Numerical Modeling of Liquefaction-Induced Downdrag: Validation against Centrifuge Model Tests

Abstract

Earthquake-induced soil liquefaction can cause soil settlement around piles, resulting in drag load and pile settlement after shaking stops. Estimating the axial load distribution and pile settlement is important for designing and evaluating the performance of axially loaded piles in liquefiable soils. Commonly used neutral plane solution methods model the liquefiable layer as an equivalent consolidating clay layer without considering the sequencing and pattern of excess pore pressure dissipation and soil settlement. Moreover, changes in the pile shaft and the tip resistance due to excess pore pressures are ignored. A TzQzLiq numerical model was developed using the existing TzLiq material and the new QzLiq material for modeling liquefaction-induced downdrag on piles. The model accounts for the change in the pile's shaft and tip capacity as free-field excess pore pressures develop or dissipate in soil. The developed numerical model was validated against data from a series of large centrifuge model tests, and the procedure for obtaining the necessary information and data from those is described. Additionally, a sensitivity study on TzLiq and QzLiq material properties was performed to study their effect on the developed drag load and pile settlement. Analysis results show that the proposed numerical model can reasonably predict the time histories of axial load distribution and settlement of axially loaded piles in liquefiable soils both during and postshaking.

Closure to “Centrifuge Model Tests of Liquefaction-Induced Downdrag on Piles in Uniform Liquefiable Deposits”

“…The writers would like to note that the instrumentation of the 0DPile measuring axial load was sensitive to bending and thus could cause errors resulting in some unusual and unexplainable axial load distribution. Laboratory tests (Sinha et al 2021b) and the subsequent SKS03 centrifuge model test (Sinha 2022;Ziotopoulou et al 2022) also confirmed this anomaly. Moreover, the inclination of 0DPile (vertically by about 1.4°) could have also resulted in bending moments affecting the axial load measurement.…”

“…The writers realize that, due to instrumentation limitations, an experiment may not provide all data of interest; thus, for a more detailed investigation, follow-on validated numerical simulations are used to extract more insights. Sinha et al (2022) developed a numerical approach (TzQzLiq analysis) for modeling the response of axially loaded piles in liquefiable soils and validated it against these centrifuge test results. Through the numerical work, Sinha et al (2022) described the mechanism of the development of drag load, neutral plane, and axial load distribution in 0DPile and 5DPile.…”

“…The modeling of the real roughness of piles [which is affected by installation type (drilled versus driven), corrosion after installation, and other factors] was not in the scope of our planned centrifuge tests. Our goal in roughening the surface was to minimize uncertainty in the estimation of interface friction angle (δ), maximize drag load to emphasize the mechanism of liquefaction-induced downdrag, simplify the interpretation of results, and, by extension of the above, enable our follow-on numerical investigations (Sinha et al 2022).…”

“…The writers agree that this would have been useful, but such instrumentation was not feasible for the scope of this project. In lieu of this, a subsequent paper by Sinha et al (2022) describes efforts to deduce soil settlement from pore pressure sensors, post-test excavations, and inverse numerical analyses. The pile settlement at any depth can be adequately determined from the pile head settlement minus the integration of axial strain measurements.…”

Centrifuge Model Tests of Liquefaction-Induced Downdrag on Piles in Uniform Liquefiable Deposits

“…The magnitude of pore pressure increase in the dense layer depends on the relative thicknesses of the layers and drainage boundary conditions. This issue is examined in more detail in Sinha (2022). When the loose sand layer was fully liquefied (between the depths of 5-12 m), the interface skin friction for the 5DPile in the liquefied layer was reduced to zero and, consequently, the drag load vanished [Fig.…”