LTPP Computed Parameter: Moisture Content

Chapter 9. Summary and Conclusions

TDR traces have been used to estimate the subsurface moisture content for unbound layers in pavement structures. In particular, the moisture content of the various roadway sublayers at SMP test sections were monitored with TDR instrumentation because it is relatively fast and accurate and provides a nondestructive in situ measurement. The TDR waveforms, however, do not provide moisture content estimates directly. In situ conditions of interest must be derived from the TDR waveforms and are largely dependent on the methodology used. However, it is clear that interpretation of electrically induced reflectometry depends not only on the material dielectric constant but also the reflectance and conductance attributes.

Moisture parameters had been estimated from TDR traces in the LTPP database previously, but significant quantities of TDR data have been collected since the completion of the original study. Therefore, one objective of this current study was to develop soil parameter estimates from TDR waveforms not previously analyzed. An additional objective was to investigate new methodologies or improvements to existing processes for interpreting TDR waveforms.

Based on the investigation conducted in phase 1 of the study, a new approach utilizing TLEs to compute dielectric constants from the TDR waveform and micromechanic models to estimate moisture and density parameters was proposed. This approach was approved by FHWA and was used to interpret 274,000 automated TDR traces in the LTPP database.

This new approach for calculation of the VMC consists of four steps:

Step 1: Calculate the dielectric constant, conductivity, and reflectivity from the TDR trace using the TLE.

• The TLE uses the shape of the trace to provide a more complete estimate of the dielectric constant.
• The solution method is the SID, which can minimize the error between the actual measurement and the calculated measurement.

Step 2: Given the moisture content and density data from the installation reports, along with the parameters calculated at step 1, backcalculate the permittivity of the solids and calibrate the micromechanics volumetric water model.

• This backcalculation is based on a theory of dielectric properties of composite materials from the micromechanics and self consistent scheme as follows:

(50)

• The dielectric constant of soil and water (e₁, e₂) and specific gravity of soil (G_s) are calibrated, based on the in situ information obtained during equipment installation, using the SID approach.

Step 3: Given the calibrated micromechanics volumetric water model, forward calculate the volumetric water content and the dry density of the soil for other times and seasons based on the TDR traces and the associated dielectric constant.

• The self-consistent model was used together with the calibration constants e₁, e₂, and G_s to calculate soil dry density (g_d) and VMC (q ).
• Systematic error was removed through consideration of the effect of individual constituent soil dielectrics.

Step 4: Compute the gravimetric moisture content using the VMC and dry density from step 3.

The TLE method used to determine dielectric constant is able to consider the soil conductivity and reflectivity influence on the dielectric value. Additionally, the micromechanics models are calibrated to site-specific conditions and equipment using ground truth measured data. These two processes work together to minimize systematic errors in the resulting moisture and density estimates.

An evaluation of the new approach was conducted by comparing moisture estimates to measured values using data from SMP Installation Reports, Klemunes' thesis, and LTPP forensic studies. The estimates were relatively accurate and were all within 10 percent of the measured values. The previous LTPP interpretation procedures did not have a mechanism for estimating dry density for the soils represented by the TDR trace, but the new method provides the capability of estimating dry density values from TDR measurements.

A key advantage to the new micromechanics-based procedure is that it incorporates the engineering properties associated with the TDR measurements, as well as the mechanical properties of the soils being measured. Beyond this, it makes use of the physical and electrical properties of the materials being measured.

In order to quickly and efficiently compute soil parameters for the large quantity of records in the LTPP database, a new program, LTPP MicroMoist, was developed. The program utilized many of the same graphical and visual features as the MOISTER program. The new program was designed to calculate all components of the soil parameters automatically. Logic and reasonableness checks were incorporated into the program to ensure anomalous data were manually reviewed and verified. Data not passing established checks were flagged as part of the process. The program was designed so that quality took precedence over the computation efficiency and ensured that the highest quality data were obtained in a practical manner.

External to the program, post-processing graphs were developed and used in beta testing and debugging of the program. These graphs were also used to perform a 100 percent review of the final set of computed parameters. Anomalous or outlier data were manually flagged in the dataset delivered to FHWA for inclusion into the database.

As a result of this study, approximately 274,000 automated TDR traces were analyzed. Some were not interpreted due to questionable TDR trace characteristics or questionable results. The vast majority of those not interpreted were from the five sites in New Jersey where ground truth moisture content data were unavailable.

The analysis team worked with FHWA contractors to deliver the data in the most efficient manner and provided table and field descriptions for their use. It is anticipated that the data will be available in the 2008 LTPP Standard Data Release.