Development and evaluation of a predictive fire risk modelling and notification system for wooden homes
Original version
Strand, R. D. (2024). Development and evaluation of a predictive fire risk modelling and notification system for wooden homes [Doctoral dissertation, Western Norway University of Applied Sciences]. HVL Open.Abstract
Accidents involving fire result in more than 300,000 deaths annually, with statistics suggesting residential fires causing above 80 % of the fire-related deaths. While these numbers differ somewhat in their statistical foundations, they succeed in expressing the imminent risk of residential fires. Still, the vast majority of fire risk modelling relates not to residential fire risk but to other areas, like forest fire risk or fire and explosion risk in the process industry. In January 2014, in Lærdal, Norway, the fire safety community was reminded of the devastating potential of conflagration events when a fire developing between wooden homes destroyed more than 40 structures. Any increased fire risk was not identified at the time of the fire, highlighting the need for novel fire risk models capable of predicting the imminent fire risk for wooden homes and possible conflagration events.
In recent years, the growing network of Internet of Things (IoT) appliances combined with the pervasive presence of cloud-based data services has provided access to a vast amount of data sources. In turn, it has increased the availability and amount of location-specific and predicted data, enabling dynamic risk assessment through data-driven applications. This thesis addresses the fire risk associated with wooden homes in cold climate regions by investigating the intersection of software engineering and fire risk modelling. The focus has been on developing and evaluating a fire risk notification software system for wooden home fire risk in cold climate regions.
A wooden home fire risk model based on first principle mathematics and physics has been further developed, emphasising the modelling concept and a justified generic modelling approach for wooden homes. A practicality assessment included the evaluation of available data sources, model output and frequency of high-risk events/days. In general, a generic approach was justified, yielding a manageable number of days with high conflagration risk, i.e., in-home drought coexisting with strong winds. Evaluating model practicability involved implementing a prototype fire risk notification system, revealing modest storage and computational needs. The prototype was implemented in conjunction with the development of a formal Coloured Petri Net (CPN) model of the software system architecture. This resulted in a formal specification of the system where specific behaviours were verified through state space analysis. Further, the development of a wooden home fire danger index involved designing and evaluating a graphical user interface (GUI) for fire risk communication as a part of a study involving a mobile application using on-device (edge) computing. The application allowed user-driven testing of the GUI among different groups of users. A second version of the application and GUI are currently undergoing prolonged testing within participant fire brigades, with promising feedback so far. The fire risk model was implemented and evaluated, emphasising robust data manipulation and third-party system integration. Overall, the fire risk modelling and developed software systems effectively meet the new regulations for Norwegian fire departments, requiring systems for detecting increased risk levels.
Has parts
R.D. Strand and T. Log. A Cold Climate Wooden Home and Conflagration Danger Index: Justification and Practicability for Norwegian Conditions. Fire, 6(10), Sept. 2023.R.D. Strand, S. Stokkenes, L.M. Kristensen and T. Log. Fire Risk Prediction Using Cloud-based Weather Data Services. Journal of Ubiquitous Systems & Pervasive Networks, 16(1):37–47, Jan. 2021.
R.D. Strand, L.M. Kristensen and L. Petrucci. Development and Verification of a Microservice Architecture for a Fire Risk Notification System. In M. Koutny, R. Bergenthum and G. Ciardo, editors, Transactions on Petri Nets and Other Models of Concurrency XVII, volume 14150 of Lecture Notes in Computer Science, pages 27–53, Berlin, Heidelberg, Nov. 2024. Springer.
R.D. Strand and L.M. Kristensen. An implementation, evaluation and validation of a dynamic fire and conflagration risk indicator for wooden homes. In Proc. of the 15th International Conference on Ambient Systems, Networks and Technologies (ANT), volume 238 of Procedia Computer Science, pages 49–56. Elsevier B.V., 2024.
R.D. Strand, L.M. Kristensen, T. Svendal, E.H. Fisketjøn and A.T, Hussain. A Mobile Application for Wooden House Fire Risk Notifications Based on Edge Computing. In A. Rocha, H. Adeli, G. Dzemyda, F. Moreira, and V. Colla, editors, Information Systems and Technologies, volume 800 of Lecture Notes in Networks and Systems, pages 238–248, Cham, Switzerland, Feb. 2024. Springer Nature.