Large Scale Triaxial Testing of Mechanically Stabilized Earth Retaining Wall Backfill

Abstract

The use of mechanically stabilized earth (MSE) retaining walls has become quite prevalent in highway embankment applications. A design criterion for these walls was originally established by the Federal Highway Administration (FHWA) and has been modified on a state by state basis. Recently, the Texas Department of Transportation (TxDOT) has recorded several wall failures mostly due to excessive settlement and lateral wall deformation and wanted to evaluate the current state design guidelines for regionally available backfill materials. Prior to numerical modeling simulations, material parameters of regionally available backfill needed to be evaluated.

As the state guidelines require 85-percent of the wall backfill material to be above the No. 4 sieve, large scale triaxial testing was an option to evaluate strength and volume change parameters. This research used cylindrical specimen 6-inches in diameter and 12- inches in height in a large scale triaxial apparatus. Three types of backfill material were tested and specimens were mixed and compacted in 4 different gradations for each material type. Each gradation was tested at confining stresses corresponding to wall heights of 10, 15, and 20 feet for a total of 36 tests.

Basic material parameters such as unit weight and friction angle were evaluated directly from testing, while more complex material parameters were selected from the data for use in the Duncan-Chang elastic constitutive model. This method utilizes hyperbolic curve fitting of both strength and volumetric test data to define soil behavior parameters which include the following: modulus number (K), modulus exponent (n), initial tangent modulus (Ei), failure ratio (Rf ), initial Poisson’s Ratio (νi), and Poisson’s Ratio Parameters G, F, and d.

Friction angles from triaxial testing ranged from 32 to 53 degrees having some uncertainty due to inconsistent compaction. The variation in sand and fine size particles in the backfill tended to reduce friction angles, except in the case of Type-B material where density increased due to the high percentage of sand and fines. Duncan-Chang parameters fit reasonably well with experimental data for strength barring some experimental errors. Volumetric parameters were inconclusive due to inconsistent compaction and membrane leakage. Additional testing is needed to provide more sound volumetric data.