· · · Institutional Repository

The TBM, not a blind mole! This thesis deals with some aspects of seismic imaging of the soft soil in front of a Tunnel Boring Machine to help tunnel constructors ``see'' the subsurface they are approaching, instead of steering the TBM forward like a ``blind mole''. The Dutch shallow subsurface has a very irregular layering and contains many heterogeneities. Geotechnical investigation of a tunnelling site, like soundings and borings, gives only local information of the soft subsurface and interpolation of the data will be inaccurate. This affects the safety during tunnel construction. The more information about the soil and obstacles in front of the TBM is known, the better tunnelling parameters can be controlled and damage to the TBM can be avoided. A seismic survey of the area will provide information on lateral variations in the subsurface. In combination with the geotechnical data, a much more accurate image of the tunnel trajectory can be made. Surface seismics will not always be easy, for example in urbanized areas. A solution would be to have a seismic system on the TBM. Seismic energy emitted from the machine reflects on contrasts ahead of the bore front and is recorded again by receivers on the TBM. The reflected energy gives the tunnel constructors an image of the subsurface in front of the TBM and therefor increases safety of the tunnelling. The source is a very important element of every seismic system. Its properties have to be optimal so that the emitted signals are suitable for the tunnelling application. In the shallow soft soils of the Netherlands, the use of shear waves is preferred over compressional waves as the former are less influenced by the water in the pores. Since the loose soils cause a high damping, the source signals should be low frequent. Some practical aspects to be investigated are the position of the source (and receivers) on the TBM and the contact with the soil for perfect transmission of the seismic energy. The design of such a seismic source falls outside the scope of this research. Instead of developping an active source, this thesis firstly examines if seismic energy is emitted by the TBM itself and, if this is the case, whether it is suitable for imaging in soft soils. The construction of the Second Heinenoord Tunnel near Rotterdam is set up as an experimental project in which many aspects of soft soil tunnelling in the Netherlands are investigated. On this site, seismic experiments are performed, on the earth surface, in the subsurface, in the TBM and in the tunnel. The first geophone lines on the earth surface show recordings of seismic energy coming from the TBM. Processing of the data indicates that the TBM emits shear waves in a frequency range between 5 and 100~Hz. As said before, these are good properties for a seismic source that is to be used in soft soils. Depending on the position of geophone lines, events with an average wavespeed varying from 100 to 200 m/s are registered. These are typical velocities for shear waves in soft soils, higher when the sand content increases, lower when more clay is present. Some recordings show distinct events with a regular amplitude and periodicity. At other locations, these can be very irregular, which makes it more difficult to analyse the data. In combination with recordings in the TBM itself, the actual source of the seismic signals is determined. Engines and pumps can be excluded quickly. Clear pulses are registered by geophones that are attached to the hydraulic jacks at the end of the TBM. These jacks push the machine forward against the tunnel elements that are already in place. The propagation of the TBM in the soil is not always a smooth movement, but it is influenced by friction effects between the machine and the surrounding soil. Most probably, the emitted seismic signals are related to the operation of the hydraulic jacks and therefor depending on the soil type around the TBM. This explains the variations in amplitude and periodicity. Experiments in the subsurface and inside the tunnel are carried out to test the suitability of the TBM as a seismic source. Performance of shear wave seismics in soft soils is very good, but for tunnelling applications new aspects have to be taken into account. Whether the TBM is used as a seismic source or an active source is developped, the optimal position for receivers when imaging the subsurface in front of the TBM is the cutter wheel. The slurry shield TBM used for the construction of the Second Heinenoord Tunnel pumps a viscous bentonite to the bore front to ensure its stability. For seismic recording on the cutter wheel, the penetrability of the bentonite by shear waves in crucial. To exclude any uncontrollable aspects of the hydraulic jacks, external compressional and shear wave sources are used in this experiment. As receivers, geophones at depth, in the path of the TBM, and pressure sensors on the cutter wheel are applied. The results are rather ambiguous. The recordings on the geophones show that the waves are highly attenuated. But not even the compressional waves are recorded on the cutter wheel. It is possible that the pressure sensors are not suitable for these kind of measurements. If, on the other hand, the seismic energy could not penetrate the bentonite, special hardware adjustments are necessary. Also the efficiency of the jacks as seismic source is tested. During construction of the second tunnel tube of the Second Heinenoord Tunnel, the first tube is used as a strong reflector. Geophones are attached to the tunnel wall of the second tube. The recordings have a very high signal-to-noise ratio. After removal of the direct waves, travelling through the tunnel wall, reflected events generated by the other tunnel tube can be traced in the data, allthough they are very weak. Reflections coming from ahead of the TBM might therefor be even harder to distinguish. Applying the TBM as a efficient, reliable and controllable seismic source for imaging requires more research on the operation of the hydraulic jacks and the influence of the soil surrounding the TBM. The theoretical part of this thesis focusses on processing of data from a seismic system for tunnelling, where receivers are installed on the cutter wheel of the TBM. Seismic surveys for oil- and gasexploration are performed over a very large area, using hundreds of receivers. Traditional seismic processing techniques are optimised for these configurations. In tunnelling applications, the receivers are limited in aperture by the diameter of the TBM and in number by the available space on the cutter wheel. Applying the existing processing techniques, like phaseshift migration, on this new configuration will introduce large artefacts into the data because of the truncation of the extrapolation operator. Short, symmetric, spatial extrapolation operators that are stable in a specified wavenumber range are designed, using a weighted-least-squares optimisation. Migration in the spatial domain instead of the frequency domain makes it possible to take small lateral soil variations (perpendicular to the cutter wheel in case of tunnelling applications) into account. For a large number of receivers, the short operators will limit computation time extensively. In tunnelling, the lenght of the operator will be based on the number of receivers on the cutter wheel. However, the symmetric operators are only optimal for the middle receiver. For the other data points, the operator is shifted and gets asymmetrically truncated, which again generates artefacts. Also, only part of the data points are used in the extrapolation, which is rather inefficient if so few data is available. Therefor the weighted-least-squares optimisation is adjusted to design short, asymmetric, spatial extrapolation operators. For each data point, a new operator is created, based on the number of receivers and the relative position of the data point, taking all other data points into account. By varying the weight and the wavenumber range in which the operator should be accurate, the optimal operator can be calculated. For the outmost points, special attention is necessary to ensure stable operators. The performance of the symmetric and asymmetric operators in both forward and inverse extrapolation is compared, for different operator lenghts. In the forward experiment, a source signal is forward extrapolated, once with symmetric and once with asymmetric operators. On both data sets, inverse extrapolation is performed with the same, symmetric method so that differences in the final result are only due to the forward extrapolation. Comparison with the original input signal gives an indication of the performance of the operators. A similar experiment is performed to investigate the inverse extrapolation. In this case, a modelled data set is inverse extrapolated once with symmetric and once with asymmetric operators. Then both data sets are forward extrapolated with the same operator set and compared with the input data. Both experiments show that the accuracy of the operators decreases with the operator lenght. In general, the symmetric and asymmetric operators have a similar level of performance. The symmetric operators have less negative effects on the inverse extrapolation. In the forward extrapolation, the asymmetric operators introduce less artefacts. Finally, the extrapolation operators are applied on simple modelled 2D data sets with homogeneous and layered backgrounds. Migration with symmetric operators for data sets with a small number of receivers introduces more artefacts. These are mainly generated at the outmost data points and are extrapolated further into the data. Especially in layered media, which is typical for the shallow subsurface, the asymmetric operators perform much better. The examples clearly show the advantage of the short asymmetric spatial operators for migration of seismic data from tunnelling applications. They contribute in creating an accurate image of the subsurface in front of the TBM and increase safety of the tunnelling activities.

The objective of this study is to gain insight into mechanisms of soil-structure interaction for buildings adjacent to deep excavations and to find a reliable method to design and monitor deep excavations in urban areas with soft soil conditions. The research focuses on typical Dutch conditions. The main questions are: How can we predict the behaviour of one or more a deep excavation will be constructed? What kind of modelling and/or measurements can be used to predict this effect? This report describes the literature reviewed for this topic and several case studies related to the topic from literature. General damage assessment procedures are also given. Assessing the response of buildings to excavation-induced deformations involves a combination of geotechnical and structural aspects, such as green field displacements (2D/3D, caused by deep excavations), building behaviour, soil -foundation -building interaction, monitoring techniques and modelling techniques. Each of these topics is described in this study. Some of the conclusions from this literature survey are: - Several, mostly empirical, relationships are available to predict green field displacements, which do not always show improvement in the amount of settlement found behind the wall over the years, especially if soft clays are present. One should expect for a deep excavation in soft clay to find a wall deflection of about 0.5 -1.0% of the retaining height (for an average system stiffness and sufficient basal stability) and a settlement behind the wall of 1%H maximum. Margins of 50%-100% should be expected. Diaphragm walls with stiff supports tend to the lower bound of these numbers or can even perform at 0.2%H if installation and other effects are strictly controlled. - Damage to buildings can be assessed by several damage criteria. The use of relative rotation and deflection ratio are both widespread, but also widely discussed. It is important to be extremely clear on how rigid body rotation and overall translation have been incorporated in the calculation. - Rigid body rotation or building tilt, is a very important parameter when discussing excavation induced damage. Real rigid body rotation should be assessed in three dimensions and it should always be made clear exactly if and in what way tilt is considered. - Soil-foundation-structure interaction should be taken into account when damage is assessed. The amount of displacement transferred to the building depends on the stiffness of the building in axial and bending modes and the interface between soil and foundation and between foundation and building. Three different case studies are presented for the insight they provide in the soil-structure interaction caused by deep excavations, tunnelling or subsidence. Both ground deformations and building deformations have been collected. These cases show aspects of the relationship between deformation of the building and damage occurring.