Case histories 189 Figure 2.25 Sequence of nine impact energy pulses from nine successive explosions in the Harbin Linen Textile Plant, Harbin, P. R. China, 15th March 1987, postulated on the basis of a seismic record of the event (From Xu Bowen et al., 1988) 190 Dust Explosions in the Process Industries 2.9.3 EXPLOSION INITIATION AND DEVELOPMENT, SCENARIO 2 This alternative scenario originates from the investigation of Zhu Hailin (1988), who found evidence of an initial smouldering dust fire caused by a live 40 W electrical portable light lamp lying in a flax dust layer of 6-8 cm thickness in a ventilation room. He also found evidence of flame propagation through the underground tunnels for the dust collection ducting. On the basis of his analysis, Zhu suggested that the explosion was initiated in the eastern dust collectors (5 in Figure 2.24) from which it transmitted to nine units of the central dust collecting plant (1 and 2 in Figure 2. 24) via the ducting in the underground tunnels. Severe room explosions were initiated when the ducting in the tunnel ruptured, and the resulting blast dispersed large quantities of dust in the workrooms into explosible clouds that were subsequently ignited. From the eastern dust collectors the explosion also propagated into the underground flax stores. It is not unlikely that even this scenario could be developed further in such a way as to agree with the evidence from the seismic recording. 2.9.4 ADDITIONAL REMARK The investigation of the Harbin disaster exposed the great difficulties in identifying the exact course of events of major explosions creating massive damage. In addition to causing pain and grief, loss of life also means loss of eye witnesses. Besides, the immediate need for fire fighting and rescue operations, changes the scene before the investigators can make their observations. Also, the explosion itself often erases evidence, e.g. of the ignition source. This problem was also shared by the experts who investigated the Harbin explosion, and it seems doubtful that the exact course of events will ever be fully resolved. However, the Harbin disaster unambiguously demonstrated the dramatic consequences of inadequate housekeeping in industrial plants where fine dust that can give dust explosions, is generated. 2.1 0 FIRES AND EXPLOSIONS IN COAL DUST PLANTS 2.1 0.1 METHANE EXPLOSION IN 17000 rn3 COAL SILO AT ELKFORD, BRITISH COLUMBIA, CANADA, IN 1982 As mentioned in Section 1.5, handling and storage of coal can, in addition to the dust explosion hazard, also present a gas explosion risk, due to release of methane from some types of coal. An account of such an explosion was given by Stokes (1986). The silo of height 48 m and diameter 21 m that exploded, was used for storage and load-out of cleaned, dried metallurgical coal. The capacity of the silo was 15000 tonnes. Case histories 19 1 Prior to the explosion accident, a methane detector had been installed in the roof of the silo. The detector activated a warning light in the silo control room when a methane concentration of 1% was detected, and an alarm light was activated when detecting 2% methane. A wet scrubber was located in the silo head house to remove dust from the dust-laden air in the silo during silo loading. A natural ventilation methane stack was also located in the silo roof to vent any build-up of methane gas from the silo. The explosion occurred early in the morning on 1st May, 1982, devastating the silo roof, head house, and conveyor handling system. Witnesses stated that a flash was noticed in the vicinity of the head house, followed seconds later by an explosion which displaced the silo top structures. This was followed by an orange-coloured fire ball that rolled down the silo walls and extinguished prior to reaching the base of the silo. Fortunately, neither injury nor death resulted, and damage to surrounding structures was minimal, although large blocks of concrete and reinforcing steel had been thrown several hundred metres from the silo. However, the plant itself had suffered substantial damage. The silo was full of coal 24 hours prior to the explosion. During the evening before the explosion, 10 OOO tonnes of coal were discharged. At the same time, conveying of deep-seam coal into the silo commenced and continued until the explosion occurred. At the time of the explosion, there were approximately 12 300 tonnes of coal in the silo, of which 7600 tonnes were deep-seam coal. Testing had shown that this quality of coal has a high methane emission rate and produced a low volatile coal dust. Clouds in air of this dust could not be ignited unless the air was mixed with methane. The ignition source was not identified, but the following three possible sources were considered: 0 Spontaneous combustion of the stored coal. An electrical or mechanical source. Hot coal from the thermal dryer. During ten years of operation, with coal being stored in different environments for varying lengths of time, spontaneous combustion had never presented a problem, and consequently was not considered to be a probable source of ignition. During demolition of the damaged silo, all electrical and mechanical components were recovered and inspected and did not show any evidence of being the ignition source. Stokes (1986) did not exclude the remaining possibility that hot coal from the thermal dryer was the source of ignition. 2.1 0.2 METHANVCOAL DUST EXPLOSION IN A COAL STORAGE SILO AT A CEMENT WORKS AT SAN BERNARDINO COUNTY, CALIFORNIA, USA This incident was reported by Alameddin and Foster (1984). A fire followed by an explosion occurred inside a coal silo of 900 tonnes capacity while the silo was nearly empty, and the remaining 85 tonnes of coal were being discharged. Prior to the explosion, a hot-spot of 0.6 m X 1.0 m had been detected on the lower part of the silo wall by means of an infrared heat detector. The hot-spot originated from smouldering combustion in the coal in the silo. This process liberated methane, carbon monoxide and other combustible gases from the coal. The explosion probably resulted from ignition of a mixture of combustible gas and airborne coal dust in the space above the bulk coal by the 192 Dust Explosions in the Process industries smouldering fire or glow when it reached the surface of the coal deposit. (See Figure 1.9 in Chapter 1. ) It was concluded that the supply of carbon dioxide from the top, which was used for suppressing the fire and preventing explosion, was insufficient to prevent the development of an explosible atmosphere in the space above the bulk coal. In order to prevent similar accidents in the future, it was recommended that a carbon dioxide system be installed in both the top and bottom of the coal silo. Sufficient inerting gas should be added for development of a slight positive pressure inside the silo. The inerting gas must be of sufficient quantity to insure a nonexplosible atmosphere above the coal and sufficient pressure to prevent a sudden inrush of fresh air into the silo. 2.1 0.3 GAS AND DUST EXPLOSION IN A PULVERIZED COAL PRODUCTION/ COMBUSTION PLANT IN A CEMENT FACTORY IN LAGERDORF IN F. R. GERMANY, IN OCTOBER 1980 According to Patzke (1981), who described this explosion accident, the explosion occurred while coal of about 30% volatiles was milled at a rate of 55 tonnes per hour. The start-up of the cement burner plant followed a compulsory break of at least 20 minutes of the milling operation to allow all airborne dust to settle out. A few seconds after the main gas valve had been opened, there was a violent explosion. The probable reason was a failure in the system for electric ignition of the gas. Within the period of six seconds before the gas valve was reclosed automatically, about 1 m3 of gas had been discharged to the atmosphere of the hot combustion chamber and become mixed with the air to an explosible gas cloud. The temperature of the walls of the chamber was sufficiently high to ignite the gas, and a gas explosion resulted. The blast and flame jet from this comparat- ively mild initial explosion was vented into the milling system where a large, turbulent dust cloud was generated and ignited, resulting in a violent secondary dust explosion. Various parts of the milling plant, some unvented and some vented, had all been designed to withstand the pressure generated in an extensive dust explosion. Furthermore, a passive device for explosion isolation of the type shown in Figure 1.82 in Chapter 1 had been installed upstream of an electrostatic dust filter. Apart from deformation of some explosion vent doors, the dip tubes of two cyclones, and the coal feeder upstream of the mill, the plant had been able to withstand the explosion without being damaged. The passive explosion isolation device effectively protected the electrostatic filter from becoming involved in the system. 2.1 0.4 FURTHER EXPLOSION/FIRE INCIDENTS INVOLVING COAL Anderson (1988) gave a step-by-step account of the process of extinction of a smould- ering fire in a 50 m3 coal dust silo in Arvika in Sweden, in August 1988. It was necessary to pay attention to the risk of explosion of combustible gases driven out of the coal by the heat from the fire. Case histories 193 First gaseous carbon dioxide was loaded into the silo at the top to build up a lid of inert atmosphere immediately above the coal deposit. Then all the coal was discharged carefully through the exit at the silo bottom. In this particular case, supply of carbon dioxide at the silo bottom was considered superfluous. Wibbelhoff (1981) described a dust explosion in a coal dust burner plant of a cement works in F. R. Germany, in March 1981. Prior to the explosion, an electrical fault had caused failure of an air blower. The explosion occurred just after restart of the repaired blower. During the period in which the blower was out of operation, dust had accumulated on the hot surfaces inside the furnace and ignited, and as soon as the blower was restarted, the glowinghurning dust deposits were dispersed into a dust cloud that exploded immediately. Pfaffle (1987) gave a report of a dust explosion in the silo storage system of a pulverized coal powder plant in Dusseldorf, F. R. Germany, in July 1985. The explosion occurred early in the morning in a 72 m3 coal dust silo. The silo ruptured and burning material that was thrown into the surroundings initiated a major fire, which was extinguished by means of water. Fortunately no persons were killed or injured in this primary accident. However, during the subsequent cleaning-up process, a worker was asked to free the damaged silo of ashes by hosing it down with water. It then appeared that a glowing fire had developed in the dust deposit that was covered by the ashes. The worker had been warned against applying the water jet directly to the smouldering fire, but for some reason he nevertheless did this. The result was an intense dust flame that afflicted him with serious third degree burns. The smouldering fire was subsequently extinguished by covering its surface with mineral wool mats, and subsequently soaking the whole system with water containing surface-active agent. 2.1 1 DUST EXPLOSION IN A SILICON POWDER GRINDING PLANT AT BREMANGER, NORWAY, IN 1972 In this serious explosion accident, five workers lost their lives and four were severely injured. The explosion that occurred in the milling section of the plant, was extensive, rupturing or buckling most of the process equipment and blowing out practically all the wall panels of the factory building. Figure 2.26 gives a flow chart of the plant. Figure 2.27 shows the total damage of the entire grinding plant building, whereas Figure 2.28 gives a detailed view of the extensive damage. Eye-witnesses reported that the flame was very bright, almost white. This is in accordance with the fact that the temperature of silicon dust flames, as of flames of aluminium and magnesium dust, is very high due to the large amounts of heat released in the combustion process per mole of oxygen consumed. (See Table 1.1 in Chapter 1.). Because of the high temperature, the thermal radiation from the flame is intense, which was a main reason for the very severe burns that the nine workers suffered. The investigation after the accident disclosed a small hole in a steel pipe for conveying Si-powder from one of the mechanical sieves to a silo below. An oxygedacetylene cutting torch with both valves open was found lying on the floor about 1 m from the pipe with the 194 Dust Explosions in the Process Industries Figure 2.26 Flow chart of dry part of plant for production of refined silicon products at Bremanger, Norway. The grinding plant that was totally damaged in the explosion in 1972 is shown to the right in the chart Figure 2.27 Totally destroyed milling section of silicon powder production plant at Bremanger, Nor- way, after the dust explosion in October 1972 Case histories 195 Figure 2.28 Bremanger, Norway, October 1972 Detailed view of the extensive material damage caused by the silicon dust explosion at hole. According to Kjerpeseth (1990) there was strong evidence of the small hole having been made by means of the cutting torch just at the time when the explosion occurred. At the moment of the explosion, part of the plant was closed down due to various repair work. However, the dust extraction system was operating, and this may in part explain the rapid spread of the explosion throughout the entire plant. The interior of the pipe that was perforated had probably not been cleaned prior to the perforation. In view of the high temperature and excessive thermal power of the cutting torch, and not least the fact that it supplied oxygen to the working zone, a layer of fine dust on the internal pipe wall may well have become dispersed and ignited as soon as the gas flame had burnt its way through the pipe wall. The blast from the resulting primary silicon dust explosion then raised dust deposits in other parts of the plant into suspension and allowed the explosion to propagate further until it eventually involved the entire silicon grinding building. The grinding plant was not rebuilt after the explosion. 2.1 2 TWO DEVASTATING ALUMINIUM DUST EXPLOSIONS 2.12.1 MIXING SECTION OF PREMIX PLANT OF SLURRY EXPLOSIVE FACTORY AT GULLAUG, NORWAY, IN 1973 The main source of information concerning the original investigation of the accident is Berg (1989). The explosion occurred during the working hours, just before lunch, while ten workers were in the same building. Five of these lost their lives, two were seriously injured, two suffered minor injuries, whereas only one escaped unhurt. A substantial part of the plant was totally demolished, as illustrated by Figure 2.29. 196 Dust Explosions in the Process Industries Figure 2.29 Scene of total demolition after aluminium dust explosion in the premix plant of a slurry explosive factory at Cullaug, Norway, in August 1973 (Courtesy of E. Berg, Dyno Industries, Cullaug, Norway) The premix preparation plant building was completely destroyed. Debris was found up to 75 m from the explosion site. The explosion was followed by a violent fire in the powders left in the ruins of the plant and in an adjacent storehouse for raw materials. The explosion occurred when charging the 5.2 m3 batch mixer, illustrated in Figure 2.30. It appeared that about 200 kg of very fine aluminium flake, sulphur and some other ingredients had been charged at the moment of the explosion. The total charge of the formulation in question was 1200 kg. The upper part of the closed vertical mixing vessel was cylindrical, and the lower part had the form of an inverted cone. The feed chute was at the bottom of the vessel. The mixing device in the vessel consisted of a vertical rubber-lined screw surrounded by a rubber-lined earthed steel tube. The powders to be mixed were transported upwards by the screw, and when emerging from the top outlet of the tube, they dropped to the surface of the powder heap in the lower part of the vessel, where they became mixed with other powder elements and eventually re-transported to the top. The construction materials of the mixer had been selected so as to eliminate the formation of mechanical sparks. This was probably why both the screw and the internal wall of the surrounding earthed steel tube were lined with rubber. During operation the 5.2 m3 vessel was flushed with nitrogen, the concentration of oxygen in the vessel being controlled by a direct reading oxygen analyser. According to the foreman’s statement, the oxygen content at the moment of explosion was within the specified limit. After the explosion, the central screw part of the mixer, with the mixer top, was retrieved, as shown in Figure 2.31, about 12 m away from the location that the mixer had prior to the explosion. More detailed investigation of the part of the screw that was shielded by the steel tube, revealed, as shown in Figure 2.32, that the screw wings had been deformed bi-directionally as if an explosion in the central part had expanded violently in both directions. This evidence was considered as a strong indication of the explosion having been initiated inside the steel tube surrounding the screw. The blast and Case histories 797 Figure 2.30 Cross section of mixer for producing dry premix for slurry explosives at Cullaug, Norway, in 7973 (Courtesy of E. Berg, Dyno Industries, Cullaug, Norway) Figure 2.31 Top of 5.2 m3 premix mixer, and 3.3 m long mixing screw with surrounding steel tube (see Figure 2.30), as found after the explosion 12 m away from location of the mixer prior to the explosion (Courtesy of E. Berg, Dyno Industries, Cullaug, Norway) 198 Dust Explosions in the Process Industries Figure 2.32 Section of screw after splitting and removal of surrounding steel tube, showing bi-directional deformation of the screw wings from the explosion centre. Part of rubber lining of steel tube removed from upper half of tube (Courtesy of E. Berg, Dyno Industries, Gullaug, Norway) flame from this primary explosion in turn generated and ignited a larger dust cloud in the main space inside the mixer, and finally the main bulk of the powder in the mixer was thrown into suspension and ignited when the mixer ruptured, giving rise to a major explosion in the workrooms. Subsequent investigations at Chr. Michelsen Institute, Bergen, Norway, revealed that clouds in air of the fine aluminium flake powder was both extremely sensitive to ignition and exploded extremely violently. The minimum electric spark ignition energy was of the order of 1 mJ, and the maximum rate of pressure rise in the Hartmann bomb 2600 bark. Both these values are extreme. The thickness of the aluminium flakes was about 0.1 pm, which corresponds to a specific surface area of about 7.5 m2/g. (See Section 1.1.1.3 in Chapter 1.) The investigation further disclosed that the design of the nitrogen inerting system of the mixer was inadequate. First the nitrogen flow was insufficient to enable reduction of the average oxygen concentration to the specified maximum level of 10 vol% within the time allocated. Secondly, even if the flow had been adequate, both the nitrogen inlet and the oxygen concentration probe were located in the upper part of the vessel, which rendered the measured oxygen concentration unreliable as an indicator of the general oxygen level in the mixer. It is highly probable that the oxygen concentration in the lower part of the mixer, and in particular in the space inside the tube surrounding the screw, was considerably higher than the measured value. This explains why a dust explosion could occur in spite of the use of a nitrogen inerting system. The final central concern of the investigators was identification of the probable ignition source. In the reports from 1973 it was concluded that the primary explosion in the tube surrounding the screw was probably initiated by an electrostatic discharge. However, this conclusion was not qualified in any detail. In more recent years the knowledge about various kinds of electrostatic discharges has increased considerably (see Section 1.1.4.6). It now seems highly probable that the ignition source in the 1973 Gullaug explosion was a [...]... discussed in Section 3.8 Figure 3.16 Velocity vJt) of a particle introduced in a gas flow of velocity v, Rudinger, 1980) + bt at t = 0 (From 2 24 Dust Explosions in the Process industries 3.5 .4 SPEED OF SOUND IN A DUST CLOUD The speed of sound plays an important role in all compressible flow phenomena, including dust explosions Rudinger (1980) distinguishes between two extreme cases In the first case the. .. elements inherent in the total problem complex Some of this will be 206 Dust Explosions in the Process Industries reviewed in the following in sufficient detail for the genuine nature of the various problems to become visible This is considered important in a new text on dust explosions because in the past, dust explosion research has often been conducted without paying appropriate attention to the central... been the only force in operation This means that the number of inter-particle bonds per unit mass of cohesive powder can be increased by compacting the powder, i.e by increasing the bulk density of the powder deposit Therefore Wminalso increases with the degree of compaction Moisture influences Wminby influencing the strength of certain types of inter-particle bonds The logical link between Wminand... explosible dust clouds from powder deposits implies that particles originally in contact in the powder deposit must be separated and distributed in the air to give concentrations within the explosible range There are two aspects to consider The first is the spectrum of forces originally acting on and between the particles in the deposit, resisting the separation of the particles The second aspect is the forces... accumulation may either be intentional, as collection of powders and dusts in silos, hoppers and bag filters, or unintentional as deposition of dust on beams, external surfaces of process equipment or walls and floors of work and storage rooms 2 04 Dust Explosions in the Process Industries Re-suspension and re-dispersion of dust may either occur intentionally, e.g by handling and transport in various process. .. any comprehensive theory of the generation of dust clouds, leading from the properties of the powder deposit, via the nature of the energy available for dispersion, to the structure of the dust cloud However, in view of the wide variation in possible boundary conditions in industrial practice, one would not expect to find one single, unified theory covering all possible situations On the contrary, each... defined by the chemical composition, the pressure and the temperature For a dust cloud, however, the state of equilibrium will be complete separation, with all the particles settled out at the bottom of the system In the context of dust explosions, the relevant ‘state’ will therefore always be dynamic In various industrial environments as well as in experiments with dust clouds, gravity and other inertia... sizes the shape of the curve of uT ( S ) will differ from that in Figure 3.3, by having its maximum in the liquid bridge range of S < 0.25 Adding liquids to dusts is sometimes used intentionally in industry for reducting dust dispersibility One application of this method is addition of soya bean oil to grain for preventing generation of grain dust clouds in grain storage plants See Section 1 .4. 10 in Chapter... lycopodium Shadowing angle 20" Generation of explosible dust clouds 2 19 Figure 3.12 Scanning electron micrograph of a single lycopodium particle showing the rough surface topography assumed in Stokes’ law, and the terminal settling velocities will be lower than predicted This is the reason for the curving of the lines in Figure 3.10 in the range of large particles The settling velocities indicated in Figure... the range However, in many situations in industry, and particularly during dust explosions, general inertia forces may dominate over the gravity force, and other flow regimes may be of primary interest The Reynolds’ number of the particle is an important indicator of the flow regime Reynolds’ number for a particle of diameter x travelling in a gas is defined as: P Re =- VreIx (3. 14) P where pg is the . elements inherent in the total problem complex. Some of this will be 206 Dust Explosions in the Process Industries reviewed in the following in sufficient detail for the genuine nature of the. coal dust in the space above the bulk coal by the 192 Dust Explosions in the Process industries smouldering fire or glow when it reached the surface of the coal deposit. (See Figure 1.9 in Chapter. transmitted to nine units of the central dust collecting plant (1 and 2 in Figure 2. 24) via the ducting in the underground tunnels. Severe room explosions were initiated when the ducting in the tunnel