Reliability-Based Stability Analysis of Rock Slopes Using Numerical Analysis and Response Surface Method


Dadashzadeh N., Duzgun H. S. B., Yesiloglu-Gultekin N.

ROCK MECHANICS AND ROCK ENGINEERING, cilt.50, sa.8, ss.2119-2133, 2017 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 50 Sayı: 8
  • Basım Tarihi: 2017
  • Doi Numarası: 10.1007/s00603-017-1206-2
  • Dergi Adı: ROCK MECHANICS AND ROCK ENGINEERING
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.2119-2133
  • Anahtar Kelimeler: Rock slope stability, Uncertainty, Numerical simulation, Response surface method, First-order reliability method, SYSTEM RELIABILITY, STRUCTURAL RELIABILITY, SOIL SLOPES, FAILURE
  • Orta Doğu Teknik Üniversitesi Adresli: Evet

Özet

While advanced numerical techniques in slope stability analysis are successfully used in deterministic studies, they have so far found limited use in probabilistic analyses due to their high computation cost. The first-order reliability method (FORM) is one of the most efficient probabilistic techniques to perform probabilistic stability analysis by considering the associated uncertainties in the analysis parameters. However, it is not possible to directly use FORM in numerical slope stability evaluations as it requires definition of a limit state performance function. In this study, an integrated methodology for probabilistic numerical modeling of rock slope stability is proposed. The methodology is based on response surface method, where FORM is used to develop an explicit performance function from the results of numerical simulations. The implementation of the proposed methodology is performed by considering a large potential rock wedge in Sumela Monastery, Turkey. The accuracy of the developed performance function to truly represent the limit state surface is evaluated by monitoring the slope behavior. The calculated probability of failure is compared with Monte Carlo simulation (MCS) method. The proposed methodology is found to be 72% more efficient than MCS, while the accuracy is decreased with an error of 24%.