Adaptive Fire Testing: A new foundation stone for fire safety

Informations projet

AFireTest

N° de convention de subvention: 101075556

DOI

10.3030/101075556

Date de signature de la CE 8 Decembre 2022

Date de début 1 Septembre 2023

Date de fin 31 Août 2028

Financé au titre de

• European Research Council (ERC)

Coût total

€ 1 500 000,00

Contribution de l’UE

€ 1 500 000,00

Coordonné par

UNIVERSITEIT GENT
Belgium

Periodic Reporting for period 1 - AFireTest (Adaptive Fire Testing: A new foundation stone for fire safety)

Période du rapport: 2023-09-01 au 2026-02-28

Résumé du contexte et des objectifs généraux du projet

Current fire safety requirements rely on a limited set of standardized fire tests. These tests do not give as a complete picture of how products and buildings behave in real fires. Fire safety in our buildings is really achieved by (i) learning from disasters; (ii) active fire and rescue service intervention; (iii) improvements in other domains such as reliability of electrical appliances; and (iv) a large degree of overinvestment, where expensive safety measures are required also where they provide limited benefit. This approach to fire safety breaks down when new materials and products are introduced in society.

Within the AFireTest project, we fundamentally rethink how fire performance is tested, how fire safety is demonstrated for individual buildings, and how authorities can adjust their regulations accordingly. Instead of relying on a fixed set of standard tests, the project introduces the idea of Adaptive Fire Testing: fire tests are specified in order to obtain a real understanding of a products behaviour in fire. We develop a methodology where test protocols are specified so that we learn as much as possible from each test. The project uses glazing as a case study, as glazing behaviour in fire determines the oxygen supply to the fire, but is still very poorly understood.
The new testing approach we develop relies on probabilistic calculations where we evaluate a very large number of possible scenarios. This quickly becomes computationally very demanding. The standard response is to adopt machine learning, but such models often defy physical laws. In safety critical environments, such as fire safety, we want to be able to trust in realism of our computer simulations. Therefore, within the AFireTest project we invest in physics-informed surrogate modelling: machine learning approaches where we are certain that the results align with physical laws.

Beyond technical innovation, AFireTest also recognizes that fire safety decisions have legal, economic, and societal dimensions. Currently, fire safety requirements are often updated following public outcry after major disasters. Such approach almost unavoidable results in avoidable societal losses (lives lost, resources spent on safety measures that do not help...). Within the AFireTest project, we develop a framework for evaluating the costs and benefits of regulatory proposals regarding fire safety. Adopting such framework will transform discussions on fire safety regulations from a meeting of subjective opinions to a fact-based argumentation on input values and model assumptions.

Travail effectué depuis le début du projet jusqu’à la fin de la période considérée dans le rapport et principaux résultats atteints jusqu’à présent

A central achievement of the project has been the development of a scientific framework to quantify the information gain of fire tests. This framework makes it possible to estimate in advance how much a specific test is expected to improve understanding of fire behaviour. The framework provides a common basis for comparing alternative test setups and supports more efficient use of experimental resources. It also forms the backbone of the project’s broader concept of Adaptive Fire Testing.

The case study on glazing is proving very challenging. Our investigations have uncovered important shortcomings in current testing practice. Current approaches determining glass strength are biased, and the glazing temperature measurements in scientific literature inaccurate temperature are unreliable (they are very much dependent on test details that are often not reported). We have developed an approach which removes the bias in glazing strength evaluation, and are investing heavily in addressing the issue of glazing temperature evaluation. If successful, this will be a major technical breakthrough.

A review of the state-of-the-art on physics-informed surrogate models for fire safety analysis has been completed. It is clear that this research area is still young, but is developing fast. A key shortcoming we have identified is that there is no common approach to confirm or quantify the degree with which models comply with physical constraints. Ongoing investigations are focussing on developing such performance metrics.

Foundational work on fire risk characterization has been carried out in collaboration with visiting researchers. This includes a conceptualization of different areas of technical knowledge, ranging from areas where prescriptive fire safety is appropriate, to areas where even our most advanced engineering tools are lacking. Such conceptualization should help engineers (and authorities) to recognize which fire safety approaches can be used for which types of buildings – and which designs are beyond our current state-of-knowledge. A second focus has been on applying systems thinking and system dynamics approaches to fire safety. System dynamics methods are found to be a very promising approach to evaluate fire risk in designs with a large number of interconnected parts.

Finally, the project has developed a conceptual framework for evaluating fire safety approaches from a law and economics perspective. This work explores how different legal frameworks for fire safety result in higher/lower costs for society.

In summary, at the time of writing (early 2026) the conceptual groundwork for AFireTest has been largely achieved. Major research challenges remain regarding operationalization with case studies.

Progrès au-delà de l’état des connaissances et impact potentiel prévu (y compris l’impact socio-économique et les conséquences sociétales plus larges du projet jusqu’à présent)

The project introduces a new way of thinking about fire testing by focusing on the value of information gained from each test. This approach has the potential to significantly reduce unnecessary or repetitive testing, saving time, money, and resources while maintaining or improving safety. It also allows experts to judge whether small changes in a product or design truly require a new fire test, rather than automatically triggering additional testing as is often the case today. If adopted, this will make fire testing more flexible, more transparent, and better aligned with real safety needs. It will also provide a large competitive advantage to companies adopting the approach for their R&D needs. Further demonstration through practical case studies is needed however to support widespread adoption. Once achieved, governmental buy-in will be required for uptake in the European market.

The project’s work on advanced modelling techniques opens opportunities for faster and more informed fire safety assessments. By combining physical fire science with modern data-driven methods, these models can support building-specific evaluations of fire risk that would otherwise be too slow or costly to perform. If successfully operationalized, this can complete remove the need to specify rigid fire safety rules, while at the same time resulting in an unambiguous increase in fire safety for all. For this potential to be realized, the simple test cases being developed now need to be expanded to larger and more challenging designs. It is likely that multi-disciplinary collaborations will be needed to fully achieve this.

The evaluation of costs and benefits of fire safety regulation provides a route for the objective comparison of regulatory alternatives on fire safety rules and compliance mechanisms. For these ideas to have impact, engagement with policymakers, standardization bodies, and professional organizations will be necessary. To enable this engagement, we are working on real-life demonstration cases.

AFireTest vision for fire safety compliance