Document 2zEpg6LOkpdkDYp4bEj9K7ML

RESOURCE ARTICLE Dust explosion causation, prevention and mitigation: An overview The current paper gives a general overview of factors that can cause a dust explosion and the means by which such an event can be prevented and mitigated. Important explosibility parameters (e.g., maximum explosion pressure and standardized maximum rate of pressure rise) are described in terms of their relation to explosion causation, prevention and mitigation. Causation is further explained by means of the fire triangle and the explosion pentagon, and standard risk reduction measures are placed in a hierarchical arrangement incorporating inherent safety, passive engineered safety, active engineered safety, and procedural safety. The importance of safety culture and a safety management system approach are emphasized by reference to an industrial case study. By Paul R. Amyotte, Rolf K. Eckhoff INTRODUCTION A dust explosion can occur when par ticulate solid material is suspended in air and a sufficiently energetic ignition source is present. The consequences are often similar to those arising from a gas explosion in terms of impact on people, physical assets and business production. While most industrial practitioners are familiar with at least the basic concepts of gas explosions (e.g., the need for a fuel, oxidant and ignition source), the same cannot be said for dust explosions. The primary distinguishing factor between dust and gas explosions is the phase of the fuel itself--solid versus gaseous. Particle size is therefore a dominant issue in efforts aimed at pre- Paul R. Amyotte is affiliated, with the Department of Process Engineering and Applied Science, Dalhousie Uni versity, 1360 Barrington Street, Hali fax, Nova Scotia, Canada (e-mail: paul.amyotte@dal.ca). Rolf K. Eckhoff is affiliated with the Department of Physics and Technol ogy, University of Bergen, Bergen, Nor way and Tyrens AB - 0SA Fire & Risk, Malmo, Sweden. venting dust explosions and mitigating their consequences. The National Fire Protection Association (NFPA) defines a dust as any finely divided solid, 420 mm or 0.017 in. or less in diameter (i.e., material capable of pas sing through a U.S. No. 40 Standard Sieve).1 Since the range of explosible particle sizes for a given material can be quite large, this definition high lights the importance of considering the particle size distribution in addi tion to a mean or median particle dia meter. Further, the shape for which a given material poses a dust explosion hazard may not be limited to spherical or near-spherical particles, but could include flakes, fibres and flocculent forms. Dust Explosion Occurrence One of the first recorded accounts of a dust explosion was written by Count Morozzo2 in 1795, and gave a detailed account of an explosion in a flour warehouse in Turin, Italy (as described by Piccinini3 and Eckhoff4). In 1845, Faraday and Lyell5 elucidated the key role of coal dust in the devastating explosion in the Haswell (UK) coal mine the previous year.6 Fast-forward ing to the 21st century, dust explosions remain a persistent and damaging industrial occurrence. The US Chemi cal Safety and Hazard Investigation Board (CSB) has recently completed a series of reports dealing with inves tigations into the causes of serious dust explosion incidents that occurred in the United States during 2003.7-9 A fourth CSB report10 gives the findings of a study of dust explosions in general industry which was initiated following the three catastrophic incidents in 2003. At the time of writing the current manuscript, the CSB is engaged in investigating the 2008 explosion and fire at the Imperial Sugar refinery near Savannah, GA. Dust explosion incidents are not, however, restricted to coal mines and food-processing facilities; nor are they restricted to the scenario of an indus trial disaster. Frank11 gives incident data reported by the US CSB and FM Global, which illustrate that dust explosions have occurred, for example, in the following industries with the indicated typical commodities: Wood and paper products (dusts from sawing, cutting, grinding, etc.); Grain and foodstuffs (grain dust, flour); Metal and metal products (metal powders and dusts); Power generation (pulverized coal, peat and wood); Rubber; Chemical process industry (acetate flake, pharmaceuticals, dyes, pesti cides); Plastic/polymer production and processing; Mining (coal, sulphide ores, sul phur); and 1871-5532/$36.00 doi:10.1016/j.jchas.2009.05.002 Division of Chemical Health and Safety of the American Chemical Society Elsevier Inc. All rights reserved. 15  Textile manufacturing (linen flax, cotton, wool). Objectives of Paper The primary objective of the current paper is to provide a general overview of the mechanisms by which dust explosions originate, and the meth odologies by which they can be pre vented and their consequences mitigated. The paper is intended for the generalist audience in industry. Other resources are available for spe cialists in industrial loss prevention and dust explosion research. For example, recent reviews cover in detail case histories, causes, consequences and control of dust explosions,12 the role of powder science and technology in understanding dust explosion phenomena,6 and the status of develop ments in basic knowledge and practical application with respect to dust explosion prevention and mitigation.13 An additional objective of the cur rent paper is to link the fundamentals of dust explosion causation, preven tion and mitigation with the concepts of safety culture, a safety management system, and inherent safety. Based on the experience of the authors, it is critically important to place technical knowledge in a context of demon strable management commitment to dust explosion risk reduction.4,14-16 DUST EXPLOSION CAUSATION As a first step to understanding dust explosion causation, it is important to distinguish between a dust explosion hazard and a dust explosion risk. While the terms hazard and risk are often used interchangeably, they repre sent very different concepts. Func tional definitions of each term are given by Wilson and McCutcheon17: Hazard: The potential of a machine, equipment, process, material or physi cal factor in the working environment to cause harm to people, environment, assets or production. Risk: The possibility of injury, loss or environmental incident created by a hazard. The significance of risk is a function of the probability (or likeli hood) of an unwanted incident and the severity of its consequences. It is clear from the above definitions that risks are determined by assess ment of the likely consequences of identified hazards. Thorough hazard identification is, therefore, key to the effective management of risk; one can not manage the risk arising from a hazard that has not been identified. With respect to dust explosions, it is essential to determine whether a given material actually constitutes an explo sion hazard, and if so-the degree of hazard as represented by various explosibility parameters. Only then can appropriate risk reduction mea sures aimed at prevention and mitiga tion be devised and implemented. Expressed from a practical, industrial perspective, hazard analysis identifies what can go wrong, whereas risk ana lysis assesses the probable conse quences of these events in terms of probable loss of life, probable injuries, and probable loss of property, produc tion capacity and market shares. The above reasoning is illustrated in Figure 1. Here, the key step of hazard identification is shown as a precursor to various explosion prevention and mitigation measures, each of which has been linked to one or more specific explosibility parameters. These points are discussed in subsequent sections of the paper, as is the issue of manage ment responsibility (as identified in the block at the bottom centre of Figure 1). Explosibility Parameters The purpose of this section is to pro vide a guide to some of the more important and commonly reported dust explosibility parameters. Para meter values for many materials can be found in the literature (e.g., NFPA 681 and Eckhoff4) or online databases (e.g., BGIA18). Such values should, however, only be used as indications, and not as the ultimate basis for design of actual safety measures in industry (which should be based on test data for the actual dust in question).4 When identifying dust explosion hazards, one is inevitably drawn to an examination of the material itself in an attempt to answer questions such as: (i) Can the dust yield dust explo sions when dispersed as a cloud in air? (ii) How high is the resulting overpres sure if the explosion occurs at constant volume? (iii) How quickly does the pressure rise if the explosion occurs at constant volume? (iv) What concen tration of airborne dust is needed for an explosion? (v) How much energy, or how high a temperature, is needed for ignition? (vi) What minimum per centage of oxygen in the atmosphere is required to sustain flame propagation in the dust cloud? These questions are addressed by determining the basic explosibility parameters of the dust in accordance with Figure 1 and as shown in Table 1. It is important to recognize that these parameters are not fundamental prop erties of a given material. They are strongly dependent on both material characteristics (e.g., moisture content and particle size, shape, and porosity) and experimental conditions (e.g., ves sel volume, turbulence of the dust cloud, and applied ignition energy). The standard test methodologies (American Society for Testing and Materials) given as examples in Table 1 represent industry-consensus approaches to ensuring the measured parameters are applicable to the pre vention and mitigation of industrialscale dust explosions. Most of the parameters listed in Table 1 are self-explanatory, with the exception of KSt, the volume-normal ized maximum rate of pressure rise. The parameter (dP/dt)max is of course dependent on the volume of the explo sion chamber, and is therefore of lim ited use on its own. For scaling to larger volumes, maximum rates of pressure rise are normalized by multi plying by the cube root of the explosion chamber volume, V: KS> = (f) V 1,3 (>) Eq. (1) is sometimes referred to as the cubic or cube root law and KSt as the dust constant. (The subscript `St' derives from the German word for dust-staub.) It is preferable, however, to refer to Eq. (1) as the cubic relation ship and KSt as the volume-normalized (or standardized) maximum rate of pressure rise, or simply as KSt. There 16 Journal of Chemical Health & Safety, January/February 2010 Does dust present explosion hazard? Check available data Yes/No Classification test Dust is not explosible, i.e. there is no dust explosion hazard Avoid dust accumulation outside process equipment. Good housekeeping! Obey rules for hot work etc. Dust is explosible Explosion prevention and mitigation are necessary Inerting of plant Maximum oxygen concentration to prevent ignition Ignition temperatures Acquire technical information about specific plant and explo sion hazard. Take appropriate actions. If required, determine ignitabiiity and explosibility parameters in professional laboratory Control process and surface temperatures to prevent ignition of dust cloud and dust deposits Maximum explosion pressure and explosion rate --------- --------- Explosion containment Explosion venting Explosion suppression Explosion isolation Partial inerting to reduce pressure and violence Minimum explosible dust concentration Control concentration of suspended dust whenever feasible Inform and motivate all levels of employees. Top management is responsible! Minimum ignition energy Earth electrically conducting plant Consider use ot antistatic materials and clothing, where appropriate Figure 1. Logic diagram for dust explosion hazard identification and risk reduction.4 Journal of Chemical Health & Safety, January/February 2010 17 Table 1. Important dust explosibility parameters and their determination and application. Typical Parameter units Description Risk component addressed Example test methodology Example industrial applications (Figure 1) Pmax (dP/dt)max KSt MEC MIE MIT LIT MOC (LOC) bar(g) bar/s bar m/s g/m3 mJ C C volume% Maximum explosion pressure in constant-volume explosion Maximum rate of pressure rise in constant-volume explosion Volume-normalized (or standardized) maximum rate of pressure rise in constant-volume explosion Minimum explosible (or explosive) dust concentration Minimum ignition energy of dust cloud (electric spark) Minimum ignition temperature of dust cloud Minimum ignition temperature of dust layer or dust deposit Minimum (or limiting) oxygen concentration in the atmosphere for flame propagation in dust cloud Consequence severity Consequence severity Consequence severity Likelihood of occurrence Likelihood of occurrence Likelihood of occurrence Likelihood of occurrence Likelihood of occurrence ASTM E1226-05 ASTM E1226-05 ASTM E1226-05 ASTM E1515-07 ASTM E2019-03 ASTM E1491-06 ASTM E2021-06 ASTM WK1680 Containment, venting, suppression, isolation, partial inerting As per Pmax As per Pmax Control of dust concentrations Removal of ignition sources. Grounding and bonding Control of process and surface temperatures (dust clouds) Control of process and surface temperatures (dust layers) Inerting (with inert gas) is nothing fundamental (in the sense of an inviolable law) or constant about either Eq. (1) or the KSt parameter. Again, KSt is not a fundamental prop erty of a given material. This observation is reinforced by the discussion in the text by Eckhoff,4 in which the need for appropriate deter mination of KSt values is emphasized. An analysis of basic considerations reveals that the cubic relationship is valid only for geometrically similar vessels giving geometrically similar flame surfaces, and if the flame thick ness is negligible compared to the ves sel radius, and if the burning velocity as a function of pressure and temperature is identical in all volumes. In view of these issues, it is clear that KSt from a dust explosion in any arbitrary vessel is a correspondingly arbitrary measure of dust explosion violence, because both the turbulence and the vessel shape are arbitrary, and because the flame thick ness is most probably considerable in relation to the vessel radius. Table 2 shows a selection of KSt values for maize starch dust clouds in air, determined in various appara tus.4 (Eckhoff4 may be consulted for the listing of investigators and original references.) The values range from 3 to 6 bar m/s to over 200 bar m/s, corre sponding to approximately an order of magnitude difference. Some of the dis crepancies may arise from differences in moisture content and effective par ticle size of the starch, and to different data interpretation (peak or mean values). However, differences in turbu lence of the dust clouds and significant flame thicknesses probably play the main roles. Therefore, when using KSt values for sizing of areas of explosion vents and Table 2. KSt values measured for clouds of maize starch dust in air in different closed vessels and arranged according to vessel volume.4. (dP/dt)max (bar/s) Volume (V) of apparatus (m3) KSt (barm/s) 680 612 220 413 320 365 10-20 60-80 272 50 72 20 136 110 55 0.0012 0.0012 0.0012 0.009 0.020 0.020 0.026 0.026 0.028 0.33 0.95 0.95 3.12 6.7 13.4 73 66 23 86 87 100 3-6 20-25 83 34 71 20 200 209 131 18 Journal of Chemical Health & Safety, January/February 2010 Table 3. Wood dust explosibility results.19. Dust Particle size distribution Coarse 50 weight% <1 mm 0.3 weight% <75 mm Fine 93 weight% < 1 mm 35 weight% <125 mm 16weight% <75 mm KSt (barm/s) 9 130 for design of explosion isolation and explosion suppression systems accord ing to current standards, it is absolutely essential to use data obtained from the authorized standard test methods for determining KSt. Applicable methods in this regard include those falling under the auspices of the American Society for Testing and Materials (ASTM), International Organization for Standardization (ISO), and Eur opean Committee for Standardization (CEN). Further guidance in the application of the KSt concept can be gained by examination of Table 3, which shows explosibility data for dust generated in a wood-processing facility19 (with KSt determined in a 20-L spherical vessel according to ASTM E 1226-05). In the actual process, although the coarse dust was predominant, pockets of the fine dust were found in a dead-space in a process unit header. Recalling that KSt is used for sizing explosion relief vents (Table 1), it is clear that a vent design based on the KSt value of the coarse dust would be inadequate for protection from the effects of a dust explosion involving the fine wood dust. The data in Table 3 also reinforce the need to distinguish between material hazard and process risk when dealing with dust explosion phenomena. As mentioned in the introduction, particle size distribution is one of the key prop erties of a dust that defines its material hazard. Knowledge of whether a par ticular size distribution is actually encountered in a given application is required to assess the process risk. This highlights the importance of gaining a thorough understanding of the dust handling process under consideration during both normal and upset condi tions. Although the preceding paragraph reinforces the difference between a dust explosion hazard and a dust explosion risk, it is equally important to recognize that the two - hazard and risk - are not independent of one another. This is evident from our pre vious comment that risk can only be assessed once hazards have been iden tified. There is thus an inevitable link between the two parameters. Selection of a dust sample for testing to see if the material represents an explosion hazard is often preceded by considera tions related to process risk. For exam ple, in plants with dust extraction systems, it is typical to select dust sam ples from filters. These samples will be finer than the main product and hence, will explode more violently and ignite more readily. Here, the reasoning behind the selection of the sample to be tested is directly or indirectly a part of the assessment of the process risk. This again is the point being made by the data in Table 3. Fire Triangle The most basic guide to understanding dust explosion causation is the familiar fire triangle (Figure 2). This simple concept indicates that three of the necessary conditions for a dust explo sion are a fuel, an oxidant and an ignition source. As presented by the IChemE20 and illustrated in Figure 3, the unique fuel requirements for a dust can be expressed as follows: (i) the dust material must be combustible, (ii) the dust must be airborne, (iii) the dust must have a particle size distribution capable of propagating flame when dispersed as a cloud in air, and (iv) the dust concentrations must be within the explosible range. Explosion Pentagon The explosion pentagon, as described by Kauffman21 and illustrated in Figure 4, expands the basic fire triangle to include mixing of the fuel and oxi dant and confinement of the mixture. The first of these additional compo nents illustrates the previously men tioned key difference between dust and gas explosions-a solid rather than a gaseous fuel. A gas explosion there fore involves a homogeneous system in which the smallest entities of fuel and air are separated only by molecular distances. Thorough mixing of fuel and oxidant is readily achieved and gravitational effects are negligible. However, in a dust/air mixture, the dust particles are strongly influenced by gravity; an essential prerequisite for a dust explosion is the formation of a dust/oxidant suspension.22 Once com bustion of the resultant mixture occurs, confinement (partial or complete) per mits an overpressure to develop, thus enabling a fast-burning dust flame to transition to a dust explosion. Primary Explosions Dust explosions usually occur in industry inside process vessels and units such as mills, grinders, and dryers-i.e., inside equipment where the conditions of the explosion penta gon are satisfied. Such occurrences are Journal of Chemical Health & Safety, January/February 2010 19 Ignition Source Figure 4. Explosion pentagon.21 I1 !I |1 | IN- ! 1 DUST-j | RIAL j HYGI- j ENE 1 i1 j1 il i------------ 1 ; EXPLOSIBLEi RANGE :'.i 1---------- 4------------1--------- L I ' :* ' I DUST DEPOSIT 1(TJ !0"3 10"' 1 10 101 10s 10' 105 10s MASS OF POWDER/DUST PER UNIT VOLUME [g/rtfl] Figure 5. Typical ranges of dust concentrations in air at normal temperature and pressure for common natural organic dusts, for maximum permissible hygienic exposure, dust explosions, and dust deposit combustion (smoldering fires), respectively.4 often called primary explosions, espe cially if they result in secondary explo sions external to the process unit (as described in the next section). The reason for the majority of dust explo sions being initiated in this manner can be understood by examining Figure 5. Here, the range of explosible dust con centrations in air at normal tempera ture and pressure for a natural organic dust (e.g., cornstarch) is compared with the typical range of maximum permissible dust concentrations that are relevant in the context of industrial hygiene, and with a typical density of deposits or layers of natural organic dusts. Clearly, the range of explosible concentrations is orders of magnitude greater than the concentrations per mitted in areas inhabited by workers. Secondary Explosions Notwithstanding the discussion in the previous section, dust explosions do occur in process areas, not just inside process units. A secondary explosion can be initiated due to entrainment of dust layers by the blast waves arising from a primary explosion. The primary event might be a dust explosion origi nating in a process unit, or could be any disturbance energetic enough to disperse explosible dust layered on the floor and various work surfaces. An example of such an energetic dis turbance (other than a primary dust explosion) would be a gas explosion leading to a dust explosion. This is a well-documented phenomenon in the underground coal mining industry, where devastating effects can result from the overpressures and rates of pressure rise generated in a coal dust explosion that has been triggered by a methane explosion. The required amount of layered dust which, once airborne, could sustain a secondary dust explosion is often grossly overestimated. For example, in the Westray mine explosion23 described later in this paper, one of the contributing factors was the pre sence of coal dust layers several centi metres thick throughout the mine workings. In fact, the amount of coal dust that could be dispersed by an aerodynamic disturbance and then combusted with the available oxygen would have been significantly less than `several centimetres'. This point is illu strated in a general manner by Figure 6, which shows the implications of the following expression: C = (r,ulk)( ) (2) Figure 6. (a-c) Illustration of the potential for dust cloud generation from a dust layer.4 20 Journal of Chemical Health & Safety, January/February 2010 where pbuik is the bulk density of a dust layer, h is the layer thickness, H is the height of the dust cloud produced from the layer, and C is the resulting dust concentration. From Eq. (2) and as shown in Figure 6, a 1-mm thick layer of a dust of bulk density 500 kg/m3 on the floor of a 5-m high room will generate a cloud of average concentration 100 g/m3 if dispersed evenly all over the room.6 Such a concentration is of the order of the minimum explosible concentra tion for many explosible dusts. Partial dispersion up to 1m above the floor yields a dust concentration of 500 g/ m3; this is a concentration that is of the order of the optimum concentration (i.e., the concentration producing the most devastating overpressures and rates of pressure rise) for many explosible dusts. Clearly, even seemingly harmless dust layers have the potential to rapidly escalate the risk of a dust explo sion. This observation helps to explain the advice given by experienced indus trial practitioners-such as: there's too much layered dust if ... you can see your footprints in the dust ... you can write your initials in the dust. These comments, although anecdotal, have a firm foundation in the chemistry and physics of dust explosions. Hybrid Mixtures Hybrid mixtures consist of a flam mable gas and a combustible dust, each of which may be present in an amount less than its lower flammable limit (LFL)/minimum explosible concen tration (MEC), and still give rise to an explosible mixture. The focus when discussing hybrid mixtures is, in fact, on admixture of a flammable gas in concentrations below the lower flam mable limit of the gas itself. If the LFL for the gas is exceeded, one soon has a situation where the worst-case sce nario for a primary explosion would be a pure gas explosion. Perhaps the most well-known hybrid mixture is the methane/coal dust system often encountered in underground coal mining. There are also several examples of hybrid mix ture formation in other industries, such as the natural gas/fly ash system in fossil-fuel burning power plants and various hydrocarbon/resin combina tions occurring in the production of plastic powders. The influence of the co-presence of a flammable gas on the explosibility parameters of a fuel dust alone is well established (e.g., Cashdollar24). These effects include higher values of max imum explosion pressure and maxi mum rate of pressure rise (and hence KSt), and lower values of minimum explosible concentration and mini mum ignition energy. There is, of course, already a hazard that exists when an explosible dust is present in a quantity above its minimum explosible concentration. With flammable gas admixture, the scenario is now one of magnification of an already existing hazard, not the creation of a problem that did not already exist in some form already. DUST EXPLOSION PREVENTION AND MITIGATION When selecting dust explosion preven tion and mitigation measures, it is helpful to employ a heuristic or frame work for making appropriate choices.25 The fire triangles shown in Figures 2 and 3, and the explosion pentagon shown in Figure 4, offer gui dance in this area in addition to iden tifying dust explosion causation factors. For example, the triangle affords industrial practitioners several approaches to explosion prevention (e.g., removal of fuel by good house keeping and removal of electrostatic ignition sources by grounding and bonding). The use of the pentagon to visualize explosion requirements leads to identification of measures for explo sion mitigation such as venting (in relief of the confinement criterion). What is missing in the above discus sion, however, is guidance on which risk reduction techniques are most effective and in what order the various techniques should be considered. Pre vention is obviously preferred to miti gation, but both are likely to be required in a given application. The question, then, is which of the various measures spanning out from the cen tre sphere in Figure 1 should receive priority? The following sections attempt to address this point by first considering a general loss prevention approach, and then tailoring that approach to prevention of loss from dust explosions. Hierarchy of Controls Industrial loss prevention is generally accomplished in three ways: (i) inher ent safety, (ii) engineered safety (pas sive and active), and (iii) procedural safety.14 Engineered, or add-on, safety involves the addition of safety devices at the end of the design. A general industrial example would be a machine guard; with respect to the subject of the current paper, an example of an engi neered safety device is an automatic dust explosion suppression system. These safety devices do not perform any fundamental operation, but are designed to act when a process upset occurs. Procedural safety measures, or administrative controls, utilize safe work practices and procedures to reduce risk. Again, a general industrial example would be standard work prac tices for confined space entry; with respect to dust explosion prevention, hot-work permitting and procedures related to grounding and bonding are pertinent examples in this category. On the other hand, inherent safety uses the properties of a material or process to eliminate or reduce the hazard. The fundamental difference between inherent safety and the other two categories is that inherent safety seeks to remove the hazard at the source as opposed to accepting the hazard and looking to prevent its occurrence or mitigate its effects. Figure 7 illustrates a systematic approach to loss prevention that has found general acceptance in industry. With this approach, the preferred order of consideration for risk reduc tion measures (from most effective to least) is inherent, passive engineered, active engineered, and procedural safety. This is akin to the layer of pro tection analysis (LOPA) concept in which inherently safer process design sits at the central core of the layers. Hopkins26 uses the phrase hierarchy of controls to describe essentially the same idea; i.e., that there is a hierarch ical ordering of controls to deal with hazards, covering the spectrum from Journal of Chemical Health & Safety, January/February 2010 21 ACTIVITIES STEPS Identify all hazards and causes of these materials, actions, and conditions Assess hazards, their causes and effects, and how these interact with the design INHERENT SAFETY PRINCIPLES Avoid or eliminate hazard by design Minimize, substitute or moderate to reduce the severity of the hazard IDENTIFY HAZARDS T UNDERSTAND HAZARDS r AVOID HAZARDS T REDUCE SEVERITY Simplify the process or plant to reduce the likelihood of the hazard occurring REDUCE LIKELIHOOD Use distance, or use sections of the plant itself as barriers to segregate/protect people and emergency systems from effects of hazards SEGREGATE ADD-ON SAFETY Use safeguards that do not need initiation, and hence have high availability Use active systems, but note these depend on timely hazard detection and initiation Operator and maintenance procedures should be the last resort, especially for control and mitigation, where the chance of error or failure is high Use findings of the hazard assessment to estimate the risks, and target and implement inherent/segregation/ add-on/procedural safeguards until risks are tolerable APPLY PASSIVE SAFEGUARDS APPLY ACTIVE SAFEGUARDS * APPLY PROCEDURAL SAFEGUARDS APPLY RESIDUAL RISK -> REDUCTION MEASURES Figure 7. A systematic approach to loss prevention.19 22 Journal of Chemical Health & Safety, January/February 2010 Table 5. A hierarchical view of various means of preventing and mitigating dust explosions. Explosion prevention Preventing explosible dust clouds Process design to prevent undesired generation of dust clouds and particle size reduction and segregation Inherent Safety-Minimization, Substitution, Moderation, Simplification Preventing ignition sources Smouldering combustion in dust, dust fires Procedural Safety-may also involve aspects of Inherent Safety or Engineered Safety Explosion mitigation Good housekeeping (dust removal/cleaning) Mitigation with respect to secondary dust explosions; prevention with respect to primary dust explosions Inherent Safety-Minimization Keeping dust concentration outside explosible range Inherent Safety-Minimization Other types of open flames (e.g., hot work) Procedural Safety-may also involve aspects of Inherent Safety or Engineered Safety Explosion-pressure resistant construction Inherent Safety-Simplification Inerting of dust cloud by adding inert dust Inherent Safety-Moderation Hot surfaces (electrically or mechanically heated) Procedural Safety-may also involve aspects of Inherent Safety or Engineered Safety Explosion isolation (sectioning) Inherent Safety-Moderation (e.g., unit segregation, product choke, etc.) if not using mechanical devices. If mechanical devices are used to isolate plant sections, classification would be Engineered Safety-Passive in the case of physical barriers, or Engineered Safety-Active in the case of isolation valves. Intrinsic inerting of dust cloud by combustion gases Engineered Safety-Active Heat from mechanical impact (metal sparks and hot-spots) Procedural Safety-may also involve aspects of Inherent Safety or Engineered Safety Explosion venting Engineered Safety-Passive Inerting of dust cloud by N2, CO2 and rare gases Engineered Safety-Active Electric sparks and arcs and electrostatic discharges Procedural Safety-may also involve aspects of Inherent Safety or Engineered Safety Automatic explosion suppression Engineered Safety-Active Partial inerting of dust cloud by inert gas Engineered Safety-Active indicators must measure the effective ness of the various measures compris ing the risk control system. In other words, safety indicators must be related to the elements making up the safety management system. Safety management systems Safety management systems are recog nized and accepted worldwide as bestpractice methods for managing risk. They typically consist of 10-20 pro gram elements that must be effectively carried out to manage the risks in an acceptable way. This need is based on the understanding that once a risk is accepted, it does not go away; it is there waiting for an opportunity to happen unless the management sys tem is actively monitoring company operations for concerns and taking proactive actions to correct potential problems. As a primary corporate objective, dust explosion prevention and mitiga tion would typically fall within the scope of a Process Safety Management system (i.e., a management system for process-related hazards such as fire, explosion, release of toxic materials, etc.) One such system widely used in industry is termed PSM, Process Safety Management-where PSM is defined as the application of management principles and systems to the identifi cation, understanding and control of process hazards to prevent processrelated injuries and accidents. The suite of PSM elements is shown in Table 6. In light of the previous discussion on the hierarchy of controls, a strong case can be made for the need to demon strate a commitment to the principles of inherent safety within each of the PSM elements listed in Table 6.16 Thus, within element 6, process risk manage ment, the hierarchical arrangement of dust explosion prevention and mitiga tion measures shown in Table 5 would find direct application. Additionally, within element 8, training and perfor mance, a strong safety culture would necessitate the provision of workplace training in dust explosion hazards, and how these hazards can be alleviated by application of, for example, the inher ent safety principle of minimization through effective housekeeping. As illustrated by the analysis that follows Journal of Chemical Health & Safety, January/February 2010 25 Table 6. Elements of Process Safety Management, PSM.16. No. Element 1 Accountability: objectives and goals 2 Process knowledge and documentation 3 Capital project review and design procedures 4 Process risk management 5 Management of change 6 Process and equipment integrity 7 Human factors 8 Training and performance 9 Incident investigation 10 Company standards, codes and regulations 11 Audits and corrective actions 12 Enhancement of process safety knowledge for element 9, similar observations can be made for the other PSM elements in Table 6. In a previous paper, Amyotte et al.15 have demonstrated that three of Hop kins' four safety subcultures (reporting, just and learning) have an explicit link to the PSM element of incident inves tigation. With respect to the subject of the current paper, a company's com mitment to just and reporting cultures will ultimately be expressed in the number of dust explosion incidents that are reported. In the spirit of a leading indicator, it would be impor tant to also measure the number of near-miss and at-risk behaviour reports involving combustible pow ders. The outputs of reporting and investigating will manifest themselves as a measure of commitment to a learn ing culture; lessons learned is a key phrase in the industrial lexicon. The idea of learning from experience extends beyond the realm of incident investigation and into other safety management system elements (e.g., process risk management, manage ment of change, and enhancement of process safety knowledge). It may be tempting to dismiss talk of safety culture and safety management systems as being unimportant in rela tion to the chemistry, physics and engi neering of dust explosions. Further, safety culture might be viewed by some as merely the current `hot topic' in industrial safety-i.e., a relatively new concept that, given time, will be replaced by something else. On the first matter, the next section will hope fully dispel any notion that well-estab lished technical knowledge, in the absence of a strong safety culture and management system, is sufficient to ensure an acceptably low dust explo sion risk. On the second point, we offer the following quote from the last para graph of Count Morozzo's report2 describing the Turin flour warehouse explosion mentioned in the introduc tion: Ignorance of the fore-mentioned circum stances, and a culpable negligence of those precautions which ought to be taken, have often caused more misfor tunes and loss than the most contriving malice. It is therefore of great importance that these facts should be universally known, that public utility may reap from them every possible advantage. The above passage makes an elo quent case for the importance of a strong safety culture, incident investi gation, and the sharing of lessons learned. It is instructive to also note that it was written over 200 years ago. Case study--Westray coal mine explosion The Westray coal mine explosion occurred in Plymouth, Nova Scotia, Canada, on May 9, 1992, killing 26 miners.23 An indication of the destruc tive overpressures generated under ground can be seen in Figure 8, which shows surface damage at the mine site. The methane levels in the mine were consistently higher than regulations, which was caused by inadequate ventilation in the mine. Dust accumulations also exceeded per missible levels due to inadequate cleanup of coal dust; additionally, there was no crew in charge of rock dusting (inerting the coal dust with limestone or dolomite). These and many other factors contributed to the poor work conditions that continually existed in the Westray mine and made it the site of an incident waiting to happen. All of these substandard con ditions and practices could be attribu ted to the lack of concern that management had towards safety issues in the mine, which was one of the primary root causes of the problem at Westray. The question may arise as to whether a coal dust explosion in an under ground mine, which was initiated by a methane explosion, is a typical case in a dust explosion context. The choice of this particular case study is, how ever, regarded as fully justified given that it has been so well-documented29 and thoroughly analysed.23 Lack of adequate loss prevention and management is generally due to deficiencies in one or more of three 26 Journal of Chemical Health & Safety, January/February 2010 dust concentration, (iii) minimum (limiting) oxygen concentration of the atmosphere for flame propagation, (iv) minimum spark ignition energy, and (v) minimum dust cloud ignition temperature. In addition, knowing the minimum ignition temperatures of dust layers and dust deposits for var ious layer thicknesses and deposit volumes is important to prevent open and smoldering dust fires. A dust explosion occurs when an explosible dust cloud (consisting of an adequately mixed fuel and oxidant) is formed and ignited by a sufficiently energetic ignition source in a confined or partially confined environment. Such explosions originate as either pri mary or secondary events, and may also involve the co-presence of a flam mable gas (leading to the creation of a hybrid mixture). Dust explosion prevention and miti gation measures can be hierarchically organized from most to least effective in terms of measures related to: inher ent safety (minimization, substitution, moderation and simplification), pas sive engineered safety, active engi neered safety, and procedural safety. Consideration of all levels in this hier archy of controls is required for effec tive dust explosion risk reduction. Equally, if not more important to dust explosion control, is the key role played by senior management in ensur ing a strong safety culture and an effec tive safety management system. Technical knowledge without the commensurate management commit ment and program to ensure imple mentation of such knowledge is doomed to failure, as evidenced by the 1992 Westray mine explosion. ACKNOWLEDGEMENTS The authors wish to gratefully acknowl edge the funding agencies and indus trial companies that have supported their research efforts over the years. REFERENCES 1. NFPA 68. Standard on Explosion Protection by Deflagration Venting, 2007 Edition; National Fire Protec tion Association: Quincy, MA, 2007. 2. Morozzo, C. Account of a Violent Explosion Which Happened in the Flour-Warehouse, at Turin, December the 14th, 1785; To Which are Added Some Observations on Spontaneous Inflammations; From the Memoirs of the Academy of Sciences of Turin. The Repertory of Arts and Manufactures: London, 1795. 3. Piccinini, N. Account of a Violent Explosion; Politecnico di Torino; Tur in, Italy, 1996. 4. Eckhoff, R. K. Dust Explosions in the Process Industries, 3rd ed. Gulf Profes sional Publishing/Elsevier; Boston, 2003. 5. Faraday, M.; Lyell, C. Philos. Mag. 1845, 26, 16. 6. Eckhoff, R. K. J. Loss Prev. Process Ind. 2009, 22(1), 105. 7. CSB. Investigation Report - Dust Ex plosion - West Pharmaceutical Ser vices, Inc., Report No. 2003-07-I-NC; U.S. Chemical Safety and Hazard In vestigation Board: Washington, DC, 2004. 8. CSB. Investigation Report - Combus tible Dust Fire and Explosions - CTA Acoustics, Inc., Report No. 2003-09-IKY; U.S. Chemical Safety and Hazard Investigation Board: Washington, DC, 2005. 9. CSB. Investigation Report - Alumi num Dust Explosion - Hayes Lemmerz International-Huntington, Inc., Report No. 2004-01-I-IN; U.S. Chemical Safety and Hazard Investi gation Board: Washington, DC, 2005. 10. CSB. Investigation Report-Combusti ble Dust Hazard Study, Report No. 2006-H-1; U.S. Chemical Safety and Hazard Investigation Board: Washing ton, DC, 2006. 11. Frank, W. L. Process Saf. Progr. 2004, 23(3), 175. 12. Abbasi, T.; Abbasi, S. A. J. Hazard. Mater. 2007, 140(1-2), 7. 13. Eckhoff, R. K. Int. J. Chem. Eng., in press. 14. Amyotte, P. R.; Pegg, M. J.; Khan, F. I. Process Saf. Environ. Protect. 2009, 87(1), 35. 15. Amyotte, P.; Khan, F.; Demichela, M.; Piccinini, N. Paper 0114, in Ninth International Probabilistic Safety As sessment and Management Conference (PSAM 9), Hong Kong, China, 2008. 16. Amyotte, P. R.; Goraya, A. U.; Hendershot, D. C.; Khan, F. I. Process Saf. Progr. 2007, 26(4), 333. 17. Wilson, L.; McCutcheon, D. Industrial Safety and Risk Management; Univer sity of Alberta Press; Edmonton, AB, 2003. 18. BGIA. GESTIS-DUST-EX, Database: Combustion and Explosion Character istics of Dusts. Available at: http:// www.dguv.de/bgia/en/gestis/expl/index.jsp [March 18, 2009]. 19. Amyotte, P. R.; Pegg, M. J.; Khan, F. I.; Nifuku, M.; Yingxin, T. J. Loss Prev. Process Ind. 2007, 20(4-6), 675. 20. IChem, E. Dust Explosion Hazards, Video Training Package 022; Institu tion of Chemical Engineers; Rugby, UK, 1993. 21. Kauffman, C. W. Fuel-Air Explosions; Lee, J. H. S. & Guirao, C. M. Ed.,FuelAir Explosions; University of Waterloo Press: Waterloo, ON, Canada, 1982p. 305-347. 22. Amyotte, P. R.; Chippett, S.; Pegg, M. J. Progr. Energy Combust. Sci. 1988,14(4), 293. 23. Amyotte, P. R.; Oehmen, A. M. Process Saf. Environ. Protect. 2002, 80(1), 55. 24. Cashdollar, K. L. J. Loss Prevent. Process Ind. 2000, 13(3-5), 183. 25. Amyotte, P. R.; Khan, F. I.; Dastidar, A. G. Chem. Eng. Progr. 2003, 98(10), 36. 26. Hopkins, A.; Safety. Culture and Risk: The Organizational Causes of Disas ters; CCH Australia Limited; Sydney, Australia, 2005. 27. Kletz, T. A. Chem. Ind. 1978, 6, 287. 28. Center for Chemical Process Safety (CCPS). Inherently Safer Chemical Pro cesses: A Life Cycle Approach, 2nd ed.; John Wiley & Sons, Inc.: Hoboken, NJ, 2009. 29. Richard, K. P. Justice. The Westray Story-A Predictable Path to Disaster, Report of the Westray Mine Public Inquiry; Province of Nova Scotia: Halifax, NS, Canada, 1997. 28 Journal of Chemical Health & Safety, January/February 2010