Preliminary prediction model of a tunnel in gravel formation

• Main Authors: Ming-Hsin Chen, 陳明新
• Other Authors: Te-Chin Ke
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/27007496661768277157

碩士 === 國立雲林科技大學 === 營建工程系碩士班 === 94 === Gravel formations are mainly distributed over several tablelands in the northwestern and western regions of Taiwan. Because of their varying texture and size effect, the mechanical behavior of such formations has not well investigated yet. Even though a large number of tunnels in Taiwan had been constructed in the past century, very few were located within the gravel formation. The thesis is intended to establish a preliminary prediction model of closure of a tunnel in such a formation primarily based on the data of site conditions and field measurements from two Chuawha tunnels of Taiwan High Speed Railroad (THSR) through the Paguashan terrace, using the neural network analysis (NNA). In this research, five evaluation factors (tunnel overburden depth Z, surface inclination(θ),ground classification index(G), ground strength index (T), tunnel support class (S) were first selected from various ones to predict the tunnel closure. More than 200 sets of data were collected and compiling from three Chuawha tunnels of THSR, followed by attaining the nonlinear prediction models of roof deflection (δF), left-wall deflection (δL), and right-wall deflection (δR) by performing the training task of NNA, under the constraint of 10 % of deviation percentage, as well as the later NNA-verification process. The linear model of NNA was also used to determine the importance order of five evaluation factors, The NNA-verification results depicted that the ANN-generated nonlinear prediction models for (δF,δL,δR) yielded the average of deviation percentage varying from 34% to 39% which is fairly small, and they can serve as a quick tool in the near future for predicting tunnel closure in the studied area. The linear model of ANN pointed out the most important factor to be the overburden depth ( Z ), implying the less variance of other factors in the studied area. The negative weighting of ( Z ) also indicates that ( G , T ) increasing with ( Z ) might override the effect of overburden pressure, i.e., the larger Z, the much larger ( G , T ). The contribution of the study may provide a fair reference to the future tunneling in gravel formations of western Taiwan.