U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-HRT-08-035
Date: March 2008
As discussed in previous parts of this report, TDR trace data as well as computed parameters developed under this project were thoroughly reviewed. Observations discovered during the project are described in this chapter.
Limited comparisons were made between data computed using the TLE micromechanics method and the moisture estimates computed previously using the apparent length method. A series of comparisons (discussed in chapter 3 of this report) were performed to validate the new procedure.
Additional comparisons were made during the development phase to ensure that the MicroMoist program was working properly. In many cases, the same trends show up for both computational processes and the resultant estimates are very similar. However, in some cases, significant differences are present, typically in higher moisture content settings. In these situations, the TLE/micromechanics method results in a moisture content that is closer to the in situ moisture content (acquired during equipment installation) as compared with the apparent length method. This is expected because the micromechanics model was calibrated to the in situ moisture content at each site/layer.
As an example, results for LTPP section 063042 from the apparent length approach and the TLE/micromechanics method can be found in figures 29 and 30, respectively. Also included in the figures are the in situ VMC obtained during equipment installation. As can be seen in figure 29, VMC from the apparent length approach ranged from 35 to 55 percent (with the majority of values falling between 35 and 39 percent). For some of the TDR sensors, the predictions are drastically higher than the measured in situ moisture content. The results from the TLE/micromechanics method shown in figure 30 correlate more closely with the measured in situ moisture content trends.
Figure 29. Graph. Results from the apparent length approach for LTPP section 063042.
Figure 30. Graph. Results from the TLE micromechanics method for LTPP section 063042.
Figures 31 and 32 provide similar information for LTPP site 313018. The data for both computational processes range from 5 to 22 percent with similar distributions.
Figure 31. Graph. Results from the apparent length approach for LTPP section 313018.
Figure 32. Graph. Results from the TLE micromechanics method for LTPP section 313018.
Differences between the two approaches are generally site specific and largely dependent on the measured in situ data. The TLE/micromechanics method is calibrated to the measured data and, therefore, yields estimates that are closer to the measured data as compared with the apparent length approach. The apparent length approach is a general empirical regression model that can vary significantly from ground truth data under certain circumstances.
Limited frost effects show up in the new data. In general, there is an expectation that the VMC values will decrease as the moisture in the ground freezes in both the wet and dry freeze regions. This can be seen in the values computed for site 274040. Where there is a distinct reduction in the volumetric moisture values during the winter months of 1993-1994, 1994-1995 and 1996-1997. There were no measurements recorded during the winter of 1996 at that location. It has been documented that frozen ground results in a TDR trace that does not have a negative slope between inflection points (i.e., an "open trace"). () These traces cannot be interpreted using the TLE/micromechanics method. Therefore, TDR traces flagged as uninterpretable during the winter months are indicative of frost effects.
The ability to calibrate to measured in situ data is one of the key advantages of the TLE/micromechanics method. Computed parameter estimates are directly affected by measurement errors in the ground truth data.
The in situ information came from limited soil tests performed during equipment installation. At that time, the apparent length approach was the accepted method for interpreting TDR traces to determine moisture estimates and was independent of the ground truth data. The soil was tested for in situ moisture content for general background information regarding the installation, not for calibration of moisture estimate algorithms. The procedure established for collection of the in situ moisture consisted of heating samples in an open pan over a propane stove in the field. The in situ dry density values were determined for each soil type based on a one point proctor test. Both of these tests have errors associated with them that directly affect soil parameters developed in this study.
A very likely source of error associated with moisture content testing in the field is soil loss due to either exploding aggregate (caused by rapid heating) or loss of fines (blown away during heating and stirring). Loss of soil during the drying process will produce moisture content results that can be up to five percent higher than the actual moisture content.
The dry density values used to calibrate the MicroMoist program often came from the one point proctor test performed on the soil samples taken during installation (when available). While the results from the one point proctor test provide reasonably accurate estimates of in situ densities, the findings are derived from disturbed samples, which can be another source of error. For some sites, dry density values reported were extremely high and not physically possible as they would result in a negative percentage of air voids. These cases were discussed in chapter 4 along with details on an adjustment procedure used to mitigate the problem. However, no correction was made for densities that were relatively low. These results were likely from deeper depths, where standard proctor densities would likely have provided more reasonable density values.
Future endeavors utilizing TDR equipment to monitor in situ conditions should focus on obtaining more accurate ground truth moisture and density data for use in the calibration of the TLE/micromechanics method. As such, soil from each layer should be sampled from a second hole and taken to the laboratory for testing. The moisture content could then be determined using a standard moisture test. The density could be estimated for each layer from a standard proctor test based on the moisture content established for each layer. For the deeper layers the density used may be 90 to 95 percent of the standard proctor density, assuming the in situ densities were below the influence area of the normal construction process. The ideal approach would be to dig test pits and perform nuclear density tests on each layer as it is excavated. However, the costs of this approach may be prohibitive and could require placement of the TDR probes in the shoulder next to the traveled lane.
|< Previous||Contents||Next >>|