Abstract : Tunnel projects within rock masses of low mechanical properties, usually described as "hard soils soft rocks" (H.S.S.R.), regularly encounter numerous difficulties. These materials show specific behaviours that make them atypical in the points of view of both rock and soil mechanics, leading to more complex designs and generating significant extra costs for the contracting authority. Considering the limited case of H.S.S.R. that can be modelled with a continuous approach, non-linearity of the failure criterion, hydromechanical coupling, and stress-dependency of the material deformability appear as distinctive behaviour features that may have a serious influence on the excavation stage (short term equilibrium). Although these properties can be adequately taken into account in some numerical models, the practical experience of tunnel calculations has long shown the interest of simplified methods such as the convergence-confinement approach. These methods allow reasonably representative designs through not very complex formulations and contribute to make sensitivity studies easier thanks to their quick and simple implementation. Thus, based on a classical "porous media" description taking into account both mineral and fluid compressibilities, and on a non-dimensional expression of the Hoek-Brown failure criterion (including edge effects), new formulations for the calculation of ground characteristic curves are presented. After a first approach considering only the non-linearity of the failure criterion, in a monophasique edium, more complex cases are studied so as to take into account biphasique drained or undrained situations. Each time, a complete solution is described, leading to explicit formulations or to differential equations that can be simply solved with a one-step numerical method. A "turnkey" design tool, in a spreadsheet form, that demonstrates the easy implementation of the developed solutions, is systematically provided. The undrained case is then completed with a calculation procedure that integrates the Fahey-Carter non-linear elasticity model, using the transfer matrices method. The last part of this work tackles the applicability of these analytical solutions on a case study, the Arbus tunnel (France). It emphasizes a few difficulties in the determination of some parameters of the mechanical model when "standard" test protocols are used, and insists on the natural variability of the ground properties in this molassic geology of the Pyrenean piedmont. This context stresses the advantages of the developed methods : besides a better description of the rock mass behaviour, they allow to identify at low cost the more influent parameters on the ground - lining equilibrium and to introduce the analysis of uncertainties in design practice. The calculations also underline the interest of non-linear elasticity models, promoting a better assessment of material deformability and also reducing the model sensitivity to elastic parameters scattering. The here-presented approaches keep some limits however, such as the necessary distinction between short term and long term equilibriums, the hypothesis of hydrostatic far-field stresses that is shown to be quite strong in the end, and the problem of lining representation (confinement line) that would deserve a deeper analysis even if a few acceptable methods are already available in the literature.