Elsevier

Journal of Hazardous Materials

Volume 318, 15 November 2016, Pages 758-771
Journal of Hazardous Materials

A dynamic approach for the impact of a toxic gas dispersion hazard considering human behaviour and dispersion modelling

https://doi.org/10.1016/j.jhazmat.2016.06.015Get rights and content

Highlights

  • A dynamic approach to assess the impact of toxic gas dispersion is introduced.

  • The approach considers gas dispersion simulation and behavioural evacuation modelling.

  • A hypothetical scenario of a toxic hazard on a crowd attending a music festival is presented.

  • Results compare methods based on the static and dynamic approach.

Abstract

The release of toxic gases due to natural/industrial accidents or terrorist attacks in populated areas can have tragic consequences. To prevent and evaluate the effects of these disasters different approaches and modelling tools have been introduced in the literature. These instruments are valuable tools for risk managers doing risk assessment of threatened areas. Despite the significant improvements in hazard assessment in case of toxic gas dispersion, these analyses do not generally include the impact of human behaviour and people movement during emergencies. This work aims at providing an approach which considers both modelling of gas dispersion and evacuation movement in order to improve the accuracy of risk assessment for disasters involving toxic gases. The approach is applied to a hypothetical scenario including a ship releasing Nitrogen dioxide (NO2) on a crowd attending a music festival. The difference between the results obtained with existing static methods (people do not move) and a dynamic approach (people move away from the danger) which considers people movement with different degrees of sophistication (either a simple linear path or more complex behavioural modelling) is discussed.

Introduction

Toxic gas dispersion can be the result of natural or industrial accidents or terrorist attacks [1], [2]. These gases can be released in the atmosphere from three prime sources: dispersion of toxic gases; dispersion following an explosion; and dispersion of combustion products following a fire of harmful chemicals [3]. Toxic gas dispersion could be hazardous for humans since, depending on the atmospheric conditions, it can affect people located at large distances from the point of the initial source [4].

The intent of risk management is to reduce the frequency of accidental releases of toxic gases and to minimize the consequences of releases that do occur [5]. According to Pietersen [6], the evaluation of the consequences of toxic gas dispersions includes three steps: (1) source term estimation; (2) transport/dispersion modelling of the toxic gas; (3) toxic gas concentration estimation; and (4) evaluation of the impact of the toxic gas concentrations on the people.

To predict the gas dispersion and gas concentrations, many different models have been developed in the past decades [7], [8]. These models can be classified in accordance with the mathematical approach in use [1]. It is possible to distinguish different categories: (1) Gaussian models; (2) Box models; (3) Lagrangian particle and puff models; (4) analytical models; and (5) computational fluid dynamics (CFD) models [9]. In the past decades, the increment in computational power has allowed the spread of the use of CFD models to investigate a number of indoor [8] and outdoor accident scenarios such as railway accidents [10], gas dispersion in urban areas [11], [12], [13], [14] or railcar releases [15]. This type of models has been in some instances more effective in representing scenarios with complex geometries, such as urban areas [14]. However, the selection of the gas dispersion model can be affected by several other factors, such as the scale of the problem.

The evaluation of the impact of the toxic gas concentrations on humans is the final step of risk assessment. To address this issue, several approaches and indices have been introduced in the literature, such as lethal dose, lethal concentration, level of concern, etc. [1], [5], [16]. The toxic absorbed dose is the amount of toxic substances that got into the body through the eyes, skin, stomach, intestines, or lungs. The absorbed dose by inhalation is generally calculated by multiplying the concentration of the gases (to the power of an empirical constant) and the exposure time [17]. This dose is often used to calculate the probability of death by the exposure to a toxic substance for a given time using probit models [18], [19], [20], [21]. Another approach available to evaluate the health impact rate of a person is the one adopted in fire safety engineering for the case of fire enclosures. In this case, the health impact rate for each person is calculated using the Fractional Effective Doses (FED) [22]. The FED is calculated by integrating the area under the concentration/time curves for the toxic products inhaled by a person and dividing this amount by the effective dose to cause incapacitation or death [23]. Therefore, incapacity or death occurs whenever the FED exceeds one.

Despite the existence of different methodologies to perform risk analysis in case of toxic gas dispersion, the main issue of the existing methods is that while sophisticated indices which may consider people behavioural activities are often used for hazard in enclosures (e.g. FED calculations) [24], [25], [26], the health impact indices in open spaces are generally calculated assuming a ‘stationary observer’, i.e. the cumulative absorbed dose over time is calculated for static points or areas of a threatened environment. In other words, this approach assumes that a person stays in a fixed point for the whole duration of the release without taking into account people attempts to escape from these areas [1], [14], [27]. For this reason, existing methods do not fully take into account human behaviour during emergencies. By contrast, risk assessment should take into account the evacuation of people from threatened areas to a safe place and calculate the health impact rate of people accordingly (considering their position over the time, i.e. their trajectories). To date, several pedestrian and traffic evacuation models have been developed to predict human behaviour and movement for both indoor and outdoor scenarios [28], [29], [30], [31]. Those models may work on different scales, ranging from small to larger city scales [32], [33]. However, despite the availability of these evacuation models, an approach for risk assessment in open spaces considering gas dispersion modelling and human behaviour modelling has not been fully investigated in the literature. To date, human factors have been investigated using modelling approaches based on assumptions on people human behaviour (i.e. the evacuees move on a predefined trajectory in a network with pre-defined constant speed without interact with each other) [21] or considering agent-based models for vehicle representation which have still not applied for human movement [34]. Furthermore, to date, no systematic comparison of different approaches is available.

The main objective of this work is to introduce a methodology considering both physical and human behaviour modelling to investigate the impact of the toxic gas concentrations on people in open spaces and compare results of static and dynamic approaches adopting different levels of behavioural complexity in the modelling assumptions. Compared with previous research on the topic [21], the proposed method uses an advanced modelling representation of human behaviour, i.e., microscopic agent-based modelling and it directly addresses people movement rather than vehicle movement [26]. Agent-based modelling is considered a high-level modelling approach in evacuation modelling if compared with simpler pedestrian flow calculations since it allows complex interactions among pedestrians (including the occurrence of emergent behaviours). This approach aims at improving the accuracy of risk assessment for disasters involving toxic gas dispersion. The proposed methodology is presented through a hypothetical exemplary scenario, which includes a ship releasing Nitrogen dioxide (NO2), affecting a crowd attending a music festival [35], [36]. In the case study, the gas dispersion is simulated using the Fire Dynamics Simulator model (FDS 6.1.2) [37]. The crowd evacuation is modelled using Pathfinder [38], and the evacuees’ health impact rate is calculated using the modified Haber’s law [17]. The results of this study are analysed using a probabilistic spatial-temporal analysis based on probit models for NO2 [27], [39], which allows identifying the areas having the highest probabilities of mortality. It is important to stress that the goal of the present study is to provide a proof-of-concept of the approach. In other words, it is not the intention to obtain the highest level of accuracy possible for CFD, nor for the evacuation modelling. In contrast, this study wants to show the feasibility and application of a dynamic approach and the differences in the results obtained in comparison with a static approach.

Section snippets

Method

This section introduces a methodology for toxic gas dispersion risk assessment. The proposed method allows performing risk analysis, including the simulation of behaviour of individuals threatened by the toxic gases, as illustrated in Fig. 1.

The methodology starts with the identification of a scenario. This scenario could either be hypothetical or represent a real accident. In case of hypothetical scenarios, different criteria can be used to identify the relevant hypothetical scenarios for risk

Case study

The approach introduced in Section 2 is applied to a hypothetical evacuation scenario, based on the case study developed in [35], [36]. The scenario concerns an outdoor music festival in an area restricted by fences. The festival area is close to a residential area, a river and major road transport infrastructure (highway and secondary roads). Only a portion of the festival is considered in the present case study. The number of attendees is calculated considering an average initial density of 3

Results

The concentrations of NO2 released from the ship have been calculated in a horizontal plane, set at 1.6 m height, as illustrated in Fig. 4b. This plane represents the closest one to the position of the nose and mouth of the evacuees [78]. Output concentrations have been recorded every second. Nitrogen dioxide (NO2) is the main harmful chemical specie based on the decomposition reaction of ammonium nitrate considered in this study hence it is considered here. As it has been previously reported,

Discussion

Including behavioural modelling in hazard risk assessment allows investigating the impact of different human factors and evacuation strategies, such as way-finding, dynamic availability of exits, flow through exits, the impact of density conditions at the beginning and during the emergency, etc. Considering human factors in a risk assessment in case of gas dispersion can help disaster managers to reduce the risk of mortality and injury and improve the people resilience in case of emergencies.

Conclusion

This work introduced an approach to perform risk assessment that considers human behaviour and gas dispersion modelling. The proposed method expands previous attempts to model human factors [21] by using an advanced modelling approach for human behaviour (i.e. a microscopic agent-based approach). The main advantage of the approach is that the absorbed dose of a toxic gas is calculated considering that people threatened by a toxic gas are not ‘stationary observers’, but they move towards a safe

Acknowledgements

This work is part of the Seventh Framework Programme EU project CascEff − Modelling of dependencies and cascading effects for emergency management in crisis situations, Grant agreement No: 607665. The CascEff project aims to improve the emergency response in incidents that involve cascading effects. The authors wish to acknowledge the members of the project consortium. The authors also thank Bart Bruelemans, Wim Van de Vijver and Kris de Troch for their help in the definition of the model case

References (87)

  • S.R. Hanna et al.

    CFD model simulation of dispersion from chlorine railcar releases in industrial and urban areas

    Atmos. Environ.

    (2009)
  • A.P. Franks et al.

    The relationship between risk of death and risk of dangerous dose for toxic substances

    J. Hazard. Mater.

    (1996)
  • W.F. ten Berge et al.

    Concentration—time mortality response relationship of irritant and systemically acting vapours and gases

    J. Hazard. Mater.

    (1986)
  • J.K.W. Davies

    The application of box models in the analysis of toxic hazards by using the probit dose-response relationship

    J. Hazard. Mater.

    (1989)
  • S. Schubach

    A measure of human sensitivity in acute inhalation toxicity

    J. Loss Prev. Process Ind.

    (1997)
  • T. Hilderman

    A model for effective toxic load from fluctuating gas concentrations

    J. Hazard. Mater.

    (1999)
  • A.R.R. Péry et al.

    Comparison of models to analyze mortality data and derive concentration-time response relationship of inhaled chemicals

    Regul. Toxicol. Pharmacol.

    (2010)
  • S. Bae et al.

    Development of a smoke effect model for representing the psychological pressure from the smoke

    Saf. Sci.

    (2015)
  • J. Wan et al.

    Research on evacuation in the subway station in China based on the combined social force model

    Physica A: Stat. Mech. Appl.

    (2014)
  • B. Zhang et al.

    Quantitative risk analysis of toxic gas release caused poisoning—a CFD and dose–response model combined approach

    Process Saf. Environ. Prot.

    (2010)
  • F.I. Khan et al.

    A criterion for developing credible accident scenarios for risk assessment

    J. Loss Prev. Process Ind.

    (2002)
  • R.F. Griffiths

    The use of probit expressions in the assessment of acute population impact of toxic releases

    J. Loss Prev. Process Ind.

    (1991)
  • Y. Mouilleau et al.

    CFD simulations of atmospheric gas dispersion using the Fire Dynamics Simulator (FDS)

    J. Loss Prev. Process Ind.

    (2009)
  • D. Purser et al.

    Quantification of behaviour for engineering design standards and escape time calculations

    Saf. Sci.

    (2001)
  • R. Lovreglio et al.

    An evacuation decision model based on perceived risk, social influence and behavioural uncertainty

    Simul. Model. Pract. Theory

    (2016)
  • M. Kobes et al.

    Way finding during fire evacuation; an analysis of unannounced fire drills in a hotel at night

    Build. Environ.

    (2010)
  • R. Lovreglio et al.

    A discrete choice model based on random utilities for exit choice in emergency evacuations

    Saf. Sci.

    (2014)
  • R. Lovreglio et al.

    A study of herding behaviour in exit choice during emergencies based on random utility theory

    Saf. Sci.

    (2016)
  • S.M. Tauseef et al.

    CFD-based simulation of dense gas dispersion in presence of obstacles

    J. Loss Prev. Process Ind.

    (2011)
  • M.J. Assael et al.

    Fires Explosions, and Toxic Gas Dispersions: Effects Calculation and Risk Analysis

    (2010)
  • P. Delmelle et al.

    Atmospheric dispersion, environmental effects and potential health hazard associated with the low-altitude gas plume of Masaya volcano, Nicaragua

    Bull. Volcanol.

    (2002)
  • H.C. Rodean

    Stochastic Lagrangian Models of Turbulent Diffusion

    (2016)
  • S.R. Hanna et al.

    Detailed simulations of atmospheric flow and dispersion in downtown manhattan: an application of five computational fluid dynamics models

    Bull. Am. Meteorol. Soc.

    (2006)
  • D.A. Purser, J.L. McAllister, Assessment of Hazards to Occupants from Smoke, Toxic Gases, and Heat, In: M.J. Hurley,...
  • D.A. Purser

    Modelling toxic and physical hazard in fire

    Fire Saf. Sci.

    (1989)
  • T. Korhonen, S. Hostikka, S. Heliövaara, H. Ehtamo, FDS+Evac: An Agent Based Fire Evacuation Model, In: Pedestr....
  • S.M. Gwynne et al.

    A review of the methodologies used in evacuation modelling

    Fire Mater.

    (1999)
  • E.D. Kuligowski, R.D. Peacock, B.L. Hoskins, A Review of Building Evacuation Models, 2nd Edition-Technical Note (NIST...
  • E. Ronchi, M. Kinsey, Evacuation models of the future: Insights from an online survey on user’s experiences and needs,...
  • A.J. Pel et al.

    A review on travel behaviour modelling in dynamic traffic simulation models for evacuations

    Transportation (Amst).

    (2011)
  • T.N. McPherson, J. Rush, H. Khalsa, A. Ivey, M.J. Brown, A day-night population exchange model for better exposure and...
  • T.N. McPherson et al.

    Estimating daytime and nighttime population distributions in U.S. cities for emergency response activities

  • J.M. Epstein et al.

    No combining computational fluid dynamics and agent-based modeling: a new approach to evacuation planning

    PLoS One

    (2011)
  • Cited by (73)

    View all citing articles on Scopus
    View full text