Finnish Case Study: Bayesian Network Modelling Workshop FMI Helsinki 16th February 2017 Dr. Julie Clarke Roughan & O’Donovan Innovative Solutions julie.clarke@rod.ie www.rain-project.eu This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no 608166. The contents of this presentation are the author's views. The European Union is not liable for any use that may be made of the information contained therein.
Finnish Case Study Implementation of RAIN risk-based decision making framework Assess the societal, security and economic impacts of critical infrastructure failure Study area: 13 municipalities in the region of Uusimaa
Focus on four types of Critical Infrastructure: Finnish Case Study Focus on four types of Critical Infrastructure: Road Infrastructure
Focus on four types of Critical Infrastructure: Finnish Case Study Focus on four types of Critical Infrastructure: Rail Infrastructure
Focus on four types of Critical Infrastructure: Finnish Case Study Focus on four types of Critical Infrastructure: Energy Infrastructure
Focus on four types of Critical Infrastructure: Finnish Case Study Focus on four types of Critical Infrastructure: Telecommunications Infrastructure
Risks Quantify the impacts of the failures and the quantifiable benefits from a societal, security and economic standpoint of providing resilient infrastructure.
January 2005, Storm Surge Event Coastal flooding Estimated damage of €20 million in costs to SAMPO insurance company
Bayesian Network (BN) Modelling A graphical method that employs Bayesian probability theory to represent complex processes or networks Facilitates the combination of information from various sources
Bayesian Network (BN) Modelling Qualitative component Each node represents a variable in the model. Presence of an edge linking two variables indicates the existence statistical dependence between them.
Bayesian Network (BN) Modelling Quantitative component Necessary to define a probability distribution for each node conditional on its parents. Used to define how strong the relationships are between the variables.
Bayesian Network (BN) Modelling A graphical method that employs Bayesian probability theory to represent complex processes or networks Facilitates the combination of information from various sources Define nodes (random variables) Construct BN model Assign probabilities to each node
Sample BN Model
Sample BN Model
Coordinator: Mr. Peter Prak Working Session Afternoon 13:15 – 15:30 Coordinator: Mr. Peter Prak Objective: To gather stakeholder input to assist in the development of a Bayesian Network (BN) model for the selected case study. Define nodes for BN model (variables) Define the dependence and independence relationships between the variables. Stakeholder opinion will be used to guide the model and to focus on the most important parts.
Working Session Hazards Infrastructure Vulnerability Consequences Source / effects Cascading effects Probabilities Infrastructure Vulnerability Failure modes Inter-network effects Consequences Societal Economic Security
Working Session
RAIN Project www.rain-project.eu Julie Clarke julie.clarke@rod.ie