Abstract

Continuous Compaction Control (CCC) systems have demonstrated great promise for improving the efficiency of field compaction and revolutionizing the compaction control process. To evaluate the effectiveness and reliability of CCC systems in the State of Delaware, a field study was performed on a local soil (a poorly graded sand with silt), with compaction being performed using an MDP-CMV equipped compactor. A variety of in-situ test methods that are currently used for compaction control were also performed as compaction progressed in the study, for purposes of comparison with the CCC results. Comprehensive analyses were performed on the data obtained from the field study using various statistical techniques.

As a first step, basic statistical analysis was performed on the recorded in-situ testing values. In general, it was concluded that there was significant scatter in the measured in-situ test results, which made it difficult to make a precise judgment on the quality of compaction. However, based on the dry unit weights measured by the nuclear density gauge (NDG), the quality of compaction was determined to be in an acceptable range, according to current DelDOT specifications. In addition, the measured water contents indicated that in general the compaction was performed on the dry side of the optimum moisture content.

Statistical analysis of the CCC roller data illustrated a promising trend for MDP and CMV values as the compaction progressed: MDP values decreased and CMV values increased as the number of passes increased. It was also realized that MDP values contained less variability than simultaneously recorded CMV values.

The ordinary kriging method was employed to determine the magnitude of CCC values at the same locations as the in-situ tests that were conducted. Comprehensive analysis of different models showed the appropriateness of the Rational Quadratic model for predicting the MDP values and the Exponential, Spherical, and Linear models for predicating the CMV values. Maximum lags of ∼ 1.5 m (3.0 ft) and ∼ 3.0 m (10 ft) were also selected as the optimum lag distances for kriging the roller data.

Univariate regression techniques were applied to the in-situ data and kriged CCC values to identify possible relationships between the data sets. It was discovered that point-by-point comparisons did not yield strong relationships between the data. However, taking the average values of each lift and pass into consideration in the regression analyses yielded much stronger correlations between the in-situ testing values and the kriged CCC data. MDP values showed stronger correlations with in-situ testing data than did CMV values. The GeoGauge did not yield strong correlations with CCC values. Strong correlations were also identified between the CCC and in-situ testing values versus the water content of the compacted soil. Therefore, it was decided to include the effect of water content in the analysis using multiple regression methods. The results showed a great improvement in the relationship between average kriged CCC values and corresponding in-situ testing data, which confirmed the effect of water content on the measured CCC and in-situ testing values.