Constitutive model for rock fractures: Revisiting Barton's empirical model