4.4 Backcalculation Methodology
This procedure is used to determine modulus values for in situ pavement materials when these materials are used as is (unmodified) in FPS design. TxDOT currently uses version 7.0 of the MODULUS software for backcalculation of deflection data collected by the FWD. Version 7.0 is comparable to version 6.0; the main difference is the ability to read current Dynatest R80 formatted data files that have GPS locations embedded. The software (v 7.0) and user’s manual (v 6.0) in PDF document format are available through the MNT link to the TTI on-line pavement design training site. Also, basic operation and discussion of inputs, cautions, and example problems is presented in the
Flexible Pavement Rehabilitation Strategies
training course and Flexible Pavement Design
workshop.The raw deflection file, pavement layer thicknesses, layer Poisson ratios, probable layer moduli ranges, and asphalt temperatures at the time of testing are all required inputs to perform backcalculation. The backcalculation process works on the assumption that the pavement structure can be modeled as a linear-elastic layered system. If the parameters of layer thickness, deflection, and Poisson ratio are known, the modulus can be approximated. A likely range of “probable” layer moduli provided by the program user facilitates the process by forming the basis of a small internal database against which mathematically generated deflection bowls are compared to the actual measured deflection bowl by the software. Once a reasonable match is made, the moduli that allow this match are reported as the individual layer moduli. In addition, the program reports a depth to stiff layer or bedrock.
4.4.1. Backcalculation Limitations and Adjustments
There are precautions and limitations to the backcalculation procedure that the user must consider. In the end, engineering judgment will be needed to decide on the veracity of solutions generated. The following are some pointers when using MODULUS 7.0:
- The modulus for layers thinner than 3.0 in. cannot be backcalculated. This situation arises most often for thin-surfaced flexible pavements. The user must assign a reasonable modulus to this layer (minimum and maximum are input as the same value in the program) based on thickness, level of distress, temperature, etc.
- The surface layer is always the layer that the load plate is in contact with, so a thickness must be entered. Where the surface is a bituminous surface treatment, it is allowable to use a nominal thickness, such as 0.5 in., and assign a nominal modulus, such as 200 ksi. Alternatively, the surface treatment can be combined with the underlying layer as the “surface,” reducing the total number of layers by one. In determining the seed moduli range for the surface, MODULUS assumes the layer is HMA and automatically fills the min/max seed values in accordance with the temperature posted in the Asphalt Temp cell. Where non-bituminous materials are the surface during testing, the user must insert seed values commensurate with the type of material tested.
- The maximum number of layers for which the modulus can be backcalculated is four (one of which is always the natural subgrade) in MODULUS 7.0. For circumstances where more layers are known to exist, the user must either consolidate or ignore layers. Consolidation is recommended for materials that are more likely to have a similar modulus and shear strength properties (i.e., different types of HMA, or flex base over reclaimed base). Ignoring layers may be reasonable in certain cases where the material’s contribution to the overall stiffness of the structure is minimal (i.e., “foundation course,” or lime treated subgrade – constructed as a working platform).
- There are times when a more reasonable solution is obtained modeling your pavement structure as a 3-layer system, even if you know there are four layers present. This situation may develop for a number of reasons, such as variable stabilization (leaching), variable depth to bedrock, etc.
- A check of reasonableness in the solutions generated shall be made. Reasonableness is more related to the in-place stiffness characteristics of the layers being modeled and not necessarily to the size of the average errors reported by the software in comparing the mathematically generated bowls to the measured bowls. While the 4-layer solution will generally give lower overall errors, the backcalculated material moduli may be unrealistic with respect to the in-place material, and the variability of reported moduli may be excessive (coefficient of variation 100% or greater). When there is doubt of reasonableness, the user should perform backcalculation runs using both 3- and 4-layer solutions (employing guidelines given in the third bullet). Additional field testing, such as with the dynamic cone penetrometer (DCP) along with engineering judgment, is necessary to ensure a valid, reliable solution.
- A check of the MODULUS summary table shall be made to detect outliers that skew the average value reported. Outliers may be the result of full-depth patches (different pavement structure) or very weak areas.
- For the purpose of using MODULUS-reported values as input to FPS 21, adjustment of the average modulus shall be considered; otherwise the performance of any pavement design solution based on these inputs could be jeopardized. As a rule-of-thumb, consider removing values that exceed one standard deviation from the unadjusted average, and then re-average. This should always be done for modulus values that are much higher than values that are more typical for the section. Consideration can be given to eliminating very low valuesonly ifthe intention is to include a bid item for repair of weak areas (i.e., Item 351, “Flexible Pavement Structure Repair,” or Item 354, “Planing and Texturing Pavement”) as part of the job.
- Shallow bedrock (typically less than 60 in. deep) will almost always result in underestimation of the subgrade modulus and overestimation of the flexible base modulus and produce very high average error (> 20%). The recommended workaround is to fix the depth to bedrock (DTB) at 120 in. or, alternatively, 240 in. if the solution using the program-generated DTB produces suspect subgrade/base moduli. Another clue that the default solution is suspect would be if the ratio of the flexible base modulus (unstabilized layers only) to the subgrade modulus is very high (> 5). If the user opts to fix the calculated DTB to a value in the 120- to 240-in. range, then this user-selected value must also be used in FPS 21 design.
- MODULUS can perceive a shallow DTB in high water table situations (water is incompressible) as can be the case in east Texas. It may be beneficial to override the program-generated DTB value by using a fixed value of 120 in. Again, check the generated subgrade/base moduli values for reasonableness.
- Soft upper subgrade can also lead to high errors in the backcalculation process. In these cases, use a 4-layer solution where the soft portion of the subgrade is modeled as the subbase layer (fix depth at 12 in.) to provide better fit and more realistic backcalculated values for the base and deep subgrade. A check can be made in MODULUS using the Boussinesq procedure to evaluate how the subgrade modulus varies with depth. Verification with a dynamic cone penetrometer (DCP) may be warranted.
4.4.2 Modulus Correction Factors
In addition to adjustments made to backcalculated average modulus values for outliers in the deflection data set, correction factors must be applied to the backcalculated HMA values. This is particularly important for HMA surfaces greater than 4 inches in thickness. FPS 21 considers the modulus of bituminous materials only at the reference temperature of 77°F. FWD data are rarely collected at the reference temperature, corrections must be made for FPS input. Since modulus values for surfaces thinner than 3 inches cannot be backcalculated, the designer must use engineering judgement in discerning the design modulus value (at design temperature 77˚F, considering condition of existing HMA) of thinner HMA surfaces to be incorporated in the FPS design. Where backcalculation of the HMA surface is possible, modulus adjustments are made using one of the following methods.
Two methods are suggested:
- Use the formulaCF = T2.81/200,000, where CF is the correction factor to be multiplied by the backcalculated HMA modulus (adjusted for outliers), and T is the average temperature over the time the FWD survey was made, or
- Use the Modulus Temperature Correction Program . MTCP can use the surface temperatures measured at each deflection location and, together with the previous day’s average temperature (available at weather underground ), predict the in-pavement temperature and compute the temperature adjusted modulus. Outliers must still be removed from the calculated temperature-corrected average.