An agent-based model that simulates how a socio-physical multilayer system recovers after a disaster.
- National Science Foundation 1638311 (CRISP Type 2/Collaborative Research: Critical Transitions in the Resilience and Recovery of Interdependent Social and Physical Networks).
- DIRSIM simulates the post-disaster recovery (PDR) of the social-physical system, enabling us to understand its details and evaluate various policies.
- It is composed of three successive components: (1) network definition; (2) agent interaction modeling; and (3) agent-based simulation.
- Network definition: build the three-layer network using mobility phone location data, Point-of-interest (POI) foot traffic data.
- Agent interaction modeling: specify how agents interact with each other during the PDR process.
- Agent-based simulation: simulate the recovery of each agent in the three-layer network temporally.
An Agent-based Model of Post-disaster Recovery in Multilayer Socio-physical Networks. Jiawei Xue, Sangung Park, Washim Uddin Mondal, Sandro Martinelli Reia, Tong Yao, Satish V. Ukkusuri*. Sustainable Cities and Society. 2024.
- Python 3.6
- Mesa 0.9.0
-
0_data. Released: the shapefile of Texas; the recovery level of physical infrastructures; Not released: mobile phone location data; POI data from SafeGraph (https://www.safegraph.com/).
-
1_data_preprocessing. extract mobile phone location data within the five counties.
-
2_network_constructor. define agents in the three-layer network.
-
Codes in "network_constructor/home/" define nodes and edges in the human layer.
-
Codes in "network_constructor/poi/" define nodes and edges in the social infrastructure layer.
-
3_model. the code under "model/" is our ABM model. It takes the outputs of previous codes and simulates the dynamic of the three-layer system.
-
4_validation. validate the simulation outcome with ground truth from mobile phone location and POI data.
-
5_results. simulate the system recovery under the nine scenarios.
-
6_figures. figures in this GitHub repository.
Overview of used data, three-layer socio-physical network, and the agent-based model (ABM) for post-disaster recovery.
The long-term PDR process. (a, b) The dynamics of recovery levels (i.e. $r_{a}(t)$) for agents representing users and POIs. (c) The flowchart that summarizes the procedure of agent interactions in the PDR process.
MIT license