USFDC Home  USF Electronic Theses and Dissertations   RSS 
Material Information
Subjects
Notes
Record Information

Full Text 
PAGE 1 1 INVESTIGATION OF THE VALIDITY OF THE ASTM STANDARD FOR COMPUTATION OF INTERNATIONAL FRICTION INDEX by Kranthi Kavuri A thesis submitted in partial fulfillment of the requirements for the degree of Master of Civil Engineering Department of Civil and Environmental Engineering College of Engineering University of South Florida Major Professor: Manjriker Gunaratne, Ph.D. Jian Lu, Ph.D. Abdul Pinjari, Ph.D. Yu Zhang, Ph.D. Date of Approval: November 6, 2008 Keywords: Runway friction testing, Calibrati on constants, Random variable, Slip speed, Speed constant. Copyright 2008 Kranthi Kavuri PAGE 2 2 DEDICATION I dedicate this work to my mother Kok ila, my grandmother Subbamma, my brother Srikanth and my friend Issaac Komminen i. Without their constant support and encouragement I would never have achieved th is. To my friends Meeta, Ramya, Qing, and Vishwanthi for their consta nt support, help and friendship. PAGE 3 3 ACKNOWLEDGEMENTS I am immensely grateful to my advisor Dr. Manjriker Gunaratne for providing me this opportunity to work on this project. Your guidance, patience and support are greatly appreciated. I would like to thank Dr. John Lu for his guidance and support throughout my Masters. I would also thank my comm ittee members Dr. Yu Zhang and Dr. Abdul R.Pinjari for their valuable s upport. I would like to extend my special thanks to Dr. Yang for his help and guidance in performing statistical tests. I am grateful to NASA Wallops funded pr oject for providing financial support. I would like to thank Mr. Issaac Kmonineni for his help and guidance in writing the computer program. I would like to thank my friends Louis, Madhur a, Sowmya and Janakre for their help and support. I would also like to thank my friends Issaac, Meeta and Qing who has been working with me and has been a great help to me throughout my Masters. PAGE 4 i TABLE OF CONTENTS LIST OF TABLES .iii LIST OF FIGURES vi ABSTRACT viii CHAPTER ONE INTRODUCTION .1 1.1. NASA Wallops Runway Friction Workshop 2 1.2. Objectives of NASA Wall ops Tire/Friction Workshop ...........3 1.3. International Friction Index ..3 1.4. ASTM Designation E 1960: Standard Prac tice for Calculating the International Friction Index (IFI) ...4 1.4.1. Summary of Practice ..4 1.4.2. Significance and Use of AS TM Standard Methodology ...5 1.5. Objectives of Proposed Research . .6 1.6. Thesis Organization ..7 CHAPTER TWO ANALYSIS OF NAS A RUNWAY FRICTION WORKSHOP DATA........................................................................................................................... .......8 2.1. Calibration of the NASA GT 8 2.1.1. Determination of the Effect of Operating Speed on the Calibration .12 2.2. Calibration for Averaged Operating Speeds ... ...13 PAGE 5 ii CHAPTER THREE VALIDATION FOR THE CURRENT METHOD OF DATA ANALYSIS 16 3.1. Validation for Individual Operating Speeds ..16 CHAPTER FOURDATA ANALYSIS WITH THE MODIFIED METHOD .21 4.1. Calibration with Modified Method ... .21 4.1.1. Procedure for Randomization of the Calibration Constants . .22 4.2. Validation for the Proposed Me thod of Data Analysis ..26 4.2.1. Speed Sensitivity 36 4.2.2. Raw Data Verification ...37 4.2.3. Check for Effectiveness of Sp and MPD Linear Relationship ..38 4.2.4. Computation of Sp a nd MPD Relationship ...39 4.2.5. Alternative Calculation for more A ccurate Sp and MPD Relationship 39 CHAPTER FIVECONCLUSIONS .43 REFERENCES ..45 APPENDICES . ..46 Appendix A: Data Analys is Results for FAA RFT .. ..47 PAGE 6 iii LIST OF TABLES Table 1 ICAO Recommendations for Preventative Maintenance of Runway 2 Table 2 ICAO Recommendations and the IFI for the Grip Test er 2 Table 3 A and B Values Obtain ed by Using Data from Individual Runs and Average Friction Values for NASA GT at 65 km/hr 9 Table 4 A and B Values Obtain ed by Using Data from Individual Runs and Average Friction Values for NASA GT at 95 km/hr 9 Table 5 The Variation of Ca libration Constants A and B for NASA GT at 65 km/hr and 95 km/hr 12 Table 6 ttest Results for Co mparison of NASA GT FR60 from 65 km/hr and 95 km/hr with DFT VT, Japan and PTI 13 Table 7 NASA GT Average Calibr ation Constants 15 Table 8 Correlation between Computed F60 and DFT F60 20 Table 9 Minimum, Maximum, M ean and Standard Deviation of A, B for NASA GT at 65 km/hr with DFTVT 23 Table 10 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 95 km/hr with DFTVT 23 Table 11 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 65 km/hr with DFTJapan 24 Table 12 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 95 km/hr with DFTJapan 24 Table 13 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 65 km/hr with DFTPTI 25 Table 14 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 95 km/hr with DFTPTI 25 PAGE 7 iv Table 15 Mean of Predicted F60 and F60VT Values for NASA GT at 65 km/hr 27 Table 16 Mean of Predicted F60 and F60VT Values for NASA GT at 95 km/hr 28 Table 17 Mean of Predicted F60 and F60Japan Values for NASA GT at 65 km/hr 29 Table 18 Mean of Predicted F60 and F60Japan Values for NASA GT at 95 km/hr 30 Table 19 Mean of Predicted F60 and F60PTI Values for NASA GT at 65 km/hr 31 Table 20 Mean of Predicted F60 and F60PTI Values for NASA GT at 95 km/hr 32 Table 21 ttest Results for NASA GT at 65 km/hr and 95 km/hr with DFTVT, Japan and PTI 33 Table 22 ttest Results for FAA RFT at 65 km/hr and 95 km/hr with DFTVT, Japan and PTI 33 Table 23 ZValues for NAS A GT at 65 km/hr and 95 km/hr with Respect to DFTVT 34 Table 24 ZValues for NAS A GT at 65 km/hr and 95 km/hr with Respect to DFTJapan 35 Table 25 ZValues for NAS A GT at 65 km/hr and 95 km/hr with Respect to DFTPTI 36 Table 26 Correlation between Predicted F60 and DFT F60 for NASA GT at 95 km/hr 37 Table 27 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 65 km/hr with DFTVT 41 Table 28 Minimum, Maximum, Mean and Standard Deviation of Predicted F60 & ZValues of DFT F60 with Respect to Predicted F60 41 Table 29 FAA RFT Calibration Constants 47 Table 30 Correlation between Computed F60 and DFT F60 47 PAGE 8 v Table 31 ZValues for FAA RFT at 65 km/hr and 95 km/hr with Respect to DFTVT, Japan and PTI 48 PAGE 9 vi LIST OF FIGURES Figure 1 NASA GT May 2007 Calibration Constants from F60 and FR60 at 65 km/hr with DFTVT 9 Figure 2 NASA GT May 2007 Calibration Cons tants from F60 and FR60 at 65 km/hr with DFTJAPAN 10 Figure 3 NASA GT May 2007 Calib ration Constants from F60 and FR60 at 65 km/hr with DFTPTI 10 Figure 4 NASA GT May 2007 Calibration Constants from F60 and FR60 at 95 km/hr with DFTVT 11 Figure 5 NASA GT May 2007 Calibration Constants from F60 and FR60 at 95 km/hr with DFTJAPAN 11 Figure 6 NASA GT May 2007 Calibration Constants from F60 and FR60 at 95 km/hr with DFTPTI 12 Figure 7 NASA GT May 2005 Average Cali bration Constants 14 Figure 8 NASA GT May 2006 Average Cali bration Constants 14 Figure 9 NASA GT May 2007 Average Cali bration Constants 15 Figure 10 DFT F60 for Validation Se t vs. F60 from Calibration Set, May 2007 (VT InstrumentsCT Meter and DFT) 17 Figure 11 DFT F60 for Validation Se t vs. F60 from Calibration Set, May 2007 (JAPAN InstrumentsCT Meter and DFT) 18 Figure 12 DFT F60 for Validation Se t vs. F60 from Calibration Set, May 2007 (PTI InstrumentsCT Meter and DFT) 18 Figure 13 DFT F60 for Validation Se t vs. F60 from Calibration Set, May 2007 (VT InstrumentsCT Meter and DFT) 19 Figure 14 DFT F60 for Validation Se t vs. F60 from Calibration Set, May 2007 (JAPAN InstrumentsCT Meter and DFT) 19 PAGE 10 vii Figure 15 DFT F60 for Validation Se t vs. F60 from Calibration Set, May 2007 (PTI InstrumentsCT Meter and DFT) 20 Figure 16 DFT F60 vs. Mean of Predicted F60 (NASA GT at 65 km/hr, DFTVT) 27 Figure 17 DFT F60 vs. Mean of Predicted F60 (NASA GT at 95 km/hr, DFTVT) 28 Figure 18 DFT F60 vs. Mean of Predicted F60 (NASA GT at 65 km/hr, DFTJapan) 29 Figure 19 DFT F60 vs. Mean of Predicted F60 (NASA GT at 95 km/hr, DFTJapan) 30 Figure 20 DFT F60 vs. Mean of Predicted F60 (NASA GT at 65 km/hr, DFTPTI) 31 Figure 21 DFT F60 vs. Mean of Predicted F60 (NASA GT at 95 km/hr, DFTPTI) 32 Figure 22 Friction Values from Ra w Data for Surface with an Excessive Z Value (R4) 37 Figure 23 Friction Values from Raw Data for Surface with a Normal Z Value (EK3) 38 Figure 24 MPD vs. Sp Using all the 14 Surfaces and Three Equipment (NASA GT, FAA RFT and VA E274) 39 PAGE 11 viii INVESTIGATION OF THE VALIDITY OF THE ASTM STANDARD FOR COMPUTATION OF INTERNATIONAL FRICTION INDEX Kranthi Kavuri ABSTRACT Runway friction testing is performed in order to enhance the safety of aircraft operation on runways. Preventative maintenance fricti on surveys are performed to determine if there is any deterioration of the frictional resistance on the surface over a period of time and to determine if there is a need for corrective maintenance. In addition operational performance friction surveys are performed to determine frictional properties of a pavement surface in order to provide correct ive action information in maintaining safe takeoff or landing performance limits. A ma jor issue encountered in both types of friction evaluation on runways is the standardization of the friction measurements from different Continuous Friction Measuring Equi pment (CFME). The International Friction Index (IFI) has been formulated to addre ss the above issue and determine the friction condition of a given runway is a standardiz ed format. The ASTM recommended standard procedure to compute the IFI of a runway surface employs two distinct parameters to express the IFI; F60 is the friction value ad justed to a slip speed of 60 km/h and correlated to the standard Dynamic Friction Tester (DFT) measurem ent. And Sp is the speed constant which is governed by the mean profile depth of that surface. The primary objective of this thesis is to i nvestigate the reliability of the current ASTM procedure to standardize runway friction meas urements in terms of IFI. Based on the ASTM standard procedure, two equipment sp ecific calibration constants (A and B) are assigned for each CFME during calibration. Then, in subsequent testing those PAGE 12 ix calibrations constants can be us ed to adjust the equipment m easurements to reliable IFI values. Just as much as A and B are presumed to be characteristic of any given CFME, they are also expected to be independent of the operational speed. The main objective of the annual NASA Runway Friction Workshop held in Wallops Island, Virginia, is to calibrate commonly used CFMEs such that all calibrated equipment would provide a standard reading (i.e. IFI) on a particular surface. During validation of the existing ASTM pr ocedure using the NASA Runway Friction Workshop data it was observed that the si ngle valuebased IFI predictions of the calibrated CFMEs were inaccurate resulting in low correlations with DFT measured values. Therefore, a landing pilot should not be left to make a safe decision with such an uncertain single standard friction value because the actual standard fr iction value could very well be much less than this value. He nce a modified procedur e was formulated to treat the calibration constants A and B as nor mally distributed random variables even for the same CFME. The new procedure can be us ed to predict the IFI (F60) of a given runway surface within a desi red confidence interval. Since the modified procedure predicts a range of IFI fo r a given runway surface with in two bounds, a landing pilots decision would be made easier based on hi s/her experience on critical IFI values. However, even the validation of the modified procedure presented some difficulties since the DFT measurements on a few validated surfaces plotted completely outside the range of F60 predicted by the modified method. Furthermore, although the ASTM standard stip ulates the IFI (F60) predictions to be independent of the testing speed, data from the NASA Runway Friction Workshop indicates a significant difference in the pred ictions from the two testing speeds of 65 km/hr and 95 km/hr, with the results from the 65 km/hr tests yieldi ng better correlations with the corresponding DFT measurements. The a bove anomaly could be attributed to the significantly different FR60 va lues obtained when the 65 km /hr data (FR65) and 95 km/hr data (FR95) are adjusted to a slip speed of 60 km/hr. Extended analytical investigations revealed that the expected testing speed independe ncy of the FR60 for a particular CFME cannot be supported by the ASTM defined general linear relationship PAGE 13 x between Sp and the mean profile depth which probably has been formulated to satisfy a multitude of CFMEs operating on a number of selected test surfaces. This very reason can also be attributed to the above menti oned outliers observed duri ng the validation of the modified procedure. PAGE 14 1 CHAPTER ONE INTRODUCTION Airport runway friction testing is performed to evaluate the coefficient of friction () on runways and it is categorized into tw o distinct operations Federal Aviation Administration (FAA) refers them to as maintenance procedures and operational procedures. Airport runway fr iction testing is performed in order to evaluate the condition of the runway surface for operational and maintenance purposes. This enhances the safety of operations on the runways. (a) Preventative maintenance friction surveys are performed to determine if there is any deterioration of the fr ictional resistance on the surface over a period of time and to determine if there is a need for corrective maintenance. Table 1 shows the thre shold friction values of airport runway surfaces with respect to diffe rent measuring devices. Meanwhile Table 2 illustrates the information specific to the Grip tester. (b) Operational performance friction surveys are performed to determine frictional properties of a pavement su rface in order to provide corrective action information in maintaining safe takeoff or landing performance limits. PAGE 15 2 Table 1 ICAO Recommendations of Values for Preventative Maintenance of Runway Design TargetIntervention LevelMinimum Level Speed (km/hr) 65 95 65 95 65 95 MuMeter 0.72 0.66 0.52 0.38 0.42 0.26 BV11 0.82 0.74 0.60 0.47 0.50 0.34 SFT 0.82 0.74 0.60 0.47 0.50 0.34 RFT 0.82 0.74 0.60 0.54 0.50 0.41 Tatra 0.76 0.67 0.57 0.52 0.48 0.42 GripTester 0.74 0.64 0.53 0.36 0.43 0.24 Table 2 ICAO Recommendations for Values and IFI for the Grip Tester Target Levels for New Surfaces Maintenance Planning Levels Minimum Allowable Friction Levels (65 km/hr) 0.74 0.53 0.43 (95 km/hr) 0.64 0.36 0.24 Sp min (km/hr) 31.0 11.6 7.7 MTD min (mm) 0.375 0.205 0.170 F60 min 0.232 0.119 0.114 1.1. NASA Wallops Runway Friction Workshop NASA Wallops Tire/Runway Friction Work shop is conducted annually at the NASA base in Wallops Island, Virginia to compar e the measurements of each friction testing device. During NASA Tire/Friction Workshops, friction data has been collected using ground vehicles on a number of textured test surfaces. These tests have been performed by various device manufactur ers and end users with a focus on preventative maintenance friction. The main ai m of comparison of friction data is to establish a basic correlation, if any does exist among different friction devices. Also the results of the above Friction Workshop can indicate various issues within the testing procedures used and th e data collected can be used to better understand the performance and evaluation of each friction measuring equipment. In this thesis the data from past three years of NASA Wa llops Tire/Runway Friction Workshop has PAGE 16 3 been analyzed and the calibration constants for each equipment used at the Friction Workshop have been computed using the re levant ASTM standard methodology [1]. Data collected on the same runway paveme nt surface generally confirms that there are differences among friction measuri ng devices because each device reports considerably different friction values on the sa me surface. It is reported that they also show inconsistent repeatability within them selves and no direct correlation to aircraft wheel braking performance. Therefore, standardization of measurements from different equipment is essential. During 1993 1998, the data for the computation of International Friction Index (IFI) at the test sites of Wallops Fli ght Facility was based on the combination of MTD (volumetric te xture depth using glass beads) and the BPN (British Pendulum Number). Based on the ASTM standards [1] Dynamic Friction Tester (DFT) is considered to be the master device for friction measurement in NASA Wallops Runway Friction Works hop, and DFT at 20 km/hr has been used for standardization instead of BPN. 1.2. Objectives of NASA Wallops Tire/Friction Workshop The following are the primary obj ectives of the above workshop (a) To obtain better understanding of diffe rent runway friction measurement procedures and factors influe ncing tire/runway friction performance. (b) Expand the existing friction measurement vehicle correlations to include new devices. (c) Provide opportunity to observe new test and pavement treatment equipment in operation. (d) Evaluate different paveme nt roughness measuring devices. (e) Identify methods to improve harmoni zation between different measurement devices and test proc edures used throughout the world. 1.3. International Friction Index The main aim of PIARC experiment [1] is to harmonize the wet friction and texture measurements which produce the Internationa l Friction Index. Similarly the goal of PAGE 17 4 the Joint Winter Runway Friction Measurement Program (JWRFMP) [2] is to harmonize the friction measurements whic h are obtained from di fferent ground test vehicles on a wide range of winter runway conditions [2 ]. There are three types of friction measuring systems in general; fixed s lip, side force and locked wheel. It is observed that macrotexture parameter is re quired in order to harmonize the results. Therefore it is clear that a friction index is represented by two numbers. One of them is related to macrotexture measurement and the other is related to a friction measurement. The main aim of all these systems is to predict the same values for these macrotexture and fricti on number on a given pavement. 1.4. ASTM Designation E 1960: Standard Practice for Calculating the International Friction Index (IFI) This practice covers the calculation of the IFI from a measurement of pavement macrotexture and wet pavement friction. The IFI was developed in the PIARC international experiment [1] to compare and harmonize texture and skid resistance measurements. This index allows for the harmonizing of friction measurements with different equipment to a common calib rated index. The above ASTM practice provides for harmonization of friction repor ted for devices that use a smooth tread test tire. (a) The IFI consists of two parameters; the calibrated wet friction at 60 km/hr (F60) and the speed constant of wet pavement friction (Sp). (b) The mean profile depth (MPD) and m ean texture depth (MTD) have been shown to be useful in predicting th e speed constant (gradient) of wet pavement friction. (c) A linear transformation of the estimated friction at 60 km/hr provides the calibrated F60 value. The estimated friction at 60 km/hr is determined from a measurement made at any speed by using the speed constant. 1.4.1. Summary of Practice This practice uses measured data of th e pavement surface on macrotexture, and measured friction (FRS) on wet pavement. PAGE 18 5 The practice accommodates the above data measured with different equipment at any measuring speed. The following steps are followed in obtaining the above. (a) Measurement of the pavement macrotext ure is used to estimate the speed constant (Sp) by using Equation (1) MPD Sp 7 89 2 14 (1) where MPD is the Mean Profile Depth which can be obtained from the Circular Texture Meter (CT Meter). (b) Determination of F60 using the DFT valu e at 20 km/hr in accordance with test method E1911 [1] for each of the test se ctions is given by using Equation (2) Spe DFT F4020 732 0 081 0 60 (2) (c) Calculation of Friction at 60 km/hr (FR 60) with the measured friction (FRS) at some slip speed (S) and the speed constant of the pavement (Sp) using Equation(3) Sp Se FRS FR6060 (3) (d) Linear regression of FR 60 (Equation (3)) and F60 (E quation (2)) is used in Equation (4) to obtain the calibration constants A and B. 60 60 FR B A F (4) (e) Reporting of F60 and Sp as IFI (F60, Sp) 1.4.2. Significance and Use of ASTM Standard Methodology (a) This is the practice for calculating the IFI of the pavement. The IFI has proven useful for harmonization of the friction measuring equipment. F60 and Sp PAGE 19 6 have proven to be able to predict the speed dependence of wet pavementrelated measurements of various types of friction measuring equipment. The two IFI parameters (F60 and Sp in Section 1.4.1) have been found to be reliable predictors of the dependence of wet pavement friction on tire slip and vehicle speed. (b) The IFI parameters F60 and Sp can be us ed to calculate the calibrated friction at another slip speed using th e following transformation equation. Sp Se F FS) 60 (* 60 (5) (c) The IFI model given in Equation (6) de scribes the relationship between the values of wet pavement friction FRS m easured at a slip speed of S and the friction values measured by differe nt types of equipment(i=1to n) For the ith equipment, ) ( ) 60 (* 60TX b a Si i i ie FRS B A F (6) (d) The significance of the IFI model is th at the measurement of friction with a device does not have to be at one particular speed us ed in the experiment. FRS can be measured at one slip speed S and is always adjusted to 60 km/hr (FR60). Thus if a device cannot mainta in its normal operating speed and must run at some higher or lower speed because of traffic, the model still works well. In that case S is determined by the vehicle speed (V) which can be converted to S by multiplying V by the per cent slip for fixed slip equipment or by multiplying V by the sine of the slip angle for side force equipment. 1.5. Objectives of Proposed Research The research program described in this thesis seeks to study the limitations of the ASTM IFI computational procedure and verify its eff ectiveness in the standardization of friction measurements from different CFMEs. In order to achieve this objective, first a PAGE 20 7 comprehensive analysis is performed on the friction data obtained from four years of NASA Wallops Tire/Friction Workshops. Then, an alternative procedures for expressing the calibration constants A and B is explored. Finally, the applicability of the standard Speed Constant (Sp) and the Mean Text ure Depth (MPD) rela tionship is also investigated. 1.6. Thesis Organization This thesis is divided into five chapters. The first chapter is the Introduction where the background of the friction measurements a nd their standardizati on is introduced. The second chapter consists of the analysis of NASA Runway Friction Workshop data using the existing methodology. The third chapter de scribes the validation of data using the current method of data analysis. Data Analysis with the modified me thod is presented in the fourth chapter while the final chapter de scribes the conclusions reached based on the research findings. PAGE 21 8 CHAPTER TWO ANALYSIS OF NASA RUNWAY FR ICTION WORKSHOP DATA 2.1. Calibration of the NASA GT At NASA Runway Friction Workshop all the friction evaluation eq uipment are operated at two different speeds; 65 and 95 km/hr. Two ki nds of equipment are used in general i.e. Dynamic Friction Tester and Continuous Friction Measuring Equipment (CFME). CFMEs are operated in repeated runs on a give n surface in order to obtain data with an improved accuracy. After the data collecti on the outlying data are removed and the average value of all the runs is computed as the friction value on a given surface at a particular speed. With known DFT20 and MPD values from NASA Wa llops Friction Workshop data, Sp is calculated using Equation (1) wh ile F60 is calculated using E quation (2). In some years both Sp and F60 values for the tested sections are provided. The friction data obtained at 65 km/hr and 95 km/hr are termed FR65 and FR95 respectively. By substituting FR65 and Sp in Equation (3) one can obtain FR60 for any equipment (ex: NASA GT) when it is operated at 65 km/hr. Similarly from FR95 and Sp one can obtain FR60 for that equipment at 95 km/hr. Then using FR60 obt ained from two different speeds and DFT F60 in Equation (4) two sets of values for A and B are obtained for the same equipment. Table 3 and Table 4 show that the calibrati on constants A and B obtained from Equation (4) by considering data from individual runs is not significantly different from those obtained from the average friction values for both speeds of 65 and 95 km/hr. PAGE 22 9 Table 3 A and B Values Obtained by Using Data from Individual Runs and Average Friction Values for NASA GT at 65 km/hr Friction Data From Individual Runs From Average Friction Values DFT VT JAPAN PTI VT JAPAN PTI B 0.554828 0.5343070.5704660.5615650.542445 0.57852 A 0.172741 0.1839460.1714660.1709840.181703 0.169304 Table 4 A and B Values Obtained by Using Data from Individual Runs and Average Friction Values for NASA GT at 95 km/hr Friction Data From Individual Runs From Average Friction Values DFT VT JAPAN PTI VT JAPAN PTI B 0.532811 0.4916540.5378190.5430630.50548 0.549053 A 0.217433 0.2329430.2189770.2127140.227517 0.213653 Regression analysis is performe d to evaluate the goodness of f it as seen in Figures 1, 2 and 3 for NASA GT at 65 km/hr and 4, 5 and 6 for NASA GT at 95 km/hr with three DFTs. NASA GTy = 0.5616x + 0.171 R2 = 0.8673 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.60.8 FR60 from 65 km/hrF60V T NASA GTat 65 km/hr Linear (NASA GTat 65 km/hr) Figure 1 NASA GT May 2007 Calibration Constants from F60 and FR60 at 65 km/hr with DFTVT PAGE 23 10 NASA GTy = 0.5424x + 0.1817 R2 = 0.8506 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.60.8 FR60 from 65 km/hrF60JAP A NASA GT at 65 km/hr Linear (NASA GT at 65 km/hr) Figure 2 NASA GT May 2007 Calibration Constants from F60 and FR60 at 65 km/hr with DFTJapan NASA GTy = 0.5785x + 0.1693 R2 = 0.8669 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.60.8 FR60 from 65 km/hrF60P T NASA GT at 65 km/hr Linear (NASA GT at 65 km/hr) Figure 3 NASA GT May 2007 Calibration Constants from F60 and FR60 at 65 km/hr with DFTPTI PAGE 24 11 NASA GTy = 0.5431x + 0.2127 R2 = 0.6879 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.60.8 FR60 from 95 km/hrF60V T NASA GT at 95 km/hr Linear (NASA GT at 95 km/hr) Figure 4 NASA GT May 2007 Calibration Constants from F60 and FR60 at 95 km/hr with DFTVT NASA GTy = 0.5055x + 0.2275 R2 = 0.6595 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.60.8 FR60 from 95 km/hrF60P T NASA GT at 95 km/hr Linear (NASA GT at 95 km/hr) Figure 5 NASA GT May 2007 Calibration Constants from F60 and FR60 at 95 km/hr with DFTJAPAN PAGE 25 12 NASA GTy = 0.5491x + 0.2137 R2 = 0.681 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.6 FR60 from 95 km/hrF60P T NASA GT at 95 km/hr Linear (NASA GT at 95 km/hr) Figure 6 NASA GT May 2007 Calibration Constants from F60 and FR60 at 95 km/hr with DFTPTI For the year 2007, the values of A, B and R2 for the year 2007 are shown in Table 5 Table 5 The Variation of Calibration Co nstants A and B for NASA GT at 65 and 95 km/hr Speed 65 km/hr 95 km/hr DFT VT JAPANPTI VT JAPAN PTI A 0.171 0.1817 0.1693 0.2127 0.2275 0.2137 B 0.5616 0.5424 0.5785 0.5431 0.5055 0.5491 R2 0.8673 0.8506 0.8669 0.6879 0.6595 0.681 Similarly the calibration constants for the FAA RFT are shown in the Appendix A 2.1.1. Determination of the Effect of Op erating Speed on the Calibration ASTM standard stipulates that the calibra tion constants A and B must be equipment constants independent of operating speed. In order to achieve this condition the equipment measurements (FRS) are adjusted to a slip speed of 60 km/hr (Equation 3). Therefore the ttest was performed to check whether FR60 values from FR65 and FR95 are significantly different from each other. Th e results of ttest are shown in Table 6 for all the three DFTs: VT, JAPAN and PTI. The ttest for paired differences between two sample for means: Null Hypothesis (H0): d=0 Alternative Hypot hesis (Ha): d 0 PAGE 26 13 Table 6 ttest results for Comparison of NASA GT FR60 from 65 km/hr and 95 km/hr with Three DF TVT, Japan & PTI FR60 ( From 65 km/hr vs. from 95 km/hr) DFT VT Japan PTI t0 4.87 5.37 5.08 Degrees of Freedom 13 13 13 LOC 90% 90% 90% tCritical 1.35 1.35 1.35 It can be seen from Table 6 that to> tcritical, the null hypothesi s is rejected at a 90 % L.O.C. Therefore FR60 obtained from FR65 and FR95 are significantly different from each other at a 90% Le vel of Confidence. 2.2. Calibration for Averaged Operating Speeds In general the equipment is operated at two di fferent speeds i.e. 65 km/hr and 95 km/hr. It was observed that the FR60 values obtai ned at 65 km/hr and 95 km/hr differed significantly. Therefore, by averaging these two FR60s a single va lue of FR60 can be obtained. Furthermore, by using this averag e FR60 and DFT F60 in Equation (4) one can obtain a single correlation for A and B as opposed to different correlations for each operating speed. This averaging is done in order to obtain single values for A and B based on the logic that A and B are constants for any given equipment. Figures 7, 8 and 9 show the regression an alysis for the linear relationship between 60 F and average FR60. The calibration constants fo r the averaged FR60 for three years (2005, 2006, and 2007) are shown in Table 7. PAGE 27 14 NASA GTy = 0.5006x + 0.1981 R2 = 0.9779 0.000 0.100 0.200 0.300 0.400 0.500 0.600 00.20.40.60.8 FR60 from averagesF60 NASA GT Linear (NASA GT) Figure 7 NASA GT May 2005 Average Calibration Constants NASA GTy = 0.7856x + 0.1484 R2 = 0.9041 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 00.20.40.60.8 FR60 from averagesF60 NASA GT Linear (NASA GT) Figure 8 NASA GT May 2006 Average Calibration Constants PAGE 28 15 NASA GTy = 0.5358x + 0.2016 R2 = 0.7699 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.60.8 FR60 from averagesF60 NASA GT Linear (NASA GT) Figure 9 NASA GT May 2007 Average Calibration Constants Table 7 NASA GT Average Calibration Constants EquipmentNASAGT Year 2005 2006 2007 A 0.198 0.148 0.202 B 0.500 0.786 0.536 R2 0.978 0.904 0.769 PAGE 29 16 CHAPTER THREE VALIDATION FOR THE CURRENT METHOD OF DATA ANALYSIS 3.1. Validation for Individual Operating Speeds During the calibration operation A and B valu es are computed for all the friction measuring equipment at their characteristic operating speeds. A and B values obtained for these friction measuring equipment can then be used for future testing. Therefore a validation procedure was performed in this thesis to validate the calibrated A and B values by predicting the measurements on the ot her four test surfaces. In this exercise the data obtained from 2007 NASA Wallops Runway Friction Workshop is used for the purpose of validation. Of the fourteen surfaces included in the entire testing program the data for the following ten test surfac es is used for computing A and B. Calibration surface set: A, B, C, D, E, F, G, Echo 1, EK 1, EK 2 Thus the data on the following four test surf aces are used for the purpose of validating the computed A and B. Validation surface set: R4, Echo 2, EK 3, and EK 4 At NASA Wallops Runway Friction Workshop all the equipment ar e operated at 65 km/hr and 95 km/hr. As shown in Section 2.2 when A and B values are calculated for these two speeds for any given equipment th eir values were found to be different. However per ASTM standards [1] A and B valu es are expected to be constants for a given equipment. For each operating speed (65 km/hr and 95 km/hr), F60 for the validation surfaces were predicted using A a nd B values obtained on the calibration set of PAGE 30 17 surfaces using three different DFT and CT meters; VT, JAPAN and PTI and FR60 obtained from the validation surfaces (Equa tion (3)). Therefore F60 computed for validation test surfaces using Equation (2) are compared with those predicted above. The results are shown in Figures 10, 11 and 12 for a speed of 65 km/hr and Figures 13, 14 and 15 for a speed of 95 km/hr. NASA GTy = 0.8749x + 0.0276 R2 = 0.501 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 00.10.20.30.4 F60 (computed)F60 (DFT) NASA GT Linear (NASA GT) Figure 10 DFT F60 for Validation Set vs F60 from Calibration Set, May 2007 (VT InstrumentsCT meter and DFT) PAGE 31 18 NASA GTy = 1.113x 0.0524 R2 = 0.6254 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 00.10.20.30.4 F60(computed)F60(DFT) NASA GT Linear (NASA GT) Figure 11 DFT F60 for Validation Set vs F60 from Calibration Set, May 2007 (Japan InstrumentsCT Meter and DFT) NASA GTy = 1.0221x 0.04 R2 = 0.5386 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 00.10.20.30.4 F60(Computed)F60(DFT) NASA GT Linear (NASA GT) Figure 12 DFT F60 for Validation Set vs F60 from Calibration Set, May 2007 (PTI InstrumentsCT Meter and DFT) PAGE 32 19 NASA GTy = 0.879x + 0.025 R2 = 0.324 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 00.10.20.30.4 F60 (computed)F60(DFT) NASA GT Linear (NASA GT) Figure 13 DFT F60 for Validation Set vs F60 from Calibration Set, May 2007 (VT InstrumentsCT Meter and DFT) NASA GTy = 1.1978x 0.0794 R2 = 0.4372 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 00.10.20.30.4 F60(computed)F60(DFT) NASA GT Linear (NASA GT) Figure 14 DFT F60 for Validation Set vs F60 from Calibration Set, May 2007 (JAPAN InstrumentsCT Meter and DFT) PAGE 33 20 NASA GTy = 0.9636x 0.0145 R2 = 0.3359 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 00.10.20.30.4 F60(computed)F60(DFT) NASA GT Linear (NASA GT) Figure 15 DFT F60 for Validation Set vs F60 from Calibration Set, May 2007 (PTI InstrumentsCT Meter and DFT) Table 8 shows the correlation results between DFT F60 and computed F60 at different speeds. Table 8 Correlation between Computed F60 and DFT F60 NASA Grip Tester DFT VT JAPAN PTI Operating Speed 65 km/hr 95 km/hr 65 km/hr 95 km/hr 65 km/hr 95 km/hr Correlation between F60 (DFT) and F60(Computed) 0.501 0.324 0.625 0.437 0.538 0.335 Similarly correlation for FAA RFT is given in the Appendix A. PAGE 34 21 CHAPTER FOUR DATA ANALYSIS WITH THE MODIFIED METHOD 4.1. Calibration with Modified Method At NASA Wallops Tire/Fricti on Workshop 14 test surfaces are provided for the purpose of runway friction testing. The main purpose of the testing program is to measure friction on the test surfaces and adjust those measurem ents to the International Friction Index IFI (F60, Sp). Various types of equipment are used for the purpose of friction testing of which the Dynamic Friction Tester (DFT) is treated as the standard device which measures IFI on the test surf ace. All other equipment are Continuous Friction Measuring Equipment (CFME) which acquire dynamic me asurements. CFMEs are operated at the two speeds of 65 km/hr and 95 km/hr. ASTM sta ndard Equation (3) is used to adjust the measured friction value FRS at a given slip speed S to a common slip speed of 60 km/hr. The current ASTM procedure calcula tes the calibration constants from the regression of the adjusted measurement (F R60) and F60 obtained from the Dynamic Friction Tester (DFT) on the same test surf ace. Two single value calibration constants (A and B) are obtained for each CFME with resp ect to DFT. The calibration constants vary from equipment to equipment while they are presumed to be constants for particular equipment. The above procedure is illustrate d for the NASA Grip Tester in Chapter 3. The application of the ASTM pr ocedure is in that when the calibrated CFME is operated on a new test surface the friction value obtaine d on that new surface is converted to FR60 using Equation (3). FR60 is then adjusted to F60 using the above calibration constants. Therefore the predicted friction value i.e. IF I on the new test surface is a single constant PAGE 35 22 value which could differ from the actual IFI on the new test surface by a certain margin of error. The validation procedure described in Chapte r 3 assumes that only ten (10) surfaces are used for testing to obtain the calibration cons tants while the remaining four (4) surfaces are provided for validation. The predicted F6 0 values for the four surfaces are then correlated with the F60 obtained from the DFT on the same test surfaces. The results of the above correlation for NASA GT are shown in Table 8 and Figures 6 to 11 in Chapter 3. The validation results indicate that by usi ng single values of A and B one can obtain a single value for F60 which can deviate in eith er direction from the actual friction value on that surface. In order to address this probl em an alternative procedure is developed to standardize the runway friction measurements. In this procedure the calibration constants A and B are treated as random variables instead of single valued vari ables. The predicted F60 will then be a random variable which woul d be within a certain confidence interval. 4.1.1. Procedure for Randomization of the Calibration Constants The ASTM standard practice fo r calibration of friction tester s recommends the use of at least 10 surfaces to compute the calibration c onstants. Therefore from the 14 available test surfaces, a sample size of 13 surfaces is us ed for calibration with 1 surface left out for validation in each trial. The reason for selecti ng 13 surfaces is to compute A and B with a better accuracy by employing more of the availa ble data. In each of such 14 trials the surface left out itself can be made the validation test surface. Therefore using this technique, 14 different validations can be performed. Since there are 13 surfaces for calibration and every time 10 surfaces are used to compute the calib ration constants, 13C10 combinations are available to compute two ranges for A and B. The minimum, maximum and mean of A, B for all the possible 14 trials are shown in Table 9 10 for 65 km/hr and 95 km/hr with DFTVT, Table 11, 12 for 65 km/hr and 95 km/hr with DFTJapan and Table 13 and 14 for 65 km/hr and 95 km/hr with DFTPTI. PAGE 36 23 Table 9 Minimum, Maximum, Mean and St andard Deviation of A, B for NASA GT at 65 km/hr with DFTVT Calibration Constants A B Surface Min Max Mean Std Dev Min Max Mean Std Dev A 0.1460 0.21320.17480.01290.41810.6939 0.5551 0.0378 B 0.1329 0.20500.16660.01230.43720.7300 0.5866 0.0432 C 0.1329 0.20370.16380.01150.44230.7300 0.5972 0.0392 D 0.1464 0.21290.17540.01250.42250.6980 0.5553 0.0374 E 0.1329 0.19380.16180.01050.45680.7169 0.5772 0.0346 F 0.1533 0.21400.17800.01210.41290.6235 0.5239 0.0349 G 0.1329 0.21400.17420.01450.41290.7300 0.5488 0.0532 Echo1 0.1403 0.19920.16580.01120.43810.6915 0.5660 0.0355 EK1 0.1462 0.21400.18220.01370.41290.6888 0.5358 0.0411 EK2 0.1419 0.20420.16950.01220.43530.6999 0.5634 0.0365 R4 0.1441 0.21360.17870.01200.43100.7300 0.5567 0.0385 Echo2 0.1422 0.20300.16860.01160.42950.6916 0.5614 0.0361 EK3 0.1431 0.20540.17100.01210.43270.7010 0.5612 0.0365 EK4 0.1365 0.21400.17060.01410.41290.7167 0.5626 0.0403 Table 10 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 95 km/hr with DFTVT Calibration Constants A B Surface Min Max Mean Std Dev Min Max Mean Std Dev A 0.1522 0.27240.21680.01990.35480.8990 0.5414 0.0843 B 0.1334 0.24390.19830.02380.37380.9906 0.6310 0.1235 C 0.1334 0.24480.19450.02230.36870.9906 0.6481 0.1163 D 0.1536 0.27000.21680.01950.36040.9185 0.5439 0.0846 E 0.1334 0.24580.20130.01870.39460.9630 0.5735 0.0853 F 0.1582 0.25530.21200.01700.34140.7886 0.5106 0.0733 G 0.1334 0.26690.21810.02060.34140.9906 0.5047 0.0994 Echo1 0.1438 0.24940.20620.01800.37720.8945 0.5571 0.0814 EK1 0.1518 0.27240.23030.01920.34140.8735 0.5013 0.0846 EK2 0.1472 0.25980.21100.01920.37050.9078 0.5519 0.0833 R4 0.1467 0.27240.21890.01950.36550.9906 0.5477 0.0885 Echo2 0.1439 0.24900.20610.01800.37710.8928 0.5570 0.0812 EK3 0.1490 0.25940.21160.01890.36960.9159 0.5510 0.0834 EK4 0.1375 0.27240.21450.02230.34140.9502 0.5443 0.0910 PAGE 37 24 Table 11 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 65 km/hr with DFTJapan Calibration Constants A B Surface Min Max Mean Std Dev Min Max Mean Std Dev A 0.1526 0.23350.18540.01430.39380.6856 0.5362 0.0422 B 0.1398 0.21710.17610.01350.41910.7253 0.5723 0.0476 C 0.1398 0.21410.17270.01220.43030.7253 0.5849 0.0418 D 0.1518 0.22960.18450.01390.40000.6895 0.5382 0.0416 E 0.1398 0.20790.17300.01150.43140.7090 0.5578 0.0381 F 0.1563 0.23700.18630.01360.36490.6406 0.5156 0.0423 G 0.1398 0.23700.18830.01620.36490.7253 0.5171 0.0559 Echo1 0.1462 0.21240.17650.01210.41470.6810 0.5473 0.0388 EK1 0.1530 0.23700.19580.01540.36490.6796 0.5094 0.0466 EK2 0.1492 0.22130.18080.01330.40880.6900 0.5434 0.0404 R4 0.1491 0.23550.18930.01370.40350.7253 0.5373 0.0435 Echo2 0.1505 0.21790.18020.01270.40080.6804 0.5404 0.0400 EK3 0.1497 0.22060.18160.01300.40630.6920 0.5417 0.0404 EK4 0.1486 0.23700.18580.01610.36490.6932 0.5339 0.0460 Table 12 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 95 km/hr with DFTJapan Calibration Constants A B Surface Min Max Mean Std Dev Min Max Mean Std Dev A 0.1608 0.29620.23100.02190.29120.8582 0.5052 0.0900 B 0.1459 0.27370.21190.02620.34530.9533 0.5978 0.1307 C 0.1459 0.27170.20790.02460.34500.9533 0.6155 0.1229 D 0.1608 0.29180.23000.02140.30280.8740 0.5086 0.0899 E 0.1459 0.27300.21680.02100.33930.9235 0.5352 0.0912 F 0.1628 0.27800.22460.01890.29070.8031 0.4900 0.0820 G 0.1459 0.29420.23480.02230.29070.9533 0.4574 0.0992 Echo1 0.1528 0.27410.22110.01990.33130.8579 0.5198 0.0864 EK1 0.1613 0.29620.24840.02110.29070.8425 0.4535 0.0898 EK2 0.1566 0.28490.22610.02110.31400.8639 0.5139 0.0887 R4 0.1576 0.29620.23320.02160.30400.9533 0.5106 0.0944 Echo2 0.1537 0.27370.22120.01970.33060.8538 0.5183 0.0860 EK3 0.1575 0.28150.22570.02060.31940.8728 0.5140 0.0885 EK4 0.1506 0.29620.23310.02460.29070.8816 0.4971 0.0968 PAGE 38 25 Table 13 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 65 km/hr with DFTPTI Calibration Constants A B Surface Min Max Mean Std Dev Min Max Mean Std Dev A 0.1429 0.21310.17150.01300.42210.7044 0.5742 0.0400 B 0.1328 0.20210.16270.01130.45890.7440 0.6141 0.0399 C 0.1328 0.21200.16400.01220.41700.7440 0.6062 0.0453 D 0.1451 0.21270.17270.01250.42500.7065 0.5729 0.0394 E 0.1328 0.19550.16010.01020.46000.7277 0.5948 0.0361 F 0.1483 0.21760.17500.01270.40890.6488 0.5456 0.0413 G 0.1328 0.21760.17600.01460.40890.7440 0.5513 0.0541 Echo1 0.1377 0.20140.16460.01120.43840.6965 0.5819 0.0376 EK1 0.1449 0.21760.18140.01390.40890.6907 0.5490 0.0446 EK2 0.1412 0.20720.16890.01220.43420.7071 0.5784 0.0387 R4 0.1427 0.21500.17680.01200.43220.7440 0.5735 0.0406 Echo2 0.1414 0.20620.16800.01160.42770.6996 0.5764 0.0386 EK3 0.1436 0.20890.17080.01210.43190.7124 0.5762 0.0389 EK4 0.1373 0.21760.16970.01410.40890.7209 0.5775 0.0426 Table 14 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 95 km/hr with DFTPTI Calibration Constants A B Surface Min Max Mean Std Dev Min Max Mean Std Dev A 0.1493 0.27230.21650.02050.34710.8893 0.5488 0.0896 B 0.1367 0.24750.19650.02320.38570.9840 0.6533 0.1213 C 0.1367 0.25480.19790.02420.34110.9840 0.6426 0.1288 D 0.1514 0.26980.21680.02000.35480.9021 0.5505 0.0895 E 0.1367 0.25030.20260.01930.38840.9522 0.5796 0.0904 F 0.1542 0.25950.21170.01780.33160.8282 0.5253 0.0820 G 0.1377 0.27240.22190.02070.33160.9840 0.4938 0.0981 Echo1 0.1431 0.25330.20750.01850.37030.8937 0.5620 0.0862 EK1 0.1507 0.27240.23270.01970.33160.8818 0.5003 0.0902 EK2 0.1470 0.26340.21280.01980.36210.8940 0.5561 0.0884 R4 0.1495 0.27240.21950.02000.35890.9840 0.5537 0.0937 Echo2 0.1438 0.25360.20790.01850.36870.8918 0.5610 0.0862 EK3 0.1490 0.26380.21390.01950.36100.9065 0.5551 0.0889 EK4 0.1367 0.27240.21580.02290.33160.9277 0.5488 0.0963 PAGE 39 26 4.2. Validation for the Proposed Method of Data Analysis The NASA GT is operated on all of the 14 test surfaces at two different speeds of 65 km/hr and 95 km/hr from which FR60 values are obtained using Equation (3). As described in Section 4.1 since the cal ibration constants are obtained from 13C10 combinations leaving out 1 surface in each trial, 13C10 values of A and B are obtained. Then the left out surface becomes the validat ion surface in that iteration. F60 on the validation surface is then predicted using the 13C10 values of A and B and FR60 obtained from each speed (Equation (4)). The above range of F60 values represent F60 as a random variable which is then compared w ith a single value of F60 measured on the validation test surface using the DFT. The pr edicted F60 can be co mpared with the DFT measured F60 on a given validation surface in two ways, as described in the Approaches 1 and 2. Approach 1: For all the 14 validation surfaces the mean of the predicted F60 can be correlated to the DFT F60 to verify the accuracy of validati on. These results are shown in the Tables 15 and 16 for 65 km/hr and 95 km/hr respectivel y with DFTVT, Tables 17 and 18 for 65 km/hr and 95 km/hr with DFTJapan, Tables 19 and 20 for 65 km/hr and 95 km/hr with DFTPTI. The correlation between the mean of predicted F60 and DFT F60 are shown in Figures 1621 for 65 km/hr and 95 km/hr with DFTVT, JAPAN, PTI DFTs. In addition the ttest was performed to check whether the mean of the predicted F60 and the DFT measured F60 are significantly di fferent from each other. The two hypotheses are: H0: Means of two samples are equal Ha: Means of two samples are significantly different from each other The results of the ttest are shown in Tabl es 21 and 22 for NASA GT and FAA RFT at 65 km/hr and 95 km/hr with three DFTsVT, Japan a nd PTI. From these results it is seen that there is no evidence to reject the null hypothesis. Therefore it is concluded that there is no significant difference between the means of predicted F60 and the DFT F60. PAGE 40 27 Table 15 Mean of Predicted F60 and F60VT Values for NASA GT at 65 km/hr NASA GT at 65 km/hr, DFTVT Surface Mean of Predicted F60 F60VT A 0.290016 0.269372 B 0.492975 0.443917 C 0.522359 0.461858 D 0.311608 0.281669 E 0.279913 0.339642 F 0.460673 0.537889 G 0.532319 0.55103 Echo1 0.326049 0.379471 EK1 0.197704 0.158143 EK2 0.307802 0.322144 R4 0.3473 0.266002 Echo2 0.347425 0.38292 EK3 0.330054 0.333793 EK4 0.235554 0.239999 y = 0.9756x + 0.0077 R2 = 0.82210 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.6 Mean of Predicted F60F60VT F60 vs Predicted F60 Linear (F60 vs Predicted F60) Figure 16 DFT F60 vs. Mean of Predicted F60 (NASA GT at 65 km/hr, DFTVT) PAGE 41 28 Table 16 Mean of Predicted F60 and F60VT Values for NASA GT at 95 km/hr NASA GT at 95 km/hr, DFTVT Surface Mean of Predicted F60 F60VT A 0.305073 0.269372 B 0.543898 0.443917 C 0.576264 0.461858 D 0.326174 0.281669 E 0.275131 0.339642 F 0.3991 0.537889 G 0.465447 0.55103 Echo1 0.326364 0.379471 EK1 0.239679 0.158143 EK2 0.31612 0.322144 R4 0.351789 0.266002 Echo2 0.327531 0.38292 EK3 0.333225 0.333793 EK4 0.253561 0.239999 y = 0.833x + 0.055 R2 = 0.5624 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.60.8 Mean of Predicted F60F60VT F60 vs Predicted F60 Linear (F60 vs Predicted F60) Figure 17 DFT F60 vs. Mean of Predicted F60 (NASA GT at 95 km/hr, DFTVT) PAGE 42 29 Table 17 Mean of Predicted F60 and F60Japan Values for NASA GT at 65 km/hr NASA GT at 65 km/hr, DFTJapan Surface Mean of predicted F60 F60Japan A 0.307876 0.289988 B 0.488881 0.435573 C 0.515786 0.449775 D 0.321829 0.307021 E 0.292267 0.347154 F 0.434425 0.49739 G 0.516698 0.550702 Echo1 0.332894 0.386467 EK1 0.21237 0.169057 EK2 0.321707 0.333995 R4 0.355028 0.279373 Echo2 0.362604 0.396476 EK3 0.350163 0.360009 EK4 0.25601 0.243078 y = 0.9727x + 0.0083 R2 = 0.7961 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.6 Mean of Predicted F60F60JAPAN F60 vs Predicted F60 Linear (F60 vs Predicted F60) Figure 18 DFT F60 vs. Mean of Predicted F60 (NASA GT at 65 km/hr, DFTJapan) PAGE 43 30 Table 18 Mean of Predicted F60 and F60Japan Values for NASA GT at 95 km/hr NASA GT at 95 km/hr, DFTJapan Surface Mean of Predicted F60 F60Japan A 0.320925 0.289988 B 0.533982 0.435573 C 0.562847 0.449775 D 0.336031 0.307021 E 0.28849 0.347154 F 0.386241 0.49739 G 0.453493 0.550702 Echo1 0.334217 0.386467 EK1 0.257956 0.169057 EK2 0.329283 0.333995 R4 0.359281 0.279373 Echo2 0.340391 0.396476 EK3 0.349241 0.360009 EK4 0.273279 0.243078 y = 0.8065x + 0.0652 R2 = 0.5137 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.6 Mean of Predicted F60F60JAPAN F60 vs Predicted F60 Linear (F60 vs Predicted F60) Figure 19 DFT F60 vs. Mean of Predicted F60 (NASA GT at 95 km/hr, DFTJapan) PAGE 44 31 Table 19 Mean of Predicted F60 and F60PTI Values for NASA GT at 65 km/hr NASA GT at 65 km/hr, DFTPTI Surface Mean of Predicted F60 F60PTI A 0.288177 0.279806 B 0.493327 0.429147 C 0.502738 0.455747 D 0.313054 0.292853 E 0.279727 0.337679 F 0.440597 0.50926 G 0.524614 0.559099 Echo1 0.328382 0.380571 EK1 0.198353 0.159461 EK2 0.311962 0.320564 R4 0.348945 0.271381 Echo2 0.354078 0.383144 EK3 0.334395 0.326104 EK4 0.242893 0.245092 y = 0.9817x + 0.0057 R2 = 0.8203 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.6 Mean of Predicted F60F60PTI F60 vs Predicted F60 Linear (F60 vs Predicted F60) Figure 20 DFT F60 vs. Mean of Predicted F60 (NASA GT at 65 km/hr, DFTPTI) PAGE 45 32 Table 20 Mean of Predicted F60 and F60PTI Values for NASA GT at 95 km/hr NASA GT at 95 km/hr, DFTPTI Surface Mean of Predicted F60 F60PTI A 0.304298 0.279806 B 0.543676 0.429147 C 0.552365 0.455747 D 0.327383 0.292853 E 0.275985 0.337679 F 0.386852 0.50926 G 0.457053 0.559099 Echo1 0.328075 0.380571 EK1 0.242745 0.159461 EK2 0.3195 0.320564 R4 0.352813 0.271381 Echo2 0.331698 0.383144 EK3 0.336572 0.326104 EK4 0.258657 0.245092 y = 0.8299x + 0.0561 R2 = 0.5456 0 0.1 0.2 0.3 0.4 0.5 0.6 00.20.40.6 Mean of Predicted F60F60PTI F60 vs Predicted F60 Linear (F60 vs Predicted F60) Figure 21 DFT F60 vs. Mean of Predicted F60 (NASA GT at 95 km/hr, DFTPTI) PAGE 46 33 Table 21 ttest Results for NASA GT at 65 km/hr and 95 km/hr with DFTVT, Japan and PTI NASA GT DFT VT Japan PTI Speed (km/hr) 65 95 65 95 65 95 to 0.024 0.125 0.043 0.155 0.021 0.125 Degree of freedom 25.86 25.71 25.8 25.64 25.83 25.65 LOC 90% 90% 90% 90% 90% 90% tcritical 1.315 1.315 1.315 1.315 1.315 1.315 Table 22 ttest Results for FAA RFT at 65 km/hr and 95 km/hr with DFTVT, Japan and PTI FAA RFT DFT VT Japan PTI Speed (km/hr) 65 95 65 95 65 95 to 0.052 0.087 0.073 0.111 0.052 0.089 Degree of freedom 25.91 25.84 25.86 25.78 25.89 25.82 LOC 90% 90% 90% 90% 90% 90% tcritical 1.315 1.315 1.315 1.315 1.315 1.315 Approach 2: In this validation procedure the predicted range of F60 is considered as a normal distribution. Although F60 does not need to follow a normal distribution, in order to predict Zvalues for DFT F60 with respect to DFT F60, F60 is assumed to follow a normal distribution. Then the Z values are computed to locate the position of the measured DFT F60 with respect to the predicted F60. It is known that Zvalues with an absolute magnitude higher than about 4.0 correspond to measured values which do not fall within the predicted domain. The Zvalues obtained for each left out surf ace are shown in Tables 23, 24 and 25 with DFTVT, Japan and PTI respectively. The ab solute Zvalues corre sponding to several surfaces (E, F, Echo1, R4, and Echo2) are grea ter than 4.0 showing that the means of measured F60 do not lie in the predicted F60 range. This anomaly in the values of Z shows the presence of an error. The possible sources of this error are discussed below. PAGE 47 34 Table 23 Z Values for NASA GT at 65 km/hr and 95 km/hr with Respect to DFTVT NASA GT, DFTVT ZValues At 65 km/hr At 95 km/hr A 2.734 3.32808 B 3.26694 2.15378 C 3.89403 2.34326 D 4.32473 4.43086 E 9.780506 6.108974 F 7.336648 9.203925 G 0.818629 2.657498 Echo1 8.425362 5.410487 EK1 3.11808 4.56925 EK2 2.070625 0.595064 R4 13.4441 8.16776 Echo2 5.154237 5.632148 EK3 0.545695 0.054914 EK4 0.428963 0.80615 PAGE 48 35 Table 24 Z Values for NASA GT at 65 km/hr and 95 km/hr with Respect to DFTJapan NASA GT, DFTJapan ZValues SurfaceAt 65 km/hr At 95 km/hr A 2.44291 2.90385 B 3.50566 2.11505 C 4.36602 2.32699 D 2.16007 2.89684 E 9.143463 5.182773 F 6.1077 8.19009 G 1.560032 3.337786 Echo1 8.892358 5.481526 EK1 3.08823 4.58592 EK2 1.835077 0.470105 R4 12.6637 7.78365 Echo2 4.932331 5.779301 EK3 1.46867 1.036551 EK4 1.1875 1.69694 PAGE 49 36 Table 25 Z Values for NASA GT at 65 km/hr and 95 km/hr with Respect to DFTPTI NASA GT, DFTPTI Zvalues Surface At 65 km/hr At 95 km/hr A 1.15533 2.32171 B 4.87454 2.6224 C 2.95866 1.96661 D 3.07658 3.53971 E 10.32233 5.871776 F 6.385201 8.448261 G 1.571269 3.407478 Echo1 8.698641 5.523046 EK1 3.05973 4.61073 EK2 1.317826 0.108849 R4 13.3833 7.86875 Echo2 4.273046 5.31191 EK3 1.2722 1.03746 EK4 0.226473 0.82307 Similar results for FAA RFT are shown in the Appendix A. 4.2.1. Speed Sensitivity The validation results sh ow that the A and B values obtai ned from 65 km/hr speed results provide a better correla tion. Therefore A and B values at 65 km/hr can be assumed as calibration constants for NASA GT. Furthermore when the A and B values corresponding to 65 km/hr are used to predict the range of F60 from 95 km/hr data, the mean of the predicted F60 values agreed better with the DFT F60 (the standard) than the F60 predicted using A and B corresponding to 95 km/hr. Table 26 shows the correlation of predicted F60 at 95 km/hr and the DFT F60 using A and B values obtained from both 95 km/hr and 65 km/hr. PAGE 50 37 Table 26 Correlation between Predicted F60 and DFT F60 for NASA GT at 95 km/hr DFT VT JAPANPTI Predicted F60 Using A,B from 95 km/hr Data 0.5624 0.5137 0.5456 Predicted F60 Using A,B from 65 km/hr Data 0.6879 0.681 0.6595 4.2.2. Raw Data Verification Continuous friction measuring e quipment measure friction values at every 1 ft intervals. Therefore raw data for each surface was checked to see whether there is any abnormal variation between the average friction value for each surface and the raw data obtained at intervals of 1 ft. Figure 22 a nd 23 show the raw data and th e averages for surfaces with excessive and normal Zvalues respectively. Since there is no obvious difference between Figure 22 and Figure 23 it was co ncluded that the variation of raw data does not explain the excessive Zvalues. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 050100150200250300350 Distance in ftFrictio n Raw Data Average Figure 22 Friction Values from Raw Data for Surface with an Excessive Z value (R4) PAGE 51 38 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 050100150200250300350 Distance in ftFrictio n Raw Data Average Figure 23 Friction Values from Raw Data for Surface with a Normal Z Value (EK3) 4.2.3. Check for Effectiveness of Sp and MPD Linear Relationship The ASTM Standard procedure emphasizes that adjusted friction at 60 km/hr slip speed must not depend on the operating speed. However results of ttest (Section 2.1.1) show that the FR60 obtained from FR65 and FR95 ar e significantly different from each other. By assuming FR60 to be a constant fo r the above two speeds a backcalculation procedure was performed to observe the speed constant Sp vs. MPD relationship. It was observed that while a linear relationship yielded a R2 values of 0.5, a polynomial of degree 6 (Equation (7)) yielde d a correlation of 0.974. 11 312 3 3323 12436 21230 17887 7172 3 10882 3 4 5 6 MPD MPD MPD MPD MPD MPD Sp (7) Hence it is suspected that one reason for obtaining abnormally high Z values during validation could be due to the in applicability of the general li near relationship to any one given equipment and a set of test surfaces as seen in Equation (3). PAGE 52 39 4.2.4. Computation of Sp and MPD Relationship In Figure 24 the author plotted the Sp for three equipment (NASA GT, FAA RFT and VA E274) vs. MPD for all 14 test surfaces. The best fit line given by Equation (8) and the line corresponding to Equa tion (3) seem to agree to some degree. MPD Sp 163 69 48 25 (8) However the R2 value corresponding to Equation (8) is only 0.10 showing that it is a very approximate general relationship for th e equipment (NASA GT, FAA RFT and VA E274). Hence the same argument applies from Equation (3) as well. y = 69.163x + 25.484 R2 = 0.1018 y = 89.7x + 14.2 0 100 200 300 400 500 600 700 800 900 1000 00.511.522.53 MPDS p Equipment (NASA GT, FAA RFT& VA E274) CT Meter Linear (Equipment (NASA GT, FAA RFT& VA E274)) Linear (CT Meter ) Figure 24 MPD vs. Sp Using all the 14 Surfaces and Three Equipment (NASA GT, FAA RFT and VA E274) 4.2.5. Alternative Calculation for more Accurate Sp and MPD Relationship In this process two equipment NASA GT and FAA RFT were considered with DFTVT data. If one assumes the FR60 resulting from FR65 and FR95 to be the same, Equation (3) yields (9) Sp FR FR LN S S ) 95 65 ( ) 1 2 ( PAGE 53 40 Where S2= 95*0.145 S1= 65*0.145 From Equation (9) MPD can be expressed in terms of a multiple linear regression equation which is developed with the Sp fr om two equipment as independent variables and MPD obtained from the CT mete rVT as the dependent variable. MPD = 0.20490.0095*(Sp)1+0.0781*(Sp)2 (10) Where (Sp)1 is obtained from NASA GT and (Sp)2 is obtained from FAA RFT. For the data analysis of NASA GT, (Sp)2 becomes zero. Then the Equation (10) is reduced to Equation (11) and Sp for NASA GT can be computed from MPD. MPD= 0.20490.0095*(Sp)1 (11) For the data analysis of FAA RFT, (Sp)2 becomes zero. Then the Equation (10) is reduced to Equation (12) and Sp for FAA RFT can be computed from MPD. MPD = 0.2049 + 0.0781*(Sp)2 (12) Then the modified method of data analysis described in Chapter 4 was performed with the Sp value obtained from the multiple li near regression equation. The minimum, maximum and mean of A, B for all the possi ble 14 trials are shown in Table 27 for 65 km/hr with DFTVT. While the minimum, maxi mum, mean and Zvalues of the predicted F60 are shown in Table 28 for 65 km/hr with DFTVT. PAGE 54 41 Table 27 Minimum, Maximum, Mean and Standard Deviation of A, B for NASA GT at 65 km/hr with DFTVT Calibration Constants A B Surface Min Max Mean Std Dev Min Max Mean Std Dev A 0.0890 0.84390.25580.15680.4394 0.5025 0.10950.1488 B 0.0376 0.89970.27440.15390.4977 0.3402 0.07540.1404 C 0.0067 0.90660.25640.15450.5038 0.3692 0.09160.1411 D 0.0890 0.85710.27780.15600.4507 0.5025 0.08600.1466 E 0.0890 0.98340.28590.16440.5833 0.5025 0.07400.1560 F 0.0470 0.80220.26050.13890.4149 0.3219 0.08180.1273 G 0.0996 0.88380.29540.14420.4980 0.2707 0.04670.1313 Echo1 0.0521 0.95800.29540.16130.5482 0.3324 0.05980.1472 EK1 0.3318 1.00150.55410.13790.5833 0.0353 0.175 0.1232 EK2 0.0890 1.00150.25460.19350.5833 0.5025 0.10830.1862 R4 0.0982 1.00150.37760.22820.5833 0.2900 0.015 0.2083 Echo2 0.0353 0.91150.30100.15400.5109 0.3551 0.05460.1425 EK3 0.0216 0.90080.29810.15570.4945 0.3740 0.06120.1441 EK4 0.0390 1.00150.31730.16940.5788 0.3628 0.04930.1560 Table 28 Minimum, Maximum, Mean and Standard Deviation of Predicted F60 and Z Values of DFT F60 with Respect to Predicted F60 F60Predicted Surface Min Max Mean Std dev Zvalue A 0.31610.51460.38730.0297 3.9717 B 0.27800.45760.34410.0289 3.4475 C 0.27010.47920.33690.0342 3.6532 D 0.33180.48220.37560.0222 4.2224 E 0.28010.51700.37510.0315 1.1249 F 0.28370.42320.33880.0227 8.7677 G 0.29740.39140.34370.0163 12.7299 Echo1 0.29000.46000.35190.0278 0.9904 EK1 0.33450.64810.44680.0635 4.5451 EK2 0.19250.59250.40140.0629 1.2601 R4 0.27530.55900.36880.0704 1.4606 Echo2 0.31170.42210.36010.0184 1.2411 EK3 0.31320.41890.36280.0186 1.5601 EK4 0.31230.43890.36600.0228 5.5190 PAGE 55 42 Comparison of Table 25 with Table 23 shows th at the Zvalues would be more realistic for the new Sp vs. MPD Equation (10). PAGE 56 43 CHAPTER FIVE CONCLUSIONS A comprehensive analytical study was perfor med to investigate th e applicability of the ASTM IFI computational procedure for standardization of fr iction measurements from different CFMEs. The following conclu sions can be reached based on the above investigation. (a) The investigation revealed that the fricti on value adjusted to 60 km/hr slip speed (FR60) based on measurements at 65 km/hr and 95 km/hr (FR65 and FR95) differed consistently for all CFMEs and r unways. A significance test conducted in this research showed that FR60 from FR65 and FR95 are signi ficantly different from each other. Therefore the calibration constants A and B would vary with the testing speed. These results also show that the ASTM calibration is speed dependent. (b) When 10 runway surfaces were selected out of 14 for calibration and the remaining 4 surfaces left for validation, the results showed that the predicted friction values obtained from the testi ng speed of 65 km/hr were more accurate than those from the testing speed of 95 km/hr. (c) ASTM standard procedure advocates the use of single A and B calibration constants for a given equipment which re sults in a single F60 prediction causing an uncertainty in the actua l friction value on the surface. This uncertainty can be addressed by treating A and B as random variables and predicting the F60 as a random variable within a cer tain level of confidence. (d) Since the random variation of the calibration constants are represented by normal distributions, the error (with respect to the corresponding DFT reading) in each PAGE 57 44 validation trial can be represented by a Zvalue. The Z values for some surfaces show excessive magnitudes indicating a poor agreement. (e) Inaccurate prediction of IFI (with respect to DFT) for some surfaces (item d) and the dependency of FR60 on the testing sp eed (item a) can be attributed to the overly simplified IFI computation protocol laid out in the ASTM IFI standard. (f) One critical simplification could be invol ved in the general linear Sp vs. MPD relationship that appears to have been derived to suit a multitude of equipment and a number of test surfaces. This relationship is equipment dependent and its inapplicability to particular equipment was amply demonstrated in this study by back computing a number of alternative Sp vs. MPD relationships. PAGE 58 45 REFERENCES 1. ASTM E196003: Standard Practice for Calculating International Friction Index of a Pavement Surface 2. Thomas J.Yager, NASA Langley Re search Center. AN OVERVIEW OF THE JOINT FAA/NASA AIRCRAFT/GROUND VEHICLE RUNWAY FRICTION PROGRAM 3. NASA Wallops Tire / Runway Friction workshop Data, May 2005 4. NASA Wallops Tire / Runway Friction workshop Data, May 2006 5. NASA Wallops Tire / Runway Friction workshop Data, May 2007 PAGE 59 46 APPENDICES PAGE 60 47 Appendix A: Data Analysis Results for FAA RFT Table 29 FAA RFT Calibration Constants Speed 65 km/hr 95 km/hr DFT VT Japan PTI VT Japan PTI A 0.155 0.167 0.189 0.179 0.196 0.204 B 0.523 0.498 0.454 0.544 0.506 0.484 R2 0.84 0.8 0.7 0.74 0.7 0.64 Table 30 Correlation between Computed F60 and DFT F60 FAA RFT DFT VT Japan PTI Operating Speed 65 km/hr 95 km/hr 65 km/hr 95 km/hr 65 km/hr 95 km/hr Correlation between F60 (DFT) and F60 (Computed) 0.50 0.391 0.6342 0.4921 0.5584 0.4363 PAGE 61 48 Appendix A: (Continued) Table 31 Z Values for FAA RFT at 65 km/hr and 95 km/hr with Respect to DFTVT, Japan and PTI NASA GT ZValues DFT VT JAPAN PTI Operating Speed 65 km/hr 95 km/hr65 km/hr95 km/hr65 km/hr 95 km/hr A 2.734 3.32808 2.44291 2.90385 1.15533 2.32171 B 3.26694 2.15378 3.50566 2.11505 4.87454 2.6224 C 3.89403 2.34326 4.36602 2.32699 2.95866 1.96661 D 4.32473 4.43086 2.16007 2.89684 3.07658 3.53971 E 9.780506 6.1089749.1434635.18277310.32233 5.871776 F 7.336648 9.2039256.1077 8.19009 6.385201 8.448261 G 0.818629 2.6574981.5600323.3377861.571269 3.407478 Echo1 8.425362 5.4104878.8923585.4815268.698641 5.523046 EK1 3.11808 4.56925 3.08823 4.58592 3.05973 4.61073 EK2 2.070625 0.5950641.8350770.4701051.317826 0.108849 R4 13.4441 8.16776 12.6637 7.78365 13.3833 7.86875 Echo2 5.154237 5.6321484.9323315.7793014.273046 5.31191 EK3 0.545695 0.0549141.46867 1.0365511.2722 1.03746 EK4 0.428963 0.80615 1.1875 1.69694 0.226473 0.82307 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 2200385Ka 4500 controlfield tag 001 002047320 005 20100113145756.0 007 cr bnuuuuuu 008 100113s2008 flu s 000 0 eng d datafield ind1 8 ind2 024 subfield code a E14SFE0002780 035 (OCoLC)497899622 040 FHM c FHM 049 FHMM 090 TA145 (Online) 1 100 Kavuri, Kranthi. 0 245 Investigation of the validity of the ASTM standard for computation of International Friction Index h [electronic resource] / by Kranthi Kavuri. 260 [Tampa, Fla] : b University of South Florida, 2008. 500 Title from PDF of title page. Document formatted into pages; contains 48 pages. 502 Thesis (M.S.C.E.)University of South Florida, 2008. 504 Includes bibliographical references. 516 Text (Electronic thesis) in PDF format. 520 ABSTRACT: Runway friction testing is performed in order to enhance the safety of aircraft operation on runways. Preventative maintenance friction surveys are performed to determine if there is any deterioration of the frictional resistance on the surface over a period of time and to determine if there is a need for corrective maintenance. In addition operational performance friction surveys are performed to determine frictional properties of a pavement surface in order to provide corrective action information in maintaining safe takeoff or landing performance limits. A major issue encountered in both types of friction evaluation on runways is the standardization of the friction measurements from different Continuous Friction Measuring Equipment (CFME). The International Friction Index (IFI) has been formulated to address the above issue and determine the friction condition of a given runway is a standardized format.The ASTM recommended standard procedure to compute the IFI of a runway surface employs two distinct parameters to express the IFI; F60 is the friction value adjusted to a slip speed of 60 km/h and correlated to the standard Dynamic Friction Tester (DFT) measurement. And Sp is the speed constant which is governed by the mean profile depth of that surface. The primary objective of this thesis is to investigate the reliability of the current ASTM procedure to standardize runway friction measurements in terms of IFI. Based on the ASTM standard procedure, two equipment specific calibration constants (A and B) are assigned for each CFME during calibration. Then, in subsequent testing those calibrations constants can be used to adjust the equipment measurements to reliable IFI values. Just as much as A and B are presumed to be characteristic of any given CFME, they are also expected to be independent of the operational speed.The main objective of the annual NASA Runway Friction Workshop held in Wallops Island, Virginia, is to calibrate commonly used CFMEs such that all calibrated equipment would provide a standard reading (i.e. IFI) on a particular surface. During validation of the existing ASTM procedure using the NASA Runway Friction Workshop data it was observed that the single valuebased IFI predictions of the calibrated CFMEs were inaccurate resulting in low correlations with DFT measured values. Therefore, a landing pilot should not be left to make a safe decision with such an uncertain single standard friction value because the actual standard friction value could very well be much less than this value. Hence a modified procedure was formulated to treat the calibration constants A and B as normally distributed random variables even for the same CFME. The new procedure can be used to predict the IFI (F60) of a given runway surface within a desired confidence interval.Since the modified procedure predicts a range of IFI for a given runway surface within two bounds, a landing pilot's decision would be made easier based on his/her experience on critical IFI values. However, even the validation of the modified procedure presented some difficulties since the DFT measurements on a few validated surfaces plotted completely outside the range of F60 predicted by the modified method. Furthermore, although the ASTM standard stipulates the IFI (F60) predictions to be independent of the testing speed, data from the NASA Runway Friction Workshop indicates a significant difference in the predictions from the two testing speeds of 65 km/hr and 95 km/hr, with the results from the 65 km/hr tests yielding better correlations with the corresponding DFT measurements. The above anomaly could be attributed to the significantly different FR60 values obtained when the 65 km/hr data (FR65) and 95 km/hr data (FR95) are adjusted to a slip speed of 60 km/hr.Extended analytical investigations revealed that the expected testing speed independency of the FR60 for a particular CFME cannot be supported by the ASTM defined general linear relationship between Sp and the mean profile depth which probably has been formulated to satisfy a multitude of CFMEs operating on a number of selected test surfaces. This very reason can also be attributed to the above mentioned outliers observed during the validation of the modified procedure. 538 Mode of access: World Wide Web. System requirements: World Wide Web browser and PDF reader. 590 Advisor: Manjriker Gunaratne, Ph.D. 653 Runway friction testing Calibration constants Random variable Slip speed Speed constant 690 Dissertations, Academic z USF x Civil and Environmental Engineering Masters. 773 t USF Electronic Theses and Dissertations. 4 856 u http://digital.lib.usf.edu/?e14.2780 