Performance of pavement crack sealants in cold urban conditions

398 Can. J. Civ. Eng. Vol. 26, 1999Table 5. Opening of routs in a 1-year cycle (in mm). *RoutMeanwidth Orientation † Min. Max. Range Mean 1 ‡ Mean 2 § opening ,%12 T –0.17 0.63 0.80 1.82 1.73 14 (6)–0.24 1.25 1.49–0.54 0.81 1.35–1.29 1.61 2.90 24–0.68 1.89 2.57L –0.00 1.24 1.24 1.62–0.24 0.66 0.90–0.95 0.99 1.94–0.24 2.17 2.4119 T –0.74 2.06 2.80 3.09 2.83 15–1.04 2.47 3.51–0.53 1.81 2.34–0.71 1.97 2.68–1.65 2.81 4.46 23–0.40 2.33 2.73L –0.33 1.70 2.03 2.06 (11)–0.55 1.54 2.0940 T –0.40 3.37 3.77 3.62 3.41 8–0.63 3.79 4.42 11–0.37 2.50 2.87–1.39 1.57 2.96–0.41 2.98 3.39–0.91 3.40 4.31L –1.12 1.25 2.37 2.79 (6)–0.61 2.60 3.21* From opening in April when the weekly average temperature was 7°C.† T, transverse; L, longitudinal.‡ Mean according to orientation and rout width.§ Mean according to rout width. Mean percent opening (mean 2 / rout width).Max. (min.)opening, %rout size combination was compared and qualified as low,medium, or high. The results are shown in Table 6. Unexpectedly,they revealed that sealants in routs of 12 × 12 mm 2debonded less than sealants in routs of 40 × 10 mm 2 .Based on numerical models (Kuri and Tons 1992; Tons1962; Wang and Weisgerber 1993), sealants in 40 mm widerouts, with a width-to-height ratio, W/H, of 4 should haveperformed better than sealants in 12 or 19 mm wide routs,with W/H = 1. Factors peculiar to field conditions, not includedin the models, are most likely responsible for the discrepancybetween the observed and expected performances.Some factors can readily be excluded. Table 5 shows, for example,that differences in percent crack opening and associatedtensile stresses cannot account for the discrepancy.Sealants in 40 mm wide routs had been exposed to only halfthe extension of other sealants. Sealant adhesion at the routbottom also does little to explain the inconsistency, as it canonly account for greater cohesive failure due to an increasein stress immediately above the crack (Wang and Weisgerber1993). Moreover, the relative adhesion area amongst therouts had been held constant by adjusting the ratio routwidth to crack width to about 3. Consequently, factors otherthan sealant adhesion to asphalt concrete and the tensile(compressive) stress induced by the crack opening (closing)influenced sealant performance.Sealant aging and unaccounted shear stresses at the sealantsurface can explain the discrepancy between the observedand expected performances. Wider sealants aresubjected to more weathering and may age more quickly.Sealants can lose a substantial amount of plasticizing oilwithin a single year (Masson and Lacasse 1997) and the relativeloss of oil may be greater in the largest sealant. In theurban setting, the shear stress at the sealant surface possiblyplay a dominant role, however. With the current practice, thesealant is most often leveled with the pavement surface.Consequently, it may be sheared by tires from vehicles thatmove slowly, stop, accelerate, and turn while in contact withthe sealant surface. The large surface and the exposure ofthe 40 mm wide sealant to shear stresses would also explainits tendency towards greater pull-out levels (Table 6).Performance trendsIn service, sealants do not remain unchanged as they areexposed to water, sunlight, and freezing and thawing cycles.They weather like other bituminous materials (Minkarah etal. 1992; Petersen 1984). As a result, they may harden(Bahia and Anderson 1991) and loose elasticity in cold temperatures.For that reason, sealants are sometimes selectedbased on field tests, the one-year test being most common(Pearson and Lynch 1992). Figure 1, with a trend typical of© 1999 NRC Canada

Masson et al. 399Table 6. Frequency of failure levels in differentrouts. *Debonding † Pull-out †Sealant 12 19 40 12 19 40A L M H L M HB L M H L H HC M L H H M LD L M H L H ME L M H H L HF M L H H L MG L M H M L HH M L H H L MJ M L H H L MK L M H L H HL H L H M H LM M L H L H HTotalsH 0 0 12 4 5 6M 5 6 0 2 2 4L 6 6 0 5 5 2* 12:12×12mm 2 ;19:19×19mm 2 ;40:40×10mm 2 .† L, M, and H represent low, medium, and high failurelevels, respectively.Fig. 1. Typical failure profile for a sealant in the various routs.the observed time-dependent performance, shows that sealantfailure did not increase linearly and that a one-year testwould have failed to accurately predict a longer-term performance.The increase in sealant failure could typically be dividedin three stages. The initial stage, a steeply rising slope,covered the first year of service. This stage was followed bya period of nearly two years where pull-out lengths remainedfairly constant. In the final stage, from 32 months to44 months in the example, the failure increased again, withthat in the 12 mm wide rout exceeding that in the others.The change in sealant performance is evidently morecomplex than anticipated. The grouping of 66 failure curves(eleven sealants, three rout geometries, two orientations)such as those in Fig. 1 allowed for highlighting recurrenttrends. These are shown in Figs. 2 and 3. Pull-out trendswere labeled P1 to P4 (Fig. 2) and debonding trends were labeledD1 to D4 (Fig. 3). In the best case, failure levels remainedlow and increased quasi-linearly (P1, P2, D3). Incontrast, failure could reach a plateau after a one- or twoyearperiod of rapid failure (P3, D4). In cases where finalfailure levels were high (D1, D2, P4), initial sealant failurewas high and the plateau was short.The exact reasons for the changes in failure rates remainunknown, but they are likely linked to specific componentsin the sealant – asphalt concrete system, namely, the interfaceand the sealant itself. Excessive sealant stiffness and aweak interface may be responsible for the initial and rapidincrease in debonding. A weak interface may result from interfacialdefects such as microcracks at the asphalt concretesurface and voids due to dust or incomplete sealant wetting.Sealant weathering and stiffening may be responsible for thelater increase in debonding. On this ground, an extendedfirst phase would be characteristic of stiff sealants, or alternativelyof an aged and brittle pavement, whereas an extendedsecond phase, a long plateau, would be characteristicof a better sealant.Sealant selection for long-term performanceTo estimate long-term performance and rank the twelvesealants, the 4-year pull-out and debonding levels were considered.For that purpose, a performance index was calculated.A simple addition of the respective debonding andpull-out levels would have implied that both types of failureswere equally damaging to pavements, and would haveunderestimated the effect of pull-outs on pavement degradation.Consequently, pull-out lengths were given more weightthan debonding lengths in the determination of a performanceindex. To that effect, the following equation wasused:PI=100–(D + nP)where PI is the sealant performance index; D is the percentdebonded length of the sealant; P is the percent pull-outlength; and n is an integer that accounts for the effect ofpull-outs over debonding on performance.Given that the absence of sealant over 1mofcrack mayallow the ingress of sand and stones that can damage thepavement during its expansion (Peterson 1982), and also allowthe penetration of much more water than a simple sealantdebonding over the same length, a value of n = 4 wasthought reasonable. The approach is admittedly simplisticand the choice of n subjective, but it allowed us to fairly appreciatethe performance of the sealants and it made for aneasy comparison of the individual performances (values of n< 4 provided a similar ranking of materials but the differenceof PI between the sealants was somewhat reduced).Accordingly, sealants could be ranked in decreasing orderof field performance and the performance compared toASTM test results (Table 7). Little correlation between fieldand laboratory test results could be found, which is consistentwith the work of others (Lynch and Janssen 1997). Seal-© 1999 NRC Canada

400 Can. J. Civ. Eng. Vol. 26, 1999Fig. 2. Pull-out trends for sealants in Montreal. The trendsoccurred with equal frequency.Table 7. Sealant acceptance based on standard andfield tests.SealantPerformanceindex4-yearperformanceH 75 Good NoB 74 YesE 72 NoF 65 Average NoM 64 YesL 54 YesD 54 YesJ 39 Poor NoK 35 YesA 33 NoC † 10 Very poor NoG 8 No* From Table 1.† After one year of service.ASTMacceptance *Fig. 3. Debonding trends for sealants in Montreal. Occurrenceswere 40%, 8%, 33%, and 19% for trends D1, D2, D3, and D4,respectively.ants H and E, which showed good field performance, failedto meet the requirements of the specification. In contrast,Sealant K, which met the standard requirements, showedpoor field performance. In the end, average performers (e.g.,Sealants M, L, D) were more likely to meet the ASTMD3405 specification than sealants at the performance limits,the good and the poor sealants. The usefulness of the specificationin selecting good sealants, those required for demandingconditions, may thus be questioned. The results forpenetration can be especially misleading. It has been shownby Lu and Isacsson (1997) that there is little correlation betweenthe performance of polymer-modified bitumens, e.g.,crack sealants, and their penetration values. The existingspecification limits the penetration to 90 dmm but three ofthe four best performers in this study exceeded that limit.ConclusionsTwelve bituminous crack sealants were installed in routsof12×12mm 2 ,19×19mm 2 ,and40×10mm 2 (width ×depth), and evaluated over 4 years in an urban setting withair temperatures ranging from –40°C to +40°C. The pavement,in which crack opening ranged from 6% to 24%, wasa bituminous overlay with a concrete base. The short-term(1-year) and mid-term (4-year) debonding and pull-out levelswere compared; the effect of rout size on performancewas determined; failure trends were identified, and overallperformance was compared with standard test results.Sealant performance was found to vary tremendouslyfrom one product to the next. The four-year debonding levelsvaried from 0 to 50% whereas pull-out levels varied from0 to 30%, depending on sealant shape (rout size) and crackorientation. Relative failure levels often correlated with sealantwidth, overall performance increasing in the order40 mm < 19 mm < 12 mm. The trend follows the increasedexposure of sealants to weathering and slow moving traffic.Sealant debonding and pull-out did not increase linearly,as three stages were identified in the time–failure curve. Instage 1, failure increased rapidly. In stage 2, little failure occurredand in stage 3, failure increases again but at a slowerrate than in stage 1. Failures in stage 1 are thought to be relatedto a weak sealant – asphalt concrete interface whereasthose of stage 3 are thought to be related to sealant weatheringand stiffening.It was also found that sealants with either good or poorfield performance failed to meet the requirements of theASTM D3405 specification. Sealants demonstrating averageperformance were more likely to meet the specification.In corollary to this study, it can be said that crack sealingis still an evolving technique. This method of preventivemaintenance has been shown to extend the service life ofrapid ways, which translates into economic benefits. These© 1999 NRC Canada

Masson et al. 401benefits are identical in the urban setting, but the serviceconditions are different. For maximum benefits, it is requiredthat sealant installation be adapted to urban conditionsand that installations suitable for rapid ways not beused indiscriminately in the city, the rout size being a casein point. The existing sealant specification also does little toprovide maximum benefits. To select sealants adapted to thedemanding cold urban conditions, a performance-basedspecification is required.ReferencesBahia, H.U., and Anderson, D.A. 1991. Isothermal lowtemperaturephysical hardening of asphalt cements. Proceedingsof the International Symposium on Chemistry of Bitumens,Rome, Italy, Vol. 1, pp. 114–147.Belangie, M.C., and Anderson, D.I. 1985. Crack sealing methodsand materials for flexible pavements. Utah Department of Transportation,Salt Lake City, Utah, Report FHWA/UT-85/1.Chai, J. 1989. Crack sealing test program. 1986 test sealants, finalreport. Pavement Managements Systems, The City of EdmontonTransportation Department, Edmonton, Alta.Chehovits, J., and Manning, M. 1984. Materials and methods forsealing cracks in asphalt concrete pavements. TransportationResearch Record 990, Transportation Research Board, Washington,D.C., pp. 12–19.City of Montreal. 1991. Specifications for crack sealing inMontreal. Public Works Service, Montreal, Que.Eaton, R.A., and Ashcraft, J. 1992. State-of-the-art survey of flexiblepavement crack sealing procedures in the United States.Cold Regions Research & Engineering Laboratory, U.S. ArmyCorps of Engineers, Hanover, N.H., Report 9218.Erickson, D.E. 1992. Crack sealing effectiveness. WashingtonState Department of Transportation, Washington, D.C., ReportWA-RD256.1.Evers, R.C. 1981. Evaluation of crack-sealing compounds forasphaltic pavements. Ontario Ministry of Transportation andCommunications, Toronto, Ont., Project No. 33, Interim ReportNo. 1.Hubrecht, L. 1986. Réparation des joints de construction dans lesrevêtements asphaltiques. Centre de recherches routières,Brussels, Belgium.Khuri, F.M., and Tons, E. 1992. Comparing rectangular and trapezoidalseals using the finite element method. TransportationResearch Record 1334, Transportation Research Board, Washington,D.C., pp. 25–37.Knight, N.E. 1985. Sealing cracks in bituminous overlays of rigidbases. Transportation Research Record 1041, TransportationResearch Board, Washington, D.C., pp. 75–81.LCPC. 1981. Scellement des fissures. Laboratoire Central desPonts et Chaussées, Paris, France.Lu, X., and Isacsson, U. 1997. Characterization of styrenebutadiene-styrenepolymer modified bitumens — comparison ofconventional methods and dynamic mechanical analyses. Journalof Testing and Evaluation, 25(4): 383–390.Lupien, C., Roireau, M., and Vézina, D. 1987. Crack sealing: anevaluation of a few compounds and a variety of applicationconditions. Paving in Cold Areas, 1(3): 689–718.Lynch, L.N., and Janssen, D.J. 1997. Sealant specifications: past,present, and future. Proceedings of the Pavement Crack andJoint Sealants for Rigid and Flexible Pavements Conference,United States Army Engineer Waterways Experiment Station,Airfields and Pavements Division, Vicksburg, Miss., pp. 5–23.Masson, J-F. 1997. Effective sealing of pavement cracks in cold urbanenvironments. Institute for Research in Construction,National Research Council Canada, Ottawa, Ont.Masson, J-F., and Lacasse, M.A. 1997. Towards a performancebasedspecification for bituminous crack sealants used in urbanconditions. Proceedings of the Pavement Crack and JointSealants for Rigid and Flexible Pavements Conference, UnitedStates Army Engineer Waterways Experiment Station, Airfieldsand Pavements Division, Vicksburg, Miss., pp. 152–167.Masson, J-F., and Lacasse, M.A. 1999. The effect of the hot-airlance on the adhesion of sealants used in roadway maintenance.Journal of Transportation Engineering, 125: In press.Masson, J-F., Lauzier, C., Collins, P., and Lacasse, M.A. 1998.Sealant degradation during the crack sealing of pavements. Journalof Materials in Civil Engineering, 10(4): 250–255.Minkarah, I., Cook, J.P., and Rajogopal, A.S. 1992. Applicabilityof artificial weathering tests for performance evaluation ofelastomeric sealants. In Science and technology of buildingseals, sealants, glazing, and waterproofing. Vol. 2. ASTM STP1200. Edited by J.M. Klosowski. American Society for Testingand Materials, Philadelphia, Pa.Pearson, D.C., and Lynch, D.F. 1992. The quality of rubberizedasphalt crack sealer for asphalt pavements supplied in 1991.Ontario Ministry of Transportation, Toronto, Ont., Report MI-154.Petersen, J.C. 1984. Chemical composition of asphalt as related toasphalt durability: state of the art. Transportation Research Record999, Transportation Research Board, Washington, D.C.,pp. 13–30.Peterson, D.E. 1982. Resealing joints and cracks in rigid and flexiblepavements. NCHRP Synthesis of Highway Practice 98,Transportation Research Board, Washington, D.C.Quebec Ministry of Transportation. 1996. Specifications for cracksealing in the Province of Quebec. Québec, Que.Smith, K.L., and Romine, A.R. 1993. Materials and procedure forsealing and filling cracks in asphalt-surfaced pavements. StrategicHighway Research Program, National Research Council,Washington, D.C., Report SHRP-H-348.Tessier, G.R. 1990. Guide de construction et d’entretien deschaussées. Association Québecoise des Travaux Routiers,Montreal, Que.Tons, E. 1962. Geometry of simple joint seals under strain. Newjoint sealants: criteria, design, and materials. PublicationNo. 1006, Building Research Institute, Washington, D.C.,pp. 41–61.Wang, C.P., and Weisgerber, F.E. 1993. Effects of seal geometryon adhesive stresses in pavement joint seals. TransportationResearch Record 999, Transportation Research Board,Washington, D.C., pp. 64–70.Ward, D.R. 1993. Evaluation of crack sealant performance onIndiana’s asphalt concrete surfaced pavements. Indiana Departmentof Transportation, Indianapolis, Ind.© 1999 NRC Canada