INTRODUCTION
This paper is based on the premise that a tunnel is not just a hole in the ground but a well-defined balance of a rock or soil mass as it occurs in nature, and the excavation and support of the underground opening. Thus, the design of a tunnel is inseparable from its construction.
The intent of tunnel design is to predict the stability of a tunnel as excavated and to define the support required if stabilization is necessary. This design process results in a formal tunnel design with technical descriptions to be incorporated in the contract documents that will ensure construction procedures that will result in the tunnel as designed.
In the Americas tunnel design almost exclusively follows the concept by TERZAGHI (1946), in which he empirically related nine rock conditions with quantitatively defined rock loads, assuming conventional tunneling procedures. Several derivatives thereof, e.g. the Rock Structure Rating (RSR) by JACOBS Assoc. (1974), have improved the original idea, are very practice oriented, and are used quite widely. In most cases design conclusions of these approaches tend to be conservative.
Numerous authors have demonstrated the use of the Finite-Element Method (FEM) in tunnel design. They have produced a wealth of information, much of which is of considerable general interest, concerning the three-dimensional stress and deformation patterns around tunnels, particularly tunnel headings. In view of limitations with respect to input data generally available, FEM's potential can rarely be fully utilized in the practice of engineering design. There are, however, noteworthy exceptions.
LAUFFER's (1958) concept of "Unsupported width of Tunnel Span vs. Standup Time" is also very practice oriented but presents the tunneling engineer with the dilemma of classifying rock based on unsafe conditions hopefully not to be encountered in the actual tunneling process.
BIENIAWSKI (1973) of CSIR, and BARTON, LIEN, and LUNDE (1974) of NGI proposed indices for determination of the tunneling quality of a rock mass. Both systems are almost exclusively based on geological and geotechnical input data, whereas the affect of the tunneling design and construction are not distinctly considered. Thus, the resulting "Rock Classes" are only empirically related to tunnel design and actual tunnel performance. In the Alpine Countries somewhat comparable classification systems such as those by RABCEWICZ and PACHER (1974), SIA (1975), and many others have proved useful over many years of application, mostly in conjunction with specific methods of tunnel construction.
This paper presents an outline of an engineering compromise, based on both practical and theoretical experiences of the authors in various technological environments, using as many aspects as possible of existing tunnel classification and design methods. An attempt has been made to present a reproducible design procedure which can be systematically reviewed and modified during construction if the need arises. Thus, this approach is intended to discourage arbitrary assessment of changed conditions by the different parties involved, without eliminating professional judgement which modern tunneling still heavily depends on.
The approach presented here covers only design of the primary support required to provide a stable and safe underground opening and stops short of the design of the final lining.
