Harpur Hll, Buxton, SK17 9JN Telephone: +44 (0)114 289 2000 Facsmle: +44 (0)114 289 2050 Potental Exploson Hazards due to Evaporatng Ethanol In Whsky Dstlleres HSL/2003/08 Crown copyrght 2003 Project Leader: H. S. Ledn Author(s): H. S. Ledn MSc PhD DIC Scence Group: Fre and Exploson Group
CONTENTS 1 INTRODUCTION.... 1 2 REPORTED EXPLOSIONS........ 2 2.1 PEORIA, ILLINOIS 1935.. 2 2.2 PEKIN, ILLINOIS 1954........ 2 2.3 AHMADNAGAR, INDIA..........2 2.4 PUERTO DE SANTA MARIA, SPAIN..........2 2.5 THONBURI, THAILAND..... 2 2.6 LAWRENCEBURG, KENTUCKY, USA....... 2 2.7 ATCHISON, KANSAS, USA.. 3 3 CALCULATIONS.... 4 3.1 ASSUMPTIONS AND SIMPLIFICATIONS..... 4 3.2 PHYSICAL PROPERTIES.. 4 3.2.1 Ethanol.... 4 3.2.2 Water.... 4 3.3 RESULTS..........5 3.3.1 70 proof whsky.... 6 3.3.2 90 proof whsky..... 6 4 DISCUSSION AND CONCLUSIONS..... 7 4.1 DISCUSSION........7 4.2 CONCLUSIONS....... 8 5 REFERENCES..... 9
EXECUTIVE SUMMARY Ethanol s a hghly flammable lqud wth a relatvely low flash pont, 21 ºC the flash pont s dependent on the alcohol concentraton. There s a concern that evaporatng ethanol could pose an exploson hazard n bonded warehouses and n stlls rooms. A lterature search was carred out to ascertan the ncdent rate of explosons n dstlleres. It appears that explosons occur only very rarely. A report from the md-1960 s suggested that explosons leadng to fataltes and casualtes as well as substantal destructon of buldngs happen, apparently as a result of a precedng fre n many cases, but that very severe ncdents occur farly rarely. Whsky s produced through a mult-stage process, of whch dstllaton of the wash s the penultmate stage, by whch the alcohol level n the whsky s rased to 70-80 %, by volume. The dstllaton takes place n copper dstllers located n a stlls room. Ethanol bols at a temperature of 78.4 ºC. The surface of the copper vessel s at a temperature at or n excess of the bolng pont of ethanol. The surface temperature s, however, well below the autognton temperature of ethanol, whch s around 426 ºC, whch precludes auto-gnton by the stlls vessel surface. The whsky s then stored n casks, made of oak, n bonded warehouses for a mnmum perod of three years, used for blends, some sngle malts are exported after fve to seven years, whle other sngle malts are stored for eght to 40 years. A consderable amount of whsky s stored n these warehouses. The warehouses are often old, possbly of Vctoran orgn, and bult wth stone. The natural ventlaton s an ntegral part n the maturng process of the whsky, lendng t a partcular qualty, whch wll be dfferent from regon to regon. It has been assumed that a typcal ar change rate s of the order of ten ar changes per hour. It s not wthn the remt of the present study to delve deeper nto the potental sources of gnton. Possble sources of gnton are however lkely to fall nto one of the followng categores - a lghtnng strke, open fre, dscarded and stll lt cgarettes, statc electrcty, sparks generated due to electrcal faults, etc. There appears to be a low probablty of an exploson due to the gnton of an ethanol/ar mxture. The evaporaton rate of ethanol at 25 ºC s too low; the natural ventlaton would almost certanly be able to dlute the gas cloud ethanol concentraton down to well below ts lower flammablty lmt. However, the present study does not take nto account the possblty of recrculaton zones or stagnant regons, where the gas cloud could, potentally, become enrched so as to fall between the lower and upper flammablty lmt. It would be necessary to use more sophstcated tools, e.g. Computatonal Flud Dynamcs, to nvestgate gas buld up n these regons. OBJECTIVES 1. To nvestgate the lkelhood of an exploson due to evaporatng alcohol n a whsky dstllery storage room 2. To nvestgate the lkelhood of an exploson due to evaporatng alcohol n a whsky stlls room 3. To comple nformaton about explosons n whsky and other sprts producng dstlleres
MAIN FINDINGS 1. Explosons n dstlleres appear to occur nfrequently. There are only a few ncdents reported n the lterature, as a search through varous databases, ncludng HSE s major accdents database, revealed. 2. The results from the calculatons suggest that there s not a problem wth the formaton of an explosve/flammable ethanol/ar mxture, at the condtons assumed n the present study, e.g. ambent temperature. It s worth notng that possble buld up of a flammable gas cloud n recrculaton or stagnant regons, caused by the natural ventlaton flow, have been gnored. MAIN RECOMMENDATION 1. There are no recommendatons.
1 INTRODUCTION The present project was ntated to nvestgate the lkelhood of an exploson due to evaporaton of ethanol from pools caused by accdental spllage of whsky n dstlleres. Fleschman, Parrs, Daley and Looby (1995) estmated the spllage at the cask fllng stage to be just under 500 kg year -1 for a medum sze bourbon dstllery, whch represents roughly a loss of the order of 0.005 % of the annual producton. Carter and Lnsky (1974) studed the emsson of ethanol from whskey fermentaton vats. Carter and Lnsky quoted average ethanol emsson fgures of 182 g m -3 of gran nput. UDV, the sprts and wne dvson of Dageo (wth labels such as Gunness, classc sngle malts, Johnny Walker, etc.), produced an envronmental report, Anon (2001), where t was suggested that an estmated 15,000 tonnes of ethanol was released nto the atmosphere from maturng whsky n Scottsh dstlleres. Mascone (1978) nvestgated the ethanol losses n the Amercan whskey ndustry. Mascone estmated the loss of ethanol to be 3.2 kg barrel -1 year -1 durng the ageng phase of whskey producton. However, of more concern s spllage n the stlls room or n the warehouse. The loss of ethanol due to the evaporaton from the casks, known as the angels share, s not consdered n the present study. The concern wth evaporatng ethanol s equally vald n dstlleres producng other sprts, such as gn, vodka, etc. Rasbash (1966) examned a number of ncdents nvolvng fre and exploson n sprts warehouses. Rasbash found that the ncdents were rare occurrences. In one case, a fre started n the warehouse, whch after several hours buld up led to an exploson, whch tore the buldng apart. The common theme n the ncdents was that a fre would precede an exploson - radaton and convecton enhancng the vaporsaton of ethanol, destroyng casks and/or structures leadng to further spllage and subsequent gnton of the explosve mxture. Rasbash dentfed fve mechansms for the producton of a flammable mxture: Evaporaton of ethanol followng a leak under normal temperature and pressure Evaporaton of ethanol from a spllage nto a hot ambent atmosphere, due to a fre Evaporaton of ethanol onto a hot surface, heated by radaton from a fre Producton of explosve mxture n a tank or cask Evaporaton of ethanol nto a oxygen defcent atmosphere, wth a subsequent ntroducton of ar The current FOD/HSE PM84 Gudance Note, Anon (2000), whch s concerned wth explosve gas mxtures n gas turbne acoustc enclosures, sets an upper lmt for the maxmum acceptable gas cloud sze (at 50 % LFL). The maxmum acceptable gas cloud sze s 0.1 % of the total enclosure volume. Experments have shown that explosons nvolvng gas clouds of that sze gve rse to neglgble over-pressures. The PM84 recommendaton has been adopted n the present study though the layout of a stlls room dffers sgnfcantly from that of an acoustc enclosure. There was some confuson over how the unt proof s defned. The Brtsh defnton, Anon (2002a) and Anon (2002b) dd not gve any explanaton of how the proof was calculated, whle Anon (2002c) dd the whsky s mxed wth gunpowder n some proporton. The mxture s then exposed to a naked flame to see f the flame would gnte the mxture. Table 4 shows the converson between proof and volume fracton of ethanol. The Amercan defnton of proof s that proof s twce the volume fracton of ethanol. The Brtsh defnton has been used n the calculatons throughout the present report. 1
2 REPORTED EXPLOSIONS A lterature search was carred out to fnd references to explosons n dstlleres. Only a few accdents could be found. In a couple of cases, there appears to be some confuson over the sequence of events, e.g. whether an ntal fre caused the exploson or the exploson preceded a fre. The nformaton obtaned s scant. 2.1 PEORIA, ILLINOIS - 1935 A bonded warehouse, owned by Hram Walker & Sons, n Peora, Illnos, had started lstng, Rasbash (1966). Remedal work had been undertaken n order to secure the buldng. However, the work was not fnshed and the buldng was stll not n plumb. The lft was not workng and the whskey casks were therefore stll stored n the buldng. Eyewtnesses clamed that they had heard an exploson. It was conjectured that a few whskey casks mght have been dslodged, but how and why they were dslodged s not known. It was never establshed what had acted as the gnton source. 2.2 PEKIN, ILLINOIS - 1954 A fre had started n Amercan Dstllng Company s dstllery n Pekn, Illnos, n 1954. The fre bult up over several hours. Two warehouses had been burned out. The radaton from the fre n an adjacent buldng led to the roof of one of the remanng warehouses gettng very hot. The ncreased roof temperature n turn led to the gnton of an ethanol/ar mxture. It was not known how the gas cloud had formed, but t s concevable that an ncreased evaporaton of ethanol due to the radaton from the fre had occurred. 2.3 AHMADNAGAR, INDIA An exploson occurred n an ndustral alcohol dstllery n Ahmadnagar, Inda, on 20 December 1985, Anon (1985). It was an exploson n an acetc acd tank; the cause s not the gnton of evaporatng alcohol so the ncdent s not drectly relevant to ths study. Nevertheless, there were fve fataltes and 45 casualtes because of the exploson. 2.4 PUERTO DE SANTA MARIA, SPAIN An exploson and subsequent fre occurred at the Alcoholes del Puerto factory on 24 August 1988, Anon (1988). The exploson affected sx tanks contanng ethanol, the dstllaton plant and offces and personnel buldngs. Seven people were klled n the accdent, whle four persons suffered burns and one person was reported mssng. Unfortunately, there was no addtonal nformaton wth regards to the cause of the exploson, or what klled the seven people. 2.5 THONBURI, THAILAND A seres of explosons rpped through a whskey dstllery n Thonbur, Thaland on 14 June 1993, Anon (1993). There were four fataltes and at least 10 njured. The explosons were caused by a fre. The report suggested that the source of gnton were sparks from weldng equpment. There were also envronmental effects as dangerous chemcals were leaked nto a rver, Anon (1993). 2.6 LAWRENCEBURG, KENTUCKY, USA A fre broke out n a warehouse at the Wld Turkey Bourbon dstllery n Lawrenceburg, USA on 9th May 2000, Chellgren (2000) and Mazza (2000). A sgnfcant amount of bourbon, 2
between 15000 and 20000 casks, each of whch held 53 gallons of whskey (3000-4000 m 3 ), was stored n the seven-story buldng. The warehouse was reduced to a ple of rubble by the fre. It was reported that workers at a nearby water treatment plant had heard a loud bang, as would be produced by an exploson. They went out to nvestgate what had happened and they then saw that one end wall and part of one of the sdewalls of the warehouse had been blown out - the warehouse then collapsed after about three mnutes. The fre brgade concentrated ther efforts on ensurng that the fre would not spread to any of the other eleven warehouses on ste. There were no fataltes - though two fremen had to be taken to hosptal sufferng from heat exhauston. A water treatment plant, whch serves Lawrenceburg and surroundngs wth drnkng water, had to be shut down as bourbon escaped nto a nearby rver and was about to be drawn nto the water treatment plant ntake. 2.7 ATCHISON, KANSAS, USA An exploson took place n the Mdwest Gran Inc. ethanol dstllery n Atchson, Kansas on 13 th September 2002. Four people were njured n the exploson. Eyewtness accounts reported that flames shot out of the wndows of the factory, after whch the freball lfted nto the ar. The roof and one sde of the buldng were completely destroyed by the blast. The strength of the blast was such that celng tles and lght lenses fell to the floor n a bowlng alley some two blocks west of the dstllery. The cause of the exploson s not known. 3
3 Calculatons 3.1 ASSUMPTIONS AND SIMPLIFICATIONS The ethanol vapour pressure s not altered by the actve ngredents n the whsky, whch are not well charactersed The age of the whsky does not affect the vapour pressure Equlbrum between the gas and the lqud phase s establshed Raoult s Law s applcable to the ethanol/water mxture The condtons n the warehouse are fully mxed The emsson rate does not change wth tme and corresponds to the maxmum rate - e.g. after the soakage phase s completed (after 6-18 months) - not needed snce t s assumed that equlbrum between the lqud and the vapour phase s establshed, see above Equpment n the stlls and storage rooms does not affect the effectveness of the natural ventlaton - nor does t affect the flud flow n the rooms Equlbrum between the gas and the lqud phase s establshed The storage room s 10 m wde, 10 m long and 5 m hgh The ar change rate s 10 h -1 The temperature n the stlls room and the storage room s 298.15 K, the humdty s 50 % and the pressure 101,325 Pa (1 atm) The temperature and pressure does not change apprecably The ambent ar s ntally made up of oxygen (20.9 % v/v) and ntrogen (71.1 % v/v) The concentratons of oxygen and ntrogen n the lqud phase s neglgble The Brtsh defnton of whsky strength has been used n the present report the measurement of strength nvolved mxng the whsky wth gunpowder and thereafter nvestgate whether a naked flame would gnte the mxture, Anon (2002C) 3.2 PHYSICAL PROPERTIES 3.2.1 Ethanol TABLE 1: Physcal propertes of ethanol, at P = 101,325 Pa and T = 298.15 K Physcal property Value Lower Flammablty Lmt (LFL) 3.5 % v/v Upper Flammablty Lmt (UFL) 19.0 % v/v Molecular weght 46.07 kg kmol -1 Flash pont 285.15 K Bolng pont 351.45 K Vapour pressure 7.9 kpa The data was obtaned from SAX, Anon (1999), except the vapour pressure, whch was calculated from Equaton 3. Olbrch (1980) quoted slghtly dfferent LFL and UFL values, but the SAX data has been used. 3.2.2 Water TABLE 2: Physcal propertes of water, at P = 101,325 Pa and T = 298.15 K Physcal property Value Molecular weght 18.04 kg kmol -1 Bolng pont 373.15 K Vapour pressure 3.2 kpa The vapor pressure for water was calculated from Equaton 3. 4
3.3 RESULTS Dalton s law relates the partal pressure of a speces to the mole fracton of sad speces n the gas phase and the total pressure: P y P, (1) where y s the mole fracton of speces n the gas phase and P tot s the total pressure. tot It has been assumed that Raoult s law apples to the bnary mxture. Raoult s law relates the partal pressure of a speces to the mole fracton of sad speces n the lqud phase and the partal pressure the speces would exert f t were a pure lqud: P x P, (2) 0 where x a s the mole fracton of speces n the lqud phase and P 0 s the partal pressure of the pure speces. The vapour pressure of speces vares wth temperature and can be calculated usng b ln 0 P a [ kpa], (3) T c where a, b and c are speces dependent constants, T s the ambent temperature, n K. Table 3 shows the values of the constants used n equaton 3. TABLE 3: Speces dependent constants n the vapour pressure equaton Speces a b c Ethanol 16.19 3424.0-55.72 Water 16.54 3985.0-39.0 The mole fracton of speces can be readly calculated usng the followng expresson, a combnaton of equatons 1 to 3: y x b 1000 exp a, (4) Ptot T c where the factor 1000 s used to convert the partal pressure to Pa. TABLE 4: Converson from proof to volume fracton of ethanol Proof Volume fracton of ethanol 70 40 75 43 80 46 90 52 100 56 105 60 5
It was necessary to carry out an teratve procedure n order to calculate the correct molecular volume, whch unfortunately vares qute consderably for the ethanol/ar mxture. A dagram n Atkns (1983), Fgure 8.1 on page 216, proved to very useful. 3.3.1 70 proof whsky The number of moles of speces n the lqud mxture can be calculated wth the followng formula: n, (5) V V m, where V s the volume fracton of speces n the mxture and V m, s the molecular volume, nterpolated from Fgure 8.1 n Atkns (1983). A mole fracton of ethanol was chosen, the correspondng molecular volumes were read from the graph, and a new mole fracton of ethanol could be calculated usng Equaton 4. A new set of molecular volumes can be read from the graph. Ths procedure s contnued untl the mole fractons have converged. There s some uncertanty n the results gven the nterpolaton from a graph. Three teratons were requred before the mole fractons could be consdered to have converged. The calculatons gave the followng result; n A = 0.85 and n B = 0.15. The mole fracton of ethanol n the gas phase was 0.0125, whch s below LFL for the ethanol/ar mxture. Hence, there s no rsk of an exploson or fre. 3.3.2 90 proof whsky The procedure outlned n Secton 3.4.1 was followed n the 90 proof case. The calculatons of the mole fractons of the two speces n the lqud phase yelded the followng result; n A = 0.774 and n B = 0.226. The mole fracton of ethanol n the gas phase was 0.0176, whch s below LFL for the ethanol/ar mxture. Hence, there s no rsk of an exploson or fre. 6
4 DISCUSSION AND CONCLUSIONS 4.1 DISCUSSION A search through the HSE ncdent database as well as other databases dd not yeld a large number of reported ncdents. The best source was a report by Rasbash (1966), whch contaned detals of a few ncdents nvolvng explosons n whsky bonded warehouses. The nformaton n the report s old t relates to ncdents, whch took place between 1935 and 1966. There appears not to have been an update of the report or any other study undertaken snce Rasbash s report was publshed, some 35 years ago. Most of the recent ncdents, whch were reported n Lloyds Casualty Reports, were descrbed n such bref detals as to make t dffcult to assess the exact nature of the ncdents. However, t would appear that the reported ncdents nvolvng explosons were not due to the gnton of an ethanol/ar mxture, formed from natural evaporaton or spllage. The ethanol evaporaton rate s low n a room at 25 ºC, whch could represent a typcal maxmum room temperature n a bonded warehouse. The calculatons, carred out n the present study, ndcate that the formaton of a gas cloud wth an ethanol concentraton between the lower and upper flammablty lmts s unlkely. A number of assumptons have been made, though t s unlkely that these would sgnfcantly alter the conclusons of the present study. However, t must also be remembered that although the bonded warehouses are naturally ventlated, assumng a typcal fgure of ten ar changes per hour, there could be regons of recrculatng flow. The ethanol vapour could fnd ts way nto these regons of recrculaton, thus gvng rse to gas clouds wth an ethanol concentraton wthn the flammablty lmts. The present study does not consder flud flow effects. A more sophstcated tool, e.g. Computatonal Flud Dynamcs, would have to be employed f t s desrable to nvestgate gas cloud buld-up n stagnant regons of the warehouse. Rasbash (1966) suggests that a fre n an adjacent buldng, f unsuccessfully checked, could lead to a temperature rse n the warehouse. The hgher temperature could then result n an ncreased ethanol evaporaton rate and, thus, to the formaton of a gas cloud. The scenaro of spllage or leakage n the stlls room s more complex. The temperature n the copper dstllers s well above 25 ºC, so that the surface temperature wll also be relatvely hgh, but well below the auto-gnton temperature of ethanol, whch s around 430 ºC. It s therefore not possble for the hot surface to auto-gnte the ethanol/ar mxture. Rasbash (1966) correctly states that a hot surface would lead to an ncreased evaporaton rate of ethanol. Moreover, Rasbash suggested that a hot surface would not necessarly lead to an exploson, f the fre were n the vcnty of the hot surface, as a relatvely small fre would ensue. Rasbash (1966) dd recognse that a flammable gas cloud could form f the fre was located a substantal dstance away from the hot surface. The ar change rate and the arflow pattern n the room would be two mportant factors n decdng whether an explosve mxture would form and where t would be located. The producton of an explosve mxture due to the heatng up of a storage tank or cask s not consdered explctly n the present study. It s noted, however, that ths scenaro s a lkely one. It s not known whether any expermental studes nto the effects of flame engulfment of oak casks have been undertaken. It s therefore not possble to comment on how long a cask would stand up to a fre. The oak used to manufacture casks s a good nsulator, see Table 5, so the temperature rse n the whsky would be relatvely slow. A possble outcome of the flame engulfment of a cask s that the metal hoops, wth hgh thermal conductvty, see Table 5, whch hold the oak staves together, would heat up frst. The transfer of heat to the contents of the cask s of course governed by the heat conductvty of the oak and not the steel. However, t s not clear what s the most lkely falure mechansm. It s nterestng to note 7
that the oak has nearly four tmes hgher thermal expanson coeffcent than the carbon steel used for the hoops, whch means that the staves and the cask ends would expand more than the metal hoops for a gven temperature dfference. The temperature n a flame could be of the order of 500 C to 1000 C, at whch steel would begn to soften. One possble falure mechansm would be that the steel hoops snap or get dslodged due to the large dfference n thermal expanson between the wood and the steel, thus relevng the pressure n the cask. The ethanol vapour would then gnte. Further study would be requred to establsh the falure mechansm of flame-engulfed whsky contanng oak casks. Table 5: Physcal Propertes of oak and carbon steel, Kaye and Laby (1989) Materal K -1 W m -1 K -1 Oak 35 10-6 - 60 10-6 0.14-0.17 Carbon steel 16 10-6 45 4.2 CONCLUSIONS The calculatons performed n the present study seem to ndcate that there s very low probablty of an exploson of ethanol/ar mxture, formed ether by a spllage or by natural evaporaton. The assumed temperature, pressure and alcohol content of the whsky would be very unlkely to yeld a gas cloud wth an ethanol concentraton n the flammable range. The falure mechansm of an oak cask s not known. There s a large dfference between the thermal expanson coeffcents of the oak and the steel hoops. The accumulaton of ethanol vapours at hgher concentratons n a slow movng, recrculatng flow, due to natural ventlaton, cannot be ruled out. However, the effects of the natural ventlaton have not been taken nto account n the present study. A more sophstcated mathematcal tool, e.g. CFD, would need to be employed n order to study the possble accumulaton of ethanol n almost stagnant regons wthn the bonded warehouse. 8
5 REFERENCES Anon, Lloyds Weekly Casualty Reports 262(12):285, 1985. Anon, Fre Preventon 189:50, 1986 Anon, Lloyds Weekly Casualty Reports 273(9):184-185, 1988. Anon, Lloyds Casualty Week 292(12):249, 1993. Anon, Emsson Factor Documentaton for AP-42, Secton 9.12.3, EPA Report 68-D2-0159, Unted States Envronmental Protecton Agency, Research Trangle Park, North Carolna, 1997. Anon, SAX S Dangerous Propertes of Industral Materals (CDROM), 10 th Edton, John Wley & Sons, 1999. Anon, Control of Safety Rsks at Gas Turbnes used for Power Generaton, Gudance Note PM84, Health & Safety Executve, UK, 2000. Anon, UDV Envronment Report 2001, (http://www.dageo.co.uk/ad/performance/envronment/udv_envr_aremssons.html? 3413), 2000 Anon, LSW and NACHO FAQ, http://www.labsafety.org/faq.html, 2002A. Anon, The makng of Whsky, http://www.nmk.co.za/products/whskes/the_makng_of_whsky.html, 2002B. Anon, Scotch Whsky, http://www.bbc.co.uk/dna/h2g2/a602939, 2002C. Atkns, P. W., Physcal Chemstry, 2 nd Edton, Oxford Unversty Press, Hongkong, 1983. Carter, R. V., and Lnsky, B., Atmospherc Envronment 8:57-62, 1974. Chellgren, M. R., Wld Turkey Fre, ABCNEWS.com, 9 th May 2000, (http://abcnews.go.com/sectons/us/dalynews/bourbon000509.html), 2000. Fleschman, M., Parrs, M., Daley, E. W., and Looby, G. P., Envronmental Research Bref: Waste Mnmzaton Assessment for a Bourbon Dstllery, EPA Report No. EPA/600/S-95/002, Unted States Envronmental Protecton Agency, Cncnnat, Oho, 1995. Kaye, G. W. C., and Laby, T. H., Tables of Physcal and Chemcal Constants, 15 th Edton, Longman Scentfc and Techncal, Sngapore, 1989. Mascone, D. C., Cost and Engneerng Study Control of Volatle Organc Emssons from Whskey Warehousng, EPA Report No. EPA-450/2-78-013, Unted States Envronmental Protecton Agency, Research Trangle Park, North Carolna, 1978. Mazza, E., Bourbon Hangover, ABCNEWS.com, 10 th May 2000, (http://abcnews.go.com/sectons/us/dalynews/wldturkey000510.html), 2000. Moon, C., Dstllery exploson leaves four njured n Atchson, Topeka Captal-Journal, 14 th September 2002, (http://www.cjonlne.com/stores/091402/kan-atchnson.html), 2002. Olbrch, H., Zum Alkohol als Gefahrenquelle n der Brennere, De Branntwenwrtschaft 120:83-85, 1980. Rasbash, D. J., A note on the occurrence of fre and explosons n sprt warehouses, Fre Research Note No. 624, Fre Research Staton, Borehamwood, UK, 1966. 9