OPTIMIZATION OF TERMINAL LAYOUTS: AN ANALYTICAL AND SIMULATIVE APPROACH BASED ON GENETIC ALGORITHMS

Every day millions of pedestrian move with different needs and objectives through spaces each of them with its functional specifications. An accurate design or revisiting of transport terminals, as for example railway stations, underway stations, airports, as well as complex buildings, open spaces and a deep analysis of public events with relevant pedestrian flows, would improve its usability at users benefit. To reach this goal is necessary a careful integration among architecture, engineering needs and transport disciplines, that, starting from the study of users behavior and pedestrian dynamics, provides the fundamental elements to be considered during design stage to ensure a major level of service. In literature nothing much is known about the optimal dimension of pedestrian transportation terminals. The aim of this study is to develop a methodology to size the functional terminal layouts, by the integration of analytical and simulative models submitted to generic algorithms, taking into account the dynamics and flows generated inside the terminals. In order to obviate the lack of requisite data for models calibration, validation and verification, as well as testing the process developed, an algorithm for data acquisition has been elaborated. It has a dedicated graphic interface, which allows to reveal the pedestrian dynamics and consequently to generate database; with these data is possible to obtain statistical and behavioral indicators about pedestrians detected. The use of analytical models, both to define the sizing of facilities inside the terminals and to model the user behavior during their paths, allows to define an objective function able to represent the performances of the terminal functional layout. Defined the dimensional ranges of each functional element inside the layout according a specific Level of Service, performed a design of experiments methodology and applied genetic algorithms to minimize the objective function, it is possible to obtain a set of optimal solutions for the terminal configuration sizing, in coherence with flows and dynamics generated inside the terminals itself. A further simulative approach, based on the application of the social force algorithm, allows, through quantitative and qualitative parameters, to identify the best solution(s) inside the domain previously identified with genetic algorithm application. Starting from the motivation that inspired this work, analyzed the existing literature and the main methods for data acquisition, it will be introduced the algorithm for the automatic acquisition of data and pedestrian database generation. The application of this tool will be illustrated in order to manifest the potentiality of the instrument same. Subsequently introduced the tool developed for the definition of the characteristic elements sizing and the model chosen for the correct estimation of pedestrian travel times, it will be explored the structure of the objective function aimed to identify the right trade-off between infrastructure and pedestrian costs. Finally, the application of genetic algorithms, resulting in the identification of Pareto front, generates the domain of optimal solutions to sift through the simulation approach. The developed methodology reveals a flexible and simple instruments, but, at the same time, accurate in the resolution of the problems for which has been structured. The potential of the developed methodology is highlighted in the course of the work thanks to a case of study.