Pavement Structural Evaluation at the Network Level: Final Report

CHAPTER 2. LITERATURE REVIEW

The objectives of the literature review were to investigate and evaluate previous, ongoing, and proposed research projects relating to available TSDDs that have the potential to meet the project objectives. Publications by the National Cooperative Highway Research Program, State transportation departments, and the Federal Highway Administration (FHWA) as well as publications discussing overseas agency practices and others related to the topic were reviewed.

During the literature review process, the technology presented was assessed for consideration under task 2, "Identification and Assessment of Capable Devices." Detailed records of each reference were kept including information on the location of the materials, relevance to the topic, and significant findings from the work. Each record contained an index, date completed, author, title, source, source document, and abstract. The relevance of each document was used to prioritize the reports to undergo further detailed review.

2.1 References Reviewed

A total of 22 references were reviewed as part of the project, most of which were recent (i.e., from the last 4 years). (See references 1 and 3–23.) Two of these references contain comprehensive literature reviews on TSDDs until 2011.^(3,19) Table 1 shows the distribution of references by reference source, and table 2 shows the distribution of references by type of equipment. Some of the references covered more than one type of equipment, while a few were non-equipment-related references covering analysis methodologies. Table 3 shows the distribution of references by subject matter; some of the references covered more than one subject matter.

2.2 Literature Review Findings

Arora et al. summarized the state of the art of continuous deflection devices by investigating an Airfield RWD, a Swedish road deflection tester, a Texas rolling dynamic deflectometer, a Danish high-speed deflectograph, and the RWD from this study.⁽³⁾ The research also investigated new methods for structural evaluation, including modulus, deflection ratios, modified modulus and deflection ratios, a method using structural numbers (SNs), an alternative method for determining SN from FWD data, and a simple approach to estimate the SN of pavements. The key findings from Arora et al. are as follows:⁽³⁾

• The best return on investment in acquiring and operating a rapid continuous deflection device is realized when measurements can be used to delineate pavement sections that are structurally marginal. This would allow the agencies to program the maintenance or rehabilitation of the sections in subsequent fiscal years.
  
  
• A high-speed device with more than one deflection sensor is desirable. Devices with one deflection sensor can delineate the pavement sections that are structurally in poor condition, but the uncertainty in delineating structurally marginal pavement sections may be high.
  
  
• The uncertainty in delineating structurally marginal pavements is related to the precision and accuracy of the measured deflections. The greater the uncertainty in measured deflections and load, the greater the risk of not properly identifying structurally marginal sections will be.
  
  
• A concerted effort is needed to quantify the variability associated with measured deflections from these devices beyond the spatial variability in deflection measurements due to changes in the natural properties of the subgrade soil and construction-related variability.
  
  
• Institutionally, the current FWD-based evaluation and scoring schemes incorporated in the Texas Department of Transportation pavement management information system should be gradually changed if a high-speed deflection device is used.

Rada and Nazarian published a report entitled The State-of-the-Technology of Moving Pavement Deflection Testing.⁽¹⁹⁾ Some of the major conclusions from their report are as follows:

• Despite the need for further improvements, the TSD investigated in this study can be used for two of the pavement applications identified: identification of pavement changes/anomalies for use at the network level and the project level and determination of overall pavement structural capacity indicators/indices at the network level. With some modifications, the TSD and RWD can be used for multiple applications within the next 5 years.
  
  
• An area of concern with the RWD and TSD seems to be the compromise between the repeatability and loss of details with spatial averaging. Depending on the application, an optimized level of repeatability should be achieved over a reasonable spatial averaging. The threshold values for the repeatability and spatial averaging should be defined.

Rada and Nazarian also provided a comprehensive list of future research needs to improve the accuracy and repeatability of the moving pavement deflection testing devices evaluated.⁽¹⁹⁾ Those needs were grouped into three major categories: equipment-related, measurement-related, and application-related issues.

Flintsch et al., as part of the Second Strategic Highway Research Program (SHRP2) Project R06(F), evaluated continuous deflection devices to support pavement management decisions.⁽¹³⁾ The study identified both the TSD and RWD as capable of meeting the criteria for speed, load, and data collection and being close to production mode. Field verification for the TSD showed that it could be used for network-level data collection with adequate repeatability and that although there was significant variation and bias between deflection measurements and surface indices computed using TSD and FWD measurements, the results were broadly comparable.⁽¹³⁾ The study recommended several improvements to the devices such as adding additional sensors to provide a more complete deflection bowl, measuring pavement layer thickness with adding a ground penetrating radar (GPR) device, and measuring the dynamic load on the loading wheel assembly.⁽¹³⁾ As is discussed in section 3.1, Device Manufacturers' Perspectives, improvements have been made to both the TSD and RWD, including increased number of sensors and ability to measure the dynamic load on the wheel assembly since the assessment was conducted during the SHRP2 study.

Many other studies have investigated TSDDs. Several studies showed consistently repeatable results for the TSD and RWD investigated in this study. (See references 5, 10, 11, 13, and 16.) Hausman and Steele evaluated the use of deflection measurements from the RWD to classify roadways structurally for inclusion in a pavement management matrix of treatment selection.⁽¹⁵⁾ They showed that the RWD could result in higher reliability and lower costs, with savings of 5 percent when used to support pavement preservation practices. For example, figure 1 illustrates a complete treatment matrix based on RWD deflections, pavement condition, and traffic levels. It should be noted that the RWD-based treatment matrix proposed relies on one single vertical deflection at the mid-point between the dual tires. This mid-point deflection represents the vertical response contribution from all pavement layers. The utilization of figure 1 requires the Pavement Condition Index rating of the pavement be known a priori.

Source: David Hein

1 mil = 0.0254 mm

Figure 1. Illustration. Treatment matrix for RWD and pavement condition.

Ferne et al. chronicled the evaluation and implementation of the TSD for network-level structural evaluation in the United Kingdom, including acceptance testing as well as assessing the effects of testing speed, road surface type, and temperature on TSD measurements.⁽¹¹⁾ The study showed that the TSD had good short-term repeatability with relatively low standard deviation and that it was capable of differentiating structural strengths equally as well as the deflectograph and FWD.⁽¹¹⁾

Thyagarajan et al. developed relationships between curvature indices (based on deflection measurements) and tensile strain at bottom of the asphalt concrete (AC) layer as indicators for progressive deterioration of pavement.⁽²¹⁾ They showed deflection measurements can be used to determine tensile strain at bottom of AC layer and used to form structural performance curves for use in PMS.

Gedafa et al. conducted a study comparing deflections measured by the RWD and FWD in Kansas and showed that the center deflections as well as the SNs computed using the devices were statistically similar.⁽¹⁴⁾ As a result of that study, they recommended network-level deflection survey using the RWD on a 4-year cycle. Elseifi et al. conducted a similar comparison between the RWD and FWD in Louisiana.⁽¹⁰⁾ The study showed that the repeatability of the RWD measurements was acceptable with a coefficient of variance of 15 percent and that both the RWD and FWD data properly reflected the pavement condition. Although the RWD measurements were in general agreement with the FWD measurements, the deflections from the two devices were statistically different at 15 out of 16 sites.⁽¹⁰⁾ The study concluded that calculating SN (independent of pavement thickness and layer properties) based on RWD measurements could be used as a screening tool to identify structurally deficient pavements at the network level.⁽¹⁰⁾ A similar comparison between RWD and FWD measurements in Virginia showed that RWD and FWD results were not well correlated, that repeated RWD measurements were not statistically similar for 8 of the 15 runs, and that RWD standard deviation measurements fluctuated with changes in surface mix type.⁽⁷⁾ Based on those findings, it was recommended that the RWD not be pursued for network-level analysis for interstate type facilities in Virginia. It was also suggested that a comprehensive review of the Virginia study is required since it is not known whether the roles of parameters, such as pavement temperature, vehicle speed, and moisture, have been appropriately included in the pavement maintenance and rehabilitation (M&R) and reconstruction decisionmaking. In addition, as previously noted, the RWD measurement of vertical displacement between the tires has a known limitation that it may not fully represent the characteristics of the bound layers alone.

Zhang et al. introduced the use of the Structural Condition Index as a screening tool to discriminate pavements that need structural reinforcement from those that do not.⁽²²⁾ Structural Condition Index is based on the SN of the pavement as determined using FWD data. The study recommended criteria of Structural Condition Index for M&R activities. Stubstad et al. described the use of deflection data to predict pavement performance for certain distresses based on FWD deflection data.⁽²⁰⁾ Bryce et al. developed a structural index for use in network-level pavement evaluation known as the Modified Structural Index (MSI).⁽⁶⁾ MSI is a modified Structural Condition Index for use in the Virginia Department of Transportation's PMS. MSI was selected as a result of the network-level predictions using MSI as the most promising index to predict project-level activities.⁽⁶⁾ MSI can be used as a network-level screening tool, for deterioration modeling, or to develop structural performance measures.⁽⁶⁾ However, MSI can only be used for flexible pavements. It is based on FWD data, and it is empirical in nature.

Austroads published a study in 2012 that evaluated the use of the TSD in Australia.⁽⁴⁾ Although results had shown consistently repeatable results in Europe, the roads in Australia are often quite different, consisting of a considerable amount of granular (unpaved) pavements. Consideration was also given to the harsher climate experienced in Australia compared to Europe. The TSD was evaluated during the Australian summer (December 2009–March 2010) by surveying around 11,185 mi (18,000 km) of the New South Wales and Queensland road networks. The assessments of the technology based on that study were that the TSD could be an effective screening tool at the network level and that it showed considerable promise for design of overlays of granular pavements.⁽⁴⁾

As a follow up to that evaluation study of the TSD, Austroads conducted a survey of eight member agencies (MAs) to determine their interests in network-level strength assessment based on deflection measurement.⁽¹⁷⁾ Seven of the eight MAs expressed interest in having network-level strength assessments and that they would use a TSD for screening out weak and vulnerable pavement sections for further evaluation. Six MAs would use a TSD to estimate pavement rehabilitation and reconstruction budgets, and three MAs would use it to evaluate the performance of maintenance management arrangements.⁽¹⁷⁾ The main concerns expressed by respondents for using the TSD for network-level evaluation were the ability to relate TSD deflection data to FWD deflection data as well as developing robust models based on TSD deflection data.⁽¹⁷⁾

Using the field data from the 2012 Austroads study, a revised approach for analyzing TSD data to predict the full deflection bowl was performed by Muller and Roberts.^(4,18) The study proposed a new methodology for fitting curves based on surface deflection velocities measured by the TSD versus the wheel offset and determining the deflected pavement profile as the cumulative area under the plot of V_V/V_H versus wheel offset, where V_Vand V_H represent the vertical and horizontal velocities, respectively.⁽¹⁸⁾ Comparisons were made between the predictions of deflection under the center of the load, D₀ and Surface Curvature Index, SCI₃₀₀ (defined as the difference between D₀ and D₃₀₀ (i.e., 12 inches (300 mm) from the center of the load (D₀ – D₃₀₀) from the proposed methodology, the original Danish analysis, and FWD measurements.⁽²³⁾ The results showed a clear correlation with the shape and magnitude for the deflection bowls of the FWD measurements and the proposed TSD methodology deflection bowls, resulting in d₀ and SCI₃₀₀ predictions from the TSD being on average 6.4 and 16.6 percent higher than the corresponding FWD measurements, respectively.⁽¹⁸⁾

Dynatest^® began redeveloping its version of the RWD based on a former Airfield RWD that was noted in the report by Rada and Nazarian to have been surplused once funding was discontinued.⁽¹⁹⁾ Based on information obtained from the manufacturer in 2011, this new RWD is supposed to utilize the process of triangulation to determine the deflection caused by a moving wheel load. Moreover, it builds on Harr's algorithm, which was developed in the 1970s for a fast-moving heavy wheel load utilizing four sensors where the sensors are equidistant apart and the measurement locations are separated by the same sensor spacing distance.⁽⁹⁾

However, Harr's algorithm only holds true if the sensor nearest the wheel load is the only sensor within the deflection basin, which is not always a correct assumption.⁽⁹⁾ Deflections calculated using Harr's algorithm were compared to simulated deflections using the layered elastic program Waterways Experiment Station Linear Elastic Analysis (WESLEA) for 27 different pavement sections with various layer thicknesses and layer moduli.⁽²⁴⁾ Results showed that the Harr algorithm underestimated the deflection.⁽⁹⁾ By using more than four sensors, one can correct the influence of the deflection basin on the outer sensors provided that the sensor spacing is greater than the equivalent thickness of the pavement layers.⁽⁹⁾ Figure 2 shows a five-sensor configuration and two measurement locations at a distance equal to the sensor spacing as the load travels down the pavement, as designated by the prime and double prime labels.

Figure 2. Illustration. Five-sensor triangulation configuration.⁽⁹⁾

The deflection under the wheel, d, at sensor E" is calculated using the equation in figure 3, where k represents the constant and h_D represents Harr's algorithm. Figure 4 contains an equation to calculate h_D from figure 3_. The constant k is based on the equivalent thickness. Comparing the deflections calculated using the equation in figure 3 to the model computed theoretical deflections for the 27 pavement sections evaluated previously results in a coefficient of determination (R²) of 0.9952.⁽⁹⁾

Figure 3. Equation. Deflection under wheel.

Figure 4. Equation. h_D.

By increasing the number of sensors up to nine, the deflection under the wheel, the deflection measured at a distance one sensor from the wheel load, the deflection measured at a distance of two sensors from the wheel load, and the subgrade modulus can be determined. Utilizing all three deflections and subgrade modulus, the vertical strain at the top of the subgrade layer and the horizontal strain at the bottom of the asphalt layer can be calculated. Comparing the vertical and horizontal strains to the theoretical stain measured based on the 27 pavement sections resulted in R² values of 0.99 and 0.91, respectively.⁽⁸⁾

2.3 Summary

As a result of the literature reviewed and the findings summarized in this chapter, it was concluded the both the RWD and the TSD are potentially viable devices, which merit further evaluation. Although the Dynatest^® RWD device also appears to be promising, until it is actually functional, it is premature to label it as a potentially viable device. It was anticipated that a functional Dynatest^® RWD would be available early in the project; however, that was not the case, and the device was dropped from further consideration in this project. Dynatest^® subsequently indicated that they expect to have a functional device by late 2017.