Abstract: In order to minimize disturbance impact in complex organizations, an agent-based decision support system is proposed using a distributed cooperative scheduling repair method. The approach is based on an isomorphic modelling of complex organizations, where agents operate cooperative repair behaviours. Agents implement solving strategies composed of atomic repair operations. The resulting multiagent system will operate the repair behaviours in order to determine the solution which minimizes the disturbance impact and to propose it to the human decision maker. This paper focuses on the description of cooperative repair operations and cooperation strategies that agents develop through their behaviours in order to find solutions to disturbances and propose them to human decision makers. Key-words—Agent modelling, decision support system, cooperative scheduling repair, Multiagent system, socio-technical organization. I. INTRODUCTION The current competitive economic context leads companies to deal with highly restrictive delays and labour fall [1]. In this context, the management of disturbances becomes a major issue to tackle in the management of socio-technical systems [2]. Disturbances lead to important cost increase. Socio-technical organization problems are mostly complex due to their topographic distribution and the variety of tasks to be achieved. This results in the difficulty of solving organizational problems especially on tasks planning. It also highlights the interest of helping decision makers in disruption management of pre-established planning. The complexity of disruption management needs to be taken into account in order to be closer to the reality of industrial systems (tasks with nonlinear constraints, human management, interruptibility of actions, etc.). Most current methods of management of schedules mainly reflect clearly defined physical and temporal constraints. The human element in the re-scheduling is generally not considered as essential. Therefore a solution can be to minimize impacts on human resources by changing as little as possible their working organizations and their forecasted frameworks. The search for impact minimization involves the implementation of cooperative decision processes that actors would naturally implement. Manuscript received March 18, 2011. S. Fournier (corresponding author), A. Ferrarini, and A. Cauvin are Associate Professors in P.Cezanne University, LSIS, UMR CNRS 6168, Domaine Universitaire de Saint-Jérôme, Avenue Escadrille Normandie- Niemen, 13397 MARSEILLE CEDEX 20, France ; (email : {firstname.lastname}@lsis.org). An approach based on agent modelling and its operationalization has been proposed. This approach has been initially developed for workshop rescheduling, supply chain management problem and building site organization [3]. This approach implements an agent-based modelling of socio-technical organizations, and runs a distributed cooperative problem solving method. With the aim of fully taking into account the complexity of present organizations, the basic axioms of the original method are redefined together with the addition of new dimensions to the problematic and new operators for solving disruptions problems. The extension of this method, based on the distributed recovery of planning, consists in developing the cooperative repair behaviours between actors based on repair operations and repair strategies. Then, the problem must be transposed and modelled thanks to the agent paradigm [4]. II. DISTRIBUTED SCHEDULE REPAIR The presented method was initiated on a rescheduling problem in production workshops inside enterprises [4]. Then it was extended and improved in the context of inter- enterprise level (for the management of supply chains) [5]. It is based on an organizational model of industrial systems whereas all the actors and resources of the system have a sufficient decisional autonomy to manage the occurrence of disruptions as soon as possible [6][7]. The principle of the proposed method comes from the cooperative methods of distributed solving of disruptions first implemented in the domain of production and logistics. This principle, extended in this issue, is based on the development of solutions by the actors themselves for limiting the impact of disruptions throughout the site while ensuring the achievement of objectives. Rather than implementing a new planning calculation, the purpose is to repair the existing situation. A wide variety of different approaches to repair schedule or rescheduling have been proposed as a part of Dynamic Scheduling approach [8]. Schedule repair refers to "local" adjustment of the current schedule and may be preferable for many reasons especially in the dynamic context of industrial systems [5]. Schedule repair can be done in a centralized way [9], but agent approaches have been highlighted as to be well suited to complex scheduling problems [10] [11]. Inherent capabilities of MAS induce reduced complexity, self-configuration, increased flexibility, and reduced costs as well as allowing heterogeneous and multiple criteria to evaluate and choose among the possible solutions. A Cooperative Agent–Based Scheduling Repair Method for Managing Disruptions in Complex Organisations Sébastien Fournier, Alain Ferrarini, Aline Cauvin