A STUDY OF RUTTING OF ALABAMA ASPHALT PAVEMENTS

Transcription

1 A STUDY OF RUTTNG OF ALABAMA ASPHALT PAVEMENTS Final Reprt Prject Number ST by Frazier Parker, Jr. E. Ray Brwn Auburn University Highway Research Center Auburn University, Alabama spnsred by The State f Alabama Highway Department Mntgmery, Alabama August 199

2 The cntents f this reprt reflect the views f the 'authrs wh are respnsible fr the facts and accuracy f the data presented herein. The cntents d nt necessarily reflect the fficial views r plicies f the State f Alabama Highway Department r Auburn University. This reprt des nt cnstitute a standard, specificatin, r regulatin. 2

3 ABSTRACT Pavement rutting is the accumulatin f permanent defrmatin in all r a prtin f the layers in a pavement structure that results in a distrted pavement surface. The verall bjecive f this study was t develp recmmendatins fr mre rut resistant asphalt cncrete mixtures which cmprise the uppermst layers f flexible pavements. T accmplish prject bjectives a plan f study was cnducted that included 1) an analysis f rutting data frm the Alabama Highway Departments pavement cnditin data base, 2) a field evaluatin and sampling prgram at thirteen test sites, 3) a labratry testing prgram, and 4) analyses f data frm the field and labratry testing prgrams. The analysis f the pavement cnditin data base indicated that rutting is increasing and that rutting susceptibility varies gegraphically because f the variable quality f lcally available aggregate. Careful cntrl f crushing f gravel and the develpment f a test t quantify and limit particle shape and texture f fine aggregate were identified as means fr imprving aggregate quality. n mst cases permanent defrmatin appeared t be cmbined t the tp three r fur inches f asphalt aggregate layers, thereby, implicating high tire pressures as the primary causative factr. A rate f rutting f 2 x 1-4 in/--iesal r 1. x 1-7in/ESAL delineated gd and pr perfrming pavements. Mix and aggregate prperties that appeared t be related t rutting ir'clude: layer thickness, vids, GS, gyratry rller pressure, percent fractured faces, percent passing N. 2 sieve and creep strain. The crrelatin f these prperties was nt very strng, but in almst every case was caused by ne r tw pints far utside the range f the bulk f the data. This illustrates the cmplexity f the rutting prcess and the necessity f cnsidering a number f prperties during material selectin and mix design. 3

4 TABLE OF CONTENTS ntrductin... 5 Objectives... 8 Scpe... 9 Plan f Study... 1 Presentatin and Analysis f Results Cnclusins and Recmmendatins References Tables Figures Appendix A, Rutting Data frm Pavement Cnditin Databases Data Srted by AHD Divisin Data Srted by Mix Type Frequency Distributins Appendix B, Layer Prfiles frm Field Test Sites

5 NTRODUCTON Pavement rutting is the accumulatin f permanent defrmatin in all r a prtin f the layers in a pavement structure that results in a distrted pavement surface. Lngitudinal variability in the magnitude f rutting causes rughness. Water may becme trapped in ruts resulting in reduced skid resistance, increased ptential fr hydrplaning and spray that reduces visibility. Prgressin f rutting can lead t cracking and eventually cmplete disintegratin. Flexible pavement rutting is nt a new prblem. As lng as flexible pavements have been used, rutting has been recgnized as a primary distress mechanism and a primary design cnsideratin. Fr high fltatin tires rutting may be cnfined t weaker materials such as subbases and subgrade sils. n the nt t distant past, the cnsensus was that rutting was generally restricted t subgrades. n fact, the 1986 AASHTO Guide fr Design f Pavement structures is based n perfrmance mdels develped at the AASHO Rad Test where tire inflatin pressures were nminally 8 psi. Asphalt pavement design cncepts are based n prviding sufficient pavement structure (rutting resistant materials) t reduce stresses in the subgrade t the point where rutting will nt develp, and n prviding asphalt quality and thickness t resist fatigue cracking. What is new regarding flexible pavement rutting is the awareness that permanent defrmatin in the high quality asphalt layers (surface, binder and base) has becme a significant cntributr t pavement rutting. N where is this mre vividly demnstrated than in rutting f asphalt cncrete verlays f Prtland cement cncrete pavements. Repetitive applicatins f heavy trucks with increasingly high pressure tires drives rut frmatin in high quality asphalt layers. The stresses induced in 5

6 near surface layers by the high pressure tires exceed the ability f the materials t resist densificatin belw critical vids (4%) and subsequent plastic flw. Recent studies (1-5) have shwn that truck tire inflatin pressures, and therefre cntact pressures, have increased dramatically frm 8 psi n which design prcedures are cmmnly based. Average truck tire inflatin pressures fr radial tires are nw arund 1 psi. This means that a significant prtin f truck tires have inflatin pressures higher than 1 psi, ften in the 13 t 14 psi range. The study by Marshek, Chen, Cnnell and Hudsn (6) ntes several additinal prblems with high tire pressures. A cmmnly made assumptin has been that cntact pressure apprximately equals inflatin pressure. The study shwed that increased tire pressure prduced prprtinally smaller grss cntact areas. This suggests that the cmmnly made assumptin f equal pressure becmes increasingly less valid. The study als shwed that cntact pressures were nt unifrm. This suggests sme cntact areas with pressures greater than a unifrm pressure based n grss cntact area. Extrardinarily high tire pressures mean that asphalt cncrete layers which are f the highest quality but nearest t the surface in a pavement structure are nt immune t rutting. Althugh recent mdificatins such as asphalt cntent selectin based n 75 blw Marshall cmpactin have increased rutting. resistance, material quality prvided by existing specificatin ccasinally is insufficient t meet the demands f tday's traffic. Assuming that truck vlume, lading and tire pressures are nt likely t decrease; the bvius slutin is t increase asphalt cncrete resistance t permanent defrmatin. As with mst simple and bvius slutins it must, hwever, be apprached with cautin. Beneficial changes in ne prperty may lead t detrimental changes in anther prperty. Fr example, decreasing asphalt 6

7 cntent will result in increased rutting resistance, but decreased fatigue resistance. ncreased asphalt cement viscsity will result in a stiffer mix, which may be mre resistant t rutting, but a mix that is mre likely t crack (traffic and envirnmental frces). n Alabama the slutin is cmplicated by available materials. Lcal natural sand and gravel are used extensively, particularly in the suthern and western prtin f the state. These materials generally have runded particle shape which is detrimental t mix stability. Even when gravel is crushed, the larger particles that result are likely t have at least ne uncrushed face since the maximum particle size f available natural gravel is arund 1-1/2 inches. The slutin t the rutting prblem in Alabama will likely nt be bvius r simple but will require careful cnsideratin and treatment f the entire prblem. Hwever, a slutin, r even small imprvemnts, are ptentially enrmusly Significant; cnsidering the dminant rle asphalt cncrete will play in rehabilitating and upgrading the states ver 1, miles f radway. Of particular significance will be the requirements f the apprximately 9 miles f interstate pavement which carries a disprprtinately large vlume f truck traffic. 7

8 OBJECTVES The verall bjective f this study was t develp recmmendatins fr mre rut resistant asphalt cncrete mixtures. T accmplish this verall bjective the fllwing five sub-bjectives were: T determine the nature and extent f rutting n Alabama Highways, T cnduct testing and evaluatin f typical asphalt cncrete mixtures, T characterize mixtures that are susceptible t rutting and thse that are nt susceptible t rutting, T review Alabama Highway Department (AHD) material selectin and asphalt cncrete mix design prcedures, and T frmulate recmmended mdificatins t current practices t enhance resistance f asphalt cncrete mixes t permanent defrmatin. 8

9 SCOPE The assessment f the nature and extent f rutting f asphalt pavements in Alabama was limited t an analysis f data frm the 1984, 86, and 88 Alabama Highway Department pavement cnditin data bases. These data bases are part f the verall pavement management system being implemented by the Alabama Highway Department. Testing and sampling was cnducted at thirteen (13) test sites. Sites were selected t prvide materials with a range f rutting resistance. Test pits were dug thrugh and cres taken f all asphalt-bund layers. Labratry testing f materials frm the cres measured prperties t characterize in-situ mix, recmpacted mix, recvered aggregate and recvered asphalt cement. Effrts t imprve rutting perfrmance f Alabama asphalt pavements were fcused n cnsistant material selectin and mix design fr asphalt cncrete. Cnsideratin f structural pavement thickness design and cnstructin aspects were beynd the scpe f the study. 9

10 PLAN OF STUDY T accmplish prject bjectives a three (3) phase study was executed. The three phases cnsisted f 1) an analysis f rutting data frm the Alabama Highway Department's pavement cnditin data base, 2) field evaluatin and sampling at thirteen test sites, and 3) a prgram f labratry testing. Data frm these studies was analyzed t develp a mdel fr describing the rutting prcess and t frmulate recmmendatins fr imprving the rutting resistance f asphalt cncrete mixes. Rutting Data frm Pavement Cnditin Database Cnditin and traffic data fr the Alabama state and interstate system are cllected frm representative 2' lng test sectins in each lane mile f pavement. Data cllectin prcedures are described in reference 7. Data cllectin was initiated by the Department in 1984 and has been repeated in 1986 and Data recrds fr each lane mile cntain identifying and descriptive infrmatin, quantitative pavement cnditin infrmatin (including rutting), pavement ratings develped frm cnditin data, present serviceability indices (PS) develped frm rughness measurement, and estimated traffic data. Eight rut depth measurements, fur in the uter wheel path and fur i'n the inner wheel path, are taken in 2' test sectins fr each lane mile f pavement. A fur ft lng straight edge is placed acrss the wheel path and the maximum rut depth measured. Fr this study these eight measurements were averaged and used t represent the rutting fr each lane mile f pavement. Reference is made in the remainder f the text t average rut depth. Measurements were made in all lanes, but rut depths in uter lanes were always larger and were used exclusively in the analyses. n the analyses, estimates f the traffic applied t the pavement were in -- terms f ttal numbers f 18 kip equivalent single axle lads (ESAL). Eighteen 1

11 kip ESAL's were cmputed fr tw and fur lane radways using equatins 1 and 2, respectively. ESAL = (AADT)(CV)(.5)(Age)(.82) ESAL = (AADT)(CV)(.5)(Age)(.82)(.85)..... (1)... (2) where AADT = Annual average daily traffic CV = Percent cmmercial vehicles as decimel.5 = Directinal split Age = [(year + mnth rated) - (year + mnth built)] = Average cnversin factr frm reference 8 = Percent traffic in uter lane fr 4-lane radway The annual average daily traffic and percent cmmercial vehicles used in the cmputatin f ESAL's were the estimated values fr the year in questin. Traffic was assumed split evenly by directin (SO/5). The pavement age, in days,. was the difference between the date rated and the date the last surface layer was placed. Fr lder pavements and pavements with significant traffic grwth, this prcedure will have ver estimated ttal applied traffic. Hwever, since mst f the cmparisns were relative this is nt cnsidered a serius prblem that wuld warrant a mre accurate estimatin f traffic. T determine the nature and extent f rutting three parameters were analyzed. These were average rut depth, mean (average rut depth/esal) and mean average rut depth/mean ESAL. The ratis f average rut depth t ESAL and mean average rut depth t mean ESAL prvide indicatrs f the rate f rut frmatin with traffic. Cmparisns f the three variables were made between Highway Department Divisins t detrmine if gegraphical differences exist, and between surface mix types. Data fr these cmparisns were gruped accrding 1 1

12 t radway type (state rutes, interstate rutes and cmbined) fr the 1984, 86 and 88 data bases. The data was als cmbined fr verall cmparisns. Frequency analyses f average rut depth and (average rut depth)/(esal) were made t determine if the distributin f rutting is changing and if rutting severity is a functin f radway type (state r interstate rutes). Field Evaluatin and Sampling Thirteen test sites were selected fr evaluatin and testing. The apprximate lcatin f these sites is shwn n Figure 1. Sites were selected t prvide a relatively unifrm statewide gegraphic distributin. Sites were als selected t prvide examples in the Piedmnt and Applachian Plateau gelgic regins where crushed stne is available, and in the Castal Plain gelgic regin where natural sands and gravels are the predminate aggregate used in ht mix asphalt. The sites were selected t prvide a range f rutting perfrmance. Five (5) sites were cnsidered by Highway Department persnnel t have prvided gd rutting perfrmance and eight (8) t have prvided pr rutting perfrmance. The develpment f rutting was investigated by cutting a trench and cring thrugh all asphalt bund layers. All test sites were n fur lane facilities and the trenches and cres were taken acrss utside lanes nly. A typical sampling layut is shwn in Figure 2. Tw lines f 12 cres each, ne f 4-inch cres and ne f 4 and 6-inch cres, were cut. Pavement surface prfiles at the trench and at cre lines were btained by measuring the distance frm a leveled 12 ft lng straightedge as illustrated in Figure 3a. Once the trench was cut, similar measurements were made t layer interfaces, as illustrated in Figures 3b and 3c, t btain a cmplete prfile fr all the asphalt-bund layers. Layer thicknesses were measured frm cres (Figure 3d) and added t surface prfile measurements as a secnd methd fr 12

13 develping cmplete layer prfiles. Cres were then used t prvide material fr the labratry testing prgram described in the fllwing sectin. Labratry Testing There are a number f labratry tests that have been used t predict rutting in asphalt pavements. These tests cannt be used individually in all cases t identify mixtures with a tendancy t rut under traffic, but all the tests cmbined have been shwn t be a fairly gd indicatr f rutting. Tests that will be used in this study t crrelate with rutting include: vids in place, vids in labratry cmpacted samples, stability and flw f labratry cmpacted samples, aggregate gradatin, fractured face cunt f carse aggregates, particle shape and texture f fine aggregate, gyratry shear index (GS), gyratry rller pressure, creep, resilient mdulus, and asphalt cement prperties. n Place Prperties. Past studies have shwn that in-place air vids are related t rutting (9,1,11). Mst data indicates that nce the air vids decrease t 3 percent r lwer plastic flw is likely t ccur. The in-place vids alne may nt be a gd predictr f rutting since sme mixes with lw vids d nt rut and sme mixes with higher vids d rut. t has als been shwn that in-place vids may decrease t a pint and then increase again as rutting prgresses. This increase f in-place vids during traffic makes it difficult t crrelate rutting with in-place vids. Fr instance rutting may have been caused by lw in-place vids but measurement f in-place vids may shw higher vids if the vids increase when rutting begins. This prcess will be cnsidered later during develpment f the mdel t describe rutting. Apprximately 24 cres were taken at each test, as illustrated in Figure 2. A number f the cres were cmbined and brken-up fr tests such as Rice specific gravity, gradatin, asphalt cntent, and recmpacting. The bulk density 13

14 f cres were then cmpared t the measured Rice specific gravity t determine in-place vids. Prperties f Labratry Cmpacted Samples. Samples f asphalt mixture were cmbined, brken-up, heated and recmpacted t evaluate prperties f labratry cmpacted samples. The prperties f labratry cmpacted samples is an indicatr f riginal mix design prperties. The Marshall stability will always be higher than the riginal mix design because the asphalt cement is nw stiffer due t xidatin, hwever, the vids and flw appear t be apprximately equal t thse measured during mix design. The use f vids f recmpacted mix des have sme advantage ver use f in-place vids. Fr instance the in-place vids will vary acrss the traffic lane due t traffic and the in-place vids may actually begin t increase as rutting ccurs. n-place vids are als a functin f riginal density and traffic. Thus, in-place vids are dynamic (changing) and will vary depending r many factrs making it difficult t predict perfrmance. Recmpacted vids n the ther hand are mre cnsistent and shuld be representative f the final in-place vids after Significant traffic. Hence, recmpacted vids may be a better indicatr f rutting ptential than the in-place vids. Samples were recmpacted in tw ways: hand hammer and gyratry. Seventy-five blws per side with the hand hammer has been shwn t prvide a density apprximately equal t that after traffic fr high vlme rads. Wrk with the gyratry testing machine (GTM) has shwn that it prvides a density apprximately equal t that f the hand hammer when it is set at 12 psi (apprximately equal t truck tire pressure), 1 degree angle, and gyrated fr 3 revlutins. Tw additinal prperties f the asphalt mixture that are evaluated when using the GTM t recmpact samples include gyratry shear index (GS) and rller 14

15 pressure (a measure f shear strength). Past studies have shwn the GS t be related t rutting. Mixes with a GS equal t apprximately ne tend t be resistant t rutting while mixes with a GS abve abut 1.3 tend t rut severely under traffic. Since the rller pressure is a rugh measure f shear strength it ptentially culd relate t rutting. Extracted Aggregate Prperties. Mix frm cres was separated int aggregate and asphalt cement cmpnents. Gradatin, fractured face cunts n carse aggregate particles, and fine aggregate particle shape and texture tests were cnducted n extracted aggregate. The aggregate gradatin f a mixture is imprtant but there is disagreement abut the gradatin that shuld be used. Mst peple agree that the aggregate gradatin shuld be apprximately parallel t the maximum density curve, but shuld be ffset t prvide sufficient VMA. Sme agencies specify that the gradatin be abve the maximum density curve while thers specify that it be belw the curve. t is difficult t quantify and evaluate an aggregate gradatin since 8-1 sieve sizes are nrmally used. n this study three sieve sizes will be used t evaluate gradatin: 3/8 inch, N. 5, and N. 2. These three sieve sizes shuld be sufficient t accurately charaterize the aggregates gradatin. Studies have shwn that the maximum aggregate size is imprtant fr rut resistant pavements. Larger maximum aggregate size mixtures are mre resistant t rutting while finer mixtures tend t rut mre. Mixtures with excessively high r lw amunts f material passing the N. 2 sieve are mre likely t rut. Mixtures with high natural sand cntent are als mre likely t rut. The three selected sieve sizes shuld be sufficient t evaluate maximum aggregate size, percent passing N. 2, and amunt f natural sand. 15

16 t has lng been knwn that crushed aggregates prduce mixes that are mre resistant t rutting than mixes with natural aggregate that has mre runded particles. t is nt clear hwever what the ptimum fractured face cunt shuld be t insure gd perfrmance. AHD 1989 specificatins have crushed particle requirements n minus 3/4" t plus N.8 material when gravel is used in binder mixes and a requirement that 8% f the plus N.4 particles used in wearing curses have tw r mre crushed faces. The prblem with runded aggregate can ccur when gravels r natural sands are used since bth f these may be the desired size withut crushing. The fractured face cunt was cnducted n the plus N.8 material. The percent f aggregate particles with zer, ne, r tw r mre crushed faces were recrded. The material smaller than the N.8 culd nt be tested since it was t small t evaluate the number f fractured faces. Resilient mdulus has becme ppular in recent years t characterize the stiffness prperties f asphalt cncrete. The test t measure resilient mdulus is repeated lad indirect tensile. Since resilient mdulus is a tensile test it shuld be affected by changes in asphalt cement prperties, but it is dubtful if it can be crrelated t rutting since this is the result f shear strain in an asphalt mixture. Tests were cnducted n cres cut frm the pavement. The test was perfrmed by applying apprximately 15% f the tensile strength fr.1 secnds and allwing the asphalt mixture t recver fr.9 secnds. Tests were cnducted at 4,77, and 14 F t establish temperature sensitivity. t is generally believe that the creep test has ptential t mdel rutting f asphalt mixes. The creep test is typically cnducted at varius temperatures. Cmpressin lads can be applied statically r repetitively t uncnfined r cnfined specimens. The mst cmmn mde is the static uncnfined mde, 16

17 because it is the easiest and simplest. Hwever, this mde prbably des the wrst jb f simulating what actually happens in the field. Tw majr prblem with the static uncnfined test is that the temperature cannt be as high as 14 F (typical in-place temperatures) and the nrmal pressure must be much lwer than 12 psi (typical truck tire pressures). High temperature r high pressures in uncnfined tests will result in unrepresentative failure f the sample. Fr this study all tests were cnducted in the static cnfined mde. Tests were cnducted at a pressure f 12 psi and a temperature f 14 F. A cnfining pressure f 2 psi was used. When a particular asphalt layer was being investigated that was less than tw inches thick, tw r mre cres were stacked t insure that a ttal thickness f at least tw inches was btained. When cres were stacked, cement mrtar was applied between the cres t prvide prper seating. The shape and texture f fine aggregate particles is thught t be as imprtant as the shape f carse aggregate particles. A test fr particle shape and texture f minus N.8 material prpsed by the Natinal Aggregates Assciatin (12) was cnducted. Uncmpacted vids f a graded samples (Methd A) were measured. n additin, 4 gm samples f ungraded minus N.8 material were run thrugh the apparatus and flw times recrded. These mdified tests were perfrmed because the uncmpacted vids frm Methd A did nt prvide a wide range f values. t was felt that flw time might prvide a wider range f values and better crrelatin with rutting susceptibility, but it did nt. Extracted Asphalt Cement Prperties. The aggregate must supprt the lad if rutting is prevented in asphalt cncrete, hwever, the asphalt cement prperties can als affect perfrmance. f the asphalt mixture has a slightly high asphalt cntent, then the asphalt cement prperties can greatly affect rate 17

18 f rutting. Asphalt cement was recvered frm cres and tested fr viscsity and penetratin. Viscsity testing was cnducted at 14 and 275 F, and penetratin testi ng at 77 and 4 F. 18

19 PRESENTATON AND ANALYSS OF RESULTS Data and analyses f these data frm the three phases f the study described in the preceding sectin are presented individually in this sectin. These analyses are then cmbined and a mdel prpsed fr describing rutting f asphalt-bund layers f flexible highway pavements. Analysis f Rutting Data frm Pavement Cnditin Database. The rutting data in the 1984, 1986, and 1988 pavement cnditin data bases were analyzed t assess the nature and extent f the rutting prblem in Alabama. As nted earlier, the data will be gruped and cmpared t islate the effects f varius parameters. Cmbined Data. Table 1 cntains a summary f all data frm the 1984, 1986, and 1988 databases. The data is gruped accrding t radway type (state rutes, interstate rutes and cmbined). Clumn 2 cntains the frequency which is indicative f the number f lane miles f pavement. Clumn 3 cntains the mean f average rut depths, Clumn 4 the rati f the m,ean average rut depth t mean EASL, and clum 5 the mean f the average rut depth t ESAL rati. Values frm Table 1 are pltted in Figure 4. Frm this figure the fllwing bservatins can be made: Mean rut depths are larger n interstate rutes than n state rutes. This is likely due t the larger traffic vlumes n interstate rutes. The average rut depth has increased frm 1984 t The average rut depth increase, frm 1984 t 1988, is larger fr interstate rutes (.1494 in.) than state rutes (.36 in.). Because f the large frequency fr state rutes, the average rut depth fr state rute and cmbined data is similar. Because f the small frequency fr interstate rutes, the average rut depth relatinship is mre erratic, i.e., the level f the veraly prgram 19

20 can have an bservable influence n mean rut depth. Values f the ratis f the means (b) are cnsiderably different than values f means f the ratis (c). This is due t the large numeric differences between numeratrs (average rut depth) and denminatrs (ESAL) and the wide range f ESAL values. Values shwn fr bth parameters are ratis multiplied by 1 7. Because f the large influence f extreme values f ESAL's, the rati f the means is cnsidered a better indicatr f rate f rut develpment. Bth ratis indicate that the rate f rut develpment is increasing. Bth ratis indicate that the rate f rut develpment is much greater n state rutes than interstate rutes. This is likely due t higher quality pavements (including quality f asphalt bund materials) n the interstate system. The ratis f the means (b) shw abut the same increase in rate f rut develpment, frm 1984 t 1988, fr state rutes (.188) and fr interstate rutes (.166). The mean f the ratis (c) shw a much larger increase in rate f rut develpment, frm 1984 t 1988, fr state rutes (1.859) than fr interstate rutes (-.52). T summarize, all but ne f the parameters examined indicated that rutting is increasing. Fr rut depth this culd be caused by an increase in pavement rutting susceptibility, an increase in traffic vlume r an increase in lading severity (truck weight and/r tire pressure). Fr rate f rut develpment, pssible causes wuld be restricted t rutting susceptibility and lading severity. Highway Department Divisin. Cmparisns by Highway Department Divisins were made fr mean rut depth, the rati f mean rut depth t ESAL's 2

21 and the mean f the rati f average rut depth t ESAL's. These cmparisns were made t determine if gegraphical variatins in rutting exists and t examine pssible reasns fr these variatins. Speculatin was that gelgy and, thus, the availability f variable quality aggregates might be a factr. As shwn in Figure 1, Divisins 1, 3 and 4 are predminately in the Piedmnt and Appalachian Plateau gelgic prvinces. Rck depsits in these areas are used fr crushed stne and are the surce f sand and gravel materials. Divisin 2 is divided between the Appalachian Plateau and the Castal Plain regin. Divisins 5-9 lie belw the Fall Line in the Castal Plain regin. Natural sands and gravels are available and are the predminate aggregate materials used in this regin. The degree f weathering and, thus, particle size and shape f sand and gravel is influenced by the distance transprted frm the surce material. Particles becme mre runded and smaller as the transprtated distances increases. mplicatins are that aggregate quality and, therefre, mix rutting susceptibility increases with mvement suthward as the distance frm the rck surce increases. Fr natural sand (fine aggregate) and uncrushed gravel (carse aggregate) in asphalt bund materials, the influence f particle size and shape is straight frward and well established. Hwever, when gravel is crushed, the influence f gravel size is nt as direct r as well dcumented. Carse aggregate fr surface and binder mixes is required t have sme crushed particles. Therefre, natural gravel requires crushing. The prblem created by using gravels is that the degree f particle fracturing will be directly related t riginal particle size. Smaller gravel particles will be less fractured and mixes cntaining these partially crushed particles will be mre susceptible t rutting. f indeed gelgy and, thus, gegraphy is a factr, rutting susceptibility shuld be less in Divisins 1-4 than in Divisins 7-9. Divisins 5 and 6 shuld be 21

22 intermediate. Figure 5 shws three histgrams which illustrate the variability f rutting susceptibility between divisins. Frm the three histgrams, it can be seen that the rut depth, rati f mean rut depth t mean ESAL's and mean f the rati f average rut depth t ESAL's are less fr Divisins 1-4 than fr Divisins 5-9. The averages f the three parameters fr Divisins 1-4 are.738 in, inches x 1-7 and 8.51 inches x 1-7. Fr divisins 5-9 the averages are inches, 2.75 inches x 1-7 and inches x 1-7. The data shwn in Figure 5 is frm the 1984 database fr cmbined state and interstate rutes. Similar histgrams were pltted fr 1986 and 1988 databases fr state, interstate and cmbined rutes. A cmplete set f these histgrams is cntained in Appendix A. Averages frm these histgrams are shwn in Table 2, and cnfirm the trends illustrated in Figure 6 fr the 1984 cmbined data, Le., that pavements in Divsins 5-9 are mre susceptible-t rutting thqn thse in Divisins 1-4. The mst cnsistant indicatr f this trend is the rati f means which is an indicatr f rate f rut develpment. Rut depth and the mean f the rati f rut depth t ESAL's is mre sensitive t pavement age. Between 1986 and 1988 the relatinship between average rut depth and the mean f the rati fr cmbined and state rute data reversed. As can be seen in Table 2, the average 1988 rut depths n state and cmbined rutes fr Divisins 1-4 are abut equal thse in Divisins 5-9. The means f the ratis fr 1988 becme larger in Divisins 1-4. This reversal in trend is thught t be primarily due t a reversal in values fr Divisins 4 and 5. Figure 6 is similar t Figure 5, but fr 1988 data. Cmparing Figures 5 and 6, it can be seen that the mean rut depths and ratis becme'larger fr Divisin 4 in 1988; which is ppsite what was bserved in 1984 and The number f ESAL's applied t the pavement is an indicatr f pavement age and, therefre, f 22

23 the verlay prgram. n 1986 the average number f ESAL's n cmbined state and interstate rutes in Divisin 4 was 36, and in 1988, 365,. The trend in Divisin 5 was just the ppsite, with applied ESAL's ging frm 45, in 1986 t 384, in This reversal in applied traffic (average pavement age) fr Divisins 4 and 5 is thught t be primarily respnsible fr the reversal in trends. Despite the exceptins nted abve, the analyses f the data supprt the cntentin that rutting susceptibility is related t gegraphic lcatin. n additin, gelgy and, thus, prperties f available aggregate prvide a lgical explanatin fr the bserved relatinship between rutting susceptibility and gegraphic lcatin. This phenmenn will be examined further in the analyses f the data frm field test sites. Mix Type. Data was gruped by existing surface mix type. Similar mix types were cmbined t btain five mix type grups identified as 41 (surface treatment), 411,416,417 (latex) and 42 (pen graded). Analyses were perfrmed fr the three parameters used previusly, Le., average rut depth, mean rut depth/mean ESAL and mean (average rut depth/esal). A cmplete set f histrgrams develped t study the influence f mix type n rutting susceptibility is cntained in Appendix A. nterstate rutes are surfaced with primarily 416 mix with sme pen graded prus frictin curse (42) mix. A minimal amunt f 417 mix shws up in 1988 but n 41 r 411 mix is used. Figure 7 shws 1988 interstate data, which is typical fr 1984 and Average rut depths are abut the same fr pavments with 416 and 42 surface mixes. The relatinship between the rate f rutting, as indicated by the rati f means, fr the mixes varies frm year t year. Hwever, the variatin is small (minimum f.434 x 1-7 inches t maximum f.672 x 1-7 inches) indicating that the rate f rutting is similar 23

24 fr bth mixes The mean f the average rut depth t ESAL rati varies cnsiderably, but the value fr 42 mix is always cnsiderably smaller than 416 mix as shwn in Figure 7. The large difference in age f surfaces with the tw mixes is thught t cause this difference. N 42 mix has been placed recently, and badly rutted sectins are verlaid. This results in a decline in the amunt f 42 mix with nly thse pavements perfrming well increasing in age. Pavements are verlaid with 416 mix, thereby, decreasing r maintaining its age. Since the mean f the ratis is sensitive t extreme values f rut depth r ESAL's (indicative f age), the large differences in values is reasnable, but prbably nt a valid indicatr f the mixes relative rutting susceptibility. Analyses f all mixes is best accmplished by cmparing data fr state rutes, cnsidering that state rutes are surfaced with all type mixes. Figure 8 shws hw usage and rate f rutting fr the varius mixes is changing with time. Figure 89 shws that the usage f 41, 411 and 42 mix is decreasing, and that the usage f 416 and 417 mix is increasing. Fr 41, 411 and 416 mix, this reflects the upgrading f state rutes t satisfy increased traffic demands. Use f 417 mix remains small, but its increased usage is, at least partially, mtivated by the belief that latex imprves rut resistance. The rate f rutting is increasing fr 41 and 416 mixes, but appears t be stabilizing fr 411 mix. Traffic lading severity is an bvius reasn fr this increase, but as discussed fr interstate rutes, changes in usage patterns will influence relative age which may in turn affect the validity f rate f rutting indicatrs. The limited use and age f 417 mix prevents drawing strng cnclusins regarding its perfrmance relative t 416 mix. Hwever, several interesting trends are apparent frm Figure 8. ts use is increasing rapidly, and its rate f 24

25 rutting is apprximately twice that f 416 mix. The relatinship between the rate f rutting f the mixes may be a result f the limited use f 417 mix. As its use and age increases, its rate f rutting shuld becme mre cmparable with 416 mix. Figure 9 shws 1986 data gruped by mix fr all three parameters. The histgrams fr mean rut depth and mean rut depth/mean ESAL are typical f 1984 and The data fr the mean f the rati f average rut depth t ESAL is mre erratic, particularly fr the 1988 data. Again, this is thught t be caused by extreme values and adversly affects the parameter as an indicatr f rutting susceptibility. The rati f means is a mre valid indicatr f rate f rutting. Figure 9 shws that mean rut depths fr 41,411 and 416 mixes are abut the same. This indicates cnsistency in the design f materials fr the traffic lading intensities that these mixes are expected t withstand. Rate f rutting, as measured by the rati f means, indicates that the rate f rutting fr 411 and 416 mixes are abut the same. Again this indicates cnsistency in material design fr expected traffic. Rate f rutting fr surface treatments (41) is higher. This is expected since rutting will develp in base, subbase, and pssibly the subgrade in these pavements. Heavy truck lads and high tire pressures will be critical fr pavements with surface treatments since they prvide nly. minimal cver fr base and subbase layers. Bases and subbases fr these type pavements are usually cmprised f unbund sil aggregate type materials which will nt be particularly rutting resistant. Rutting Frequency Distributins. Frequency distributins f average rut depth and rate f rutting, as measured by the rati f average rut depth t 18 kip ESAL, were analyzed fr trends with time and pavement type. A cmplete set f frequency distriputins are cntained in Appendix A. 25

26 Figures 1 and 11 shw, respectively, 1988 state and interstate rut depth frequency distributins and 1988 state and interstate rate f rutting frequency distributins. The shapes illustrated are typical fr 1984 and 1986 data. The rut depth distributins, Figure 1, shw, as did Figure 4a, that rut depths are larger n interstate pavements. The rate f rutting distributins, Figure 11, shw, as did Figure 4b, that the rate f rutting is smaller n interstate pavements. Figure 11 als shws a much wider range f rate f rutting n state rute pavements. This is expected because f the greater diversity f materials, pavement structures and traffic n state rutes. Figures 12 and 13 shw, respectively, rut depth and rate f rutting frequency distributins fr 1984, 86, and 88 state rute data. The trends illustrated fr state rutes were generally applicable fr interstate rutes. Figure 12 shws a change in shape fr the rut depth distributin between 1984 and This appears t be an indicatin that the magnitude f rutting was significantly increasing, but cmparisn f the cumulative frequencies and percentages fr 1984 and 86 reveal nly minr changes. The shape f the 1988 distributin reverts back t the 1984 shape. Figure 13 shws cnsistent shapes and nly minr changes in frequency fr rate f rutting. Althugh significant changes with time in rut depth and rate f rutting are nt visually apparent frm the frequency distributins, mean values did increase with time as demnstrated in Figures 4a and c. Analysis f Field Data. Data frm the thirteen (13) field sites was analyzed t determine where in the pavement structure permanent defrmatin was develping and the relatinship between rut develpment and traffic. The lcatins f the 13 test sites are shwn in Figure 1 and descriptins f the pavement structures given in Table 3. 26

27 Layer Prfile Analysis. The prfiles f the asphalt bund layers were analyzed t determine where rutting was develping. When trenches were pened, string lines were stretched alng layer interfaces, as illustrated in Figure 13, t defect depressins in the lwer layer surface. These depressins wuld be indicative f permanent defrmatin in the layer itself r lwer layers. As can be seen frm the pavement structure descriptins in Table 3, nine f the 13 pavements were cmprised f an riginal structure plus at least ne verlay. This made the determinatin f the lwer limit f rutting mre difficult, but measurements in the trenches indicated that permanent defrmatin was primarily cnfined t near surface (apprximately 4 inch depth) asphalt bund layers. n mst pavements this limited permanent defrmatin t surface and binder layers. The interface between binder and black base layers were usually relatively depressin free. At nly Site 9 was there evidence f rutting in base r subbase layers belw asphalt bund layers. There was evidence f rutting in r belw the sand-clay-shell base at this site. When this rutting ccurred culd nt be determined. t may have develped in the riginal pavement when cver was nly the "AKG" treatment and apprximately ne inch f asphalt cncrete. Or, it may have develped later when the structure was thicker, but lads and tire pressures greater. At nly Site 2 was there evidence that stripping may have cntributed t rutting. At this site several cres in wheel paths disintegrated and culd nt be cmpletely recvered. Stripping was cnfined t the riginal binder and base layer. Prfiles were als analyzed t determine the magnitude f rutting and where in the pavement structure permanent defrmatin was ccurring. This analysis is illustrated in Figure 14. Prfiles fr all thirteen sites are cntained in Appendix B. Ttal rutting was determined by averaging rut depths (Ri and R) 27

28 in inner and uter wheel paths at the trench and tw cre line lcatins. Rut depths are cmpiled in Table 4. Als shwn in Table 4 are rut depths fr the test sites cmpiled frm the 1988 pavement cnditin database. Database rut depths are smaller primarily because f different measuring methds. A 4-ft straightedge was used fr data base measurements and a 12-ft straightedge was used at test sites. Data base rut depths are the average f eight measurements per lane mile fr design prjects in which test sites were lcated. Design prjects were several miles lng. The rut depths frm Table 4 are pltted as a histgram in Figure 15. Frm this histgram the fllwing can be nted: Rut depths at sites selected fr gd rutting perfrmance are generally less than rut depths at sites selected fr pr perfrmance. Rut depths (12 ft straight edge) are generally less than.4 inches fr sites with gd perfrmance and greater than.4 inchs fr sites with pr perfrmance. Rut depths (4 ft straight edge) are generally less than.2 inches fr sites with gd perfrmance and greater than.3 inches fr sites with pr perfrmance. While rut depth is an indicatr f pavement perfrmance, it is influenced by traffic (vlume and lad) which must be cnsidered when assessing rutting susceptibility. The effects f traffic will be cnsidered in the fllwing sectin. Permanent defrmatin in varius layers was determined analyzing the shape f layer interfaces. As indicated by stringlining layer interfaces in trenches, the permanent defrmatin ccurred primarily in upper layers. As illustrated in Figure 14, strait edges alng binder/base interfaces indicated 28

29 minimal rutting in lwer layers (the exceptin being Site 9). T get sme idea f the permanent defrmatin in the varius layers, layer thickness in wheel paths (T W1 and T W2 ) were cmpared with layer thicknesses utside the wheel paths (T1' T2 and T3)' The summatin f the permanent defrmatin in all layers (TO - T W) shuld apprximate the ttal rutting (R). This analysis was nt successful fr several reasns. Accuracy f measurements was likely ne reasn, but mre imprtantly was inapprpriatness f the methd. Overlays create tw prblems. When rutted pavements are verlaid (withut milling), layers in the wheel paths will be thicker, and nt as well cmpacted. Secndly, permanent defrmatin in the existing pavement will nt be related t rutting f the verlay. Finally, at thse sites where plastic flw has ccurred, upheaval utside the wheelpaths will distrt the measurements. This prcess will be examined mre clsely in the sectin n the mdel fr rutting, but material simply mves frm the wheel path t adjacent areas giving a false impressin f layer thickness. Sme dilatin may als ccur in cases f severe rutting, causing further distrtin. T summarize, analysis f the layer prfiles prduced gd measures f ttal rutting and gd qualitative indicatins that permanent defrmatin was limited t near surface layers (surface and binder). Hwever, the anlaysis t quantify permanent defrmatin in individual layers was nt successful. Ttal rutting will be cmbined with traffic and analyzed in the fllwing sectin. Rutting vs Traffic Analysis. T study the relatinship between traffic and rutting at test sites, traffic was cnverted t ttal18 kip ESAL's applied t the pavement since cnstructin r since verlay. Equatin 2 was used t cnvert MDT and percent cmmercial vehicles t 18 kip ESAL's. Traffic data frm the 1986 data base with n grwth factrs was used fr this purpse. Since pavements were cnstructed frm 1974 t 1985 and als rated in 1988 and 29

30 1989, cmputed 18 kip ESAL's are apprximatins. Traffic data fr the thirteen test sites is cmpiled in Table 5. All test sites were n uter lanes f fur lane facilities and the 18 kip ESAL's are estimates fr these lanes. Traffic vlumes ranged frm.3 x 1 6 ESAL's at Site 12 t 6.6 x 1 6 ESAL's at Site 1. This represents a 22 fld difference and must be cnsidered when evaluating the influence f traffic n rutting. The mdel that will be subsequently prpsed t describe the rutting prcess with traffic is highly nnlinear and the rate f rutting develpment will be dependent n lcatin f traffic vs rutting alng the curve. The rati f rut depth t 18 kip ESAL's prvides a measure f rate f rutting. Using rut depths frm measurements at the test sites and frm the 1988 pavement cnditin database, ratis were cmputed and cmpiled in Table 4. These ratis are pltted as a histgram in Figure 16. Except fr Site 13, the histgram prvides a clear distinctin between the gd and pr perfrming pavements. The histgram als suggests a 1. x 1-7 inch/esal rate f rutting criteria fr delineating rutting and nnrutting pavement. Rate f rutting will be used with labratry data in the fllwing sectin t develp crrelatins with aggregate, asphalt and mix prperties. General Mdel fr Rutting in Asphalt Bund Layers Observatin f pavement crss sectins at test sites and examinatin f in-place mix prperties indicates rutting in asphalt pavements develps in tw phases. This prcess is graphically depicted in Figure 17. n the first phase repeated lad applicatins causes densificatin frm as cnstructed vid cntent (8% r less). n prperly designed mixes, densificatin stabilizes at abut 4% and in gd perfrming pavements, rut depth develpment ceases r decreases t very lw rates as illustrated in Figure 17a. Mst mixes are designed t have apprximately 4/ vids, but are nrmally 3

31 cmpacted during cnstructin t 7-8% vids. After cnstructin, the pavement surface shuld be flat and free f ruts. Traffic will cntinue t cmpact a well designed mix t the 4% design vids. Vids may stabilize at higher vids, but if much higher durability prblems may develp. The additinal cmpactin will result in small ruts. Fr example, an.8 inch rut will develp in a 2-inch thick layer with a 4% reductin in vids. At abut 4% vids, the ability t resist permanent defrmatin in prperly designed mixes is ptimum. t is critical that the aggregate skeletal structure have the ability t resist further densificatin. This is best accmplished with well graded aggregate with angular rugh textured particles. Asphalt cntent is als critical as the mix reaches abut 4% vids. Excess asphalt will decrease intergranular cntacts weakening the aggregate skeletal structure and leading t further densificatin. Excess asphalt can weaken therwise very stable aggregate structures. This emphasizes that aggregate prperties and ptimum asphalt cntent are equally imprtant aspects f the mix design and cnstructin prcess. Fr pavements that experience severe rutting, densificatin cntinues and secnd phase cnditins develp. When vids reach 2-3%, the mix becmes very unstable and plastic flw will develp with cntinued traffic, as illustrated in Figure 17b. Rut depth increases rapidly and upheaval utside wheel paths will begin. Carried t extremes, pushing and shving may develp causing a dramatic increase in rughness. Dilatin may ccur as the material shears and flws plastically frm whe"elpaths causing an apparent increase in vids. Large vids, Figure 18, were visible in cres adjacent t wheel paths at Sites 5 and 6 where advanced secnd phase cnditins had develped. At sites selected fr gd rutting perfrmance (1,4,7, 11 and 13), nly first phase densificatin had ccurred and vid cntent was stable as illustrated 31

32 in Figure 17a. Vids in the surface layer within wheel paths at these sites were 5.,4.7,4.1,9.4 and 3.2%, respectively. At the remainder f the sites, selected fr pr perfrmance, rutting was at several stages f develpment, as illustrated in Figure 17b. Site 12 had received nly a small amunt f traffic and was cnsidered at the beginning f first phase densificatin (vids were 6.8%). Sites 2, 3 and 1 were still experiencing first phase densificatin, with vids f 3.,2.1 and 3.7%, respectively, but appeared abut t g int secnd phase plastic flw. Site 5 and 6 were well int secnd phase plastic flw with vids f 2.3 and 2.1 %, respectively. t is f interest t nte here that the last verly at Sites 5 and 6, as well as Site 2, was thin ( b/sy) asphalt cncrete ver a su,rface treatment. Asphalt cement frm the surface treatment, particularly fr heavy applicatins, may migrate upward, sftening the asphalt cncrete and cntributing t rutting. Site 8, where traffic vlume was lw, appeared t be int the initial stages f secnd phase plastic flw develpment. The vid cntent f the surface layer in the wheel paths was 1 %, but very high asphalt cntent (7.8 % ) is thught t have cntributed t this lw value. Vid cntent at Site 9 was 3.8%, but n cnclusins culd be drawn regarding the stage f rutting develpment because f evidence that lwer layers might als be cntributing. A prperly designed and cnstructed asphalt-aggregate mixture will 'have 7-8% vids after cnstructin. t will slwly cmpact t apprximately 4% vids and stabilize. An imprperly designed mix, ne that will result in rutting, will usually initially have vids abve 4-5%, but will cmpact under traffic t 2-3% vids. Since the prpsed mdel fr describint rut develpment with traffic is nnlinear, ratis f rut depth t several functins f ESAL's were examined. The rati f rut depth t 18 kip ESAL's seemed t prvide the best measure f rate 32

33 f rutting. Using rut depths frm measurements at the test sites ratis were cmputed and cmpiled in Table 4. These ratis are pltted as a histgram in Figure 19. The histgram prvides a clear distinctin between the gd and pr perfrming pavements. The histgram als suggests a.2 x 1-3 inch/esal rate f rutting criteria fr delineating rutting and nnrutting pavement. Analysis f Labratry Data. After cmpletin f all labratry testing a detailed statistical analysis using SAS prgram was perfrmed t determine thse prperties that are related t rutting. n-place mix prperties, Table 6 included asphalt cntent, vids, resilient mdulus and creep strain. Prperties f recmpacted mix, Table 7, included vids, stability and flw f samples cmpacted with a manual Marshall hammer and with a gyratry testing machine. During cmpactin in the gyratry, rller pressure was measured and gyratry shear index (GS) was cmputed. Prperties f recvered asphalt, Table 8, included penetratin and viscsity. Prperties f recvered aggregate, Table 9, included gradatin, fractured face cunts n carse aggregate particles, and uncmpacted vids and flw time fr fine aggregate fractins. T be useful a mdel must include rut depth and traffic. Three relatinships were cnisdered: (Rut Depth)/ESAL, (Rut Depth)/ESAL, and (Rut Depth)/n ESAL. t was determined that (Rut Depth)/ESAL was the parameter that crrelated best with labratry prperties. Crrelatin cefficients frm the linear regressin between (Rut Depth)/ESAL and varius parameters are tabulated in Table 1. A crrelatin cefficient clse t 1 indicates a gd crrelatin and a crrelatin cefficient clse t a indicates a pr crrelatin. Since mst f the rutting that was bserved ccurred in the tp fur inches and generally in the tp layer, the analysis was made cnsidering nly the prperties f the tp layer. The crrelatin cefficient between 33

34 (Rut Depth)/-v'ESAL and thickness was high enugh (-.59) t indicate that the thickness f the tp layer was an imprtant factr fr evaluating rutting ptential. The relatinship between tp layer thickness and (Rut Depth)/-v'ESAL is shwn in Figure 2. A discussin f the results fr varius prperties that affect rutting is presented belw. The results are based n samples thrughut the State f Alabama and may nt be apprpriate in ther states r even in Alabama when materials, thicknesses, envirnment, traffic, and ther factrs are different than thse analyzed in this study. n-place Vids. t has been knwn that rutting is a functin f in-place vids. Brwn and Crss (9), Frd (1), and Huber and Herman (11) shwed that nce in-place vids drp belw apprximately 3 percent rutting is likely t ccur. Table 1 shws that the crrelatin cefficient fr in-place vids and (Rut Depth)/-v'ESAL is The vids in wheel paths are usually lwer due t cmpactin, hence, it appears that these lwer in-place vids shuld be cmpared t rutting. Sme prjects, hwever, shw that the lwer vids are in between the wheel paths. One explanatin fr the cause f this is that the vids in the wheel path may decrease t sme minimum amunt at which rutting ccurs. Once rutting begins t ccur it is likely in sme cases that the density in the wheel path actuatly decreases due t plastic flw resulting in an increase in vids. The analysis in this study was made by first cmputing the average cre density and the standard deviatin. The critical in-place vids were then calculated at the 8th percentile. n ther wrds 8 percent f the vids wuld be higher than the selected value and 2 percent wuld be lwer. This is an acceptable minimum vid level fr cmparing t rutting. Six pavements had in-place 8 percentile vids belw 3 percent in the tp 34

35 layer. These pavements were at sites 2,3,5,6,8, and 9. These six sites alng with site 1 had the highest {Rud Depth)/-VESAL values and hence rut at a faster rate. Site 1 had very lw in-place vids (1.6 percent) in the secnd layer which likely explains why it had a high rate f rutting. Althugh the in-place vids are clsely related t rutting there is n way t use this infrmatin in the initial mix design and cnstructin cntrl f asphalt mixtures. The in-place vids can nly be measured after the mixture has been placed which makes this prperty useless fr mix design and quality cntrl. Labratry cmpactive effrt has been calibrated in the past t prvide a density equal t the in-place density after traffic. f this crrelatin is crrect then the in-place density can be predicted with labratry cmpacted samples. Figure 21 graphically shws the effect f in-place vids n rut depth. This figure shws a general trend f increasing rut depth with decreasing in-place vids. n place vids near fur percent and higher typically result in a (Rut Depth)/-VESAL f apprximately.2 r less. This means that the expected rut depth fr these mixes after 1 millin ESALs wuld be n mre than.2 inches and after 4 millin ESALs, it wuld be n mre than.4 inches. Resilient Mdulus (MR)' The crrelatin cefficient between MR and (Rut Depth/-vESAL was determined t be -.65 (Table 1). This is a relatively high crrelatin and shws that an increase in MR shuld result in a decrease in rut depth. The data in Figure 22 des shw a definite trend. Since the MR was cnducted n field samples, it is likely that the mixes with higher vids aged mre rapidly than ther mixes and thus, resulted in higher MR values. Since MR changes with age f the asphalt mix, it wuld be impssible frm this study t determine minimum MR values t specify fr new cnstructin. Creep. N crrelatin analysis was perfrmed fr the creep strain data because six f 13 samples failed during testing (Table 6). The remaining 35

36 seven samples had measurable creep strains ranging frm.15 t.39. Five f the six samples that failed during testing had{rut Depth)/v'ESAL greater than.2, but nly fur f the samples with measurable creep strain had {Rut Depth)/v'ESAL greater than.2. Creep strain in general identifies mixes that are susceptible t rutting, and samples that defrm excessively when laded with 12 psi cmpressin at 14 F are particularly unstable and susceptible. Recmpacted Vids (Hand Hammer). Sme f the mix taken frm the in-place pavement was heated, brken up and recmpacted using 75 blws with the Marshall hand hammer. This prcess shuld prvide an estimate f the riginal labratry cmpacted mix prperties. Table 1 shws that the crrelatin cefficient between recmpacted vids and (Rut Depth)/v'ESAL is This is nt as high as the crrelatin fr in-place vids but is still a reasnable crrelatin. Figure 24 shws that there is cnsiderable scatter, but there is a general trend fr lwer rut depth with higher vids. Fr example tw f the fur pavements with (Rut Depth)/v'ESAL less than.2 had vids abve fur percent. Only ne f the remaining nine samples with (Rut Depth)/-vESAL greater than.2 had vids abve fur percent. While recmpacted vids d nt relate as well with rutting as in-place vids, it des relate well enugh t be effective in minimizing rutting. f labratry vids are fur percent r higher, rutting shuld nt be a majr prblem prvided all ther prperties are acceptable. Recmpacted Vids (Gyratry Testing Machine). Samples recmpacted in the gyratry prvide similar results as thse recmpacted with hand hammer. The crrelatin cefficient fr samples cmpacted in the GTM is -.57 which is slightly better than that fr hand hammer. Figure 25 shws that 36

37 there is again cnsiderable scatter in the data, but it des indicate that vids d affect rutting. Seven f the nine pavements with (Rut Depth)..JESAL greater than.2 had recmpacted vids less than fur percent. Three f the fur pavements with (Rut Depth)..JESAL less than.2 had recmpacted vids greater than fur percent Marshall Stability (Recmpacted with Hand Hammer). The crrelatin between (Rut Depth)..JESAL and Marshall stability f samples recmpacted with a hand hammer is very high (crrelatin cefficient = -.74). This indicates that an increase in stability will result in a decrease in rutting. This crrelatin is smewhat surprising since mst pavement engineers believe that Marshall stability is nt related t rutting. The actual Marshall stability values measured are generally very high since the asphalt has xidized and stiffened. Even if Marshall stability is clsely related t rutting it wuld be difficult t establish stability requirements frm the data in this reprt since the stability values reprted are much higher than that which wuld be expected fr new mix. The relatinship between rutting and stability fr hand hammer cmpactin is shwn in Figure 26. There is a definite trend shwing a reductin in rutting fr an increase in stability. There are at least tw explanatins fr the strng crrelatin between stability and rutting even if a gd crrelatin des nt exist. t is knwn that mixes with high vids will xidize faster than mixes with lw vids. Hence, it is likely that mixes with high vids (mre rut resistant) xidized mre and artifically increased the Marshall stability. Hence, it is reasnable t expect mixes with high vids t have high stability and mixes with lw vids have lw stability. A statistical evaluatin f the data shws that the crrelatin cefficient between Marshall stability (hand cmpactin) and in-place vids is 37

38 .7 and Marshall stability (gyratry cmpactin) and in-place vids is.67. These high crrelatins between vids and stability likely explains part f the reasn that rutting and stability appear t be clsely crrelated. Anther explanatin fr the gd crrelatin is the way sites were selected fr this prject. Thse sites with mre rutting had generally been in-place a shrter time than thse sites with little rutting. Everything else being equal, this wuld result in the least rutted pavements having higher stbility that the mre rutted pavements simply because f pavement age. The lder pavements being mre xidized wuld have higher stabilities. Cnsidering the abve discussin it is still likely that there exist sme crrelatin between stability and rutting. Marshall Stability (Recmpacted with Gyratry). The crrelatin cefficient between the stability (gyratry cmpacted) and (Rut Depth)/--JESAL is Again this is a high degree f crrelatin. The explanatin prvided fr stability (hand hammer cmpactin) als applies t the stability measured here. A plt f the data is shwn in Figure 27. Again there is a definite trend that indicates an increase in stability results in a decrease in rutting. Marshall Flw (Hand Cmpacted). Flw is cnsidered a reasnably gd indicatr f rutting. The crrelatin cefficient between Marshall flw (hand cmpacted) and rutting fr this study was measured t be.5 (Table 1). A flw f 16 is specified by mst agencies as the maximum allwable flw. Figure 28 shws that fur f the five mixes with a flw abve 16 had (Rut Depth/--JESAL greater than.2. Five f the mixes with flw 16 r less had (Rut Depth)/--JESAL greater than.2. Only tw f the seven mixes with flw 16 r less had (Rut Depth)/--JESAL greater than.3. Again, there is a trend which indicates that mixes with a flw abve 16 are mre likely t rut. 38

39 Marshall Flw (Gyratry Cmpacted). The crrelatin cefficient between Rut Depth/"ESAL and Marshall Flw (Gyratry Cmpacted) is.41 (Table 1). These results are pltted n Figure 29. Fur f the five mixes with a flw greater than 16 have a (Rut Depth)/"ESAL greater than.2. Only fur f eight mixes with flw less than sixteen have (Rut Depth)/"ESAL greater than.2. Gyratry Shear ndex (GS). The GS has been shwn t be a gd indicatr f rutting (9). As shwn in Table 1 the crrelatin cefficient between GS and Rut Depth/"ESAL was.64. Figure 3 shws that there is significant scatter abut the best fit line. The data that plts well abve the best fit line generally has lw fractured face cunt and the data belw the line generally has a high fractured face cunt. Much data is gruped next t GS = 1: since this is the lwest value that a mix can have. Previus studies have shwn that mixes with a GS greater than 1.3 are expected t exhibit severe rutting (9). The data shws that 4 f the 5 mixes with GS abve 1.3 had experienced a Rut Depth/"ESAL greater than.2. Gyratry Rller Pressure. The crrelatin cefficient between rller pressure and (Rut Depth)/"ESAL is -.61 (Table 1). The rller pressure is that frce required t prduce a 1 degree gyratin angle in the asphalt mix. A mix that is mre resistant t defrmatin shuld require a higher pressure t defrm it during the cmpactin prgress. Figure 31 shws that there is cnsiderable scatter abut the best fit line but there is a definite trend. Lwer rller pressure is typical fr thse mixtures that have high rutting. Seven f the eight mixes with a rller pressure f 14 psi r less had Rut Depth/"ESAL greater than.2. Three f the five mixes with a rller pressure greater than 14 psi had Rut Depth/"ESAL less than.2. Aggregate Gradatin. The aggregate gradatin definitely affects the 39

40 rutting resistance f an asphalt mixture but this is a difficult prperty t analyze. Studies have shwn that the maximum aggregate size is imprtant as well as percent passing N. 2 sieve are imprtant (13,14). Hwever, the verall evaluatin f individual gradatins is difficult. Fr this prject the percent passing 3/8 inch sieve, percent passing N. 5, and percent passing N. 2 sieve were analyzed t determine their affect n rutting. As shwn in Table 1 the crrelatin cefficient between Rut Depth/"ESAL and percent passing the 3/8 inch sieve is This indicates that an increase in percent passing the 3/8 inch sieve will decrease rutting. This is ppsite frm the expected trend. A plt f the data in Figure 32 shws that there is cnsiderable scatter in the data with nly limited range. The high crrelatin cefficient is basically the result f ne data pint that has a very lw percent passing the 3/8 inch sieve and very high rutting. Based n the findings by thers and the data scatter it is cncluded that clear trend between Rut Depth/"ESAL and percent passing the 3/8 inch sieve is nt shwn in this study. The secnd sieve size that was investigated was the perent passing the N. 5 sieve. The crrelatin cefficient f.17 (Table 1) and the data scatter shwn in Figure 33 indicate very little trend between Rut Depth/"ESAL and percent passing N. 5 sieve. The crrelatin cefficient f.37 and the data scatter in Figure 34. indicate a pr crrelatin between (Rut Depth)/"ESAL. Figure 28 des shw, hwever, that the tw mst severely rutted pavements had mre than seven percent passing the N. 2 sieve. Fractured Faces. The fractured face cunt f an aggregate shuld affect its ability t resist rutting. Sme percentage f fractured aggregate is almst always specified fr high vlume rads but there is very little field data t supprt r cntradict this type specificatin. The crrelatin cefficient 4

41 between fractured face cunt and (Rut Depth)/ -vesal fr the study was -.13 (Table 1). This is a very lw crrelatin that shws a slight trend tward less rutting fr higher fractured face cunt. The crrelatin appears t be much better than this after reviewing Figure 35. The tw mixes with highest (Rut Depth)/-vESAL (Site 5 = 77.1 x 1-5 and Site 8 = 79.2 x 1-5 ) als had high fractured face cunts (Site 5 = 81.% and Site 8 = 98.1 %). These mixes had lw in-place vids (Site 5 = 2.3% and Site 8 =.6%). Site 5 was the mst severely rutted site studied (Rut Depth = 1.9 inches) with rutting well int plastic flw. Plastic flw had nt started at Site 8, but the mix was characterized by verh high asphalt cntent (7.8%) and very lw in-place vids (.6%). f the data fr Site 8 is eliminated, fr unrealistically high asphalt cntent, the crrelatin cefficient becmes -.41 indicating a much strnger trend. The data in Figure 35 shws that all six mixes with fractured face percentages f 8 r less had a (Rut Depth/ -vesal greater than.2. The data als shws that fur ut f seven mixes with a fractured face cunt greater than 8 had (Rut Depth/ -V ESAL less than.2, including the tw mixes discussed abve. Fine Aggregate Shape & Texture. Uncmpacted vids frm the NAA flw test (12) and time frm the mdified test measure particle angularity and' texture. Higher vids and flw times indicate rugher textured and mre angular particles. The crrelatin cefficients in Table 1 shws that flw time frm the mdified NAA test has very little crrelatin (.5). Uncmpacted vids frm the NAA test have better (.15), but still very pr crrelatin. Figure 36 shws the weak trend fr uncmpacted vids, but the trend indicates that an increase in vids will result in an increase in rutting. This is ppsite f the expected trend. t appears frm Figure 36 that the data frm 41

42 Sites 5 and 8, which have the tw highest rates f rutting, d nt fllw the pattern f the data at ther sites. f the data pint fr Site 8 (45.9, 79.2 x 1-5) is mitted, fr unrealistically high asphalt cntent, the crrelatin cefficient becmes -.25 indicating a strnger trend. Mre imprtantly the sign f the crrelatin cefficient is reversed and indicates, as expected, that rate f rutting decreases as uncmpacted vids increases. Figure 37 illustrates the very weak crrelatin fr flw time. Again, if the data pint fr Site 8 (23.9, 79.2 x 1-5) is mitted, the crrelatin cefficient becmes This nt nly represents a dramatic increase in magnitude, but the change in sign means that the trend is in the expected directin, i.e., rate f rutting decreases as flw time increases. The perfrmance f the mix at Site 8 demnstrates the multiplicity f factrs that can influence rutting perfrmance, and the imprtance f bth aggregate prperties and asphalt cntent during material selectin and mix design. Asphalt Penetratin. The data in Table 1 shws that the crrelatin cefficient between (Rut Depth/..JESAL and penetratin is.46. The data pltted in Figure 38 shws an bvius trend indicating an increase in penetratin wuld result in an increase in rutting. Since mst asphalt pavements in Alabama begin with similar penetratin, it is nt clear what this trend indicates. As befre, it may be that larger vids result in mre xidatin f the asphalt and better' resistance t rutting. At any rate it is reasnable t expect mre rutting when using an asphalt with higher penetratin. Viscsity. The crrelatin cefficient between (Rut Depth)...jESAL and viscsity is -.5 as shwn in Table 1. The trend is shwn graphically in Figure 39. This indicates that an increase in viscsity wuld result in a decrease in Rutting. The discussin under asphalt penetratin will als be true fr viscsity. Predictive MOdel. n develping a cmbined predictive mdel, thse 42

43 prperties that were independent f each ther that appeared t crrelate best with rutting and thse easily measured were selected. After evaluatin f several cmbinatins it was determined that the best cmbinatin included the fllwing three prperties: vids in labratry cmpacted samples, percent f fractured faces, and percent f material passing the N. 2 sieve. This mdel which has an R2 f.35 is shwn in Figure 4. The equatin can be used t estimate (Rut Depth/"ESAL) frm the three aggregate and mix prperties. Mixes resulting in estimates greater than.2 shuld be examined carefully fr redesign. Gegraphic and Aggregate Prperty Relatinships. n the sectin n the analysis f rutting data frm the AHD pavement cnditin database, it was cncluded that pavement rutting susceptibility was related t gegraphical area and that variable gelgy and, thus, variable quality aggregate was the mst prbable cause. Specifically it was cncluded that pavements in Divisins 5-9 which are lcated in the Castal Plain, where natural sands and gravels are used, are mre susceptible t rutting than pavements in Divisins 1-4 which are lcated predminately in the Appalachian Plateau and Piedmnt gelgic regins, where crushed stne is available. Particle shape and texture are assumed t be indicatrs f aggregate quality and several parameters will be cmpared fr the extracted aggregates frm Sites 1-4 and 5-6. Average values fr the parameters are cntained in Table 11. Fr the fine aggregate fractin (minus N.8, vids frm the NAA test and flw time frm the mdified NAA test were cmpared. Average NAA vids and flw times are larger fr the sites in Divisins 1-4 indicating mre angular particles with rugher surface texture. This was the expected respnse since bth angularity and surface texture are indirectly related t weathering r distance transprted. 43

44 Crushed face cunts are a measure f particle angularity and were made n carse (plus N.8) aggregate particles. These cunts can als be used t indicate angularity f carser fractins, i.e., plus N.4 particles. The plus N.4 fractin is cnsidered because AHD specificatins fr surface mix aggregate have the requirement that 8% f these particles have tw (2) r mre crushed faces. Percentages f particles having tw r mre crushed faces fr the plus N.8 and plus N.4 fractins are shwn in Table 11. The numbers are smewhat different but bth sizes indicate the same differences between the aggregate frm sites in Divisins 1-4 and the aggregate frm sites in Divisins 5-9. Fr surface mixes the percentages f particles with tw r mre crushed faces is nly slightly larger in Divisins 1-4 than in Divisins 5-6. This is prbably due t the widespread use f crushed gravel carse aggregate statewide with nly slightly mre granite and slag used in Divisins 1-4. Fr binder mixes the percentages in Divisins 1-4 are 1 % and apprximately twice thse in Divisins 5-6. This si due t the sidespread use f crushed limestne in Divisins 1-4. Fr cmbined mixes the percentages fr bth size fractins are larger in Divisins 1-4 than in Divisins 5-6, indicating mre angular curse aggregate particles in Divisins

45 CONCLUSONS AND RECOMMENDATONS Analyses f the Department's pavement cnditin database indicate that rutting in Alabama is increasing, and that this increase is attributible t either increased lading intensity r increased asphalt cncrete rutting susceptibility. The analyses als indicate that rutting varies gegraphically and that this variatin can be explained by quality f lcally available aggregate. Thse areas with crushed stne and angular natural sands are less susceptible t rutting. Analyses f data frm field test sites indicate that permanent defrmatin causing rutting is generally cnfined t the tp 3 t 4 inches (surface and binder curses). There was little evidence that lwer base/subbase curses r subgrade were significant cntributrs t rutting. At nly ne site was there evidence that stripping may have cntributed t rutting. There was sme evidence that surface treatment layers used in cnjunctin with thin verlays may have cntributed t rutting susceptibility. A rate f rutting f 2 x 1-4 in/..jesal r 1. x 1-7 in/esal delineated gd and pr perfrming pavements. The prperties measured in this study that appear t be useful in minimizing rutting include: layer thickness, vids, GS, Gyratry rller pressure, percent f fractured faces, percent passing N. 2 sieve, Marshall flw, and creep strain. The R2 fr mst f these prperties were lw, hwever, there appeared t be a definite trend shwn in the figures. The lw R2 in almst every case was caused by ne r tw f the data pints being far utside the range f the ther data. This shws that rutting is a very cmplicated prcess and is affected by a large number f factrs and hence, t use nly a small number f prperties t accurately predict rutting is impssible. The best mdel selected fr predicting rutting had an R2 equal t.35 and included the vids in labratry cmpacted samples, percent carse aggregate with tw r mre faces, and percent f materials pasisng the N. 2 sieve. 45

46 The crushing f gravel shuld be mre carefully cntrlled t insure that the 8/ requirement fr particles with tw fully fractured faces is met. This may require limitatins n minimum particle size fr crushing. On heavily traveled radways (state primary and interstate rutes) crushed particle requirements fr binder mixes shuld be the same as surface mixes. The use f a test, such as the Natinal Aggregate Assciatin's uncmpacted vids methd, t quantify and limit particle shape and texture f fine aggregate shuld recieve additinal study but this study shwed n crrelatin. The use f surface treatment interlayers, particularly in cnjunctin with thin verlays, shuld receive further study. This study shuld fcus 1) n identifying cnditins where surface treatment interlayers shuld and shuld nt be used, and 2) n cnstructin cntrl prcedures that will prevent excess asphalt cement that culd sften the verlay and increase rutting susceptibility. Prcedures shuld be adpted fr better cntrl f asphalt cntent during cnstructin. Target jb mix frmula asphalt cntent shuld nt be changed withut sufficient test results t justify changes. 46

47 References Middletn, D.R., F.L. Rberts and T. Chira-Chavala, "Measurement and Analysis f Truck Tire Pressures n Texas Highways," in Transprtatin Research Recrd 17, Transprtatin Research Bard, Washingtn, D.C., 1986, pp Thmpsn, M.R., "Analytical Methds fr Cnsidering Tire Pressure Effects in Pavement Design," Prceedings, in A SympsiumlWrkshp n High Pressure Truck Tires, American Assciatin f State Highway and Transprtatin Officials and Federal Highway Administratin, Austin, Texas, February Kim, O. and C.A. Bell, "Measurement and Analysis f Truck Tire Pressures in Oregn," in Transprtatin Research Recrd 127, Transprtatin Research Bard, Washingtn, D.C., 1988, pp Hudsn, S.W. and S.B. Seeds, "Evaluatin f ncreased Pavement Lading and Tire Pressures," in Transprtatin Research Recrd 127, Transprtatin Research Bard, Washingtn, D.C., 1988, pp ;--_-;-, "High-Pressure Truck Tires," in Pavement Newsletter, Federal Highway Administratin, Washingtn, D.C., January Marshek, K.M., H.H. Chen, R.B. Cnnell and W.R. Hudsn, "Experimental Determinatin f Pressure Distributin f Truck Tire Pavement Cntact," in Transprtatin Research Recrd 17, Transprtatin Research Bard, Washingtn, D.C., 1986, pp =--:-:----: _, "Pavement Cnditin Survey, Distress dentificatin and Rating Manual," Bureau f Materials and Tests, Alabama Highway Department, Mntgmery, Alabama, July ;--=-:==-, "Truck Weights as Related t Pavement Design in Alabama," HPR N. 96, Prject N , Bureau f Materials and Tests, Alabama Highway Department, Mntgmery, Alabama, June Brwn, E.R. and Stephen A. Crss, "A Study f n-place Rutting f Asphalt Pavements," Prceedings, Assciatin f Asphalt Paving Technlgi.sts, Vlume 58, 1989, pp Frd, M.C., "Pavement Densificatin Related t Asphalt Mix Characteristics," paper presented at 67th Annual Transprtatin Research Bard Meeting, January Huber, G.A. and G.H. Herman, "Effect f Asphalt Cncrete Parameters n Rutting Perfrmance: A Field nvestigatin,' Prceedings, Assciatin f Asphalt Paving Technlgists, Vlume 56, 1987, pp Meininger, R.C., "Prpsed Methd f Test fr Particle Shape and Texture f Fine Aggregate Using Uncmpacted Vid Cntent," Natinal Aggregates Assciatin, Silver Springs, Maryland, March

48 Brwn, E.R., J.L. McRae and A. Crawley, "Effect f Aggregate n Perfrmance f Bituminus Cncrete," mplicatin f Aggregate in the Design, Cnstructin, and Perfrmance f Flexible Pavement, ASTM Special Technical Publicatin 116, Schreuders and Marek, ASTM, 1989, pp Brwn, E.R. and Charles E. Bassett, "The Effects f Maximum Aggregate Size n Rutting Ptential and ther Prperties f Asphalt-Aggregate Mixture," Paper presented at TRB Meeting, January

49 TABLES 49.

50 TABLE 1. SUMMARY RUT DEPTH DATA, OUTER LANES Year Frequency MRD(in.) MRD/MESAL (in. x 1-7) M(RD/ESAL)(in. x 1-7) State & nterstate , , , State , , , nterstate Frequency - MRD- MRDMESAL- M(RD/ESAL) - Number f test site. Eight rut depth measurements at each test site, fur in inner and fur in uter wheel paths. Mean average rut depth. Rati f mean average rut depth t mean 18 kip equivalent single axle lads. Mean f the rati f average rut depth t 18 kip equivalent single axle lads. 5

51 TABLE 2. SUMMARY OF RUTTNG SUSCEPTBLTY BY AHD DVSON QataYeS[ Rutes Divisins Average Rut Depth (inches) State & nterstate State nterstate Mean Rut Depth/Mean ESAL (inch x 1-7 ) State & nterstate State nterstate Mean (Average Rut Depth/ESAL) (inch x 1-7 ) State & nterstate State : nterstate

52 TABLE 3. TEST STE PAVEMENT DESCRPTON. Site Rute Directin Mile Pst Pavement Structure 1-59 S OO#/SY 328 Mix placed 5/75 3#/SY 327 Mix placed #/SY 227 Mix placed 5175 Sil-Aggregate Base 2-59 N #/SY 416 Mix placed 1/83 "G" Treatment placed 1/83 1 OO#/SY 328 Mix placed #/SY 327 Mix placed #/SY 227 Mix placed 1975 SOil-Aggregate Base E #/SY 416 Mix placed 1/84 3#/SY 414 Mix placed 1/84 "L" Treatment placed 1/84 1 in. Plain PCC placed E #/SY 42 Mix placed 9/81 9#/SY 416 Mix placed 9/81 3#/SY 414 Mix placed 9/81 "L" Treatment placed 9/81 1 in. Plain PCC placed N #/SY 416 Mix placed 9/83 "H" Treatment placed 9/83 1 OO#/SY Suriace placed #/SY Leveling placed #/SY Leveling placed #/SY Suriace placed #/SY Bi nder placed 1962 Cement Treated Base S #/SY 416 mix placed 6/82 "G" Treatment placed 6/82 1 OO#/SY 328 Mix placed #/SY 327 Mix placed #/SY 227 Mix placed 1967 Select Sil Subbase N #/SY 416 Mix placed 1/79 "G" Treatment placed 1/79 1 OO#/SY Suriace placed #/SY Suriace placed #/SY Binder placed 196 Crushed Stne Base 52

53 TABLE 3 cntinued. TEST STE PAVEMENT DESCRPTON. Site Rute Directin Mile Pst Pavement Structure 8. US28 W #/SY 416 Mix placed 5/85 3#/SY 414 Mix placed 5/85 432#/SY 327 Mix placed 5/85 Crushed Stne Base 9. US231 S #/SY 416 Mix placed 2/83 1#/SY 416 Mix placed 198 "G" Treatment placed #/SY Surface placed #/SY Surface placed 196 "AKG" Treatment placed 196 Sand-Clay-Shell Base 1. US78 E #/SY 416 Mix placed 12/83 1#/SY 416 Mix placed layers (7") Asphalt Cncrete 11. US78 E #/SY 328 Mix placed 124 3#/SY 327 Mix placed OO#/SY 411 Mix placed 1966 "F" Treatment placed layers (5") Asphalt Cncrete Curshed Stne Base 12. SR157 S #/SY 416 Mix placed 11/85 325#/SY 414 Mix placed 11/85 "G" Treatment placed 11/85 Crushed Stne Base 13. US72 E #/SY Surface placed 11/76 3#/SY Binder placed #/SV.Black Base placed 11/76 Crushed Aggregate Base 53

54 TABLE 4. TEST STE RUTTNG ANALYSS DATA. But Dej;llb BDESAL BOdESAL. 18kip Test" Test 1988 Test Site ESAL Site DB Site DB Site (in) (in) (in) (in) (in) 1 6.6x x1-7.S8x x1- S 2 2.6x1 6 O.4S x Sx x1-S 3 2.9x x x x1- S 4 4.3x x1-7.47x Sx1-S S 2.x S S.4Sx Sx x1-S 6 3.7x x x x1-S 7 3.6x x1-7.36x x1- S 8.Sx1 6.S x1-7 4.x x1-S 9 1.6x x x1-7 S2.2x1-S 1 2.x1 6.S Sx x Sx1- S 11 S.9x1 6.3S.18.S9x1-7,3x x1-S 12.3x x1-7.67x1-7 2S.6x1- S 13 1.Sx x x x1- S Sites selected fr gd rutting perfrmance. "Rut depth measured with 12 ft straight edge acrss lane. +Rut depth measured with 4 ft straight edge acrss wheel path. Average f 8 measurements per mile fr entire design sectin in which test site lcated. 54

55 TABLE 5. TRAFFC DATA FOR TEST STES. %Cmm Date Date 18Kip Site Rute Directin MDT Vehicles Built Sampled ESAL S /75 11/88 6.6x N /83 11/88 2.6x E /84 3/89 2.9x E /81 3/89 4.3x N /83 3/89 2.x S /82 1/89 3.7x N /89 3.6x US28 W /85 1/89.5x US231 S /83 1/89 1.6x US78 E /83 11/89 2.x US78 E /89 5.9x SR157 S /85 12/89.3x US72 E /89 1.5x1 6 Ntes: traffic data used with n grwth factrs applied. 2. Age is dnference between date sampled and date placed. 3. Traffic distributin fr 4 lanes is 85% uter lane and 15% inner lane (Ref. 8). 4. Truck distributin factr f.82 used t cnvert traffic t 18 kip ESAL's (Ref. 8) 55

56 TABLE 6. PROPERTES OF N-PLACE ASPHALT MXTURES Site Layer Thickness (in) Asphalt Cntent (%) Average Vids (%) 8th Percentile Vids (%) Resilient Mdulus at 14 (ksi) Cnfined Creep Strain (14 F) Layer 1 was frictin curse. Sample failed during test. Resilient mdulus tests culd nt be perfrmed n samples frm sme layers. 56

57 TABLE 7. PROPERTES OF RECOMPACTED ASPHALT MXTURES 75 Blw 75 Blw 75 Blw GTM GTM GTM Rller Vids Stability Flw Vids Stability Flw Pressure Site Layer (%) (bs) (.1 in) (%) (bs) (.1 in) GS (psq Layer 1 was frictin curse. 57

58 TABLE 8. PROPERTES OF ASPHALT RECOVERED FROM ASPHALT MXTURES Penetratin Penetratin Penetratin Viscsity Viscsny Site Layer (77 F,1g,58) (4 F,2g,68) (4 F,1g,58) 14 F(P) 275 F(Cst) , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Layer 1 was a frictin curse 58

59 TABLE 9. PROPERTES OF AGGREGATE RECOVERED FROM ASPHALT MXTURES. NAA Flw Fractures Faces n Vids Time Plus NQ. a Material (.(QL PerQent PaSSing Sieve Size (.(12) Sne Layer (%) (Sec) 1 2 r mre 3/8 in. N. 5 N Layer 1 was a frictin curse. Minus NO.8 material. 59

60 TABLE 1. CORRELATON COEFFCENTS BETWEEN RUT DEPTH/-vESAL AND VAROUS AGGREGATE, ASPHALT AND MX PROPERTES. Prperty Crrelatin Cefficient Thickness, inches Vids (n-place), % Resilient Mdulus, psi Recmpacted Vids (Hand), % Recmpacted Vids (Gyratry), % Marshall Stability, bs Gyratry Stability, bs Marshall Flw,.1 inches Gyratry Flw,.1 inches GS Gyratry Rller Pressure, psi Passing 3/8 inch, % Passing N. 5, % Passing N. 2, % Fractured Faces, % NAAVids, % Flw Time, sec Asphalt Penetratin (77 F) Asphalt Viscsity (14 F)

61 TABLE 11. AVERAGE AGGREGATE PROPERTES FOR DVSONS 1 4 AND DVSONS 5-9. Divisins 1-4 Divisins 5-9 Aggregate Prperty Sites 7,8 & 1-13 Sites 1-6 & 9 Fine Aggregate (-#8) NAA Vids 44.1'% 42.3% Flw Time 22.3 sec. 2. sec. Carse Aggregate (+#8) Tw r Mre Crushed Faces Surface Mixes 78.6% 77.7% Binder Mixes 1.% 52.9% Cmbined Mixes 85.2% 7.4% Carse Aggregate (+#4) Tw r Mre Crushed Faces Surface Mixes 87.4% 82.7% Binder Mixes 1.% 52.7% Cmbined Mixes 91.3% 73.9% 61

62 FGURES /62.

63 1-9 AHD DVSONS CD TEST STES r7'l PEDMONT & l.l.:.lj APPALACHAN PLATEAUS - - FALLLlNE FGURE 1. LOCATON OF FELD TEST STES. 63

64 Radway nside Lane Outside Lane - : Cres spaced <a apprximately ' c tc : inch Cres 6 inch Cres FGURE 2.. TYPCAL STE SAMPLNG AND EVALUATON LAYOUT. 64

65 a. Measurement f Pavement Surface Prfile b. Measurement t Layer nterface. FGURE 3. LAYER PROFLE DEVELOPMENT. 65

66 c. Measurement t Layer nterface. d. Layer Thicknesses Frm Cres. FGURE 3. LAYER PROFLE DEVELOPMENT. (cntinued) 66

67 (J) -...J g '" i= OJ2 fu a OJO a: z.8 --<>.6", YEAR,:- 3. x..., ti :J <t 13 2.!!. '" <D Z i= 1. n. w a l- ii! a. Mean Rut Depth w Oll = YEAR b. Men Rut Depth Men ESAL 14,: x c 12 u c '" :J 1 w!!. '" <D - 8 J: l- n. w a 6 l- ii! w C!) 4 <t ffi > z 2 State nterstate Cmbined ',, YEAR c. Mean (Rut Depth ESAU FGURE 4. COMBNED DATA FROM PAVEMENT CONDTON DATABASES.

68 VM..UE FREQUENCY PERCENT (C1VER.'U.) VM..UE FREQUENCY PERCENT (x 1"7) D V S N D 3 Y 4 T 5 N B B B MEAN RUT DEPTH (inches) MEAN (RUT DEPTH/18 KP ES/>J...) (x 1"7) ') CO VM..UE FREQUENCY PERCENT (x 1"7) OJ (C1VER.'U.) D B Y T N MEAN RUT DEPTH/MEAN 18 KP ESM.. (x 1"7) FGURE DATA STATE AND NTERSTATE OUTER LANES RUTTNG AND RATE OF RUTTNG WTH AHD DVSON, 1984 STATE AND NTERSTATE DATA.

69 2 - VALUE FREQUENCY lb61 PERCENT B (OVER.Al.) VALUE FREQUENCY PERCENT (x 1"7) lb61 B.7 D 3 Y 4 S BO.97B B D 3 Y 4 T 5 N B Bl 2.B BO B B B B lbb6 B leb6 B.B MEAN RUT DEPTH (nches) MEAN (RUT DEPTH/B KP ESAL) (x 1"7) m CD 2 D 3 Y 4 T (O'VER.'LL) , 5 MEAN RUT DEPTH/MEAN lb KP ESAL (x 1"7) VALUE FREQUENCY PERCENT (x 1"7) B lb61 B.7 26BO B lbb6 B.B FGURE DATA STATE AND NTERSTATE OUTER LANES RUTTNG AND RATE OF RUTTNG WTH AHD DVSON, 1988 STATE AND NTERSTATE DATA.

70 41 VALUE FREQUENCY PERCENT l 41 VALUE FREQUENCY PERCENT (x 1'7) M 411 X T Y p E M 411 X T 416' Y P E MEAN RUT DEPTH (inches) t MEAN (RUT DEPTH / 18 KP ESAL) (x 1"7) '-l 1988 DATA NTERSTATE OUTER LANES 411 M 411' X VALUE (x 1'7) T Y P E] FREQUENCY PERCENT FGURE 7. RUTTNG AND RATE OF RUTTNG WTH MX TYPE, 1988 NTERSTATE DATA O.BO 1. MEAN RUT DEPTH / MEAN 1 B KP ESAL (x 1"7)

71 " "" M X T416 Y P E 417rl V.AL.UE FREQUENCY PERCENT (x 1"7) V.AL.UE FREQUENCY PERCENT (x 1"7) 4C M X MEAN RUT DEPTH / MEAN 18 KP ES/>L (x 1"7) 1984 DATA MEAN RUT DEPTH / MEAN 18 KP ES/>L (x 1"7) 1988 DATA STATE OUTER LANES -...J...L V.AL.UE (x 1"7) M X T Y 6.41 P E] FREQUENCY PERCENT FGURE 8. RATE OF RUTTNG WTH MX TYPE 1984, 1986, 1988 STATE ROUTE DATA T---' MEAN RUT DEPTH / MEAN 18 KP ES/>L (x,1"7) 1986 DATA

72 VAlUE FREQUENCY PERCENT VAlUE FREQUENCY PERCENT (x 1"7) ' M M 411' X 1 X T T 416' Y P Y P E E t= ---, MEAN RUT DEPTH (inches) MEAN (RUT DEPTH / 18 KP ESAL) (x 1 Ql\7) -...J VAlUE FREQUENCY PERCENT \) (x 1"7) M FGURE 9. RUTTNG AND RATE OF RUTTNG 1 WTH MX TYPE, 1986 STATE X T ROUTE DATA. y E] p ,,,, MEAN RUT DEPTH / MEAN 18 KP ESAL (x 1Q1\7)

73 FREQ CUM. PERCENT CUM. FREQ PERCENT UJ - - )..c c:: - = a.. UJ ::: FREQUENCY 1988 DATA STATE RaLTES OUTER LANES FREQ CUM. PERCENT CUM. FRED PERCENT "OJ' - Q)..c c:: = a.. w f-- ::> ::: : FREQUENCY 1988 DATA NTERSTATE ROUTES OLTER LANES FGURE RUTTNG FREQUENCY DSTRBUTONS. 73

74 FREQ CUM. PERCENT CUM. FREQ. PERCENT , x <ll c u c = <i! (fj W a s:: CO F a W :::J :: FREQUENCY 1988 DATA STATE ROUTES OUTER LANES g FREQ CUM. PERCENT CUM. FREQ. PERCENT x <ll.c c = w a s:: CO F a w :::J :: FREQUENCY 1988 DATA NTERSTATE ROUTES OUTER LANES FGURE RATE OF RUTTNG FREQUENCY DSTRBUTONS. 74

75 meq CUM. PERCEN" CUM. FREQ CUM. PERCEN" CUM. FREQ PERCEN" meq PERCEN" B ;;; c: u g - c: <= <= i!: h a:: J (J BODO FREQUENCY FREQUENCY 19B4 DATA FREQ CUM. PERCEN" CUM. meq PERCEN" DATA SfAT ROU1ES OU1ER LANES FGURE 12. RUT DEPTH FREQUENCY DSTRBUTONS ;;;- -.c: " u c: <= ,66 5 a:: BODO FREQUENCY 1986 DATA

76 --..J (J) 62.5 R Q;' b 312.5! ] : 562.5l S g 812.S = 937.S 162. w a S S CO 1562.S S F a gj 262.S ' OC S4q:;::;::;::;,".."..rr-r-.r..,..., FREQUENCY 1984 DATA ;!1t::::::::::::::::::::::: ' x g 812.S c <f Ul - w a S CO 1562.S F '3 c FREQUENCY DATA 6 5 FREa CUM. PERCENT CUM. FREa. PERCENT FREa CUM. PERCENT CUM. FREa. PERCENT , sll S p', ---' w a F a gj ' c J7.sl, iii i ii ii iii ii iii iii ii i, ] iii iii ii iii iii iii FREQUENCY 1988 DATA FGURE 13. STATE ROUTES OUTER LANES FREa CUM. PERCENT CUM. FREa. PERCENT ' RATE OF RUTTNG FREQUENCY DSTRBUTONS.

77 .-.-:-,.;;,::.=-: ',-... ::... :"). - 3='".t:j;-":,:::".;,;,i:,';;;; FGURE 13. STRNGLNNG LAYER NTERFACES TO DETECT PERMANENT DEFORMATON. 77