Analysis and prediction of mechanical characteristics of corrugated plate as primary support in tunnels. (May 2021)

• Record Type: Journal Article
• Authors: Sun, Keguo; Hong, Yiqin; Xu, Weiping; Hou, Zonghao; Liu, Xu; Yu, Mingzhao; Yuan, Ziyi
• Is Part Of: Tunnelling and underground space technology. Volume 111(2021)
• Publication Date: 2021-05
• Subjects: Corrugated steel plate -- Size -- Finite-element simulation -- BP neural network -- Verification experiment; Tunneling -- Periodicals; Underground construction -- Periodicals; Tunnels -- Periodicals; Underground areas -- Periodicals; 624.193
• DOI: 10.1016/j.tust.2021.103845
• Languages: English
• ISSNs: 0886-7798

Abstract

Highlights: Outcomes of beam-element and shell-element model are similar. Adding thickness is not so effective when corrugated plate is thick enough. Decreasing width is still effective though it is already small. Increasing height is well effective when it is around 100 mm. BP neural network is able to predict maximum deformation and stress by sizes. Abstract: Traditional primary support made of reinforced concrete is widely applied but can bring many problems. Therefore, to study the mechanical behavior of corrugated-plate structure as primary support, several finite-element simulations in different sizes, shapes and elements were carried out. Besides, to ensure the exactness and practicability, a verification experiment and a BP neural network for prediction were done. The result shows that the outcome of beam-element and shell-element model is similar and the main deformation and stress is respectively down and compressive. And the difference between horseshoe-shaped and circular tunnel is small in aspect of vertical displacement and stress. Then, when t (thickness) increases, d (maximum vertical deformation) and s (maximum stress) decrease more and more slowly. When w (width) ascends, d and s improve more and more slowly. When h (height) becomes bigger, d declines more and more slowly and s goes down slowly, fast and slowly again. The outcome of the experiment conforms with the simulation. Next, by BP neural network, using top load and t, w, h to predict d and s is feasible and efficient. It performs better when predicting d . This paper can provide reference for the application of corrugated-steel-plate structure as new-type primary support.