Improving Safety and Security using Bayesian Belief Networks
What is the project about?
Bayesian belief networks (BBNs) are frameworks to model safety and security related problems within the public and private sectors. It involves the identification of causal factors that may lead to economic damage and loss of life within these sectors. These cause–consequence chains are typically described in terms of probabilistic formulations to capture uncertainties in these chains. The framework allows for optimizing over all possible decision variables in order to improve safety and security levels. It also allows for updating the uncertainties, by the evidence propagation process, as soon as new data are observed or elicited from expert judgments.
What problems do these projects intend to solve?
Policymakers and decision makers face many challenges in making difficult investment decisions in order to reduce the risks in societies, companies and businesses. This requires a thorough understanding of all factors that influence these risks, not only technical factors, but also human-related factors and management factors, as well as their interrelationships. The effectivity of these factors on the resulting risk level is typically probabilistic in nature. BBN’s provide a suitable framework to model these probabilistic interdependencies in order to derive optimal decision rules for problem owners, such as managers of electricity, rail, road, aviation, chemical and supply chain networks.
What research does TPM do and how does it impact society?
TPM has carried out a number of projects on BBNs that have not only led to new scientific insights but have also had a considerable social impact. An example is the PLATYPUS project, in which an integrated model for risk in a real-time environment for the hydrocarbon industry was built for Shell. Here, BBN is used to calculates the frequency of loss of containment accidents for an existing or a proposed plant.
Another project is Risk Analysis of Infrastructure Networks (RAIN) in response to extreme weather. It will aid decision making for infrastructural managers in improving the vulnerable spots within their networks (road, rail, electricity and telecommunication). Technical and logistic solutions are being developed, which will include novel early warning systems, decision support tools and engineering solutions to ensure the rapid reinstatement of the infrastructure network. Various European partners are involved: four universities as well as consultancy businesses and research institutes.
In the Causal model for Air Transport Safety (CATS) project, a strategic perspective of aviation risk is provided that can be used as input into the policy formulation process. This project is carried out in cooperation with Ministry of Environment and Infrastructure, NLR (Netherlands Aerospace Centre) and safety consultancy businesses. The Ministry intends to use the model as a support tool for implementing a regulatory risk-based safety oversight system.
BBN modelling is also used to increase effective border controls in the interests of public safety by improving the exchange of data between the various partners in the chain of international trade. It makes inspections more effective and vastly improves the safety, security and efficiency of freight and cargo transport. The Dutch government intends to promote internationally the innovation knowledge of TU Delft and its partners in the area of trade facilitation. SAtIN is a joint project on Supply Chain Control and Compliance involving TU Eindhoven, business partners ASML, IBM, Philips, Seacon Logistics and research partners TNO and CWI.
‘I want to contribute to a safer society by building and implementing better models that understand multiple causes that lead to unwanted incidents.’
|Coverage at the Dies Natalis public lecture on safety and security of TU Delft in 2014.|
Intention of the Ministry of the Environment and Infrastructure to use the CATS model as a support tool for implementing a regulatory risk-based safety oversight system.
Equipping managers with Bayesian decision support tools to improve the safety and security levels within their organisations.
Who is involved?
|Department of Values, Technology and Innovation|
Pieter van Gelder, Noel van Erp, Genserik Reniers, Nima Khakzad, Yaohua Tan, Ben Ale (Emeritus Professor), Simone Sillem
How is the project funded?
NWO and EU, the Ministry Infrastructure and the Environment, and Shell.