A simulation-based tool for environmentally hazard-conscious urban design

The research is part of the scientific and methodological framework concerning new support tools for urban planning oriented toward the prevention of risk due to natural disasters. In this context, the first results of the prototype of a digital (agent-based) simulation system of crowd movement based on configured risk scenarios are reported. In general, agent-based modelling (ABM) is a computational technique that can simulate systems by representing individual agents with specific behaviours and rules. It allows for the study of emergent phenomena in various spatial contexts, offering insights into complex systems’ dynamics and interactions. From the point of view of pre-disaster planning, agent-based modelling can be a tool that helps simulate crowd behaviour in emergencies. It factors in individual actions and interactions to predict collective responses, aiding in crisis management and policy development. Experiments that use agent-based modelling to simulate situations such as those described, now, focus mainly on scenarios linked to the evacuation of single buildings. Few, however, have been carried out at the neighbourhood or urban scale. Of those encountered in the process of bibliographical investigation of the subject, none concern Italian territory. In this context given, the objectives of the research are multiple. On the one hand, there is, in fact, the understanding of the potential reaction of the crowd in a built environment in case of natural disasters, for now at a neighbourhood scale, based on predetermined boundary conditions that can also be changed during the simulation itself in real-time. On the other hand, there is, consequently, the implementation of urban design methodologies capable, based on simulation results, of optimizing the spatial organization of places and urban shapes. In this regard, a methodology composed of five blocks is proposed. The first one concerns knowledge: the construction of an information system capable of collecting and managing data of different types, formats, and origins is carried out. The logic lies in understanding, also through the representation of the data themselves, the current state of the area studied based on scientific evidence. The second step concerns, in fact, the choice of case studies at different scales of analysis and types of contexts. This aspect is emphasized because on the Italian territory and especially in regions affected by calamitous events such as earthquakes for the past 15 years, the orography of the territories plays a role of primary importance. The third block of the methodology concerns the implementation from the software perspective of an algorithmic simulation ecosystem to which the logic of agent-based modelling can be applied. Then (fourth step), based on the results of condition-driven simulations capable of realistically defining different crisis scenarios (depending on the level of depth and logical implementation of the components of the simulation itself such as, in this case, people trying to reach safe areas following the disaster event), we move on to the implementation of urban design techniques capable of collecting and translating the results themselves into design proposals. The final step is to define new models and forms of resilient urban spaces with resources available for each type of user. These updated models are then fed back into the simulation environment so that they can be validated and tested before their actual implementation using the same boundary conditions, thus moving away from the purely self-justifying logic of the project itself given their data-driven nature. Partial results of the research can be traced to an initial experimentation of agent-based simulation, carried out through a specially compiled script, in a pilot urban context. This allows testing potentials and limitations of the methodology from both computational and operational perspectives. From a purely practical point of view, the results consist of the development of a three-dimensional virtual simulation environment capable of generating georeferenced maps (aimed to optimize urban shapes towards safety), data and diagrams according to the set boundary conditions. Figure 1 and 2 show the first test carried out on the “Coppito Campus” of the University of L’Aquila. The model is compiled considering the actual maximum capacity of the four main buildings and the actual location of the gathering areas, to be used in case of an actual evacuation. The capacity of the latter is defined by calculating the actual available area considering an occupancy of 2.5 square meters per person. The accuracy of the model is ensured by the ability of the simulation environment itself to handle georeferenced data (on which classical analyses proper to urban planning practice are based). The next steps in the research, also thanks to an ongoing collaboration between the working group and the Civil Protection Corps, concern the scaling up of the simulation model to the urban level, to validate the actual potential that such a tool can have in real-life risk situations.