U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

 
REPORT
This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-16-010    Date:  March 2017
Publication Number: FHWA-HRT-16-010
Date: March 2017

 

Using Falling Weight Deflectometer Data With Mechanistic-Empirical Design and Analysis, Volume II: Case Study Reports

CHAPTER 1.

CASE STUDY 1: FLEXIBLE PAVEMENT EVALUATION AND OVERLAY DESIGN

PROJECT OVERVIEW

Project 30-0100, located on I-15 near Great Falls, MT, consisted of 12 test sections. Test section 30-0113 was selected as a flexible pavement rehabilitation case study. Test section 30-0113 was a flexible pavement cross section with a hot-mix asphalt (HMA) surface and aggregate base layer over subgrade. This test section is representative of the following selection factors:

  • Flexible pavement.
  • Dry-freeze climate zone.
  • Principal interstate—rural functional class.
  • “Fair” pavement condition.
  • Sand subgrade classification.
  • No (or deep) rigid layer.
 

The original pavement, nominally a 4-inch (102-mm) HMA surface on 8-inch (203-mm) aggregate base, was constructed in 1997. Rehabilitation and repair work—consisting of crack sealing and an aggregate seal coat—were performed in 2003 and 2004, respectively (construction numbers 2 and 3, respectively).

PAVEMENT CONDITION/PERFORMANCE

Pavement condition and performance data can be used to customize (or calibrate) the performance models within the Mechanistic-Empirical Pavement Design Guide (MEPDG) software for the specific State using the design procedure.(1) However, the calibration of performance models for specific States was beyond the scope of this study. The design inputs after performance model calibration used relatively little distress data. The distress data used for design inputs included the percentage of cracking, rutting depth, and overall condition, depending on design level being used.

The pavement distress data used in the rehabilitation design depend on the design level hierarchy selected. The distress data used for the level 1 analysis are summarized in table 1 and presented in figure 1. Level 1 analysis uses the rutting attributed to each layer of the cross section. Because trenching data were not available, the individual layer rutting contributions were estimated.

Table 1. Summary of existing pavement distress inputs for level 1

Laboratory Test Rutting (inches)
Existing HMA 0.25
Granular base 0.2
Subgrade 0.15
1 inch = 25.4 mm.

 

Click for description
1 inch = 25.4 mm.
Figure 1. Screen Capture. Summary of condition inputs for level 1.

 

Level 2 design includes the percent cracking in the existing HMA surface in addition to the layer rutting, and level 3 uses the overall rut depth (not individual layer contributions) and the overall pavement rating (good, fair, or poor), as shown in figure 2.

Click for description
1 inch = 25.4 mm.
Figure 2. Screen Capture. Summary of condition inputs for level 3.

TEST SECTION DATA

The MEPDG design program requires a significant number of inputs, particularly for level 1 analysis. The required design data for the 30-0100 project section were obtained from the Long-Term Pavement Performance (LTPP) Program DataPave database. The required data were not generally complete for any one specific test section within project 30-0100; therefore, results for the entire project were used to obtain the necessary design inputs. Although the test sections had slightly varying cross sections and maintenance histories, the research team concluded that the overall initial data were sufficient for this study.

Deflection-Testing Data

Deflection data for test section 30-0113 were available from the LTPP Program database for several years of testing, including 1998, 1999, 2001, 2003, and 2005. Deflection testing was conducted following the LTPP Program protocols.

Equipment

Deflection testing was conducted with a Dynatest® falling weight deflectometer (FWD) (SN8002-131 for 1999 through 2005).

Sensor Configuration

Sensors were located at 0, 8, 12, 18, 24, 36, and 60inches (0, 203, 305, 457, 610, 914, and 1,524mm) for the 1999 datasets and 0, 8, 12, 18, 24, 36, 48, and -12 inches (0, 203, 305, 457, 610, 914, 1,219, and -305 mm) for data collected after 1999.

Number of Drops and Load Levels

Four load-level targets—6,000, 9,000, 12,000, and 16,000lb (2,724, 4,086, 5,448, and 7,264 kg)—with four drops at each load level were performed, and data were recorded. Seating drops were also performed, but data were not recorded.

Test Locations/Lanes and Increments

Testing was conducted in the outer wheelpath and mid-lane at 50-ft (15.3-m) intervals, with the outer wheelpath and mid-lane locations at the same stations (i.e., not staggered).

Temperature Measurements

Temperature measurements were taken using drilled holes in the pavement at 1-, 2.2-, and 3.3‑inch (25.4-, 55.9-, and 83.8-mm) depths at prescribed time intervals during deflection testing.

Material Properties Data

Laboratory testing data for unbound materials were available from 2000 and 2002 and for HMA materials from 1999.

Subgrade

Eleven subgrade samples were retrieved from the 30-0100 project location as part of the LTPP Program. The subgrade was classified as A-2-6 under the American Association of State Highway and Transportation Officials (AASHTO) soil classification system.(2)

Laboratory resilient modulus testing results were available for several samples of the subgrade materials obtained as part of the LTPP Program data collection. Ten subgrade samples were tested for the project 30-0100 test sections; however, 2 samples were removed from this data analysis because the samples were classified as different soil types than the remaining samples, as discussed previously. Laboratory resilient modulus testing results for the subgrade samples (identified as BS ##) are illustrated in figure 3. Additional subgrade properties, including Atterberg limits and sieve analysis, are summarized in table 2 and table 3.

Click for description
1 psi = 6.89 kPa.
Figure 3. Graph. Summary of LTPP Program laboratory-measured subgrade resilient modulus from field samples.

 

Table 2. Summary of subgrade Atterberg limits.
Laboratory Test Average Test Result (percent)
Liquid limit 33
Plasticity index 16

 

Table 3. Summary of subgrade sieve analysis.

Sieve Size Average Percent Passing
No. 4 100
No. 10 99
No. 40 98
No. 80 61
No. 200 27.8

 

Base Aggregate

Three coarse aggregate samples were obtained as part of the LTPP Program data collection and were classified as AASHTO A-1-b. Laboratory resilient modulus testing results were also available for three samples of coarse aggregate materials in the project 30-0100 test sections. Laboratory resilient modulus testing results for the coarse aggregate samples (identified as BG##) are illustrated in figure 4.

Click for description
1 psi = 6.89 kPa.
Figure 4. Graph. Summary of LTPP Program laboratory-measured base aggregate resilient modulus from field samples.

 

The average thickness of the base from available data was 8.4 inches (213 mm). Additional aggregate properties are summarized in table 4 and table 5.

Table 4. Summary of base aggregate Atterberg limits.
Laboratory Test Average Test Result (percent)
Liquid limit 11
Plasticity index 1

 

Table 5. Summary of base aggregate sieve analysis

Sieve Size Average Percent Passing
1.5 inch 96.7
1.0 inch 93.6
3/4 inch 90.8
1/2 inch 85.8
3/8 inch 82.3
No. 4 74.2
No. 10 64.0
No. 40 37.6
No. 80 20.8
No. 200 13.4
1 inch = 25.4 mm.

 

HMA Surface

Based on the data for the section, the average HMA thickness was 4.3 inches (109 mm). Construction data indicate the HMA was placed in two approximately equal lifts of the same material. The LTPP Program database contained laboratory resilient modulus testing results for six HMA surface samples from the 30-0100 project section. Dynamic modulus testing data (AASHTO Test Procedure 62) were not available.(3) Testing was performed on 1.5‑inch (38-mm)-thick test samples. The average resilient modulus testing results—instantaneous and total—are summarized in figure 5. Additional HMA properties for design are summarized in table 6 and table 7.

Click for description
1 psi = 6.89 kPa.
°F = 1.8 × °C + 32.

Figure 5. Graph. Summary of LTPP Program laboratory-measured average HMA resilient modulus.

 

Table 6. Summary of existing HMA material properties.
Variable Value
Asphalt grading Performance grade (PG) 70-28
Asphalt content (percent) 9.8
Air voids, percent 7.4
Total unit weight (lb/ft3) 143.5
1 lb/ft3 = 0.0160 g/cm3.

 

Table 7. Summary of existing HMA aggregate sieve analysis.

Sieve Size Average Percent Passing
3/4 inch 100
3/8 inch 78
No. 4 50
No. 200 6.5
1 inch = 25.4 mm.

 

Depth to Rigid Layer/Water Table

The depth to the water table is also required by the MEPDG. Unfortunately, no project-specific data were found for project 30-0100. One record was obtained from the LTPP Program database for test section 30‑8129, with measurements obtained during May and June 1997. The depth to the water table at that time averaged approximately 9.4 ft (2.9 m). Historical groundwater levels from the United States Geological Survey (USGS) indicate an average of approximately 12 ft (3.7 m) below land surface.(4) Again, this monitoring location was not at the project 30-0100 site but was in the Great Falls, MT, area.

Climate/Environment Data

Climate data were obtained from the updated climate files on the MEPDG Web site.(5) The weather station at Great Falls, MT, was used for this study. The general weather category for the case study location was dry-freeze.

Traffic Data

Traffic data were obtained from the LTPP Program database. Because evaluating the effect of traffic data on design results was not a primary goal of this study, only basic information was used from the available data (i.e., total volume, growth, and vehicle class distribution), with the remaining inputs (i.e., monthly distribution, hourly distribution, and wheel spacing) kept at their default values in the MEPDG software. A 0.7-percent compounded growth rate was estimated based on traffic data. An average annual daily truck traffic (AADTT) volume of 657vehicles per day was estimated. The vehicle class distribution obtained from the LTPP Program database is summarized in table 8.

Table 8. Traffic distribution by vehicle class.

Vehicle Class Traffic Distribution (percent)
4 1.8
5 24.6
6 7.6
7 0.5
8 5.0
9 31.3
10 9.8
11 0.8
12 3.3
13 15.3

 

ANALYSIS AND INTERPRETATION OF FWD TESTING DATA

This section presents the data checks and backcalculation analysis of the FWD data, as well as a comparison of the backcalculation results with laboratory testing.

Preprocessing Deflection Data

In addition to general data checks (such as non-decreasing data, zeroes, etc.), FWD deflection data were checked for linearity. Figure 6 compares the load versus sensor deflection for a few selected locations within the project. The measurements show slight stress hardening in 2005 as opposed to very slight stress softening behavior in 1999. This may reflect the effect of traffic compaction on unbound layers. Nevertheless, the data were deemed to be acceptable for linear analysis. Depth to bedrock calculations indicated no rigid layer was likely within 7.6 m (25 ft) of the surface.

Backcalculation Analysis

Backcalculation of the test section 30-0113 outer wheelpath deflection data was performed using linear, static layered elastic analysis, as discussed in chapter 4, volume I of this report. Three-layered elastic backcalculation programs were used for this analysis: (1) MODTAG, (2)MICHBACK©, and (3)EVERCALC© to look into the effect of different inverse routines on backcalculated parameters and ultimately on rehabilitation design. In addition, the following layer combinations were used to determine the most realistic design inputs for the MEPDG software:

The seed, minimum, and maximum values for layer moduli along with assumed Poisson’s ratio values are summarized in table 9.

Click for description
1 lb = 0.454 kg.
1 mil = 0.0254 mm.

Figure 6. Graphs. Comparison of sample plots of FWD load versus sensor deflection.

 

Table 9. Input layer modulus and Poisson’s ratio values for backcalculation analysis.
Layer Thickness
(inches)
Seed Modulus
(psi)
Minimum
Modulus, (psi)
Maximum
Modulus, (psi)
Poisson’s
Ratio
HMA 4.3a 300,000 50,000 2,500,000 0.35
Base 8.4a 30,000 5,000 150,000 0.3
Top 2 ft subgrade 24 15,000 3,000 100,000 0.4
Subgrade Infinite 7,500 3,000 100,000 0.4
aAverage thickness determined from LTPP Program inventory data.
1 psi = 6.89 kPa.
1 inch = 25.4 mm.

 

The backcalculation was performed for each FWD test location using the latest available test data (2005). This allowed investigation of the effect of construction and site variability on backcalculation results. The average layer thicknesses shown in table 9 were used in the MODTAG analysis, while the actual layer thicknesses for each FWD test location, summarized in table 10, were used in the backcalculation analyses using MICHBACK© and EVERCALC©.

Table 10. Input layer thicknesses for different FWD test locations.
Station (ft) Asphalt Concrete Thickness (inches)a Base Thickness (inches)a
0 4.2 9.3
50 4.2 8.1
100 4.2 9.5
150 4.4 8.2
200 5.0 7.5
250 4.0 7.1
300 4.2 8.4
350 4.4 8.8
400 4.2 8.6
450 4.2 8.2
500 4.1 8.4
aObtained from LTPP Program inventory data.
1 inch = 25.4 mm.

 

Backcalculation Results

Backcalculation results for the various layer models and programs are summarized in table 11 through table 13. Figure 7 illustrates the backcalculation results from MODTAG for the various stations and load levels using several layer combinations, as described previously. The HMA layer moduli were not affected much by the layer combinations for this project (all results are within approximately a 2-percent difference). Introducing a 0.6-m (2-ft) top subgrade layer to the semi-infinite subgrade model led to a lower compacted subgrade modulus value (approximately 10percent), and a higher base layer value (approximately 15 percent). However, the base layer results tended to be lower than expected for both models.

Table 11. Summary statistics of backcalculation results for three-layer system with semi‑infinite subgrade (modulus, psi).

Backcalculation Tool Statistic HMA Base Subgrade Error (percent)
MICHBACK© Mean 595,965 12,181 25,592 1.86
Standard deviation 88,487 3,760 1,339 0.84
COV 0.15 0.31 0.05 0.45
N 40 40 40
EVERCALC© Mean 598,400 12,100 25,600 1.89
Standard deviation 87,300 3,700 1,400 0.79
COV 0.15 0.31 0.05
N 40 40 40 40
MODTAG Mean 608,472 16,727 21,461 9.10
Standard deviation 133,362 5,595 1,622 4.40
COV 0.22 0.33 0.08 0.48
N 36 36 36
1 psi = 6.89 kPa.
— Indicates not applicable.
COV = Coefficient of variation.
N = Number of samples.

 

Table 12. Summary statistics of backcalculation results for four-layer system with semi-infinite subgrade (modulus, psi).

Backcalculation Tool Statistic HMA Base Top 2-ft Subgrade Subgrade Error (percent)
EVERCALC© Mean 732,278 15,072 21,839 25,272 1.28
Standard deviation 66,300 2,400 149,400 1,215 0.71
COV 0.21 0.34 1.05 0.05 0.56
N 36 36 36 36
MODTAG Mean 591,139 19,652 17,961 23,706 4.31
Standard deviation 135,146 6,017 3,077 1,188 1.64
COV 0.23 0.31 0.17 0.05 0.38
N 36 36 36 36
1 psi = 6.89 kPa.
— Indicates not applicable.

 

Table 13. Summary statistics of MODTAG backcalculation results with rigid layer (Modulus, psi).

Pavement System Statistic HMA Base Subgrade Error (percent)
Three-layer system Mean 592,611 19,051 19,081 6.76
Standard deviation 132,024 6,292 1,486 1.47
COV 0.22 0.33 0.08 0.22
N 36 36 36
Four-layer system Mean 608,083 16,493 23,272a 17,439
Standard deviation 135,993 5,178 5,022a 892
COV 0.22 0.31 0.22a 0.05
N 36 36 36a
aTop 2-ft (610 mm) subgrade.
1 psi = 6.89 kPa.
— Indicates not applicable.

 

Click for description
1 psi = 6.89 kPa.
1 ft = 0.305 m.

Figure 7. Graphs. Comparison of backcalculated layer moduli (MODTAG) for different layer combinations and load levels along the section.

 

As shown in table 13, introducing a rigid layer at a fixed depth had some effect on the base and subgrade moduli for the three- and four-layer system, with a more significant effect on the subgrade and base layer moduli for the four-layer system (35- and 19-percent decrease, respectively).

Figure 8 shows the root mean square (RMS) values from MODTAG for the deflection basins at each station along the section using the different layer combinations. The RMS values were observed to be very high (much greater than 2 to 3 percent) and would typically be considered unacceptable for design purposes. The four-layer system with a 2-ft (0.6-m) compacted subgrade layer on top of an infinite halfspace gave the lower RMS values among the four-layer combinations considered. However, the top subgrade modulus was less than the lower subgrade modulus. This may suggest a compensating layer effect, but similar trends were observed when a rigid layer was and was not included in the analysis. It is possible the upper layer was not compacted as well as the lower layer.

Click for description
1 ft = 0.305 m.
Figure 8. Graph. RMS values (MODTAG) along the section for different layer combinations and load levels.

 

Figure 9 through figure 12 show the backcalculation results and corresponding RMS values obtained from EVERCALC© and MICHBACK©, respectively, for the various stations and load levels using a three-layer system. It is noted that the two programs produced very similar results, within approximately 1 percent. They also led to lower RMS values (mostly below 3 percent). However, the subgrade modulus was consistently higher than the base modulus (generally about twice as much).

Click for description
1 psi = 6.89 kPa.
1 ft = 0.305 m.

Figure 9. Graph. EVERCALC© backcalculated layer moduli for a three-layer system and different load levels along the section.

 

Click for description
1 psi = 6.89 kPa.
1 ft = 0.305 m.

Figure 10. Graph. EVERCALC© RMS values for a three-layer system and different load levels along the section.

 

Click for description
1 psi = 6.89 kPa.
1 ft = 0.305 m.

Figure 11. Graph. MICHBACK© backcalculated layer moduli for a three-layer system and different load levels along the section.

 

Click for description
1 psi = 6.89 kPa.
1 ft = 0.305 m.

Figure 12. Graph. MICHBACK© RMS values for a three-layer system and different load levels along the section.

 

Figure 13 summarizes the backcalculated layer moduli (averaged over different load levels) for a three-layer system obtained from the three different programs. The results from MICHBACK© and EVERCALC© were essentially the same for each layer, while those from MODTAG generally differed from them. This could be in part owing to the difference in layer thickness used (average versus location-specific). The base modulus was the most variable along the section, followed by the HMA modulus.

Click for description
1 psi = 6.89 kPa.
Figure 13. Graphs. Summary of backcalculated results for a three-layer system (average of load levels).

 

Figure 14 summarizes the backcalculated layer moduli from MODTAG and EVERCALC© for a four-layer system (average of load levels). The MICHBACK© program was unable to converge for this case. The variability in modulus was higher for this case. Also, the compacted subgrade modulus was still higher than the base layer modulus.

Backcalculation Modeling Issues and Recommendations

The RMS values obtained from the three backcalculation programs were generally very high, especially for the MODTAG program. However, the backcalculated layer moduli agreed reasonably well if they were averaged across FWD test locations. Some locations gave unreasonable backcalculation results, the results of which should be excluded from the analysis. The base layer modulus was consistently lower than the subgrade modulus even when using a four-layer system, which included a 2-ft. (0.6-m) compacted subgrade layer on top of the semi-infinite subgrade. The use of correction factors can force the moduli to align along the expected trend of having the subgrade modulus lower than the base modulus. This is addressed in the next section.

Comparison of Backcalculation and Laboratory Testing Results

To assist in evaluating which layer characteristics were appropriate to use in the MEPDG software, the results of the backcalculation were compared with results obtained from laboratory testing conducted as part of the LTPP Program. While it was outside of the scope of this study to develop new correlations or conversion factors between the two, it was still beneficial to evaluate these relationships and how they may influence MEPDG input selection.

Unbound Materials

The first general argument when comparing field with laboratory results of unbound materials is that the stress state of the material is different for the two conditions. Therefore, correction factors are typically applied to unbound layers. The guidance in the MEPDG, as of the time of this report, was to use the previously established coefficients (summarized in table 3.6.8 of the MEPDG) to adjust backcalculated layer moduli for use in design.(1) These included a factor of 0.35for subgrade material below flexible pavement with a granular base layer and 0.62 for a granular base below a flexible surface or base layer.

To compare laboratory results to those obtained from FWD testing (or vice versa), the resilient modulus was estimated for the stress conditions at the time of FWD testing. The stress conditions accounted for the overburden pressure of the pavement and the stress due to loading. Load stresses were determined using WinJULEA, the load applied during deflection testing, and initial backcalculated layer properties. Overburden stresses were estimated using layer densities, moisture contents, thicknesses, and at-rest earth pressure coefficients (Ko). A Ko value of 0.5 was used to remain consistent with MEPDG documentation.

Click for description
1 psi = 6.89 kPa.
Figure 14. Graphs. Summary of backcalculated results for a four-layer system (average of load levels).

 

To compare the resilient modulus at the stress conditions of FWD testing (assuming a 9,000-lb (4,086-kg) load), the constitutive model in figure 15 (contained in the MEPDG) was used.(6)

Click for description

Figure 15. Equation. Constitutive model for determining resilient modulus.

Where:

MR = Resilient modulus, kPa (psi).
θ = Bulk stress, kPa (psi).
τOct = Octahedral shear stress, kPa (psi).
pa = Atmospheric pressure, kPa (psi).
k1, k2, k3 = Regression constants.

Developed constitutive models are illustrated in figure 16 and figure 17 for the subgrade and base materials, respectively. The estimated subgrade resilient modulus based on laboratory testing and using the constitutive model at the stress conditions for a 9,000-lb (4,086-kg) FWD load was 3,900psi (26,900kPa). The resulting laboratory-to-field value ratios ranged from approximately 0.15 to 0.22. For this project, this range of values was lower than the typical correction factor (0.35) provided in the MEPDG.(1) Consequently, the equivalent modulus was much lower than the default value used in the MEPDG (20,500 psi (141,340 kPa) for AASHTO A-2-6 subgrade).

Click for description
1 psi = 6.89 kPa.
Figure 16. Graph. Estimated constitutive model for subgrade resilient modulus.

 

Click for description
1 psi = 6.89 kPa.
Figure 17. Graph. Estimated constitutive model for base resilient modulus.

 

The estimated base resilient modulus based on laboratory testing and using the constitutive model at the stress conditions for a 9,000-lb (4,086-kg) FWD load was 13,200psi (91,000 kPa). The resulting laboratory-to-field value ratios range from approximately 0.67to 1.09. This range was greater than the typical correction factor (0.62) in this case. The equivalent modulus value was much lower than the default value used in the MEPDG—38,000 psi (262,000kPa) for AASHTO A-1-b base).

Bound Materials

The laboratory results for the HMA layer modulus for several temperatures are shown in figure 18. The pavement temperature at the time of the FWD testing was recorded as 70 °F (21 °C) based on borehole temperature data. Using figure 18, a corresponding laboratory HMA modulus of about 350,000 psi (2,413,000 kPa) was obtained, compared with average backcalculated moduli ranging from approximately 591,100 psi (4,075,000 kPa) to 732,300 psi (5,049,000kPa).

Click for description
1 psi = 6.89 kPa.
°F = 1.8 × °C + 32.

Figure 18. Graph. Approximate HMA resilient modulus based on LTPP Program laboratory testing.

 

An additional consideration in comparing laboratory and deflection testing is the frequency of loading: laboratory resilient modulus testing was conducted with a 0.1-s load pulse, whereas FWD testing was conducted with a 0.03-s load pulse (approximately). Although dynamic modulus testing data were not available, reviewing the approach for shifting the HMA modulus due to frequency and temperature was worthwhile. Based on the available HMA properties, recommended MEPDG default values, and predictive equations, the estimated dynamic modulus (E*) master curve using the MEPDG software is illustrated in figure 19.

Click for description

Figure 19. Graph. Illustration of HMA dynamic modulus master curve from MEPDG design software.

 

Based on the results of the predictive equations incorporated in the MEPDG and case study inputs, a dynamic modulus of 924,000 psi (6,371,000 kPa) was estimated for the conditions during FWD testing (approximately 70 °F (21.1 °C) and 0.03-s loading time). A dynamic modulus of 730,400psi (5,036,000kPa) was estimated at 70 °F (21.1 °C) and 0.1-s loading time (resilient modulus testing conditions). In this case, it appears the predictive model results in moduli greater than the resilient modulus testing results and possibly slightly high for HMA material.

REHABILITATION DESIGN RESULTS

An HMA overlay rehabilitation design was selected for this case study to assess design differences between backcalculated inputs and laboratory-based inputs. It was assumed that 0.75inches (19.1mm) of the existing HMA overlay would be milled and that additional pre‑overlay repairs (patching and crack sealing) would be performed as necessary. The overall design level selected for this rehabilitation (HMA over HMA) was level 1. The required layer inputs used in the MEPDG depend on the overall design level selected for design. When level 1 is selected, all unbound layer inputs need to be level 1, but the HMA layers can be levels 1 through 3. When the overall design levels 2 or 3 are selected, unbound layer inputs can be level 2 or 3, and the HMA layers can be level 1 through 3. Level 3 was used here for the HMA layers.

Design Criteria and General Inputs

The MEPDG calls for determining the HMA overlay thickness by trial and error. A trial thickness is assumed, and the program is executed to predict the future performance in terms of the different performance measures (cracking, rutting, International Roughness Index (IRI), etc.). For a given desired rehabilitation design life, the level of distress at the design life should not exceed a prescribed limiting value, summarized in table 14 for this case study. A reliability level of 90 percent was selected as well as a 20-year design life.

Table 14. Summary of analysis parameter inputs.
Variable Value
Initial IRI (inches/mi) 63
Terminal IRI (inches/mi) 172
HMA surface down cracking, long cracking (ft/mi) 2,000
HMA bottom up cracking, alligator cracking (percent) 25
HMA thermal fatigue (ft/mi) 1,000
Chemically stabilized layer fatigue fracture (percent) N/A
Permanent deformation—total pavement (inches) 0.75
Permanent deformation—HMA only (inches) 0.25
1 inch/mi = 0.0158 m/km.
1 inch = 25.4 mm.
1 ft/mi = 0.19 m/km.
N/A = Not applicable.

 

For new HMA materials, level 1 analysis requires conducting E* (complex modulus) laboratory testing (ASTM D3496) at loading frequencies and temperatures of interest for the given mixture.(7) Level 2 analysis does not require E* laboratory testing; instead, the user can input asphalt mix properties (gradation parameters) and laboratory binder test data (from G* testing or other conventional binder tests). The MEPDG software calculates the corresponding asphalt viscosity values; it then uses the modified Witczak equation to predict E* and develops the master curve for the HMA mixture. The same procedure is used for level 3 analysis to estimate the HMA dynamic modulus except that no laboratory test data are required for the binder, and typical values for the selected binder grade are used. The properties used for the new HMA overlay are summarized in table 15 and table 16.

Table 15. Summary of new HMA material properties.

Variable Value
Asphalt grading PG 70-28
Asphalt content (percent) 12.5
Air voids (percent) 4.0
Total unit weight (lb/ft3) 148
1 lb/ft3 = 0.0160 g/cm3

 

Table 16. Summary of new HMA aggregate sieve analysis.

Sieve Size Average Percent Passing
3/4 inch 100
3/8 inch 78
No. 4 50
No. 200 6.5
1 inch = 25.4 mm.

 

Incorporation of Backcalculation Results With the MEPDG Software

Several analyses were executed with the MEPDG design software (summarized in table 17) to evaluate the influence of the unbound and bound layer inputs, as discussed in the following sections. Version 1.003 of the software was used, as well as the nationally calibrated performance models.

Table 17. Input base and subgrade moduli for five MEPDG runs.
Analysis Run Number Subgrade Modulus (psi) Base Modulus (psi) HMA Modulus (psi)
1a 14,500 38,000 N/A
2b 21,500 16,700 608,500
3c 7,500 10,400 608,500
4d 9,000 7,500 597,000
5e 3,900 13,200 350,000
6f 6,300 (24 inches)
8,300 (infinite)
12,200 591,000
aAnalysis 1: Default layer properties; HMA modulus determined internally based on material properties.
bAnalysis 2: Uncorrected backcalculation values based on MODTAG results for three-layer system.
cAnalysis 3: Corrected backcalculation values based on MODTAG results for three-layer system.
dAnalysis 4: Corrected backcalculation values based on EVERCALC© and MICHBACK© for three-layer system. eAnalysis 5: Based on LTPP Program laboratory testing results.
fAnalysis 6: Corrected backcalculation values based on MODTAG results for four-layer system.
1 psi = 6.89 kPa.
1 inch = 25.4 mm.
N/A = Not applicable

 

The matrix of design analyses incorporated the use of default layer properties and internally calculated HMA moduli based on mixture properties (analysis 1), backcalculation-based values (analyses 2–4), and LTPP Program laboratory testing results (analysis 5).

HMA Layers

For rehabilitation design, the determination of the existing HMA layer dynamic modulus follows the same general concepts described for new HMA, except that the software allows a modified procedure to account for damage incurred in the HMA layer during the life of the existing pavement. Therefore, the procedure determines a field-damaged dynamic modulus master curve, as discussed in chapter 4, volume I, of this report. When the overall design hierarchy is level 1, FWD data can be used at material layer levels 1 through 3. For the overall design levels 2 and 3, FWD data are not used, regardless of the layer input level.

Level 1 analysis requires entering the backcalculated HMA modulus with the corresponding temperature at the time of testing and an equivalent frequency for the FWD pulse, as shown in figure 20. As previously indicated, the pavement temperature at the time of the FWD testing was 70 °F (21 °C) based on borehole temperature data. Based on the HMA over HMA rehabilitation example in the MEPDG appendix, an equivalent frequency of 30 Hz was used for FWD-based inputs. This was calculated as 1/(FWD pulse duration) = 1/0.033 s = 30 Hz. Although this formula technically does not yield frequency (based on sinusoidal wave), it is compatible with the equivalent frequency used for calculating E* in the MEPDG, which is calculated as equivalent frequency = 1/equivalent time. The equivalent time is calculated using the width of the tire, an equivalent depth based on the Odemark concept of equivalent thickness and a 1:1 stress dissipation slope.

Click for description
1 inch =25.4 mm.
Figure 20. Screen Capture. HMA input based on FWD testing and backcalculation for overall design level 1 and layer input level 3.

 

Unbound Layers

For unbound materials (and bedrock, if used), only level 1 analysis calls for FWD testing in rehabilitation and reconstruction designs. The resilient modulus, MR, for each unbound layer (including the subgrade) can be either determined in the laboratory using cyclic triaxial tests or backcalculated using standard backcalculation procedures. Although the MEPDG does allow use of the generalized nonlinear, stress-dependent model in the design procedure, this approach is not recommended at this time because the performance models in the software have not been calibrated for nonlinear conditions. Therefore, the option of backcalculating the k1, k2, and k3 parameters in the nonlinear model is not discussed in this report. Consequently, the discussion includes only the backcalculation and use of “effective” moduli that would account for any stress-sensitivity, cracks or any other anomalies in any layer within the existing pavement.

For level 2 analysis, correlations with strength test data were used. For level 3, the MEDPG lists typical modulus values based on soil classification but warns that they are very approximate and strongly recommends some form of testing, especially using FWD testing and backcalculation (level 1). The guide notes that the reason for caution is related to using the wrong assumptions: either a fairly strong subgrade material may be erroneously assumed to be semi-infinite while it may actually be less than 3 ft (1 m) thick (e.g., as part of an embankment), or conversely, a weak subgrade soil may be assumed to be semi-infinite while it may, in reality, be overlying a stronger soil or bedrock.

The MEDPG also notes that for granular materials, moduli values that matched FWD backcalculated results were 50 to 70 percent higher than the typical laboratory-tested values, while for subgrade soils, they were two to three times the typical laboratory determined values. The guidance in the MEPDG as of the time of this report is to use the previously established coefficients (summarized in table 3.6.8 of the MEPDG) to adjust backcalculated layer moduli for use in design.(1)

The MEPDG allows for entering the backcalculated base and subgrade layer moduli directly, together with a correction factor to account for the difference between laboratory-determined and backcalculated moduli, as shown in figure 21. However, the version of the MEPDG used in the analysis (version 1.003) did not allow the base modulus to be changed for the level 1 rehabilitation module (i.e., it reverted to the default values). Therefore, a suitable correction factor had to be used to obtain the desired input value. For example, the default modulus for AASHTO A-1-b material was 38,000 psi (262,000 kPa). To obtain an input of 10,400 psi (71,700 kPa) (16,700 psi × 0.62 (115,100kPa × 0.62)), a factor of 0.27 (or 0.62 × 0.44) was used. The same was done for the subgrade modulus.

Click for description
1 psi = 6.89 kPa.
1 inch = 25.4 mm.

Figure 21. Screen Capture. Base layer input for overall design level 1.

 

Unbound layers can be entered as “Representative value,” or the user can select the “Integrated Climatic Model (ICM)” analysis. Because backcalculated values were not available throughout the year for this case study location, the ICM analysis was used. ICM inputs (gradation, Atterberg limits, etc.) were obtained from the LTPP Program database, as summarized earlier.

Evaluation of Design Results

As noted previously, the analyses summarized in table 17 were run with the MEPDG software (version 1.003) to evaluate the influence of varying the backcalculation-based inputs; the nationally calibrated performance models were used. The HMA overlay thickness requirement was determined for each set of inputs, and the change in distress predictions and reliabilities were compared.

The required HMA overlay thickness for all analyses was 3 inches (76 mm), except for analysis4, which was 3.5 inches (89 mm), to achieve a 90-percent reliability level. The required HMA overlay thickness was controlled primarily by top-down cracking (summarized in table 18 through table 20 and illustrated in figure 22), with other distress predictions showing minimal differences between the analyses.

Analysis 4 had the greatest predicted top-down cracking. While analysis 4 had the weakest base layer modulus, it had one of the higher subgrade moduli. The existing HMA modulus for analysis4 was also nearly the same as analyses 2 and 3. Even though analysis 5 had the lowest subgrade modulus, the top-down cracking was lower than analyses 2 and 3 (uncorrected and corrected backcalculation-based, respectively). However, the permanent deformation for this analysis increased compared with analysis 3, which was partly attributed to the lower HMA modulus. Analysis 6 (four-layer system) was comparable to the three-layer systems.

Table 18. Summary of design results for 76-mm (3-inch) HMA overlay (analyses 1 and 2).
Pavement Performance Measure Default Values Analysis1 Uncorrected Backcalculation-Based Analysis 2
Quantity Reliability Quantity Reliability
Terminal IRI (inches/mi) 109.3 98.06 97 99.67
AC surface down cracking (longitudinal cracking) (ft/mi) 5.2 99.91 62.7 91.65
AC bottom-up cracking (alligator cracking) (percent) 0 99.999 0 99.999
AC thermal fracture (transverse cracking) (ft/mi) 1 99.999 1 99.999
Chemically stabilized layer (fatigue fracture) N/A N/A N/A N/A
Permanent deformation (AC only) (inches) 0.13 99.43 0.08 99.999
Permanent deformation (total pavement) (inches) 0.38 99.999 0.08 99.999
1 inch/mi = 0.0158 m/km.
1 ft/mi = 0.19 m/km.
1 inch = 25.4 mm.
N/A = Not applicable.

 

Table 19. Summary of design results for 76-mm (3-inch) HMA overlay (analyses 3 and 4).
Pavement Performance Measure Corrected Backcalculation-Based Analysis 3 Corrected Backcalculation-Based Analysis 4
Quantity Reliability Quantity Reliability
Terminal IRI (inches/mi) 97 99.67 97.1 99.67
AC surface down cracking (longitudinal cracking) (ft/mi) 69.5 91.01 165 85.5
AC bottom up-cracking (alligator cracking) (percent) 0 99.999 0 99.999
AC thermal fracture (transverse cracking) (ft/mi) 1 99.999 1 99.999
Chemically stabilized layer (fatigue fracture) N/A N/A N/A N/A
Permanent deformation (AC only) (inches) 0.07 99.999 0.08 99.999
Permanent deformation (total pavement) (inches) 0.08 99.999 0.08 99.999
AC = Asphalt concrete.
N/A = Not applicable.
1 inch/mi = 0.0158 m/km.
1ft/mi = 0.19 m/km.
1 inch = 25.4 mm.

 

Table 20. Summary of design results for 76-mm (3-inch) HMA overlay (analyses 5 and 6).
Pavement Performance Measure Laboratory-Based Analysis 5 Corrected Backcalculation-Based Analysis 6
Quantity Reliability Quantity Reliability
Terminal IRI (inches/mi) 100.7 99.39 97.9 99.62
AC surface down cracking (longitudinal cracking) (ft/mi) 49.4 93.07 50.4 92.95
AC bottom-up cracking (alligator cracking) (percent) 0.3 99.999 0 99.999
AC thermal fracture (transverse cracking) (ft/mi) 1 99.999 1 99.999
chemically stabilized layer (fatigue fracture) N/A N/A N/A N/A
Permanent deformation (AC only) (inches) 0.12 99.72 0.08 99.999
Permanent deformation (total pavement) (inches) 0.16 99.999 0.12 99.999
AC = Asphalt concrete.
N/A = Not applicable.
1 inch/mi = 0.0158 m/km.
1ft/mi = 0.19 m/km.
1 inch = 25.4 mm.

 

Click for description
1ft/mi = 0.19 m/km.
Figure 22. Graph. Top-down cracking distress prediction from MEPDG design program for analysis 3.

 

SUMMARY

Three backcalculation programs (MODTAG, MICHBACK©, and EVERCALC©) were used to analyze FWD deflection data from various test locations (stations) within the project. The deflection data showed considerable variability within the project. The RMS values obtained from the three backcalculation programs were also generally very high. It is recommended that one should conduct FWD testing at multiple locations and use the average backcalculated layer moduli. This should provide consistent recommended overlay designs.

For this case study, the MEPDG results indicated that surface-down cracking was critical in rehabilitation design of HMA overlay over existing HMA pavements; although this was minimal even considering a 3-inch (76-mm) HMA overlay was satisfactory for nearly all of the input combinations. Within the ranges identified, the selection of inputs was more critical as one approached lower values for any layer.

In addition, the following procedures are recommended:

 

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101