Vertical displacement estimation in roof and floor of an underground powerhouse cavern. (August 2018)

• Record Type: Journal Article
• Authors: Rezaei, Mohammad; Rajabi, Morteza
• Is Part Of: Engineering failure analysis. Volume 90(2018)
• Page Start: 290
• Page End: 309
• Publication Date: 2018-08
• Subjects: Powerhouse coven -- Vertical displacement -- Numerical analysis -- Neural network -- Fuzzy logic; System failures (Engineering) -- Periodicals; Fracture mechanics -- Periodicals; Reliability (Engineering) -- Periodicals; Pannes -- Périodiques; Rupture, Mécanique de la -- Périodiques; Fiabilité -- Périodiques; Fracture mechanics; Reliability (Engineering); System failures (Engineering); Periodicals; Electronic journals; 620.112
• DOI: 10.1016/j.engfailanal.2018.03.010
• Languages: English
• ISSNs: 1350-6307

Abstract

Abstract: Vertical displacement (VD) estimation is one of the most important topics in construction of an underground powerhouse cavern. This phenomenon commonly depends on excavation progress and corresponding geometrical, geological and geomechanical conditions of surrounding rock masses. Although many researchers investigated this issue, however, application of the new powerful predictive models is essential due to the topic complexities and limitations of previous methods. This study focuses on the estimation of VDs at the middle and key points on the roof and floor of a powerhouse cavern using the numerical analysis (NA), neural network (NN), fuzzy logic (FL) and statistical analysis (SA) models under various conditions. Rock geomechanical characteristics and geometrical variables of powerhouse related structures regarded as inputs in numerical analysis of VDs. On the basis of a large number of numerical simulations, a new predictive equation considering seven effective basic factors was fitted to predict the elasto-plastic vertical displacement at the key points on roof and floor of powerhouse cavern. Through the numerical analysis, an enough datasets was introduced to construct and develop the NN, FL and SA models. Based on the testing datasets, the achieved results from these models were compared with each other as well as with the results of NA-based equation using the relevant statistical indices. This comparison confirmed that the performance of FL and NN models are relatively better than the NA-based equation. Moreover, these three models are more relevant than the SA model. At the end of modeling, conducted sensitivity analysis proved that overburden depth, lateral stress coefficient are the most effectual variables on the VDs whereas the tensile strength is the least ones. Highlights: NA-based equation and FL, NN and SA models were suggested. Models were developed using the numerical simulations datasets. High accuracy of NA-based equation, FL and NN models have proved. Overburden depth has proved as the most effectual variable on VDs.