Numerical Modeling Of Ground Movements Associated With Trenchless Box Jacking Technique

Abstract

The Trenchless Technology (TT) is defined as a family of methods used to install, renew, replace or renovate new pipe/box or existing pipe/box underground with little or no surface disruption. TT is divided into two main categories including Trenchless Construction Methods (TCMs) and Trenchless Renewal Methods (TRMs). TCMs are used to install new utilities and pipes underground while TRMs are used to renew, renovate and replace an existing utility or pipe. Box jacking (BJ) is a TCM used to install rectangular box culverts under existing facilities such as highways and railroads. In this method, box culverts are pushed through the ground using the thrust power of a hydraulic jack. Due to excavation methods and space requirements, the box culverts need to be large enough to provide adequate space for excavation. The sizes of box culverts usually range from 1.2 m x 1.2 m (4 ft x 4 ft) to 24.4 m x 12.2 m (80 ft to 40 ft).Installing box culverts underground, like other trenchless methods, may cause surface settlement and consequently damage existing road pavement or railroad bed. As a result, the need to better understand ground movements induced by the BJ process is important to minimize damage to adjacent infrastructures and facilities. Settlement in BJ projects are divided into two main categories including advance settlement, and trailing settlement. This research is focused on trailing settlement. The main objective of this research is to develop a surface vertical displacement prediction model using Artificial Neural Network (ANN). In this research, the ANN model was developed and trained using eight parameters including 1) modulus of elasticity (E), 2) friction angle, 3) unit weight, 4) cohesion (c), 5) box culvert height (h), 6) box culvert width (w), 7) overcut size (s), and 8) depth of box culvert from surface (H1). Exactly 300 finite element models were generated using PLAXIS 2D and used to train the ANN model. Twenty-two new finite element models were generated to verify the final ANN model. Moreover, the final ANN model was verified by collected data from two case studies and by new finite element models. The secondary objectives of this research are to evaluate the effects of different parameters on surface vertical displacement, to study arching effects and to evaluate applicability of available methods to estimate vertical stress at the top of box culverts.Results obtained from the final ANN model was in a good agreement with collected data from case studies and new finite element models. It was observed that the empirical method, suggested by Milligan and Marshal (1995) and originally developed for tunneling and pipe jacking (PJ), overestimated the maximum surface vertical displacement and underestimated the width of settlement trough (channel). Results showed that soil cohesion and box depth from ground surface had the highest impact on determining maximum surface vertical displacement, and soil friction and dilation effect have negligible impact on surface vertical displacements. Analysis of stress redistribution due to soil collapsing into the overcut area showed that stress was reduced due to arching effect above the box culvert. However, both Terzaghi and Marston's theories underestimated vertical stress.