SIMULACRA: a systemic multi-agent architectural framework for organizational simulation
University of New Brunswick
This dissertation introduces, realizes, and evaluates SIMULACRA (SystemIc MULti-Agent arChitectural fRAmework), a novel tool to improve the ability of researchers and practitioners to investigate multi-organizational systems, by providing a multi-dimensional, multi-paradigm simulation environment in which the effects of model parameters can be observed, modified, and evaluated. Traditionally, organizations have been studied in isolation, from a single perspective, such as for business process modelling or IT integration. However, the nature of organizations|underscored by investigating organizational ecosystems, as seen, for example, in emergency response|is too complex to be captured by a single perspective. What is needed, instead, is a multi-disciplinary perspective that examines an organization systemically. This research considers organizations from such a systemic viewpoint, introducing a seven-dimensional organizational modelling methodology composed of the structural, functional, normative, social, psychological, physical, and informational dimensions. It further sets forth, using the proposed modelling methodology as a basis, an architectural framework to simulate organizations, their interrelationships, and the environment using a combination of simulation techniques (e.g., discrete-event, system dynamics, and BDI-agent simulation). The realization of this architectural framework in the form of the interactive and visual SIMULACRA tool is described, along with its implementation using AnyLogic and Brahms. Lastly, SIMULACRA is applied to a case study from emergency preparedness and response, wherein the policies and practices of a harbour-security organizational ecosystem are examined and their impact on achieving threat-level consensus analyzed. Compared with existing approaches, the results demonstrate that SIMULACRA offers a meaningful test-bed for exploring various what-if scenarios by more fully and naturally representing and simulating the system of interest, thereby facilitating the analysis and investigation of organizational systems.