as they vary with strain, rate of strain and temperaturemust be known. This work was performed by Zankin [ll] using a temperaturecontrolled apparatus capable of measuring damping in rubber sliding at up to80 km/hr. Taneerananon [10] modified Reynold's equations for sliding and sinkageto use with the mechano lattice stress-strain analysis so that masking water filmthicknessescouldbedetermined.The author's friction prediction was based on the concept that a profile of the roadsurface texture could be broken up into a number of components ranging from courseto fine. While large volumes of rubber were expending energy as they flowed overthe coarsest scales, smaller shallower volumes of rubber simultaneously expendedenergy as they flowed overthe finer scales of texture.The total hysteretic friction wasthe sum of frictions generated on each scale of texture.The friction on a scale was afunction of the effectivedamping factor of the rubber and the average absolute slopeof that scale.Thedampingfactor is the energylost divided bythe energyapplied indeforming rubber in a load-unload operation [3]. The average absolute slope is afunctionoftextureroughness.The next stage in the development of a system for predicting wet friction from roadsurface texture involved measuring the texture of a number of roads of diverse surfacetexture with either bituminous or concrete surfacing [12]. The coarse texture wasmeasured with a profile former (row of needles) the fine by Ziess light sectionmicroscope.The total texture was divided into four scales. The dry hysteretic frictionwas determined from the average absolute slope of that scale of texture and thedamping factor of the tread rubber using the mechano-lattice analysis [12]. Thecoefficients of wet sideways force and locked wheel braking friction for speeds of16,48 and 80km/hrwere computed and compared withpredicted values measured

by a multi-mode friction measuring truck. An example of a correlation for lockedwheel braking is shown in figure 1. The R-squared was 0.7.The above process though reasonably accurate was clumsy and time consuming.Accordingly the author devised a portable device that would do the same job inseconds. It was called the Yandell-Mee Texture Friction Meter. Dr Mee designed thecircuit boards and wrote the Pascal programs that controlled the original meter'soperation. The meter simulates the behaviour of a smooth pneumatic passenger cartyre travelling on a wet pavement. A later version (mk 2) of the Y-M Texture FrictionMeter was developed with the assistance of S. Sawyer and will now be described2.THEYANDELL-MEETEXTUREFRICTIONMETERMARKIThe first portable YM Texture Friction Meter was built under the sponsorship ofPavement Management Services Ltd in Sydney.They incorporated it in theirAustralian Road Evaluation Vehicle (AREV) where friction measurements were madesimultaneously with other pavement characteristics in Australia and Indonesia. TheYM Texture Friction Meter Mark II was developed from the Mark I model at theUniversity of N.S.W. with the N.S.W. State Road Authority's financial support.Mark II is superior to Mark Iin that it is operator independent, has a 60 cm longtexture profile sample, is surface brightness independent and is faster.2.1.GeneralDescriptionThe portable instrument has two main components -the compact surface texturemeasuring unit and the P.C. with screen that controls the entire operation.Figure 2 isa very much summarised flow chart showing the operation of the device

The 60 cm long profile is read with an accuracy of 0.05 mm by means of aA.black and white video camera viewing the image of a laser line projected at anangle onto the surface (figure 3).Any gaps in the profile are filled in. Thisprofile, recorded digitally as 12,000 ordinates, is divided into 4 bands with afifthorderBesselFilter.If the average absolute slope of the coarsest component is greater than anB.arbitrary 0.1 the surface is regarded as "rough" and the program goes to C fordrain path length computation. If the surface is "smooth" the program goes toD where a longer drainage path is computed.The number of large asperities in the hypothetical contact patch of the tyre onC.the rough surface, that part of the texture in contact with the tyre and thedrainagepath lengths are computed. Then move toE andF.D.Thedrainage path length on the smooth surface is assumed equal to the radiusof the contact patch.Then moveto E and FE.The average absolute slope of the texture, is computed, of that part of eachscale that is in contact with the tyre. (The average absolute slope of the twosides of an equilateral triangle for example is 3)The average water film thickness and the ratio of a/d is computed.a' is the asperityF.height 'd' of the second finest scale of texture minus the water film thickness.This is done for each of the speeds 18, 48 and 80 km/hr and for sidewaysforce coefficient and the locked wheel braking force coefficient.Then move toI

The damping factor of the rubber is determined for each of the three speeds ofG.sliding and for sideways and for locked wheel friction. The effect oftemperature rise during locked wheel braking is accounted for. Zankin [11]provided this information.H.The coefficient of dry hysteretic friction C, is calculated for each scale oftexture using the damping factor of the rubber and the average absolute slopeofthatscaleoftexturetogiveCo,C,CCI.The coefficient of wet (not flooded) hysteretic friction is equal to the sum ofthe coefficients of dry hysteretic friction each modified by the effect of surfacewater film thusThe Coeff of Wet Friction = Co + 三 ci +Cwhere a/d is the film thickness ratio, a is the height of the coarsestcomponent of "microtexture" not masked by the water film, d is theaverage depth of the coarsest component of "microtexture"and C arethe coefficients of hysteretic friction computed from average absoluteslope and tread damping factor using the mechano-lattice analysis [10]Theexpression is Tanueerananon's hypothesis.2.2TheTextureMeasuringDeviceThe texture measuring device is housed in a light weight case measuring 40 cm by50 cm and 40 cm high. A slot in the base of the case allows the laser to be projectedonto the road surface and viewed by the video camera from inside the case. Thisportable unit is connected by a long cable to a 486 Compac P.C. which controls the

operation. The components of the texture measuring device are the laser source, thecamera and thelaser-camera transport.2.2-1 The TransportThe laser and the camera are fitted to a cross carriage which is sequentially moved intothree alternate positions by fixed slides situated at each end of the main slide.Themain carriage carrying the cross slide is driven by a motor through a lead screw for adistance of 20 cm shown in figures 4 and 5. In this way three parallel profiles each 20cm long can be recorded automatically in the one operation.2.2-2TheLaserSourceThe laser is a 5 mw 670 nm laser diode.Its fine cylindrical beam is changed to a flatknife by passing it through a cylindrical lens. It impinges on a miror which reflects itOnto the road surface in view of the video camera.See figures 4 and 5.2.2-3TheCameraandLensThe black and white video camera views the laser line impinged on the road surfacethrough powerful magnifying lenses. The laser beam and the line of sight of thecamera are mutually at right angles so the line is always in focus. The aperture of thelens is adjusted automatically. The magnitifcation of the lens is such that onecentimetre of the surface is viewed ata time

2.3TheComputerOperation andOutputThe texture measuring device is placed on the road.Upon iniation the computerquickly tracks the edge of the laser line to give a profile of texture 10 mm long asshown in figure 6.60 of these profiles are sequentially shown on the monitor andrecorded whilethecarriageis transportedautomaticallyone centimetre ata time.Oncethe profile data is stored in the P.C. the processing is effected as shown in section2.1.The 60 parts of the profile are accurately connected and missing pieces are flledin.All the ordinates are divided by 2 to give the vertical resolution of the 45°view of theprofile.Figure 7shows thetotal texture and its four components as shown onOutputthe monitor screen.Figure 8 shows the main screen display with coefficient offriction values average and peak texture depths and the menu for other operations.The inappropriate three decimal places will be modified to a more appropriateaccuracy.The contents of other files can also be shown on the screen and or printed.Forexample figure9 is an example of a block file showing the chainage,texture depth,locked wheel and sideways force friction for any number of readings along a road.Figure 10 is an example of a pointfile which gives the six friction readings plus filmthickness ratios anddryhystereticfrictionvalues.CORRELATIONWTTHDIRECTLYMEASUREDFRICTION3.There are a great number of devices which measure pavement friction directly with atesttyre.Thereis seldomcompleteagreementbetween anytwo,measuringfrictionon the same surfaces.For example a Runway Friction Tester [15] and a Pavement

4.CONCLUSIONThe main advantage in predicting tyre-road friction from total texture measurements isbased on the fact that the smooth pneumatic tyre the behaviour of which is beingsimulated has fixed characteristics. The initial hypothical water film thickness is alsofixed at O.5mm.Any variation in the predicted friction for a particular speed is duesolely to a change in the road surface texture. The surface texture is under the controlof the road authority and so can be used a trigger in pavement maintenancemanagement. The recorded texture can also be used for other investigations such astyre-road noise generation. A disadvantage, of course, is the need for the measuredsurface to be free from water and detritis where tire contact occurs.5.ACKNOWLEDGEMENTSThe Roads and Traffic Authority of N.S.W. supplied the funds for the authors toupdate the TF meter. The authors would like to thank Dr W.H. Cogill for hiscontinued help and advice.REFERENCES1. YANDELL, W.O., "A Mathematical Simulation of Hysteretic Sliding Friction"Proc. 4th Conf. ARRB, October, 1968.2.YANDELL,W.O.,"The Effect of Surface Geometry on the Lubricated SlidingFriction and Polishing of Roadstones ". Aust. Road Research, Vol 3, No 10,1969