What is a suitable modelling approach for socio-technical systems? The answer to this question is of great relevance to decision makers in large scale interconnected network systems. Typical examples are the (inter)national and (inter)continental infrastructure networks found in the public utility sectors and network industries which provide, for example, energy, telecommunication and transportation services. These systems are characterized by both technical and social complexity. Each actor finds himself in a dynamic, multi-actor, multi-objective and multi-level jungle. Models to support an actor in making better decisions should be able to capture both the physical and social reality of the system, their interactions with one another and the external dynamic environment. Moreover, they must allow users to experiment with changes in both the physical and the social network configuration.
The agent-based modelling framework developed in the Next Generation Infrastructures project “Modelling Multi-Agent Systems”, executed as part of the Intelligent Infrastructures program, presents a suitable generic modelling approach for socio-technical systems. Agent-based models are particularly appropriate to experiment with different scenarios and to answer what if questions. This gives valuable support for decision makers in dealing with, for example, disturbances in the physical system or with new regulations imposed. The framework offers a flexible bottom-up approach, allowing users to experiment with changing system elements and configurations. The cornerstone of the framework is a shared language formalised in an ontology, which forms the interface needed to bring different aspects of the system (both social and physical) together and to interconnect different models. Besides interconnectivity, the ontology ensures interoperability of models and their building blocks, and therewith allows for re-use of available building blocks in building new models.
Following a well-defined benchmarking process and after comparing the agent-based approach with other modelling paradigms, it can be concluded that agent-based models are particularly interesting to use for decision support when a change in structure (either social, technical, or both) is required, when social elements (i.e. actors performing a specific task) have to be combined with technical elements, or when a natural representation of interaction between system elements (at different levels) is important Another distinct advantage lies in the re-use of elements of the model. Not only the conceptualisation could be re-used, but because of the bottom-up nature of the agent-based approach and the possibility to use of an ontology as an interface between “building blocks”, it is also possible to directly re-use source code from previous models, as well as input data. When it comes to explaining the model and the model results, agent-based modelling offers a natural representation of the decision making processes and interactions between the entities in the system.
Illustrative case studies of, for example, an oil refinery supply chain and an intermodal freight hub, demonstrate how agent-based simulation models developed with the new framework can successfully be applied to decision support. This way it contributes to the goals of the Intelligent Infrastructures programme, which has a focus on the short-term aiming at developing new, intelligent modes of operation for existing infrastructures.
