Scholarlink

<p>Geological conditions encountered during the Rishikesh–Karnaprayag railway tunnel construction deviated significantly from contractual predictions across rock classes C2 to B1. Stable B1 rock was absent, while squeezing (C2/C3) and rolling (B3) ground were encountered 4.1 times more than forecast. To address this uncertainty, pull length was modelled using data from over 200 blast rounds. Multiple linear regression correlated pull length to specific drilling, specific charge, and tunnel area. Specific charge decreased from 1.68&#xa0;kg/m³ in B1 to 0.70&#xa0;kg/m³ in C2 and further with increasing tunnel area, confirming a conservative, low-energy blasting strategy in poor ground. Rock-class-specific models achieved coefficient of determination (R²) values of 0.18–0.54, with the highest predictability in B3. Monte Carlo simulation (10,000 iterations) quantified prediction uncertainty, yielding 90% prediction intervals ranging from 0.98 to 1.47&#xa0;m in C2 to 1.95–3.90&#xa0;m in B1. These intervals provide a risk-informed basis for advance planning. The results demonstrate that New Austrian Tunnelling Method’s observational adaptability, guided by real-time blasting and support calibration, is essential for safe and efficient excavation in geologically unpredictable Himalayan terrain. The models offer field-validated, quantitative guidelines for blast design where conventional assumptions fail.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Applications of NATM in Garhwal Himalayan Tunnelling based on Field Evidence from Rishikesh–Karnaprayag Railway Tunnel Project

  • 期刊论文

  • 发表日期：

    2026年4月7日

查看全文
  • Khem Singh,
  • Siddharth Garia

摘要

Geological conditions encountered during the Rishikesh–Karnaprayag railway tunnel construction deviated significantly from contractual predictions across rock classes C2 to B1. Stable B1 rock was absent, while squeezing (C2/C3) and rolling (B3) ground were encountered 4.1 times more than forecast. To address this uncertainty, pull length was modelled using data from over 200 blast rounds. Multiple linear regression correlated pull length to specific drilling, specific charge, and tunnel area. Specific charge decreased from 1.68 kg/m³ in B1 to 0.70 kg/m³ in C2 and further with increasing tunnel area, confirming a conservative, low-energy blasting strategy in poor ground. Rock-class-specific models achieved coefficient of determination (R²) values of 0.18–0.54, with the highest predictability in B3. Monte Carlo simulation (10,000 iterations) quantified prediction uncertainty, yielding 90% prediction intervals ranging from 0.98 to 1.47 m in C2 to 1.95–3.90 m in B1. These intervals provide a risk-informed basis for advance planning. The results demonstrate that New Austrian Tunnelling Method’s observational adaptability, guided by real-time blasting and support calibration, is essential for safe and efficient excavation in geologically unpredictable Himalayan terrain. The models offer field-validated, quantitative guidelines for blast design where conventional assumptions fail.