Koen H. van Dam

Agent-Based Modelling of Socio-Technical Systems

Decision makers in large scale interconnected network systems require simulation models for decision support. The behaviour of these systems is determined by many actors, situated in a dynamic, multi-actor, multi-objective and multi-level environment. How can such systems be modelled and how can the socio-technical complexity be captured? Agent-based modelling is a proven approach to handle this challenge. This book provides a practical introduction to agent-based modelling of socio-technical systems, based on a methodology that has been developed at TU Delft and which has been deployed in a large number of case studies. The book consists of two parts: the first presents the background, theory and methodology as well as practical guidelines and procedures for building models. In the second part this theory is applied to a number of case studies, where for each model the development steps are presented extensively, preparing the reader for creating own models.

Optimizing the Location of Intermodal Freight Hubs: An Overview of Agent Based Modelling Approach

This paper aims at developing an integral model for the evaluation of road-rail intermodal freight hub location decisions. The model comprises four dominant agents, namely, hub owners or operators; transport network infrastructure providers; hub users; and communities. An agent based modelling approach is introduced to allow such negotiation to happen to achieve a global objective. The paper outlines the methodology to be used. It also presents an initial location selection process, a testing with individual objective functions, and a design for an agent based model using a case study of intermodal freight hub location decisions in South East Queensland of Australia.

Agent-Based Control of Distributed Electricity Generation with Micro Combined Heat and Power: Cross-Sectoral Learning for Process and Infrastructure Engineers

For the distributed control of an electricity infrastructure incorporating clusters of residential combined heat and power units (micro-CHP or ?CHP) a Multi-Agent System approach is considered. The network formed by households generating electricity with ?CHP units and the facilitating energy supplier can be regarded as an electricity production system, analogous to a (flexible) manufacturing system. Next, the system boundary is extended by allowing the trade of electricity between networks of households and their supplier. A methodology for designing an agent-based system for manufacturing control is applied to both cases, resulting in a conceptual design for a control system for the energy infrastructure. Because of the analogy between production systems and infrastructures Process Systems Engineering (PSE) approaches for optimisation and control can be applied to infrastructure system operations. At the same time we believe research on socio-technical infrastructure systems will be a valuable contribution to PSE management strategies.

Benchmarking numerical and agent-based models of an oil refinery supply chain

Abstract Todays integrated refinery supply chains embrace two distinct types of complexities—(i) a complex production processes that can operate in various regimes, handling different raw materials and producing a variety of products, and (ii) a complex network of intelligent plan-source-move type elements that synchronize among the far-flung supply chain entities to ensure smooth, efficient, and profitable operation. Modelling such socio-technical systems poses a significant challenge. Traditionally numerical modelling has been the preferred choice, especially in process systems engineering; but recently agent-based modelling, which take an actor-centric perspective, has begun to be considered as an alternative. In this paper, we critically evaluate the choice of modelling paradigms for an integrated oil refinery supply chain. Initial experiments confirm that the behaviour of the two models is the same—thus validating the conjecture that a problem can be adequately described in both paradigms. The equation-based model appears to be better suited for describing complex physical and chemical phenomena; the agent-based model allows efficient ways to describe actions and behaviours of human and decision-making elements where cooperation and negotiation between intelligent entities come to the fore.

Re-use of an ontology for modelling urban energy systems

The use of ontologies for the interoperability of software models is widespread, with many applications also in the energy domain. By formulating a shared data structure and a definition of concepts and their properties, a language is created that can be used between modellers and—formalised in an ontology—between model components. When modelling energy systems, connections between different infrastructures are critical, e.g. the interaction between the gas and electricity markets or the need for various infrastructures including power, heat, water and transport in cities. While a commonly shared ontology of energy systems would be highly desirable, the fact is that different existing models or applications already use dedicated ontologies, and have been demonstrated to work well using them. To benefit from linking data sources and connecting models developed with different ontologies, a translation between concepts can be made. In this paper a model of an urban energy system built upon one ontology is initialised using energy transformation technologies defined in another ontology, thus illustrating how this common perspective might benefit researchers in the energy domain.

Abnormal Situation Management in a Refinery Supply Chain Supported by an Agent-Based Simulation Model

Abstract Oil refineries are of high importance for global economic health and energy supply; any disruptions to their operations may have major worldwide impact. In this paper the application of an agent-based refinery supply chain model to abnormal situation management is described. Agents represent the various decision makers in the supply chain. They own, operate and manage the elements of the physical network of the supply chain. A disruption in ship arrival is used to illustrate the applicability of the decision support system. The decision support system derives a suitable course of action for a given situation based on the outcomes of a number of simulation runs according to the Nelder-Mead zero-order optimization method. This method is based on identification of the best, the worst, and the second worst outcomes in each iteration for the pre-defined experiment. It can be concluded that the decision support system can interact with multiple actors in the supply chain to diagnose and compensate for unanticipated disruptions, with a substantial impact on refinery productivity.

Modelling an electricity infrastructure as a multi-agent system — Lessons learnt from manufacturing control

Abstract To model the control of an electricity infrastructure incorporating domestic level combined heat and power units (micro-CHP) a Multi-Agent System (MAS) approach is considered. This approach has already successfully been used to control manufacturing systems in the process industry. Because similarities between manufacturing and electricity generation exist it is interesting to investigate how a MAS methodology designed for manufacturing systems is applied to an electricity infrastructure. The interaction between energy companies and households is viewed here in a novel way, namely as a production process. By using an existing methodology for manufacturing control to design an agent-based controller for an electricity infrastructure, issues can be identified that have to be addressed in a control methodology specific for infrastructures.

Distributed intelligence in autonomous multi-vehicle systems

To make better use of existing infrastructures, new control methods are under development. In the Intelligent Infrastructures programme, an infrastructure is seen as a multi-agent system, with more or less autonomous subsystems that are related to each other in hierarchical, coordinated, cooperative, or non-cooperative way. Current controls for multi-vehicle systems are based on the hierarchical control concept. In this paper, it is shown how the incident handling, efficiency, and flexibility of multi-vehicle systems can be improved by applying a cooperative control strategy. An existing multi AGV application in a seaport illustrates that efficiency of operations can be improved considerably with smarter control. Finally, a research project is introduced concerning cooperative multi-agent control of true free-ranging automated guided vehicles.

Simulating residential electricity and heat demand in urban areas using an agent-based modelling approach

Cities account for around 75% of the global energy demand and are responsible for 60-70% of the global greenhouse gasses emissions. To reduce this environmental impact it is important to design efficient energy infrastructures able to deal with high level of renewable energy resources. A crucial element in this design is the quantitative understanding of the dynamics behind energy demands such as transport, electricity and heat. In this paper an agent-based simulation model is developed to generate residential energy demand profiles in urban areas, influenced by factors such as land use, energy infrastructure and user behaviour. Within this framework, impact assessment of low carbon technologies such as plug-in electric vehicles and heat pumps is performed using London as a case study. The results show that the model can generate important insights as a decision support tool for the design and planning of sustainable urban energy systems.

An integrated optimisation platform for sustainable resource and infrastructure planning

Abstract It is crucial for sustainable planning to consider broad environmental and social dimensions and systemic implications of new infrastructure to build more resilient societies, reduce poverty, improve human well-being, mitigate climate change and address other global change processes. This article presents resilience.io, 2 a platform to evaluate new infrastructure projects by assessing their design and effectiveness in meeting growing resource demands, simulated using Agent-Based Modelling due to socio-economic population changes. We then use Mixed-Integer Linear Programming to optimise a multi-objective function to find cost-optimal solutions, inclusive of environmental metrics such as greenhouse gas emissions. The solutions in space and time provide planning guidance for conventional and novel technology selection, changes in network topology, system costs, and can incorporate any material, waste, energy, labour or emissions flow. As an application, a use case is provided for the Water, Sanitation and Hygiene (WASH) sector for a four million people city-region in Ghana.