Performance Comparison of Mechanical and Chemical Stabilization of Undercut Subgrades

Abstract

This study evaluated the performance of undercut subgrade stabilization measures during construction traffic loading prior to final paving. Twenty-two simulated undercut sections with different stabilization configurations over a Coastal Plain subgrade typically undercut in North Carolina were built in a large-scale test pit. The subgrade was placed at a California Bearing Ratio (CBR) of approximately 2.0% and stabilized with granular layers, granular layers reinforced with geosynthetics, and lime. Granular layers consisted of either aggregate base course (ABC), sandy select fill, or a multi-layer system with both soil types. The four geosynthetics tested were a woven reinforcement geotextile, a woven separation geotextile, and two biaxial polypropylene geogrids. A circular steel plate loaded the sections statically, as well as with simulated proof-roll inspection and construction traffic equipment pulses. After initial cycles, deformations were refilled to simulate rut repair before final paving. The sections were then re-loaded to simulate paving traffic. Resulting surface displacement and subgrade stress increase were recorded using electronic instrumentation. Cyclic loading showed that, in general, tests with thicker granular layers had less surface displacement. Some differences occurred likely as a result of reusing the same Coastal Plain subgrade throughout testing. Remolding caused a reduction in the measured undrained shear strength, which was reflected in an increasing dry unit weight and DCP Index at a given water content as testing progressed. Typical material prices were used to calculate the unit cost of each stabilization configuration. A performance-cost analysis was performed by coupling these values with the cyclic load displacements to determine the most economical stabilization alternatives. The results showed that tests with lime stabilized subgrade (LSS) test were the most economical over initial and post-rut repair cycles. The LSS had low construction cost and when the LSS had higher tested strength, loading caused less surface displacement.

Unreinforced ABC (between 14 and 20 inches) was economical during initial cycles. Geosynthetic-reinforced tests showed that when the ABC layer was approximately as thick as the load plate diameter (12 inches), high displacements caused significant geosynthetic mobilization during initial cycles that proved economical during long-term loading applications. This was particularly true of tests with the reinforcement geotextile, as it was the geosynthetic with the highest tensile strength and also provided layer separation. At these depths, the less stiff separation geotextile and both geogrids were not strong enough to be economically viable. When the ABC layer was thicker (between 18 and 20 inches), these geosynthetics proved moderately economical, but differences in the reinforcement type at these depths were less significant. Tests with thirty-six inch select fill and three inch ABC stabilization had a high unit cost, but were moderately economical during all cycles. Inclusion of separation geotextile in deep select fill tests did not significantly reduce surface deformation because the fabric was far enough from the load plate to have little influence. Tests with less (fourteen to seventeen inches) of select fill and three inches ABC reinforced with separation geotextile experienced more displacement because the load plate punched through the thin ABC layer at low cycles. The displacements likely caused lateral spreading of the select fill without mobilization of the fabric. Thus, the tests were not as economical as the other tested configurations.