Use of Combustible Dust Risk Assessments in the Agricultural and Food Processing Industries. by Douglas Moilanen

Transcription

1 1 Use of Combustible Dust Risk Assessments in the Agricultural and Food Processing Industries by Douglas Moilanen A Research Paper Submitted in Partial Fulfillment of the Requirements for the Master of Science Degree ill Risk Control Approved: 3 Semester Credits Dr. Elbert Sorrell The Graduate School University of Wisconsin-Stout December, 2010

2 2 The Graduate School University of Wisconsin-Stout Menomonie, WI Author: Moilanen, Douglas J. Title: Use of Combustible Dust Risk Assessments in the Agricultural and Food Processing Industries Graduate Degree/ Major: Master of Science in Risk Control Research Adviser: Dr. Elbert Sorrell Month/Year: December, 2010 Number of Pages: 69 Style Manual Used: American Psychological Association, 6 th edition Abstract Combustible dust has been recognized as a hazard in the agricultural and food processing industries for many years. However, the methods of assessing and controlling hazards in a particular facility are not widely recognized. The goals of this study were to provide a brief summary of dust explosion incidents; indentify the physical and chemical properties that affect the degree of hazard posed by combustible dusts; discuss the properties and behavior of dust clouds; outline the strategies available to control combustible dust hazards; and provide a framework for conducting combustible dust hazard assessments. To accomplish these goals, a literature review was conducted and recognized combustible dust experts were consulted. The research identified a number of steps necessary for conducting a combustible dust hazard assessment, a number of combustible dust parameters that affect the degree of hazard posed by a combustible dust, and a several strategies available for controlling combustible dust hazards.

3 3 Acknowledgements I would like to express my appreciation to Dr. Elbert Sorrell for his assistance in completing this graduate thesis. I would also like to extend thanks to Mr. Steven Luzik, Mr. Bill Stevenson, and Dr. Robert Zalosh for sharing their combustible dust expertise with me.

4 4 Table of Contents Page Abstract...2 Chapter I: Introduction...7 Statement of the Problem...9 Purpose of the Study 10 Goals of the Study 10 Definition of Terms..10 Limitations of the Study..13 Chapter II: Literature Review 14 The Fire Triangle and the Explosion Pentagon 14 Fuel...14 Particle Size and Shape 14 Moisture Content.17 Chemical Composition.18 Oxygen.21 Dust Cloud Concentration (Dispersion of Dust).21 Sources of Ignition...23 Types of Equipment Involved in Combustible Dust Explosions (Confinement).25 Hazard Mitigation Techniques.28 Controlling Sources of Ignition...28

5 5 Venting of explosive pressure.32 Explosion suppression.33 Prevention of Secondary Explosions...34 Chapter III: Methodology.36 Method of Study...36 Literature Reviewed.36 Personal Communications 37 Chapter IV: Results 40 Literature Review.40 Consultation with Combustible Dust Experts..42 Findings 44 Goal one...44 Goal two...45 Goal three.46 Goal four..47 Chapter V: Discussion..50 Summary..50 Problem 50 Purpose and Goals...50 Background and significance...50 Research design...51 Limitations...51

6 6 Conclusions..51 Recommendations 52 References..55 Appendix - Interview Summaries Mr. Steve Luzik 60 Mr. Bill Stevenson 63 Dr. Robert Zalosh.67

7 7 Chapter I: Introduction Several recent high profile combustible dust incidents have resulted in an increased awareness of the risks posed by combustible dusts. The most notable of these recent incidents being the February 2008 explosion and fire at the Imperial Sugar Company in Port Wentworth, Georgia. The explosion and subsequent fire at Imperial Sugar resulted in 14 fatalities, 36 injuries, and a total loss of the facility (U.S. Chemical Safety and Hazard Investigation Board [CSB], 2009). Catastrophic events involving combustible dusts, however, are not a newly recognized phenomenon. The hazards of combustible dust have been recognized since at least 1785 when Count Morozzo documented a wheat flour explosion in Turin, Italy (Eckhoff, 2003). Since then, thousands of additional events have occurred throughout the world. Many of these incidents have occurred in a segment of the economy often thought to not handle hazardous substances, the agricultural and food processing industries. However, many if not most materials, including food stuffs, can become explosive given the right conditions (Eckhoff). Between 1980 and 2005, the CSB (2006) identified 281 combustible dust incidents in the United States. Twenty four percent of the incidents occurred in the food products industries (CSB). As a result of the frequency of events in the food industry, the United States government s first regulatory attempt to reduce combustible dust hazards, the 1987 Occupational Safety and Health Administration (OSHA) grain handling standard, focused on a segment of the food industry. According to OSHA s 2003 assessment of the grain handling standard, the rule resulted in a substantial reduction in grain elevator explosions (OSHA). As a result of the

8 8 limited scope of the standard, however, combustible dust incidents continued unabated in other industries, including other segments of the food industry. In an attempt to fill the regulatory gap, OSHA issued a Combustible Dust National Emphasis Program (NEP) in 2007, reissued the NEP in 2008, and initiated combustible dust rulemaking activities in 2009 (OSHA, 2008; Office of Information and Regulatory Affairs, 2009). The NEP and proposed rule are largely based upon the contents of National Fire Protection Association (NFPA) standards including: NFPA 51B - Standard for Fire Prevention During Welding, Cutting, and other Hot Work. NFPA 61 Standard for the Prevention of Fires and Dust Explosions in Agricultural and Food Processing Facilities. NFPA 68 Standard on Explosion Protection By Deflagration Venting. NFPA 69 Standard on Explosion Prevention Systems. NFPA Recommended Practice for the Classification of Combustible Dusts and of Hazardous (Classified) Locations for Electrical Installations in Chemical Process Areas. NFPA 654 Standard for the Prevention of Fires and Dust Explosions from the Manufacturing, Processing, and Handling of Combustible Particulate Solids. NFPA 664 Standard for the Prevention of Fires and Dust Explosions in Wood Processing and Woodworking Facilities. However, the NFPA standards are complex, poorly understood, occasionally conflicting, and often misapplied. For example, the definition of combustible dust varies among the various standards that address combustible dust (NFPA, 2006; NFPA, 2007a; NFPA 2007b; NFPA,

9 9 2008b; NFPA, 2008c). In addition, the OSHA Combustible Dust NEP (2008) instructs compliance officers to enforce portions of NFPA 654 in agricultural and food processing facilities under the general duty clause without recognizing that NFPA 654 does not apply to agricultural or food processing facilities unless specifically referenced by NFPA 61 (NFPA, 2006; NFPA 2008b; OSHA, 2008). The complexities, poor understandings, generalizations, and conflicts result from the many, often transient, factors that affect the hazards associated with a particular combustible dust. Humidity, particle shape, particle size, and particle size distribution can change as a substance is handled and all can significantly affect the type and degree of hazard that a particular dust presents. Without an understanding of the factors that affect the hazards of combustible dust, abatement of combustible dust hazards and compliance with applicable standards would be difficult. Statement of the Problem The NFPA standards and most OSHA regulations permit the use of site specific hazard assessments in lieu of the prescriptive standards. However, most readers of the NFPA standards focus on the prescriptive portions of the NFPA standards. Site specific hazard assessments have the potential to offer more effective and lower cost hazard abatement options than prescriptive standards. However, many companies lack the understanding of the concepts behind combustible dust hazards required to perform a combustible dust hazard assessment.

10 10 Purpose of the Study The NFPA standards and most OSHA regulations permit the use of site specific hazard assessments in lieu of the prescriptive standards. The use of site specific hazard assessments will, potentially, offer lower cost hazard abatement options along with a better understanding of a company s dust hazards. The purpose of this study was to provide the basis of understanding required to conduct a site specific combustible dust hazard assessment. Goals of the Study The goal of this study was to: 1. Identify the physical and chemical properties that affect the degree of hazard posed by combustible dusts; 2. Discuss the properties and behavior of dust clouds; 3. Outline the strategies available to control combustible dust hazards; and 4. Provide a framework for conducting combustible dust hazard assessments. Definition of Terms Agricultural Dust - any finely divided solid agricultural material that is 420 micrometers (µm) or smaller in diameter (material passing through a U.S. No. 40 Standard sieve) that presents an explosion hazard when dispersed and ignited in air (NFPA, 2007, p. 5). Combustible Dust variable, see Chapter II dp/dt the maximum rate of pressure rise in a constant-volume explosion (Amyotte & Eckhoff, 2010, p. 18).

11 11 Explosion Severity - an empirical index comparing the explosion severity of a sample to the explosion severity of Pittsburgh coal dust (Committee on Evaluation of Industrial Hazards, 1980). It is determined using the formula: (P max (Sample) ) X (dp/dt Sample ) (P max (Pittsburgh Coal Dust) ) X (dp/dt Pittsburgh Coal Dust ) K st the volume-normalized (or standardized) maximum rate of pressure rise in a constant volume explosion (Amyotte & Eckhoff, 2010, p. 18). K st =(dp/dt) max X {Test Chamber Volume} 1/3 (CSB, 2009b) Ignition Sensitivity an empirical index comparing the ignition sensitivity of a sample to the ignition sensitivity of Pittsburgh coal dust (Committee on Evaluation of Industrial Hazards, 1980). It is determined using the formula: (MIT Pittsburgh Coal Dust ) X (MIE Pittsburgh Coal Dust ) X (MEC Pittsburgh Coal Dust) (MIT Sample ) X (MIE Sample ) X (MEC Sample) Layer Ignition Energy (LIE) - a measure of the sensitivity of a dust deposit to ignition by an electric spark (Chilworth Global, n.d.). It is conducted by placing a sample of material on a metal plate and introducing a spark of known energy from above (Chilworth Global).

12 12 Layer Ignition Temperature (LIT) a measure of the surface temperature necessary to ignite a dust layer (Chilworth Global, n.d.). ASTM E2021 describes the LIT testing procedure. In the procedure, a 5 mm thick, 100 mm in diameter dust sample is heated on a hot plate for ½ hour. The sample and hot plate temperature are monitored until the temperature where the dust layer is ignited is determined (Chilworth Global). Minimum Explosible Concentration (MEC) the minimum concentration of a combustible dust cloud that is capable of propagating a deflagration through a well dispersed mixture of the dust and air (American National Standards Institute, 2007a, p. 486). Minimum Ignition Energy (MIE) - the electrical energy discharged from a capacitor, which is just sufficient to effect ignition of the most easily ignitable concentration of fuel in air (American National Standards Institute, 2007b, p. 747). MIE is determined by dispersing a sample in a Plexiglas tube and introducing a spark of known energy (Janes, Carson & De Lore, 2008; Chilworth Global, n.d.). Tests are generally conducted in either a Harmann Tube or Mike 3 apparatus (Janes et al.) in accordance with American Society for Testing and Materials (ASTM) E2019, British Standard 5958, 1991 and European Standard: IEC : 1994 (Chilworth Global). Minimum Ignition Temperature (MIT) the minimum temperature at which a dust cloud will self ignite (American National Standards Institute, 2006, p. 474). MIT is determined by dispersing a sample in a Godbert-Greenwald Furnace (Eckhoff, 2003) (Chilworth Global, n.d.). Tests are generally conducted in accordance with American Society for Testing and Materials (ASTM) E1491 or European Standard : 1994 (Chilworth Global). P max maximum explosion overpressure generated in the test chamber (CSB, 2009b, p. 16).

13 13 Limitations of the Study The focus of this study was limited to the experiences of and implications upon agricultural and food processing facilities. Conditions unlikely to be found in an agricultural or food processing facility were not addressed. The effects of pressure on MIE and explosive violence, for example were not explored. Reactive chemistry, particulate toxicity, and conductive metals were similarly not addressed.

14 14 Chapter II: Literature Review The Fire Triangle and the Explosion Pentagon. In order for combustion to occur, fuel, oxygen and a source of ignition are all required. Without all three elements, a fire can not occur. This familiar concept is known as the fire triangle. In order for a dust explosion to occur, two additional elements, dispersion of dust particles and confinement of the dust cloud, are required (Cashdollar, 2000; OSHA, 2005). The following sections discuss each of the five elements. Fuel In a combustible dust incident, the fuel in the dust explosion pentagon is a combustible dust. As was noted by Eckhoff (2003), and The Center For Chemical Process Safety Of The American Institute Of Chemical Engineers (2005) most materials are capable of being explosive. Whether a particular material is explosible depends on a number of factors. According to Eckhoff (2003), a list of these factors would include: the particle size or specific surface area, the chemical composition of the dust, the moisture content of the dust, the distribution of particle sizes and shapes in the dust, and the degree of agglomeration. Particle Size and Shape. According to Amyotte and Eckhoff (2010), particle size is a dominant factor in dust explosion prevention and mitigation. Eckhoff (2003) used the example of wood to describe how the velocity of fuel oxidation is increased as the degree of material subdivision increases and the surface area available to react with the oxygen in the air increases. A large piece of wood will

15 15 burn slowly; small pieces of kindling will burn quickly; while dry wood dust, if suspended in air in the right concentration, can explode. Dusts can propagate a flame in two ways. They can propagate a flame directly or can vaporize and/or release pyrolysis gases that are flammable (Cashdollar, 2000). Both mechanisms tend to increase in velocity as particle size is reduced. The reason for the effects described by Cashdollar is the increase in reactive surface area available for oxidation, vaporization, and/or pyrolysis gas release. Particle size also has a significant effect upon the propensity for dust cloud formation (Cashdollar; Center For Chemical Process Safety Of The American Institute Of Chemical Engineers, 2005; Eckhoff, 2003). Finer particles are both easier to be put aloft and remain aloft for a longer period of time. Because of their inability to stay in suspension and the amount of surface area available for oxidation, particles larger than 500 µm generally do contribute to dust explosions (Calle, Klaba, Thomas, Perrin, & Dufaud, 2005). Historically, most definitions of combustible dusts have used a size definition near the size threshold noted by Calle et al. Most definitions have focused on material that will pass through a Number 40 US Standard Mesh Screen (a Number 40 US Standard Mesh Screen is designed to allow material 420 micrometers (µm) or smaller in diameter to pass through). For example: NFPA 61 (2008b, p. 5) defines agricultural dust as any finely divided solid agricultural material that is 420 micrometers (µm) or smaller in diameter (material passing through a U.S. No. 40 Standard sieve) that presents an explosion hazard when dispersed and ignited in air; NFPA 499 (2008a, p. 5) and NFPA 704 (2007c, p. 17) both define combustible dust as any finely divided solid material that is 420 micrometers (µm) or smaller

16 16 in diameter (material passing through a U.S. No. 40 Standard sieve) that presents an explosion hazard when dispersed and ignited in air; NFPA 664 (2007a, p. 7) defines deflagrable wood dust as wood particulate with a median diameter of 420 micrometers (µm) or smaller in diameter (i.e., material passing through a U.S. No. 40 Standard sieve) having a moisture content of less than 25 percent (wet basis) and defines dry nondeflagrable wood dust as wood particulate with a median diameter greater than 420 micrometers (µm) or smaller in diameter (i.e., material that will not pass through a U.S. No. 40 Standard sieve) having a moisture content of less than 25 percent (wet basis). However, starting with NFPA 654 in 2006, size designations have started to be removed from some of the NFPA definitions. NFPA 68 (2007b, p. 8), NFPA 69 (2008c, p. 8), and NFPA 654 (2006, p. 7) now define a combustible dust as a combustible particulate solid that presents a fire or deflagration hazard when suspended in air or some other oxidizing medium over a range of concentrations, regardless of particle size or shape. The OSHA National Emphasis Program (2008) has also adopted this definition. Prior to 2006 the definition of combustible dust in NFPA 654 (2005) was limited to material 420 micrometers (µm) or smaller in diameter (Technical Committee on Handling and Conveying of Dusts, Vapors, and Gases, 2005). Similarly, prior to 2008, the definition of combustible dust was limited to material 420 micrometers (µm) or smaller in NFPA 69 (2002a). Perhaps the most confusing definitions are found in NFPA 68. Prior to 2007, NFPA 68 did not define combustible dust (NFPA, 2002b). The 2002 version of NFPA 68 did, however, contain a definition for dust. Dust was defined as any finely divided solid, 420 µm or in. or less in diameter (that is, material capable of passing through a U.S. No. 40 Standard Sieve) (NFPA

17 b, p.7). In the 2008 version of the standard, the definition for combustible dust found in the 2006 version of NFPA 654 (in which the size designations have been removed) was added while the definition for dust from the 2002 version (with a size designation) remained unchanged. The NFPA Technical Committee Report (2005) and the explanatory material in NFPA 654 (2006) describe the changes in definition as necessary to address the hazards posed by platelet, flake, or fiber shaped particles. Eckhoff (2003) explained how flakes and fibers can fail to pass through a 40 mesh screen, yet be capable of forming explosive dust clouds and showed how the mass to surface area ratio decreased as a cube or sphere is compressed into a flake or stretched into a fiber using the concept of the specific surface area. According to Eckhoff, however, the effect is usually moderate except where the shapes are extreme such as in the case of thin flakes or long fibers. Moisture Content. The effects of moisture content on combustible dust hazards have long been known. In his account of the 1785 flour dust explosion in Turin, Count Morrozo wrote that the baker had never had flour so dry as in that year [1785], during which the weather had been remarkably dry (Eckhoff, 2003, p. 158). More recently, Traore, Dfaud, Perrin, Chezelet, and Thomas (2009) reported on the work of Li et al. (1995) that the dust explosion statistics in the United States between 1979 and 1986 show that most incidents occur during the low atmospheric humidity winter months. According to Eckhoff (2003), there are three ways that dust ignition sensitivity, explosive severity and propensity for dispersal are affected by moisture:

18 18 First, an internal heat sink is represented by the heating and vaporization of water; Second, oxygen and pyrolysis gasses are diluted by water vapor; and Third, water prevents dispersion by facilitating inter-particle cohesion. Eckhoff (2003) provided a summary of the work of van Laar and Zeeuwen showing the effect of moisture content on the measured MIE for several organic dusts. The results showed that increasing moisture caused a significant increase in MIE for tapioca, maize starch, and flour dust clouds. The MIE results were presented in the form of a small log scale graph and the original work by van Laar and Zeeuwen was not obtained. As a result, only approximate values are presented here. The approximate MIE results were approximately 20 mj, 200 mj, and 7000 mj for 1%, 7% and 10% moisture tapioca respectively; approximately 40 mj, 80 mj, 150 mj, and 300 mj for 1%, 7%, 10%, and 15% moisture flour respectively; and approximately 200 mj, 400 mj, and 2000 mj for 1%, 7% and 14% moisture maize starch respectively. Similarly, Traore et al. (2009) looked at the effect of moisture content on the MIE of an artificial organic ingredient in hard candy, magnesium stearate. The results showed that the MIE of magnesium stearate was increased by a factor of three when its moisture content was increased from near zero to 7%. The aforementioned work of van Laar and Zeeuwen summarized by Eckhoff (2003) also showed the effect of moisture content on the MIT of organic dusts. The results showed an increase in MIT with increasing moisture for starch and flour dust clouds although the effect of moisture on MIT was less pronounced than the effect on MIE. The MIT for 14% moisture flour was 470 C compared to 440 C for dry flour. The MIT for 14% moisture starch was 460 C compared to 400 C for dry starch. It is not known what percent moisture was considered dry. Traore et al. (2009) also looked at the effect of moisture content on the MIT of magnesium

19 19 stearate and showed that the MIT of magnesium stearate increased from approximately 420 C to approximately 570 C when its moisture content was increased from near zero to 7%. Eckhoff (2003) and Traore et al. (2009) also showed how increasing moisture content decreases the severity of an explosion for maize starch and for magnesium stearate and icing sugar respectively. The Eckhoff data showed that the maximum rate of pressure rise for Maize starch decreased approximately 100 bar/s as the moisture content was increased from near zero to about 28%. The maximum rate of pressure rise for the dry starch ranged from slightly less than 100 bar/s to slightly less than 200 bar/s depending on the ignition delay time. For the 28% moisture material, the maximum rate of pressure rise ranged from approximately 20 bar/s to slightly less than 100 bar/s. The reduction was more pronounced at lower moisture percentages than at higher percentages. The Traore et al. data for magnesium stearate showed a decrease in the maximum rate of pressure rise from 1035 bar/s to 640 bar/s as moisture content was increased from near zero to 4%. The data showed very little change as moisture content was increased from 4% to 7%. The Traore et al. data for icing sugar was more complex. Their data for icing sugar showed an increase in the maximum rate of pressure rise as the moisture content rose from near zero to 0.34% percent followed by a subsequent decrease as the moisture content is increased further. The authors hypothesized that the initial rise in explosion severity of sugar could be the result of water reactions that transition the sucrose to glucose and fructose. Traore et al. acknowledged that, generally, humidity decreases explosion severity and ignition sensitivity but warn that water can also cause reactions such as those shown with sugar or in the release of combustible gas during the fermentation of grain dust and should not be overlooked. Eckhoff (2003) also described how agglomerations of small particles can be very difficult to break apart and that persistent agglomerations behave like larger particles in dust explosions.

20 Bryant (1973) showed that stable particle agglomerates behave like a single particle the size of the agglomerate and result in long burn times. 20 Chemical Composition. Dust explosions, in general, follow the chemical formula Fuel + Oxygen Oxide + Heat (Abbasi & Abbasi, 2007) The pressure rise experienced in a dust explosion will be governed by the ideal gas law and the heat of combustion of the product (Abbasi & Abbasi, 2007). The Committee on Evaluation of Industrial Hazards (1980) published ignition sensitivity, explosion severity, P max, dp/dt, MIT, LIT, MIE, and MEC for 72 agricultural and food products. Most of the listed substances were either classified as moderate or strong with respect to ignition sensitivity; but the values ranged all the way from weak to severe (0.1 to 8.5). Similarly, most of the listed substances were either classified as moderate or strong with respect to explosion severity; but the values ranged from weak to severe (0.1 to 5.4). P max ranged from 0.5 bar to 8 bar. The dp/dt ranged from 10 bar/s to 700 bar/s. MITs ranged from 350 C to 720 C. LITs ranged from 180 C to 470 C. The majority of the MIEs listed fell between 25mJ and 320mJ. One of the substances was an order of magnitude higher than 320mJ and five substances were not successfully ignited with an electric spark. Oxygen. In chemical processes, reactive chemistry may provide oxygen or other oxidizers. In food processing facilities, however, the only source of oxygen is likely to be atmospheric oxygen. Elevated oxygen will enhance the explosion severity and ignition sensitivity of

21 21 combustible dusts while reduced oxygen will reduce explosion severity and ignition sensitivity (Eckhoff, 2003; Center For Chemical Process Safety Of The American Institute Of Chemical Engineers, 2005). Dust Cloud Concentration (Dispersion of Dust) Flammable gasses will only burn at concentrations where the concentration of fuel gas is high enough to support combustion but not so high that there is no longer enough oxygen to support combustion. This concept, expressed as the lower flammable limit (LFL) and upper flammable limit (UFL), is well established. Combustible dusts, similarly, are only explosive in a narrow concentration band. The lower boundary of this band, analogous to the LFL, is referred to as the minimum explosive concentration (MEC). As with the LFL for flammable gasses, the MEC varies for different dusts. NFPA 61 (2008b) and NFPA 68 (2008c) show MEC for various combustible dusts. The MEC for the listed dusts were generally between 60 g/m 3 and 250 g/m 3. There were two outliers on the list in NFPA 61, however. The listed MEC for dehydrated parsley was 26 g/m 3 while the listed MEC for oat grain dust was 750 g/m 3. It is worth noting that it unclear from the information provided in the table if the MEC for either outlier have MECs outside of the range other agricultural commodities in general or if the MECs observed were the result of a property exhibited by the specific samples tested. Moisture percentage data was only provided for a small fraction of the substances for which MEC data was listed. Particle size data was provided for the majority of the substances for which MEC data was listed; but was not provided for the dehydrated parsley. Eckhoff (2003) noted that the MEC for combustible dusts is several orders of magnitude higher than the concentrations of concern to industrial hygienists. The threshold limit value

22 22 (TLV ) for inhalable particulates not otherwise specified established by the American Council of Governmental Industrial Hygienists (2005) is 10 mg/m 3. As was previously noted, the lowest MEC reported for the substances listed in NFPA 61 (2008) was dehydrated parsley. The reported value of 26 g/m 3 is three orders of magnitude greater than the 10 mg/m 3 TLV value, which supports Eckhoff s assertion. As a result of the high concentrations required for a dust explosion to occur, explosible concentrations are not normally present outside of processing equipment. However, if accumulations of settled dust are present in the workplace, a primary explosion can dislodge and elevate enough combustible dust to create an explosible dust cloud outside of processing equipment. Because the explosible range of a dust cloud ranges from around 50 g/m 3 to several kg/m 3 while the bulk density of dust settled in a layer or a heap, range from several hundred kg/m 3 and above, a seemingly small accumulation of settled dust is capable of forming a very large dust cloud if suspended (Eckhoff, 2003). If the dust suspended by the initial event is ignited, a secondary explosion may result. As a result of the potentially larger quantity of dust involved, the secondary explosion can be far more destructive than the initial event (OSHA, 2005). Numerous examples are described in the literature. A few examples include: In February 1999, a fire in a molding machine spread to the ductwork of a Massachusetts foundry. The deflagration that resulted inside the ductwork shook dust that had settled on the ductwork free. Ignition of the dust lofted by the initiating event provided sufficient fuel to lift the roof and cause the failure of the walls. Three people were killed and nine were injured (OSHA).

23 23 In February 2003, a small fire in an unattended oven ignited a dust cloud created by the cleaning of a nearby line in a Kentucky fiberglass plant. A cascade of explosions resulted throughout the plant that killed nine and injured 37 (OSHA). In October 2003, an explosion inside a dust collector traveled through the ductwork and suspended and ignited aluminum dust inside an Indiana wheel manufacturing plant. One employee was killed and several were injured (CSB, 2006). In February 2008, an explosion occurred in a conveyor beneath several large sugar silos in a Georgia sugar refinery. The explosion caused a series of secondary explosions that progressed through several packaging buildings, a palletizer room, and a bulk storage facility. The blasts blew out walls and buckled concrete floors. Fourteen were killed and 36 were injured (United States Chemical Safety and Hazard Investigation Board, 2009a). Sources of Ignition In order for a dust explosion to occur, a source of ignition and dust cloud must be simultaneously present (Cashdollar, 2000). According to Eckhoff (2003) the most common sources of ignition for combustible dust explosions are: burning or smoldering material, heat from mechanical impacts, electrical arcs or discharges, hot work and hot surfaces such as those found on dryers heaters, overheated bearings, etc. Burning or smoldering material can occur as a result of exothermic reactions in stored material, as a result of one of the other ignition sources noted by Eckhoff (2003), or as a result of intentional heating. In the case of food stuffs, one possible exothermic reaction is biological

24 24 degradation. (Eckhoff) Insulation by overburden material can allow biologically produced heat to raise internal temperatures above a specific material s LIT and/or MIT. The resulting smoldering nest may ignite a dust cloud when it is exposed to oxygen, when it is conveyed out of storage, or when overburden product is otherwise removed. (Eckhoff, 2003; Center For Chemical Process Safety Of The American Institute Of Chemical Engineers, 2005) Single impacts are capable of creating an impact spark. However, the evidence does not indicate that a single impact is capable of producing enough energy to ignite a dust cloud or layer (Abbasi & Abbasi, 2007; Eckhoff, 2003). Repeated impacts, such as those seen in a hammer mill or a damaged bucket elevator where a single spot is repeatedly impacted can eventually generate enough heat in the metal to ignite a dust cloud or layer (Abbasi & Abbasi; Eckhoff). Tramp metal in hammer mills can be rapidly heated by impacts and transported downstream where clouds can be ignited. The fact that electric discharges are capable of igniting dust clouds and layers has been long known and is well established. Electrical discharges vary widely in both source and energy (Eckhoff, 2003). Sparks generated as a result of an inductive spark or break flash that occurs when a live electric circuit is broken can be energetic enough to ignite dusts (Eckhoff). Some types of electrostatic discharges are also capable of igniting dust clouds (Eckhoff). Electrostatic discharges in industrial facilities occur when electrostatic charges are not safely dissipated prior to a discharge (Glor, 2003). Glor described eight different types of electrostatic discharges: spark discharges, brush discharges, corona discharges, propagating brush discharges, brush and corona discharges, lightning like discharges, and cone discharges. According to Glor, the evidence suggests that brush discharges, corona discharges, brush and corona discharges, and lighting like discharges may be capable of igniting flammable gasses and hybrid mixtures; but

25 25 are not capable of igniting pure dust clouds. Spark discharges, propagating brush discharges, and cone discharges, however, are capable of igniting combustible dust clouds (Glor, 2003). Hot work such as welding and flame cutting are also capable of igniting a dust cloud or dust layer. The risk of ignition as a result of hot work is greatest when a dust with a MIT below 200 C is located nearby (Abbasi & Abbasi, 2007). Incidents frequently occur as a result of a lack of hazard recognition and failure to remove dust deposits from inside equipment before hot work is initiated. Hot surfaces such as hot bearings, heating units, and dryers are also capable of igniting dust accumulations and or dust clouds. According to the United States Chemical Safety and Hazard Investigation Board (2009a) an overheated bearing on a conveyor initiated the February 2008 explosion at Imperial Sugar in Port Wentworth, GA. Types of Equipment Involved in Combustible Dust Explosions (Confinement) As was explained in an earlier section, the MEC for combustible dusts is several orders of magnitude higher than the concentrations of concern to industrial hygienists (American Council of Governmental Industrial Hygienists, 2005; Eckhoff, 2003). Eckhoff also presented a rule of thumb based upon obscured vision. It states that a 25 watt bulb is not visible at 2 m through a 40 g/m 3 dust cloud. These conditions would certainly not be safe or pleasant in an occupied work environment. As a result, explosible dust clouds are not normally present outside of processing equipment. It is not unusual for explosible concentrations to be present inside processing equipment, however. The number of explosion events caused by different equipment types was listed in Factory Mutual Insurance Company (2009), Center For Chemical Process Safety Of The

26 26 American Institute Of Chemical Engineers (2005), and Eckhoff (2003). The Factory Mutual Insurance Company report lists explosions occurring at its insured customer locations between 1983 and The Center For Chemical Process Safety Of The American Institute Of Chemical Engineers data includes insurance company data (including Factory Mutual) for explosions in the United States from 1985 through 1995 as well as Health and Safety Executive (HSE) data for explosions in the United Kingdom from 1979 through The Eckhoff data includes explosions in the Federal Republic of Germany from 1965 through 1985.

27 27 Figure 1 Equipment Type United States ( ) Factory Mutual ( ) United Kingdom ( ) Federal Republic of Germany ( Total* Number Number Number Number Number Percent Dust collectors Mills, grinders, ect Silos and bunkers Dryers, ovens, ect Conveyor systems Mixers and blenders Other Total * United States and Factory Mutual Information includes overlapping data The data in Table 1 includes only a small fraction of the actual incidents occurring during the given time in the indicated countries (Center For Chemical Process Safety Of The American Institute Of Chemical Engineers, 2005). The Center For Chemical Process Safety Of The American Institute Of Chemical Engineers noted that only three of 84 incidents reported in a voluntary survey in the United Kingdom had been reported to the HSE. The average of one dust explosion per day in Germany reported by Glor (2003) supports the Center For Chemical Process

28 28 Safety Of The American Institute Of Chemical Engineer s assertion that the number of reported incidents is low; as does the Calle et al. (2005) report that France s experience is similar to that of Germany. The data in Table 1, however, can be used to illustrate some of the equipment types that have been involved in combustible dust explosions. According to the data provided in Table 1 and the United States Chemical Safety and Hazard Investigation Board (2006), dust collectors are involved in more combustible dust incidents than any other type of equipment. The Center For Chemical Process Safety Of The American Institute Of Chemical Engineers (2005) provided three likely explanations for the frequency of dust collector involvement. 1. Dust collectors are present in nearly all facilities that handle particulate solids; 2. Dust collectors, by design concentrate the finer fractions of the material being handled; and 3. Dust collectors are often not as structurally robust as other equipment in a given process. Most of the equipment found in industrial processes, however, is not strong enough to withstand the pressure generated by an unvented dust deflagration or explosion (Eckhoff, 2003). Hazard Mitigation Techniques Controlling Sources of Ignition. As was previously discussed, a fire can not occur without three things: fuel, oxygen or other oxidizer, and a source of ignition. These three things, along with confinement and dispersion of a dust cloud, are necessary for a dust explosion (Cashdollar, 2000; OSHA, 2005). Control of potential ignition sources is, therefore, important. Several common sources of

29 29 ignition were also discussed previously. The sources included: burning or smoldering material, impact heating, electrical discharges, hot work, and hot surfaces. Control of intentionally heated product and control of storage conditions is paramount in preventing burning or smoldering material (Eckhoff, 2003). Material that is left in storage for extended periods of time can self heat as a result of decomposition processes (Eckhoff). These decomposition processes can be accelerated by elevated initial temperatures or by the presence of excess moisture (Traore et al.,2009). Overburden material can insulate material undergoing the decomposition process and allow temperatures to exceed the MIT of the material being stored. The resulting smoldering nest can ignite a dust cloud when it is removed from storage or overburden material is removed. Smoldering nests can be prevented by ensuring that product has cooled sufficiently prior to storage; that product stockpiles are adequately turned (Eckhoff, 2003); and that stagnant zones are eliminated in bins and silos (Eckhoff, 2009). Impact heating is capable of generating enough heat to ignite a dust cloud. The evidence suggests, however, that single impacts generally do not generate sufficient energy for ignition (Abbasi & Abbasi, 2007; Eckhoff, 2003). Repeated impacts, however, can. As a result, removal of tramp metal ahead of hammer mills and implementation of certified preventive maintenance programs are required by the OSHA grain handling standard (OSHA, n.d.). Electrostatic discharges of certain types are capable of producing sufficient energy to ignite a combustible dust cloud (Eckhoff, 2003; Glor, 2003). According to Glor, most organic substances are capable of being charged even if handled in properly grounded bonded metallic equipment. However, relatively low currents are generally required to safely dissipate static charges that develop in industrial processes (Glor). As a result, resistances to ground as high as

30 ohm are usually sufficient to allow safe dissipation (Glor). Static discharge problems can be avoided by ensuring that equipment is properly bonded and grounded. Sparks generated as a result of an inductive spark or break flash that occurs when a live electric circuit is broken or as a result of electrostatic discharges are capable of igniting dust clouds (Eckhoff, 2003). NFPA 499 (2008a) provides guidance on Hazardous Area Classification for electrical installations where combustible dust is present. It defines a Class II, Division 1 location as a location: 1. In which combustible dust is in the air under normal operating conditions in quantities sufficient to produce explosive or ignitable mixtures, or 2. Where mechanical failure or abnormal operation of machinery or equipment might cause such explosive or ignitable mixtures to be produced, and might also provide a source of ignition through simultaneous failure of electric equipment, through operation of protection devices, or from other causes, or 3. In which combustible dusts of an electrically conductive nature may be present in hazardous quantities (NFPA, 2008a, p. 5). It defines a Class II, Division 2 location as a location: 1. Where combustible dust is not normally in the air in quantities sufficient to produce explosive or ignitable mixtures, and the dust accumulations are normally insufficient to interfere with the normal operation of electrical equipment or other apparatus, but combustible dust may be in suspension in the air as the result of infrequent malfunctioning of handling and processing equipment and 2. Where combustible dust accumulations on, in, or in the vicinity of the electrical equipment may be sufficient to interfere with the safe dissipation of heat from

31 31 electrical equipment or may be ignitable by abnormal operation or failure of electrical equipment (NFPA, 2008a, p. 6). NFPA 499 (2008a) requires that electrical installations in classified locations must conform to the protection requirements in article 500 of the National Electric Code. Work that is capable of producing sparks, flame, or heat is referred to as hot work. NFPA 61 (2008b) and NFPA 654 (2006) both recognize hot work as a source of ignition and reference NFPA 51B as a source of guidance. NFPA 51B (2009, p. 4) applies to the following hot work processes: 1. Welding and allied processes 2. Heat treating 3. Grinding 4. Thawing pipe 5. Powder driven fasteners 6. Hot riveting 7. Torch applied riveting in conjunction with the requirements of NFPA 241, Standard for Safeguarding Construction, Alteration, and Demolition Operations 8. Similar applications producing or using a spark, flame, or heat OSHA s Welding, Cutting, and Brazing Standard incorporates the 1962 version of NFPA 51B by reference (OSHA, n.d.). NFPA 51B (2009) requires that hot work be conducted using a permit system that ensures that combustible dust be removed from a 10m radius surrounding the work area. NFPA 51B also requires that conveyor or dust systems that could carry sparks or hot material to remote locations be shut down prior to conducting hot work.

32 32 Surfaces that are heated to temperatures above the LIT or MIT of materials handled in a facility can also be a source of ignition. Surfaces that are intentionally heated or inherently hot may need be insulated or removed from a process area. Malfunctioning equipment, however, is also capable of producing enough heat to ignite combustible dusts (Eckhoff, 2003). It is suspected that the 2008 Imperial Sugar Dust Explosion was caused by an overheated conveyor belt bearing (CSB, 2009a). As a result of the hazard, NFPA 61 (2008b), NFPA 654 (2006), and OSHA s Grain Handling Standard (OSHA, n.d.) all contain guidance and/or requirements intended to prevent, reduce the frequency, or quickly identify heat generating equipment malfunctions. Venting of explosive pressure. Previously in this document, it was discussed that, due to the high concentration of dust required to form an explosive dust cloud, primary dust explosions usually occur inside processing or storage equipment. Often, this equipment is not robust enough to withstand the internal pressures that can be generated in a dust explosion (Center For Chemical Process Safety Of The American Institute Of Chemical Engineers, 2005). According to Eckhoff (2003), with certain exceptions, strengthening equipment to withstand a dust explosion is generally not the most cost effective method of hazard abatement. Exceptions mentioned by Eckhoff included, certain types of mills that are already heavy and cylindrical ducting. One method of mitigating this hazard is to provide a means to vent explosion gasses to a safe location (Factory Mutual Insurance Company, 2009). A vent is defined by NFPA 68 (2007b, p. 9) as an opening in an enclosure to relieve the developing pressure from a deflagration. Types of explosion venting described in NFPA 68

33 33 include: louvered openings, hanger-type doors, hinged doors, hinged panels, hinged widows, shear panels, and rupture diaphragm devices. In order to properly size a vent system, the K st of the product must be known (NFPA, 2007b). It is preferable that explosion vents lead directly to a safe, normally unoccupied location outdoors without the use of ductwork (Factory Mutual Insurance Company, 2009). Venting ductwork reduces venting efficiency and must be taken into account if present (Eckhoff, 2003). In addition, the concept of pressure piling as described by the Center For Chemical Process Safety Of The American Institute Of Chemical Engineers (2005) must be considered. According to the Center For Chemical Process Safety Of The American Institute Of Chemical Engineers, pressure piling occurs when an explosion occurs inside a member of a series of process equipment interconnected by relatively narrow passages. Pressure piling results in pressures greater than expected because explosions in adjacent equipment can raise internal pressures prior to the arrival of a flame front. In many cases pressure increases as the explosion travels from one location to the next, increasing the damage (CSB, 2006, p. 14). Explosion suppression. Another method of mitigating the hazard of internal dust deflagrations is to suppress an explosion in its early stages (Factory Mutual Insurance Company, 2009). Explosion suppression systems consist of an explosion detection device or system, a control device or system, and a suppressant delivery device or system (Factory Mutual Insurance Company, 2001). Detection options include optical detectors and pressure sensors (Factory Mutual Insurance Company, 2001). Upon detection of a deflagration in the incipient stage, the detection system generally triggers a high rate discharge of fire suppressant (Factory Mutual Insurance Company, 2001).

34 34 Common suppressants include water, halon gas, and powder suppressants (Factory Mutual Insurance Company, 2001). NFPA 69 (2008c) contains guidance on the design, installation, testing, and maintenance of explosion suppression systems. Prevention of Secondary Explosions. Many if not most of the catastrophic dust explosions experienced by industry have involved secondary explosions (CSB, 2006; Eckhoff, 2003). Secondary explosions can be far more destructive than the initial event if large amounts of dust are lofted by the initiating event (OSHA, 2005). Prevention of secondary explosions is generally accomplished by ensuring that dust accumulations that could become fuel for a secondary explosion are removed via a housekeeping program. The OSHA Grain Handling Standard (n.d.), OSHA s Combustible Dust National Emphasis Program (2008), NFPA 61 (2008b), and NFPA 654 (2006) all include housekeeping requirements. The OSHA Grain Handling Standard requires that employers develop written housekeeping plans to reduce surface accumulations of dust. The Grain Handling Standard also requires that, in grain elevators, accumulations of grain dust be kept free of dust accumulation exceeding 1/8 in depth in priority areas. NFPA 654 requires that areas where greater than 1/32 of dust accumulation is present be considered to posses a dust explosion hazard unless the bulk density of the dust is less than 75 lb/ft 3. For dust accumulations less than 75 lb/ft 3 the allowable thickness under NFPA 654 is calculated using the formula: Allowable Thickness = (1/32)(75) Bulk density (lb/ft 3 )

35 35 NFPA 61 (2008b) requires that dust accumulations be removed concurrently with operations but does not specify acceptable depths of accumulation. The explanatory material in NFPA 61, however, points the reader to NFPA 654 for more information. According to the explanatory material in NFPA 61, most agricultural dusts have a bulk density much lower than 75 lb/ft 3. The Combustible Dust National Emphasis Program references the 1/32 standard and bulk density calculations found in NFPA 654 as well.

36 36 Chapter III: Methodology Method of Study The primary objective of this study was to provide a basis of understanding required to conduct site specific combustible dust hazard assessments. In order to accomplish this objective, it was necessary to: identify the primary physical and chemical properties that affect the degree of hazard posed by combustible dusts, discuss the properties and behavior of dust clouds, and outline some of the strategies available to control combustible dust hazards. The primary method used to accomplish the objective was a review of relevant literature. Literature reviewed included: textbooks, scholarly journals, OSHA regulations and background information, current and out dated NFPA standard documents, ASTM standards, CSB technical reports, and insurance industry guidance. In addition, telephone interviews were conducted with several recognized combustible dust experts. Each of the experts had at least 30 years of experience dealing with combustible dust. In addition, each of the experts sat on one or more NFPA technical committees. Literature Reviewed Two text books and several journal articles were reviewed that provided detailed discussions of the broad range of topics that are required to understand combustible dust hazards and combustible dust hazard abatement techniques. The broad range of topics included but was not limited to: combustible dust incident analysis, combustion science, powder technology, ignition energy, control technology, and laboratory analysis techniques. In addition, numerous journal articles that focused more narrowly on the aforementioned topics were reviewed.