INTERNATIONAL JOURNAL FOR NUMERICAL AND ANALYTICAL METHODS IN GEOMECHANICS Int. J. Numer. Anal. Meth. Geomech. 2008; 32:559–577 Published online 29 June 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/nag.637 The probabilistic distinct element method P. Moarefvand 1 and T. Verdel 2, ∗, † 1 Mining Engineering Department, Zanjan University, 5 km Tabriz road, Zanjan, Iran 2 Laboratoire Environnement G´ eom´ ecanique et Ouvrages, Nancy Universit´ e, Ecole des Mines - Parc Saurupt, 54042 Nancy Cedex, France SUMMARY The distinct element method (DEM) that appeared in the 70s is now widely used in the field of Rock Mechanics to study the stability of fractured rock masses, the field of application it has been dedicated to. Until now, most of the engineers and researchers who have been using this method with a code such as Udec (Itasca Consulting Group), ignore in their model the uncertainties that come with the physical and mechanical properties of the studied rock masses. Nevertheless, because collected data on rock masses are usually very limited in number, and variable as well as imprecise in nature, investigation on such media is suffering from uncertainty. To take into account the uncertainties in data when carrying numerical models with the DEM and to analyse their effects on model results, we have developed a ‘probabilistic distinct element method’ based on the Udec code. This method allows entering most of the necessary physical and mechanical data under a probabilistic form and returns results also expressed in this form. Its main interest is then to provide a range of probable results instead of a non-guaranteed unique solution. We have implemented this method by using the technique of approximation of the two first statistical moments to build a probabilistic calculation code named Pudec. It takes into consideration rigid or fully deformable block as well as joint mechanical behaviour and does not consider fluid and heat flow calculations. Copyright 2007 John Wiley & Sons, Ltd. Received 5 May 2007; Accepted 5 May 2007 KEY WORDS: distinct element method; uncertainties; probabilistic analysis; sensitivity approach; statistical moments 1. INTRODUCTION Because the physical reality is spoilt by uncertainties, decision-makers should introduce extra information in their decision process. There are a lot of sources of uncertainty but only uncertainties ∗ Correspondence to: T. Verdel, Laboratoire Environnement G´ eom´ ecanique et Ouvrages, Nancy Universit´ e, Ecole des Mines - Parc Saurupt, 54042 Nancy Cedex, France. † E-mail: thierry.verdel@mines.inpl-nancy.fr Copyright 2007 John Wiley & Sons, Ltd.