Performance of Recycled Asphalt Concrete Overlays in Southwestern Arizona

Transcription

1 3 TRANSPORTATION RESEARCH RECORD 1427 Performane of Reyled Asphalt Conrete Overlays in Southwestern Arizona MusTAQUE HossAIN, DWIGHT G. METCALF, AND LARRY A. SoFIELD Reyled asphalt onrete overlay is a routine rehabilitation strategy for most highway agenies. In 1981 the Arizona Department of Transportation onstruted an experimental asphalt onrete overlay projet on Interstate 8 in southwestern Arizona. The projet onsisted of eight test setions omparing long-term performane of reyled and virgin asphalt onrete overlays in an arid limate. Mays meter roughness, Mu meter skid number, and perentage raked data were olleted on different test setions over the servie life of the projet. A visual distress survey was onduted on eah setion at the end of servie life. Rut depth and falling weight defletometer measurements were also taken before rehabilitation of the 1981 projet. Analysis of roughness, skid, and raking data indiates that the reyled and virgin asphalt onrete overlays have performed similarly. The poor ondition of one virgin overlay setion at the end of the servie life appeared to be aused by weaker subgrade support. The thiker overlays performed better than did the thinner overlays. Considerable rutting was observed on the thiker overlay setions partly beause of densifiation of asphalt onrete mixes under traffi load. The average annual maintenane osts for different test setions were similar. Asphalt onrete overlays are the most widely used method for rehabilitation of asphalt onrete pavements in the United States. Pavement surfae preparation and type and thikness of overlay are the most important details of suh rehabilitation methods (J). The Arizona Department of Transportation (ADOT) has been ative in reyling asphalt onrete pavements sine the late 197s. Although experiene with reyled pavement was mixed, the pratie was ontinued beause the proess was eonomial. In 1981 ADOT onstruted an experimental overlay projet inorporating both virgin and reyled asphalt onrete mixes and different overlay thiknesses with and without surfae preparation by milling. The projet is in southwestern Arizona approximately 64 km ( 4 mi) west of Casa Grande extending from milepost to milepost 145. on both diretions of Interstate-8. Interstate-8 is one of the major east-west truking routes in Arizona. It is a divided highway, 11.8 m (38 ft) wide with 1.2-m (4-ft) inner and 3.1-m (1-ft) outer asphalt onrete shoulder, with two lanes in eah diretion. During a distress survey of this projet in summer 199, all seti.ons exhibited varying degrees of strutural distresses. The projet was rehabilitated in fall This paper desribes and ompares the performane of reyled asphalt onrete overlays with virgin asphalt onrete overlays after 9 years in servie. M. Hossain, Department of Civil Engineering, Kansas State University, Seaton Hall, Manhattan, Kans D. G. Metalf and L. A. Sofield, Arizona Transportation Researh Center, Arizona Department of Transportation, Phoenix, Ariz LAYOUT AND DESCRIPTION OF TEST SECTIONS Layout and Pavement Setions Eight test setions inorporating virgin and reyled asphalt onrete were onstruted on the projet. Table 1 presents the physial desription of the test setions, and Figure 1 shows the pavement ross setions. Eah type of mix was represented by four setions. Preoverlay surfae preparation was represented by "milling" and "no milling." Four setions are "mill and replae," and the rest are overlays over existing pavements. The overlay thikness on two setions is 12 mm (4 in.), and the other six have 51-mm (2-in.) overlays. Test Setions 1, 2, 5, and 6 have granular aggregate bases. Setions 3, 4, 7, and 8 have ement-treated bases. The overlays were designed for a 1-year life. As-Built Asphalti Mix Design Information Seleted information from the as-built mix designs for both reyled and virgin mixes are shown in Table 2. The binder ontent of the reyled mix was approximately.5 perent higher than.the ontent of the virgin mix, but the design virgin bitumen ontent for this mix was 3. perent. The binders used on this projet were AR-4 and AR-8 for the reyled and virgin mix, respetively. The reyled mix had 5 perent reyled asphalt pavement materials in the mix. The bulk and maximum densities of both mixes were omparable. Both mixes had 1 perent liquid antistripping agent. Soils and Geology The geologial formation of the area is weakly to moderately onsolidated alluvium of late Tertiary and early Quaternary ages. The vegetation is Sonoran desert srub in lower Colorado subdivision. The soils in this area are mostly arid soils with pedogeni horizons and low organi matter. These soils are mostly fine silts to poorly graded sand with some gravels. Climate The limate in the viinity of the projet is arid. The Strategi Highway Researh Program has designated this region as "dry no-freeze" (1). The 3-year average annual maximum temperature is 32 C (89 F), and the minimum is 13 C (56 F) (2). The 3-year annual rainfall is 148 mm (5.8 in.). The referened meteorologial data were reorded at the National Weather Servies station at Gila Bend, Arizona.

2 Hossain et al. 31 TABLE 1 Test Setion Loations and Features OVERLAY RECYCLED AC VIRGIN AC THICK. (mm) MILL and NO MILL.:ING MILL and NO MILLING REPLACE REPLACE Setion: 3 Setion: 5 Setion: 4 Setion: 6 Diretion: EB Diretion: WB Diretion: EB Diretion: WB M.P.: M.P.: M.P.: M.P.: Length: 1.61 Length:1.67 Length: 1.61 Length:1.66 km km km km Setion (mm): Setion (mm): Setion (mm): Setion (mm): 152 SB 23 SB 152 SB 23 SB 152 CTB 12 AB 152 CTB 12 AB 76 AC 76 AC 76 AC 76 AC Setion: 7 Setion: 8 Diretion: WB Diretion: WB M.P.: M.P.: Length: 1.61 Length: 1.61 km km Setion (mm): Setion (mm): 152 SB 152 SB 152 CTB 152 CTB 76 AC 76 AC Setion: 1 Setion: 2 Diretion: EB Diretion: EB M.P.: M.P.: Length: 1.67 Leng_th: 1.66 km km Setion (mm): Setion (mm): SB 23 SB 12 AB 12 AB 76 AC.76 AC Notes: 1. AB = Aggregate Base 2. AC = Asphalt Conrete 3. CTB = Cement Treated Base 4. SB = Subbase (Selet Material) 5. 1 mm =.4 in, 1 km =.63 miles Traffi History The average annual daily traffi on this projet was approximately 7,9 in 199. Figure 2 shows the umulative 18-kip equivalent single-axle loads (ESALs) on this projet sine onstrution. The umulative traffi arried by this projet was approximately 7 million ESALs. FUNCTIONAL PAVEMENT PERFORMANCE The funtional performane of the roadway an best be desribed by its servieability. In this paper, the present servieability is desribed in terms of roughness, measured by the Mays ride meter, and fritional harateristis, assessed using the Mu meter. ADOT performs an annual inventory of its highway network and reords this information on a route-milepost basis for the pavement management system (PMS). Roughness, skid, raking, pathing, rutting (for flexible pavements), and faulting (for portland ement onrete pavements) are measured in the travel lane and entered into the PMS data base. Roughness is determined by a Mays ride meter (ar) traveling at 5 mph, whih obtains ontinuous readings between mileposts. The readings are summarized in in./mi and the results assigned to the milepost loation at whih the readings begin. When the field data is obtained it is normalized to 1972 alibration values to provide onsisteny with time. Skid (frition) is determined by a Mu meter, whih is a ontinuous, reording, frition-measuring trailer. Continuous readings are obtained for a 153-m (5-ft) setion of wet pavement starting at a milepost loation. The readings are averaged and assigned to the milepost. The perent raking data are olleted at every milepost loation by omparing the pavement ondition of a 93-m 2 (1-ft 2 ) pavement setion with a standard set of photographs showing different perentages of raking. To use these data, the following table shows milepost loations hosen to be representative of the test setions (where EB is eastbound and WB is westbound). Test Setion (Number and Diretion) 1, EB 2, EB 3, EB 4, EB 5, WB 6, WB 7, WB 8, WB Milepost

3 32 TRANSPORTATION RESEARCH RECORD 1427 Test Setions 1, 2, 5, and 6 are situated within a milepost. The Mays roughness data for these setions are tainted beause of the data olletion proess. The years for whih the performane data are available in the PMS data base for different attributes are shown in the following table. Attribute Mays roughness Mu meter skid Perent raking Period of Data , 1987, 1989, To find the rate of hange of the attributes over the servie lives of the test setions, linear regression models were developed for eah attribute (exept rut depth) taking the orresponding attribute as the dependent variable and age sine onstrution (in 1982) as the independent variable. The linear model was hosen on the basis of the value of F-ratio in the analysis of variane for signifiane of regression. These models were subsequently used for analysis of roughness, skid resistane, and raking history of different setions. Roughness (a) (b) () (d) (e) (f) Nole: AC = Asphalt Conrete, AB = Aggregate Base, SM = Selet Material /Subbase CTB = Cement Treated Base FIGURE 1 Pavement ross setions: (a) Setion E, (b) Setion 2, () Setion 5, (d) Setion 6, (e) Setions 3 and 7, (f) Setions 4 and 8. Table 3 gives the roughness history of the test setions. The average Mays meter roughness before rehabilitation (in 199) for setions with reyled overlay is 139 m/km (88 in./mi) ompared with 151 m/km (96 in./mi) for setions with virgin overlay. However, before onstrution of experimental test setions, average roughness of virgin setions (336 m/km) was muh higher than that for reyled setions (267 m/km). The average as-built Mays roughness for the reyled setions was between 65 m/km (41 in./mi) and 9 m/km (57 in./mi) with an average of m/km (5 in./ mi). The orresponding values for virgin overlay setions are 52 m/km (33 in./mi), m/km (49 in./mi), and 66 m/km (42 in./mi). The rate of inrease of roughness for all setions was between 9 and 17 m/km/year (6 and 11 in./mi/year) with the virgin setions showing slightly higher rate of inrease. The 12-mm ( 4-in.) overlay setions were built to be onsiderably smoother than 51-mm (2-in.) overlays and remained so at the end of the servie lives. The mill and replae setions had average roughness omparable to simple overlay setions before rehabilitation in Average preoverlay roughness of mill and replae setions was onsiderably higher: 357 m/km (226 in./ mi) ompared with 226 m/km (143 in./mi) for nonmilled setions. TABLE 2 As-Built Mix Design Information Aggregate Density (kn/m 3 ) Marshall Flow lmmer- Asphalt Cement Stability (%) sion - (N) Com pression Retention Coarse Inter- Fine Passed Bulk Maximum (%) mediate (%) #2 (%) (%) (%) Content Type Visosity (%) Range (Pa.sl (a) Virgin Mix (b) Reyled Mix AR , Notes: i) Both mix used 1 % liquid anti-stripping agent (pave bond speial) ii) Aggregate for reyle mix used 5% reyled aggregate. iii) Design bitumen ontent in reyled mix is 3.% iv) 1 kn/m 3 = lb/ft 3, 1 N = lbs, 1 poise=.1 Pa.s AR

4 Hossain et al. Skid Resistane There is no apparent differene between 199 Mu meter values of setions with virgin and reyled overlays (Table 4). The Mu meter values for setions with virgin and reyled overlays at the end of servie lives are 66 and 65, respetively. The rate of derease of skid resistane, as indiated by the rate of derease of the Mu meter value, is similar for both types of overlay. As an be expeted, the skid resistane of 12-mm (4-in.) and 51-mm (2-in.) overlays as well as setions with mill and replae and no-milling were similar at the end of servie lives STRUCTURAL PAVEMENT PERFORMANCE 7 6 :E. 5 <( Cf) w 4 to (]) 3 > :; E 2 ::J (.) Year Sine Constrution FIGURE 2 Traffi history of projet The strutural ondition of the pavement was evaluated by pavement distress surveys and falling weight defletometer testings. In addition to the historial rak data available in ADOT's PMS data base, pavement distress surveys were onduted using the PA VER (3) in 199. Rut depth measurements were also taken on eah setion with a 3-m (1-ft) straight edge. Craking Table 5 shows the raking history of all setions. The two reyled setions in the eastbound diretion had more raking in 199 than did the virgin setions of idential struture. However, in the westbound diretion, the reyled and virgin overlays with idential strutural setions had similar perentages of raking. The average raking for 12-mm ( 4-in.) TABLE 3 Test Setion Roughness History Test Pre-ex pt. As-built Rate of Current Setion Overlay Roughness Roughnss Roughness Roughness (1982) Inrease (199) (m/km) (m/km) (m/km) (m/km) (1981) 1, EB , EB , EB (eyled) 4, EB , WB , WB , WB , WB reyled virgin mm mm No-milling Mill & Pave

5 34 TRANSPORTATION RESEARCH RECORD 1427 TABLE 4 Test Setion Skid Resistane History Test Pre-expt. As-built Mu- Rate of Mu- Current Setion Overlay meter Value meter Value Mu-meter Mu-meter Value (1981) (1982) Derease (units/year) Value (199) 1, EB 72 2, EB 75 3, EB 69 4, EB 67 5, WB 73 6, WB 7, WB 8, WB reyled 73 virgin mm mm No-milling 73 Mill & Pave overlays was muh less than for 51-mm (2-in.) overlays. On the average, the rate of inrease of raking was slightly higher for setions with virgin overlays than for setions with reyled overlays. Although the setions with mill and replae strategy had higher preoverlay raking, they showed less raking in 199 ompared with setions with no milling. Visual Distress Survey A PA VER ondition survey was onduted on eah test setion in May 199. Approximately 3 perent of the area in eah test setion was surveyed. The survey onsisted of observing 19 distress types on asphalt onrete pavements on 61-m (2-ft) sample units The sample units were hosen systematially on eah setion; the first sample unit was hosen at random. Five rut depth measurements were taken on eah sample unit at 15-m (5-ft) intervals using a 3-m (1-ft) straight edge and a ruler. On average, 43 readings were taken on eah setion. The predominant distresses were found to be alligator. raking, longitudinal and transverse raking, blok raking, and weathering. Pavement ondition index (PCI) was alulated for eah sample unit using the Miro-PA VER program (3). Rut depths were omitted from the alulation of PCI beause of the diffiulty assoiated with defining the atual rutted area. Paver Survey Results Table 6 shows the results of the PA VER survey for eah test setion. From the results it is evident that, one rutting is ignored, the load-assoiated distresses fairly dominated the PCI values for most of the test setions. The limate,.assoiated dedut values (for blok raking and weathering) also ontributed onsiderably to the lower PCI values of all setions. The.average PCI value for Setion 1[reyled12-mm (4-in.) overlay] is 71, slightly higher than for Setion 2 [virgin 12- mm (4-in.) overlay]. For 51-mm (2-in.) overlays, the mean PCI value for reyled setions is 42 ompared with 29 for virgin overlays. Students' t-tests onduted between the means of PCI values for virgin and reyled overlays showed no signifiant differene at 5 perent level of signifiane. The 12-mm (4-in.) overlays have performed better than 51-mm (2-in.) overlays. A signifiant differene was found between the means of PCI values for 51-mm (2-in.) and 12-mm (4- in.) overlays at a 5 perent level of signifiane. The mean PCI value for the mill and replae setions with 51-mm (2- in.) overlays 3 and 4 is 41, whih is almost-equal to the mean value, 39, for the existing Setions 5 and 6 with 51-mm (2- in.) overlays. Again, the mean PCI value for setions with reyled mix with mill and replae strategy was 44 ompared with 24 for overlay setions with virgin mix on the existing pavement.

6 Hossain et al. 35 TABLE 5 Test Setion Craking History Rut Depth Measurements Test Setion Pre-expt. Overlay Craking (1981) (%) 1, EB 6 2, EB 15 3, EB 5 4, EB 15 5, WB 15 6, WB 25 7, WB 3 8, WB 12 reyled 25 virgin mm mm 25 No-milling 15 Mill & Pave '27 Rate of Craking Progression {%/year) Current Craking (199) (%) The mean rut depth values for different setions are also shown in Table 6. The average values vary from 4 mm (.15 in.) to 13 mm (.5 in.). The 12-mm (4-in.) overlay setions have higher rut depths than do 51-mm (2-in.) overlays. Densifiation of asphalt onrete in the wheelpaths of the 12- mm ( 4-in.) overlays was suspeted to be responsible for exessive rutting on these setions. Test setions with 12-mm ( 4-in.) overlays exhibited higher rut depths than did setions with 51-mm (2-in.) overlays. The mean rut depths reorded on Setions 1 and 2 were 13 mm (.5 in.) and 12 mm (.48 in.), respetively. The orresponding values for 51-mm (2-in.) reyled and virgin overlay setions were 6 mm (.22 in.) and 5 mm (.2 in.), respetively. Bulk Density and Air Voids As mentioned earlier, the 12-mm ( 4-in.) overlay setions had higher rut depths than did 51-mm (2-in.) overlay setions. Densifiation of asphalt onrete mix in the wheelpaths on the 12-mm ( 4-in.) overlays was suspeted to be responsible for this exessive rutting. To verify this, bulk density test results from five ores obtained from the between wheelpath loations were ompared with the test results from five ores retrieved from the outer wheelpath loations for Setions 1, 2, 5, and 6. The summary statistis for bulk density of the reyled mix for Setions 1 and 5 for both loations are shown in Table 7. Table 7 also presents the summary statistis of perent air voids of the virgin mix in Setions 2 and 6 for both loations. Students' t-tests between the means of these pa- TABLE 6 PA VER and Rut Depth Survey Results Load Setion PCl 1 Asso. Number Mean S.D. Dedut.v. Values (%) (%) 1, EB , EB , EB , EB , WB , WB , WB , WB Climate Other Average Asso. Dedut Rut Dedut Values Depth Values (%) (mm) (%) Notes: 1. Exludes rutting 2. 1 mm =.4 in 3. Reyle PCI = 49 Virgin PCI = mm overlay PCI = 7 51-mm overlay PCI = 36 No-milling PCI = 55 Mill & Pave PCI = 34

7 36 TRANSPORTATION RESEARCH RECORD 1427 TABLE 7 Perent Air Voids and Bulk Density for Setions with Aggregate Base Setion No. Outer Wheel Path Air Voids (%) Between Path x u n x u 1, EB (Reyle) Bulk Density (kn/m 3 ) Wheel Outer Wheel Path Between Wheel Path n x u n x u n , EB (Virgin) , WB (Reyle) , WB (Virgin) Note: 1 kn/m 3 = lb/ft 3 rameters for the two loations show signifiant differene at 1 perent level of signifiane. The t-test results indiate the densifiation in the outer wheelpath loations for all setions. Densifiation of asphalt onrete mixes on these setions might have ontributed to the observed higher rut depths. Again, Table 7 indiates that lower amount of densifiation had ourred on Setion 6, whih had also the lowest mean rut depth value (4 mm, or.15 in.). Defletion Testing Defletion data were olleted at five random loations on eah of the eight test setions with a Dynatest 8 falling weight defletometer in May 199. Seven sensors were used, with the first sensor at the enter of the loading plate and six others at a uniform radial distane 35 mm (12 in.) apart. Three drops of falling weight defletometer load were used for target loadings of 27 kn (6, lb), 4 kn (9, lb), and 67 kn (15, lb). Bakalulation Results The subgrade moduli were bakalulated from falling weight defletometer data using an elasti layer analysis bakalulation sheme, BKCHEVM (4). Table 8 gives the average and range of subgrade moduli for eah test setion. The estimated subgrade moduli were unusually high. The effet of rigid bottom was not fully taken into aount in the bakalulation proess. This may have had an effet on the bakalulated subgrade moduli. The table indiates that test Setion 8 has the lowest average subgrade modulus among all test setions. The lower range of subgrade moduli on this test setion might be responsible for the extensive alligator raking observed on this setion during the PA VER survey. TABLE 8 Test Setion No. and Diretion Test Setion Subgrade Moduli Subgrade Modulu., (MPal Mean Range 1, EB , EB , EB , EB , WB , WB , WB , WB Note: 1 MPa =. 145 ksi Material Related Problems Beause ADOT's PMS does not report any material related distress suh as stripping, no reords are available on the development of this distress. However, prerehabilitation inspetion of this projet did not reveal any stripping problem. MAINTENANCE COSTS Signifiant maintenane ativities were performed on this projet just before the rehabilitation projet. A fog seal oat was applied to the projet in Figures 3 and 4 show the reported annual average maintenane ost per mile from the PMS data base for different strategies from 1981 through 199. As noted, very high maintenane osts are evident during 199 just before the rehabilitation. The average annual maintenane osts for all strategies were similar.

8 Hossain et al. 37.E iii C) Q) (.) o.2l ro E ii5 C) (J) C) o (J) ro Year I 4-in ovly ml 2-in ovly FIGURE 3 Average annual maintenane osts for overlays with different thiknesses Year J Reyle - Virg_in D_ _:_N_o_-_m_il_lin_g_._ - _M_il_I&_p_av_e, FIGURE 4 Average annual maintenane osts for overlays with different mixes. CONCLUSIONS The test setions for the experimental overlay projet, designed for a 1-year life, have performed satisfatorily throughout the servie period. The funtional performanes of reyled and virgin mix overlays were similar. The 12-mm (4-in.) overlays were built to be smoother than 51-mm (2-in.) overlays and remained so at the end of their servie lives. The 12-mm ( 4-in.) overlays performed slightly better than did the 51-mm (2-in.) overlays when evaluated by the load-assoiated distresses (exept rutting). However, the amount of raking on the thiker overlays was onsiderably less than on the thinner overlays. The average rut depth on the 12-mm ( 4-in.) overlays was about twie that on the 51-mm (2-in.) overlays. Densifiation of both types of mixes under traffi wheel load was mostly responsible for rutting on the thiker overlays. The virgin mix had slightly less rutting than the reyled mix. The mill and replae strategy before overlay plaement did not appear to provide more added life than did the simple overlay strategy. Although the setions using mill and replae strategy were originally in worse ondition, these setions performed better than did the simple overlay setions. For the design using the mill and replae strategy before overlay, the reyled mix outperformed the virgin mix. After 1 years, both the 12-mm (4-in.) and 51-mm (2- in.) overlays required rehabilitation, whih suggests that the 51-mm (2-in.) overlay design strategy was the most ost effetive. The average annual maintenane osts for different strategies were similar. ACKNOWLEDGMENTS The authors aknowledge finanial help for this study from the Arizona Transportation Researh Center. The PA VER survey was performed by Sylvester Kalevela and Greg Rollins of Arizona Transportation Researh Center. REFERENCES 1. Speifi Pavement Studies; Experimental Design and Researh Plan for Experiment SPS-5: Rehabilitation of Asphalt Conrete Pavements. Strategi Highway Researh Program, Washington, D.C., Climate of the States ( ). National Oeani and Atmospheri Administration, U.S. Department of Commere, Silver Spring, Md., Shahin, M., and S. D. Kohn. Pavement Maintenane Management for Roads and Parking Lots. Tehnial report M-294. Constrution and Engineering Researh Laboratory, Champaign, Ill., Mamlouk, M. S., W. N. Houston, S. L. Houston, and J.P. Zaniewski. Rational Charaterization of Pavement Strutures Using Defletion Analysis. Report FHWA-AZ88-254, Vol. 1. Arizona Department of Transportation, Phoenix, De The ontents of this paper reflet the views of the authors, who are responsible for the fats and the auray of the data presented herein. The ontents do not neessarily reflet the offiial_ views or poliies of ADOTor FHWA. Publiation of this paper sponsored by Committee on Charateristis of Bituminous-Aggregate Combinations To Meet Surfae Requirements.