Strategic Highway Research Program Binder Rheological Parameters: Background and Comparison with Conventional Properties

Transcription

1 32 TRANSPORTATON RESEARCH RECORD 1488 Strategic Highway Research Program Binder Rheological Parameters: Backgrond and Comparison with Conventional Properties HUSSAN U. BAHA AND DAVD A. ANDERSON As a reslt of the research condcted for the Strategic Highway Research program (SHRP), a set of new testing methods to characterize the rheological, failre, and drability properties of asphalt binders has been developed. These methods se testing devices that either are completely new or that have been sed before only for research prposes. The methods also call for measring mechanical response parameters that are not very common for asphalt pavement engineers or for many asphalt prodcers. The prpose is to discss the following points: (a) the viscoelastic natre of asphalts and its relation to pavement performance, (b) the concepts behind selecting the new test methods and the new characteristic properties, and (c) how the new measred properties compare with the conventional properties. These points are addressed by providing a theoretical-conceptal backgrond abot the conventional and new tests and by comparing new and conventional data measred for a large nmber of asphalts that vary in their sorces and their grades. The comparison identifies the advantages of the new testing methods and the need to implement the proposed testing and specification system. One of the main objectives of the Strategic Highway Research Program (SHRP) A2A project was to identify the physical properties of asphalt cement binders that are related to pavement performance and the methods of reliably measring these properties (). The first step to achieving this objective involved an extensive review of the literatre related to asphalt materials and pavements. The review reslted in more than 5 abstracts and inclded information pblished since the beginning of the centry (2). The review indicated that there is no lack of realization of the types of pavement failres; rtting, fatige cracking, and thermal cracking are the main failre modes that asphalt researchers have commonly related to the physical properties of binders. Also, it was clear that age hardening is a main factor in changing the properties of asphalts dring pavement service life and that ths affects performance. Moistre damage, althogh a major distress mode, was known to be the reslt of the interaction between asphalt binders and aggregate, and hence cannot be appropriately addressed by binder properties alone. The review, however, indicated that there is a significant confsion abot important binder properties and which of these properties can more reliably be related to pavement performance. The confsion mainly comes from nderestimation of the complexity of the binder properties and the empirical natre of the methods sed to measre H. U. Bahia, Department of Civil and Environmental Engineering, The University of Wisconsin-Madison, Madison, Wis D. A. Anderson, The Pennsylvania Transportation nstitte, The Pennsylvania State University, University Park, Pa these properties. Althogh many asphalt researchers and practitioners realized the viscoelastic natre of the material and the need for fndamental rheological methods for proper characterization, few had the resorces or the backgrond to stdy asphalt binders by rigoros rheological methods. The findings of the SHRP asphalt binder research indicated that to better select asphalts there is no sbstitte for fndamental rheological and failre characterization. The findings indicated that the existing methods are handicapped by empiricism and simplifications to levels that are not acceptable to meet the present needs of the indstry. As part of the research new methods and parameters were introdced to measre more fndamental properties that can be easily related to pavement performance on the basis of sond engineering concepts. The new testing and aging methods inclde the dynamic shear rheometer (3), the bending beam rheometer ( 4), the direct tension test (5), and the pressre aging vessel (6). The new parameters inclde complex shear modls (G*), phase angle (S), creep stiffness [S(t)], logarithmic creep rate [m(t)], and failre strain (E 1 ). The prpose of this paper is to provide a theoretical-conceptal backgrond abot the conventional and new tests and a comparison of properties measred by conventional and new methods for a large nmber of asphalts. The backgrond and the data comparison identify the advantages of the new parameters and the importance of the properties measred with the new testing systems. VSCOELASTC NATURE OF ASPHALT BNDERS The most niqe behavior of viscoelastic materials is the dependency of their mechanical response on time of loading and temperatre. At any combination of time and temperatre, viscoelastic behavior, within the linear range, mst be characterized by at least two properties: the total resistance to deformation and the relative distribtion of that resistance between an elastic part and a viscos. part. Althogh there are many methods of characterizing viscoelastic properties, dynamic (oscillatory) testing is the best techniqe to explain the niqeness of the behavior of this class of materials. n the shear mode G* and angle Sare measred. G* represents the total resistance to deformation nder load, whereas S represents the relative distribtion of this total response between an in-phase component and an ot-of-phase component. The in-phase component is an elastic component and can be related directly to the energy stored in a sample for every loading cycle, whereas the ot-of-phase com-

2 Bahia and Anderson 33 ponent represents the viscos component and can be related directly to the amont of energy lost per cycle in permanent flow. The relative distribtion of these components is a fnction of the composition of the material, loading time, and temperatre. Rheological properties can be represented either by the variation of G* and 8 as a fnction of freqency at a constant temperatre (commonly referred to as a master crve) or by the variation of G* and 8 with temperatre at a selected freqency or loading time, commonly called an isochronal crve. Althogh time and temperatre dependency can be related by a temperatre-freqency shift fnction (7), for practical prposes it is mch easier to present data with respect to one of the variables. Figre depicts typical rheological properties of an AC-4 and an AC-5 asphalt binder at a wide T 1 -o- AC-4 (G ) - AC-4 (Delta) 1 ;:: - AC-5 (G ) - - AC-5 (delta) Q_ Ol 8 Cl 9 _J 8 l{) \ () N 7 6 N 6 (/) ::l -s 5 4 "() c <( L 4 CD Ol x (1) r. a. 3 2 a.. E 2 (a) Loading Freqency, log rod/s () O Q Ci range of temperatres and freqencies. Figre l(a) shows master crves at 25 C, and Figre (b) shows isochronal crves at 1 rad/sec. Some common niqe characteristics of the rheological behaviors of asphalts can be seen in the typical plots of Figre 1: At low temperatres or high freqencies both asphalts tend to approach a limiting vale of G* of approximately 1. GPa and a limiting vale of 8 of. degrees. The 1. GPa reflects the rigidity of the carbon hydrogen bonds as the asphalts reach their minimm thermodynamic eqilibrim volme. The. vale of 8 represents the completely elastic natre of the asphalts at these temperatres. As the temperatre increases or as the freqency decreases, G* decreases continosly, whereas 8 increases continosly. The first reflects a decrease in resistance to deformation (softening), whereas the second reflects a decrease in elasticity or ability to store energy. The rate of change is, however, dependent on the composition of the asphalt. Some will show a rapid decline with temperatre or freqency; others will show a gradal change. Asphalts within this range may show significantly different combinations of G* and 8. At high temperatres the 8 vales approach 9 degrees for all asphalts, which reflects the approach to complete viscos behavior or the complete dissipation of energy in viscos flow. The G* vales, however, vary significantly, reflecting the different consistency properties (viscosity) of the asphalts. From this simplified description of asphalt properties it is clear that withot the distinction between types of asphalt response in terms of total resistance to deformation (G*) and relative elasticity (8) and withot measring properties at the temperatre or loading freqency ranges that correspond to pavement climatic and loading conditions, selection of asphalt binders for better-performing pavements is not possible. One of the main problems with the methods sed crrently is their failre to measre properties at application temperatres and to distingish between elastic and nonelastic binder behavior o- AC-4 (C.) - AC-4 (Delta) ll. 1 --AC-5 (G ) AC-5 (delta) - - _J, Cfi, / "" """- "() 8 O L O 7 6 a) Q 6 c <( 5 4 Ul Gl ::l "5 "().c 4 a... x 3 "\.v 2 a. E 2 / _,..; " (b) Temperatre, deg C FGURE 1 Typical rheological spectra for AC-5 and AC-4 asphalt binders: (a) freqency master crves; (b) isochronal crves. ASPHALT PROPERTES AND PAVEMENT PERFORMANCE Figre 2 is an isochronal plot that depicts the rheological properties of an asphalt in its naged condition and after aging in the field nder a moderate climate for approximately 16 years. To relate asphalt properties to pavement performance reference can be made to for temperatre zones. At temperatres above C mixing and constrction take place, and ths, the binder consistency needs to be controlled. At temperatres above 1 C most asphalt binders behave like Newtonian flids, whose response is totally viscos. Therefore, a measre of viscosity is sfficient to represent the workability of the asphalt dring mixing and constrction of hot-mix asphalt. At temperatres in the range of 45 C to 85 C, which are typical of the highest pavement in-service temperatres, the main distress mechanism is rtting, and therefore, G* and 8 need to be measred. A measre of viscosity alone cannot be sfficient, since viscosity measrements are done on the assmption that asphalt response has only a viscos component. For rtting resistance a high G* vale is favorable becase it represents a higher total resistance to deformation. A lower 8 is favorable becase it reflects a more elastic component of the total deformation.

3 34 TRANSPORTATON RESEARCH RECORD 1488 a.. Ol...J i "O ""... (...) * - Unaged (C.) --- Unaged (delta) - Field Aged (G ) - Field Aged (delta) , ro 6 ro m 1 /::," 1 ---; THERMAL CRACKNG ATGUE CRACKNG RUTTNG r------r--ir--..., ,---r Temperatre, deg C FGURE 2 Typical rheological behavior of asphalt binders before and after aging in the field in relation to pavement main distress modes. Within the intermediate temperatre zone asphalts are generally harder and more elastic than at higher temperatres. The prevailing failre mode at these temperatres is fatige damage, which is cased by repeated cycles of loading at levels lower than the static strength of a material. For viscoelastic materials like asphalt binders, both G* and S play a role in damage cased by fatige. They are both important becase dring every cycle of loading the damage is dependent on how mch strain or stress is developed by the cyclic load and how mch of that deformation can be recovered or dissipated. A softer material and a more elastic material will be more favorable for resisting fatige damage becase the stress developed for a given deformation is lower and the asphalt will be more capable of recovering to its preloading condition. Similar to the case for rtting, a single measre of hardness or viscosity cannot be sfficient to select better-performing asphalts with respect to fatige resistance. Rtting and fatige damage are both fnctions of freqency of loading, and therefore, the rate of loading of the pavement nder traffic needs to be simlated in measrement to obtain a reliable estimate of the binders contribtion to pavement performance. The forth and last temperatre zone is the low-temperatre zone, at which thermal cracking is the prevailing failre mode. Thermal cracking reslts from the thermal stresses generated by pavement shrinkage as a reslt of thermal cooling. Dring thermal cooling asphalt stiffness increases continosly and ths reslts in higher stresses for a given shrinkage strain. Simltaneosly, thermal stresses relax becase of the viscoelastic flow of the binder. To reliably predict the binders contribtion to cracking both the stiffness of a binder and its.rate of relaxation need to be evalated. The stiffness of the binder is directly proportional to G*, and the rate of relaxation is directly related to S. A lower stiffness and higher rate of relaxation are favorable for resistance to thermal cracking. As with other temperatre zones, a single measre of the stiffness or viscosity of the binder is not sfficient to select better binders that will resist cracking at the lowest pavement temperatres. This discssion of the relation between asphalt binder properties and pavement performance is frther complicated by the aging phe- Ol 1) i 1) "... nomenon. Asphalts are hydrocarbon materials that oxidize when they come into contact with oxygen from the environment. This oxidation process changes the rheological and failre properties of the asphalt. As shown in Figre 2 the rheological master crve becomes flatter with aging, which indicates higher G* vales and lower S vales at all temperatres. These changes translate into less sensitivity of G* and S to temperatres or loading freqency and into a more elastic component (lower S vales). Significant oxidation effects sally appear after considerable service life. ncreased G* vales and lower S vales are favorable changes with respect to rtting performance, bt they are nfavorable for thermal cracking performance. For fatige cracking the increase in the G* vale is not favorable, whereas the decreased S vale is generally favorable, depending on the type of pavement and the mode of fatige damage. NEW PROPOSED MEASUREMENTS AND PROPERTES The properties that are proposed for the new SHRP binder specification were derived and selected by addressing each type of pavement failre, nderstanding the failre mechanism, nderstanding the contribtion of the binder to resistance to that failre, and selecting the reqired measre that will best reflect that contribtion of the binder (J). The new binder specification is based on climatic conditions; the criteria that a binder mst meet do not change, bt the temperatre at which the property is measred depends on the specific field climate and on the failre mode being considered. Three failre modes were identified as critical pavement distress modes in which binder plays an important role: rtting, fatige. cracking, and thermal cracking. Oxidative aging and physical hardening were considered drability factors that case changes in properties of binders and that ths affect performance. For types of tests were selected (8): The rotational viscometer, to measre flow properties at temperatres that mimic temperatres at which the pmping and mixing of binders occr. The dynamic shear rheometer, to measre properties at temperatres that mimic high and intermediate pavement temperatres and to mimic loading rates typical of traffic loadings. The bending beam rheometer, to measre properties at the lowest pavement temperatres and to mimic loading conditions that reslt from thermal cooling. The direct tension test, to measre failre properties at the lowest pavement temperatres and to mimic loading that reslts from thermal cooling. High-Temperatre Consistency Althogh workability is not directly related to pavement distress modes, asphalt binders mst be workable enogh at high temperatres sch that pmping, mixing with aggregates, and compaction can be done efficiently to prodce the reqired mixtre properties. To ensre binder workability the Brookfield viscometer has been selected to measre steady-state viscosity at one or more temperatres. n the binder specification the binder is reqired to have a maximm viscosity of 3. Pa sec to ensre workability. Althogh the rotational viscometer is already a standard test (ASTM D442),

4 Bahia and Anderson 35 its se will reslt in two changes with respect to the existing specification practices (ASTM 3381). First, it will replace the kinematic viscosity measrement; second, a maximm limit, instead of the minimm limit reqired in the crrent specification, is specified. The rotational viscometer is more sitable for modified asphalts, which are difficlt to test with the capillary tbes sed for kinematic viscosity. Also, a maximm limit is more appropriate to ensre the workability of asphalts. Contribtion of Binder to Rtting Resistance Opinions abot the contribtion of binder to rtting resistance differ. t is a fact, however, that soft asphalts are not sed to constrct pavements in hot desert climates. t is also a fact that dring the past decade more and more engineers are specifying polymer-modified binders, at mch higher costs, to mitigate rtting problems. Aggregate properties are withot dobt very important. Many engineers, however, agree that it is not good engineering practice to ignore binder properties. Rtting is cased by the accmlation of permanent deformations cased by the repeated applications of traffic load!ng. Assming that pavement rtting is mainly cased by deformations of the srface layer, rtting can be considered a stress-controlled, cyclic loading phenomenon. Dring each cycle of traffic loading a certain amont of work is being done in deforming the srface layer. Part of this work is recovered in elastic rebond of the srface layer, whereas the remaining work is dissipated in permanent deformation and heat. To minimize rtting the work dissipated dring each loading cycle shold be minimized. For a viscoelastic material the work dissipated per cycle (We) is calclated in terms of stress (cr) and strain ( E) as follows: We = 7T a E sin 8 Rtting within the asphalt concrete layer can be assmed to be a stress-controlled (cr ) repetitive phenomenon. Therefore, the following sbstittion can be made:.. We = 7T a e sin 8 since <To E=- G*. w ) c - 7T CTo ( G*/sin 8 This relationship indicates that the work dissipated per loading cycle is inversely proportional to the parameter G*/sin 8, which is the parameter selected for the SHRP specification. The parameter combines the total resistance to deformation, as reflected by G*, and the relative nonelasticity of the binder, as reflected by sin 8. Sin 8 is the ratio of the loss modls (G") to the complex modls G*. G" is directly related to the work dissipated dring a loading cycle, and ths, its ratio to G* gives a relative measre of the nonelastic (permanent) component of the total resistance to deformation. The logic associated with the parameter is that the contribtion of the binder to rtting resistance can be increased by increasing its total resistance to deformation (G*) or decreasing its nonelasticity (sin 8). G* and 8 are fnctions of temperatre and freqency of loading. Therefore, to relate the measrements to pavement conditions, the specification reqires testing at the average 7-day maximm pavement design temperatre and at a freqency of rad/sec. The proposed measrements take into accont the viscoelastic natre of the material, the climatic conditions of the specific application, and the loading condition (traffic) that is casing the pavement distress. Figre 3 depicts the vales of G* and 8 for a large nmber of asphalts that were tested for SHRP. All measrements were done at a freqency of rad/sec, which is assmed to simlate the average freqency of a stress wave in the srface layer of a typical pavement as cased by a vehicle moving at 5 to 6 mph. Figre 3 shows no specific trend between G* and 8 vales. This indicates that at a certain freqency and temperatre asphalts vary significantly in their G* and 8 properties. t is therefore necessary to measre both and to consider both in estimating the contribtions of binders to rtting resistance. The grop of datm points to the left of the plot are data measred for polymer-modified asphalts that were formlated to increase the elasticities of specific binders. The testing was done at temperatres in the range of 72 C to 82 C. This set of data points ot the importance of measring 8 to characterize the elasticity at high temperatres that may significantly contribte to resistance to rtting. Figre 3 represents the properties of naged binders and binders aged by the thin film oven test (ASTM D 754). For all nmodified asphalts and most modified asphalts, oxidative aging reslts in increased G* vales and decreased 8 vales. These changes reslt in more resistance to deformation and more elasticity, which means more contribtion to rtting resistance. The initial properties of the binders (early pavement life) are therefore more critical than the aged properties, and that is why the SHRP specification minimm limits are reqired for G*/sin 8 measred on the naged and the oven-aged binders. Figre 4 compares the new measre with the absolte viscosity. Althogh the figre shows that there is a fair correlation between the two measres, it indicates that there is a wide range of vales of G*/sin 8 for each vale of viscosity and vice versa. For example, at a vale of 2 P, typical of an AC-2 asphalt, the vale of G*/sin 8 may vary between 17 and 32 Pa, which is a range of -15 to + 6 percent. n fact the standard error for the estimate by sing a (}._ 12 i E Bl K le * 1 P H 6... () y 1111 Bl x BE J F ::J 8l :i P1l wii!b 8l L Hl K P6!B x P8 P2 "(!9 G 2 111!BEii Bl a. P1 al. P11P7 fbli 8 E P9 PJ Bl A P4 PS " v Phase Angle at 1 rad/s, deg FGURE3 Typical vales of G* and 8 at high pavement temperatre.

5 36 TRANSPORTATON RESEARCH RECORD 1488 Cl. cti - " !2 6 c 4 "iii (j * Absolte Viscosity at 6 C, P FGURE 4 Comparison between capillary viscosity and G*/sin o at 6 C. linear relation is estimated at approximately 75 Pa, a range that cannot be considered acceptable for many engineering prposes. The discrepancy between the G*/sin 8 and absolte viscosity is a reflection of the difference in the two measres; absolte viscosity is measred at a different shear rate, a different mode of loading, and a different strain and stress level. Absolte viscosity also does not consider the elastic response of a binder. All of these factors point ot the advantages of the new parameter and indicate that absolte viscosity cannot be sbstitted for G*/sin 8. Contribtion of Binder to Fatige Cracking Resistance As shown in Figre 2, at intermediate pavement temperatres the main distress mode is fatige cracking. Fatige of pavement can be a controlled-stress phenomenon (typical for thick pavement layers) or a controlled-strain phenomenon (typical of thin pavement layers). Fatige cracking, however, is known to be more prominent in pavements with thin layers. Based on the assmption that the fatige cracking mechanism is mainly driven by relatively large deformations of thin srface layers nder traffic loading, it can be considered predominantly a strain-controlled phenomenon. The large deformations sally reslt from the low level of spport of sbsrface layers that may reslt from poor design and constrction or becase of the satration of base layers dring the spring season. Based on these assmptions the dissipated work concept can be sed to derive the parameter sed in the SHRP specification, G* sin 8. For a strain-controlled cyclic loading the work per cycle eqation can be rewritten as follows:.. We = 7T er Eo sin 8 where Eo is the strain amplitde being applied. Since stress (er) is related to strain by G*: cr =.Eo X G* Sbstittion leads to the following eqation, which shows that We nder strain-controlled conditions is directly related to G* sin 8: :. We= 7T E6 (G* x sin 8) The work done dring a loading cycle can be dissipated in one or more damage mechanisms: cracking, crack propagation, dissipated heat, or plastic flow. Althogh the dissipation in heat or plastic flow may be better than dissipation in cracking, heat and plastic flow are only other types of damage that may case permanent deformation, allow faster crack propagation, or permit detrimental distortion of the asphalt mixtre strctre. To prevent all types of damage it is therefore best to limit the energy dissipation by limiting the vale of the parameter G* sin 8. This is the concept on which the SHRP specification is based. The logic associated with the parameter is that the amont of work dissipated is directly proportional to G* sin 8; asphalts with lower G* vales will be softer and ths can deform withot developing large stresses. Also, asphalts with lower 8 vales will be more elastic and ths will recover to their original condition withot dissipating energy in any fashion. Figre 5 depicts the typical range of vales for G* and 8 for a large nmber of asphalts tested for SHRP. Similar to the rtting parameter, the vales are measred at l rad/sec to simlate traffic loading. They are, however, measred on binders after aging in the pressre aging vessel (PAV), which has been shown to simlate long-term oxidative aging in the field. The PAV-aged condition is assmed to be more critical becase for most binders it reslts in a significant increase in G*, which offsets the effect of the decrease in 8. The tests are done. at intermediate pavement temperatres as determined from the average of the 7-day maximm and the lowest pavement design temperatres. The only conventional measre that is in the same temperatre range of average pavement temperatres is the penetration at 25 C. Figre 6(a) and 6(b) show plots of G* sin 8 vales before and after PAV aging verss the penetration vales of the naged asphalts. The plots show that asphalts with G* sin 8 vales ranging between.45 and 1.8 MPa, a range of more than forfold, can have penetration vales that vary only between 5 and 6 dmm. Similarly, after PAV aging asphalts with approximately the same penetration vales may have a range of G* sin 8 vales of between.6 and 7. Cl. 2 >" <( Cl {) en... * T fd J ; x z 9 f E.. R Bf 1y BA?t s D r v ct.a Phase <Jlgle at 1 roo/s (25 C), deg FGURE 5 Relationship between G* and phase angle at 25 C for PAV-aged binders. * 65

6 Bahia and Anderson 37 3 a oi 1J " 2 L..! l 81 &.8.,,. - "U m P H c "iii...5 () (a) + Q5 "O c "iii () * 1B m & Penetration at 25 C,.1 rrm ;i i { _,... \G!Hr :m T Hl 4 i z J ::-.1.:-.:,,J " Bl i Q T--,-.--,-T--,-..--,-.;--, 5 15 (b) Penetration at 25 C ( naged),.1 mm FGURE 6 Relationship between penetration of naged asphalts and G* sin 8 (a) before and (b) after PAV aging at 25 C. MPa. The plots demonstrate the failre of the penetration measre to give a reasonable indication abot the critical properties of asphalts. The selection of the fatige parameter, similar to the rtting parameter, is derived on the basis of sond engineering principles; it considers the climatic conditions by testing at average pavement temperatres; it is measred by sing a loading mode that simlates traffic loading; and it considers the viscoelastic natre of the asphalt material. Contribtion of Binder to Resistance to Thermal Cracking Thermal cracking is the reslt of stresses developed in pavement layers becase of thermal shrinkage cased by environmental cooling. Althogh cracking may be cased by rapid thermal cycling in relatively moderate climates, low-temperatre cracking in cold 2 regions is the predominant failre distress. Dring a thermal cooling cycle shrinkage of the asphalt layer in a pavement is restrained by friction with the nderlying layers that either are warmer or ndergo less shrinkage becase of a smaller coefficient of thermal contraction. This restraint will reslt in the development of tensile stresses that, if not relaxed by the flow of the asphalt layer, will eventally exceed the tensile strength of the material and case cracking. The total amont of stresses developed depends on the stiffness (resistance to deformation) of the asphalt binder and on its ability to relax stresses by dissipating energy in permanent flow. Traditionally, thermal cracking has been correlated with the stiffness of asphalts measred or estimated at certain loading times (8,9). Stiffness, however, does not reflect the stress relaxation ability of a binder. To be able to relax stresses an asphalt shold be able to flow readily nder stress and to have a less elastic component in its response. By measring the creep response of asphalts with the bending beam rheometer the stiffness of asphalts, S(t), can be measred at the lowest pavement temperatres. Also, by measring the logarithmic creep rate, m( t ), the ability of an asphalt to relax stresses can be evalated. A higher S(t) reflects more stresses reslting from a given thermal strain (shrinkage), and a higher m(t) reflects a higher creep rate and ths a faster relaxation rate. S(t) and m(t) are, however, both fnctions of loading time; therefore, a certain loading time mst be selected to reflect pavement thermal cracking. n the asphalt literatre loading times ranging between 3,6 and 2, sec have been correlated with thermal cracking (2). Sch loading times are not practical for laboratory testing. To shorten the testing time the time-temperatre sperposition principle was sed to perform tests at a higher temperatre, bt for a shorter loading time. Dring SHRP the stdies of low-temperatre properties have indicated that time-temperatre eqivalency factors are approximately the same for most asphalts ( 4). This finding was sed to calclate the temperatre shift reqired to redce the 7,2-sec loading time, most commonly recommended in the asphalt literatre, to a loading time within a 24-sec loading time range. A 1 C increase in temperatre was fond to be eqivalent to a time shift from 7,2 sec to approximately 6 sec (4). n the crrent specification a maximm limit of 3 MPa is placed on S(6), and a minimm m(6) of.3 is reqired. The logic associated with the low-temperatre measres is that by placing a maximm limit on S(t) the level of stresses developed in the pavement is limited, and by placing a minimm limit on m(t) the rate of relaxation is kept above a certain limit. Figre 7 is a plot of S(6) verss m(6) for a large nmber of asphalts varying in sorce and physical properties. The measrements were done on binders aged in the PAV with the bending beam rheometer. Oxidative aging always reslts in illcreased S(t) and decreased m(t). Therefore, testing in the specification is reqired on PAV-aged binders to represent the critical aging condition. The testing is done at the minimm pavement temperatre pls 1 C to reflect pavement climatic conditions. The loading condition in the pavement is also simlated by sing a transient loading mode to measre the extensional stiffness. Figre 7 reflects the wide variation in m(6) vales for a given S(6) and vice versa. One other factor related to the low-temperatre behavior of asphalts is the newly discovered physical hardening at low temperatres (1). Physical hardening is the increase in S(t) and the decrease in m(t) that occrs as a reslt of time-dependent volme shrinkage of asphalts. The phenomenon is cased by the deviation from thermodynamic eqilibrim becase of the lag of moleclar adjstment behind the thermal change dring cooling of the material throgh its glass transition temperatre range. The phenomenon

7 38 TRANSPORTATON RESEARCH RECORD i ld + Cl) Cl).E... a !Biii s.... m.. S m: iii ilb - i! l 1l11 m ;- - - ml lmmlbllm... m _ Logaritrmic creep rate "m 11 at 6 s..._...._1..._ FGURE 7 Relationship between creep stiffness and logarithmic creep rate at 6 sec for a large set of asphalts after PAV aging..6 \5 c..5 () c 5.. ; R 5 <).. 5 LL m Penetration at 4 C (naged),.1 mm FGURE 8 Relationship between strain at failre measred at -1 C and 1-mm/min test rate and penetration at 4 C. resembles what is known as physical aging, which is a case of time-dependent hardening for many plastics and polymers (8,9). For many asphalts physical hardening was fond to increase S( t) by 5 to 1 percent within 24 hr. ts conseqences on asphalt concrete mixtre properties and pavement performance are not yet known. Therefore, the SHRP specification reqires the measrement and reporting of the vales of S(6) and m(6) at 1 and 24 hr to provide an indication of the potential of the binder for hardening. The reported vale is intended to make engineers aware of the existence of this phenomenon and to encorage the research and engineering commnity to evalate the effect of this phenomenon on pavement thermal cracking. As for the measrements sed in controlling S(6) and m(6), a constant conditioning time of 6 ::!:: 5 min is reqired to test all asphalts at an eqal isothermal conditioning time. Physical hardening has been fond to be a strong fnction of asphalt chemical composition and moleclar strctre. ts conseqences, if proved to be carried at the same level as the mixtres properties, Can be very important in estimating binder contribtion to thermal cracking resistance. n addition to the creep measrements the new specifications inclde the direct tension test, which is a tre strength test. For a wide variety of materials prefailre properties do not necessarily correlate very well with the failre properties. Unmodified asphalts, however, have been shown to have failre properties that correlate well with the stiffness vales at low temperatres (1). Based on this finding the specification does not reqire testing for strain at failre if the S(t) and m(t) criteria are met. For some modified asphalts the relation between the stiffness and failre properties can be different. Therefore, to accommodate those modified asphalts that may have high S(t) vales bt that can still show high strain tolerance, the direct tension test can be sed to measre the strain and failre and override the S(t) criterion if S(6) is between 3 and 6 MPa. The direct tension test is another new test that gives the opportnity to directly evalate the strain tolerance properties of specially modified binders. Figre 8 gives typical vales of strain at failre for a wide variety of asphalts tested at temperatres resembling low pavement temperatres. To compare creep measrements with conventional measres, Figre 9 shows penetration at 4 C with S(6) and m(6) measred at - l C. At a vale of S(6) of 1 MPa asphalts can have penetration vales ranging between 1 and O dmm. Similarly, at a vale of m(6) of approximately.35 asphalts can have penetration vales ranging between. and 13 dmm. The data in Figre 9 are a clear indication of the insensitivity of penetration to the tre rheological properties of asphalts at low temperatres. SUMMARY AND CONCLUSONS A conceptal analysis of the viscoelastic natre of asphalt binders and the relation between their rheological properties and pavement performance has been presented. The types of conventional measres that are sed at present and their failre in reflecting the crit-... > <(...,..._ ld -" l/) en (/) <) :;:; (/) Q , ! S(O) m(6) : i ;...!! -l O B l i..._...) i <> ; : () a () () () -f- i ! ! S! m t :tst-.---f!!! > <( a ld i -.3 E ,---.._,---., ,--.--,-,--.--,--t.2 o s n m Penetration at 4 C (ll1aged),.1 mm FGURE 9 Relationship between S(6) and m(6) measred at - l C and penetration at 4 C.

8 Bahia and Anderson 39 ical properties of binders were discssed. The concept on which the new testing and characterization methods introdced by SHRP are based has been presented with emphasis on how these new measres relate to pavement behavior and failre mechanisms. The conceptal analyses and the typical data comparing the conventional and the new SHRP measrements lead to the following conclsions: 1. Asphalt binders vary in their rigidities (total resistance to deformation) and their relative elasticities (distribtion of that resistance between an elastic and a viscos part). For pavement applications both of these characteristics need to be measred at temperatres and loading rates that resemble climatic and traffic conditions. 2.; Conventional measrements of the physical properties of asphalt inclde empirical measrements, viscosity measrements, and ssceptibility parameters. These measrements cannot be considered reliable for characterizing the asphalt properties that are critical for pavement performance becase of the empiricism involved and becase of the engineering complications related to the method by which they are interpreted. 3. The new measrements proposed by SHRP are based on sond engineering principles derived from an nderstanding of the mechanisms of failre or damage in the pavement. They reflect.the best estimate of the contribtion of binders to pavement performance. They cover the main distress mechanism of flexible pavements, and they are measred nder conditions that mimic climatic and loading conditions in the field. 4. No strong relationships between the new SHRP measrements and the conventional measrements exist. These two sets of measrements differ in the material characteristics that they represent and the test conditions at which they are obtained. ACKNOWLEDGMENTS The work reported herein is a part of Project A2A of SHRP. Project A2A was condcted by the Western Research nstitte, Laramie, Wyoming, in cooperation with the Pennsylvania Transportation nstitte (PT). Acknowledgments go in particlar to D. Christensen, R. Dongre, and M. G. Sharma, who were principal members of the research team at PT. Acknowledgments also go to T. Kennedy, E. Harrigan, J. Molthrop, and D. Jones for their contribtions in disgssions and their feedback dring the development of the specification concepts and parameters. Acknowledgments also go to all members of the FHW A Expert Task Grop for their valable discssions and comments on the work condcted in the project. REFERENCES 1. Anderson, D. A., H. U. Bahia, D. W. Christensen, R. Dongre, M. G. Sharma, C. Antle, and J. Btton. Binder Characterization and Evalation, Vol. 3. Physical Characterization Report SHRP-A-369. Strategic Highway Research Program, National Research Concil, Washington, D.C., Bahia, H. U. Bibliographies for Physical Properties of Asphalt Cement. Report SHRP-A-626. Strategic Highway Research Program, National Research Concil, Washington, D.C., Anderson, D. A., and H. U. Bahia. Evalation of Operating Conditions for Dynamic Shear Rheometers Used for Asphalt Binder Specification Testing. Presented at ASTM Symposim on Physical Properties of Asphalt Cements, Dallas/Ft. Worth, Tex., Dec. 7, Bahia, H. U., and D. A. Anderson. The Development of the Bending Beam n Rheometer: Basics and Critical Evalation. n Physical Properties of Aspalt Cement Binders: ASTM STP 1241 (J.C. Harden, ed.) ASTM, Anderson, D. A., and R. Dongre. The SHRP Direct Tension Specification Test-ts Development and Use. Presented at the ASTM Symposim on Physical Properties of Asphalt Cements, Dallas/Ft. Worth, Tex., Dec. 7, Bahia, H. U., and D. A. Anderson. The Pressre Aging Vessel (PAV): A Test to Simlate Rheological Changes De to Field Aging. n Physical Properties of Asphalt Cement Binders: ASTM STP 1241 (J.C. Harden, ed.) ASTM, Ferry, J. D. Viscoelastic Properties of Polymers. John Wiley and Sons, nc. New York, Bahia, H. U., and D. A. Anderson. "Physical Hardening of Asphalt Binders and Relation to Compositional Parameters," Preprints of the 24th American Chemical Society National Meeting, Division of Fel Chemistry, Vol. 37, Nos. 3 and 4, Bahia, H. U., D. A. Anderson, and D. W. Christensen. The Bending Beam Rheometer; A Simple Device for Measring Low Temperatre Rheology of Asphalt Binders. Jornal of the Association of Asphalt Paving Technologists, Vol. 61, 1992, pp Bahia, H. U., and D. A. Anderson. Glass Transition Behavior and Physical Hardening of Asphalt Binders. Jornal of the Association of Asphalt Paving Technologists, Vol. 62, 1993, pp