Colombian Experience with Accelerated Testing of ...

Colombian experience with Accelerated Testing of Geogrid-Reinforced Flexible Pavement

Fredy Reyes Associate Professor Universidad Javeriana Bogota- Colombia Tel 57-1-3208320 ext 5270 [email protected]

Erwin Kohler Project Scientist Department of Civil Engineering University of California Davis One Shields Ave. Davis, CA 95616 Tel.(530) 754.8699 [email protected]

Transportation Research Board 84rd Annual Meeting January 2006 Washington, D.C.

TRB 2006 Annual Meeting CD-ROM

Paper revised from original submittal.

Colombian experience with Accelerated Testing of Geogrid-Reinforced Flexible Pavements

ABSTRACT Four experimental pavements section were tested at a new full-scale accelerated loading facility in Colombia. The sections consisted on three flexible pavements with geogrid reinforcement at different depths in the structure, plus a control section. The four pavement structures were constructed using the same materials under a unique design, varying only by the arrangement of the geogrid, which was placed on the top of the subgrade, at middle of the base layer, and on a double configuration on top of the subgrade and the middle of the base. The loading equipment is known as the Fatigue Carousel, and for this test it applied a total of 100,000 repetitions over the four sections, which are arranged in a circular track. Extensive material characterization was carried out and is presented in the paper. The results indicate that the best fatigue response in a geogrid reinforced flexible pavement is obtained when the geogrid is placed on the base or on both the base and top of the subgrade. The paper includes performance evaluation in terms of deflection and surface distresses.



INTRODUCTION Accelerated pavement testing provides valuable information for improving the mechanistic design of pavement structures. Analytical models and laboratory testing are often insufficient to reasonably obtain an accurate prediction of pavement life, mainly because of the difficulties involved in simulating the effects of wheel and environmental loads on material behavior and layer interactions. Because of this limitation there has been a worldwide trend toward the use of accelerated pavement tests, particularly fullscale testing (Hugo and Epps-Martin 2004; Romero 1996).

As part of this trend,

Colombia’s University of Los Andes constructed a pavement test track in Bogotá, and subjected four experimental sections to accelerated wheel loading with a circular pavement loading device. Three of the four sections include geogrid reinforcement at different depths in the structure while the fourth section is a conventional flexible pavement utilized as control. All sections have a 180mm surface course asphalt concrete placed on 300mm of granular base, and 300mm of granular subbase. Details on material characterization are presented later in this paper. In the first pavement section the geogrid was placed at middepth in the granular base. The second section has the geogrid located on top of the subgrade, instead of in the base. The third pavement section uses two geogrid layers, one placed on top of the subgrade and the other at middepth in the granular base. The same type of geosynthetic reinforcement material was used in the three sections, and since materials and thicknesses are the same in all cases, the only variable is the reinforcement location. The objective of the experiment was 1) to determine actual pavement life in terms of the number of load repetitions to failure and compare it to the design service life, and 2) Evaluate the benefit of geogrid reinforcement at the different locations. This paper describes the test track, pavement materials and construction of the experimental sections, and presents the results of deflection testing and distresses observed along the study.



EXPERIMENTAL SET UP Material characterization The first stage in the experimental set up consisted in characterizing materials subgrade, granular subbase and base, asphalt concrete, and geogrid reinforcement. The subgrade was classified as MH A-7-5(82) soil, and its resilient modulus adopted at 19,600 kPa based on laboratory test results. Figure 1 shows resilient modulus for subgrade, subbase, and base. The granular subbase was GM A-1-a(0) (P200 = 13.6%, LL=19.9, IP=5.5), with optimum moisture content of 8.6 percent. The LA test indicated for this material 27 to 33 percent abrasion. The base material, obtained from the same quarry, was also GM A-1-a(0), with 96 percent fractured faces. The flatness- and elongation ratios were found to be 18 and 24 percent respectively. The base resilient modulus was 343,200 kPa. Maximum aggregate size for subbase and base was 20 and 31.5mm respectively.



a) Resilient modulus (kPa)

1,000,000 100,000 10,000 1,000 y = 229338x -0.5988 100

b) Resilient modulus (kPa)

1,000,000

10

100

1,000

Suma de esfuerzos principales (kPa) 100,000 10,000 1,000 y = 120605x 0.0724 100 1,000,000 10

Resilient modulus (kPa)

c)

100

1,000

Suma de esfuerzos principales (kPa) 100,000 10,000 1,000 y = 48726x 0.3504 100 10

100

1,000

Sum of principal stresses (kPa)

Figure 1. Resilient modulus versus sum of principal stresses for a) subgrade, b) subbase, and c) base materials

The bitumen used for the AC had a penetration value of 70-72 (mm/100). The softening point was 48.9 to 50 °C, and the ductility greater than 1500 mm. The optimum asphalt content for the mix from the Marshall method was 6.3 percent, with 5.35 percent voids. The dynamic modulus of the mix is presented in Figure 2 at three temperatures and three frequencies. The fatigue relationship is presented in Figure 3.



7,000

Modulus (MPa)

6,000 5,000 16 Hz 4Hz 1 Hz

4,000 3,000 2,000 1,000 0

10

20

30

40

50

Temperature ºC

Figure 2. Dynamic modulus of the mix

Tensile deformation (microstrain)

1,000

100

y = 8977x-0.3651 10 10,000

100,000 No. of cycles

1,000,000

Figure 3. Tensile strain versus number of cycles

The geosynthetic reinforcement material utilized in the test is a Tensar BX1200, which is a polypropylene grid with apertures size of 25 by 36 mm. Strength at failure, in the machine and cross-machine directions of the material, are 21 and 31 kN/m respectively, as reported by the manufacturer. Design traffic of 100,000 repetitions was assumed for the test sections, at the standard load level of 80 kN (1 pass = 1 ESAL). Specific climatic conditions for the city of Bogotá were used, and consisted on 15°C of average temperature and the rainfall for the three months of expected testing. The admissible tensile strain at the bottom of the



asphalt and compression strain at the subgrade were determined as 118 and 1,242 microstrain respectively, using the pavement design software applications Alizé III (Autret et al., 1982; LCPC-SETRA, 2003)

Equipment for Accelerated Loading The test track consists of a ring 3.5 meter wide by 35 meter long (measured at the center of the lane). The carousel equipment can load the pavement at a rate of 200,000 reps per week, simulating the wheel load produced by a dual-tire, single axle. The carousel’s self weight is 8,000 kg, and it can apply loads up to almost 150 kN. Wheel speed can be set up to 40 Km/h, and the lateral position of load application is adjustable within a range of 1.0 m. For this test the lateral position was adjusted by 100 mm every 1,000 repetitions. Pressure cells and other types of instrumentation were installed in the pavement. Figure 4 shows the carousel.

Figure 4. Carousel for accelerated pavement loading



Test track construction The construction work began with the removal of existing material at the carousel facility, followed by compaction of subgrade, placement of reinforcement geogrid, and construction of the granular subbase and base layers. Density tests indicated average compaction on the subbase of 97 percent modified Proctor, with a 0.9 percent coefficient of variation. For the granular base the compaction was 99.3 percent, with 1.6 percent CV. In both cases the results were obtained from nine tested locations. Figure 5 shows part of the construction process related to geogrid and granular layers placement. Asphalt laying was carried out with standard local practices and equipment, as shown in Figure 6.

Figure 5. Reinforcement and granular layers placement



Figure 6. Placement and compaction of asphalt layer

STRUCTURAL EVALUATION AND ANALYSIS OF RESULTS A Benkelman beam was routinely used to assess structural condition of the four test sections, by means of measuring deflection at the pavement surface directly under the wheel load and at 250 mm (D0 and D250 respectively). The deflection measurements took place at the locations where the ribs of the grid are perpendicular to the wheel path, which match how the grid is used in the field. The radius of curvature (RC) was also computed from deflection measurements. Results are presented in Table 1.



Table 1. Surface deflections at 0 and 250 mm from the load (microns), and radius of curvature (meters) during loading Section

Measurement

Control (conventional flexible pavement)

D0 D250 RC D0 D250 RC D0 D250 RC D0 D250 RC

Section 1 (reinforcement within the base) Section 2 (reinforcement on top of the subgrade) Section 3 (reinforcement within the base and on top of the subgrade)

Cumulative number of repetitions 0 50,000 75,725 10 106 159 8 78 128 1,586 112 100 6 115 125 4 92 93 1,578 140 100 7 140 209 5 101 150 1,586 80 53 5 134 137 3 90 87 2,016 70 63

100,139 165 131 92 126 95 99 216 154 51 141 90 61

As the loading test progressed, the following surface distresses were observed: •

At 50k repetitions only section 2 presented signs of distress, revealed by longitudinal cracks at the wheel path of about 1 to 2 mm wide. Some rutting was also observed on section 2.

•

At 75k repetitions, the control section exhibited longitudinal cracking (1 to 2 mm), while very fine longitudinal cracks were observed on section 1. Cracking in section 2 had widened to 3 -10 mm. There was no cracking on section 3. Rutting on section 2 remained as before, and no rutting was observed on the other sections.

•

By the end of the test (100k reps), cracking on the control section and on section 1 remained same as before, at 1-2 mm. Cracking on section 2 had increased to be between 5 and 12 mm wide, and there was no cracking on section 3. Some rutting was observed on the control section and on section 2.

A picture of deflection measurements is presented in Figure 7. Surface distresses, the progression with loading, and the overall condition of the sections are presented in Figure 8 to Figure 10.



Figure 7. Benkelman beam being used on a test section

a)

b)

Figure 8. Surface condition on section 2 at a) 50k reps, and b) 100k reps



a)

b)

Figure 9. Surface cracking at 75k reps on a) section 2 and b) control section

a)

b)

Figure 10. Surface overall condition at 100k reps a) section 1, b) control section



Initial deflections measured under the wheel (D0) and at 250mm from the wheel, (D250) are slightly smaller in the sections that contain geogrid compared to control. However, as the number of repetitions increases, the section with geogrid on the subgrade (section 2) start to develop higher deflections than the control section, as presented in Figure 11.

Deflections under the load (microns)

250 200 Control 150

Section 1 Section 2

100

Section 3 50 0 0

20

40

60

80

100

Load repetitions (Thousands)

Figure 11. Pavement deflection in each section versus load repetitions

By the end of the test, i.e. at 100k repetitions, the best structural performance corresponded to section 1, followed by section 3, and then the control section. The highest deflections were observed on section 2. These results are in agreement with findings by Perkins et al (2004) in the sense that thick sections do not benefit as much from geogrid reinforcement as do thin sections. Placement of geogrid closer to the surface reduces the strains in the unbound layer directly under the AC layer, while the reinforcement on the subgrade does not seem to act in the pavement system. Another way of presenting results is by the number of repetitions each structure had supported by the time a deflection limit of 100 microns was observed. In this case interpolation of data in Figure 11 indicates that such a limit is reached between 35k and 46k repetitions of the standard 80-kN axle. Under this approach, section 1 and the control performed better than section 2 and, surprisingly section 3.



Excellent agreement exists between surface distresses, i.e. cracking and deformation, and the deflections measured in section 2 and the control section. However, in spite of similar deflections, sections 1 and 3 did not present elevated distress levels. Failure in sections 1 and control is believed to be associated to fatigue rather than permanent deformation. Unfortunately, no forensic investigation was conducted that could have helped to reveal differences in pavement thickness, pavement density, and subgrade rutting. It must also be mentioned that early surface deformations on section 2 could be associated to construction issues. Asphalt placement began on section 2, and for this section the paver was not fed directly from the trucks, but through a mini front loader.

SUMMARY AND CONCLUSIONS Three experimental pavement sections with different configurations of geogrid reinforcement were compared to a control section at the full-scale testing facility at the Universidad de Los Andes in Colombia. The following conclusions can be drawn from the study. 1. Failure on the sections occurred, as expected, due to fatigue on the AC mix instead of under compression on the subgrade. 2. Best performance is obtained when geogrid is placed in the granular base and on the two-layer configuration on top of the subgrade and in the base. From these results, it appears that the fatigue performance of a flexible pavement is improved as long as the geogrid reinforcement is placed on the upper layers. 3. The advantage of using full-scale testing is evident considering that the entire 100k-repetition test was completed in about two weeks. 4. The total thickness of 780mm for the pavement structure was the design required by the rational methodology to withstand 100,000 ESALs. This design seems exaggerated compared to typical paving projects, but it is the result of the method when the dynamic characteristics of the materials are applied, and for the highly compressive subgrades typical of Bogotá, which have very low bearing capacity.



5. AC thickness of 180mm resulted from a calculated admissible tensile strain at the bottom of the layer of 118 microstrain, considering the dynamic modulus of 3,530 MPa for the surface course AC and the specifically determined fatigue relationship. Had an AC base course been used, which has twice the elastic modulus of the AC surface course, a thinner AC layer would have been obtained.

6. REFERENCES 1. Hugo, F.; Epps-Martin, A. (2004), “Significant Findings from Full-Scale Accelerated Pavement Testing”. NCHRP Synthesis of Highway Practice 325, 211p 2. Romero, R. (1996), “Test Tracks, Applications to Pavement Design and Evaluation”. XII International Course for Pavement Roads and Highways. CEDEX, Spain (in Spanish). 3. Munevar, S. and Perez, S., (2000), “First Colombian Experience with the Fatigue Carousel”. M.Sc. thesis, Universidad de Los Andes. (in Spanish) 4. Reyes, F.(1998), “Rational Pavement Design” First Edition, Universidad de Los Andes, Colombia (in Spanish) 5. LCPC, SETRA, (2003), “ALIZÉ-LCPC Routes. Software for structural design of roadways”, Itech-soft edition and distribution.(in French) 6. Autret P., De Boissoudy A., Marchands J.-P. (1982) “Utilization of Alizé III”. 5th International Conference for Pavement Design, Delf, Netherlands. (in French) 7. Perkins and Cortez (2004). “Evaluation of Base-Reinforced Pavements Using a Heavy Vehicle Simulator”. Proceedings TRB 2004 8. Perkins, SW; Christopher, BR; Cuelho, EL; Eiksund, GR; Hoff, I; Schwartz, CW; Svano, G; Watn, A (2004). “Development of Design Methods for Geosynthetic Reinforced Flexible Pavements”. FHWA Report FHWA DTFH61-01-X-00068, 284p

