Preventing and resolving design conflicts for collaborative convergence in distributed set-based design.

Summary In distributed design, in the product sizing phase, inconsistencies may emerge between design goals and between the work procedures of heterogeneous subsystems. In this phase, the design actors must collaborate in a concurrent manner, as their tasks are linked to each other by the design couplings between their sub-problems. Inconsistencies can cause design conflicts due to these couplings. The question is how to achieve collaborative convergence to satisfy the global and individual goals of design actors when these actors make design decisions under uncertainty. The objective of this thesis is to propose a model to prevent and resolve design conflicts, while overcoming the problem of design uncertainty with the "set-based design" (SBD) approach. For this purpose, design attitudes are modeled with the "Beliefs-Desires-Intentions" paradigm to explore inconsistencies and manage conflicts in design processes. The conventional bottom-up approach is thus extended with multi-agent modeling techniques. In this approach, design agents can set requirements directly on their "well-being" indicators. These indicators represent how their design goals are likely to be satisfied at a given point in the process. Monte Carlo simulations are performed to evaluate the performance of this approach, offering a variety of agent attitudes. Compared to traditional bottom-up and top-down design approaches, the results reveal fewer design conflicts and reduced conflict intensity. Constraint satisfaction problem (CSP) techniques and design attitudes are applied to detect and justify design conflicts between heterogeneous agents. A new form of the Cooperative CSP (CoCSP) model is thus developed to resolve design conflicts by detecting the trade-off between constraints. The conflict resolution system can be adopted through different proposed strategies that take into account the solidarity architecture of the agents. Simulation results show that the intensity of conflicts in distributed design is reduced by promoting solidarity that triggers help to suffering agents.