JID: JPECS

ARTICLE IN PRESS [m5G;October 20, 2017;8:28]

Progress in Energy and Combustion Science 000 (2017) 1 2

Contents lists available at ScienceDirect

Progress in Energy and Combustion Science

journal homepage: www.elsevier.com/locate/pecs

Perspective

Fires and explosions

TagedPTrygve Skjolda,*, Claude Souprayenb, Sergey Dorofeevc
a
TagedP Gexcon, R&D Department, Fantoftvegen 38, 5072 Bergen, Norway
b
Fluidyn France, 7 boulevard de la Liberation, 93200 Saint-Denis, France
c
FM Global, 1151 Boston-Providence Turnpike, Norwood, MA 02062, USA

TagedPKeywords: Fires, Explosions, Safety, Loss prevention, Risk assessment, Scaling, Knowledge gaps

1. Introduction tTagedP hat solve conservation equations for mass, momentum, energy and
chemical species. It is essential for the quality of loss-prevention
TagedPAccidental fires and explosions cause severe losses in society. standards, risk assessments, and hence for loss prevention, safety
Examples of such events range from domestic accidents, such as and security, that engineers understand the underlying assumptions
dwelling fires, to devastating fires in densely populated areas and and inherent limitations of the tools they use, as well as the level of
high-rise buildings, such as the recent fire in the Grenfell Tower accuracy they can expect in the results.
(14 June 2017, about 80 fatalities), and large industrial disasters, TagedPAlthough the governing equations for turbulent fluid flow with
such as Flixborough (1 June 1974, 28 fatalities), Piper Alpha (6 July chemical reactions are well established, there are still many gaps in
1988, 167 fatalities), Buncefield (11 December 2005), Deepwater the fundamental understanding of fire and explosion phenomena.
Horizon (20 April 2010, 11 fatalities) and Fukushima Daiichi (11 This understanding is being explored using analytical methods,
March 2011). Many of the largest property losses in industries over Direct Numerical Simulation (DNS), and high-fidelity Large Eddy
the last century, as well as numerous less devastating accidents Simulation (LES) methods. LES models have gained increasing popu-
associated with the handling, processing and storage of flammable larity in recent years for developing loss prevention solutions and
material, in gaseous, liquid or solid form, involved fires and explo- modelling consequences of fires and explosions. However, within
sions. The oil spill following the Deepwater Horizon disaster and the the context of simulating accident scenarios for Quantitative D8X X D9X X
Risk
release of radioactive material in Fukushima demonstrate the devas- D10XAssessment X D1X X
(QRA), most commercially available Computational
tating effects such events can have on the environment. Common D12XFluid X D13X X
Dynamics (CFD) tools still rely on the Reynolds-averaged
denominators for accidental fire and explosion scenarios are com- Navier Stokes (RANS) equations. The challenges for modelling con-
plex transient fluid flow phenomena, with chemical reactions, cover- sequences of fires and explosions are related to the fact that gener-
ing a wide range of spatial scales. Finally, fire scenarios may escalate ally there is limited time and resources available to produce relevant
to explosion scenarios, and vice versa. simulations for regulatory or design purposes. Therefore, results
from fundamental research must D14X X be integrated in dedicated practical
2. Risk assessment, consequence modelling and scaling numerical models with reasonable performance to account for the
complex coupling between turbulence, chemistry, radiative and con-
TagedPOrganizations that routinely assess the hazard posed by fires and vective heat transfer, and multi-phase effects, such as thermal
explosions, such as facility owners, operators, safety consultants decomposition and phase change (for liquids and solids). On the
and insurance companies, are often involved in the development of other hand, the steady increase in the availability of computing
loss prevention and safety standards. A wide range of methods and power, accompanied by modern parallel numerical solvers and
models have been developed for performing quantitative or qualita- Adaptive Mesh Refinement (AMR), allows for faster calculations on
tive risk analysis as a part of risk management in these organiza- larger computational domains. However, there is still a need for fur-
tions. While the development of robust loss prevention solutions ther speed-up, improved accuracy, and reduced sensitivity of the
and safety standards requires fundamental understanding of fire results with respect to the spatial resolution used in the simulations.
and explosion phenomena, this understanding is also needed for the To this end, fundamental and applied combustion research is a key
development of simplified models applicable for risk assessments for a wide range of applications, from estimating accident initiation
that involve numerous accident scenarios. The types of models used probability (i.e. release, dispersion and ignition), to assessing the
for developing loss-prevention solutions and assessing the conse- consequences of fire and explosion events, and finally to the design
quences of accidental fires and explosions range from analytical and optimization of risk-reducing measures.
expressions and empirical correlations, to phenomenological tools
of varying complexity, and finally to sophisticated numerical models 3. Challenges

*
Corresponding author. TagedPThe primary challenge with respect to predicting the consequen-
E-mail address: [email protected] (T. Skjold). ces of accidental fire and explosion scenarios, including the effect of

http://dx.doi.org/10.1016/j.pecs.2017.09.003
0360-1285/© 2017 Elsevier Ltd. All rights reserved.

Please cite this article as: T. Skjold et al., Fires and explosions, Progress in Energy and Combustion Science (2017), http://dx.doi.org/10.1016/j.
pecs.2017.09.003
JID: JPECS
ARTICLE IN PRESS [m5G;October 20, 2017;8:28]

2 T. Skjold et al. / Progress in Energy and Combustion Science 00 (2017) 1 2

TagedPmitigating measures, is arguably the scaling from controlled experi- cTagedP losures/models, for turbulent burning rates as functions of
ments that can be simulated with reasonable accuracy by LES, and fundamental combustion properties, for all types of fuel-air
sometimes even DNS, to realistic full-scale systems characterised by mixtures.
complex initial and boundary conditions. Both numerical models TagedP Combustion models that are applicable for laminar to turbu-
(LES/RANS) and physics-based correlations are under development lent transition and perform well up to the fully turbulent
for addressing this challenge. Yet, there are important features in regime and with spatially varying mixture stoichiometry.
the underlying physical phenomena in fires and explosions where TagedP Models for reliable prediction of Deflagration-to-Detonation-
fundamental understanding is lacking. There are also areas where Transition (DDT), especially in large-scale systems.
well controlled experiments, that can be used for model validation, TagedP Reliable models for describing dust lifting from layers by flow
are needed. or shock waves in dust explosions.
TagedPSpecific to fire scenarios, the areas where better understanding TagedP Models for the mitigation/suppression effects of explosion
and predictive practical models are required include: D16X X
protection D17X X
devices, such as explosion venting, explosion sup-
pression and explosion isolation systems.
agedPT Turbulent (momentum dominated and buoyant) gas phase com-
bustion models, including description of flame extinction and TagedPImproved understanding of the mechanisms involved in tran-
re-ignition for premixed, non-premixed and partially premixed sitions between fire and explosions scenarios are also of interest.
flames. Furthermore, D18XUncertainty
X D19XQuantification
X (UQ) should be an
TagedP Soot generation and radiation of large scale fires involving gas- integral part of future models to give predictions of quantifiable
eous, liquid and solid fuels, including the effect of complex reliability. Such treatment should include model parameter
geometries and boundary conditions, such as wind. uncertainties and model-form uncertainty, i.e. errors due to trun-
TagedP Heat transfer (convective and radiative) in solid and liquid fires, cating physics.
such as wall fires, pool fires, etc.
TagedP Thermal decomposition (pyrolysis) of solids (e.g. cables, building 4. Perspectives
materials and combustible commodities in warehouses), and
related pyrolysis gas output. TagedPFurther progress in the modelling of fire and explosion phe-
TagedP Mitigation effect of sprinklers and water mist, including the nomena that can support loss prevention and safety in industry
interaction of water sprays with fire plumes, radiation attenua- may be achieved through systematic collaboration between
tion and scattering, and the physics of fire suppression. Data and experimental and modelling groups. By designing and conducting
modelling of sprinkler/nozzle droplet size, velocities and mass controlled fire and explosion experiments, from laboratory scale
flow rate distributions. to large scale, and through the sharing and discussion of results,
TagedP Reliable prediction of the effect of other mitigating measures, it is possible to assess and improve the performance of models,
such as firefighting foam. and at the same time gain improved understanding of physical
TagedP Structural behaviour of construction materials under fire, and chemical phenomena. Blind-prediction studies, where mod-
including fire spread over facade cladding and penetration of ellers submit their predictions before the actual experiments are
fires through structural separations. performed, are particularly valuable for assessing the predictive
capabilities of consequence models. Such activities facilitate
TagedPThe challenges specific to explosion scenarios include: cooperation within the global research community, with moder-
ate requirements for funding. Improvements in data quality and
agedPT High-quality data from experiments or numerical models for open access to data will certainly benefit the progress. Finally,
the laminar burning velocity and other fundamental combus- from a practical point of view, it is important for safety engi-
tion properties (e.g. Markstein numbers), for a wide range of neers in industry that experienced researchers take the time to
gaseous fuels, over the entire flammable concentration range. collect information and publish review articles and books that
TagedP Reliable mixing rules for the laminar burning velocity and describe the current state-of-the-art within selected fields.
other fundamental combustion properties for mixtures of
several gaseous fuels, including the effect of inert species. 5. Conclusions
TagedP Specification of fundamental parameters characterising the
reactivity of two-phase fuel-air mixtures (e.g. dust clouds, TagedPThe interaction between fundamental and applied combustion
liquid sprays, mists, and hybrid mixtures), such as laminar research is the key to D20X X providing reliable engineering models for
burning velocity. assessing the consequences of accidental fire and explosion
TagedP Fundamental understanding of the effect of flame instabilities, scenarios.D21X X Such models have important applications in safetyD2X X and
spatial scales, pressure, temperature and turbulence parame- loss prevention, including risk analysis and design of mitigation
ters on flame propagation, for all types of fuel-air mixtures. measures. Despite significant progress, there are numerous areas of
TagedP Understanding of the effects of mixture composition gradients combustion science where our understanding of the underlying
on flame propagation for all types of fuel-Dair
15X X mixtures. physical phenomena, as well as the quality and predictive capabili-

TagedP Reliable correlations for the turbulent burning velocity, ties of models, should be significantly improved.
and more generally, reliable and scale/mesh independent

Please cite this article as: T. Skjold et al., Fires and explosions, Progress in Energy and Combustion Science (2017), http://dx.doi.org/10.1016/j.
pecs.2017.09.003