A Field and Numerical Evaluation of Blast Densification at a Test Section

For more than 70 years, blast densification has been employed as a ground improvement technique to densify loose granular soils. Explosives are detonated at depth to induce liquefaction in the target sand and subsequent pore pressure dissipation causes the soil to consolidate into a denser configuration. Typically, improvement is verified via measured surface settlements and increased penetration resistance with in-situ tests, such as the standard penetration (SPT) or cone penetration test (CPT). However, often times no increase or a decrease in penetration resistance is observed, despite significant amounts of densification in the target sand layer. Researchers have suggested several theories for this paradox, including arching, decreased stress levels, destruction of cementation bonds, and the presence of gas released by the explosives. In some instances, a time-delay on the order of months or years was needed before an increase in penetration resistance was measured with in-situ tests. Evidence used to support these theories has been limited, and many questions still remain.;This study focused on a single blast zone where no increase in penetration resistance was measured with the CPTu following densification of a 3.8 m thick layer of loose liquefiable sand. BAT probes (piezometers) at 10 locations were used to verify liquefaction and evaluate the spatial generation of excess pore pressure inside and outside the blast zone. Measured surface settlements and CPTu soundings taken before and after blasting were used to verify densification. Pore fluid samples were collected with BAT probes and tested using gas chromatography techniques to measure the concentrations of N2 and CO2 in the soil. This was done to address theories concerning the presence of gas released by the explosives, and any impact it might have on the penetration resistance measured after blasting.;A numerical simulation of blast densification was performed with the finite element code PLAXIS 2D to address theories concerning possible changes to the stress-regime as a result of blasting. Measured pore pressures from the field were applied to the pore pressure degree of freedom in the target layer to simulate the "generation" of pore pressures. A coupled flow-deformation formulation in PLAXIS was used to simulate pore pressure dissipation thereafter. The hypoplastic sand model (SH-model) and SH-model with the intergranular small-strain extension were used to model the constitutive response of liquefied and non-liquefied sand, respectively.;Results from the field and numerical analysis were combined to draw conclusions as to why no increase in penetration resistance was observed after blasting. It was found that the presence of gas released by the explosives was mainly responsible, especially at the center of the zone. However, the redistribution of stress is partially responsible for decreases in penetration resistance observed near the zone perimeter, in addition to gas.