Several carefully scaled physical models (scale 1:300) were used to study the two-dimensional deformation resulting from excavation of very large span openings in near-surface jointed rock masses. The models consisted of at least 20,000 discrete blocks. Joint orientations and stress levels were varied, and some models were dynamically loaded to simulate elements of strong earthquake's (peak horizontal acceleration of 0.2–0.7 g). The purpose of these model studies, which were performed at the Norwegian Geotechnical Institute many years ago, was to provide deformation data to compare with monitored data from future engineering projects. Recently, since the discontinuum code UDEC-BB (BB for the Barton-Bandis joint model) is used at NGI as a design tool, the code has been applied to simulate the same dynamic load on similar numerical models. Not unexpectedly, where the joint orientation was unfavourable for the wall stability the dynamic loading on both physical and numerical model caused considerable distress. Where joint geometries were favourable to stability, dynamically loaded physical and numerical models suffered no block falls or collapse, despite the lack of support. In general there has been good correlation between the results obtained from the physical and numerical models.


Plusieurs modèles reduits à l'echelle du 1/300 ont ete utilises pour etudier les deformations bi-dimensionnelles consecutives à la raalisation d'ouvertures de grande ampleur et à faible profondeur dans des massifs rocheux fractures. Les modèles etaient constitues d'au moins 20,000 blocs unitaires. Differentes orientations des joints et niveaux de contraintes ont ete etudies, et quelques modèles ont ete soumis à des sollicitations dynamiques afin de simuler de violents tremblements de terre (acceleration horizontale maximale de 0.2 a 0.7 g). L'objectif de ces etudes, realisees au NGI il y a plusieurs annees, etait de fournir des valeurs previsionnelles de deformation à comparer avec des valeurs mesurees sur des projets futurs. Recemment, Ie code de calcul discontinu UDEC-BB (BB: modelisation des joints Barton-Bandis), qui est utilise au NGI comme outil de dimensionnement, a ete applique à la simulation des mêmes chargements dynamiques sur des modèles numeriques similaires aux modèles reduits.


Die zwei-dimensionale Verformung, verursacht durch das Auffahren sehr groβer und oberflachennaher Kavernen, wurde anhand von mehreren Modellen im Maβstab 1:300 untersucht. Die Modelle bestanden aus mindestens 20 000 einzelnen Blöcken. Es wurden unterschiedliche Kluftorientierungen und Spannungszustande untersucht und mehrere Modelle wurden dynamischen Belastungen ausgesetzt, um starke Erdbeben zu simulieren (maximale horizontale Beschleunigung: 0.2–0.7 g). Diese Modellversuche wurden vor vielen Jahren am NGI durchgefuehrt, um Verformungsdaten fuer Vergleiche mit zukuenftigen in situ Messungen und numerischen Berechnungen zur Verfuegung zu stellen. Das diskontinuierliche Programm UDEC-BB (BB - Barton Bandis Kluftmodell) wird am NGI als Entwurfswerkzeug angewandt, und wurde verwendet um das Verhalten der Modellkavernen unter dynamischen Belastungen zu berechnen.

Comparison of physical and numerical models

Both numerical and physical models can be invaluable in correctly interpreting the results of displacement monitoring of important large span, near-surface openings. Physical models have a double role in that they also provide a check for numerical model results. The number of blocks that can be physically modelled is closer to that in practice. The number of blocks that can be presently handled by the Universal Distinct Element Code (UDEC) is directly dependent on the RAM memory of the computer involved. It must be stated though that running numerical models with more than 10,000 discrete blocks is still very time consuming and considered to be inefficient.

This content is only available via PDF.
You can access this article if you purchase or spend a download.