Improving Agent-Based Simulation of Major Incident Response in the United Kingdom through Conceptual and Operational Validation

Introduction

Traditionally, in the United Kingdom (UK), practising and testing the response of the emergency services to major incidents is performed via live, table-top, or discussion-based exercises (UK Cabinet Office, 2011). However, computer simulation of the response to major incidents has been acknowledged as being able to play a role in emergency preparedness (Longo, 2010). Whereas traditional exercises often evaluate only the response according to the plans set out in documents detailing how the UK emergency services are to respond to a major incident such as those referred to in Figure 1, computer simulations ‘allow evaluation of alternative strategies to respond to a disaster event’ (Jain & McLean, 2003). Furthermore, these disaster events can be ‘scenarios that would be prohibitively expensive, dangerous, environmentally damaging, or even physically impossible to re-create in reality’ (Straylight, 2010). Thus computer simulation benefits emergency preparedness by expanding the scope of what is possible through the three traditional types of exercise.

Figure 1.

UK emergency services’ literature related to responding to major incidents

Computational methods which have been used to simulate emergency response include systems dynamics (Zhong & Kim, 2011), discrete-event simulation (Connelly & Bair, 2004), stochastic modelling (Mukherjee & Gupta, 2009), queuing networks (Au-Yeung et al., 2006) and agent-based simulation (Hawe et al., 2013). Agent-based simulation (ABS) is particularly popular for simulating the response to emergencies, however ‘there are hardly any agent systems used in disaster management practice’ (Fiedrich & Burghardt, 2007). Engaging practitioners and convincing them that ABS is a useful tool involves: (1) providing them with easy-to-use ABS software (“a good interface is important in engaging stakeholders in the process” (Hoad & Watts, 2012)); (2) capturing their domain at an appropriate level of detail (“in order to improve engagement and make ABM more useful to decision makers, the models first needed to be less abstract, that is, based on documentary evidence or stakeholder-generated scenarios captured in the language of the stakeholders” (Hoad & Watts, 2012)); (3) removing objections to use through validation (‘after all, if a model runs perfectly well and correctly reflects the workings of an important real-world system, why wouldn’t someone want to use the model?’ (North & Macal, 2007)).

Previously, our efforts towards addressing point (1) through the development of the STORMI (‘Simulation of the Tactical and Operational Response to Major Incidents’) Scenario Designer have been reported in (Hawe et al., 2012a); this is a Graphical User Interface (GUI) designed specifically to make our ABS software usable for emergency planning officers. In this paper, the focus is on our efforts to address point (3), which involves providing enough details regarding point (2) to do so. That is, identifying and addressing those weaknesses in STORMI which concern practitioners most through a process of conceptual and operational validation in order to improve the credibility of our ABS of major incident response in the UK. Achieving this will contribute towards removing potential barriers to adoption by practitioners and the long term goal of agent-based modelling and simulation being accepted as playing a useful role in emergency preparedness.