xml version 1.0 encoding UTF8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchemainstance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 crbnuuuuuu
008 s2009 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14SFE0003284
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Fuentes, Luis.
0 245
Investigation of the factors influencing skid resistance and the international friction index
h [electronic resource] /
by Luis Fuentes.
260
[Tampa, Fla] :
b University of South Florida,
2009.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
Includes vita.
502
Dissertation (Ph.D.)University of South Florida, 2009.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: This dissertation is compiled of the findings of several phases of a detailed research study that was aimed at investigating the Skid Resistance phenomenon. In the first phase of the dissertation research a study was performed to evaluate the different factors that influence frictional measurements obtained using the Dynamic Friction Tester (DFT). A temperature calibration factor that would account for temperature effects on DFT readings and IFI computations was developed. In addition, other variables that also affect the friction measurements obtained using the DFT are identified. In the next phase of the dissertation research the effect of pavement roughness on the Skid Resistance was investigated. The variation of the normal load and its nonlinear relation to SN was used to explain lower SN values measured on relatively rougher surfaces. The feasibility of using the International Roughness Index (IRI) and the Dynamic Load Coefficient (DLC) as predictors of the reduction in SN due to pavement roughness was also investigated. In the final phase of the dissertation research an indepth investigation was carried out to better understand the principles underlying the concept of the International Friction Index (IFI), and specifically the role played by the Speed Constant (Sp) parameter in the IFI computations. The parameter Sp dictates the speed variation of the wet friction measurements taken on a given pavement surface. The results of the current investigation suggest the revision of the procedure for computation of the Sp parameter to incorporate device specific properties. Furthermore, the incorporation of vehicle characteristics in the Sp parameter computations would help address a well known deficiency of the IFI, which is the inconsistent FR60 (predicted friction at 60 km/h) obtained from the friction values measured at two different slip speeds on the same surface. This study also showed that the modification of the Sp parameter reduces significantly the slip speed dependency of the device calibration parameters A and B. Finally, a modified IFI procedure that incorporates device specific slip conditions is presented. The modified IFI procedure showed consistently better predictive capability than the conventional ASTM procedure on all the different devices considered in this study.
590
Advisor: Manjriker Gunaratne, Ph.D.
653
Skid resistance
IFI
Roughness
Speed constant
Macrotexture
690
Dissertations, Academic
z USF
x Civil and Environmental Eng.
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.3284
PAGE 1
Investigation of the Factors Influenc ing Skid Resistance and the International Friction Index by Luis G. Fuentes A thesis submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Civil and Environmental Engineering College of Engineering University of South Florida CoMajor Professor: Manjriker Gunaratne, Ph.D. CoMajor Professor: Daniel Hess, Ph.D. Gray Mullins, Ph.D. Jian John Lu, Ph.D. George Yanev, Ph.D. Date of Approval: November 6, 2009 Keywords: Skid resistance, IFI, Ro ughness, Speed constant, Macrotexture Copyright 2009, Luis G. Fuentes
PAGE 2
DEDICATION To my parents, Luis Guillermo and Maria L ourdes who supported me all these years, and to Mercy who stood by my side the whole way.
PAGE 3
ACNOWLEDGMENTS I wish to thank my mayor ad visor, Dr. G, for his constr uctive advice, and patience throughout my graduate study. I w ould also like to express my most sincere appreciation to Dr. Gray Mullins who was always there for me when I needed advice. I also wish to extend my gratitude to Dr. Daniel Hess, whose guidance helped me go thru many problems I encounter during my studies. During my graduate study, fi nancial support provided by the National Aeronautics and Space Administration (NASA) is gratefully acknowledged. I feel indebted to my parents for their co ntinuous support. Special appreciation and love go to my fianc, Mercy Gomez, who has been supportive during these past years. Without their help, patience, and encouragement, I could not complete my doctoral studies.
PAGE 4
i TABLE OF CONTENTS LIST OF TABLES............................................................................................................. iv LIST OF FIGURES......................................................................................................... viii ABSTRACT ... xi CHAPTER 1 PROBLEM STATEMENT........................................................................... 1 1.1 Review of Literature on Friction...............................................................................2 1.2 The Classic Laws of Friction....................................................................................3 1.3 Mechanics of Tire P avement Friction...................................................................... 4 1.4 Models for Tire Pavement Interaction...................................................................... 7 1.4.1 Models of Highway and Aviation Industries.....................................................7 1.4.2 Models of Automobile Industry......................................................................... 7 1.5 Friction Measuring Devices...................................................................................... 7 1.5.1 Pavement Friction Measuring Vehicles ............................................................. 8 1.5.1.1 Locked Wheel Trailer (LWT)..................................................................... 9 1.5.1.2 Runway Friction Tester (RFT).................................................................... 9 1.5.2 Laboratory Methods.........................................................................................10 1.5.2.1 Dynamic Friction Tester (DFT)................................................................ 10 1.5.2.2 British Pendulum Tester (BPT)................................................................ 10 1.6 Parameters Affecting TireP avem ent Friction Interaction......................................11 1.6.1 Pavement Surface Characteristics.................................................................... 11 1.6.1.1 Parameters Used for Texture Characterization......................................... 13 1.6.1.1.1 Mean Profile Depth (MPD)............................................................... 14 1.6.1.1.2 The International Roughness Index (IRI)..........................................14 1.6.1.1.3 Root Mean Square (RMS).................................................................15
PAGE 5
ii 1.6.2 Vehicle Operational Parameters......................................................................16 1.6.2.1 Slip Ratio.................................................................................................. 16 1.6.2.2 Vehicle Speed........................................................................................... 17 1.6.3 Tire Characteristics..........................................................................................17 1.6.3.1 Tire Tread.................................................................................................. 17 1.6.3.2 Tire Inflation Pressure............................................................................... 18 1.6.4 Environmental Factors..................................................................................... 18 1.6.4.1 Pavement Surface Temperature................................................................ 19 1.7 International Friction Index (IFI)............................................................................19 1.8 Scope of Investigation.............................................................................................21 1.9 Organization of Dissertation...................................................................................22 CHAPTER 2 FACTORS INFLUENCING FRICTIONAL MEASUREMENTS USING THE DYNAMIC FRICTION TESTER (DFT) ............................. 23 2.1 Introduction............................................................................................................. 23 2.2 Experimental Setup.................................................................................................25 2.3 Results of Experiments........................................................................................... 26 2.4 Variation of Friction Measurements due to Environmental Conditions................. 31 CHAPTER 3 EVALUATION OF THE EFFECT OF PAVE MENT ROUGHNESS ON SKIDRESISTANCE............................................................................33 3.1 Significance and Standardization of Pave ment Friction Measurements................. 33 3.2 Limitations of the Current Friction Models............................................................ 34 3.3 Objectives of the Current Study.............................................................................. 36 3.4 Experimental Program............................................................................................36 3.4.1 Equipment Used in the Study.......................................................................... 37 3.4.1.1 Circular Track Meter (CT Meter)............................................................. 37 3.4.1.2 Dynamic Friction Tester (DFT)................................................................ 37 3.4.1.3 Locked Wheel Tester (LWT).................................................................... 37 3.4.2 Selection of Test Sections................................................................................38 3.5 Results of Texture Analysis.................................................................................... 40
PAGE 6
iii 3.6 Results of Friction Testing...................................................................................... 43 3.6.1 Effect of Roughness on Friction......................................................................43 3.6.2 Effect of the Normal Load on Friction............................................................46 3.6.3 Explanation of the Abnormal SN vs. Load Behavior ....................................... 51 3.7 Modeling the Dynamic Effects of Pavement Roughness....................................... 52 3.7.1 Model Development and Validation................................................................ 52 3.7.2 Simulation of Friction Measurements.............................................................. 56 CHAPTER 4 EVALUATION OF THE SPEED CONS TANT (Sp) AND ITS EFFECT ON THE CALIBRATION OF FRIC TION MEASURING DEVICES....... 61 4.1 Standardization of Friction Measurements............................................................. 61 4.1.1 Investigation of the Validity of the IFI Concept..............................................63 4.1.2 Assumptions Governing the IFI Concept........................................................64 4.2 Objectives of the Current Investigation.................................................................. 66 4.3 Data Collection.......................................................................................................66 4.4 Analysis of Data......................................................................................................67 4.4.1 Effect of the Slip Speed on FR60 ....................................................................67 4.4.2 Speed Constant ( Sp) and the Significance of the a and b Parameters.............. 72 4.4.3 Device Dependency of the a and b Param eters............................................... 73 4.4.4 Slip Speed Sensitivity of the A and B Param eters............................................ 79 4.4.5 Effect of the Use of a Modified Sp Parameter in IFI Standard Correlation..... 81 4.4.6 Prediction Capabilities of the Proposed Models.............................................. 88 4.4.7 Modified International Friction Index............................................................. 93 CHAPTER 5 CONCLUSIONS........................................................................................ 95 REFERENCES...............................................................................................................100 ABOUT THE AUTHOR.104
PAGE 7
iv LIST OF TABLES Table 1 Friction level classifica tion of runway pavem ent surfaces (Adapted from FAA, 1997).................................................................................. 9 Table 2 Testing temperature combination........................................................................ 25 Table 3 All possible regressions....................................................................................... 29 Table 4 Summary of statistics........................................................................................... 30 Table 5 Analysis of variance of microtexture ( DF T20) on pavement A........................... 42 Table 6 Analysis of variance of microtexture ( DF T20) on pavement B............................ 42 Table 7 Analysis of variance of macrotexture ( MPD) on pavem ent A............................ 42 Table 8 Analysis of variance of macrotexture ( MPD) on pavem ent B............................. 42 Table 9 Analysis of variance of friction measurements ( SN ) including the interaction variable Speed*Roughness on pavement A....................................................... 45 Table 10 Analysis of variance of friction measurements ( SN ) including the interaction variable Speed*Roughness on pavement B....................................................... 45 Table 11 Analysis of variance of friction measurements ( SN ) on pavem ent C based on Speed and Normal Load..................................................................................... 47 Table 12 Analysis of variance of friction measurements ( SN ) on pavem ent D based on Speed and Normal Load..................................................................................... 47 Table 13 Pairwise comparison among Skid Num bers means at different load configuration using Tukeys HSD Test for pavement C...................................48 Table 14 Pairwise comparison among Skid Num bers means at different load configuration using Tukeys HSD Test for pavement D...................................49 Table 15 Parameters characterizing the frictionload de pendency of pavements C and D ..............................................................................................................51 Table 16 Dynamic model parameters............................................................................... 53 Table 17 Texture characteristics of tested pavement surfaces on 2007 ............................ 68
PAGE 8
v Table 18 Comparison of FR60 m eans obtained from different slip speeds using the FAA RFT07....................................................................................................... 70 Table 19 Comparison of FR60 m eans obtained from different slip speeds using the DND GT07......................................................................................................... 70 Table 20 Comparison of FR60 m eans obtained from differe nt slip speeds using the Illinois E27407.................................................................................................71 Table 21 Comparison of FR60 m eans obtained from different s lip speeds using the TC SFT8507..................................................................................................... 71 Table 22 Statistical analysis of the a and b param eters of the model presented in Figure 37............................................................................................................73 Table 23 Texture characteristics of tested pavement surfaces on 2008 ............................ 74 Table 24 Statistical analysis on the a and b param eters for different friction measuring devices.............................................................................................. 77 Table 25 A and B param eters calculated in accordance with ASTM standards................ 80 Table 26 A and B param eters calculated using revised a and b parameters..................... 80 Table 27 Evaluation of th e correlatio n between FR60 obtained at different slip speeds and F60 values using the ASTM method.............................................. 83 Table 28 Evaluation of th e correlatio n between FR60 obtained at different slip speeds and F60 values using the modified Sp method......................................84 Table 29 Evaluation of the corr elation between transform ed FR60 using the logarithm transformation and F60 values using the modified Sp method at different slip speeds.......................................................................... 85 Table 30 Evaluation of the corr elation between transform ed FR60 using the square root transformation and F60 values using the modified Sp method at different slip speeds.......................................................................... 85 Table 31 Evaluation of the corr elation between transform ed FR60 using the cube root transformation and F60 values using the modified Sp method at different slip speeds.......................................................................... 85 Table 32 Evaluation of the corr elation between transform ed FR60 using the fourth root transformation and F60 values using the modified Sp method at different slip speeds.......................................................................... 86
PAGE 9
vi Table 33 Evaluation of the corr elation between transform ed FR60 using the fifth root transformation and F60 values using the modified Sp method at different slip speeds.......................................................................... 86 Table 34 Evaluation of the corr elation between transform ed FR60 using the sixth root transformation and F60 values using the modified Sp method at different slip speeds.......................................................................... 86 Table 35 Evaluation of the corr elation between transform ed FR60 using the seventh root transformation and F60 values using the modified Sp method at different slip speeds.......................................................................... 87 Table 36 Summary of predicted F60 values for the different m ethod used for the VTTI GT............................................................................................................ 88 Table 37 Summary of predicted F60 values for the different m ethod used for the DND GT08......................................................................................................... 88 Table 38 Summary of predicted F60 values for the different m ethod used for the Dynatest RFT.....................................................................................................89 Table 39 Summary of predicted F60 values for the different m ethod used for the FAA RFT08....................................................................................................... 89 Table 40 Summary of predicted F60 values for the different m ethod used for the USF RFT............................................................................................................89 Table 41 Summary of predicted F60 values for the different m ethod used for the USF E274........................................................................................................... 89 Table 42 Summary of predicted F60 values for the different m ethod used for the VDot E27408....................................................................................................89 Table 43 Summary of predicted F60 values for the different m ethod used for the PTI E274............................................................................................................90 Table 44 Summary of predicted F60 % Errors for the different m ethod used for the VTTI GT............................................................................................................ 90 Table 45 Summary of predicted F60 % Errors for the different m ethod used for the DND GT08......................................................................................................... 90 Table 46 Summary of predicted F60 % Errors for the different m ethod used for the Dynatest RFT.....................................................................................................90
PAGE 10
vii Table 47 Summary of predicted F60 % Errors for the different m ethod used for the FAA RFT08....................................................................................................... 91 Table 48 Summary of predicted F60 % Errors for the different m ethod used for the USF RFT............................................................................................................91 Table 49 Summary of predicted F60 % Errors for the different m ethod used for the USF E274........................................................................................................... 91 Table 50 Summary of predicted F60 % Errors for the different m ethod used for the VDot E27408....................................................................................................91 Table 51 Summary of predicted F60 % Errors for the different m ethod used for the PTI E274............................................................................................................91 Table 52 Summary of the average predicted % Errors of the devices that operate in the range of 1020% slip condition .................................................................... 92 Table 53 Summary of the average predicted % Errors of the devices that operate at 100% slip condition .......................................................................................92
PAGE 11
viii LIST OF FIGURES Figure 1 Mechanism of friction at the tire pa vem ent interface (Adapted from Moore, 1975)............................................................................... 6 Figure 2 Friction components: Adhesion and Hysteresis (Adapted from Moore, 1975)............................................................................... 6 Figure 3 Texture effect on friction.................................................................................... 13 Figure 4 QuarterCar Model............................................................................................. 14 Figure 5 Coefficient of friction vs. Slip ratio on different surfaces (Adapted from NCHRP, 2009)..........................................................................16 Figure 6 Effect of tire inflati on pressure on tire stiffness ................................................. 18 Figure 7 Typical out put from the DFT............................................................................. 24 Figure 8 Average DFT measurements. Friction Coeffi cient vs. W ater Temperature in the range of 90 to 100 F (Hot Medium).. Figure 9 Average DFT measurements. Friction Coe fficient vs. W ater Temperature in the range of 65 to 80 oF (Medium).27 Figure 10 Combined effect of surface and water tem perature on the coefficient of friction......................................................................................... 27 Figure 11 Change in dynamic viscosity of water with temperature................................. 28 Figure 12 3D plot of model (Equation (13)) fitted surface an d actual measurement of coefficient of friction..................................................................................... 31 Figure 13 Plot of fitted data vs. measured DFT20 (Equation (13))................................... 31 Figure 14 Typical seasonal vari ation of Skid Number for an asphalt pavem ent site located in Lucas County, Ohio (A dapted from Bazlamit et. al., 2005)....... 32 Figure 15 Skid Number vs. Temperature for an asphalt pavem ent site located in Lucas County, Ohio...........................................................................................32 Figure 16 Evaluation of the effect of repe ated m easurements on the Skid Number........ 39 Figure 17 BoxPlots of texture comparison between test and c ontrol sections of pavem ents A and B............................................................................................41
PAGE 12
ix Figure 18 Measured roughness characteristics of subsections of pavement A............... 43 Figure 19 Measured roughness characteristics of subsections of pavements B.............. 43 Figure 20 Effect of roughness on the Skid Numbers of (a) pavem ent A and (b) pavement B................................................................................................... 44 Figure 21 Frequency decomposition of the pavem ent profiles......................................... 45 Figure 22 Effect of normal load on Skid Numbers on: (a) pave m ent C and (b) pavement D.................................................................................................. 46 Figure 23 Confidence intervals for all differences in m eans of all pair of load combination on: (a) pavement C and (b) pavement D....................................... 49 Figure 24 Nonlinear model representing the e ffect of Nor mal Load on Skid Numbers... 50 Figure 25 Measured vs. Predicted LoadSN relationship at different speeds ...................50 Figure 26 LWT half trailer vibration model..................................................................... 53 Figure 27 Instrumentation on the LWT............................................................................ 55 Figure 28 Input profile for the validati on of the system (pavement bump)...................... 55 Figure 29 Frequency spectrum of the accelerometer readings corresponding to Mt........56 Figure 30 Measured and pred ic ted velocities of: (a) Mw and (b) Mt.................................56 Figure 31 SN vs. IRI on the sim ulated profile of (a) pavement C at 30 mph, (b) pavement C at 55 mph, (c) pavement D at 30 mph, and (d) pavement D at 55mph.................................................................................. 57 Figure 32 3D representation of IRI with respect to frequency and am plitude.................. 58 Figure 33 SN vs. Frequency of sim ulated pavement D with amplitude of 20 mm........... 59 Figure 34 SN vs. DLC on (a) pavement C and (b) pavem ent D........................................ 60 Figure 35 Relation among FRS, S, FR60 and Sp...............................................................65 Figure 36 FR6 0 values obtained from different devices on the same pavement surface E.............................................................................................................68 Figure 37 MPD vs. Sp obtained experimentally for all friction measuring devices using the 2008 NASA Wallops data..................................................................72 Figure 38 Measured coefficient of friction vs. slip speed relationships of friction m easuring devices on test pavement surfaces.................................................... 75 Figure 39 FR60s calculated from different slip speeds on surface Echo2 for the 2008 FAA RFT........................................................................................................... 76
PAGE 13
x Figure 40 Sp vs. MPD for the different specific devices used in the 2008 NASA workshop................................................................................................78 Figure 41 Dynatest RFT correlations for Equation 11...................................................... 82 Figure 42 PTI 274 correlations for Equation 11............................................................... 83 Figure 43 Power Transformation vs. R2 on the friction measuring devices that operate in the 1020% slip condition range.......................................................87
PAGE 14
xi Investigation of the Factors Influencing Skid Resistance and the International Friction Index by Luis G. Fuentes ABSTRACT This dissertation is com piled of the findings of several phase s of a detailed research study that was aimed at investigati ng the Skid Resistance phenomenon. In the first phase of the dissertation res earch a study was performed to evaluate the different factors that influence frictiona l measurements obtained using the Dynamic Friction Tester (DFT). A temperature cal ibration factor that would account for temperature effects on DFT readings and IFI computations was developed. In addition, other variables that also affect the fric tion measurements obtained using the DFT are identified. In the next phase of the dissertation rese arch the effect of pavement roughness on the Skid Resistance was investigated. The vari ation of the normal load and its nonlinear relation to SN was used to explain lower SN values measured on relatively rougher surfaces. The feasibility of using the In ternational Roughness Index (IRI) and the Dynamic Load Coefficient (DLC) as predictors of the reduction in SN due to pavement roughness was also investigated.
PAGE 15
xii In the final phase of the dissertation research an indepth investigation was carried out to better understand the principles underlying the concep t of the Inte rnational Friction Index (IFI), and specifically the role played by the Speed Constant ( Sp) parameter in the IFI computations. The parameter Sp dictates the speed variation of the wet friction measurements taken on a given pavement surfac e. The results of the current investigation suggest the revision of the pro cedure for computation of the Sp parameter to incorporate device specific properties. Furthermore, the incorporation of vehicle characteristics in the Sp parameter computations would help address a well know n deficiency of the IFI, which is the inconsistent FR60 (predicted friction at 60 km/h) obtained from the friction values measured at two different slip speeds on the same surface. Th is study also showed that the modification of the Sp parameter reduces significantly the slip speed dependency of the device calibration parameters A and B Finally, a modified IFI procedure that incorp orates device specific slip conditions is presented. The modified IFI procedure showed consistently better predictive capability than the conventional ASTM pr ocedure on all the different devices considered in this study.
PAGE 16
1 CHAPTER 1 PROBLEM STATEMENT Skid resistance is the force developed when a tir e that is fully or partially prevented from rolling slides along a pavement surface unde r lubricated conditions. Accidents due to skidding on pavement are a major concern of the aviation and highway industries. These accidents are generally attributed to skid re sistance deficiencies on the pavement surface. Hence the tire pavement friction interacti on mechanism is one of the most important issues of safe vehicle/aircra ft operations on pavements. Different models are available to simulate the tire pavement interaction and predict pavement friction. These models have di fferent formats and outputs depending on the industry in which they were de veloped. The investig ator was able to identify clearly two different industries in which different mode ls were used. These are; the highway and aviation industries and the automobile indust ry. Different friction m easuring devices have been developed by these industries to evaluate the frictio nal properties of a pavement surface. These devices operate under different principles; therefore, direct comparison between equipment is inappropriate. The American Society for Testing and Materi als (ASTM) provides specification for the standardization of different friction measur ing devices and computation of different indices for comparison of friction values meas ured by different equipment on the same surface. Specifically, the International Friction Index (IFI) defined in ASTM E 1960 (ASTM E196007, 2009) is used as the stan dard for comparison of friction values measured by different equipment. IFI has b een developed in the Permanent International Association of Road Congresses (PIARC) In ternational experiment (Wambold et. al.,
PAGE 17
2 1995) for the purpose of harmonizing friction m easurements from different equipment to a common calibrated index. Furthermore, The National Aeronautics and Space Administration (NASA) has held Annual Runway Friction Workshops at the Wallops Flight Facility since 1993 on the eastern shore of Virginia. Th e main objective of this workshop is to calculate the IFI index from different devices that particip ate in it by evaluating the standardization parameters for each equipment. The workshop al so serves to create an extensive friction database that would be used for subsequent research purposes. 1.1 Review of Literature on Friction A com prehensive history of research on tire pave ment interaction is available, because of the diverse efforts and background of many inve stigators interested in this particular topic. Many aircraft accidents have occurred around the world, particularly related to the aviation industry due to skidding. In some of th ese cases the aircraft had not been able to stop properly on the landing strip due to the improper understandin g of the braking operation. There are many variables that are involved in this phenomenon. The understanding of each of these variables will help one to realize the magnitude of the problem that one faces and eventually design mo re reliable friction measuring techniques. The coefficient of friction is defined as follows: S NF F (1) The coefficient of friction is an abstract quantity used to expr ess the proportionality between the normal force (FN) and the shear (frictional) force (FS) of two parallel surfaces that are compressed together. Many devices operating under different mechanisms have been developed to measur e the coefficient of friction of pavement surfaces. Unfortunately, different coefficient of friction values are obtained on the same surface when different devices are used; ther efore direct comparison between coefficient
PAGE 18
3 of friction obtained from different devices is inappropriate. The coefficient of friction is not a material property. In other words, one cannot state that a certain surface has a specific coefficient of fricti on. Instead, it is a system property with its magnitude depending on both two surfaces that are in cont act. This is the reason why one must study the tirepavement interaction as a system, instead of characteri zing the surface only. 1.2 The Classic Laws of Friction The cla ssic laws of fricti on evolved from the early work of Amontons and Coulomb. These laws were based on empirical observa tions and can be summarized as follows: (1) Friction is independent of the a pparent or nominal contact area. (2) Friction force is proportional to the normal load. (3) Static coefficient of friction is greater that the kinetic coefficient of friction. (4) Kinetic friction is independent of the sliding speed. The friction at the rubber ti repavement interface constitutes a complex phenomenon due to the viscoelastic nature of the rubber. Empirical work conducte d by many investigators shows that the classical laws of friction ar e not valid on viscoelastic materials. Denny (1953) conducted laboratory experiments on r ubberlike materials and showed that under contaminated conditions the coefficient of friction decreases with increasing contact pressure. Thirion (1946) confirmed the load dependence of rubber friction and proposed an empirical relationship between the coefficient of friction and pressure. Schallamach (1952) showed that the load dependence of rubber friction can be explained by assuming spherical surface asperities and elastic behavior of rubber in compression. Although the mechanisms of tirepavement fr iction interaction are not fully understood, the Molecular Attraction Theory, developed by Tomlinson and Hardy in the 1930s, seems to be the most accepted (Moore, 1975). The cu rrent investigation is limited to friction or skid resistance on wet pavement surfaces only.
PAGE 19
4 1.3 Mechanics of TirePavement Friction Although the m echanisms of tirepavement friction intera ction are not fully understood, it is agreed in the literature that the frictiona l force is composed mainly of adhesion and hysteresis components (NCHRP, 2009). Tire rubber shear is another component that contributes to the frictional force, but its magnitude is negligible when compared to the adhesion and hysteresis force components. So one can express the frictional force as: F = Fadhesion + Fhysteresis (2) If one divides both sides of the Equation (2) by the normal load, the following result is obtained: f = fa + fh (3) Where f is the total coeffi cient of friction, and fa and fh are the components of the coefficient due to adhesion and hysteresis respectively. It can be seen that both fa and fh depend on the viscoelastic properties of the rubber: tan' 1 r ap E sKf (r < 1) (4) and tan' 2 n hE p Kf (n >=1) (5) Where tan is the tangent modulus of the elasto meter, defined as the ratio of energy dissipated to energy stored per cycle, p is the normal pressure, E is the storage modulus or stressstrain ratio for the component of strain in phase with the applied stress, s is the effective shear strength of the sliding interface, r is an exponent with a value of about 0.2 and n is an index greater that the unity.
PAGE 20
5 ''tan E E (6) and '''*jEEE (7) Where E* is the complex modulus (equal to the stressstrain ratio in a viscoelastic body) and E is the loss modulus which is equal to th e stressstrain ratio for the component of strain 90o out of phase with the applied stress. The adhesion component of friction is due to the molecular bonding of exposed surface atoms of both surfaces (tire and pavement), followed by a stretch, break and relaxation cycle. Rubber has an elastomeric structure wh ich is composed of flexible chains which are in constant state of thermal motion. A bond is produced between the separate chains in the surface of the tire and molecules of the pavement during tirepavement interaction. Essentially, the rubber molecu les jump a molecular distan ce to their ne w equilibrium position during the above cycle. On the other hand, hysteresis forces are due to continuous dr aping of rubber over pavement aggregate asperities. The pressure distribution about the asperity depends on two distinct conditions: (1) No sliding (no relative motion) (2) In the presence of relative sliding When there is no relative motion, the drap ing around the contact area and hence the pressure distribution is symmetrical about the asperity giving rise to no net horizontal frictional force. As the sliding begins, rubbe r accumulates in the leading edge of the asperity creating an asymmetrical pressure distribution producing a net friction force (unbalanced force) opposing the motion. At higher sliding speeds, the extent of the contact area decreases and approaches symme trical conditions thus reducing hysteresis.
PAGE 21
6 Figure 1 illustrates both mechanisms of a dhesion and hysteresis; while Figure 2 shows the variation of adhesion and hysteresis with sliding speed. Figure 1 Mechanism of friction at the tire pav ement interface (Adapted from Moore, 1975) Based on the Figures 1, and 2, one can conclu de that hysteresis is relatively independent of sliding speed (operational sp eed), but highly dependent on the pavement macrotexture. On the other hand, adhesion is dependent on both the operational sliding speed and the microtexture. Figure 2 Friction components: Adhesion a nd Hysteresis (Adapted from Moore, 1975) The knowledge acquired by studying the mechanic s of tirepavement interaction will enable one to better understand how the fricti onal measurement devices work, what their operational concepts are and hence wh at they eventually measure.
PAGE 22
71.4 Models for Tire Pavement Interaction The investigators recognize that the tire pave ment interaction is viewed from different perspectives by different indu stries. Customized models de veloped by these industries provide specific information to those industries in areas of their general interest. Consequently there are two main industries that have proposed different model to predict the frictional properties of a pavement su rface, namely: (1) the highway and aviation industries, and (2) the automobile industry. 1.4.1 Models of Highway and Aviation Industries The highway and the aviation industries have cr eated different empiri cal models in which they have tried to simulate the coefficien t of friction using some measurable texture parameters of the pavement as explanat ory variables (NCHRP, 2009). The most popular models are the Penn State University (PSU ) model and the Rado model. These models serve as a basis for the PIARC model (Wambol d et. al., 1995), which ultimately is used in the computation of IFI (Section 1.7). 1.4.2 Models of Automobile Industry The automobile industries have also developed different models that are significant from the vehicle control point of vi ew. Of these, two particular models stand out, namely, (1) The LuGre model and (2) Lumped model (S eneviratne et. al., 2009). These are dynamic models that focus on the properties of the tir e itself rather than those of the pavement. Further details of the above models are found in the cited literature. 1.5 Friction Measuring Devices As mentioned in Section 1, there is a neces sity to evaluate accurately the frictional conditions of a pavement surface in order to prevent accidents and ensure safe highway and aviation operations. Reliable pavement su rface friction information can be obtained from friction measuring vehicl es or by laboratory methods.
PAGE 23
81.5.1 Pavement Friction Measuring Vehicles Different types of vehicles are capable of evaluating the frictional properties of a pavement surface. These devices can be subdi vided into four different groups, depending on their operating mechanism. These mechanisms are: (1) the locked wheel, (2) side force, (3) fixed slip, and (4) variable slip. These vehicle subgroups operate under different principles simulating the relevant scenarios. This fact makes direct comparison between devices inappropriate. The following are the different types of scenarios that these vehicles are designed to simulate: (1) Locked wheel trailer Emergency breaking situation without an AntiLock Braking Systems (ABS). (2) Side force stability in highway curves. (3) Fixed slip and variable slip simulated braking action with AntiLock Braking Systems (ABS). One can observe that each device measures a different coefficient of friction on the same surface making difficult the decision making process about the exact or representative frictional conditions of a pavement surface. Th ere are different friction measuring devices that are approved by the Federal Aviation Administration (FAA). Table 1 lists them and shows the different coefficient of friction th resholds specified by the FAA for different devices. The different threshol d friction levels incorporated in Table 1 for different devices clearly shows that the pavement management co mmunity has acknowledged the incompatibility among different devices. The data in Table 1 suggest that different devices may be correlated by using a linear model. This correlation is more or less ach ieved by the Internationa l Friction Index (IFI). Of the above, the two most commonly used vehicles used in the industry and the ones that would be the subject of this study are presented in the following sections.
PAGE 24
9Table 1 Friction level classification of runway pavement surfaces (Ada pted from FAA, 1997) 1.5.1.1 Locked Wheel Trailer (LWT) The Locked Wheel Trailer (LWT) is an equi pment or device which is the most popular vehicle used by different Departments of Transportation (DOTs) to evaluate pavement condition. It operates under 100% slip conditions, which means that the wheel that is used to measure the coefficient of friction is completely prevented from rolling during testing. It is used to simulate the emerge ncy braking condition without an ABS system. A more detailed operation standard can be found in the ASTM E 274 (ASTM E 27406, 2009). 1.5.1.2 Runway Friction Tester (RFT) The Runway Friction Tester (RFT) is a device that is typically used to evaluate the frictional properties of runways. It operates at approximately 15% of slip, in order to simulate the ABS action on the braking operation of aircrafts. The RFT is an approved continuous friction measuring device for wh ich the threshold values for evaluating runway pavement condition can be seen in Table 1.
PAGE 25
101.5.2 Laboratory Methods Laboratory methods are available as alternatives for evaluating the frictional properties of a pavement surface. The cost of one of these devices is much lower compared to that of field friction measuri ng vehicles. There are tw o commonly devices used in the industry to evaluate surface frictional properties of a pavement in th e laboratory. These are: The Dynamic Friction Tester (DFT) and th e British Pendulum Tester (BPT). 1.5.2.1 Dynamic Friction Tester (DFT) The Dynamic Friction Tester (DFT) can be employed to evaluate the surface frictional properties of a pavement. The measuring mechanism of the DFT is based on energy concepts with the loss of kinetic energy of a rotating disk resting on rubber sliders converted to an equivalent frictional force exerted by the pavement. DFT is capable of measuring friction over the sliding speed ra nge of zero to 90 Km/h. A more detailed operation standard can be found in th e ASTM E 1911 (ASTM E 191109, 2009). ASTM E 1960 advocates the use of DFT for the calibra tion of friction testing devices due to the high repeatability of DFT in IFI computations (Henry et. al., 2000). The DFT is used in conjunction with the Circular Track Meter (CT Meter) to calculate the IFI of a pavement surface. The CT Meter is a device used to evaluate texture properties of a surface, specifically the Mean Profile Depth ( MPD ) which is used to explain the frictionvelocity dependency in the IFI model. A more detailed operation standard on the CT Meter can be found in the ASTM E 2157 (ASTM E 215701, 2009). 1.5.2.2 British Pendulum Tester (BPT) The British Pendulum Tester (BPT) measures the frictional properties of pavement surfaces. The BPT measures friction at a lowsliding speed contact between a standard rubber slider and the pavement surface. Th e elevation to which the pendulum swings after contact provides an indicator of the fr ictional properties of the pavement surface
PAGE 26
11 (NCHRP, 2009). The standard practice for m easuring surface frictional properties using the British Pendulum Tester can be found in the ASTM E 303 (ASTM E 30393, 2008). 1.6 Parameters Affecting TireP avement Friction Interaction In practice, engineers have tried to identify different characteristics of pavement texture that affect the tirepavement interaction. Different parameters have been identified in the literature to have an effect on the tire pave ment friction interact ion. Generally these factors can be grouped into four different categories: (1) Pavement surface characteristics. (2) Vehicle operational parameters. (3) Tire Characteristics. (4) Environmental factors. A detailed study illustrating the effect of th ese pavement surface characteristics on the friction measurement will be covered in the following sections. 1.6.1 Pavement Surface Characteristics Pavement texture is perhaps the most important parameter related to the tire pavement friction interaction. A pavement surface shou ld provide enough skid resistance to stop a vehicle in a panic braking situation. However friction should not be too excessive to produce mechanical wear in th e tire structure. The pavement designer should find an optimum point where it would satisfy both requirements. Seve ral studies performed at the PIARC (Wambold et. al., 1995) es tablished three texture le vels on pavements which describe different effects of frictional perf ormance of a pavement surface. These levels are,
PAGE 27
12 (1) Microtexture: it is a function of the aggr egate asperities. Its magnitude ranges from 1 to 500 m (0.5mm). Microtexture is related to o, which has been correlated to the friction value obtained at ze ro sliding speed. It is also associated with the friction measurements of the Dynamic Friction Tester (DFT) obtained at a sliding speed of 20 km/h (DFT20). The function of microtexture is to provide adhesional friction at th e tire pavement interface under light contamination conditions, where there is still contact be tween the tire and the asperity tips. (2) Macrotexture: it is a function of the arrangement and orientation of aggregates at the pavement surface. Its magnitude range s from 0.5mm to 50 mm. Macrotexture is related to the Speed Constant ( Sp). Specifically Sp has been linearly correlated to MPD, and it can be calcula ted by using the CT Me ter and the numerical correlation expressed in the ASTM E 1960 (Equation (12)). The macrotexture facilitates rapid drainage of water arrest ed in the pavement surface under the tire patch which could lead to hydroplaning conditions. The macrotexture of a pavement surface perform the same function performed by the treads in the tire. (3) Megatexture: it results from pavement surface distress. Generally roughness with amplitude of 50 mm and larger is define d as megatexture (roughness). In the past, megatexture has been related to passenger comfort. One objective of the current study is to quantify the effect of mega texture on the normal load at the tire pavement interface. The magnitude of th e megatexture varies depending on the nature of the profile. On a given pa vement, although the microtexture and macrotexture remain more or less constant, possible changes in the normal load due to variable megatexture would be refl ected in the frictional resistance, which at times could lead to longer braki ng distances. Megatext ure (roughness) is evaluated by the Intern ational Roughness Index ( IRI ), following the specifications established in the ASTM E 1926 (ASTM E 192608, 2009).
PAGE 28
13 Different pavement surfaces may present the sa me frictional properties at a certain sliding speed and yet have different frictional properties at other sliding speeds. Figure 3 illustrates two different surfaces that present the same frictional value at a sliding speed of 30km/h, but different frictionspeed rela tionships due to different macrotexture properties. This illustrates the importance of defining all the text ure parameters when reporting frictional values. Hen ce all of the above parameters must be determined in order to characterize the texture condition of a pavement surface. Figure 3 Texture effect on friction As mentioned in Section 1.3 when one describes the details of the tirepavement interaction mechanism, one has to consider the two friction al components, adhesion and hysteresis. These components are directly re lated to the different texture components of the pavement surface. Adhesion defined by microtexture is responsible for the frictional force at relatively low speeds (adhesion), wh ile hysteresis defined by macrotexture is responsible for the frictional for ce at relatively high speeds. 1.6.1.1 Parameters Used for Texture Characterization There are a number of different parameters us ed to quantify pavement texture. These are: (1) Mean Profile Depth ( MPD) (2) International Roughness Index ( IRI) (3) Root Mean Square ( RMS )
PAGE 29
141.6.1.1.1 Mean Profile Depth (MPD) The MPD is a parameter used to quantify paveme nt macrotexture. The standard practice for calculating the pavement MPD is established in the ASTM E 1845 (ASTM E 184501, 2009). A minimum of ten evenly space segments of 100 mm are needed for every 100 m of the test section for the computation of the MPD Every segment is divided into two equal parts of 50 mm, and the peak value of th e profile is determined for each of the 50 mm sub segments. The average of the two peaks is evaluated to obtain the Mean Segment Depth ( MSD ) of every segment. Finally, the average value of the MSD for all the segments of the measured profile is used to obtain the MPD The MPD is linearly correlated to Sp (Equation (12)), which is one of the parameters used to report the IFI. The MPD quantifies the drainage capabilities of a pavement surface. For instance, a surface with higher MPD value will produce a more stable frictionvelocity relationship (Figure 3(a)) than that w ith a surface with lower MPD in which the friction will decrease rapidly with speed, under wet conditions (Figure 3(b)). 1.6.1.1.2 The International Roughness Index (IRI) The International Roughness Index ( IRI ) was established by the World Bank with the intention of standardizing the longitudinal pavement profile evaluation (Sayers, 1995). IRI is based on a QuarterCar Model, which is a "two degrees of freedom system" used to simulate the suspension system of a vehicle (Figure 4). Figure 4 QuarterCar Model
PAGE 30
15 IRI is a portable quantity, which means that it can be computed independently of the device used to obtain the profile. IRI is defined as the cumulative displacement between the sprung and unsprung mass ove r the length of the profile. IRI is stable with time because it is based on the concept of a true longitudinal profile, rather than the physical properties of a particular vehicle. The standard practice for calculating the IRI is established in the ASTM E 1926 (ASTM E 1192608, 2009). S L usdtzz L IRI01 (8) Where uz and sz are the velocities of the unsprung and sprung masses respectively. L is the length of the profile, and S is the velocity of the vehicle. S is fixed at 80 Km/h for IRI computations. The QuarterCar Model can be used to evaluate the dynamic response of a particular vehicle to a given pavement profile using the appropriate model parameters for that vehicle (Figure 4). Then one would be able to evaluate the roughness effects of that pavement on the normal load at the tire pavement interface. 1.6.1.1.3 Root Mean Square (RMS) The Root Mean Square (RMS), also known as the quadra tic mean, is a statistical parameter used to characterize a pavement profile. N iiY N RMS1 21 1 (9) Where Y(i) is the elevation of the profile at the ith sample point, and N is the sample size.
PAGE 31
161.6.2 Vehicle Operational Parameters 1.6.2.1 Slip Ratio Friction researchers use the slip ratio term to indicate the difference between tire velocity and vehicle velocity, as i ndicated in Equation (10). S wRS Slip (10) Where S is the velocity of the vehicle, w is the angular velo city of the tire and R is the nominal radius of the tire. It is seen from Equation (10) that when the tire is rolling freely the slip must be 0 (S = wR). On the other hand when the tire is locked up the slip ratio is 1 (wR = 0). Locked wheels suffer severe localized wear under dry cond itions since there is no rolling and subsequent uniform wear in the wheels when locked. Thus the material at the contact area between the wheel and the pavement surface is subjected to a frictional force that can lead to permanent deformation localized into one point only of the wheel. On the other hand, a rolling wheel distri butes these effects in a uniform manner throughout the circumference; therefore the wear is considerably lower than that in the locked wheel condition. Experimental work (NCHRP, 2009) shows that the maximum coefficient of friction for most surfaces is generally reached in a range between 0.1 and 0.2 slip ratio, depending on the type of surface, as shown in the Figure 5. Figure 5 Coefficient of friction vs. Slip ratio on different surfaces (Adapt ed from NCHRP, 2009)
PAGE 32
17 This is the principle on which Antilock Brake Systems (ABS) work. An ABS system recognizes that the maximum coefficient of friction is reached at a certain slip range, and hence controls the rotation of th e tires for the slip ratio to be around that slip range. Thus ABS prevents the tires from locking, which provides vehicle stabilit y, steerabi lity and improves stopping capabilities. The ABS is an independent system in that only the wheels that are about to be locked will be pumped and slipcontrolled, while the others will be subjected to the full braking pressure Consequently this system would allow one to stop a vehicle within the s hortest possible distance. A com puter monitors the speed of each wheel, which is fed in to the ABS syst em (Mauer, 1995). When the system detects that one or more tires have locked up or are turning rela tively slower compare to the remaining tires, the computer sends a signal to momentar ily remove and reapply the braking pressure to the affected tire to allow it to continue turning. This "pumping" of the brakes occurs at ten or more times a second, far faster then a human can pump the brakes manually. 1.6.2.2 Vehicle Speed In general, the friction coefficient decreases with speed on wet conditions. This phenomenon is attributed to the facilitation of drainage under the ti re. The higher the speed, the less time the water under the patch ha s to drain off. Pavement macrotexture (MPD) is usually used to explain the fricti onvelocity dependency. High macrotexture improves the drainage properties of the tire patch area, avoiding hydroplaning conditions. Figure 3 shows both the effects of pavement macrotexture and vehicle speed on the coefficient of friction. 1.6.3 Tire Characteristics 1.6.3.1 Tire Tread The tire tread is a major factor when cons idering friction on contaminated pavement surfaces. Tire tread provides a drainage sy stem to evacuate contaminants at the tire
PAGE 33
18 pavement interface; thus having the same function as pavement macrotexture. The use of smooth tires is recommended when performi ng friction tests on a pavement surface, because then information specific to drainage capabilities and texture of the pavement can be obtained. 1.6.3.2 Tire Inflation Pressure Tire inflation pressure is directly related to the tire stiffness. Hence frictional characteristics of a tire are re lated directly to its inflation pressure. Low tire pressures will be reflected in higher rolli ng resistance. Figure 9 shows the results from a loaddeflection test performed by the investigator using the sm ooth tire of the Locked Wheel Tester. It is seen that the tire stiffness increases with increasing inflation pressure. Figure 6 Effect of tire inflation pressure on tire stiffness From Figure 6 one can infer the effects that tire inflation pressure would have on the vertical load at the tire pavement interface. As an example, a stiffer tire is more sensitive to a vertical displacement due to a profile than a softer tire. 1.6.4 Environmental Factors A significant variation is obser ved in friction values measur ed on the same pavement surfaces at different times of the year. Several studies have suggested that this variation can be attributed to different environmental f actors, such as rainfall, dry days preceding
PAGE 34
19 the measurement, temperature, cumulative vehicles passes on test lane; and grease deposits, etc (Jayawickrama et. al., 1998). 1.6.4.1 Pavement Surface Temperature Temperature has a significant effect on the fric tional behavior of the tires, due to the viscoelastic nature of rubber. Friction in rubber like ma terial generally decreases with increasing temperature. Temper ature effect on friction is th e main parameter responsible for seasonal variations of fric tion measurements. Therefore, it is necessary to apply a temperature correction to friction measurem ents in order to perform comparisons between those at different temperatures or di fferent seasons of the year (Fuentes et. al., 2009). 1.7 International Friction Index (IFI) As discussed in Section 1.5 different devices are available in the industry to measure the frictional properties of a pavement surface. As mentioned before, one of the issues surrounding these devices is that when the fric tional characteristics of a pavement surface are evaluated using different fri ction measuring devices one sees significant deviations in their measurements. These friction measuri ng devices operate on different mechanisms and under different physical condition. Thus the observed differences would be obvious. As stated in Section 1.2 the coefficient of fr iction is not a material or a pavement surface property but is a property of the entire measuring sy stem, which includes vehicle characteristics such as, slip ratio and tire pressure and waterfilm thickness. All these parameters change from device to device, so one should not be surprised of the difference on the measurements obtained from di fferent devices on the same surface. The PIARC International experiment (Wambold et. al., 1995) was conducted in Europe with the objective of developi ng a model that would be used to harmonize measurements obtained from different devices into a comm on calibrated index. Forty seven different friction measuring devices participated on the experiment and fifty fo ur different sites, which covered a wide variety of pavement texture characteristics, were used to perform
PAGE 35
20 the experiment. It was found that a simple lin ear regression could be used to correlate measurements from different devices. Cons equently a friction index was proposed and later standardized by ASTM to be device independent, which implied that a certain pavement surface would have a specific inde x independent on the device used to obtain it. This index is the Interna tional Friction Index (I FI) as established in the ASTM E1960. IFI defined in ASTM E 1960 is used currently as the standard for comparison of friction values measured by different equipment. The PIARC International experiment served as the base for harmonizing friction measuremen ts from different equipment to a common calibrated index through IFI. IFI consists of two parameters: (1) Friction Number (F60) and (2) Speed Constant (Sp). It is typically reported as IFI (F60, Sp) and defined by Equations (11) and (12). MPDC S S EXPFRSBAFp* 60 **60 (11) MPDbaSp* (12) Where FRS is the friction measurement obtained from a specific device at slip speed S (slip speed in km/h). MPD is the Mean Profile Depth is defined in Section 1.6.1.1.1. F60 is the prediction of the calibrated Friction Number at 60 km/h and Sp is the prediction of the calibrated Speed Number, which was found in the PIARC experiment to be linearly correlated to the MPD (Wambold et. al., 1995); A, B and C are parameters specific to the friction measuring device while a and b are parameters considered to be specific to the texture measuring device used to measure the MPD A, B, C, a and b are obtained by simple linear regression involving the relevant measured parameters in Equations (11) and (12), with the parameter C being used only when a ribbed tire is used for friction testing. ASTM 1960 stipulates the use of DFT and CT Meter as standard equipment for the calibration of the IFI. Therefore the measurements obtained from these equipment ( DFT20 and MPD ) should be used as dependent variab les when performing the simple linear
PAGE 36
21 regression to obtain the characteristic pa rameters of the different equipment on calibration. 1.8 Scope of Investigation A significant variation has been observed in coefficient of friction values measured on the same pavement surfaces at different times of the year. The objective of the first phase of this investigation is to evaluate the eff ect of the pavement temperature on the observed friction seasonal variations and propose a standardization pro cedure to correlate readings obtained at different times of the year using a given fricti on measuring device. Different friction measuring devices availabl e around the world are capable of evaluating the skid resistance properties of paveme nt surfaces. However, friction measurement values obtained from different devices on th e same pavement surface are different, which makes the direct comparison between frictio n values obtained from different devices inappropriate. The current practice used to co rrelate friction measurem ents from different friction measuring devices is the Internationa l Friction Index (IFI). IFI is a mathematical model used to compare and harmonize frictio nal measurements taken from different equipment to a common calibrated index. The IFI assumes that there exists a linear correlation between measurements obtaine d from different frictional measuring equipment. IFI model uses DFT measurements obtai ned at 20 km/h as the standard friction measurement. However the DFT is a spot tester; which evaluates the frictional properties of a surface taking in to account only a limited range of texture of the pavement (only micro and macro texture). On the other hand, conventional full scale friction testing equipment such as the Locked Wheel Tester, Runway Friction Tester, Grip tester etc. evaluate the frictional characteristics of a pavement over a specific length and hence their measurements are affected invariably by the dynamic effects arising from longwave pavement roughness. The fact that DFT measur ements, used for standardization, are not affected by long wavelength roughness, whic h affects all other full scale friction measuring vehicles, is a definitive issue in the current IFI standard. A second objective of
PAGE 37
22 this investigation is to ev aluate the effects that pave ment roughness have on friction measurements. Furthermore the IFI model assumes that the friction vs. speed dependency found in frictional measurements expres sed by the Speed Constant (Sp) is accounted for by the pavement texture characteris tics only. The parameter Sp dictates the spee d variation of the wet friction measurements taken on a given pa vement surface. Therefore another phase of the investigation would be directed at better understanding the principles underlying the concept of the International Friction Index (IFI) in general, and specifically the role played by the Speed Constant (Sp) in the IFI computations. Th e author believes that once the different factors that have an effect on friction measurem ents are identified, the IFI model can be revised to better correlate friction measurements obtained on the same pavement surface through time and among different friction measuring devices. 1.9 Organization of Dissertation The following chapters represent a synthesi s of research papers which address the respective issues described in Chapter 1. Chapter 2 presents a study of the different factors influencing the frictional measuremen ts obtained using the DFT, while Chapter 3 describes the evaluation of the effect of roughness on skid resistance. Chapter 4 outlines an investigation evaluating the principles underlying the IFI, and a modified methodology proposed to evaluate the IFI parameters. Finally, Chapter 5 details the different conclusions obtained in this investigation.
PAGE 38
23 CHAPTER 2 FACTORS INFLUENCING FRICTIO NAL MEASUREMENT USING THE DYNAMIC FRICTION TESTER (DFT) 2.1 Introduction Skid resistance on runways produced by the tir e pavement friction interaction is an important element in aviation safety. Study of aircraft braking performance on runway surfaces has become so vital that organi zations like NASA and FAA have developed research programs to evaluate the skid resistance available on runway pavement in order to ensure safe landing operations. Aircraft landing accidents continue to occur in many runways around the world. In some of these cas es the aircraft had not been able to brake properly in the landing strip due to the improper understanding of the tirepavement friction mechanism. There are many variables that are involved in this phenomenon. The understanding of each of these variables will help one to realize the magnitude of the problem that one faces and eventually design mo re reliable friction measuring techniques. The International Friction Index (IFI) defined in ASTM E 1960 is used to harmonize friction measurements obtained from different equipment to a common calibrated index. The IFI would provide the means to evaluate the frictional characteri stics of a pavement surface. ASTM E 1960 advocates the use of DFT for the calibration of a given friction testing equipment, as describe d in detail in Section 1.7. The Dynamic Friction Tester (DFT) can be employed to evaluate the surface frictional properties of a pavement. A typical output from DFT is illustrated in Figure 7. In addition to its significance in IFI computations, many researchers have recognized the importance of the DFT in the area of skid resist ance measurement (Wambold et. al., 1995). The above reasons led the author to conduct an i nvestigation of the opera tion of the DFT, in
PAGE 39
24 order to study the factors that could affect its measurements. The primary objective of this phase of the study was to identify the parameters that affect the friction measurements obtained from the DFT. Although this chapter is focused primarily on the temperature effects on the measured coefficien t of friction values, other parameters that have significant impacts on DFT measurements were also identified in this study. These parameters are: (1) Temperature of surface (2) Temperature of water (3) Site variability (4) Water tank height (5) Velocity (6) Rubber slider Figure 7 Typical output from the DFT The author felt that seasonal variations in DFT measurements could mostly be explained by temperature effects, although the former ha s been proposed as an independent factor affecting friction measurements (Bazlamit et. al., 2005, and Jayawickrama et. al., 1998).
PAGE 40
252.2 Experimental Setup The author performed a field i nvestigation within the Univers ity of South Florida campus to select a uniform test site for the experiment. Twelve (12) preliminary DFT measurements were performed at four (4) di fferent sites. Finally the USF Laurel Drive site was selected because of its lowest spa tial variation of friction, with a standard deviation of 0.01 at 20 km/h. This test site has an asphalt pavement that consists of a friction course of FDOT type S3. DFT measurements were ta ken at the site at four (4) different levels of surface temperature and thr ee (3) different levels of water temperature. Thus, twelve (12) different combinations of te mperatures were used for testing. Four (4) DFT measurements were taken for each temperature combination to account for the spatial variation of friction. Surface temperatur es were measured using an infrared sensor and every DFT measurement was performed on a dry spot. The di fferent levels of temperature are seen in Table 2. Table 2 Testing temperature combination Surface Water Temperature oF Temperature Classification Temperature oF Temperature Classification 120 97 78 54 Hot (H) Hot Medium (HM) Medium (M) Cold (C ) 135 75 34 Hot (H) Medium (M) Cold (C ) Effects of contaminants such as dust, sa nd, oil, grease and rubber particles were controlled by an intense clean up of the test site prior test ing using a brush only with no chemicals, to avoid their effect on the fr iction measurement (Jayawickrama et. al., 1998). Any possible effects due to the height of the water level were eliminated by maintaining the water tank at an elevation of 0. 6m above the DFT (ASTM E 191109, 2009). Moreover after each measurement the water tank was refilled back to the same level in order to maintain a constant water pre ssure throughout the experimentation.
PAGE 41
26 Rubber sliders were replaced after twelve r uns, as recommended in ASTM E 1911. It was observed in the present study that wear in th e sliders was affected by the texture of the surface with rougher surfaces producing more wear. The DFT is able to evaluate the coefficien t of friction at different sliding speeds. However, DFT20 is used for IFI computations (Equation (11)) since DFT20 has provided the highest correlation to the IFI model in the PIARC International Experiment (Wambold et. al., 1995). By observation, the author was able to identify two different phenomena that could cause the change of th e waterfilm thickness with speed. The first one is the rate of rotation of the disk, which at high angular speeds would induce a centrifugal force caused by the air movement under the disk, and the other one being the wiping effect produced by rubber sliders. 2.3 Results of Experiments The results obtained from the 48 DFT me asurements under different temperature combinations shown in Table 2 were subjecte d to a statistical study. Different software packages were used to perform the analys is, including Matlab, Excel and R. Figure 8 shows graphically the effect of the water te mperature on the measured coefficient of friction at different speeds. Figure 8Avera g e DFT measurements. Friction Coefficient vs. Water Temp erature in the range of 90 to 100 oF (Hot Medium)
PAGE 42
27 Figure 9Avera g e DFT measurements. Friction Coefficient vs. Water Temp erature in the range of 65 to 80 oF (Medium) Four readings at each temperature combination were considered in the statistical analysis. It is observed in the Figures 8 and 9 that th e coefficient of friction value decreases with increasing water temperature and surface temperature. Therefore it was necessary to perform a 3 parameter analysis to understand the combined effect of both surface and water temperature on coefficient of friction. Th is is illustrated in the 3D plot shown in Figure 4. Figure 10 Combined effect of surface and wate r temperature on the co efficient of friction In order to be able to explain the phenome non of frictional variat ion due to temperature theoretically, one needs to unde rstand the properties of the two types of materials that interact in producing pavement friction. The stiffnesses of both materials in contact, rubber and asphalt, are expected to decrease with increasing temperature (Bazlamit et.al., 2005).
PAGE 43
28 Although the exact mechanisms of tirepave ment friction inter action are not fully understood yet, it is generally agreed in the li terature that the frictional force is composed mainly of adhesion and hysteresis components (Moore, 1975). Shearing of tire rubber is another factor that contributes to the frictional force, although at a negligible magnitude when compared to the adhesion and hystere sis components. Moore, 1975, showed that the coefficient of friction decreases due to an increase of T for any given sliding speed S in the operational speed ranges (Li et. al., 2004). The decrease of coefficient of friction with the increase in water temperature could be attributed to the sensitivity of the hydr odynamic properties of water to temperature (Figure 11). The viscosity of water decreases as temperature increases thereby decreasing the boundary layer shear stress. This is becaus e the shear stress in a Newtonian fluid is equal to the product of the viscosity and the ti me rate of strain (B azlamit et.al., 2005). Figure 11 Change in dynamic visc osity of water with temperature The least square regression method was used to analyze the relation between the two temperatures variables and the DFT20. The friction measurements obtained on the different temperature combinati ons were used to obtain the re gression presented in this paper. All possible regression s using the different potentia l explanatory variables are presented in Table 3. Different statistical parameters such as R2 R2 pred and Cp where used as criteria for the selec tion of the proposed model presented in Equation (13).
PAGE 44
29Table 3 All possible regressions p R2 R2 pred Cp p + 1 Variables 1 0.842 0.827 55.72 2 . . SW 1 0.740 0.714 97.06 2 . W2 1 0.729 0.702 101.53 2 W . 1 0.250 0.175 295.45 2 S . . 1 0.220 0.142 307.33 2 . S2 . 2 0.945 0.933 16.05 3 S . W2 2 0.923 0.906 25.04 3 S W . 2 0.923 0.906 25.06 3 . S2 W2 2 0.900 0.878 34.38 3 W S2 . 2 0.868 0.839 47.39 3 . W2 SW 2 0.858 0.826 51.50 3 W . SW 2 0.843 0.809 57.35 3 S . SW 2 0.842 0.807 57.72 3 . S2 SW 2 0.741 0.683 98.80 3 W W2 2 0.302 0.147 276.24 3 S S2 . 3 0.981 0.974 3.61 4 S S2 W2 3 0.965 0.952 10.18 4 SW S2 . 3 0.953 0.935 15.20 4 S . W2 SW 3 0.946 0.926 17.86 4 S W W2 3 0.931 0.905 23.92 4 S W . SW 3 0.925 0.897 26.39 4 . S2 W2 SW 3 0.924 0.895 26.82 4 W S2 W2 3 0.901 0.864 35.98 4 W S2 SW 3 0.896 0.857 37.94 4 S S2 SW 3 0.878 0.832 45.28 4 W W2 SW 4 0.984 0.975 4.40 5 S S2 W2 SW 4 0.981 0.970 5.61 5 S W S2 W2 4 0.969 0.951 10.53 5 S W S2 SW 4 0.953 0.926 16.95 5 S W W2 SW 4 0.925 0.882 28.38 5 W S2 W2 SW 5 0.985 0.973 6.00 6 S W S2 W2 SW S* 0.00001992 W2 0.00000572 S 0.004404 1.002 2 2 20 TDFT (13) (R2 = 98.1%, R2 (pred) = 97.4%)
PAGE 45
30 Where DFT20T = DFT20 measurement at temperature T W = Water temperature in oF; and S = Surface temperature in oF. R2 pred is determined by systematically removing each observation from the data set and estim ating a regression equation, and finally determining how well the model predicts th e removed observation. It is known that higher R2 pred ensure models of greater predictiv e ability. The summary of the above statistics is shown in Table 4. In order to develop the temperature correction the author used a standard surface temperature of 98 oF and a standard wate r temperature of 70 oF. After substituting the above standard values in Equa tion (13) to obtain standard DFT20 and subtracting the result from Equation (13) one obtains Equation (14). Equation (14) can be used to adjust F60 (Equations (11)) to account for the temp erature effect on DFT measurements. S* 0.00001992 W2 0.00000572 S 0.004404 0.2684 2 2 20 TDFT (14) Thus, DFT20 can be considered as a correction fact or that must be used on the Equation (11) to calculate the IFI parameter. Table 4 Summary of statistics Coefficients: Estimate Std. Error t value Pr(>t) Significance level (Intercept) 1.00E+00 3.62E0227.6463.16E09 0 S 4.40E03 8.87E044.9660.00110.001 S2 1.99E05 5.11E063.8970.004560.001 W2 5.72E06 3.37E0716.9941.46E07 0 Residual standard error: 0.008496 on 8 degrees of freedom Multiple RSquared: 0.9812, Predicted Rsquared: 0.9741 Fstatistic: 139.1 on 3 and 8 DF, pvalue: 3.057e07 Figures 12 and 13 illustrate the values corr esponding to the fitted Equation (13) with respect to the actual measurements.
PAGE 46
31 Figure 12 3D plot of model (Equation (13)) fitted surface and actual measurement of coefficient of friction Figure 13 Plot of fitted data vs. measured DFT20 (Equation (13)) 2.4 Variation of Friction Measurement s due to Environmental Conditions A significant variation has been observed in coefficient of friction values measured on the same pavement surfaces at different times of the year. In the literature, two different phenomena have been reported to have an effect on frictional measurements due to environmental conditions. They are (1) short term variation of fric tion, which has been attributed to different parameters such as rainfall, dry days preceding the measurement, cumulative vehicles passes on test lane; and grease deposits, etc (Jayawickrama et. al., 1998), and (2) seasonal variation (long term effects) which has been attributed to temperature and wear effects (Bazlamit et. al., 2005). Based on the findings reported in
PAGE 47
32 Section 2.3, the author believe s that a temperature correcti on factor should be used to account for the variation of friction measuremen ts due to the seasonal variation. Figure 9 shows the seasonal variation of the skid number measured on an asphalt pavement site in Lucas County, Ohio (Bazlamit et. al., 2005). One can observe a peri odic behavior of the skid number with respect to the time of the year. Figure 14 Typical seasonal variat ion of Skid Number for an asphal t pavement site located in Lucas County, Ohio (Adapted from Bazlamit et. al., 2005). By determining the general air temperature tr ends of Lucas County, Ohio, from relevant sources (Weather Underground, (2007)), the aut hor was able to plot the variation of the skid number with respect to the average ai r temperatures on the dates reported on Figure 14. It is seen from Figure 15 that the skid va ries almost linearly with respect to only the temperature. This has been shown by the author in Figure 9 and can be predicted by Equation (13) formulated by the author. Figure 15 Skid Number vs. Temperature for an asphal t pavement site located in Lucas County, Ohio
PAGE 48
33 CHAPTER 3 EVALUATION OF THE EFFECT OF PAVEMENT ROUGHNESS ON SKIDRESISTANCE 3.1 Significance and Standardization of Pavement Friction Measurements Accidents due to pavement skid resistance deficiencies are a major concern of the aviation and highway industries. Skid resistan ce is developed when vehicle tires are fully or partially prevented from rolling under lubri cated conditions and start to slide along a pavement surface. Hence modeling the mechanis m of skid resistance generation due to the tire pavement friction interaction is a ma jor issue in safe operation of vehicles and aircraft. Lubricated mechanical systems operate in four different regimes of lubrication, namely: static friction, boundary lubricat ion, partial fluid lubrication and full fluid lubrication. Of these, the mechanics of static friction, boundary lubrication and full fluid lubrication are well understood as indicated in the literatur e (Armstrong, 1991). Skid resistance is particularly critical when wate r or other contaminants are pres ent serving as lubricants at the tire pavement interface. Hence, the genera tion of pavement skid resistance must be investigated within the regime of partial fluid lubrication. However, the understanding of partial fluid lubrication is va gue and only empirical studies have been conducted to study this phenomenon. The variation of skid re sistance under partial fluid lubrication is complicated by its slip speed dependency. In the numerous empirical models that have been developed to address this variation, pavement texture pa rameters have been used as independent variables (Henry, 2000).
PAGE 49
34 Since a multitude of devices are available for measurement of pavement skid resistance, there has been an imminent need for standa rdization of skid resistance measurements. The International Friction Index (IFI) is used to harmonize friction measurements obtained from different friction measuring devices to a common cal ibrated index (ASTM E196007, 2009). The IFI concept is based on th e assumption that the friction value of a given surface depends on the slip speed at which measurements are taken, the texture properties of the pavement surface (both micr o and macrotexture) and characteristics of the device used to obtain the measurements. Hence, the ASTM E196007 stipulates the use of Dynamic Friction Tester (DFT) and th e CT Meter as the standard equipment for the calibration of the IFI. In the IFI method, DFT measurement obtained at 20 km/h ( DFT20) is considered the standard skid resistan ce value of a pavement and CT Meter is recommended as the standard instrument for evaluating Sp. DFT20 and Sp can be correlated to microtexture and macrotex ture respectively (A STM E196007, 2009). The IFI concept also assumes the existence of a linear correlation between measurements obtained from different frictional measuring equipment. Hence any given friction measuring device can be calibrated ag ainst the DFT using two parameters; A and B inherent to the given device. The parameters A and B represent the intercept and slope, respectively, of the simple linear regres sion between the friction measurements of a specific device on different pavement su rfaces and the corresponding measurements obtained using the DFT. These parameters coul d be used later for IFI computations using the given device or in the standardization of the r eadings of that device. 3.2 Limitations of the Current Friction Models In the current pavement friction evaluation mode ls, coefficient of friction is defined for a finite area of the pavement accounting onl y for micro and macrotexture. This is exemplified by the use of a spot tester such as DFT as a standard device in the IFI computation. On the other hand, in field evaluations, conventional full scale friction testing equipment are employed to evaluate the frictional characteristics of a pavement over a significant length. Hence, their measurements are affected invariably by the
PAGE 50
35 dynamic effects arising from longwave pa vement roughness or megatexture. More specifically, in full scale friction measuring devices, a significant dynamic normal force is generated due to the mechanical vibrati on of its individual components in response to pavement roughness. The overall magnitude of megatexture (considered to be of amplitudes of 50 mm and larger) depends on the nature of the profile and it is typically evaluated by the International Roughness Index ( IRI ), following the specifications established in the ASTM E 192608 (ASTM E 192608, 2009). Dynamic changes in the normal load on a given pavement with variable megatexture and presumably more or less constant micro and macrotexture, would be ma nifested as reductions in skid resistance, leading to longer braking distance at times. In this regard, a number of studies have attributed skid related accide nts to rough pavements (with high IRI ) (AlMasaeid, 1997, Tighe et. al., 2000, Davies et. al., 2004). Al Masaeid observed that multiplevehicle accidents increase as IRI increases and Davies et. al. also observed that skid related accidents involving multiple vehicles would increase with IRI The current friction models which form the basis for IFI obviously do not incorporate the effect of pavement roughness on friction measurements. In this study, an explanation or mechanism for the reduction in friction is shown to be due to megatexture and the resulting variat ions of normal load at the tire pavement interface and the well documented phenomenon of the reduction in rubber friction due to increased normal loads. Roth et. al. (1942) conducted an investigation on the friction produced by soft rubber compound commonly us ed in tire treads observing that the coefficient of friction decreases as the norma l load and the pressure increase. However Roth et. al. have not proposed a viable mechanism for the observed reduction on the frictional force. Thirion (1946) also studied the influence of normal load in rubber friction being the first investigator to in troduce adhesion as a friction generating mechanism on rubber like materials. Furtherm ore, Thirion observed that the coefficient of friction of rubber decreases hyperbolic ally with increasing normal pressure. Schallamach (1952) presented experimental evidence on the normal load dependency of rubber friction subsequently hypot hesizing that the proportionalit y that exists between the frictional force and the true area of contact would be re sponsible for this phenomenon.
PAGE 51
36 Therefore Schallamach proposed the follo wing empirical equation to quantify the dependence of the coefficient of friction of rubber ( ) on the normal load ( W ). 3 1 cW (15) The constant c has to be determined experimentally for a given velocity and interacting material types. Conventional frictional mode ls do not also account for the normal load dependency of the coefficient of friction. Equation (15) reveals that for rubber the coefficient of friction decreases as the normal load increases. 3.3 Objectives of the Current Study This phase of the investigation is focu sed on evaluating the im pact of pavement megatexture on fullscale friction measur ing devices in particular, through an understanding of the normal load variation ca used by megatexture and the dependency of frictional coefficient on the normal load. The specific objectives of this phase of the study documented in this chapter are; (1) Experimental verification of the effect of pavement roughness on skid resistance. (2) Experimental verification of the reducti on in the coefficient of friction with increased normal load. (3) Formulation of a simplified vibration mode l to interpret and quantify the changes in skid resistance due to pavement roughness. 3.4 Experimental Program A detailed experimental program was pla nned and executed to ach ieve the first two objectives in the Section 3.3.
PAGE 52
373.4.1 Equipment Used in the Study The following stateoftheart equipment were us ed to measure the parameters relevant to this study. 3.4.1.1 Circular Track Meter (CT Meter) The CT Meter (ASTM E2157 01(2005), 2009) is a device used to evaluate the macrotexture properties of a given surface. It specifically measures the Mean Profile Depth (MPD) which is used to express the frictionslip speed dependency in the IFI model (Equation 12). 3.4.1.2 Dynamic Friction Tester (DFT) As described in Section 1.5.2.1 the Dyna mic Friction Tester (DFT) (ASTM E 191109, 2009) can be employed to evaluate the surface frictional properties of a pavement. The measuring mechanism of the DFT is based on energy concepts with the loss of kinetic energy of a rotating disk resti ng on rubber sliders converted to an equivalent frictional force exerted by the pavement. DFT is capa ble of measuring friction over the sliding speed range from zero to 90 km/h. 3.4.1.3 Locked Wheel Tester (LWT) As described in Section 1.5.1.1 the Lo cked Wheel Tester (LWT) (ASTM E 27406, 2009) is the most popular device used by the U.S. Departments of Transportation (DOT) to evaluate skidresistance of highway pa vements. It operates at full (100%) slip conditions, whereby the wheel used to measure the coefficient of friction is completely prevented from rolling during testing. Thus, the LWT is used to simulate an emergency braking condition without an antilock braking system (ABS). The specific LWT used in this investigation was equipped with a profilometer capab le of measuring IRI and MPD of the test wheel path. Smooth tires were used in this study since they allow one to better
PAGE 53
38 evaluate the drainage capability of a pavement surface facilitating the comparison of pavement surfaces for skid resistance. 3.4.2 Selection of Test Sections The first phase of experimentation was conduc ted to evaluate the effect of roughness on friction measurements. The specific goal of this phase was to show that significantly different friction measurements would be obtained using the LWT on two distinct sections possessing the same levels of micr otexture and macrotexture, but different degrees of roughness. In keeping with this objective, two pavement types (A and B) with different levels of roughness (IRI) were selected. Surface A was an asphalt pavement that consists of a friction course of Florida Department of Transportation (FDOT) type S3, while surface B was an asphalt pavement consisting of a fric tion course of FDOT type FC5. The next task was to select two subsections on each section (A and B) so that a significant difference in roughness (megatexture) was apparent between them. In each section A and B, a subsection relatively rougher than the rest of the section was de signated as the test site and another regular subsection in close proximity to the test site was selected as the control site. Then after meticulous visual survey of both pavement sections, two subsections of each section were chosen as the test sites and the control sites. Both the DFT and the CT Meter were used to verify that the micro and macrotexture characteristics on both subsections were similar. Then five repeated friction measurements were obtained at three predetermined speeds on each subs ection using the LWT. After every friction measurement, an air blower (leaf blower) wa s used to remove the excess water from the wheel path of the LWT. The order in which the readings were ta ken was randomized to control all of the noise variables that were not of interest in this st udy, but that could have an effect on the measurements. The randomi zation also served to ensure that the measurements were unbiased with respect to the testing sequence. Friction measurements were reviewed with respect to the order in which they were performed to observe whether any specific trend would be observed as the friction measurements proceeded. No
PAGE 54
39 definitive pattern such as a decreasing trend in the friction measurements at a specific speed was observed as measurements were repeated (Figure 16). Figure 16 Evaluation of the effect of re peated measurements on the Skid Number The second phase of the experiment was to evaluate the effect of the normal load on friction generated at the tire pavement interface. The specific goal of this phase was to illustrate that different friction measuremen ts would be obtained by changing the load configuration of the LWT on the same surface. Four different levels of normal load were configured in the LWT to quantify its effect on friction measurements. Each time the desired variation in normal load was achieve d by adding or removi ng appropriate weights from the LWT trailer. The modified weight configuration was designed so as to prevent any eccentricity at the tire pavement interf ace and maintain the static download force between the truck and the trailer within the range specified by ASTM E 27406. The main aim of changing the static load on th e skidtrailer was to expand the range of variation of the normal load induced by pavement roughness alone a nd provide a picture of the relationship between fr iction and roughness at a much higher resolution. In this phase of the experiment, since friction had to be measured in each test under the designated normal load, testing was limited to relatively smooth surfaces in order to prevent dynamic changes in the normal load caused by pavement roughness from becoming a confounding factor.
PAGE 55
40 To achieve the above conditions, two additi onal pavements surfaces (C and D) were selected. Surface C is an asphalt pavement th at consists of a friction course of FDOT type FC5, while surface D is an asphalt pavement that consists of a friction course of FDOT type FC12.5. These friction course s are the most abundant in Florida's highway network. FC5 is an open graded surface t ype generally used on highways with speed limits higher than 50 mph, and FC12.5 is a dense graded surface type used on highways with speed limits lower than 50 mph. Five friction measurements were performed using the LWT at each of the two predetermined sp eeds for selected design load combinations on each surface. After every friction measuremen t, a leaf blower was used to remove the excess water from the wheel path of the LWT. The order in which these readings were obtained was randomized to control all the extr aneous (noise) variables that could affect the friction measurements. After every fric tion measurement, the LWT was recalibrated for the updated load configuration. 3.5 Results of Texture Analysis Figure 17 present different descriptive statistics of MPD (from CT Meter) and DFT20 (from the DFT) used for the texture comparis on between the control and the test sections of both pavement surfaces A and B. From Figure 17 it is observe d that there is no significant difference in macrotexture (indicated by MPD) and microtexture (indicated by DFT20) between the test and control subsections on both pavements A and B.
PAGE 56
41 Figure 17 BoxPlots of texture co mparison between test and contro l sections of pavements A and B In order to verify that the random variations in MPD and DFT20 within each (A and B) section are not statistically significant, an effects model of the following form was used: 4,3,2,1 2,1j i yiji ij (16) where yij is the jth observation of the ith treatment, is a parameter common to all treatments called the overall mean. In this case would be the overall DFT20 or MPD mean of the pavement surface; i is a parameter unique to the ith treatment called the ith treatment effect, which in this case would be roughness. The null hypothesis would be that i = 0, implying that the effect of roughness on the test a nd control subsections is zero for the measurements obtained using the DFT and the CT Meter. Tables 58 present the analysis of variances (ANOVA) of texture (MPD and DFT20) for both pavements A and B with roughness as the treatment factor.
PAGE 57
42Table 5 Analysis of vari ance of microtexture ( DFT20) on pavement A Source of Variation Degrees of Freedom Sum of Squares Mean Square F0 PValue Roughness 1 0.000008 0.000008 0.0213 0.8887 Residuals 6 0.0022515 0.00037525 Table 6 Analysis of vari ance of microtexture ( DFT20) on pavement B Source of Variation Degrees of Freedom Sum of Squares Mean Square F0 PValue Roughness 1 0.0000061 0.000006125 0.3411 0.5805 Residuals 6 0.000107 0.000017958 Table 7 Analysis of vari ance of macrotexture ( MPD ) on pavement A Source of Variation Degrees of Freedom Sum of Squares Mean Square F0 PValue Roughness 1 0.00005 0.00005 0.0698 0.8005 Residuals 6 0.0043 0.0007167 Table 8 Analysis of vari ance of macrotexture ( MPD ) on pavement B Source of Variation Degrees of Freedom Sum of Squares Mean Square F0 PValue Roughness 1 0.000112 0.000112 0.0184 0.8965 Residuals 6 0.036675 0.006113 From Tables 58, it can be observed that in all four cases the Pvalue is significantly high at a significance level () of 0.1, which leads to ac ceptance of the null hypothesis (Equation (16)) that the difference in roughness is not significant (i = 0) for the DFT20 and MPD measurements within the test and cont rol subsections of pavements A and B. This leads to the conclusion that the test and control si tes only differ in roughness (megatexture) and not in micro and macrotexture properties.
PAGE 58
433.6 Results of Friction Testing 3.6.1 Effect of Roughness on Friction Figures 18 and 19 show the IRI (International Roughness Index), RN (Ride Number), which is another indicator of roughness, MPD and the DFT20 evaluations of the test and control subsections of pavements A and B respectively. It must be noted that the parameters IRI and RN are computed in accordance with the ASTM E 192608 and ASTM E 148908 (ASTM E 148908, 2009) respectiv ely, using the laser profiler of the LWT. Figure 18 Measured roughness characteri stics of subsections of pavement A Figure 19 Measured roughness characteri stics of subsections of pavements B Figure 20 depicts the friction measurements on pavement A and B at subsections of regular (control) and high (test) roughness. It can be observed that on both pavements, at a given speed, the skid number d ecreases as the roughness increases.
PAGE 59
44 Figure 20 Effect of roughness on the Skid Numbers of (a) pavement A and (b) pavement B It is also seen from Figure 20 that the roughness ranges encountered in pavements A and B contribute to Skid Number reductions of 6% and 20% respectively. In addition an ANOVA study was also performed to investig ate the significance of the reduction in SN due to roughness and the interaction betw een speed and roughness. The results of ANOVA are given in Tables 910. It can be concluded from Tables 910 that the roughness variation has a statistically significant effect on frictional measurements, at a 90% level of confidence (Pvalues << (= 0.1)). It also appears that the plots in Figure 20 have para llel trends with respect to speed, indicating that there is no signifi cant interaction betw een roughness and speed. However, one would expect the interaction between speed and roughn ess to affect the friction measurements, since the speed of the moving LWT determines how the roughness of the pavement surface is felt by th e LWT in terms of the rate of change in elevation with time (Gillespie et. al., 1993). In this respect, it can be observed from Table 9 that the interaction variable (speed*roughness) has no significant effect on SN in pavement A, where as Table 10 shows that the Pvalue for the intera ction variable is much lower than (0.1), indicating its significance on SN measurements. One possible explan ation for this anomaly could be that a specific predominant roughness wavelength encountered on pavement B excites one of the natural frequencies of the suspensi on system of the LWT at a specific speed, a condition that does not occur on pavement A.
PAGE 60
45Table 9 Analysis of variance of friction measurements ( SN ) including the interaction variable Speed*Roughness on pavement A Source of Variation Degrees of Freedom Sum of Squares Mean Square F0 PValue Speed 1 2449.43 2449.43 186.2531 8.34E13 Roughness 1 40.3 40.3 3.064 0.09282 Speed*Roughness 1 0.02 0.02 0.0012 0.97237 Residuals 24 315.63 13.15 Table 10 Analysis of variance of friction measurements (SN ) including the int eraction variable Speed*Roughness on pavement B Source of Variation Degrees of Freedom Sum of Squares Mean Square F0 PValue Speed 1 856.41 856.41 408.967 2.20E16 Roughness 1 546.13 546.13 260.797 4.524E15 Speed*Roughness 1 43.88 43.88 20.955 0.0001025 Residuals 26 54.45 2.09 In order to provide support for the above hypothesis the profiles of pavements A and B were plotted in Figure 21 in the frequency dom ain. First a Fast Fourier Transform (FFT) was performed on the profiles to obtain the wave number (wn) spectrum and then the frequency spectrum, f, corresponding to any evaluation speed (S) was determined using Equation (17). Swfn* (17) where f is frequency (cycles/sec), and wn is the pavement wavenumber (cycle/length). Figure 21 Frequency decomposition of the pavement profiles
PAGE 61
46 It can be seen from Figure 21 that the prin cipal frequency components of the profiles of pavement A and B (for the three speeds used) lie in the ranges of 0. 2 0.8 Hz, and 0.5 2 Hz, respectively. Since one of the natural frequencies of the LWT is found to be 1.8 Hz (Section 3.7), the dynamic response of the LWT is magnified on pavement B. This illustrates that the relationship between fr iction and roughness can be confounded by speed effects when testing pavements where one predominant roughness wavelength can interact with the operational speed to pr oduce a resonant condition of the LWT. 3.6.2 Effect of the Normal Load on Friction In the second phase of the experimental prog ram, testing was conducte d to investigate the effects of the normal load on friction meas urements. Figure 22 presents the frictional measurements taken on pavements C and D (Section 3.4.2) by varying the normal load at two distinct speeds. The typical static normal load of the LWT is 1085 lb, as stipulated in the ASTM standards. From Fi gure 22 one can observe that SN follows an inverse S shape, where SN remains constant for low normal loads until a certain limiting normal load is reached at which SN starts to decrease as the load increases, and finally reaches a residual SN. It must be noted that at a fi xed normal load the reduction in SN at higher speeds (Figure 22) is an established fact as seen in Equation (27) (Henry et. al., 1978, ASTM E196007, 2009). Figure 22 Effect of normal load on Skid Numbers on: (a) pavemen t C and (b) pavement D ANOVA was also performed to verify the sign ificance of the dependence of normal load on SN. From the results of this analysis show n in Tables 1112 one can see that the P
PAGE 62
47 value representing the factor, load, is approxima tely zero, leading to the conclusion that the normal load is statistically sign ificant in friction measurements. Table 11 Analysis of variance of friction measurements (SN ) on pavement C based on Speed and Normal Load Source of Variation Degrees of Freedom Sum of Squares Mean Square F0 PValue Speed 1 376.86 376.86 237.714 2.20E16 Load 3 238.72 79.57 50.193 9.17E13 Residuals 35 55.49 1.59 Table 12 Analysis of variance of friction measurements (SN ) on pavement D based on Speed and Normal Load Source of Variation Degrees of Freedom Sum of Squares Mean Square F0 PValue Speed 1 2670.45 2670.45 921.39 2.20E16 Load 3 2077.02 692.34 238.88 2.20E16 Residuals 35 101.44 2.9 However, the ANOVA results shown in Ta bles 1112 do not provide adequate information to distinguish the normal load levels at which the significant differences in the measured average SN values occur. In order to verify that the mean SN values measured under two selected normal loads ar e significantly different from one another, one could carry out multiple independent samp le ttests, each time performing a pairwise comparison of the corresponding SN means. However when one performs n simultaneous independent sample ttest comparisons, one would expect the Type I error ( ) to accumulate n times, thus inflating it to the point that the studies would no longer produce meaningful results. In order to address th is issue, a techniqu e known as Tukey's HSD (Honestly Significant Difference) test (Mon tgomery, 2008) was executed to control the familywise error rates. In this investiga tion, the HSD test was used specifically to determine as to which of the SN means measured at selected load level pairs were significantly different from each other. Tukey's test is a hypothesis test where a pairwise comparison of means is performed in which the overall significance level is exactly (Type error I). Tukeys test turns out to be conservative when compared to results obtained by performing multiple independent sample ttests, making the conclusions
PAGE 63
48 drawn from this analysis more meaningful The results of the Tukeys HSD test are demonstrated in Tables 1314 and Figure 23. Table 13 and 14 show that the Pvalues for co mbinations #2, 3, 4 and 5 are equal to zero, enabling one to reject the null hypothesis that the SN is the same at those load combinations. This observation is further supported by the confidence intervals obtained using the Tukeys HSD test which are pl otted in Figure 23( a) and Figure 23(b). Furthermore, the above results also show that that there is no significant difference between the mean SN measured at load combination #1 in both Tables 13 and 14; and in combination #6 in Table 13. Although th e very low P value corresponding to the combination #6 in Table 14 suggests that there is a difference in SN for that combination, one can observe in Figure 23(b) that th e corresponding confiden ce interval for the difference in means almost contains the valu e of zero. This slight anomaly is pronounced only at the speed of 50 mph as seen in Figure 23(b). Results of the data analysis provided in Tables 1314 and Figure 23 further support the inverse S shape proposed for the SN vs. load variation seen in Figures 22(a) and 22(b) Hence, the following mathematical representation, which captures Sshap e dependencies, is used to model SN with respect to normal load. Table 13 Pairwise comparison among Skid Numbers means at different load configuration using Tukeys HSD Test for pavement C Combination # Load Pair Combination (lb) Difference of SN means between Load levels Lower SN Limit of Confidence Interval Upper SN Limit of Confidence Interval PValue 1 1085900 0.2374102 1.79763 1.32281 0.9763147 2 1400900 4.76 6.278607 3.241393 0 3 1600900 5.2015174 6.685209 3.717826 0 4 14001085 4.5225898 6.08281 2.96237 0 5 16001085 4.9641072 6.490364 3.43785 0 6 16001400 0.4415174 1.925209 1.042174 0.8527597
PAGE 64
49Table 14 Pairwise comparison among Skid Numbers means at different load configuration using Tukeys HSD Test for pavement D Combination # Load Pair Combination (lb) Difference of SN means between Load levels Lower SN Limit of Confidence Interval Upper SN Limit of Confidence Interval PValue 1 1085900 0.8330 2.8863 1.2203 0.6954 2 1400900 13.4076 15.4137 11.4015 0 3 1600900 16.2307 18.3402 14.1211 0 4 14001085 12.5746 14.5807 10.5685 0 5 16001085 15.3977 17.5072 13.2881 0 6 16001400 2.8231 4.8867 0.7594 0.0040 Figure 23 Confidence intervals for all differences in means of all pa ir of load combination on: (a) pavement C and (b) pavement D b wW SNSN SNSNi or oexp1 (18) where SNo is the SN at relatively low loads, SNr is the residual SN at relatively high loads, W is the normal load, b is the width of the inverse S shape curve along the w axis, and wi is the inflection point of the curve, as shown in Figure 24. For a particular LWT, the parameters b and wi can be determined using nonlinear regression or curve fitting techniques. It can be envisione d that Equation (18) would have a wider field applicability to include any vehicle compared to an equivalent laboratory format developed by
PAGE 65
50 Schallamach (1952) in Equation (15). Figure 25 shows the results of fitting Equation (18) for pavements C and D. An interesting observation from Figure 25 is that the values of the parameters wi and b do not seem to depend on either the operational sp eed or the pavements, suggesting that they could be inherent parameters of the measur ing device (LWT). Equation (18) will be used in Section 3.7 to simulate the friction m easurements under different dynamic conditions. The parameters of Equations (1 8) that are listed in Table 15 represent th e two pavements C and D. Figure 24 Nonlinear model representing the effect of Normal Load on Skid Numbers Figure 25 Measured vs. Predicted LoadSN relationship at different speeds
PAGE 66
51Table 15 Parameters characterizing the fric tionload dependency of pavements C and D Surface Speed (mph) b (lb) wi (lb) SNo SNr C 30 50 1300 47.541.5 C 55 50 1300 42.536.4 D 30 50 1300 51.734.5 D 55 50 1300 35 21 3.6.3 Explanation of the Abnormal SN vs. Load Behavior Skid resistance is commonly explained theore tically by the molecular attraction theory (Moore, 1975) that states that the friction ge nerated at the tire pa vement interface is composed of two main components; adhesi on and hysteresis (Equation (3) on Section 1.3). From Section 1.3 it can be seen that both fa and fh depend on the viscoelastic properties of rubber. Also one can see from Equation (4) th at the adhesion component decreases as the normal pressure increases. On the othe r hand, Figure 2 illustrates how the two components contribute to the overall coefficien t of friction as the speed changes. Li et. al., 2003, showed that the adhe sion component dominates the frictional coefficient at general operational speeds (20 60 mph). Hence, one could use the above facts to formulate the following explanation for the SN vs. Load behavior presented in Figures 24 and 25. Although Equations (4) and (5) show opposing trends of f as the normal load changes, based on Figure (2) one could expect the adhesion component to dominate the overall f within normal operational speeds (Li et. al., 2003). Therefore f would decrease with the normal pressure ( p ).
PAGE 67
523.7 Modeling the Dynamic Effects of Pavement Roughness 3.7.1 Model Development and Validation To simulate the dynamic response of the LW T trailer, and specifically assess the variation of the normal load at the tire pa vement interface due to road roughness, a rigidbody vibration dynamic model with twodegrees of freedom wa s formulated for one half of the LWT (Figure 26). For this purpose, six model parameters: two masses Mw and Mt, the two spring stiffnesses Kw and Kt, and the two damping coefficients Cw, and Ct are defined. As seen in Figure 26, Mw (unsprung mass) is the comb ined mass of one half of the vehicle axle, and the si ngle trailer tire, while Mt (sprung mass) is half the total mass of the trailer without Mw. Kt is the stiffness coefficient of the suspension of the trailer as provided by the manufacturer and verifi ed later from laboratory testing, and Kw is the stiffness coefficient of the tire established from tire testing in the laboratory. Ct is the damping coefficient of the tra iler (the shock absorber), and Cw is the damping coefficient of the tire. It must be noted that the a bove model does not include the additional normal force resulting from the back torque produced on the testing wheel due to the frictional force. The two damping coefficients were de termined using experimental modal analysis followed by backcalculation from measured response of the LWT when subjected to a defined profile input. The experimental measurements will be described in the validation procedure. Table 16 defines the values of the above parameters. The proposed model (Figure 26) constrains th e motion in the vertic al direction and the vertical displacements are defined using the variables qt(t) and qw(t) measured from a predetermined baseline. The displacement input to the syst em is provided by the road profile. Equation (19) can be used to transform the spatially defined profile, y = f(x) to a time dependent vertical displacement input y(t) S x t (19)
PAGE 68
53 where x is the longitudinal distance along a given pavement profile, and S is the operational speed of the LWT. By providing th e facility of converti ng the varia tions in the pavement to a forced displacement on a time scale, Equation (19) enables one to investigate the interac tion between speed and ro ughness (Gillespie, 1993). Figure 26 LWT half trailer vibration model Table 16 Dynamic model parameters Parameter Description MagnitudeUnits Mt Sprung mass 440 Kg Mw Unsprung mass 60 Kg Ct Suspension damping 3.5 kNs/m Cw Tire damping 0.5 kNs/m Kt Suspension stiffness 70 kN/m Kw Tire stiffness 265 kN/m The equations of motion can be written in state space variable form (Cauchy form) as ycybzAz (20) whereyis the first derivative of the profile with respect to time and zis the array of state variables (Equation (21)) that defi nes the motion of the system. A, band carrays are defined as follows using the model parameters:
PAGE 69
54 T wwttqqqqz (21) w wt w wt w t w t t t t t t t t tM CC M KK M C M K M C M K M C M K A 1 000 0 010 (22) T w wM K b 00 0 (23) T w wM C c 00 0 (24) RungeKutta method is used to solve Equation (20) for any given input profile expressed in the time domain as y(t). In order to validate the above model and ev aluate its damping parameters, the LWT was instrumented with two accelerometers attached at two specific loca tions of the LWT (as shown in Figure 27) to measure the correspond ing accelerations (motion) of the trailer frame (Mt) and the axle (Mw). Then, a simple test was designed in which LWT was moved rapidly over a pavement bump (Figur e 28) at a speed of 4.2 mph inducing a sudden excitation on it. During this test, the acc eleration records of the trailer frame and the axle that approximate the motion of the two lumped masses of the model (Figure 26) were recorded. Profile measur ements of the separate pavement bump needed for the prediction model in Equation (20) wa s measured using a rod and level.
PAGE 70
55 Figure 27 Instrumentation on the LWT Figure 28 Input profile for the valid ation of the system (pavement bump) Figure 29 presents the frequency spectrum of the acceleration corresponding to Mt. The spectrum reveals the two natural frequencies of the system, which represent the principal modes of motion (degrees of freedom) of th e LWT. These can be calculated by solving the eigenvalue problem of Equation (20). The co mputed natural frequencies of the system are 1.8 Hz and 11.9 Hz, which shows excelle nt agreement between the predicted and measured natural frequencies. Figure 30 displays a comparison between the actually measured and the model predicted vertic al velocities of the two masses or the corresponding components of the system when the LWT traveled over the bump described on Figure 28 at 4.2 m ph. The vertical velocity wa s obtained by integrating the measured acceleration with time. One can obser ve in Figure 29 that the theoretical model captures both the high and the low frequency modes of the response. Accelerometer corresponding to Mt Accelerometer corresponding to Mw
PAGE 71
56 Figure 29 Frequency spectrum of the accelerometer readings corresponding to Mt Figure 30 Measured and predicted velocities of: (a) Mw and (b) Mt 3.7.2 Simulation of Friction Measurements Using the solution of the above validated model (Equations (20)(24)) and Equation (25), one can determine the dynamic normal load indu ced at the tire pavement interface due to roughness as yqCyqKWWww ww static (25) where W and Wstatic are the dynamic load during operation and the load caused by only the masses at static equilibrium at the tire pavement interface. With the statistically established SN vs. W relationship for LWT (Equation (18)) and the model for predicting the dynamic variations of the normal load W along a given profile (Equation (25)) one can simula te the effect of profile roughness on pavement skid resistance. In this process, for a given input pavement profile, one can predict W at any
PAGE 72
57 time using Equation (25) and use Equation (18) to obtain the instantaneously effective SN. The average SN can then be calculated for a suffi ciently long profile length. Figure 31 shows average SN values obtained from the above procedure for the simulated pavements profiles using the friction characteri stics of pavements C and D at two specific speeds, plotted with resp ect to the corresponding IRI. The IRI values for the simulated profiles were calculated in accordance w ith the algorithm provided by the ASTM standards. Figure 31 SN vs. IRI on the simulated profile of (a) pavement C at 30 mph, (b) pavement C at 55 mph, (c) pavement D at 30 mph, and (d) pavement D at 55mph Figure 31 shows a highly scatter plot between SN and IRI. It can be seen that at relatively low values of IRI (0 to 10 m/km), which are usuall y encountered in real pavement profiles, there is no appa rent correlation between SN and IRI. However, at high IRI values, as expected, SN shows a decreasing trend with increasing IRI. The illdefinition of the SN vs. IRI relationship, seen particul arly at usually encountered IRI values, can be attributed to the dependency of the IRI parameter on the natural frequencies of the Quartercar model used in the IRI computation. IRI is a standard parameter that is used to
PAGE 73
58 evaluate the roughness condition of a pavement profile calculated from a dynamic model that simulates the motion of a Quartercar (ASTM E 192608, 2009). In order to verify this explanation, profiles were simulate d over a large range of wavenumbers and amplitudes. Figure 32 shows the IRI obtained from the simulate d profiles plotted against the corresponding amplitude and the frequenc y. The frequency in Figure 32 was obtained using the operational speed of 80 km/h, as stipulated in the ASTM E1926 for IRI computations. Figure 32 3D representation of IRI with respect to frequency and amplitude In Figure 32 one can observe that at two specific frequencies the IRI gets magnified. These frequencies correspond to the natural freq uencies of the Quartercar model used to compute the IRI. Therefore, it is seen that IRI is quite sensitive to the dynamic characteristics of the Quarte rcar model and exhibits a nonmonotonic relationship with respect to roughness, with many conditions of roughness producing the same IRI. Therefore one can conclude that although IRI is used as a standard to evaluate the roughness of pavements it is not an appropriate parameter for expressing the SN vs. roughness relationship. The nonmonotonic nature of the IRI vs. roughness relationship (Figure 32) clearly explains why multiple values of SN correspond to a given IRI in Figure 31. Furthermore, even if one expresses the variation in the SN values measured by LWT with respect to more appr opriate roughness parameters of the pavement profile, e.g. wavelength and amplitude, one could expect a nonmonotonic relationship because the natural frequencies of the LWT itself can produce a resonance effect. To illustrate this,
PAGE 74
59 the vibration model (Equation (18) and (19)(25)) was used to predict the SN values that would be measured by the LWT on si mulated pavement profiles with SN vs. W characteristics similar to that of pavement D, as seen in Table 16. The simulated profiles were obtained by using simple sine waves of constant amplitude of 20 mm and a wide range of wavelengths. Figure 33 shows the predicted variation of SN throughout the frequency spectrum corresponding to the range of input wavelength and the sharp recess in SN corresponding to the predominant LWT frequency of 1.8 Hz. Figure 33 SN vs. Frequency of simulated pavement D with amplitude of 20 mm This study shows that the Dynamic Load Coefficient (DLC) is appropriate for representing the effect of roughness on SN. DLC is used typically to assess the dynamic variation of the normal load at the tire paveme nt interface of a given vehicle for a specific combination of pavement roughness and vehicle speed (Gillespie, 1993). It is defined as Static WW DLC (26) where W is the standard deviation of the normal load (W) variation, obtained from Equation (25) in response to a given pavement profile. Since the variation of SN with roughness is triggered by the dynamic variation of the normal load, SN vs. DLC provides a singlevalue relations hip for a given pavement and speed. Consequently Figure 34 shows SN vs. DLC plots to be of monotonic inverse S shapes as seen before in Figures 24 and 25. Figure 34 was constructed as follows: The
PAGE 75
60DLC for a given pavement with a known profile (y) can be determined from Equations (20)(26). Also, using Equation (18) and th e corresponding parameters found in Table 15, one can estimate SN values measured by the LWT on that pavement. Hence, this study suggests the use of the welldefined curves in Figure 34 to define the effect of roughness and speed on the measured SN. Figure 34 SN vs. DLC on (a) pavement C and (b) pavement D
PAGE 76
61 CHAPTER 4 EVALUATION OF THE SPEED CONSTANT (Sp) AND ITS EFFECT ON THE CALIBRATION OF FRICTI ON MEASURING DEVICES 4.1 Standardization of Friction Measurements Accidents due to skidding on pavements are a major concern of the aviation and highway industries. These accidents are generally attri buted to skid resistance deficiencies on those pavement surfaces. Skid resistance force is developed when a vehicle tire that is fully or partially prevented from rolling sl ides along a pavement surface under lubricated conditions. Skid resistance has been associated with pavement texture, starting with Henry et al., 1978, who used different pavement texture ch aracteristics for the prediction of skid resistance variation w ith slip speed. In the above work, pavement microtexture, which depends on the surface of aggregate aspe rities with its magn itude ranging from 1 to 500 m (0.5 mm), is associated with the skid resistance at low vehicle slip speeds. Furthermore, pavement macrotexture, which depends on the arrangement and orientation of aggregate particles on the pavement surf ace with its magnitude ranging from 0.5 mm to 50 mm, is used to evaluate the skid resistan cespeed relationship by correlating it to the rate at which surface wetting water can es cape from the tire footprint during skid resistance measurements. Penn State Model (NCHRP, 2000) employs the Henry et. al., 1978, concept and introduces a variable called the percent normalized gradient (PNG) to express the skid resistancespeed dependency as; S PNGeSNSSN100 0*) ( (27)
PAGE 77
62 where SN is the Skid Number measured at slip speed S, and SN0 is the Skid Number at zero slip speed. SN0 has been shown to be correlated to pavement microtexture. PNG is the percent normalized gradient which describe s the rate of decrease of skid resistance with the slip speed. Henry et. al., 1978, have shown that PNG is more or less constant on a given surface, and that it is correl ated to the pavement macrotexture. SNo and PNG can be evaluated by performing a simple linear regression of frictiona l measurements on a given surface at different slip speeds. It must be noted th at Locked Wheel Testers (LWT) had been used primarily to gather the frictional data that lead to these original formulations. Since then, several equipment capable of eval uating frictional characteristics of pavement surfaces have been developed around the world. It has been observed frequently that different devices would produce different friction measurements when used on the same pavement surface. Hence, a problem arises when frictional measurements obtained from different devices are compared by personnel from different agencies in runway operational and management decision maki ng. During the Permanent International Association of Road Congress (PIARC) international experime nt held in Europe in 1992 a series of experiments were conducted with one of its objectives being the development of a model that would address the differen ces in frictional measurements among various friction measuring devices. The International Friction Index (IFI) (Secti on 1.7) was developed consequent to the PIARC international experiment, and it is used as the standard to evaluate frictional characteristics of pavement surfaces. IFI is used to harmonize measurements obtained from different frictional measuring devices to a common calibrated friction index. The IFI concept is based on the assumption that the friction value of a given surface depends on the slip speed at which the measurements ar e taken, texture properties of the pavement surface (both micro and macrotexture) and charac teristics of the device used to obtain the measurements.
PAGE 78
63 The PIARC model (Wambold et. al., 1995) introduces the Speed Constant (Sp) to replace the PNG/100 parameter of Equation (27) as follows: pS SeFRFRS *0 (28) where FR0 is the friction at zero slip speed. By rearranging Equation (28) one can obtain Equation (29), to express FR60 in terms of FRS and S, pS SeFRS FR60*)60 ( (29) where FRS is the friction measurement obtained fr om a specific device at slip speed S and FR(60) is the predicted value of friction corresponding to a slip speed of 60 km/h. Sp has been considered as a constant with units of speed that characterizes the drainage properties of a given surface which is correlated to the pavement macrotexture. On the other hand, the IFI of a given pavement consists of two parameters: (1) Friction Number ( F60) and (2) Speed Constant ( Sp). It is typically reported as IFI ( F60, Sp) and defined by Equations (11) and (12). 4.1.1 Investigation of the Va lidity of the IFI Concept Different researchers have reviewed the standardization or harmonization procedure defined by the International Fr iction Index (IFI) concept. Fl intsch et. al., 2009, conducted an investigation using the data collected from the devices used by the consortium members in the 2008 Virginia Smart Road Rodeo on different pavement surfaces. The results obtained from this investigation did not produce harmonious results among the devices used when the original PIARC calibration parameters were used. Flintsch et. al., 2009, were able to improve the agreement between F60 values obtained from their devices by performing a renewed correlation and modifying the A B and C coefficients originally proposed in the PIARC experiment. In addition, Flintsch et. al., 2009, proposed that the discrepancies
PAGE 79
64 seen in the IFI values calculated from differ ent devices used in their investigation could be solved by reevaluating the original coefficients ( A, B, and C ) that were determined in the PIARC international experiment and used in IFI computations. Furthermore Flintsch et. al., 2009, also proposed the implementa tion of equations specific to each friction measuring devices to calculate the Sp but claimed to have obtained relatively weak correlations (coefficients of determination) associating Sp to pavement macrotexture for all the devices used. However Flintsch et al., 2009, obtained improved correlations between Sp and MPD using power models, with the highe st coefficient of determination (R2) being 0.56. National Cooperative Highway Re search Program (NCHRP), 2009, summarizes the guidelines for management of pavement fric tion on existing pavement surfaces and the development of the IFI model. Although S ection 4 of NCHRP, 2009, reaffirms the ASTM E 1960 stipulation that the Sp depends on pavement texture parameters and the texture measuring device only, in its Appendix E it is clearly stated that on the same surface, different friction measuring devices would produce a different set of parameters, F60 and Sp. This reinforces the hypothesis that different friction measuring devices would produce different frictionspeed dependencies on the same pavement surface as defined by their characteristic Sp parameters. Transport Canada Publication No TP 14289E 2003, presents a comprehensive method for modelling the braking performance of airc raft tires on wet pavements. This model acknowledges the importance of the coeffici ent of friction as a system property by incorporating additional parameters such as the vertical load, tire parameters and the waterfilm thickness to predict the frictionspeed relationship. 4.1.2 Assumptions Govern ing the IFI Concept Based on ASTM E 1960 the Speed Constant ( Sp) in Equation (12) depends on the macrotexture characteristics of the paveme nt surface and the method used to determine the macrotexture. In other words, the current International Friction Index (IFI) concept
PAGE 80
65 implies that the frictionspeed relationship that is established based on friction measurements on a given pavement surface at di fferent slip speeds is a characteristic of that pavement only. It must be noted that Sp defines the gradient of the semi log (friction) vs. speed relationship. Hence, for a given pa vement surface any friction measuring device should produce a frictionspeed gradient defined by Equation (12). If Equation (12) truly represents the FRS vs. S gradient for any device on a given pavement surface, then, as illustrated in Figure 35, FR60 predicted from Equations (12) and (29) would be invariant of the slip speed used for its calculation. Figure 35 Relation among FRS S FR60 and Sp The coefficient of friction () is a system modeling property used to express the proportionality between the normal force (FN) and the shear (frictional) force (FS) on two parallel surfaces that are in contact. is known to be dependent on both surfaces that are in contact as well as several other exte rnal factors (Booser, 1989). The predominant factors that affect the interaction include th e normal load, contact pressure, slip speed, and the thickness of the waterfilm when fric tion is evaluated under wet conditions. These factors are generally different from one friction measuring device to another and therefore, one can expect the assumption of the frictionspeed gradient that solely depends on the pavement surface texture to have a limitation when the measuring device changes.
PAGE 81
664.2 Objectives of the Current Investigation The objective of this phase of th e investigation is to evaluate the principles underlying the IFI concept, especially the definition of the Sp parameter. The author also needed to understand the role played by Sp in producing different FR60 values when FR60 is evaluated from friction measurements obtaine d at different slip speeds on the same pavement surface, contrary to the IFI implica tion that it would be independent of the slip speed used for its calculation. Finally the au thor expects to illustrate how the existing Sp evaluation procedure can be modified to produce more consis tent IFI calibration parameters ( A and B ). The data used for the investigation would be obtained from the data acquired during the Wallops Runway Friction workshops conducted in May 2007 and May 2008. 4.3 Data Collection Over more than a decade runway pavement fr ictional data have b een collected on an annual basis at the friction workshops held at the Wallops Flight Faci lity in Virginia, with the purpose of improving the understanding of the concept of skid resistance. An additional aim of the above workshop has been to produce an extensive friction database for harmonization of the friction measurem ents from the different devices used worldwide. At this workshop, frictional and texture measurements have been obtained over 14 different surfaces available at Wall ops using a variety of friction and texture measuring devices. These surfaces have b een constructed covering a large range of microtexture and macrotexture for the purpose of investigating the intricate behavior tire pavement frictional interaction. Wallops Runway Friction Workshop conduc ted in May 2007 included 12 full scale friction measuring devices, namely, 2 E274, SFT, 3 GT, SARSYS, 2 BV11, RFT, Mu METER, and NAC DFT. The plan was for every vehicle to take a total of 10 measurements at two different operational speeds (65 and 95 km/h) on the above
PAGE 82
67 surfaces. Some of the devices we re unable to accomplish this task and hence appropriate adjustments have been made in the analysis. In addition, the Wallops Runway Fricti on Workshop conducted in May 2008 included 15 full scale friction measuring devices, name ly, 3 E274, SFT, 2 GT, SARSYS, 2 BV11, 3 RFT, Mu METER, TWO, and NAC DFT, which we re used to evaluate the same surfaces at 4 different vehi cle speeds (50, 65, 80 and 95 km/h ). At both the above workshops the Circular Track Meter (CT Meter) was used to evaluate the texture ch aracteristics of the tested pavement surfaces and provide the means to compute the Sp of every surface using Equation (12). 4.4 Analysis of Data 4.4.1 Effect of the Slip Speed on FR60 A preliminary study was performed using the 2007 NASA Wallops data to investigate the effect of the slip speed on FR60. At this stage four devices were selected to perform the analysis; Illinois E274, TC SFT 85, FAA RFT and the DND GT. The Sp was calculated on different surfaces (Table 17) according to ASTM standards using the CT Meter and Equation (12), with the recommended a and b parameters of 14.2 and 89.7 respectively. Then the friction measurements taken at th e two different measuring speeds (65 and 95 mph) were converted to FR60 using Equation (29) based on the assumption that the frictionspeed variation would be defined accurately by the Sp parameter. Appropriate slip speeds were used for the FR60 calculations for the devices that operate at slip ratios different from 100%. It was mentioned in Section 4.1.2 that based on the fundamental concepts that were used in formulating the IFI that one would expect the same value of FR60 to be calculated from different slip speeds, when using the Sp parameter obtained from the CT Meter and the a and b values recommended by the ASTM E 1960 procedure. In this regard Figure 36 presents some descriptive statistics of interest on the average FR60 values calculated
PAGE 83
68 for the four devices on pavement surface E at the two measuring speeds (65 and 95 km/h) employed in the Wallops workshop in 2007. Table 17 Texture characteristics of tested pavement surfaces on 2007 SITE MPD Sp (km/h) A 0.433 53.01 B 1.764 172.42 C 2.232 214.44 D 0.512 60.16 E 0.415 51.46 F 1.333 133.77 G 1.929 187.20 ECHO 10.690 76.06 ECHO 20.706 77.50 EK 1 0.176 30.02 EK 2 0.446 54.18 EK 3 0.653 72.74 EK 4 0.340 44.67 R 4 0.956 99.95 Figure 36 FR60 values obtained from different devices on the same pavement surface E
PAGE 84
69 One can observe from Figur e 36 that the values of FR60 do not show a constant trend. In order to prove statistically that the FR60 magnitudes obtained from different slip speeds are not equal, a hypothesis test was performed. A pooled ttest was used to compare the mean of the FR60 values obtained from different slip speeds and a tdistribution was used for statistical inference. The standard de viations of the populations were unknown and hence they were assumed to be different. The test statistic of the tdistribution, Td, for d degrees of freedom can be expressed as 21 60602 2 2 1 2 1nn FR FR TSS S S d (30) where FR60S1 and FR60S2 are the average values of the FR60s obtained from testing at slip speeds S1 and S2 respectively. 2 1 S,2 2 S, and n1, n2 are the corresponding sample standard deviations and the sample sizes respectively. d can be expressed by; 12 2 11 1 212 2 2 2 2 1 2 2 2 2 1 n n n n nn dS S S S (31) Tables 1821 present the statistical summary of this study of FR60 values from the 2007 NASA Wallops database. A pvalue lower than 0.05 would indicate that the null hypothesis ( FR60S1 = FR60S2) should be rejected at 95% confidence level. Since one can observe that the vast majority of such pvalu es are less than 0.05 in Tables 1821, it can be concluded that the null hypothesis defined as the FR60 values computed from different slip speeds are similar can be rejected at a 95% confidence level.
PAGE 85
70Table 18 Comparison of FR60 means obtained from different s lip speeds using the FAA RFT07 Section FR60S1 FR60S2 n1 n2 S1 S2 d Td pvalue A 0.529 0.665 10 10 0.0190.05811 6.995 0.00001 B 0.842 0.913 10 10 0.0290.02618 5.683 0.00001 C 0.808 0.834 10 10 0.0370.02617 1.799 0.04489 D 0.504 0.570 10 10 0.0260.03318 4.970 0.00005 E 0.634 0.700 10 10 0.0410.08214 2.304 0.01854 F 0.827 0.790 10 10 0.0190.04213 2.556 0.01196 G 0.945 0.900 10 10 0.0230.03416 3.446 0.00166 Echo 1 0.626 0.659 10 10 0.0460.05118 1.517 0.07335 EK 1 0.256 0.285 10 10 0.0720.08618 0.823 0.21067 EK 2 0.677 0.852 10 10 0.0390.06216 7.520 0.00000 R 4 0.471 0.494 10 10 0.0440.02615 1.404 0.09038 Echo 2 0.646 0.692 10 10 0.0350.05416 2.247 0.01956 EK 3 0.555 0.557 10 10 0.0500.07017 0.100 0.46088 EK 4 0.510 0.537 10 10 0.0540.04818 1.188 0.12513 Table 19 Comparison of FR60 means obtained from different s lip speeds using the DND GT07 Section FR60S1 FR60S2 n1 n2 S1 S2 d Td pvalue A 0.600 0.733 10 10 0.0350.19410 2.149 0.02859 B 0.704 0.769 10 10 0.0180.08610 2.326 0.02115 C 0.685 0.736 10 10 0.0190.05112 2.913 0.00651 D 0.615 0.724 10 10 0.0290.19110 1.772 0.05343 E 0.594 0.598 10 10 0.0260.32510 0.034 0.48670 F 0.706 0.690 10 10 0.0250.17510 0.296 0.38652 G NA NA NA NA NA NA NA NA NA Echo 1 0.603 0.566 10 10 0.0320.04816 2.043 0.02893 EK 1 0.271 0.364 10 10 0.0300.03518 6.355 0.00000 EK 2 0.667 0.723 10 10 0.0350.06115 2.536 0.01142 R 4 NA NA NA NA NA NA NA NA NA Echo 2 NA NA NA NA NA NA NA NA NA EK 3 NA NA NA NA NA NA NA NA NA EK 4 NA NA NA NA NA NA NA NA NA
PAGE 86
71Table 20 Comparison of FR60 means obtained from different slip speeds using the Illinois E27407 Section FR60S1 FR60S2 n1 n2 S1 S2 d Td pvalue A 0.210 0.303 8 7 0.0300.05510 3.965 0.00133 B 0.521 0.519 10 10 0.0220.01416 0.198 0.42274 C 0.542 0.506 10 10 0.0260.00711 4.261 0.00067 D 0.251 0.363 10 10 0.0330.02818 8.166 0.00000 E 0.204 0.273 10 8 0.0200.0599 3.115 0.00621 F 0.469 0.454 10 8 0.0760.03013 0.597 0.28026 G 0.584 0.518 10 9 0.0320.02317 5.178 0.00004 Echo 1 0.325 0.337 10 7 0.0110.0437 0.701 0.25310 EK 1 0.075 0.082 10 5 0.0080.0185 0.886 0.20815 EK 2 0.247 0.252 10 10 0.0180.01618 0.616 0.27270 R 4 0.236 0.277 10 10 0.0240.01113 4.964 0.00013 Echo 2 NA NA NA NA NA NA NA NA NA EK 3 0.233 0.337 10 10 0.0460.02715 6.122 0.00001 EK 4 0.150 0.103 10 10 0.0420.00910 3.489 0.00292 Table 21 Comparison of FR60 means obtained from different slip speeds using the TC SFT8507 Section FR60S1 FR60S2 n1 n2 S1 S2 d Td pvalue A 0.702 0.912 10 10 0.0150.03413 17.824 0.00000 B 0.892 0.981 10 10 0.0140.01618 13.183 0.00000 C 0.862 0.912 10 10 0.0280.02117 4.455 0.00017 D 0.655 0.824 10 10 0.0220.03416 13.269 0.00000 E 0.771 0.916 10 10 0.0420.07315 5.404 0.00004 F 0.899 0.960 10 10 0.0140.03912 4.679 0.00027 G 0.981 1.031 10 10 0.0070.02411 6.425 0.00002 Echo 1 0.752 0.826 10 10 0.0340.03018 5.133 0.00003 EK 1 0.428 0.525 10 10 0.0550.04418 4.360 0.00019 EK 2 0.849 1.081 10 10 0.0410.06016 10.102 0.00000 R 4 0.650 0.660 10 10 0.0190.02717 0.981 0.17012 Echo 2 0.741 0.803 10 10 0.0330.05316 3.118 0.00331 EK 3 0.733 0.780 10 10 0.0330.03918 2.932 0.00446 EK 4 0.638 0.777 10 10 0.0380.05017 7.037 0.00000 The results presented in Tables 1821 and th e discussion in Section 4.1.2 lead to the conclusion that the Sp values computed for each surface based on Equation (12) and the ASTM E 1960 recommended parameters a and b do not represent the true frictionslip speed dependency of that surface with resp ect to all the fricti on measuring devices.
PAGE 87
724.4.2 Speed Constant ( Sp) and the Significance of the a and b Parameters The conclusions drawn in Section 4.4.1 certai nly illustrates the ne ed for revision of Equation (29) especially with respect to the Sp proposed by ASTM E1960 which stipulates the Sp to be computed using Equation (12). To illustrate this need further, a correlation analysis was performed between the MPD values obtained on the tested pavement surfaces using the CT Meter at the 2008 NASA Wallops workshop, and the Sp parameters that were obtained by fitting the model expressed in Equa tion (32) (linearized form of Equation (28)) to th e experimental friction data obtained by diffe rent friction measuring devices at different slip speed, on each tested pavement surface. The results from this analysis can be seen in Figure 37 and Table 22. S S FR FRSp* 1 ln ln0 (32) From Figure 37 one can observe the weak correlation that exists between the MPD and the experimental Sp values obtained from all the friction measuring devices. Although one can observe an increasing trend of Sp with MPD in Figure 37, the coefficient of determination (R2) which is an indicator of the proporti on of the variabilit y in the data set that is accounted for by the statistical model, is only 0.257. This indicates that most of the variability present in the data is not explai ned by the predictor included in the linear model which in this case is the MPD Figure 37 MPD vs. Sp obtained experimentally for all fric tion measuring devices using the 2008 NASA Wallops data
PAGE 88
73 An additional summary of the statistical analysis performed on the a and b parameters of the linear model in Figure 37 is presented in Table 22. One can see that the 95% confidence intervals of the parameter b do not contain the value recommended in the ASTM standard. It is also noted that the b parameter is the one that represents the rate of change of the Sp parameter with the pavement macr otexture. The above results further reinforce the need realized in Section 4.4.1 to revise the current values ( a =14.2 and b = 89.7) used in the ASTM for the computation of the Sp parameters in Equation (12). Table 22 Statistical analysis of the a and b parameters of the model presented in Figure 37 Coefficient Magnitude Confidence Interval Lower 95% Value Confidence Interval Upper 95% Value ASTM Values a 4.036 16.797 24.871 14.2 b 63.961 41.927 85.995 89.7 4.4.3 Device Dependency of the a and b Parameters From the scatter plot and the low R2 seen in Figure 37 one coul d conclude that there is a weak correlation between MPD and Sp, when the data from all friction measuring devices is used for regression analysis indiscri minately. Consequently, 2008 NASA Wallops runway friction workshop data was used to further investigate the device dependency of a and b parameters. Table 23 presents the textur e conditions of the different pavement surfaces that were included in the above workshop. The MPD values in Table 23 are measured using the CT Meter, and the Sp values were computed using Equation (12), using the ASTM recommended a and b values. When Tables 23 and 17 are compared one can observe that there have been no major changes in the texture conditions on the Wallops pavements surfaces between years 2007 and 2008. Figure 38 presents the friction measurements of three representative friction measuring devices (FAA RFT, VTTI GT and USF 274) obtained during the 2008 Wallops workshop plotted against the corresponding slip speeds. It is noted that alt hough the testing speeds were similar, the slip speeds varies from one device to another since each device operates at its characteristic slip ratio. The line la beled as the standard device in Figure 38
PAGE 89
74 represents the frictionspeed trend that a device should follow in accordance with Equations (28)(29) and (12) using Sp values presented in Table 23. However from Figure 38 one can also observe that, on any given pave ment surface, different friction measuring devices plot frictionspeed rela tionships that deviate signifi cantly from the standard plot defined by the ASTM. On the other hand, one can observe that the only vehicle that seems to follow the standard device is the USF LWT which operates at a slip ratio of 100%, while the other devices (FAA RFT and VTTI GT) that operate at slip ratios in the range of 1020% present a seemingly steeper slope, with friction decreasing faster with slip speed. The reason for the observed deviati on from the standard trend could be that the actual frictionspeed behavi or of a given friction measuri ng device is different than the general trend suggested by Equation (12), with the use of the ASTM recommended parameters a = 14.2, and b = 89.7. It is important to realize that these parameters ( a and b) have been evaluated statistically, based on the results of all the devices used in the PIARC experiment (Wambold et. al., 1995). The data dependency of the Sp vs. MPD relationship was illustrated in Figure 37 in a minor scale, where different values (a and b) were obtained. Hence th e standard values of a and b would not be very much applicable to any particular device. Table 23 Texture characteristics of tested pavement surfaces on 2008 Surface MPD Sp (k/hr) A 0.55 63.535 B 1.753333171.474 C 2.09 201.673 D 0.64 71.608 E1 0.42666752.472 E2 0.46 55.462 F 1.163333118.551 G 1.896667184.331 Echo1 0.82 87.754 EK1 0.19 31.243 EK2 0.52666761.442 R4 0.89333394.332 Echo2 0.74 80.578 EK3 0.67 74.299 EK4 0.37333347.688
PAGE 90
75 0 0.2 0.4 0.6 0.8 1 020406080100 Slip Speed (km/h)Coefficient of Friction FAA RFT VTTI GT USF 274 Standard Device (a) Surface A 0 0.2 0.4 0.6 0.8 1 1.2 020406080100 Slip Speed (km/h)Coefficient of Friction FAA RFT VTTI GT USF 274 Standard Device (b) Surface B 0 0.2 0.4 0.6 0.8 1 020406080100 Slip Speed (km/h)Coefficient of Friction FAA RFT VTTI GT USF 274 Standard Device (c) Surface D 0 0.2 0.4 0.6 0.8 1 020406080100 Slip Speed (km/h)Coefficient of Friction FAA RFT VTTI GT USF 274 Standard Device (d) Surface E2 0 0.2 0.4 0.6 0.8 1 1.2 020406080100 Slip Speed (km/h)Coefficient of Friction FAA RFT VTTI GT USF 274 Standard Device (e) Surface F 0 0.2 0.4 0.6 0.8 1 020406080100 Slip Speed (km/h)Coefficient of Friction FAA RFT VTTI GT USF 274 Standard Device (d) Surface Echo2 0 0.2 0.4 0.6 0.8 1 020406080100 Slip Speed (km/h)Coefficient of Friction FAA RFT VTTI GT USF 274 Standard Device (f) Surface EK3 0 0.2 0.4 0.6 0.8 1 020406080100 Slip Speed (km/h)Coefficient of Friction FAA RFT VTTI GT USF 274 Standard Device (g) Surface EK4 Figure 38 Measured coefficient of friction vs. slip speed relationships of friction measuring devices on test pavement surfaces
PAGE 91
76 The trends seen in Figure 38 and the illustration in Figure 39 could be used to explain the anomaly found in Tables 1821; where it was proven statistically that the FR60 values for a given device would not be the same when co mputed from different slip speeds. As an example, if one were to extract FAA RFT data from Figure 38(d) and calculate the FR60 from two different slip speeds using the Sp calculated from Equation (12) with the standard parameters associated with pa vement macrotexture, one would obtain two completely different numbers (0.47 and 0.32) instead of the actual value of 0.18 as illustrated in Figure 39. Figure 39 FR60s calculated from different slip speeds on surface Echo2 for the 2008 FAA RFT Based on these findings, if one were to correlate the data presented in Figure 37 using carefully selected clusters of data specific to each device, a and b parameters that would be more representative of that device could be obtained. Then these revised parameters could be used in Equation (12) to capture th e actual frictionspeed dependency for that particular device. Figure 40 presents such individual correlations between Sp vs. MPD for different devices used in the 2008 NASA Wa llops Runway Friction workshop. One can see from Table 24 that the coefficients of determination found when associating Sp to MPD for each specific device are significantly higher than that seen in Figure 37, where all the devices were included in the regression analysis. Table 24 presents the summary of the statisti cal analysis obtained from the tests results plotted in Figure 40. 95% confident intervals for the magnitude of the parameters a and b is provided to illustrate their variation from one device to another. Table 24 also includes a summary of the statistical analysis for the devices associated with Figure 40 which were
PAGE 92
77 also used for testing in 2007, to illustrate how the relevant parameters vary over time. In summary, by using the a and b parameters specific for each device in Equation (12) one would be able to better predict the fric tion measurement at 60 km/h based on a single friction measurement at any other slip speed This would certainly address the issue illustrated in Tables 1821 where different FR60 values were obtained when they were calculated from different slip speeds. Table 24 Statistical analysis on the a and b parameters for different friction measuring devices Coefficient a Coefficient b Device Magnitude CI Lower 95% CI Upper 95% Magnitude CI Lower 95% CI Upper 95% R2 VTTI GT 1.265 3.325 5.855 16.116 11.168 21.063 80.76% DND GT08 5.642 19.976 31.261 59.812 29.492 90.133 65.89% DND GT07 5.848 3.409 8.288 20.342 15.843 24.841 80.05% Dynatest RFT 7.506 1.138 13.874 9.555 3.171 15.940 49.66% FAA RFT08 5.919 4.022 7.816 8.470 6.251 10.688 89.23% FAA RFT07 3.680 0.406 7.766 11.752 7.771 15.734 79.33% USF RFT 2.241 5.409 0.927 25.109 21.271 28.947 95.51% USF E274 28.386 5.101 61.873 90.640 58.632 122.64874.22% VDot E27408 29.981 19.442 79.404 110.592 53.304 167.88071.24% VDot E27407 37.071 48.793 122.936126.055 2.270 249.83940.80% PTI E274 31.200 2.130 60.269 75.577 46.394 104.76079.23%
PAGE 93
78 (a) VTTI Grip Tester (b) DND Grip Tester (c) Dynatest Runway Friction Tester (d) FAA Runway Friction Tester (e) USF Runway Friction Tester (f) USF Locked Wheel Tester (g) VDot Locked Wheel Tester (h) PTI Locked Wheel Tester Figure 40 Sp vs. MPD for the different specific devices used in the 2008 NASA workshop
PAGE 94
794.4.4 Slip Speed Sensitivity of the A and B Parameters The final phase of the investigation was to assess the impact of device specific Sp vs. MPD relationship on the device calibration parameters A and B First, in order to evaluate the sensitivity of the parameters A and B to the slip speed, these parameters were calculated from two different sets of meas urements taken at specific slip speeds S1 and S2 for the friction measuring de vices presented in Table 24. A and B parameters listed in Table 25 were determined by first computing the FR60 for ten different surfaces from friction measurements made at the specified slip speeds using Equation (12) with the ASTM suggested a and b (14.2 and 89.7) values relevant to the CT Meter and then linearly correlating th em to the corresponding F60 values obtained using the Dynamic Friction Tester (DFT) (Equation (11)). Since no ribbed tires were used for this analysis C was consider to be 0. By comp aring the values obtained by using FR60S1 and FR60S2 in Table 25, one can observe clearly the effect of the measuring slip speed on the parameters A and B It is evident that the slip speed is a crucial variable in determining the calibration parameters A and B based on the current ASTM standard. The deviation within each parameter ( A and B ) due to the difference in the slip speed can be expressed by Equation (33), were the average A and B values were used to compute the percent deviation. 2 %260 160 260 160 SFR SFR SFR SFRAA AA Deviation (33) where AFR60S1 and AFR60S2 are the parameters A computed from the two different sets of measurements taken at specific slip speeds S1 and S2 respectively. The corresponding form of Equation (33) was used to co mpute the deviation in the parameter B. The effect of the operational slip speed seems to lead to deviations of as much as 164% for parameter A (for the PTI E274) and 30 % for B (for the VDot E27407). Considering all the devices used, the average deviations for parameters A and B computed according to ASTM standards were 37% and 13% respectively.
PAGE 95
80 On the other hand, the modified A and B parameters shown in Table 26 were calculated using the conventional form at, but using the revised a and b parameters specific to every friction measuring device shown in Table 24. It is evident from Table 26 that the modified A and B are less dependent on the slip speed. The deviations associated with the modified values are significantly less with the highest being a 56% for the parameter A (for the PTI E274) and 15% for the parameter B (for the VDot E27407). The revised a and b parameters reduced the average deviation of A and B to 11% and 6% respectively, for all the devices used. Table 25 A and B parameters calculated in accordance with ASTM standards FR60S1 FR60S2 %Deviation Device A B A B A B VTTI GT 0.145 0.727 0.229 0.692 45% 5% DND GT08 0.135 0.771 0.165 0.726 20% 6% DND GT07 0.153 0.726 0.187 0.849 20% 16% Dynatest RFT 0.155 0.658 0.200 0.708 25% 7% FAA RFT08 0.180 0.458 0.217 0.505 18% 10% FAA RFT07 0.178 0.589 0.199 0.677 11% 14% USF RFT 0.166 0.675 0.203 0.699 20% 4% USF E274 0.123 0.680 0.088 0.728 33% 7% VDot E27408 0.132 0.572 0.155 0.466 16% 20% VDot E27407 0.067 0.699 0.099 0.517 39% 30% PTI E274 0.099 0.562 0.010 0.708 164% 23% Table 26 A and B parameters calculated using revised a and b parameters FR60S1 FR60S2 %Deviation Device A B A B A B VTTI GT 0.322 1.878 0.323 1.961 0% 4% DND GT08 0.207 0.784 0.216 0.758 5% 3% DND GT07 0.315 1.539 0.315 1.760 0% 13% Dynatest RFT 0.298 2.878 0.300 2.900 1% 1% FAA RFT08 0.329 2.485 0.329 2.483 0% 0% FAA RFT07 0.309 2.253 0.309 2.387 0% 6% USF RFT 0.325 1.031 0.323 1.096 1% 6% USF E274 0.122 0.677 0.130 0.674 7% 1% VDot E27408 0.119 0.604 0.136 0.565 13% 7% VDot E27407 0.075 0.699 0.112 0.600 41% 15% PTI E274 0.102 0.556 0.057 0.626 56% 12%
PAGE 96
814.4.5 Effect of the Use of a Modified Sp Parameter in IFI Standard Correlation The data presented in Tables 25 and 26 illustrates the im provements that the use of a modified Sp parameter would provide to the stability of the parameters A and B with respect to the measuring slip speed. Howeve r this information does not indicate whether the improved approach predicts better F60 than the current approach. On the other hand, it is clear that the use of a modified Sp parameter would capture the actual friction vs. speed gradient of a given devi ce and therefore predict better FR60 independent on the FRS used on the computations, as illustrated on Figure 39. This fact alone is a significant improvement in the current procedure since it provides a better tool to evaluate the actual FR60 of a given pavement to be used in the decision making process of a pavement management system. Furthermore, in order ev aluate Equation (11) and illustrate which method (ASTM or modified Sp method) produces a better correlation between the two variables F60 and FR60, Figures 4142 and Tables 2728 are presented. One can observe from Figure 41 that even though the parameters A and B are speed invariant when using the modified Sp procedure, the ASTM method produces a better coefficient of determination (R2). One can also observe that the R2 obtained using the ASTM method is significantly sensitive to th e measuring speed, producing lower values at higher measuring speeds (R2 S1> R2 S2 with S1 < S2). The %Errors in Tables 2735 were calculated from the average fitt ed error obtained for the different pavement surfaces used at the calibration stage, where the corresponding A and B parameters were obtained. The corresponding calibration for Dynate st RFT is seen in Figure 41. The above observation is further supported by the results in Table 27 where the same trend occurs in different independent devices used in the analysis. When Tables 27 and Table 28 are compared, one can see that only the devices that ope rate at 100% slip condition (USF E274, VDot E27408 and PTI E274) show a significant improvement when the modified Sp method is used (as illust rated in Figure 42).
PAGE 97
82 The phenomenon observed with the devices that operate at a slip range of 1020% occurs because these devices present a relatively stee p friction speed gradient as illustrated in Figure 38. Given the slip condition of the devi ces that operate at this slip range, the operational slip speed would be significantly low compared to a device that operates at 100% slip condition. When friction values obta ined from any of the devices at any given slip speed ( FRS ) is used to predict the FR60 using Equation (29), with the appropriatly modified a and b parameters, one realizes that th ese values would asymptotically approach zero given the steep friction slip gr adient that these devi ces present (Figure 38). It is realized that FR60 is the friction value predicted for 60 km/h, which is significantly higher than the normal operating slip speeds of these devices. This condition is further illustrated in Figures 41(c) and 41(d) where it can be seen that the FR60 values concentrate in the region close to zero. (a) Using the ASTM procedure on FR60S1 (b) Using the ASTM procedure on FR60S2 (c) Using the modified procedure on FR60S1 (d) Using the modified procedure on FR60S2 Figure 41 Dynatest RFT correlations for Equation 11
PAGE 98
83Table 27 Evaluation of the correlation between FR60 obtained at different slip speeds and F60 values using the ASTM method %Deviation FR60S1 FR60S2 %Error Device A B R2 R2 FR60S1 FR60S2 VTTI GT 44.9% 5.0% 0.835 0.690 8.7% 13.5% DND GT08 19.7% 6.1% 0.833 0.794 8.1% 9.9% Dynatest RFT 25.2% 7.2% 0.866 0.749 10.2% 15.3% FAA RFT08 18.2% 9.8% 0.918 0.888 8.0% 9.7% USF RFT 20.2% 3.6% 0.928 0.932 6.6% 6.8% USF E274 32.7% 6.8% 0.893 0.895 8.7% 9.1% VDot E27408 15.8% 20.4% 0.856 0.598 7.5% 13.6% PTI E274 164.4% 23.0% 0.856 0.835 10.1% 11.5% (a) Using the ASTM procedure on FR60S1 (b) Using the ASTM procedure on FR60S2 (c) Using the modified procedure on FR60S1 (d) Using the modified procedure on FR60S2 Figure 42 PTI 274 correlat ions for Equation 11
PAGE 99
84Table 28 Evaluation of the correlation between FR60 obtained at different slip speeds and F60 values using the modified Sp method %Deviation FR60S1 FR60S2 %Error Device A B R2 R2 FR60S1 FR60S2 VTTI GT 0.1% 4.3% 0.568 0.551 20.6% 21.0% DND GT08 4.5% 3.3% 0.810 0.778 11.0% 12.0% Dynatest RFT 0.9% 0.8% 0.638 0.639 18.9% 18.9% FAA RFT08 0.1% 0.1% 0.616 0.614 15.8% 15.8% USF RFT 0.6% 6.2% 0.611 0.634 18.5% 18.2% USF E274 6.8% 0.5% 0.894 0.893 8.9% 8.6% VDot E27408 13.0% 6.6% 0.884 0.763 6.6% 11.0% PTI E274 56.4% 11.8% 0.862 0.883 9.9% 8.8% In the light of the above information, one could suggest that th e relationship observed between FR60 and F60 is not linear when values obtaine d from devices operating at slip range of 1020% are used. At this poi nt it must be realized that the F60 values used for the standardization of any given friction m easuring device are obtai ned using the DFT, which is also a device that ope rates at 100% slip conditions. Alternatively, one could use a power transformation on the FR60 values to obtain a linear relation between the F60 and the correspond ing transformed FR60 data. Tables 2935 present results of the analys is of implementing different power transformation to the FR60 values. It can be seen from Tables 2935 that the use of power transformation on the FR60 values improves significantly the correlati ons of the devices th at operate at slip ranges of 1020%. The devices that operate at 100% slip conditions did not show any improvement when power transformations were used compared to the results presented in Table 28. One can conclude that this was b ecause there was already a linear relationship between FR60 values of the devices that operate at 100% slip condit ions and the DFT F60, and hence no transformation of the da ta would be necessary for further improvement on these particular devices.
PAGE 100
85Table 29 Evaluation of the correlation between transformed FR60 using the logarithm transformation and F60 values using the modified Sp method at different slip speeds %Deviation FR60S1 FR60S2 %Error Device A B R2 R2 FR60S1 FR60S2 VTTI GT 0.3% 1.0% 0.728 0.718 12.2% 12.1% DND GT08 0.4% 3.8% 0.735 0.732 12.1% 11.5% Dynatest RFT 0.1% 2.0% 0.824 0.781 9.2% 11.9% FAA RFT08 0.1% 0.2% 0.946 0.943 5.0% 5.1% USF RFT 0.7% 4.7% 0.678 0.689 13.9% 13.8% USF E274 0.6% 1.7% 0.877 0.886 8.7% 7.9% VDot E27408 7.6% 22.9% 0.812 0.638 9.3% 14.1% PTI E274 4.8% 21.7% 0.815 0.848 10.0% 10.0% Table 30 Evaluation of the correlation between transformed FR60 using the square root transformation and F60 values using the modified Sp method at different slip speeds %Deviation FR60S1 FR60S2 %Error Device A B R2 R2 FR60S1 FR60S2 VTTI GT 1.4% 0.6% 0.728 0.695 16.0% 16.7% DND GT08 15.6% 3.6% 0.821 0.800 8.1% 8.0% Dynatest RFT 5.1% 3.2% 0.768 0.738 14.3% 15.8% FAA RFT08 0.3% 0.3% 0.808 0.805 12.1% 12.1% USF RFT 0.7% 2.4% 0.816 0.831 12.4% 11.9% USF E274 13.1% 1.5% 0.905 0.907 6.9% 7.0% VDot E27408 69.8% 13.9% 0.852 0.706 7.8% 12.5% PTI E274 60.2% 16.7% 0.846 0.875 9.5% 8.5% Table 31 Evaluation of the correlation between transformed FR60 using the cube root transformation and F60 values using the modified Sp method at different slip speeds %Deviation FR60S1 FR60S2 %Error Device A B R2 R2 FR60S1 FR60S2 VTTI GT 3.2% 2.5% 0.789 0.749 12.0% 13.0% DND GT08 236.3% 3.6% 0.804 0.789 9.3% 8.5% Dynatest RFT 13.4% 3.7% 0.802 0.762 12.1% 14.2% FAA RFT08 0.6% 0.3% 0.881 0.877 9.6% 9.7% USF RFT 0.9% 1.6% 0.899 0.911 8.7% 8.3% USF E274 5.6% 1.6% 0.900 0.904 7.1% 6.6% VDot E27408 36.9% 16.7% 0.839 0.684 8.3% 13.1% PTI E274 37.7% 18.4% 0.837 0.868 9.6% 8.9%
PAGE 101
86Table 32 Evaluation of the correlation between transformed FR60 using the fourth root transformation and F60 values using the modified Sp method at different slip speeds %Deviation FR60S1 FR60S2 %Error Device A B R2 R2 FR60S1 FR60S2 VTTI GT 5.1% 3.1% 0.810 0.770 10.0% 11.2% DND GT08 20.6% 3.6% 0.791 0.779 9.9% 9.2% Dynatest RFT 71.0% 3.7% 0.814 0.771 11.1% 13.6% FAA RFT08 1.3% 0.3% 0.913 0.908 8.3% 8.4% USF RFT 1.3% 1.4% 0.927 0.938 7.1% 6.6% USF E274 4.1% 1.6% 0.896 0.901 7.5% 6.7% VDot E27408 31.1% 18.2% 0.833 0.673 8.5% 13.4% PTI E274 31.9% 19.2% 0.832 0.864 9.7% 9.1% Table 33 Evaluation of the correlation between transformed FR60 using the fifth root transformation and F60 values using the modified Sp method at different slip speeds %Deviation FR60S1 FR60S2 %Error Device A B R2 R2 FR60S1 FR60S2 VTTI GT 7.1% 3.3% 0.815 0.776 9.1% 10.3% DND GT08 12.5% 3.6% 0.782 0.772 10.4% 9.7% Dynatest RFT 35.2% 3.5% 0.819 0.775 10.5% 13.2% FAA RFT08 4.1% 0.3% 0.928 0.923 7.4% 7.6% USF RFT 1.7% 1.4% 0.929 0.939 5.9% 5.6% USF E274 3.4% 1.7% 0.893 0.899 7.8% 6.9% VDot E27408 28.7% 19.1% 0.829 0.666 8.7% 13.5% PTI E274 29.2% 19.7% 0.829 0.861 9.8% 9.3% Table 34 Evaluation of the correlation between transformed FR60 using the sixth root transformation and F60 values using the modified Sp method at different slip speeds %Deviation FR60S1 FR60S2 %Error Device A B R2 R2 FR60S1 FR60S2 VTTI GT 9.7% 3.2% 0.813 0.777 8.5% 9.8% DND GT08 9.7% 3.6% 0.775 0.766 10.7% 10.0% Dynatest RFT 16.1% 3.4% 0.822 0.777 10.2% 12.9% FAA RFT08 5.1% 0.3% 0.936 0.931 6.8% 7.1% USF RFT 2.4% 1.4% 0.919 0.929 7.0% 6.8% USF E274 3.0% 1.7% 0.891 0.897 7.9% 7.1% VDot E27408 27.4% 19.7% 0.826 0.662 7.9% 13.6% PTI E274 27.6% 20.0% 0.827 0.859 9.8% 9.4%
PAGE 102
87Table 35 Evaluation of the correlation between transformed FR60 using the seventh root transformation and F60 values using the modified Sp method at different slip speeds %Deviation FR60S1 FR60S2 %Error Device A B R2 R2 FR60S1 FR60S2 VTTI GT 13.6% 3.1% 0.809 0.775 8.6% 9.5% DND GT08 8.2% 3.7% 0.770 0.762 10.9% 10.2% Dynatest RFT 11.1% 3.3% 0.823 0.778 9.9% 12.7% FAA RFT08 1.6% 0.2% 0.941 0.936 6.4% 6.7% USF RFT 3.2% 1.5% 0.905 0.914 8.0% 7.8% USF E274 2.8% 1.7% 0.889 0.896 8.0% 7.2% VDot E27408 6.8% 7.0% 0.824 0.658 8.0% 13.7% PTI E274 45.8% 33.1% 0.825 0.858 9.8% 9.4% Figure 43 provides a tool to analyze the effect of the power transformation, on the average coefficient of determination (R2) of the correlation between F60 and transformed FR60 values for friction measuring devices that operate in the 1020% slip range. It can be seen from Figure 43 that the fifth root power transformation ( = 1/5) maximizes the R2 value. Therefore, the information provided in Figure 43 would help one to select the optimum power transformation th at would enhance the fitting capabilities of the proposed transformed model. It must be noted that a value of zero corresponds to the logarithmic transformation. However the results of Figure 43 would not pr ovide one with the predictive capabilities of each power model. Figure 43 Power Transformation vs. R2 on the friction measuring devices that operate in the 1020% slip condition range
PAGE 103
884.4.6 Prediction Capabilities of the Proposed Models In order to evaluate which of the proposed power transformation yields the highest predictive capability, F60 values were calculated for different friction measuring equipment on three independent pavement surfaces. The following computations were performed: (1) Calculate FR60 values using Equation (29) using the corresponding modified Sp parameter for the different devices on the respective surfaces. (2) Calculate F60 values using Equation (11) with the appropriate A and B parameters that are characteristic to both the fric tion measuring device and transformation used on the corresponding transformed FR60 values. F60 values were also computed using the trad itional ASTM procedure. The result of this analysis is presented in Tables 364 3, which also contain the standard F60 value obtained using the DFT. Of these, Tables 4143 only compare the general ASTM method with the linear modified Sp method, since it has been shown be fore that a transformation of the data was not necessary for the Locked Wh eel Testers that operate at 100% slip. Table 36 Summary of predicted F60 values for the different method used for the VTTI GT F60 Values Surface DFT ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 0.318 0.270 0.3710.322 0.328 0.336 0.341 0.346 0.349 D 0.322 0.283 0.3890.340 0.350 0.359 0.365 0.369 0.372 E1 0.340 0.235 0.3270.295 0.288 0.290 0.293 0.296 0.299 Table 37 Summary of predicted F60 values for the different meth od used for the DND GT08 F60 Values Surface DFT ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 0.318 0.371 0.3840.365 0.372 0.375 0.377 0.378 0.379 D 0.322 0.368 0.3870.370 0.376 0.379 0.381 0.382 0.383 E1 0.340 0.299 0.3260.297 0.304 0.309 0.312 0.314 0.316
PAGE 104
89Table 38 Summary of predicted F60 values for the different method used for the Dynatest RFT F60 Values Surface DFT ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 0.318 0.280 0.3370.324 0.326 0.328 0.330 0.331 0.332 D NA NA NA NA NA NA NA NA NA E1 0.340 0.264 0.2940.298 0.294 0.293 0.293 0.293 0.293 Table 39 Summary of predicted F60 values for the different meth od used for the FAA RFT08 F60 Values Surface DFT ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 0.318 0.285 0.3510.335 0.336 0.338 0.340 0.341 0.342 D 0.322 0.289 0.3680.346 0.350 0.353 0.355 0.357 0.359 E1 0.340 0.290 0.3200.319 0.314 0.313 0.313 0.313 0.314 Table 40 Summary of predicted F60 values for the different me thod used for the USF RFT F60 Values Surface DFT ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 0.318 0.268 0.3860.321 0.327 0.336 0.344 0.351 0.356 D 0.322 0.283 0.3980.341 0.353 0.364 0.372 0.377 0.381 E1 0.340 0.246 0.3610.295 0.287 0.289 0.296 0.303 0.309 Table 41 Summary of predicted F60 values for the different me thod used for the USF E274 F60 Values Surface DFT ASTMModified A 0.3180.292 0.304 D 0.3220.306 0.312 E1 0.3400.314 0.307 Table 42 Summary of predicted F60 values for the different method used for the VDot E27408 F60 Values Surface DFT ASTM Modified A 0.3180.368 0.355 D 0.3220.353 0.343 E1 0.3400.430 0.406
PAGE 105
90Table 43 Summary of predicted F60 values for the different me thod used for the PTI E274 F60 Values Surface DFT ASTMModified A 0.3180.296 0.303 D 0.3220.285 0.298 E1 0.3400.290 0.288 In order to evaluate the pr edictive capabilities of the proposed methods predicted F60 values were compared to the corresponding standard F60 obtained using the DFT following the stipulations of the ASTM standards. The results of this analysis are presented in Tables 4451. Table 44 Summary of predicted F60 % Errors for the different me thod used for the VTTI GT %Error Surface ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 15.0% 16.8% 1.3% 3.3% 5.5% 7.4% 8.8% 9.9% D 12.0% 20.6% 5.4% 8.7% 11.4% 13.2% 14.5% 15.5% E1 30.9% 3.8% 13.2% 15.2% 14.9% 13.9% 12.9% 12.0% Table 45 Summary of predicted F60 % Errors for the different me thod used for the DND GT08 %Error Surface ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 16.5% 20.7% 14.8% 16.9% 17.9% 18.5% 18.9% 19.2% D 14.3% 20.2% 14.8% 16.7% 17.7% 18.2% 18.6% 18.9% E1 12.1% 4.2% 12.8% 10.5% 9.1% 8.2% 7.6% 7.1% Table 46 Summary of predicted F60 % Errors for the different meth od used for the Dynatest RFT %Error Surface ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 11.9% 6.1% 1.8% 2.6% 3.2% 3.6% 4.0% 4.2% D NA NA NA NA NA NA NA NA E1 22.5% 13.6% 12.3% 13.5% 13.8% 14.0% 14.0% 14.0%
PAGE 106
91Table 47 Summary of predicted F60 % Errors for the different me thod used for the FAA RFT08 %Error Surface ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 10.4% 10.4% 5.3% 5.6% 6.2% 6.8% 7.2% 7.6% D 10.3% 14.3% 7.2% 8.4% 9.5% 10.3% 10.9% 11.3% E1 14.9% 6.0% 6.3% 7.7% 8.0% 8.0% 7.9% 7.7% Table 48 Summary of predicted F60 % Errors for the different me thod used for the USF RFT %Error Surface ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 A 15.6% 21.4% 0.8% 2.8% 5.6% 8.2% 10.3% 12.0% D 12.1% 23.4% 5.8% 9.7% 12.9% 15.3% 17.0% 18.2% E1 27.8% 6.0% 13.4% 15.8% 14.9% 13.0% 11.0% 9.1% Table 49 Summary of predicted F60 % Errors for the different method used for the USF E274 %Error Surface ASTM Modified A 8.1% 4.3% D 5.0% 3.3% E1 7.6% 9.7% Table 50 Summary of predicted F60 % Errors for the different meth od used for the VDot E27408 %Error Surface ASTM Modified A 15.9% 11.6% D 9.4% 6.4% E1 26.3% 19.4% Table 51 Summary of predicted F60 % Errors for the different me thod used for the PTI E274 %Error Surface ASTM Modified A 6.8% 4.8% D 11.7% 7.5% E1 14.8% 15.3%
PAGE 107
92 A summary of the results in Tables 4448 is presented in Table 52, in terms of the average %Errors for different devices that operate in the slip range of 1020%. In addition, Table 53 summarizes the average %Erro rs of the devices that operate under full braking locked wheel (slip = 100%) conditions. Table 52 Summary of the average predicted % Errors of the devices that operate in the range of 1020% slip condition Average %Error Device ASTM = 0 = 1/2 = 1/3 = 1/4 = 1/5 = 1/6 = 1/7 VTTI GT 19.3% 13.7%6.6% 9.1% 10.6% 11.5% 12.1% 12.4% DND GT08 14.3% 15.0%14.1%14.7% 14.9% 15.0% 15.0% 15.1% Dynatest RFT 17.2% 9.9% 7.1% 8.0% 8.5% 8.8% 9.0% 9.1% FAA RFT08 11.9% 10.2%6.3% 7.2% 7.9% 8.3% 8.6% 8.9% USF RFT 18.5% 17.0%6.7% 9.4% 11.1% 12.2% 12.8% 13.1% Overall Average 16.2% 13.2% 8.1% 9.7% 10.6% 11.2% 11.5% 11.7% Table 53 Summary of the average predicted % Erro rs of the devices that operate at 100% slip condition Average %Error Device ASTM Modified USF E274 6.9%5.8% VDot E27408 17.2%12.5% PTI E274 11.1%9.2% Overall Average 11.7% 9.1% Table 52 shows that the power transformation with = produces a consistently lower average %Error for all the friction measuring de vices that operate in the slip range of 1020%. It is also seen in Table 52 that the overall average of the averages %Errors using the conventional ASTM method yield a value of 16. 2% when all the devices are considered, while the corresponding value for the square r oot transformation is 8.1%. This shows that the proposed transformation reduces the predicti on error significantly. On the other hand, Table 53 shows that the modified linear Sp method produces consiste ntly better results for the Locked Wheel testers (slip = 100%) comp ared to the conventional ASTM method.
PAGE 108
934.4.7 Modified International Friction Index Based on the findings of this investigati on, a modified methodology is proposed to compute the International Fr iction Index (IFI). The modified methodology is summarized by the following steps: (1) Select a set of at least te n pavement surfaces that contain a wide variety of both macrotexture and microtexture. The macrotexture and the microtexture are to be evaluated using the MPD parameter and the DFT20. MPD values must be in the range of 0.251.5mm, while DFT20 values are to be in the range of 0.300.90. (2) Measure the coefficient of friction at different speeds on the selected pavement surfaces. (3) Fit the friction data to Equation (32) fo r every pavement surface to obtain the different experimentally obtained Sp values. Determine the calibration constant ( a and b) from a linear regression between the experimentally obtained Sp values and the MPD (4) Using the friction device to be calibrated, determine the friction values ( FRS ) of the test pavements and calculate the FR60 using Equation (29), with the appropriate Sp parameter determined in step (3). (5) Use Equations (34) and (35) to obtain the standard friction parameters of the test pavement surfaces: MPD Sp7.892.14 (34) pS EXP DFT F 40 **732.0081.06020 (35) If the friction device to be calibrated operate s in the slip range of 1020%, then follow step (6a);
PAGE 109
94 (6a) Determine the calibration constant ( A and B ) from a linear re gression using Equation (36), where is equal to 60*60 FRBAF (36) On the other hand, if the friction device to be calibrated operates at 100% slip conditions, then follow step (6b); (6b) Determine the calibration constant ( A and B ) from a linear re gression using Equation (11).
PAGE 110
95 CHAPTER 5 CONCLUSIONS The ASTM standards dictate DFT to be used for calculating IFI (ASTM E196007, 2009). Therefore an experimental investigatio n was performed to study the factors that influence the DFT measurements. The results from the investigation indicated that a number of parameters must be controlled in order to obtain consistent results on the same pavement surface on different trials. The results also indicate a strong dependence among DFT20 and the temperatures of water and the pavement surface. Generally, as the temperature increases DFT20 values decrease. It is also concluded that the temper ature effect on friction is the main parameter responsible for seasonal variations of friction measurements. Therefore, it is necessary to apply a temperature correction to DFT readings in order to perfor m comparisons between measurements taken at different temperatures or different seasons of the year. The authors suggest that a standard temperature be established in order to perform the above correction more systematically. The average temperature of the summer season seems to be the most fitting standard temperature. This is because coefficient of friction values are generally lower during high summer temperatur es and in the absence of a temperature correction the risk of ove rprediction of coefficient of friction exists. It is suggested in ASTM E1911 that the wate r tank must be maintained 0.6 m above the DFT to eliminate possible effects due to the elevation of the water tank height. However, when performing multiple tests at the same location, the water level inside the tank subsides thus changing the water pressure of the system. Therefore, in order to maintain constant water pressure, the water in the tank must be continuously replenished during continued testing.
PAGE 111
96 In the second phase of this research program an experimental and analytical investigation was performed to study the effect of pa vement roughness on friction measurements. Significant experimental evidence was presented in this investigation to illustrate that increased pavement roughness decreases the measured SN (skid number or 100*coefficient of friction). Although the investigation was limited to LWT measurements, this finding could be extended to other vehicles as well as to explain the results of previous investigations attri buting skid related accidents to pavement roughness. Furthermore, the current International Friction Index (IFI) standard for characterizing the frictional properties of a pavement surface does not consider pavement roughness as a required parameter for estimation of IFI. This is because it advocates the use of the Dynamic Friction Tester (DFT), which is a spot tester, as the calibration device. In this research it was concluded that even when the DFT measures the same friction value on different pavement sections with similar te xture conditions, full scale friction measuring device such as the LWT would produce differe nt friction readings if the roughness condition changes from one section to another. In order to account for this phenomenon, an additional parameter accounting for roughness effects could be included in the regression analysis used in IFI computati ons. A parameter appropriate for this purpose can be found by extending this research further. Since the primary factor that c ontributes to the variation of SN with roughness is the dynamics of the normal load, a rigidbody tw odegreeoffreedom system was used successfully to model the variat ion of normal load at the ti re pavement interface of the LWT. The stiffness and damping properties of the model were estimated by direct laboratory measurements and backcalculation from field expe rimental data. The natural frequencies and velocities pred icted by the theoretical model showed excellent agreement with the corresponding experimental m easurements performed subsequently. It was also shown experimenta lly that a quantifiable nonlinea r relationship exists between rubber friction and the normal load. This relationship was combin ed with the above
PAGE 112
97 vibration model to explain the dependency of friction ( SN ) on roughness. In addition, a procedure was also formulated to predict the pavement friction response (measurable SN ) of LWT on any known profile. It can be visu alized that the same general trend of SN vs. roughness would also be exhibited by other vehicles as well. Different parameters that have the potential to define the effects of roughness on friction measurements were studied. It was f ound that the Internat ional Roughness Index ( IRI ) is sensitive to the properties of its inherent Quartercar model, and hence it is not an accurate predictor of the SN vs. roughness relationship. Ho wever, the Dynamic Load Coefficient ( DLC), commonly used to evaluate the e ffect of heavy vehi cles on pavements structures, was found to be an adequate es timator of the roughness effects on measured friction. This preliminary investigation show ed that isolated testing of rough pavement sections with excessively high DLC values produce relatively lower SN values than smoother pavement sections with the same texture characteristics. It is the authors belief that SN values should only be used as a reference for maintenance purposes. A SN value measured at 100% slip does not necessarily represent the actual friction conditions encountered by any given vehicl e that could very well opera te under a different braking mechanism. Therefore the conc lusions of this research cannot be used to associate measured SN values with the potential for skidrelated accidents of automobiles. The findings of this investig ation would provide better ch aracterization of pavement friction in current models by incorporating pa vement roughness. They would also lead to formulation of more meaningful and sa fer Pavement Management System (PMS) decisionmaking criteria with respect to the rehabilitation of rough pavements, which will not only address the serviceability, but also the safety issues. In the final phase of this research a comp rehensive study was performed using the data collected from 2007 and 2008 Wallops Friction Workshops to understand the concept of Speed Constant and investigate its effect on the IFIbased calibration of friction measuring devices. The following were the main findings of this study; (1) different FR60 values were calculated from different slip speeds on the same pavement surface
PAGE 113
98 using the Sp vs. MPD relationship recommended in the ASTM. (2) a and b parameters specific to both the friction measuring device and the pavement w ould capture the true friction speed dependency, enabling one to predict more accurate FR60 values for a given surface. The modified Sp parameter captures the real fric tion speed gradient of a given device, hence predicting better FR60 independent on the FRS used on the computations. This fact alone is a significant improvement in the current procedure since it provides a better tool to eval uate the actual FR60 of a given pavement to be used in the pavement management decisionmaking process. By not using the a and b parameters specific for the friction measuring device one could mischar acterize the frictional characteristics of a pavement surface and hence disseminate misleading information for the pavement management decisionmaking process. (3) A and B device calibration parameters which were calculated in accordance with ASTM standa rds were sensitive to the slip speed used in their calculation. Different A and B parameters were obtained when friction measurements at different slip speeds were used for its calculation. Furthermore by using device dependent Sp parameters the slip speed dependency of the parameters A and B can be reduced significantly. This investigation also indicated that there is a li near correlation between F60 obtained from devices that operate at 100% slip cond itions and the standard friction measuring device DTF, which also operates under the same slip conditions. On the other hand the relationship between F60 values obtained from the standard device, DFT, and the F60 values obtained using friction measuring devices that operate in the slip range of 1020% was found to be nonlinear. However, it was found that the use of a power transformation on the F60 values obtained using friction measuring devices that operate in the slip range of 1020% would produce a linear relationship between the standard F60 values obtained using the DFT and the co rresponding transformed F60 values. The revised A and B parameters corresponding to the different devi ces used in this investigation were then calculated using the proposed modifications. The above values presented were less sensitive to the speed at which the tests were performed, and also possessed a much higher coefficient of determination than th e corresponding parameters obtained using the conventional ASTM method.
PAGE 114
99 Based on the results of the inve stigation the authors felt that the current procedure used by ASTM for calculation of the Internationa l Friction Index (IFI) could be modified slightly, particularly with respect to (1) th e procedure used for the calculation of the Speed Constant and (2) the met hodology used for correlating the FR60 values to the standard DFT F60 measurements. Although it has been well documented that a direct relationship between the fricti onal speed dependency and macrotexture exists, the results of this study indicate that the a and b parameters used in the ASTM procedure should be specific to the device used for testing. This is necessary to compensate for the effects of the specific frictional measuring device, since it has been observed that different frictional measuring devices generate their own characteristic frictional speed trends on the same pavement surface. Therefore, the authors suggest these parameters be calculated by correlating the experimental Sp parameters obtained from data clusters corresponding to specific friction measuring devices to the MPD of different pavement surfaces covering a wide variety of macrotexture. Finally, a modified IFI procedure that incorporates device specific slip conditions into its co mputations is proposed. The modified IFI procedure consistently produc ed more accurate predictions than the conventional ASTM procedure on all the diffe rent devices considered in this study.
PAGE 115
100 REFERENCES AlMasaeid, H. R. (1997) Impact of Pave ment Condition on Rural Road Accidents. Canadian Journal of Civil Engineering, Vol. 24, No. 4, 1997, pp. 523532. ArmstrongHlouvry, B (1991) Control of Machines with Fric tion, The Springer International Series in Engineeri ng and Computer Science, Vol. 128. ASTM: Standard Test Method for Skid Resi stance of Paved Surfaces Using a FullScale Tire, Standard No E27406, ASTM 2009. ASTM: Standard Test Method for Measuri ng Surface Frictional Properties Using the British Pendulum Tester, Sta ndard No E30393, ASTM 2008 ASTM: Standard Practice for Computing Ri de Number of Roads from Longitudinal Profile Measurements Made by an Inertial Profile Measuring Devi ce, Standard No E1489 08, ASTM 2009 ASTM: Standard Practice for Calculating Pave ment Macrotexture Mean Profile Depth, Standard No E184501(2005)e1, ASTM 2009 ASTM: Standard Test Method for Measuring Paved Surface Fricti onal Properties Using the Dynamic Friction Tester, Standard No E191109, ASTM 2009 ASTM: Standard Practice for Computing In ternational Roughness I ndex of Roads from Longitudinal Profile Measurements, Standard No E192608, ASTM 2009. ASTM: Standard Practice for Calculating In ternational Friction Index of a Pavement Surface, Standard No E196007, ASTM 2009 ASTM: Standard Practice for Measuring Pavement Macrotexture Using the Circular Track Meter, Standard No E 215701 (Reapproved 2005), ASTM 2009 Bazlamit, S. Reza, F. (2005), Changes in Asphalt Pavement Friction Components and Adjustment of Skid Number for Temperature. Journal of Transportation Engineering, Volume 131, Issue 6, pp. 470476. Booser, E. R. (1989).CRC Handbook of Lubrica tion: Theory and Practice of Tribology Volume II. CRC Press, Boca Raton, Fl, pp 3947.
PAGE 116
101 Cenek P.D. and Davies R.B. (2004) Crash Risk Relationships for Improved Safety Management of Roads, Proc Towards Su stainable Land Transport, Wellington, New Zealand. Denny, D. F., (1953) The Influence of Load and Surface Roughness on the Friction of RubberLike Materials, Proc. Phys Soc., Series B, 66, 721727. ESDU International plc, (2003). Development of a Comprehensive Method for Modelling Performance of Aircraft Tire s Rolling or Braking on Dry and PrecipitationContaminated Runways TP 14289E. Transportation Deve lopment Centre: Transport Canada. Federal Aviation Administration (1997). Mea surement, Construction, and Maintenance of SkidResistant Airport Pavement Surfaces, U.S. Department of Transportation, AC No: 150/532012C. Flintsch, G. W; de Leon Izeppi, E. D; McGh ee, K. K and Roa, J. A. Evaluation of the International Friction Index Coefficients fo r Various Devices. Transportation Research Board Annual Meeting 2009 Paper #093240, Washington, D.C. Fuentes, L. Gunaratne, M. (2009) Factors Influencing Frictional Measurement Using the Dynamic Friction Tester (DFT), Presented at 88th Meeting of the Transportation Research Board .Paper #090100. Washington, D.C. Gillespie, T D; Karamihas, S M; Sayers, M W; Nasim, M A; Hansen, W ; Ehsan, N ; Cebon, D. (1993) Effects of HeavyVehicle Characteristics on Pavement Response and Performance, NCHRP Report 353, Transpor tation Research Board, Washington DC. Henry, J.J., Hironari Abe, Shulchi Kameya ma, Akinori Tamai, Atsushi Kasahara, Kazuo Saito (2000) Determination of the International Friction Inde x (IFI) Using the Circular Texture Meter (CTM) and the Dynamic Friction Tester (DFT). PIRAC 109. Henry, J. J., (2000) Evaluation of Pavement Friction Characteristics. NCHRP synthesis 291. Transportation Research Bo ard. Washington, D.C. 66pp. Henry, J. J., Leu, M C. ( 1978). Prediction of Skid Resi stance as a Function of Speed from Pavement Texture Measurements. In Transportation Research Record: Journal of the Transportation Research Board, No. 666, Transportation Research Board of the National Academies, Washington, D.C., pp. 7. Jayawickrama, P.W., and Thomas, B., (1998) Correction of Field Skid Measurements for Seasonal Variations in Texas, Transporta tion Research Record 1639. Transportation Research Board, National Research Council, Washington, D.C., pp. 147154. Li, S., Noureldin, S., and Zhu, K. (2004) Upgrading the INDOT Pavement Friction Testing Program. Joint Transportation Res earch Program, Perdue Libraries: 69.
PAGE 117
102 Mauer, G. F. (1995). "A Fuzzy logic contro ller for an ABS braking system", IEEE Transaction on Fuzzy Systems, 3, 4, 381388. Montgomery, D.C, (2008). Design and Analys is of Experiments, 7th. New York: John Wiley & Sons Inc. Print. Moore, D. F, (1975). The Friction of Pneuma tic Tyres New York: Elsevier Scientific Pub. Co. NCHRP, (2009) Guide for Pavement Friction Project No. 0143. Tran sportation Research Board. National Research Council. February. Roth, F.L., Driscoll, R.L., and Holt, W.L., (1 942) Frictional Prope rties of Rubber, J. Res. Nat. Bur. Stds. 28, 439. Sayers, M. (1995) On the Calculation of International Roughness Index from Longitudinal Road Profile, Transportati on Research Record 1501, Transportation Research Board Business Office, Washington, D.C., pp. 112. Schallamach, A. (1952) The Load Dependenc e of Rubber Friction Proc. Phys. Soc., Section B, Volume 65, Issue 9, pp. 657661. Seneviratne, H. N. Rajapakshe, M.P. Gunara tne, M. (2009). Fie ld Calibration of an Analytical Model for Pavement Friction Tes ting Applications, Journal of Testing and Evaluation, 37(1), pp. 110. Thirion, P., (1946) Les coefficients d' adherence du caoutchouc., Revue Generale du. Caoutchouc 23(5), 1016. Tighe, S., Li, N.Y., Falls, L.C., & Haas, R. (2000) Incorporating Road Safety into Pavement Management. Trans portation Research Record: Jour nal of the Transportation Research Board (1699), 110, Washington, D.C. Wambold, J. C., C. E. Antle, J. J. Henr y, and Z. Rado (1995) International PIARC Experiment to Compare and Harmonize Text ure and Skid Resistance Measurements, Final Report submitted to the Permanent International Association of Road Congresses (PIARC), State College, PA. Weather Underground, (2007) "T oledo, OH Typical Weather.,. Website. . Last Accessed on 20 November 2007.
PAGE 118
103 Weather Underground, (2007) "History for Toledo Metcalf, OH.", Website. . Last Accessed on 20 November 2007.
PAGE 119
ABOUT T HE AUTHOR Luis Fuentes received a Bachelors Degree in Civil Engineering from Universidad del Norte, Colombia, in 2005 and a Master in Civil Engineering (MCE) from University of South Florida in 2006. He entered the Ph.D. progr am at the University of South Florida in 2006, where he worked on a project sponsored by the National Aeronautics and Space Administration (NASA) to investigate the skid resistance phenom enon. He has also coauthored publications on differe nt journals such as the AS CE Journal of Transportation Engineering and the TRR Journal of the Tran sportation Research Board; and made paper presentations at the annual meetings at the Transportation Research Board (TRB). 104
