Pavement Structural Evaluation at the Network Level: Final Report

CHAPTER 9. SUMMARY AND CONCLUSIONS

The goal of the project was to establish a reliable measure of the structural condition of bound pavement layers over time based on moving pavement deflection technology measured at posted traffic speeds. The specific project objectives were as follows:

• Assess, evaluate, and validate the capability of traffic speed devices that measure deflection or other pavement responses for pavement structural evaluation at the network level for use in pavement management application and decisionmaking.
  
  
• Develop methodologies for enabling the use of the device(s) capable of meeting the first objective in pavement management or alternatively develop recommendations to further develop promising device(s) and/or technologies if the devices do not meet the first objective.

To accomplish the stated goal and objectives, 10 tasks were carried out under 2 phases. The initial phase focused on identifying and assessing capable devices, while the second phase focused on evaluating and validating the field capable devices.

Much of the first phase focused on gathering information related to potentially viable devices. A literature review was performed to investigate and evaluate previous, ongoing, and proposed research projects related to available traffic speed pavement deflection devices that have the potential to meet the project objectives. Based on the literature, an RWD and TSD were found as potentially viable devices that merited further evaluation.

To augment the literature review findings, questionnaires were developed and sent to the device manufacturers as well as owners and users of the devices. Follow-up interviews with specific or clarifying questions were also conducted. These activities further reinforced the RWD and TSD as potential devices capable of meeting the project objectives.

In light of these findings, a work plan (driven by the planned analyses) for the conduct of field trials was developed to perform the following:

• Confirm that the two TSDDs in question met a minimum set of specifications related to the structural evaluation of pavements at the network level, including accuracy and precision requirements.
  
  
• Propose processes to incorporate pavement structural information from the TSDDs into network-level PMS applications.

The field trials were performed at the MnROAD facility near Albertville, MN, because it provided a multitude of test sections in one location and readily available information, including environmental and dynamic load response data. Field trial testing was also planned on an 18-mi (29-km) loop located near the MnROAD facility in Wright County, MN.

In addition to the existing sensors, four geophones and one accelerometer were installed to measure deflection velocity and displacement parameters at four MnROAD cells (three flexible cells covering a range of stiffnesses and one rigid pavement cell). Data from these sensors were used to estimate the accuracy of the TSDDs by statistically comparing the results measured with the newly installed sensors with those reported by the TSDDs at the four cells on three separate repeat passes.

Conversely, the precision analysis included almost all cells of the MnROAD facility as well as the 18-mi (29-km) loop to account for different pavement structures and other factors such as vertical and horizontal curves. To better evaluate the precision of the TSDDs, they were tested at the MnROAD facility at different speeds and at different times of the day. Data were collected up to five times and at two different speeds. Deflection data for the replicate passes at similar speeds and times were statistically compared to estimate precision.

Given the amount of data collected to facilitate the project data analyses, an online database was developed for ease of update and instant access to various data. The raw, reduced, and analyzed data from the accuracy and precision analyses were placed in the database. This database was also populated with other relevant data such as cell and sensor inventory, ambient conditions, pavement structure, and pavement condition (e.g., IRI measurements).

While the performance of the RWD and TSD varied under different field trials and conditions, it was found that both devices were capable of providing reasonably accurate and precise pavement response measurements. The findings from the accuracy and precision analyses were used later in the project to recommend the optimum operational conditions and device limitations. It is important to recognize that the conclusions and recommendations derived from the accuracy and precision analyses were limited by the amount of data available to the project and the precision and accuracy of the sensors used in this study.

Having established that the TSDDs measurements were acceptable, the 3D-Move software, which estimates dynamic pavement responses at any given point within the pavement structure using a continuum-based finite-layer approach, was calibrated using data from the MnROAD facility field trials. The objective of this calibration was to enable the use of the 3D-Move software in the development of methodologies for incorporating TSDD measurements into network-level PMS applications. A key element in the calibration was simulating pavement surface deflections using numerical models with a focus on understanding the parameters that affect the TSDD measurements. Those parameters include changes in TSDD vehicle speed, pavement layer properties (e.g., age and moisture), and vehicle loading (e.g., tire configuration, load, and inflation pressure).

The 3D-Move analyses were calibrated using inputs derived based on the following considerations:

• Existing MnROAD database of pavement layer properties (thicknesses and physical properties) and recent FWD deflection data.
  
  
• Representative layer material properties based on backcalculation of FWD deflection data and, subsequently, viscoelastic AC properties as a function of frequency extrapolated for the TSDD field trial temperatures using Witczak's model and AC middepth layer temperatures derived using the BELLS equation.
  
  
• Static tire loads measured at the MnROAD facility weighing station.

Numerous 3D-Move analyses were performed to bracket the computed deflection time histories (peak and basin) with the measured ones from the project geophones. The 3D-Move software was further calibrated using strain measurements taken by the MnROAD strain gauges at various interior pavement locations. Since load-induced strains are critical inputs to pavement performance predictions, this effort was considered critical in ascertaining the applicability of the 3D-Move for pavement response predictions to be used in identifying the most promising indices from TSDD measurements that best relate to pavement structure.

The 3D-Move maximum strains correlated well with the MnROAD sensor measurements. Accordingly, it was further concluded that 3D-Move captures the pavement strain responses well, and therefore, can be used to evaluate pavement responses under TSDD loadings.

Pavement structural capacity can be estimated from performance prediction equations, which relate load-induced pavement responses to one or both of the following pavement distresses: AC fatigue cracking and rutting subgrade rutting. The critical load-induced pavement responses that relate to these two distresses are the maximum tensile strain at the bottom of the AC layer and the vertical compressive strain on top of the subgrade, respectively. The focus of this project was on the AC fatigue cracking and therefore maximum tensile strains at the bottom of the AC layer. An analytical investigation was undertaken using a calibrated 3D-Move analysis to explore the relationships between load-induced structural-related responses of a pavement system and the corresponding surface deflection basin-related indices.

A number of previous studies have suggested that deflection-based indices relate reasonably well to the structural capacity of pavements. For the purposes of this project, it was important to identify indices that best correlate with the critical pavement responses. The selection of the best indices was conducted using the following three-step process:

1. The surface deflection indices that correlated well with the critical pavement responses were initially identified using the 3D-Move calibration results (43 datasets), which were based on the TSDD field trials carried out at the MnROAD facility.
   
   
2. A sensitivity analysis of the correlations associated with various deflection indices (not limited to those indices identified in step 1) was undertaken using a set of 36 pavement structures (different combinations of layer thicknesses and moduli) at several vehicle speeds as input to 3D-Move.
   
   
3. The robustness of the correlations identified in the first two steps was further explored by considering a much larger database of approximately 15,000 pavement structures generated using the layered elastic program JULEA.

Based on the results from the first step, the indices that best relate to the maximum horizontal strains at the bottom of the AC layer were as follows:

• R1₁₂ based on the FHWA equation and deflections measured at radial distances of 0 and 12 inches (0 and 304.8 mm) from the load center.
  
  
• R2₁₈ based on Horak's equation and deflections measured at radial distances of 0 and 18 inches (0 and 457.2 mm) from the load center.
  
  
• SCI₁₂ based on deflections measured at radial distances of 0 and 12 inches (0 and 304.8 mm) from the load center.
  
  
• SD₁₂ based on deflections measured at radial distances of 0 and 12 inches (0 and 0.3 m) from the load center.

In addition, the following were determined:

• Using D₀ or D₄ instead of D_max to calculate SCI is reasonable.
  
  
• The proposed indices (i.e., R1₁₂, R2₁₈, SCI₁₂, and SD₁₂₎ were not sensitive to the TSDD load characteristics (i.e., tire spacing, tire load, and inflation pressure).

Similarly, the following major observations and conclusions relating to deflection indices and AC tensile strains (study focus) were made from the combined results from the first two steps:

• SCI₁₂ based on deflections measured at radial distances of 0 and 12 inches (0 and 304.8 mm) from the load center best reflects the role of the upper portion of the pavement system, which was significantly influenced by the AC thickness and modulus.
  
  
• TS, which is the ratio between the vertical particle velocity and the vehicle speed, is analogous to deflection slope directly reported by the TSD. TS showed high sensitivity to changes in the pavement structural properties while avoiding potential errors in the deflection basin computed from the TSD measurements. Even though TS₈ was one of the most appropriate indices, sensitivity analyses showed that TS₄ correlated well with the horizontal strains for AC thicknesses of 3 to 6 inches (76.2 and 152.4 mm), while TS₈ correlated well for AC thickness greater than 6 inches (152.4 mm).
  
  
• AC thickness is a critical parameter that should be used to categorize pavements into groups. Subgrade modulus, AC modulus, and loading speed only marginally affected the indices in table 77. R2₁₈, based on Horak's equation and deflections measured at radial distances of 0 and 18 inches (0 and 457.2 mm) from the load center, is a good index only minimally affected by the AC thickness.

Based on the results of the first two steps, it was concluded that for estimating the maximum horizontal strains at the bottom of the AC surface layer, pavements can be grouped into the following categories:

• AC surface layer thickness less than 3 inches (76.2 mm).
  
  
• AC surface layer thickness between 3 and 6 inches (76.2 and 152.4 mm).
  
  
• AC surface layer thickness greater than 6 inches (152.4 mm).

In the third step, a wider range of pavement structures was analyzed using the layered LE program, JULEA.⁽²⁶⁾ A database of 15,000 pavement structures over a wide range of layer moduli and thicknesses were simulated using the Monte Carlo technique. The corresponding pavement responses (strains and deflections) were computed for each simulated pavement structure. A procedure similar to the one used in the 3D-Move analyses was used to compute the maximum fatigue strain. Similarly, surface deflections at the same locations used for the first step were computed. The database was used to identify the most sensitive pavement properties that affect the critical responses and the most sensitive indices that correlate well with the critical responses.

The thickness of the AC layer was found to be the most sensitive pavement property that affected the pavement responses in question. The pavement structures in the JULEA database were then grouped based on the AC layer thickness. For pavement structures with a thin (i.e., less than 3 inches (76.2 mm)) AC layer, the stiffness of the base, AC, subgrade, and base thickness significantly influenced the critical pavement responses. A weak correlation between the deflection indices and critical strains was observed only when the AC thickness was considered in the development of correlations. However, for network-level applications, a relationship involving several material properties was not practical and thus was not pursued further.

For pavement structures with AC layer thicknesses between 3 and 6 inches (76.2 and 152.4 mm), it was found that the most sensitive indices for the maximum fatigue strain response were as follows:

• R1₈ based on deflections measured at radial distances between 0 and 8 inches (0 and 203.2 mm) from the load center.
  
  
• SCI₈ based on deflections measured at radial distances between 0 and 8 inches (0 and 203.2 mm) from the load center.

For pavement structures with an AC layer thickness between 6 and 16 inches (152.4 and 406.4 mm), the indices found to be most sensitive for the maximum fatigue strain response were as follows:

• DSI_{4 - 18} based on deflections measured at radial distances between 4 and 18 inches (101.6 and 457.2 mm) from the load center.
  
  
• SCI₁₈ based on deflections measured at radial distances between 0 and 18 inches (0 and 457.2 mm) from the load center.

For pavement structures for which the AC layer thickness was unknown, the R2₁₂ index based on Horak's equation and deflections measured at radial distances of 0 and 12 inches (0 and 304.8 mm) from the load center was found to be a reasonable predictor of maximum fatigue strain response for both thin and thick pavements. Accordingly, when the AC thickness information is not available, R2₁₂ can be used to estimate the maximum fatigue strain.

Because incorporating TSDD measurements into network-level PMS applications requires establishing a relationship between the computed or measured indices and the critical pavement responses, the critical responses were then related to the corresponding indices deemed sensitive. For thin AC layers, separate relationships were developed for those cases where the AC layer was thicker and thinner than the base layer.

The final phase of the selection of the indices was to conduct a balance evaluation of the uncertainties associated with the models and relationships developed and the effect of precision and accuracy of the devices on the computed indices. The accuracy of the devices was defined as the median accuracy of the index computed from all tests carried out for that purpose at MnROAD. Precision is defined as the ratio of the median of the SEE among replicate runs from all tests carried out in Minnesota divided by the corresponding median deflection index. The uncertainty of the model was assessed by dividing the SEE of the strains for each index divided by the median of the strain estimated. The R² values were also used as a second parameter related to the appropriateness of the relationships.

Based on the criteria, the most optimized indices for each pavement thickness range are reported in table 80. The recommended indices include the following:

• DSI_{4 - 12}, based on deflections at 4 and 12 inches (101.6 and 304.8 mm) from the load center, which was formulated as part of the study, was found to be the most appropriate index regardless of AC thickness (whether known or unknown).
  
  
• SCI₁₂, based on deflections at 0 and 12 inches (101.6 and 304.8 mm) from the load center, performed nearly as well as DSI_{4 - 12}.

Data analyses have shown that indices that can be derived from TSD measurements provide robust assessment of pavement structural condition at the network level. Improvements in the number of sensors and their locations are needed to use the recommended analyses methodologies with the RWD, but these should not be difficult to achieve.

1 inch = 25.4 mm

Recommendations for incorporating the project findings and conclusions into network-level PMS applications are detailed in the next chapter, including data requirements, selection of indices, computation of strains, temperature correction of strains, and various approaches for relating strains to pavement structural capacity.