Engineering Safety : With applications to fire safety design of buildings and road tunnels

Abstract

A continuously changing and increasingly complex society leads to new challenges in safety design. Modern buildings and road tunnels are being packed with new technology that creates new failure modes, multiple subsystem interactions and tight couplings between different socio-technical systems. Meanwhile, safety is largely designed into these systems using prescriptive design rules that have evolved through reactions to accidents in systems with limited resemblance to modern systems. The traditional prescriptive approach to safety design was developed to avoid the re-occurrence of previously experienced accidents. New types of systems and accidents need a different design philosophy. The focus should be on the future instead of the past. Hence, the following question was outlined as the major issue of this thesis: what promotes and inhibits performance-based safety management of design processes? Performance-based design principles and regulations are nothing new. In Norway, performance-based fire safety legislations were introduced in the onshore building industry in 1997, and the international fire safety science community had a great focus on promoting these issues during the 1990s. However, experience with the performance-based legislation regime shows that the majority of fire safety designing activity is still based on prescriptive design rules, even in the most novel and complex cases. This is an unfortunate practice, considering that the prescriptive design rules have a boundary of validity associated with historically appropriate designs. Another matter is the restricted empirical foundation for the prescriptive design rules. Accidents are relatively rare events in socio-technical systems. Hence, the ‘test of time’ is a rather weak test in terms of determining the appropriateness of the prescriptive design rules. Strengthening the performance-based alternative to safety management of design process is thus of major importance. Four research questions were developed to support the major issue. The research questions were associated with: (1) understanding current fire safety engineering practice, (2) investigating the scientific foundation of the concepts of fire safety level and safety margin, (3) investigating methodological challenges associated with current practice, and (4) SUMMARY VI transforming the understanding associated with current challenges into proposals for improvement. The research was limited to issues associated with engineering practice, safety science and safety regulation, explored through six case studies: A. A study of fire safety engineering practice in Norway in the period from 1997 to 2012. B. A study of fire safety science’s treatment of major concepts associated with the measurement of safety levels and safety margins. C. A study of 40 different technical fire safety strategies (combinations of safety measures) for multi-story residential apartment buildings. D. A study of the application of an engineering methodology to a specific design example: a concert hall. E. A study of the risk analyses and uncertainty management process in the Rogfast road tunnel project. F. A study of the application of a Bayesian Network model for risk analysis in road tunnels generally and in the Rogfast tunnel specifically. The data the case studies dealt with has mainly been written documents, either collected from the different projects or through literature surveys associated with the topic. Documents have been analyzed using qualitative text analyses, except for case studies C, D and F, which also include quantitative risk and fire modeling approaches. The major finding of the project is that there is a mismatch between current fire safety engineering practices and fire safety science. Fire safety engineering practice builds largely on the application of prescriptive design rules. Deviations from these design rules are often made, and the consequences of these deviations are often documented qualitatively using engineering judgment and argumentation. Fire safety science, on the other hand, builds on a rather narrow scientific framework, greatly inspired by the natural sciences. Fire safety is preferably measured by the application of quantitative relationships and models. The type of qualitative knowledge reflected by the fire safety engineering practice is poorly reflected in fire safety science, and the type of quantitative rigor reflected in fire safety SUMMARY VII science is poorly reflected in fire safety engineering practice. Obviously there is a need to increase the common understanding. I argue that the scientific framework for fire safety science is too narrow to capture the essence of the concept of fire safety. The traditional framework builds on scientific reductionism, which leads to great simplifications in the treatment of systems and environmental complexity and excludes critical issues that are difficult to quantify dependably. Examples of the latter are human and organizational behavior. Similar conclusions are drawn with regards to the risk concept from the Rogfast cases. Overemphasis on model concepts, such as relative frequencies or universal causal structures, excludes the individual knowledge safety experts may bring to the table in novel designs. An alternative scientific framework is suggested, which builds on a constructivist systems thinking perspective. A fundamental assumption is that complex socio-technical systems, such as certain modern buildings and road tunnels, are modeled as social hierarchies. The macro-level includes social institutions, such as national safety authorities and fire departments, while the micro-levels include the building’s components, sub-systems, and nuts and bolts. Fire safety, then, is a property of the system as a whole and cannot be associated with any lower layer in the hierarchy, for instance by only considering the technical infrastructure or the reliability of an automatic sprinkler system. Moreover, complex socio-technical systems are constantly adapting to changes within themselves and in the environment. Hence, safety design is a matter of creating a control structure that enables the system to change in a safe manner. Application of the proposed framework would lead to a more holistic approach to safety design, regardless if one applies a risk-based methodology or a systems safety methodology. For instance, it would broaden the view on what knowledge is relevant in design processes and what measures could be used to achieve safety. Knowledge associated with the individual engineer’s experience would become more important. This knowledge may be tacitly known, and works, for instance, in terms of how the engineer creatively frames and reframes design problems to the stakeholders’ needs. A holistic perspective on safety measures includes, in principle, all thinkable measures, SUMMARY VIII and not only those measures associated with quantitative knowledge. A consequence of this would be that mathematical rigor would have to give way to more qualitative and discursive decision processes. Alternative processes and supplementing methods to traditional quantitative modeling and analysis for determining quality and coherence of the documentation would have to be developed.