Analysis of transformations to socio-technical systems using agent based modeling and simulation
Shah, Anuj P.
MetadataShow full item record
Transformations to socio-technical systems may be enacted at fairly low levels, i.e., through changes in technology, processes, information and workers, to obtain a change in system level performance. This thesis provides a conceptual framework and a simulation platform to a priori computationally analyze the impact of such transformations. This thesis builds on the principles of cognitive engineering to describe the components of the work environment, i.e., technology, processes and information, in work relevant ways and using a structure-preserving model, i.e., a model form that describes the models using the same attributes and structure as used by system designers. This thesis also builds on the principles of agent-based modeling to model workers and their interactions with the work environment. These models are specified through a conceptual framework that includes declarative models describing which components are included within the system and their interrelations, and object-oriented models of those complex, dynamic behaviors that cannot be adequately described declaratively. Declarative modeling enables easy composition and modification of component models and, by declaring all required components collectively, a system model. This thesis established an XML representation for the declarative models, and developed a mechanism that automatically assembles, from the individual components' specifications and interrelations, a network-level model of the entire system in XML, which can serve to analyze network dependencies between components. The combination of the declarative and object-oriented models also enables computational simulations to predict the system performance that will emerge from a network of components when placed in a given scenario. Thus, this thesis also developed an agent-based simulation platform to support analysis of emergent behavior. The theoretical contributions of this thesis include the conceptual framework as a broadly applicable and structure-preserving representation of the important aspects of socio-technical system behavior, and associated extensions to cognitive engineering descriptions of the work environment. These insights, combined with the simulation platform, also enable computational modeling, analysis and prediction of socio-technical system performance with a comprehensiveness and detail not possible before. The theoretical and practical utility of these developments is demonstrated through a case study in air traffic control.