In this paper we present an agent-based model which reproduces transport choices of a sample of 5000 citizens of the city of Varese (northern Italy) and the corresponding PM emissions of their daily commutes. The aim of the model is testing the impact of public policies willing to foster commuting choices with lower PM emissions. Agent-based models are simulation techniques that allow to represent and analyze scenarios containing features of complexity, with heterogeneous agents, complex interactions among them and between them and the environment. In our model the agents represent the commuters. The model considers one main process, the commuters‘ decision about what means of transport to utilize. The agents start with a set of preferences—one for each mode of transport (private motorized vehicle, private non-motorized vehicle, public transport)—that have been assigned to them. Throughout the process, these preferences are influenced by the relative price of the different means of transport, by social influence and by the intensity of the policies applied. Each agent is embedded in a social network: neighborhoods are formed according to the closeness of the initial preferences of the agents. Each means of transport has an absolute cost per kilometer, a relative cost with respect to the other means, an average PM emission per kilometer and an environmental index which is proportional to the emissions produced by each mean of transport for an average-length commute. The agent decides its means of transportation on the basis of two factors: total need satisfaction and uncertainty. In turn, the total need satisfaction is composed by the sum of personal and social needs, divided by the relative price of the means of transport. The personal need satisfaction is made by the relation between the past and present transport choice. The social need satisfaction is made of the proportion of members of the agent’s network using the same means of transport of the agent. The uncertainty consists of the variation over time of agents’ satisfaction. Two kinds of policies are implemented in order to give incentives to the agents to shift to means of transportation with lower PM emissions. The first is market-based and is represented by a parameter that increases the prices of the means of transportation, with a larger increase of the most polluting ones. The second policy is preference-based and is represented by a parameter that increases the preference for less polluting means of transportation independently from their price. This parameter may either affect all agents in the same way or may target agents choosing more polluting means of transportation. The intensity of both policies is decided by the modeler, who, moreover, may test each policy alone, or different combinations of the two. We model two scenarios. In the first one the initial preferences for each modes of transport which are assigned to the agents are derived from distributions inspired to the data from the Italian National Institute of Statistics 2011. In the second one the initial preferences are derived from normal distributions. We utilize the opensource software NetLogo. Preliminary results suggest that price-based policies are more effective than preference-based policies for the generation of more environmental friendly behaviours. However price-based policies are usually more controversial to implement from a political point of view. The aim of the work is to identify the best combination of the two kinds of policies, considering also the possibility of targeting particularly agents with the worst environmental performance in the starting scenario.
An application of agent-based models on urban passenger mobility: the case study of Varese
MAGGI, ELENA;VALLINO, ELENA
2016-01-01
Abstract
In this paper we present an agent-based model which reproduces transport choices of a sample of 5000 citizens of the city of Varese (northern Italy) and the corresponding PM emissions of their daily commutes. The aim of the model is testing the impact of public policies willing to foster commuting choices with lower PM emissions. Agent-based models are simulation techniques that allow to represent and analyze scenarios containing features of complexity, with heterogeneous agents, complex interactions among them and between them and the environment. In our model the agents represent the commuters. The model considers one main process, the commuters‘ decision about what means of transport to utilize. The agents start with a set of preferences—one for each mode of transport (private motorized vehicle, private non-motorized vehicle, public transport)—that have been assigned to them. Throughout the process, these preferences are influenced by the relative price of the different means of transport, by social influence and by the intensity of the policies applied. Each agent is embedded in a social network: neighborhoods are formed according to the closeness of the initial preferences of the agents. Each means of transport has an absolute cost per kilometer, a relative cost with respect to the other means, an average PM emission per kilometer and an environmental index which is proportional to the emissions produced by each mean of transport for an average-length commute. The agent decides its means of transportation on the basis of two factors: total need satisfaction and uncertainty. In turn, the total need satisfaction is composed by the sum of personal and social needs, divided by the relative price of the means of transport. The personal need satisfaction is made by the relation between the past and present transport choice. The social need satisfaction is made of the proportion of members of the agent’s network using the same means of transport of the agent. The uncertainty consists of the variation over time of agents’ satisfaction. Two kinds of policies are implemented in order to give incentives to the agents to shift to means of transportation with lower PM emissions. The first is market-based and is represented by a parameter that increases the prices of the means of transportation, with a larger increase of the most polluting ones. The second policy is preference-based and is represented by a parameter that increases the preference for less polluting means of transportation independently from their price. This parameter may either affect all agents in the same way or may target agents choosing more polluting means of transportation. The intensity of both policies is decided by the modeler, who, moreover, may test each policy alone, or different combinations of the two. We model two scenarios. In the first one the initial preferences for each modes of transport which are assigned to the agents are derived from distributions inspired to the data from the Italian National Institute of Statistics 2011. In the second one the initial preferences are derived from normal distributions. We utilize the opensource software NetLogo. Preliminary results suggest that price-based policies are more effective than preference-based policies for the generation of more environmental friendly behaviours. However price-based policies are usually more controversial to implement from a political point of view. The aim of the work is to identify the best combination of the two kinds of policies, considering also the possibility of targeting particularly agents with the worst environmental performance in the starting scenario.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
