Ever since safety started to be investigated in a consistent manner, around 150 years ago, there has been a tremendous improvement, both in our understanding of accident processes, and in reduction of harm and damage caused by these occupational and major accidents. Major improvements in safety theories, models and metaphors were made after World War II, with the late 1970s till the late 1990s as the ‘golden years’. But still these major accidents occur and they will keep prompting future scientific developments in safety, as they have done in the past. Reducing the frequency of major accidents remains challenging. Improving design and automation, as starting point for safety has its limits due to the complexity of processes and the inability to foresee all safety related conflicts. The modern emphasis to assure the capacity to handle unforeseen events, such as resilience promises to deliver, will become even more important in the future. Inherent safe design on the other hand make a sensible approach when designing production processes for emerging and future technologies, like nano- and biotechnology. Also, it will remain difficult for small and medium sized enterprises to adhere to complicated laws and regulations. In addition, an increased participation of stakeholder groups makes future safety decision-making even more challenging than it already is today. Yet we foresee that there may be grounds for change in which safety rules, laws and regulations are set aside, the bureaucratic approach towards safety is stopped and the focus is on dynamic accident processes detection. Today, methods are developed to automatically assess time-dependant advancement of accident scenarios and barrier degradation. This direction will contribute substantially to a future higher level of safety in different industrial sectors and might alleviate the emphasis on bureaucracy. We end with developments in two countries where safety and safety science is emerging.

Objective: What is the influence of general management trends and safety research on managing safety? Method: A literature study which is limited to original English and Dutch books, documents, and articles in relevant scientific journals, for the period 1988–2010. Results and conclusions: Safety science does not yet have a unifying theory, which betrays its young age as a scientific discipline. In the period concerned, well-known theories, models and metaphors are established or re-issued, including the High Reliability Theory, the Man-Made Disasters and the corresponding Disaster Incubation Theory, and the Normal Accident Theory. The Swiss cheese metaphor takes its final form, the bowtie metaphor and the Drift into Danger model are published. All these theories, models and metaphors emphasize organisational aspects of major accidents in high-tech-high-hazard sectors. General management trends highlight the importance of external stakeholders, which are only reflected in the Drift into Danger metaphor. These developments must be considered in the context of a dynamic influence of external factors, like a decrease in government influence coinciding with strong market and technology developments, which can conflict with safety requirements for high-tech-high-hazard companies. Organisational/safety culture and risk/safety management systems take off during this period, both in terms of academic research and consultancy activities for companies. Whether these concepts will have a lasting influence on safety levels in companies is yet to be seen, given the unclear relationship with major accident processes. Research findings show that many companies suffer from sloppy management, having only a limited insight into possible disaster scenarios.

Research question: What is the influence of general management trends and research into causes of accidents on safety management? Method: The literature study is limited to English and Dutch books, documents and articles in the scientific, professional, and technical literature from the period 1988–2010. Results and conclusions: Quite some developments occurred in the occupational safety domain. During the period concerned three models are developed, the Dutch Tripod Model, the Swedish Occupational Risk Unit Model (QARU), and the Dutch Occupational Risk Model (QRM), a barrier based model founded on the bowtie metaphor. These models address occupational accidents from different perspectives, and surprisingly similar factors. While terminology differs, these factors are called basic risk factors, situational, or management factors. Self-regulation of companies has been a strong stimulus for research on safety management systems and audits. Traditionally research in management related topics has not been part of safety research, and thus it has to be developed. While the quality of this type of research is rather low, a general structure of safety management systems is related to the Rhineland management concept. Such evidence is found in new management models such as the EFQM/INK and, to a lesser extent, Corporate Social Responsibility (CSR). While organisational learning, its quality and effectiveness on occupational safety is not researched in this period, research interests are focussing on other organisational aspects like safety culture and climate, including a renewed interest in human behaviour.

In facilities handling combustible dusts, the isolation of propagating deflagrations requires great attention due to the potential catastrophic consequences of secondary dust explosions. While the ability of dust explosions to propagate is widely recognized, some misconceptions still exist. One of the common myths is that a dust explosion cannot propagate through small diameter pipes and that explosion isolation may not be required in that case. This article first presents a simplified theory of flame propagation in pipes. Dust explosion experiments performed in industrial-scale pipes smaller or equal to 4 in (or 100 mm) in diameter are then reviewed. The findings of the experiments are interpreted in the light of the simplified theory. Our study reveals that dust explosion propagation has been consistently observed in pipes with a diameter as small as 1 in. While the likelihood of flame propagation seems to decrease with pipe diameter and other “chemical” and “engineering” factors, it remains a realistic scenario and therefore should be addressed in the design and operation of powder handling systems.

Many industrial processes handle, use, or produce metallic particles small enough to explode in air, thus posing severe explosion hazards. Finishing operations, for example, create very fine particles and have been involved in a growing number of accidents in recent years. New emerging processes, such as 3D printing, are being rapidly developed and directly use micrometric particles to create complete objects by welding layers of material together. Finely divided metals also enter into the composition of plastics, rubber, fibers, paints, coatings, inks, pesticides, detergents, and even drugs; additionally, they are used as catalysts for major industrial chemical reactions, and are even being explored as possible clean alternatives to fossil fuels. Metal dusts are of special concern due to their peculiar combustion properties, including their higher heat of combustion and pyrophoric nature,. As a result, metal dusts explosions are often much more devastating than explosions involving organic materials. Additionally, due to their high reactivity, many fine and most ultra-fine metal powders can burn in carbon dioxide, water vapor and even nitrogen. Whereas preventive measures may reduce explosion risks efficiently, they rarely are sufficient to eliminate explosions completely, especially when dealing with highly reactive metallic particles. Therefore explosion protection measures usually also need to be considered. The high energetic content of metal dusts poses new challenges to conventional explosion protection systems in terms of robustness and response time. This paper reviews the special hazards of metal dusts and presents the state-of-the-art in terms of explosion protection.

The Chemical-Engineering model describes how the synergetic combination of chemical and engineering factors determines gas, particulate, and hybrid explosion behavior. Explosion behavior is expressed as explosion sensitivity (e.g. probability of explosion), explosion severity (e.g. adverse effects of explosion), and explosion types (deflagration or detonation). Chemical factors are ranked using NFPA methodology. Engineering includes whether explosion occurs in open or enclosed spaces, equipment shape/size, and initial T, p, and flow conditions. The model is semi-quantitative; its virtues are its unique integration of chemical and engineering factors in determining explosion risk, its didactic qualities, provision of insight, and practical utility to engineers who are not experts in combustion science.

Objective: Which general management and safety models and theories trends influenced safety management in the period between Three Mile Island in 1979 and Piper Alpha in 1988? In which context did these developments took place and how did this influence Dutch safety domain? Method: The literature study was limited to original English and Dutch documents and articles in scientific and professional literature during the period studied. Results and conclusions: Models and theories of human errors, explaining occupational accidents were still popular in the professional literature. A system approach was introduced into mainstream safety science, starting in process safety, and subsequently moving into occupational safety. Accidents were thought to be the result of disturbances in a dynamic system, a socio-technical system, rather than just human error. Human errors were also perceived differently: they were no longer faults of people, but consequences of suboptimal interactions during process disturbances. In this period quality of safety research increased substantially, also in the Netherlands. Major disasters in the 1980s generated knowledge on process safety, and soon process safety outplaced developments in occupational safety, which had been leading before. Theories and models in this period had advanced sufficiently to explain disasters, but were still unable to predict probabilities and scenarios of future disasters. In the 1980s ‘latent errors’ appeared in safety literature, and in The Netherlands the concept of ‘impossible accidents’ appeared. Safety management was strongly influenced by developments in quality management.

Metal dust deflagrations have become increasingly common in recent years. They are also more devastating than deflagrations involving organic materials, owing to metals' higher heat of combustion, rate of pressure rise, explosion pressure and flame temperature. Aluminum finishing operations offer a particularly significant hazard from the very small and reactive aluminum particles generated, and thus require high attention to details of operation and explosion safety management. This paper presents available statistics on metal dust explosions and studies the specific explosion hazards of aluminum finishing operations. The analysis of seven case studies shows that the proper design, monitoring and maintenance of dust collection systems are particularly important. Furthermore, the isolation of deflagrations occurring in dust collection systems, as well as good housekeeping practices in buildings, are critical safeguards to avoid the occurrence of catastrophic secondary explosions.

Dust explosibility is traditionally described by two parameters, namely the maximum explosion pressure, P_max, and the deflagration index, K_St, usually determined through testing in a closed, pressure-resistant spherical vessel, either 20 L or 1 m³ in volume. These parameters constitute key variables in the design of explosion protection systems, such as venting, suppression or isolation systems. The potential for overdriving dust combustion with pyrotechnical igniters in the 20-l sphere has been recognized, discussed and analyzed for many years, notably in the determination of the minimum explosible and limiting oxygen concentrations, which has led to specific guidelines regarding the ignition source strength in ASTM standards. The current paper presents new experimental evidence that the energy provided by pyrotechnical igniters may, in some instances, physically alter the dust being tested in the 20-l sphere. K_St values can be several times greater in the small vessel compared to those measured in the 1-m³ chamber. Further visual evidence is provided to show that high energy ignition can produce a turbulent flame region, possibly consisting of a hybrid mixture of flammable gas (or vapor) and dust, which can propagate faster than the corresponding pure dust. The experiments suggest that K_St values measured in the 20-l sphere may no longer be representative of a dust deflagration in a real process environment. We recommend additional tests in a 1-m³ chamber when a dust exhibits a low flash point, or when it's K_St is above 300 bar m/s in the 20-l sphere.

Objective: What influence has research conducted by general management schools and safety research had upon the causes of accidents and disasters in relation to the managing of safety between 1970 and 1979? Method: The study was confined to original articles and documents, written in English or Dutch from the period under consideration. For the Netherlands, the professional journal De Veiligheid (Safety) was consulted. Results and conclusions: Dominant management approaches started with (1) classical management starting from the 19th century incorporating as a main component scientific management from the early 20th century. The interwar period saw the rise of (2) behavioural management which was based on behaviourism, this was followed by (3) quantitative management from the Second World War onwards. After the war it was (4) modern management that became important. A company was seen as an open system, interacting with an external environment with external stakeholders. These management schools of thought were not exclusive, but existed side by side in the period under consideration.Early in the 20th century, it was the U.S. 'Safety First' movement that marked the starting point of this knowledge development in the sphere of safety managing, with cost reduction and production efficiency as the key drivers. Psychological models and metaphors were used to explain accidents resulting from 'unsafe acts'. Safety was managed by training and targeting reckless workers, all in line with scientific management. Supported by behavioural management, this approach remained dominant for many years until long after World War II.Influenced by quantitative management, potential and actual disasters occurring after the war led to two approaches; loss prevention (up-scaling in the process industry) and reliability engineering (inherently dangerous processes in the aerospace and nuclear sectors). The distinction between process safety and occupational safety became clear after the war when the two evolved as relatively independent domains.In occupational safety in the 1970s human error was thought to be symptomatic of mismanagement. The term 'safety management' was introduced to scientific safety literature alongside concepts such as loosely and tightly coupled processes, organizational culture, disaster incubation and the notion of mechanisms blinding organizations to portents of disaster scenarios. Loss prevention remained technically oriented. Until 1979 there was no clear link with safety management. Reliability engineering that was based on systems theory did have such a connection with the MORT technique that served as a management audit. The Netherlands mainly followed Anglo-Saxon developments. In the late 1970s, following international safety symposia in The Hague and Delft, independent research finally began in the Netherlands.

Serious incidents in the 1970s and continuous growth of factories producing and/or using hazardous substances formed the basis of a quantitative approach to risk. While discussions of risk were conducted in all industrialised countries they were particularly important in The Netherlands due to space limitations and short distances between industrial plants and residential areas. This article is part of a series covering the history of the safety science discipline (Swuste et al., 2015; Van Gulijk et al., 2009; Swuste et al., 2010). The concept risk entered the Dutch safety domain before the 1970s in relatively isolated case studies and in managing flood defences in The Netherlands. Since the 1970s these case studies paved the way for the development of mathematical models for quantitative risk analysis that were based on experience from nuclear power plants, the process industries and reliability engineering from operations research. 'External safety' was a focal point for these early developments in the process industries: adverse effects of dangerous goods outside the factory's property boundaries. The models were documented in standardised textbooks for risk analysis in The Netherlands, the so-called 'coloured books'. These works contributed to the development of the Seveso Directive. For internal safety (taking place within property boundaries) semi-quantitative approaches were developed simultaneously. The models for quantitative risk analysis were deemed reliable, but the acceptability of a quantified risk was another matter. Making decisions on risk relates to complex societal issues, such as ethics, stakeholder perception of risks, stakeholder involvement, and politics, all of which made the decision making process far from straightforward. With the introduction of the abstract concept of risk in the Dutch safety science domain, the question of risk perception became important in Dutch safety research. The concept risk and methods for quantitative risk analysis first entered into Dutch law in environmental risk regulations. It took a while for risk to be accepted by occupational safety experts, but just before the turn of the century 'occupational risk inventory and evaluations' or RI&E methods were introduced into Dutch occupational safety legislation. This finalised the paradigm shift to risk-based safety-decision making in the Dutch safety science domain. While methods for quantifying risk are now widely applied and accepted, the proper use of risk perception and risk in the political decision process are still being debated.