Settlement Damage to Masonry Structures

This project consists of the study of damage to existing structures caused by tunnelling or other earthworks in soft ground conditions. The project is prompted by the extensive current interest in new underground developments, particularly in London. The Jubilee Line extension, and other projects, have involved tunnelling under many important masonry buildings. The accurate identification of buildings at risk, and the assessment of possible damage is essential if appropriate procedures for settlement control are to be designed.

How do settlements arise from tunnelling?





Settlements arise from the installation of a tunnel for three reasons. Firstly, there is a difference between the volume necessary to house the tunnel and lining, and the volume of the void created during excavation. This produces tail loss, to the rear of the shield. Excavation also reduces the horizontal stresses at the tunnel face, leading to soil deformations and face losses. Finally, longer term movements occur due to changes in pore pressures initiated by the installation of the tunnel. These soil movements around the tunnel lead to (smaller) movements at the surface.

This research uses numerical modelling to study the problem of tunnelling-induced settlement damage. It is thought that only by the use of a 3D model, including tunnel installation and surface structures, can a reasonable simulation of the real situation be made.A three-dimensional finite element model in which the tunnel, the soil and the building are all included in a single analysis was developed during an initial phase of this research. This work was carried out between 1994-6 with support from EPSRC . The components of the model are shown in the figure below.


The model is currently restricted to undrained soil behaviour. A typical FE mesh and solid model created to demonstrate the technique are shown here.

Research is continuing in two areas, improvement and extension of the model, and verification and use of the model.

Improvement and extension of the finite element model

The current numerical model is based on direct methods to solve the finite element equations. It is thought, however, that iterative solvers, may lead to algorithms which are more efficient. The possibility of using preconditioned conjugate gradient methods in this application is currently being investigated.

The arrival of the Oxford Supercomputer, Oscar, followed by its replacements Oswald and Oswell (resources managed by the Oxford Supercomputing Centre) has allowed larger analyses to be undertaken. The non-linear finite element code, OXFEM developed in-house, has been ported to run in parallel using OpenMP directives. It is intended that further parallelisation will use MPI or BSP paradigms.

Further planned developments to the numerical model include

  • numerical methods to model compensation grouting
  • improved model simulation of tunnel installation
  • procedures to model the long term effects on surface buildings due to the construction of nearby tunnels.

Any enquiries about this research should be directed towards Prof. Guy Houlsby, or Dr H.J. Burd