Compaction models for predicting moisture-density-energy relationships for earth materials

This thesis describes an evaluation of the use of curve-fitting methods to develop field compaction specifications in earthwork projects with respect to optimum moisture content and maximum density. Five (5) curve-fitting methods were evaluated for forty-two (42) soils over a total of 763 lab compaction tests. The investigation indicated that of the five methods, two were found to be more reliable than others. All models were found to be unable to provide an adequate fit of the data for granular soils with a bulking moisture condition. A statistical evaluation of the mean square error for two of the models indicated that the Caterpillar model provides similar prediction reliability while using fewer parameters for a larger sample size. This investigation is meant to provide earthwork professionals insight into limitations associated with the use of curve-fitting applied to field compaction specifications.