Predicting Tunnel Squeezing Using Multiclass Support Vector Machines

Table 1

Summary of previously published empirical correlations for predicting tunnel squeezing.

Proposed by	Correlations	Required parameters	Sources	Type
Jethwa et al. [10]			—	Semiempirical
Singh et al. [11]			Thirty-nine case histories	Empirical
Aydan et al. [12]			Cases from Japan tunnels	Semiempirical
Barla [13]			—	Semiempirical
Bhasin and Grimstad [14]			Tunnel case histories	Semiempirical
Hoek [15]			Finite element models	Semiempirical
Jimenez and Recio [6]			Sixty-two case histories	Empirical
Dwivedi et al. [16]			Sixty-three case histories	Empirical

N_c (or α): competency factor (also called “strength stress ratio (SSR)”), σ_cm: rock mass uniaxial compressive strength (MPa), γ: rock mass specific weight (MN/m³), H: overburden or depth of tunnel (m), c_p: rock mass peak cohesion (MPa), ϕ_p: rock mass peak friction angle (degree), Q: rock tunneling quality index, σ_c: uniaxial compressive strength of intact rock (MPa), σ_θ: tangential stress (MPa), ε: percentage strain (ratio of tunnel closure to tunnel diameter), p₀: in situ vertical stress at tunnel depth (MPa), p_i: internal support pressure (MPa), σ_v: vertical in situ stress (MPa), and K: support stiffness (MPa).