Soil-Structure Interaction Modelling of High-Rise Building Settlements due to Compressible Soil Layers below Foundation Level

Abstract

In the field of structural and geotechnical engineering, a uniform approach to predict and model foundation settlements during the design phase of a high-rise building appears to be missing. In the Netherlands, pile foundations of tower structures underlain by compressible soil layers are challenging to model due to different stiffness and load distribution effects. As a result, the Dutch building code currently used for foundation design, the NEN9997-1, does not include realistic soil-structure interaction (SSI) effects. Instead, the NEN defines a simplified approach for high-rise buildings as the sum of two types of foundation settlements: individual pile head settlements (s1) and pile group settlements (s2) due to compressible layers below pile tip level.

Numerical models were used in this thesis to predict the individual contribution of different soil layers to measured subsidence of tower structures. By running several simulations using Tomlinson’s load spread method and the new embedded beam formulation (EB-I) in Plaxis 3D, it was found that approximately 65% of the total (s2) settlements are caused by the compression of clay layers below foundation level. Moreover, the effects of different pile factors (αs, αp) on the load distribution (more pile shaft resistance versus base resistance) from superstructure to subsurface were investigated. This research concluded that updated pile factors - in accordance with recent pile load tests on the Maasvlakte (Gavin, 2020) - influenced the predicted and modelled pile head settlements (s1) slightly for a Fundex 560 pile. Nonetheless, the change in load distribution due to different pile factors did not affect the vertical effective stresses or resulting (s2) settlements at depth.

Further, to accomplish a more uniform modelling approach for high-rise building settlements, this thesis provides insights for an automated soil-structure interaction mattress methodology as illustrated in Figure 1. A model verification is proposed for the mattress model approach using finite element software commonly used by geotechnical (Plaxis 3D) and structural engineers (SCIA Engineer) in daily practice. In essence, it is based on a simplified (s2) settlement analysis from Plaxis 3D (step 1) and mattress fit model in SCIA Engineer (step 2) consisting of multiple springs with linear stiffness (k_bedding) connected by a plate (E_plate) and a simplified surface load on top. The surface load represents the quasi-permanent building loads. An apparent limitation of the Plaxis 3D model (step 1) was the missing building stiffness or load redistribution within the superstructure due to differential settlements over time. However, a modelling discrepancy of only 1% was found for both the peak and differential settlements between SCIA Engineer (step 3) and Plaxis 3D (step 4) for a theoretical, symmetric high-rise building of 69 m in the North of Amsterdam. Thus, a model verification was accomplished by comparing the settlements from Plaxis 3D with the building on top of EB-I embedded beams (step 4) to the deformations of the fitted mattress model (k_bedding + E_plate) representing the compressible soils underneath the structure in SCIA Engineer (step 3). Altogether, this thesis provides a solid foundation towards a more universal design methodology between multiple stakeholders while including SSI effects for settlement predictions of high-rise buildings in daily practice.