Building response to ground movements induced by tunnelling and excavation

In urbanized regions, two important infrastructure activities; subway projects and deep foundation excavations can cause significant negative effects on neighboring surface structures. The increase in demand for transportation, residential and office buildings in high-density areas increases the number of such infrastructure investments. Since lateral urban growth is limited in these areas, tall structures with deep foundation pits have been built to use vertical spaces efficiently. As a result, vertical and horizontal ground movements due to tunneling or deep excavation activities pose a serious risk of damage to existing surface buildings and infrastructure services. Some characteristic building damages that can arise from tunnelling- and excavation- induced ground movements are classified in the literature as cracking, jammed doors, failure of beams, defects of some sensitive devices, and service pipes. In most cases, the first cracks appear around the corners of the window and door openings; since these are the most vulnerable parts of the wall. Later, as the differential effects of the settlement increase, they can multiply and expand. Severe vertical cracks in the upper part of the masonry walls and subsequent separation along these cracks have also been observed in some cases. Damage caused by tunnelling and excavation can cause severe loss of strength and stiffness. In order to determine an effective protection method to prevent or mitigate building damage, anticipated magnitude and distribution of tunneling and excavation-induced ground movements must be accurately estimated, and then building response should be predicted. However, this bi-directional problem that involves excavation, soil and building is often not easily solvable due to the uncertainties in material properties, excavation details, construction steps and soil-structure interactions and the difficulties in modelling all these components and the interactions between the ground and the building. In practice, the risk of building damage from tunneling- and excavation-induced ground movements is assessed at three levels: (i) preliminary, (ii) second-stage, and (iii) detailed assessment. At the preliminary level, the risk is assessed against the limit values of the maximum settlement and rotation of the building, which have been specified based on the previous empirical observations. In the second-stage assessment, an elastic beam analogy is used for buildings and an analytical solution is performed. As the risk level is determined by comparing with the pre-defined empirical strain ranges, the second-stage risk assessment is also based on a quasi-experimental technique. This method, which is formed using the elastic beam approach and uses the pre-defined limit values of the tensile stresses, is also called the limiting tensile strain method (LTSM). At the preliminary and second-stage assessments, building is assumed to follow the deformed ground profile without any relative movement in-between. During the detailed assessment level, a comprehensive examination is carried out using modern numerical tools such as finite element method (FEM). During, such an analysis, the non-linear properties of the materials (for soil and building), the soil-structure interaction, and the construction stages for the excavation or tunnel can be taken into account. Of course, this latter method offers the most realistic results, but it is apparently more complex, time consuming and costly. Modelling also requires extensive data. According to the building risk assessment procedure; generally, buildings with moderate or high risk at the preliminary assessment level are re-evaluated at the second-stage assessment level. Similarly, buildings with moderate or high risk at the second-stage assessment level are re-evaluated at the detailed assessment stage. The final risk level is achieved by following these steps. In a typical influence zone, for only few of the buildings and for important historic buildings, detailed assessments are found necessary. The previous experiences showed that for the majority of the buildings the final risk decision is usually made at the second-stage assessment level using the LTSM. Therefore, if the accuracy and precision of the LTSM used in the second assessment stage is increased, not only the accuracy of the final risk decisions determined by this method will increase but also the number of buildings to be evaluated using more costly and more time consuming methods during the detailed risk assessment will be reduced. Due to the simplicity of the method, LTSM offers rapid evaluation. However, this situation causes some important factors such as soil-structure interaction to be ignored. Or it has been shown, for example, that the damage class of the masonry buildings with large openings in the load-bearing walls are underestimated due to the inability of this method to consider stress concentrations. Although it is often used in practice, the number of studies in which the LTSM has been analyzed using the real site data is still limited. In addition, the number of projects where the project information is recorded in detail in a way that allows an in-depth analysis of this method is very few. After a brief introduction (Chapter 1), this study first focuses on the existing methods which dealt with the prediction of the building damage caused by tunneling or deep excavations. A very extensive and comparative literature review is presented for them to convey the current state-of-the-art. Reported advantages and disadvantages of the LTSM are clearly addressed (Chapter 2). Chapter 3 is devoted to the development of a more independent sight for the LTSM that will ease the interpretation of the relationships between deflection ratio, structure's geometry, longitudinal/shear stiffness ratio and the limiting tensile strain. For this purpose, a transfer function view is proposed. Another original contribution is the inclusion of the effect of average horizontal strain in the modified beam equations. It used to be considered separately in the original LTSM. Using reported data and observed damage classes of real and simulated case studies available in the literature, back-calculations for the coefficients of the transfer function are made. The results showed that the observed damage and measured crack widths are more or less compatible with the limiting tensile strain boundaries. However, for the cases in which moderate or higher damage was observed due to excavations, the use of the LTSM results in underestimation of the resultant damage class. Also, it was shown that the use of the LTSM can cause underestimated damage predictions for the masonry buildings with large openings. In Chapter 4, a detailed examination of a case, in which a masonry structure has been affected by the ground movements caused by a deep shaft and tunnel excavations, is presented. The structure has a historical importance and possesses several pre-existing cracks in the load-bearing walls. It is shown that the LTSM underestimates the damage class due to presence of large openings in the load-bearing walls. This case is also a good example for which several information gaps and uncertainties are demonstrated along with the considered assumptions and simplifications. The objective is to demonstrate the added value of a noval two-stage numerical analysis approach for the assessment of the settlement response of the building. The first stage comprises analyses of the structural behaviour using the monitored settlements for each wall. The second stage examines the behaviour of the complete building as a whole. The effects of soil-structure interaction and the pre-existing cracks are considered through discrete interface elements. The parametric analyses executed in the second stage also lead to generalizable results of potential use beyond the specific case. The analysis results showed that using a two-stage analysis approach, the actual response of the building can be satisfactorily represented. Including existing cracks into modelling through the discrete interfaces reduces tensile strains in the close vicinity but results in an increase in the tensile strains of other neighbouring sections. Excluding both bedding and existing crack interfaces during modelling results in more severe cases in terms of tensile strain levels and a higher concentration of tensile strains. As generalizable results of this numerical study, previous findings in the literature was also confirmed: Increased structural stiffness through the increase of the Young's modulus of the masonry walls reduces the conformity of the structure to the applied prescribed settlements and leads to lower tensile strains. Increase of the soil stiffness is simulated with the increase of the bedding stiffness. The higher the bedding stiffness is the more the structure conforms to the applied prescribed settlements. This eventually results in increased tensile strains. As an extreme condition, excluding bedding interface corresponds to a case with infinite bedding stiffness. In Chapter 5, a holistic conclusion of the findings and suggestions for the future studies are presented.