Mines and Bunkers: Volume 10, Fire Safety Aspects of Polymeric Materials (1980)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

CHAPT E R 3 FIRE D Y NAMICS AND SCE NARIOS 3. 1 I ntrod uction Man has experi enced fire for many thousands of years . As with other usefu l but potential l y dangerous an d destructive forces, it became i m perative that he learn to u se and control fire. Practica l solutions for many fi re situati ons were devel oped gradua l l y and empirica l l y, l argely on the basis of post-fire i nvestigation and without f u l l unde rstandi ng of the processes associated with ign ition, com bustion, fi re s pread, detection, and exti nguishment. This approach, however, is no longer accept­ able given society's i ncreased technologica l capabil ity and awareness of the value of h uman l i fe and safety. Thus, a more soph isticated approach to fi re preven tion and control is requi red especially i n the h igh ly h az ardous mine environmen t. Although the number of fata l i ties resu lti ng from mine fires is re l atively smal l , fi res together with explosion s are a dreaded h azard i n mining, particu larly coal m i n i ng, and the potential for loss of l i fe and property from these disasters is s u bs tantial. ( Statistics on mine fi res in the U nited States an d the Un ited Kingdom a re presented i n Appendix A) . It therefore is of the utmost i mportance that the best avail a bl e methods be used to assess an d analyze the sou rces of the h azards . The conti n u i ng mechani zation of coal mining, wh ich i nvolves the use of hydrau l ic equipment, plastics and large quanti ties of electricity , has intensified the ch ance of fi re wh ile m i n i mi z i ng l oss of h u man l i fe from non ·fuel accidents. The concept of using fi re scenarios as a tool for gai n ing a better understanding of m ine fires is i ntroduced in this chapter. Fire scenario developmen t and analysis a re descri bed and selected scenarios are presented. The state of the art of fi re dynam ics a l so is discussed in terms of the characteristics of mine fires, the properties of fu m es behind the fire zone, the forces developed by these fum es, an d the venti l ation disturbances caused by fi res. G aps in knowledge are iden ti fied an d approaches for developing improved fire prevention and control measu res are proposed. 3.2 Mine and Bunker Fire Scenarios Real f i re situations in m ines, especi a l l y during the initial stages of devel opmen t, a re sel dom observed by trained person nel. How a fire was i n i ti ated often can be deduced by carefu l ly exa mining the remaining evi dence, and the sequence of events often can be reconstructed with reasonable re l ia bi l i ty ; however, i n some c�ses, most evidence is completely destroyed an d any attempt at analysis is restricted to mere s pecul ati on. The developmen t and ana lysis of fire scen arios therefore can be of great v a l u e by permitti ng alternative fire devel opment sequences to be considered, 11 

M I N ES AN D B U N K E R S by lending support to speculations and deductions, and by providing gu idance for the design of usefu l experi ments. 3.2. 1 G uideli nes for Development Mine fi re scenarios are best based on real fi re incidents that lend themselves to plausi ble extrapol ation of the i m portant elements. F ortunately, m ine fire reports are compi led by the U.S. Bu reau of Mines ( Bu M i nes ), and there is a l arge number of real incidents from wh ich to choose. In thi s section, the importan t physica l e le­ ments of a mine fi re that belong i n a scenari o are considered. Major emphasis is placed on the physical behavior of fi re rather th an on the human element even though i t is recogni zed that people enter the scenario by preventi ng, detecti ng, exti ngu ishi ng, or starting the fire and by escaping from or being inju red or k i l led by the fire. 3.2. 1 . 1 Pr• Fire Conditions The i m portant elements of a fire incident general ly are establ ished long before the incident. The physical l ayout of the m i ne , the structu ral elements used, and th e equi pment selected decisively affect the events th at lead to a fi re inciden t, and a great deal of attention must be given to these pre-fire conditions. Consequently, the fi rst step in the development of a m ine fi re scenario shou ld be to gather al l i m portant pre-fi re data. I ncl uded may be i n formation on the general l ayout of the mine; the production rate ; the number of personnel em ployed per shift; the location of conveyors and vents; vent flow di rection and vol ume; the type and phys i ca l condition of the mining, transportation, and auxi l iary equ i pmen t em ­ ployed ; mai ntenance records and housekeepi ng conditions (e.g., the accu m u l ation of coa l dust/oil mi xtu re on work ing mach inery ) ; and the location and condition of permanent and mova ble electric power distri bu tion l i nes, switch gear, and fai l u re protection equi pment. The basis for materials selecti on also shou ld be exam ined ; where and h ow materi als were used and stored shou l d be established; and compli­ ance with appli cable codes, regul ations, an d procedural i nstructions shou l d be con­ sidered. 3.2. 1 .2 Ignition Source The ty pical fi re scenario starts with ign i tion, which may be characterized as the bringi ng together of an energy source and a combusti ble su bstance i n th e presence of an oxidizing atmosphere so that a sel f-sustaining exotherm ic reaction occu rs . I gn ition sou rces in m i ne fires usual l y i nvolve el ectric arcs ; overheated conductors ; o r the frictional heat that results from electrical o r mechanical fai l u res, malfun� tions, or accidents. Spontaneous combustion of materials also occurs with som e frequency, and human error (welding or improper or u nauthori zed use of fi re o r electric heat) i s another contri butor. If possi ble, the ign ition sou rce shou l d be characteri zed quantitatively in the fi re scenario in te rms of : 12 

F I R E DY NA M I CS A N D SCENAR I O S 1 . Maximum temperature ( ° C) , 2. Ene rgy release rate (cal /sec or watts ) , 3. Time o f appl ication t o target (sec ) , and 4. Area of contact (cm 2 ). I n consideri ng the i gn ition of sol i d materials (e. g., pol ymers, coal , coal dust and oil mixtu re), it is most i m portant to recogni ze that a "strong" source wi l l ign ite the target whereas a "wea k" one may not. The "strength " of the source depends on the e nergy fl ux and the ti me of appl icati on to the target or on the product of these two. I gn ition of gaseous targets, on the other hand, can occu r with a very weak sou rce even i f the flashpoi n t temperatu re is reached on l y for a moment. The meth· a ne and air mixtu re frequentl y encou ntered in coal m i nes is such a target and its flash poi nt, wh ich depen ds on i ts concentration, is i mportant. 3.2. 1 .3 Material Fint Ignited Of major i mportance in developing a m i ne fi re scenario is defin ition of the target m ateri al fi rst ignited by the energy sou rce . This fi rst step in the f i re ch ain - and a favora ble point at which to break it - represents a transition from a transient ( accidenta l ) energy release to an uncontrol led exotherm ic reacti on of com busti ble fuel and oxygen that is capable, if not checked, of accelerated growth to cata­ stroph ic proportion. Whether ign ition occu rs under a given energy release depends on the physical and chemical properties of the target materi al ; therefore, a deta i l ed description of these properties is essentia l if the probabil ity of ign ition is to be assessed p roperly. M ost organic gases, l i qu i ds, and sol i ds wi l l ign ite if heated to a sufficiently high temperature in the presence of an adequate oxygen supply. Com bu stible gas m ix­ tures and certai n d ust an d air mixtu res (e.g., coal dust) a re ignited more easily th an l i qu ids and sol i ds. In coa l m i n es, where such mi xtu res are u n avoidable, the concen­ tration of the gaseous (or dust) fuel must be reduced below the flammable l i m it by a wel l designed, high-volume forced venti l ation system . Thus, it is i m porta nt i n mine fire scenarios t o define the ty pe, concentrati on , a n d tem peratu re o f va rious fuels (or dust) and the ventilation air velocity ( ft./m i n . ) or rate of replacement (ft.3 /m in . ) . Liquid organ ic fuel s a lso can be ignited qu ite easily depending on their physical s tate (pool, foam, mist, or spray) and temperatu re, and these parameters are i m por­ tant in the fire scena rio. I f the temperatu re of an organ ic pool is incre ased above the flash poi nt, ign ition of the fuel vapors above the pool w i l l occur and the pool wil l sustai n burning. In a m i ne envi ronment, the most common organic liquids are h y drau l i c flu ids, l u bricating oils and greases, and diesel fuel. It is qu ite common , for i ns tance, to find an accumulation of mixtu res of oi l and grease and coal dust on work equ i pment (cutters) and transport equ i pment (tro l l eys and conveyors ) . Al­ though the total prevention of such an accumul ation is rather difficult (if not i m possible), good housekeeping practices shou l d req u i re frequent remova l . All these 13 

M I N ES AND B U N K ER S elements are of i m portance a n d shou l d be considered i n the developm en t o f f i re scen arios. Quite frequently sol i d pol ymeric materials are the fi rst mate rials ignited in a m ine. These combusti bles com prise electric insu l ation, hydra u l ic hoses, rubber veh i· cle tires, conveyor belts, venti l ation cloth or polyurethane foam used for tem porary or permanent seals, and structura l timber. The coal bed itself ( although not consid· ered as polymeric material ) and coal dust, its fragmented product, also can be early targets of ignition. H ow easi ly a sol i d pol ymeric target wil l ign i te depends on the chemical composition and physical form of the materi a l . Generic terms such as pol ystyrene and polyurethane are not adequate to de­ scri be synthetic pol y mers. Most of these materials contain a variety of additives; some are composed of several polymers ; and some ex hi bit an altered chem ical com positi on as a result of reaction with the environment. All of these factors can substanti a l l y modify the com busti bil ity and fi re-sustaining ch aracteristics of the base polymer. Su rface texture (s mooth or frayed ) , form (sol id, rigid , or flexi ble foam ) , structure ( open or closed ce l l ) , and density and l ayer th ickness (surface to vol u me ratio) a re physical properties that, together with geometric configu ration, h ave a sign ificant effect on the behavior of a polymeric materi al in a fire situation. The thermal properties of sol i d targets play a vita l role i n determ ining ease of i gn i tion. Since the ignition of a sol id requ i res that the tem perature of its su rface be raised to some critical value ( i .e., the ignition tem peratu re) , heat conduction from the exposed surface to the inte rior wi l l affect the ti me of ign ition. Th is heat· transfe r mechan ism may become crucial to a sce nario if the heat flux is of relative· l y short duration. Material properties of th ickness and thermal diffusivity ( i.e., the ratio of therma l conductivity to heat capaci ty ) as wel l as heating rate and physical thickness determine whether a target material be haves i n a "therm ally th in " or "thermally thick" manner (see Vol u me 4 of the comm ittee 's report ) . The ignition time of a "thermal ly thick" material is relatively independent of physica l th ickness; it is control led by "thermal inertia" the product of heat capacity ( per unit vol ume) and thermal conductivity. The igni tion time of "thermally th i n " materials (e.g., ven t cl oth fa brics used in mines) is proportional to the product of heat capacity and physical thickness. In the case of compos i te structures, the properties of the layers used u nder surface fi l ms or sheets (e.g., the materials supporting belts) also w i l l affect ease of ign i tion. To be considered are the strength of the bond between the ex posed and the supporting layer, the thickness of the layers, and the thermal conductivity of the supporting material . The geometric confi guration o f a material also can i nfl uence ease o f ignition. F lat su rfaces and single sol i d members general l y a re more difficul t to ignite than closely stacked pieces or mem bers having fol ds and crevices. Si m i l arly, vertical or downward-faci ng su rfaces are more suscepti ble to ignition than those facing upward because of increased heat transfer from a risi ng convective heat plume. 14 

F I R E DYNAMICS A N D SCENAR IOS 3.2.1 .4 Combustion Types - F laming and Smoldering Some combustible materials bu rn in either a smoldering or a fl aming mode, but genera l l y on l y sol i ds with very l ow thermal conductivi ty (e.g., p lastic foam or fi be r pad ) can smolder. Whether a materi al burns i n the smol deri ng or fl am ing mode may be determined by the ignition source. A h igh-temperatu re ignition sou rce (e.g., an open flame) usua l ly wi l l i n i tiate fl aming combustion whe reas a low-temperatu re sou rce (e.g., an overheated wi re) is more l i kely to result i n smol de ring combustion. A restricted air supply l i ke that in a closed compartment or the interior of parti· tions also is more l i kely to result i n smoldering com bustion. Smol deri ng combustion is characterized by a slow spread rate, a rel atively l ow temperature, the absence of visible flame, and the production of smoke and gas. I n developing m ine fire scenarios i t is i mportant to note that the products o f smolder­ ing combustion are d ifferent from those of flaming com bustion and that a transi­ tion to flami ng combustion after a long smo l dering period m ay resu lt i n a rapidly spreading fi re because of the preheating of the fuels and the accum u l ation of com busti ble gases du ring smoldering. In addition, smol dering or deep-seated fi res are difficu l t to extinguish ( i .e., a gaseous extingu ish ing agent m ay exti ngu ish flam es but the residual charcoal may continue to bu rn by "glowing com bustion" and the flame mi ght rekindle after the extinguishant has dissi pated). F l a m ing com bustion is characteri zed by visi ble flames, a h igh temperatu re, and a ra pid spread rate. I ts presence ususal l y does n ot g o undetected for l ong. Thus, smol dering combustion is general ly the more insidious h azard, and the possibility of i ts occu rrence shou ld be consid ered careful l y in i nvestigating acci dents and developing scenarios. 3.2.1 .5 F i re Propagation The course of a fi re after igni tion is determi ned by the rate of fi re growth and the time at which various defensive actions are i n itiated. These factors are therefore very i m portant elements in the m i ne fi re scen ario. F i res grow by spreading over the surface of an ign i ted fuel element, by spreading from one contiguous target to the next, or by jumping across a gap from one fuel e lement to the next. The rate of fi re propagation over a hori zontal or downward­ faci ng sol i d surface general l y is rather slow. However, the fire can spread rapi dly if the material is "therm a l l y thi n " or has been preheated by convection or radiation or if there i s forced ventilation. I f the physical l ayout perm its, upward propagation (e.g., through shafts and vertical ducts ) wi l l occu r very rapidly and at a progres­ sively i ncreasing rate . If the origi na l l y ign ited materi al is separated by a gap from the nearest secondary combustible target, it wi l l either die out or spread across the gap. "Ju m pi n g the gap" can occu r by a variety of modes . The secondary target can be preheated ( by convection or radiation ) unti l it pyrolyzes and em its flammable vapors th at spread across the gap and a re ign i ted. I f the pri mary burn ing material is a thermopl astic, i t may spread the fi re by melting and dri pping bu rning droplets onto the secondary materi a l . 15 

M I N ES AN D B U N K ERS The specific envi ronment prevai l i ng i n m i nes and bu n kers wi l l h ave a great effect on the rate of fi re spread. The normal forced vent i l ation in m i nes w i l l greatly accelerate fi re spread ; however, when forced venti l ation is stopped, oxygen is con· sumed rapidly and fi re spread is greatly dece lerated. An effective, frequently prac· ticed means of contro l l i ng fi res in m i nes consists of inhi biting oxygen flow to the fuel by seal ing off the fi re·affected a rea. When th is is done, however, the concentra­ tion of combustible gases increases and an explosion or reignition of the fire at an accelerated rate can occur when venti lation is restarted . Add i tiona l ly, if the bu rn i ng material is a structural s upport and is mechanica l l y weakened by the fire, the mine shaft or duct can col lapse and alter the fi re spread situation by l i miting or en· h ancing vent i l ation and curtai l ing extinguishing or seal i ng operations. The foregoing i l l ustrates why it is unwise to make any change to the mine venti lation system wi thout carefu l consideration by and agreement among com pe­ tent, responsible persons. In coal mi nes, m iscontrol of venti l ation du ring a fire h as resu lted i n explosi ons and i n rapid, upwi n d burn i ng o f the coal ri bs and roof. The subject is fu rther discussed be l ow. 3.2. 1 .6 Evol ution of Smoke and Toxic Gases Du ring the early stages of a m i ne fire, the forced ventilation system usual ly provides abundant oxygen, but when venti l ation is interrupted, the fire environ· ment rapi dly becomes oxygen lean. Th is s ituation results i n i ncom plete com bustion and the production of highly tox ic gases and fu mes . Depen ding u pon the types of pol ymeric material present, the typica l products of i ncom plete combustion m ay i ncl ude carbon monoxide, hydrogen cyanide, n itrogen oxides, am monia, hydrogen sul fide, phosgene, and many other compounds . Most of these are h igh ly tox ic to human l i fe. Smoke (suspended sol i ds) is also dangerous to l ife in th at it can plug ai rways and carry adsorbed gases, liquids, and residual heat to the respi ratory tract. U nderground m i ne access and egress routes usua l l y are very l i m ited and the spread vel ocity of toxic fumes and gases genera l l y is more important than the spread velocity of the fire itself. Thus, the response to a fire s ituati on in a m ine is different from th at i n a bui lding or aboard a sh ip or a i rplane where the first response usu a l l y is an attempt to exti ngu ish it with someth i ng i m m ediate ly avai l · a b l e (e.g., a c u p of coffee, water, or a cloth ) . I n a m i ne, t h e fi rst response must be an immediate start of evacuation wh i le simul taneously cutting off the ignition sou rce ; on ly then can an attempt to exti ngu ish the fire be made. In assessing the effects of fu mes and tox ic gases in a mine fi re scenario, the fol l owing points shou l d be considered : 1 . Because of thei r physical and chem ical properties, sm oke and ga ses offer an o pportunity for early fire detection and initiation of evacuation and exti n· guish ment operations. Thus, an efficient al arm and communications system i s extremely i m po rtant. 2. Smoke and gases can rapidly incapacitate person nel ; therefore, evacu ati on 16 

F I R E DYNAM I CS A N D SCENAR IOS operati ons are of paramount i mpo rtance. 3. The high concentrations of smoke and fu mes present i n a mine fi re dictate that fi refighters carry complete oxygen su pport systems when downwind of the fire. Dense smoke also may curtail rescue and fi re control efforts on the downwi nd s i de . 4. The h i gh concentrations o f gases a n d fumes also can have residual effects on mine equ i pment and materials. Exposed machi nery m ay corrode or be cov­ ered with a fl amma ble, corrosive and often electrica l l y conductive f i l m . 3.2. 1 .7 Detection The first detection of the fire is a critical element in a m ine f i re scenario because of the time required to evacuate personnel through l i m ited escape routes. Auto­ matic detection equi pment may be sensitive to heat or to the particu l ate or gaseous products of combustion. F i re detectors are most effective in mi nes when located on or near face-cutti ng equ i pment and i n the vent system downstream of active opera­ tions. How automatic detection equ i pment responds to products of com bu stion is i mportant and may depen d on the concentration of gases, the particle size of smo ke, the velocity of smo ke and gases flowing past the detector, the orientation of the detector chamber to the fl ow, and the sensi tivity setti ng and operating charac­ teristics of the detection instrument. In scenario developmen t it is i m portant to remem ber th at under differen t venti l ation conditions the same m aterials w i l l pro­ duce smo ke having d i fferent particle si zes and wi l l have d ifferent com busti on char­ acteristics. I n addition, the characteristics of the smo ke m ay change as it "ages" and smaller particles aggl omerate into l arger ones. 3.2. 1 .8 Extinguish ment At some point in a m i ne fi re scenario, extinguishment and fire control activ ities w i l l be i n i tiated. A consi deration of extingu ish ment tech n iques is beyond the scope of th is study, but it should be noted that the effectiveness of control and extin­ guishment efforts depen ds on the burn i ng characteristics of the polymeric materi­ als, the spread rate of the fi re , and the ti me l a pse between first ignition and detecti on . The accessi bi l ity of the fi re scene to firefighters and the time req u i red to reach the scene also a re factors to consider in mine fires. 3.2. 1 .9 Summary of Essential Scenario Elements The m i ne fi re scenario shoul d descri be a l l significant factors and events in the development of the fi re and should cover as many as possible of the fol l owing points : 1 . The pre-fire situation shou l d be descri bed. 2. The source of ign i tion energy shou l d be i dentified and descri bed i n quantita­ tive terms. 17 

M I N ES AN D B U N K ERS 3. The first material to be i gn i ted shou l d be identified and ch aracterized i n terms o f its chemical a n d physica l properties. 4. Other fuel materials that play a significant role in the growth of the fire shou l d be i dentified and descri bed. 5. The path and mechanism of fi re growth shou l d be determined ; particu lar attention shou l d be given to fuel element location and orientation, venti la­ tion, compartmentation, and other factors that affect fire spread. 6. The possi ble role of smoke and toxic gases i n detection, fi re spread, and casualty production shou l d be determi ned. 7. The possibi l i ty of smo l dering com bustion as a factor in the fire i ncident should be considered. 8. The means of detection, time of detection, and the state of the fire at the time of detection shoul d be descri bed. 9. Defensive actions and evacuation procedu res shou l d be descri bed, and the i r effects on fi re control shou l d be determ i ned. 1 0. I n teracti ons between personnel in the m i ne and the fire sh ou l d be detailed . 1 1 . T h e t i me and sequence o f events, from the fi rst occ u rrence o f the ign i tion energy f l ux to the final resolution of the fi re incident, shou l d be establ i shed . The complete scenario shou l d permit genera l i zation from the particu lar i ncident descri bed, and it shou l d provide the basis for expl orati on of alternative paths of fire i n itiation and for analysis of the effects on fi re con tro l of changes in materials, design, and operating procedures. As noted earl ier, the scenario shou l d be based on real fire incidents or shou l d take some of i ts elements from real fire incidents . These i ncidents should be fu l l y docu mented by a post-accident i nvestigation report and an analysis designed to determ ine how and where the fire started and progressed unti l its term i n ation. The scenari o also can be based on a report of a fu l l y instru mented, fu l l -scale test burn. I n either case, the development of a f i re scenario wil l remain an art rather than a scienti fic presentation of i rrefutable evidence unti l existing fi re dyn amics knowledge is augmented by add i tional research . Nevertheless, if con­ structed to be as complete and as accurate as possible, the fire scen ario can be an effective tool for use i n improving the fire safety of m i n i ng operati ons by i ncreasing man's abi l ity to visua l ize and comprehend the events. 3.2.2 Guidelines for Analysis Ana lysis of we l l devel oped, plausible scen ari os is an effective methodology for developing econom ical and efficient methods of fi re prevention and control in m ines and bun kers. One method of fire scenario ana l ysis involves the carefu l ex­ a m i nation of each element of the fire inci dent, the posing of rel event questions regarding each, and the identification of alternatives that coul d have prevented the i ncident. I n con ductin g such an analysis, the fol l owing shou l d be considered : 1 . Pre- F i re Situation - Were existing codes, standards, and operating procedu res adequate? Were they enforced? If not, what shou l d be done to correct the 18 

F I R E DYNAM I CS A N D SCENAR IOS s i tuati on? Were the materi als and equ i pment used properl y selected? Were the materials in good physical con dition and was the equ ipment properl y m ai n­ tai ned? What housekeepi ng practices and cond itions preva i led prior to the fi re i ncident, and what actions shou l d be taken to correct any deficiencies? 2. Ign i tion Sou rce - What was the ignition source and for how l ong was it in contact with combusti ble material? Cou l d the ign i tion sou rce be e l i m inated by edu cation or by design? 3. I gn i ted Material - What we re the chemica l and physical characteristics of the ign i ted material ( its shape, form, and appl ication ) ? Where was it l ocated in rel ation to wal ls, equ i pment, and other com busti ble m aterials? Did melting or dripping of the ignited material affect fi re spread? Did the ign ited material col lapse? Wh at was the heat-release rate at the ti me the ign ited m ateri al was fu l l y i nvolved? Could another less flammable or more fi re- reta rdant materi al h ave been used i n pl ace of the ign i ted materi a l ? Were flammabil i ty tests on materials i ntended for th is appl ication avai lable and adequ ate? 4. Combustion Type - Did smol dering precede fl aming? I f not known, was the i gn i ted material capable of smol deri ng? What were the vol u me and com posi­ tion of gases generated by the smo l dering fi re? G iven the prevai l i ng venti l a­ tion conditi ons, how quickly did a dangerous concentrati on of smoke and tox ic gases deve lop an d spread? 5. F i re Spread - How much time elapsed before the origi nal l y ign ited target was fu l l y i nvolved in the fire? What was the mechanism of i gnition transfer to a second combustible material ? What was the flame spread rate and how did it change as vent i l ation conditions were altered? What effect did the material properties of the fi rst two materials ign i ted have on the rate of fire spread? Wou l d material su bstitutions and desi gn mod i ficati ons have affected fi re spread? 6. Smoke and Tox ic Gases - Of what value were smoke detectors? Did their absence or presence and location have an effect on the fi nal outcome? D id dense smo ke h i nder escape. Which fuels contri buted sign ificantly to the de­ crease in vis i bi l i ty? What effects did toxic su bstances have on person nel (i.e., extent of i njuries, i nterference with escape efforts, and deaths ) ? Which tox ic substances were most respons i bl e for injuries and from whi ch fue ls did they evo lve? 7. Ext i nguishment - How much time elapsed between i gn i tion, fi rst detection, start and fin ish of evacuation, and i n i tiation of exti nguishment efforts? Which esti ngu ishment or fi re control techn i ques were used and how successfu l were they ? Wou l d the use of di fferent extingu ish ment techni ques or materia ls or better trai ned fi refi ghters h ave improved efficiency or reduced i nju ries and damages. 8. Secondary E ffects - D i d any secon dary occu rrences such as a rekindling of the fire, expl os i ons, fl ashovers, and post-fl ashovers occu r? What caused these 19 

M I N ES A N D B U N K ER S i ncidents? D i d structural col lapse occu r? I f so, was any code violated? Was any ch ange in ventil ation con ditions i nvolved? Were establ ished procedures fol l owed? 3.2.3 Selected Mine F ire Scenarios In order to demonstrate that fire scenario devel opment and analysis is a produc­ tive methodology for improvi ng the fi re safety perform ance of polymeric m aterials i n m i nes, a number of scenarios are presented below. I t m ust be emphasi zed that these scenarios are not considered to be perfect or, in some cases, even satisfactory. Most of these incidents were based on actu al mine f i res ; h owever, sufficient docu mentation was not always ava i lable, exa m ination was not al ways adequ ate, an d fol l ow-up ex peri mental demonstration of undeterm ined causes, occu rrences or consequences was someti mes lacking. Thus, these scenari os a l so i l l ustrate the importance of in-depth data col l ection and fol l ow-u p exam ina­ tion , including experi mental demonstration. 3.2. 3. 1 Insulation and Hydraulic Hose F i re Caused Short Circuit (Sorrel and Lyon, 1 973) Summary The fi re occu rred when a tra i l ing ca bl e short-circu ited on the reel of a cutt i ng machine at the face of the m ine. The nearest office of the U . S. Bureau of M i nes was notified by the superi ntenden t: he stated that all men, except pe rsons engaged i n firefighti ng activities, were o n their wa y t o t h e surface. Another i nspector was sent to the neigh boring coal m i ne, wh ich was con nected with the m i ne i nvolved , to issue a n order requ iring that a l l persons be withdrawn from and be proh i bi ted from e nteri ng the neigh boring m i ne. At the time of the fire, 1 50 employees were unde rgroun d at the m ine i nvolved, and 1 1 2 were underground at the neighboring m i ne. Al l em ployees escaped u nas· s isted and there were no i nju ries. Property damage was confined to the cutting machine. The i nvestigation was completed the fol lowing day . Pre- Fire Conditions The m i ne was opened by three shafts and one slope. Of the 301 men employed, 255 worked un derground and produced an average of 4,000 tons of coal per day. The room-and-p i l l a r system of m i n i n g was used, but p i l l a rs were not extracted. At l east two separate and d isti nct travel able passageways, one of wh ich was venti l ated by i nta ke air, were ma intained between the working sections and the su rface . The m i ne was ven tilated by an ax ial-flow fan that was properly instal led on the surface and equ ipped wi th the necessary safety dev ices . The m i ne s urfaces ra nged from damp to dry; loose coal and coal dust were not perm itted to accumulate i n active workings. R oc k dust, i n a m p l e quantities, was applied t o with i n 4 0 feet of the working faces, including open crosscuts . 20 

F I R E DYNAM I CS A N D SCENAR I OS Coal was transported from the areas in shuttle cars , disch arged onto bel t convey­ ors, and transported to a mi ne-car loading fac i l i ty . Men were transpo rted u nder· grou nd in battery-powered, trac k-m ou nted personnel carriers. U ndergrou nd, the coal was discharged from drop-bottom mine cars into a hopper and then trans­ ported by bel t conveyor to the su rface . Electric power (41 60 volts ac ) was pu rchased from a local uti l ity company to operate a 300-k i lowatt rectifier l ocated on the coal-producing section. This suppl ied 275 vo lts de power for the electric face equ i pment. The 4 1 60 volts ac was red uced to 480 volts for the undergrou nd be lt conveyor motors. The fram es of 80 percent of the el ectric face equ i pment were grounded by means of s i l icon d iodes ; the remai n i ng 20 percent were grounded by means of tra i l i ng cables. The tra i l ing cables were of the flame-resistant ty pe, an d a l l were equ i pped with suitable sh ort-c i rcu it protection. The mach ine i nvolved i n the fire was a permissi ble type rubber-ti red cutting machi ne equ ipped wi th a 1 25-horsepower cutt i ng motor, a 4 5-horsepower pump motor, and two 500-foot lengths of No. 4/0, single conductor, Type W flame­ resistant cable. The cable contained one permanent spl ice approxi mately 8 feet from the ree l entrance gland and was protected by a short-c i rcuit re l ay . The rel ay was adjusted to operate at an instantaneous cu rrent of approx imately 1 ,825 am pe res and to l ock out if a sol i d short circuit occurred. The hydrau l ic system of the cutt i ng machine contai r:ed approx i mate l y 95 gal l ons of flammable hydrau l ic fl u id . The approved record book in dicated that the mach ine h ad been i nspected by a qua l i fied person 1 2 days before the f i re. F i refighti ng equ i pment was readily available. It consisted of dry -chem ica l fire exti ngu ishers, water u nder pressu re with sufficient fire h ose to reach each working face, a nd roc k dust. Description When the fire occu rred, the cutting mach i ne had finished cu tting the face of an entry on the intake s i de of the working section , and the mach i ne operator's helper sumped the ba r of the mach i ne in the right ri b of the left crosscut. The trai l i ng cabl e then short-c ircu ited on the reel and ignited the jacket of the cable. The fire spread to the outer jackets of the hydrau l ic hoses. The hel per and the regu l ar mach ine operator i mmed i ately left the m achine without actuati ng the fire· suppress i on system installed on the mach i ne, deenergized the power, and notified the men on the working section to proceed to inta ke a i r. A trained mi ne-rescue team was d ispatched to the fire area. The l i ne brattice had bee n i nsta lled to with i n 1 0 feet of the face with 1 3,000 cubic feet of air reach i ng the en d of the brattice. The fi re hose, wh ich was con nected to the end of a 2-i nch waterl i ne, was exten ded to the entry and water was applied di rectly to the fi re . During th e fi refighti ng operation, the fire-su ppression system on the m ach ine was thermal l y actuated. The fire was extingu ished approx imately 1 -3/4 hou rs after it started. 21 

M I N E S AND BUNKERS Analysis The mach ine invo lved was provi ded with an operative fire-su ppression system that consisted of 40 pounds of dry chemicals and two manual actuators. The actuators were readi l y access i ble, but the system was not manua l l y actuated. The system contained eight spray nozzles, fou r of wh ich were directed toward the reel com partment. Adequ ate firefighting equ i pment was read i l y ava i l able. The mach i ne i nvolved was exami ned by a qua l i fied electrician 1 2 days before the f i re. The tri p rel ay for the tra i l i n g cable i n the face box was in operative co ndition. The only spl ice i n the tra i l ing cable was a permanent spl ice 8 feet from the reel entrance gl and. The ca ble insulation on the cable next to the reel was brittle and cracked due to excessive heat. Th is was possi bl y due to ox i dati on or m igration of the polymeric i nsu l ation pl asticizer. Approx i m ately 50 feet of the tra i l ing cable was on the reel at the time of the fire. The hydrau l ic oil tan k remai ned intact, and the o i l in the system was not a factor in the fi re. The fi re started as a result of a hi gh-resistance fault in the trai l i ng ca ble. The resistance of the faul t l i mited the cu rrent to a value l ower than needed to actuate the circu it protective device. The resultant arcing ign ited the i nsu l ation. Analysis of the scenario suggests that : 1 . Electrical equi p ment shou l d be inspected by qual i fied persons as often as necessary to ensure safe operating conditions. The inspection should include an exami n ation of the tra i l ing cable on ree ls. 2. Ope rators of electrica l equ i pment shoul d be tra ined to actuate fi re­ suppression devices. 3.2.3.2 F i re Caused by Spontaneous Combustion - Smoke, Heat, Toxic Gases (Matekovic, 1 97 1 ) Summary D u ring his normal duties, the day-sh i ft fire boss detected smoke, heat, and carbon monox i de gas at the base of a pi l lar located two entries (abou t 200 feet) from a previ ously sealed area i n the m i ne. He notified the mine superi ntendent who teleph oned a m i ne inspector making a spot safety i nspection i n another section of the mi ne. An i nvestigation was started immediately. The cause of the inci pient fire was determi ned to have been spontaneous com bustion ; there were no inju ries. Pre- Fire Conditions The m i ne was opened by fou r slopes, one drift, and one shaft i nto the coal bed, wh ich range d from 20 to 25 feet in thickness. A total of 206 men, 1 79 u n der­ ground, were employed on one mai ntenance and two coa l-producing shifts per day, 5 days pe r week. Average daily production was 4,000 tons of coal. The en tries were developed by the room-and-pi l l ar method. Entries were driven on 80-foot centers and crosscuts were tu rned on 90-degree angles at 80-foot i n ter- 22 

F I R E DYNAMICS A N D SCENAR IOS vals. D u r i ng initial devel opment, approxi mately 8 feet of coal was extracted adja· cent to the i mmediate roof, leav i ng between 1 2 and 1 7 feet of bottom coal, 7 feet of wh ich was recovered during second m i n i ng. D uring deve l opme nt of one of the entries that was close to the fire, excessive heating was encou ntered. This forced m i n i ng operations to be discontinued and the entries to be sealed. Due to natu ral physica l conditions, the area was su bject to heavy bumps, heav ing, sloughage, and fractures in the bottom coal. F i refighti ng materi als consisting of portable fi refighti ng units, waterl ines, high· pressure rock-dusting mach i nes, and rock dust materials for construction of sea ls were ava i l able at the mine. Twelve 2-hou r and six 1 -h ou r self-contai ned breath ing units were avai lable on the surface. Sixteen 4 5-m inute Chemox u n its also were ava i l able at fire stations l ocated undergrou nd. The use of these units was not requ ired during sealing operations. The l ast regu lar federal i nspection of this mine had been completed two months before the fi re. A spot safety inspection was i n progress in another area of the m i ne at the time. Description The day-shift fire boss discovered smo ke and detected ca rbon monoxide gas when approach ing the seals during his regu l a r i nspection. He cal led the m ine super­ intendent who, with a sma l l crew, proceeded to the area. Tests were m ade to measu re the carbon monox ide and methane content in the mine atmosphere. Tem­ porary fi re-resistant pl astic cu rtains were instal led to excl ude the air from the affected area. The mi ne superi ntendent ordered all pe rsons not engaged in correct­ ing the condition to leave the mine. The company and the federal m i ne ins pectors a rrived and tested the ai r w ith a carbon monoxide gas detector; the carbon monox ide content was 250 parts per mil l ion. A 2- i nch fire hose was insta l led and water was appl ied to the hot area. An hour l ater the air was tested aga i n and only 50 parts pe r m i l l ion of carbon mon­ oxi de were present. Three temporary seals of 3 inch by 1 2 inch by 4 foot planking were started to seal off the area. These we re cored by th ree permanent seals constru cted of 6 inch by 8 i nch by 3 foot ties laid longitudinally "s kin to skin," fi l led with i nert material, and pl astered on the outside. Sealing operations were completed on the fo l lowing day. Analysis The smo ke, heat, an d carbon monoxi de were cau sed by spontaneous com bus­ tion. This scenario suggests only that the dai ly patrol of th is area shou l d be main­ tained. 3.2.3.3 Major Mine Fire and Fatal ities Caused by Arcing Trolley Cable (O' Rourke et al. 1 971 ) 23 

M I N ES A N D B U N K ERS Summary The fire occu rred in the straight-mains section of the mine. Of the 1 25 persons u n dergrou nd, 1 1 were worki ng in this secti on. N i ne of these persons escaped and two were k i l led. One additional person died acci denta l l y 2 1 days l ater fighting the fire. The fire began when the end of a tro l ley wi re fel l in an entry (empty track) of the straight ma i ns secti on and came in contact with the grounding clamp attached to the track rai l. Pre- Fire Conditions A tota l of 349 persons, 263 worki ng u ndergrou nd, were employed on 3 shifts per day, 5 and 6 days per wee k. Coal production averaged 5, 850 tons per day. The m i ne was opened by one slope and five shafts into the coal bed, wh ich averaged 84 i nches in thickness in the fi re a rea. The i mmediate roof consisted of wi l d coal, lami nated shale, and sandstone. The floor was h ard shale and f i re clay. The volati le ratio of the coa l i n th is m i ne was 0.39, indicating that the coal dust was explosive. The m i ne was devel oped by the room ·an d·pi l l a r method and p i l l ars were recovered. Entries were driven in sets of fou r to seven , 1 6 feet wi de, with crosscuts at su itable interva ls. The m i ne was venti l ated by three propel ler-type exh aust fans that prov i ded 940,000 cu bic feet of air per m i nute. Aux i l iary fans were used to ventil ate the work ing faces. Requi red tests for methane and other hazards were made. Conti nuous-type m i n ing machi nes were used and loaded into sh uttle cars, trans­ ported to loading ramps, and transfe rred onto steel m i ne cars. The loaded cars were gathered by sectio n trolley locomotives and placed in si detracks for fu rther trans· portation by tandem locomotives to the rotary dumps at the sh aft bottom . Coal was hoisted to the su rface preparation plant from a bi n at the bottom. E lectric power was pu rchased at 26, 000 and 28, 000 volts ac and transformed to 7, 200, 3,000, 440, 220, and 1 1 0 volts ac for use on the surface and u n derground. D i rect-cu rrent power at 290 volts was provi ded for underground use by 1 1 co nver· s i on un its provided with the requ i red safety devices. The de power circuit in the strai ght ma i ns was suppl i ed with 290 volts by two 7 50 ki lowatt, s i l icon di ode rectifiers. These recti fiers were l ocated approx i m ately 2,500 feet and 3,500 feet from the working faces. The circu it brea ker on the closer recti fier was set at 3, 500 a mps and on the fa rther rectifier, at 3,300 amps. The d i rect-cu r rent power was transmitted over a 400 m i l l i ci rcu l ar m i l ls coo per trolley wi re along the entry to the working secti on. The negative ci rcu it consisted of 7�pou nd ra i l track extended along the road haul ageway. Th is track was paral lel to a negative 1 ,000 mi l l i ci rcu lar m i l ls copper cable that extended to the working section. Up to th is point, s i ngle-bonded 4�pou nd rai l track was used for entering branches. A 7, 200 volt ac powe r cable, supported by messenger cable su spended from the roof in the entry, su ppl ied the power centers and porta ble recti fiers which prov i ded 550 volts ac and 250 volts de for use by the face equi pment. A trained an d fu l l y equ i pped mine rescue tea m was ava i l a ble at the m ine. Water- 24 

F I R E DYNAM I CS A N D SCENAR I OS l i nes, fire h ose, h igh- pressure rock-dusti ng mach i nes, and 2,000-ga l lon-capacity water cars, properl y equi pped, were ava i lable underground. Description A stoper operator, in the course of traci ng a pressure fai l u re on a roof bol ting machine, opened a canvas check instal led in an entry and observed yel l ow smoke. He cl osed the check and went to notify the foreman of the fi re. The two men then traveled th rough several entries and crosscuts unti l they met a shuttl e car operator. Due to the dense s mo ke, the foreman i nstructed the other two men to notify the crew to leave the section. He then attempted to l ocate the sou rce of the smoke but dense smoke i n several entries forced h i m back to a section sti l l having clear a i r. He was able to reach and open a trol ley swi tch i n a crossover that was one of the power sou rces to the tro l ley wi re in the track leadi ng to the smoke-fi l led secti on. He then ran to a m i n e phone, noti fied mi ne officials and the crew in a nearby m i ne section and i nstructed a motorman to open the other trol ley switch that suppl ied power to the f i re area. Meanwh i l e, the other two men notified the uti l i ty man and the men in the adjacent entries. They a l l assembled at the check curtain of their entry. A continu­ ous-mi ner operator and a roof bolter decided to retu rn for thei r safety lamps. When the roof bolter did not retu m, the operator fol l owed h i m as fa r as he cou ld, cal led several times but recei ved no answer, an d retu rned to the asse mbled group. The group then proceeded out of the mi ne. On the way they met the foreman and i nformed him th at the roof bolter was sti l l i n the section. A mason who had been i n sta l l ing a stoping on the ri ght side of the mains also was not accou nted for at th is ti me. The section forema n of the adjacent mains a rrived and the two foremen started to force fresh a i r th rough the affected entries. These efforts proved futi le. The assistant su perintenden t issued orders to evacuate all other areas of the m i ne and to sh ort ci rcu it the venti l ation at the crosscut. Checks were erected across the i nta ke entries at this point. A mandoor was opened into the left retu rn, an d a hole was made in th e stoping in the right retu rn. Efforts to locate the missing men also were futi le. Several state and federa l i nspectors arrived and the fi rst s i gn of an active fi re soon was observed by two inspectors an d two fo remen when the pl asti c check across the entry melted. A waterl ine para l l e l ing the haulage road was broken to al low the water to flow into the fi re area and the quantity of air passing over the fire was fu rther reduced by open i ng additional mandoors. A 2, 000-ga l lo n water car arrived at the scene and the fi re was attacked di rectly. A sampl i ng station was establ ished to mo n i tor the air return i ng from the fi re area in the left return and an hour l ater it was fo und that com busti bles had reached a dangerous poi nt. Al l persons we re withdrawn from the m i ne. Offici a ls of the coal mine company, the United M ine Workers of America 25 

M I N ES A N D BU N K ER S ( U MWA), a l ocal rescue organization, and the U . S. Bu reau of Mines jo intly planned rescue of the two missi ng men. Accordi ngly, the dri l l ing of boreholes into the face a rea was started i mmediate ly, and as these holes penetrated the m ine worki ngs, efforts were made to contact the trapped men by lowering phone communications. Geo-phones also were used to detect sound vi brations underground; however, these efforts were futi le. U l timately, approx i mately 90 boreholes were dri l l ed into the m i n e workings in and arou nd the fi re area, and various materials were introduced through these holes in an effort to control the fire. F ive days after the start of this incident, it was deci ded that the trapped men cou l d not have su rvived the gases produced by the fire . Since d i rect attack was i neffective, it was agreed that the fire area wou l d be flooded with water pumped from the su rface th rough boreho les into the face areas. Eight days later, ex pansion foam was i ntroduced fro m the su rface through 1 3 boreho l es into the i ntake ai rways near the fire area. The max i mu m hi gh-water level was reached on the eighteen th day. On the twenti eth day, m i ne rescue teams entered the area and di rect firefighting was re­ sumed; however, the fi re had spread beyond its l ast known l ocation. Due to i n· creased concentrations of combusti bles, al l persons were aga i n withdrawn from the m i ne. On the twenty-fi rst day, a state m i ne inspector was accidental ly drowned during the fi refighting operati ons. D u ri ng the next six months, underground dams were erected remotely to raise the water level, the mine was repeatedly i nspected, roof sections were rei n forced, bu l kheads were erected, and the m i ne was pu rged with nitrogen and flooded with water. After an i nspection made at the company's request 1 68 days after the fi re was detected, reha bi l i tation work was perm itted in seve ral sections of th e m ine. Attempts to recover the entombed men were to be pursued without interruption. After all information indicated that the fire was exti ngu ished 281 days after the original i ncident, it was decided to de-water the area behind the bu l kheads and it was estimated that approx i mately 50 m i l l ion ga l l ons of water were i mpou nded beh i nd these bu l kheads. Recovery operations proceeded slowly because many seals had to be removed, crosscuts made, and sam pl i ng stati ons for gas analysis establ ished as each area was reopened. The ventilation system also had to be modified conti nuously. The body of the roof bol ter was fou nd in 3 feet of water, several days after recovery operations started an d al most 1 8 months after the fi re began. The body of the mason was fou nd in a crosscut the next day. At that ti me, it became necessary to modi fy the vent i l ation system significant l y . Concu rren tly, the en ti re m i ne was completely ins pected to determ ine whether it was feasi ble to resume operations. Al most 26 months after the i n itial i nci dent, production was resumed in a l l but one of the active worki ngs of the m ine . Analysis The fi re most proba bly began when the end of the trol ley wire fe l l and came i n 26 

F I R E DYNAM I CS A N D SCENAR I OS contact with the grounding clamp attached to the track rai l . The tro l ley wire and i nsu lating anchor were found l y i ng on the m i ne floor at the compressor nipping station. Examination of the trol ley wire and the attached anchor indicated th at the clamp attach ing the hanger to the trol l ey wi re was parti a l l y consumed by electrical short circuit. Evidence i ndi cated that heat had softened the vinyl-type insu l ation in the hanger be l l used to anchor the trol ley wire. The trol ley wire and the soft i nsu l ation then pul led free from the hanger bel l , al lowi ng the end of the trol ley wire to drop to the mine fl oor and stri ke one of the com pressor ground cl am ps. Th is resu lted in the parti al destruction of the anchor and the ground clamps. The second ground clamp with a negative and frame grou nd conductor was i ntact. Ana lyses of samples of coa l and coke co l lected in the area of th e com pressor nipping station indicated that temperatu res had been higher in the roof area than near the bottom. After removing the trol ley wire anchor bolt assembly, samples were col lected from the i nside of the hole and at the roof l i ne . The analyses of these samples i ndicated that the temperatu res had been h igher inside the hole th an at the roof l ine outside the hole. Th is i ndicated that the hanger probably h ad been grounded an d had generated heat wh ich softened the i nsu lation in the be l l and thereby al lowed the trol ley wire to drop to the m i ne floor and stri ke the negative ground cl amp, thus in iti ating the fi re . Analysis of t h i s scenario suggests that : 1 . I n addition to the anchoring device, an i ns u l ated hange r shou l d be provi ded at the ends of trolley wi res and trol ley feeder wires to prevent the wires from making contact wi th the mine floor or m i ne track rai ls if the anchoring device fails. 2. Circuit breaker sh ort-ci rcui t trip settings shou l d be consistent with the power transmission system. The power transmission system shou l d be eval uated peri­ odical ly to dete rmi ne if adequ ate short-circuit protection is provi ded during norma l mining and idle periods. 3. When any smoke or abnormal amounts of fumes are detected in a m ine, a thorough sea rch shou l d be i n itiated an d conti nued until the source is deter­ m i n ed and e l i mi nated. 3.2.3.4 Cutti ng Mach ine Caused Fi� (Jarvis, 1 967 ) Summary The fi re occu rred on a m i n i n g machine at the face of a work ing pl ace i n the mine. The fi re apparently started in the area of the cu tting motor and the resu lting arcs an d flame ign ited accu mu lations of oil and coal dust on the mach ine, some of the hose in the hydrau l ic system, the front ti res, and wiring. The fi re was fought directly with d ry-che mical fi re exti n gu i shers and water and was complete ly extin­ gu ished the same day. The 1 8 men working in the only secti on active at the time of the occurrence assisted i n fighti ng the fire. There were no injuries, and dam age was 27 

M I N ES AN D B U N K ERS confined to the mach i ne, wh ich was considered to be a total loss. Pre- F i re Conditions The m i ne was opened by drifts i nto the coal bed. Of the 43 men employed, 40 worked u ndergrou nd. An average of 700 tons of coal per day was loaded by a m obi le loadi ng machine i nto ru bber ti red m i ne ca rs. The m i ne was devel oped by the room-an d-pi l lar method. The i mmed i ate roof i n the fire a rea varied from firm to fragi l e shale. The mi ne was cl assified a s nongassy. Venti l ation was i n duced by a propel ler exhaust fan i nsta l led on the surface. Loose coal and coal dust had not accumul ated in dangerous quantities at the time of the l ast federal i nspection before the fire and rock dust had been appl ied to with i n 30 feet of a l l faces. The m in i ng machi ne i nvolved in the fi re was equ i pped with a 5o-horsepower cutti ng motor and a 3D-horsepower pu rnp moto r. Except fo r the accumul ations of oil and coal dust on the machine and the con dition of the cutt i ng motor, it was mai ntained in good mechanical condition. The mach ine was equ i pped with tw o para l le l 30Q-foot lengths of single conductor, N o . 1 /0 flame-resistant tra i l ing cables that were without short-ci rcu it protection. The hydra u l i c system for the m ac h i ne , wh ich ru ptu red d u r i n g the fi re, h a d a capaci ty of 9 0 gallons and was fi l l ed with flammable hydrau l i c f l u i d. F i refighti ng equ i pment consisted of a 3Q-pou nd dry chem ical system on the m i n ing mach ine, dry-chemical fire exti ngu ishers on each mobi l e unit and at strate­ gic l ocations, and a hi gh-pressure rock-dusti ng mach i ne with ample quantities of rock dust. Description The fi re occurred when the min ing mach ine operator and a hel per completed u ndercutting the face of a working place and began tu rn ing the machine to start a crosscu t to the left. I ntens ive arcs and flame suddenly issued from the area of the cutting motor and ign ited accumu l ations of oil and coal dust on the m achi ne, som e of the hose on the hydraulic system, the front ti res, and wiring. The operator actuated the firefi ghting syste m on the mach i ne and had the power removed before the rapidly spreadi ng fi re and dense smoke forced retreat from the i m medi ate area. A l i ne curtain was erected and the contents of al l fi re exti ngu ishers at the mine were emptied on the mach ine ; however, the fi re continued to bu rn. Meanwh i le, a pump was instal led and 500 feet of 2-inch plastic water l i ne was ru n to the fire area. Water was appl ied to the fi re, and it was exti ngu ished about 1 - 1 /2 hours l ater. A sma l l dam was bui l t with bags of rock dust to contai n the water to cool the mach i ne and material i n the fire area. State and federa l inspectors assisted in the firefighti n g operations and kept the area under constant supervision unti l it was determined that the area had cooled to the extent that there was no danger of rekind l i ng. 28 

F I R E DYNAM ICS A N D SC ENAR I OS Analysis The fi re apparently started in the cutti ng motor, wh ich was known to be in poor operating condition. The resulting arcs and fl ames ignited accu m u l ations of o i l and coal dust on the machine, some of the hydrau lic hoses, the front ti res, and the wiri ng. This analysis suggests that : 1 . Tra i l ing cables should be prov i de d with suitable short-ci rcu it protection, and there shou l d be some means for disconnecting power from the cable. 2. E lectric equ i pment shou l d be i ns pected as often as necessary to ensure safe operating condi tions, and any defect shou l d be corrected pro mptly . 3. M i n ing equi pment shou l d b e cleaned as necessary t o prevent oi l a n d coal dust from accu mu lating on its su rface. 4. Consi deration shou l d be given to the use of fi re-resistant hydrau lic fl u i d in min ing equi pment. 3.2.3.5 Polyurethane Foam Fire Caused by Spontaneous Combustion ( Freemen 1 969) Summary The secon�shift m ine fo reman detected the fi re in a crosscut between the i ntake and return a i r courses of the m i ne. The fi re occurred when u reth ane foam appl i ed to a ci nder block and wood stoppi n g ignited spontaneously. Of the 64 men in the mine, 1 5 were in areas ventilated wi th air that passed by the fi re. An hou r after the fire was detected compa ny officials ordered al l men except those attempti ng to control the fire out of the mi ne. The fi re was exti ngu ished com pletely with i n the next hour. No i nju ries occurred and property damage was confined to the stopping and foam mach ine. Pre- Fire Conditions The mine was opened by two drifts, three manways, three retu rn ai rways, a rock tunnel, and two openings. The coal was of high-vol ati le bitu m inous ran k and the coal dust was h igh ly explos ive . A total of 244 men were employed and an average of 4,000 tons of coal per day was produced with ri pper-ty pe conti n uous m i ners. Develo pment was by the room-an � p i l l a r method. The m i ne was cl assed as gassy. Venti l ation was i n duced by two ax i al-flow fans driven electrica l l y and exh au sting 4 20,000 cubic feet of air per m i nute. Materi als an d tools for firefighti ng pu rposes were pl aced at strategic locations th roughout the m i ne. Self-generati ng oxygen breath i ng units and gas mas ks were mai ntained u nderground, and sel f-rescuers were avai lable in a l l active work ing sec­ tions. Description On the day of the fi re, three ci nder block stoppings and one com bination ci nder 29 

M I N ES AN D B U N K E R S bloc k and wood stopping separating the i ntake and return ai rways had been coated with u rethane foam on the fi rst shift. When the crew finished coating the stoppings, they cleaned the equ i pmen t, left the portable foam mach i ne near the stoppings, placed the spray gu n in a bucket contai ning acetone near the mach ine, an d departed for the su rface. The 8 foot by 20 foot sl ant was provi ded with two ci nder block stoppi ngs equ i pped with 6 foot by 8 foot wooden doors and was erected to form an air lock. About 45 minutes later, the foreman detected bl ack ro l l i ng smo ke in the hau l age entry, an i ntake ai rway. Upon investigating, he fou nd the smoke com ing from the s l an t conn ecting with the right back rai se, an idle section . Unable to approach close enough to determine the cause of the smoke, he suspected it m i gh t be from the 2, 30o-volt power ca ble. He immediatel y telephoned the su rface and noti fied the superi ntendent. The crew worki n g near the fi re was contacted and instructed to retreat an d remain on intake air u nti l given fu rther i nstructions. The ge neral m i ne foreman, superi nten dent, and safety committeemen departed i mmed i ately for the trou ble area. En route they discon nected the 2,300-volt power. The mine offic ials entered the smo ke area and discovered flame issu ing from the 5-ga l l on bucket containing acetone in which the u rethane foam spray gu n had been l eft to soa k. The fl ame was exti nguished i mmediately with a 2Q- pound dry chemical fire extingu isher and the bucket was removed from the area. A smal l fi re detected at the l ower left corner of the wood and ci nder bloc k stopping was exti ngu ished with th ree bags of rock dust. The general m i ne foreman and an employee entered the return a i r cou rse and traveled with the air to the back of the stopping. They fou nd the wood door and the frame around the door on fire. This fi re was co n· trol led but not extingu ished using two 2Q-pou nd dry-chem ica l fire exti ngu ishers. At this ti me, officials ordered a l l men, except those engaged in contro l l i ng the fire, to l eave the mine. F i re control conti nued by water bucket bri gade and rock dust wh i l e a fire hose was obtained from fire stati ons and con nections made to the nearby waterl ine outlet. When the fire house was placed in service, the enti re area o n the intake side of the stopping was soa ked thoroughly. The hose then was passed through a hole bro ken in the stoppi ng and the fi re on the back side of the sto pp i ng was exti n· gu ished immediately. The fi re ign i ted a sma l l amount of coa l near the stoppi ng, burned a l l the foam from the wood and cinder block stoppi ng, burned part of the wooden door and frame, destroyed all hoses and pl astic con nections on the foam mach i ne, and charred a ti mber set about 7 feet away from the stoppi ng. The th ree nearby ci nder block stoppings also coated with u rethane foam that day were exam ined and charred foam materi al was found at several pl aces. Foam materi al on these stop­ pi ngs ranged up to 1 4 inches in th ickness. Analysis Prior research h as shown that a th ick mass of u rethane foam can ign ite spontane· 30 

F I R E DYNAM I CS A N D SC E N A R I O S ousl y. The charred foam material fou nd on the stopp ings i nvolved i n th is fi re indicated that the fire started spontaneously in foam appl ied in a thick m ass. Analysis of the scenario suggests that : 1 . Permanent stoppi ngs should be constructed of sol id, su bstantial, i ncom busti· ble material. 2. The th ickness of the foam l ayer on a si ngle pass shou l d not exceed 4 inches. Su bsequent appl ications sho u l d not be made unti l the fi rst l ayer is cu red and is n o l onger tacky when touched. Foam app l i cati ons of 1 i nch are adequate for seal ing pu rposes. 3. Workmen assigned to apply urethane foam shou l d be instructed in the proper appl ication procedu re and shou l d be informed of the h azards involved. 4. Areas freshly coated with u rethane foam shou l d be i nspected for heat. 3.2.3.6 Conveyor Belt and Coal Oust Fire Presumably Caused by F rictional Heat (Gay 1 970) Summary The fi re of u ndetermi ned origin occurred near the belt conveyo r head where the conveyor discharged onto the main slope be l t in the m i ne. Smoke was discovered com ing from the a i r shaft at the main fan by the second-sh i ft m i ne foreman, who was work ing the third sh ift as a watch man at the surface faci l ities of the m ine. No i nju ries were susta i ned by persons fi ghting the fi re, and property damage was con· fined to the conveyor be l t and accessories. Pre- Fire Situation The m i ne was opened by a 40-foot shaft and a 350-foot slope i nto the h i gh· volati l e coal bed, wh ich averaged 50 i nches in thickness. A total of 39 men were employed, and an average of 500 tons of coal was produced per day. The m i ne was being developed usi ng the room-and-pi l lar method and conven· tional equ ipment. Pi l lars were not being extracted. The i mmedi ate roof was hard shale of u ndetermined thickness. Ventil ation was induced by a propel ler fan ex­ h austing 50, 000 cubic feet of air per minute. The m ine su rfaces ranged from wet to dry, and dangerous accu mu l ati ons of l oose coal and coal dust were not present in the active un dergroun d workings. Three be l t conveyors were used. One was i nstal led i n the slope an d extended from the slope bottom to a coal-storage bin on the su rface, a distance of approx i· mately 400 feet. The other two conveyors extended i nto the mine approx i m ately 3,240 and 1 ,800 feet, respectively. The conveyor belts were 36 i nches wide and were equ i pped with s l i ppage and sequence switches. Electric power at 1 3, 200 volts ac, reduced by a ba n k of three transformers, was uti l i zed to su pply 440 and 220 volts ac to the surface motor i nsta l l ations an d the No. 2 bel t co nveyor drive motor. 31 

M I N ES AND B U N K ERS Description When the m i ne ceased operations for the wee k, the belt conveyors were emptied of coal i n transit to the su rface. The m ine forman had observed no unusu al condi­ tions at this time. At m i dn i ght, the second-sh ift mine foreman, who was assi gned the du ty of su rface watch man, saw smo ke com i ng ou t of the fan open ing and i mmediately notified the mine presi dent who assigned another forem an to accom­ pany the fi rst i n i nvestigating the fire. The two men wal ked down the slope and, at the bottom, discovered an active fi re in the vicinity of the No. 2 be l t head. They reported thei r findi ngs to the presi dent and requested aid from the fi re department of the two nearby towns. Water was pumped from fi re trucks and from a nea rby river i nto the m ine at several locations. The prol onged bu rning of the belt ignited wooded roof supports along the belt entry and a third coal pil l ar nea r the bel t head . The vol ume of a i r was reduced by opening the explosion door at the fan. Line brattices were used to d i rect the venti l ation air at the sight of the fire. Several permanent stoppings were ope ned to pe rmit smo ke and fu mes to enter the mai n retu rn. The heat generated by the fire wea kened the bolted roof and it fel l along the length of the entry from the slope to the entrance crosscut. Loadi ng of the bu rning material was started at three l ocations. The fire was considered total l y extingu ished when the fal len materi al was co mpletely removed 1 1 days after i nception of the fire. Approx i mately 800 l i near feet of conveyor and 400 feet of rope-su pported stru ctu re were destroyed by the fire . Analysis The ex act cause of the fi re cou l d not be determ i ned because of the extensive damage. I t is be lieved, however, that the fire began after the belt was stopped at the end of the last produ ction shift when frictional heat from a defective bottom belt rol ler i gn i ted the conveyor bel t and accu mul ated coal dust. Ana lysis of the scenario suggests that: 1 . Coal dust and l oose coal sh ou l d not be permitted to accu mul ate i n dangerous quanti ties under, along, or around be l t convey or structures. 2. When stopped prior to idle periods, belt conveyors shou l d be i nspected care­ ful ly in thei r entirety. 3. Belt rol lers and beari ngs shou l d be lu bricated frequently and shou l d be i n­ spected frequently and thorough ly to determ i ne whether there are any bro­ ken beari ngs or "frozen " rol lers. Any defects fou nd shou ld be corrected promptly. 4. The ve locity of air al ong bel t conveyors shou l d be l i m ited. 3. 2.3.7 Coal Bed F i re Caused by Blasting ( fanok and McMonies, 1 970) Summary 32 

F I R E DYNAM I CS A N D SC ENAR I OS The fi re occu rred i n the coal bed after 40 charged boreholes were fi red simulta­ neously at the faces of the east and north approaches in a new shaft bei ng devel­ oped. All of the wo rkmen were on the surface when the fire occu rred. The fi re, wh ich was confined to the east approach, was exti ngu ished by flooding the bottom of the shaft with water. There were no i nju ries and no property damage. Pre- F ire Situation A total of 31 men were employed constructing a new ven ti lation shaft at th is l ocation. The shaft was 1 8 feet i n d i ameter and 483 feet deep. At the time of the f i re, four approaches in the coal bed were being deve l oped Each 1 7 foot by 8 foot approach was to extend 30 feet i nto the shaft bottom. Ventil ation i n the shaft was i nduced by two aux i l i ary-type bl ower fans located on the su rface and powered by 20-horsepower motors. Corrugated tu bing, 20 i nch es in diameter, extended from each fan to with in 9 to 5 feet of the sh aft bottom. F i refighting equ i pment was read i l y ava i l able on the su rface. It consisted of water from three 1 ,000-ga l lon storage tan ks, six 20-pound m u l ti pu rpose fire exti n­ guishers, and an adequate supply of rock d ust. Description Prior to the fi re, six men were d ri l l i ng boreholes i n the coal face of the east and n orth approaches. In order to e l i m i nate an overhang at the face of the east ap­ proach, boreho les, about 2 feet deep, were dril led - two in the roof strata (one in each corner) and three a bout 4 feet apart in the coal bed near the roof. A total of 3 5 holes, 6 feet deep, we re dril led in the coal face of the north approach . Each hole in the east approach was ch arged wi th a half stick of exp losive and each hole in the n orth app roach was charged with four or five sticks of ex pl osive. Al l charged holes were stem med with incombusti ble material and the leg wi res of the deton ators were con nected in series. After the entire crew was hoisted to the su rface, the power was disconnected from the sh aft, the two fans were stopped, and the ends of the fan tu bings were c l osed and raised about 5 feet up the shaft. The 40 charged holes then were fired s i m u ltaneously from a 220-volt ac ci rcu it on the su rface . The fan tu bings were opened and l owered from the su rface and the two fans restarted . About 30 m in utes l a ter, the men were bei ng lowered i nto the sh aft when they encou ntered smoke a bout 50 or 60 feet from the top of the shaft. The men were hoisted to the su rface i mmed i ately. Shortly thereafter the bottom of the shaft was fl ooded with water from three 1 ,000-ga l l on storage tan ks and a nearby creek. About 1 0 tons of rock d ust also were dropped i nto the sh aft. The foll owing day, after approx imately 1 50,000 ga l lons of water had been pumped i nto the shaft, no smo ke cou l d be observed and no carbon monox i de was 33 

M I NES AND B U N K E R S detected i n the air returning from the shaft. The wate r level was 1 0. 5 feet above the bottom of the coal bed when pumping was disconti nued. F ive d ays after the bl asti ng, wi th the water removed from the shaft, the two fans operati ng and no i nd ication of fire in the shaft, com pany officials and personnel of the U. S. B ureau of M ines were lowered in a bucket to the bottom of the shaft. A careful exam i nation i n dicated that the fire was extingu ished. I n the east approach, coke was found on the roof and ri bs for a distance of 10 feet out from the face; methane was bei ng freel y li berated from the u pper right and l eft corners . Analysis The fire probably started when a fi red shot or series of shots, which were overbu rdened and/or unde r bu rdened, ignited gas being emitted by the feeders. Analysis of the scenario suggests that: 1 . Ventil ating fans shou l d not be stopped and tu bing providing face vent i l ation shou l d not be removed d u ring blasting operations. 2. Boreholes shou l d be properly pl aced, charged, and stem med to prevent m is­ fires or bl own-out shots. 3. An exami nation for fi res shou l d be made as soon as the smoke and dust from blasting operations is removed by ve nti l ation in the shaft. 3.2.3.8 Coal F ire Caused by an Electric Arc and a Subseq uent Explosion (Dobis, et al. 1 965) Summary The fire occu rred al ong the north mains track hau l ageway of the mine. An explosi on of the d isti l l ate by- products from the burn i ng coal and other com busti­ bles i n the h igh-temperatu re, lean-a i r env ironment occu rred about an hour l ater, shortly after the main ventilating fan, wh ich had been stopped by a foreman for 1 5 to 20 m i n utes, was restarted. When the fire began two fo remen and five workmen were engaged in m isce l l ane­ ous work near the scene of the occu rrence and a foreman and two workmen were tram m i ng the continuous-min ing mach ine i nvolved in the fire. Six men died of asphyx iation and one man, who was fou nd unconscious, died en route to a hospital. The foreman and two workme n engaged i n tramming the continuous m i ner escaped through the drift portal . The fi re was in itiated b y a short c i rcuit when the top o f the traction pu m p drive on the stripped-down continu ous mi ner being trammed on the north mains track h au l ageway contacted the energi zed trol ley and/or trol ley feeder wires . The resu l t­ i ng electric arc and flame ign ited the ru bber belting used for i ns u l ation on top of the traction pump drive, head coal and ri bs, hydra u l i c hoses, and o i l . U. S. Bureau of M i nes i nvestigators be lieve the explosion originated at the scene of the fire because the intentional stoppi ng of the main venti l ating fan permitted disti l l ate by-products from the burn i ng coal a nd other com busti bles to accumulate in the h igh- 34 

F I R E DYNAM I CS A N D SCENA R IOS tempe ratu re, lean-air env i ronment. These gases were ign ited when the fan was re· started and they were enriched and moved into the fi re zone. The force of the explosi on ex tended through the north mains entries, a distance of a bout 1 ,800 feet, and was dissi pated as it traveled toward the d rift openi ngs and through the fan shaft. Pre- Fire Conditions A tota l of 1 60 me n were employed, 1 40 underground, and an average of 3 , 1 00 tons of coal was produced per day. The mine was opened by th ree drifts and a shaft into a h i gh-volati l e coal bed, wh ich averaged 80 inches in thic kness. Analysis of a raw coal sample taken from the coal bed in the m ine showed a volati l e ratio of 0.46 percen t, indicati ng that the coal dust was ex plosive. A b l oc k system of m i n i ng was fol l owed. M u l tiple entries i n sets of fou r to nine were 1 2 to 1 5 feet wide, and crosscuts were made at intervals of 80 to 1 05 feet. M i n i ng in two sections was accomplished with conventional mechanical equ ipment, a ripper, and a borer-type con ti nuous m i n i ng mach ine. P i l l ars were bei ng partia l l y extracted in tw o working sections. R oof bolts were be ing used in all active areas of the m i ne. The m i ne was classified as gassy. Ventilation was induced by an ax ial-flow ex­ haust fan insta l led on the su rface. Overcasts and permanent stopp ings were con­ structed of i ncom bustible material. Main doors were not used or needed . Check curtains and l i ne brattices were used to conduct a i r to the face areas. The air from all parts of the mi ne was returned to the upcast fan shaft. Each section was venti la­ ted by a separate spl i t of air. The m i ne su rfaces ranged from d ry to definitely wet. Water sprays were used on the two contin uous m i ners, on al l roof-bolting machines, and at several of the main belt heads to al lay the dust at its source. Uniform dust su rveys had been made in the m i ne since it was opened. Apparently the appl ication of rock dust prevented further spread of the explosion. Electric power at 1 1 0, 220, 440, 4, 1 60, 7, 200, and 23,000 volts ac and 3,200 volts de was used on the su rface and at 440, 4, 1 60, and 7, 200 volts ac and 300 volts de was used u nderground. The trolley wire was installed on bel l -type insu lators, 8 to 20 feet a pa rt, at least 6 inches outside the ra il and had a vertical cl earance ranging from 54 to 69 i nches from the top of the rai l along the haul age. Ci rcu it protection for the de trol ley and feeder system was provided by automatic t i me delay recl osing c i rcuit breakers at the rectifier stations. A state-trai ned and ful l y equ i pped mine rescue tea m com posed of com pany person nel was ma intained at the m i ne. Sel f-rescue rs were prov ided for employees undergrou nd and were kept in boxes in the wo rking areas of the various sections. Emergency escapeways from each working section to the su rface were in safe condition for travel and reasonably free from obstructions. 35 

M I NES A N D B U N K E R S Description Prior to the fire, production in the mine was discontinued and a borer-type continuous mi ner was being trammed from the su rface to the west mains faces, a distance of approxi mately 1 0, 800 feet. The machine was partly dismantled to faci l itate tramming through restricted a reas . The tramming operation was per­ formed by two men and su pervised by the shift foreman. Seven other men were deeper inside the m i ne. One secti on foreman and two men were rock dusting in the west mains area ; another section foreman and th ree men were mov ing equ ipment in the north ma ins area. As the continuous mi ner approached the ju nction of west mains, after being moved some 4,000 feet towards its destination, the top of the traction pu m p drive of the mach i ne came i n contact with the energi zed trol ley and/or trolley feeder wi res. The resu lting short circuit caused i ntense arci ng and flame that ign i ted the coal roof and ri bs and the ru bber be l ting used as insulati o n ; the hydrau lic hoses and oil ignited l ater. The sh i ft foreman, who was standing 30 to 40 feet beh i nd the conti nuous m i ner when the fire started, instructed the two men to try to ex tingu ish the fi re wh i l e he proceeded by personnel carrier to the power switches to de­ e nergi ze both the ac and de power ci rcu its ; however, since the personnel carrie r received i ts power from the trol ley wi re, the shift forem an h a d t o recl ose the d e switch. H e tried to contact the trapped m e n by pagephone but rece ived no re­ sponse ; he then tel ephoned the superintendent and suggested that the main venti lat· ing fan be stopped to prevent smo ke from entering the sections where the seve n men were worki ng. The superi ntendent agreed and shut down the main venti lating fan. Meanwh i le, the two men at the continuous m i ner expended the contents of a sma l l fire extinguisher and appl ied 1 5 to 20 bags (30 pou nds) of rock dust to the fire. These efforts we re i neffective, and one of the men, wh ile getti ng a hose, also tried unsuccessfu l l y to contact the trapped men by pagephone. Afte r the h ose was con nected to a nearby waterl i ne, water was appl ied to the bu rning materials, but both men were forced to retreat because of the rol l back of dense smoke. They and the shift foreman then retu rned to the su rface . An hour after the fire bega n or 30 minutes after the fan was restarted , a team e ntered the mine and fou nd the first evidence that an ex plosion had occu rred. The y attempted t o contact the trapped m e n without success, exam ined the situation , a n d c u t the trol ley wire t o remove the sh ort ci rcu it. Rescue teams su bsequently found that several mandoors and stoppi ngs had bee n bl own out by the explosion, causing short circu iting of venti l ation. Twenty-one i nci pient fi res also were fou nd and exti ngu ished . As the rescue operation pro­ ceeded, stoppings were erected, the venti l ation system modified, and air hoses were con nected . Approx imately 1 1 hours afte r the original incident, one secti on fo reman was fou nd unconscious some 4, 600 feet from the fi re area. He was removed immed i ate­ ly but died on the way to the hospital. Shortly thereafter, the bodies of two men 36 

F I R E DYNAM I CS A N D SCENAR IOS were fou n d 2,000 feet from the fire area, and the bodies of four men, 1 ,600 feet from the f i re area. When the seven v icti ms were fou nd, i t was discovered that they h ad th i rteen self-rescuers. The i nner and outer l i ds of two of these u n i ts were detached from the cartridges i n d icating that they coul d have been used ; however, the nose cl i ps were not in place. Two extra unused sel f-rescuers also were carried by the afore­ mentioned men. A self-rescuer removed from the cannister but not activated for use was found in the carrying receptacle on the bel t of the section foreman , who was fou nd unconscious; an unused self-rescuer foun d nearby may have been d i scarded by h i m . The fou r victims in the north mains had seven sel f-rescuers ; however, no evidence was fou nd to indicate that the victims had attempted to u se them . I m med iately after the bodies of the victims were brought to the su rface, all work was d i rected to fighting the fi re by di rect methods. While water was being appl ied at several points, wo rk was started to rei nforce and tighten the plastic stoppings that h ad been erected during the initial ex ploratory tri ps and to replace the tem po­ rary stoppi ngs with concrete bloc k stoppi ngs . Effo rts to move out the fal l s and hot materi a l were started on the fi fth day after the ex ploration . On the tenth day, the contin uous mi ner was u ncove red. Duri ng the next seven days, the mov i ng out of the h ot material contin ued, and the roof and ri bs were supported by cross bars, posts, and n umerous roof bolts. Moving out of the hot m ateri al was completed 27 days afte r the fi re occu rred. The fi re area then was cooled with water and kept u nder su rvei l lance for several shifts. No one was injured d u ring the recovery opera­ tions. Analysis The ci rcumstances under which the gases accumul ated and the soot residue from the fire and/or ex plosion was deposited indicate that the gases entering the explo­ s i on were prima ri ly the disti l l ate by products of coal i n a high-temperatu re envi ron­ ment with a l ow or diminishing oxygen-ai r content. The gases contai ned several h ydrocarbons, carbon monox ide, and possibly a sma l l amount of free hydroge n . Soot and co ke were observed i n a l l eight entries for distances of approx i m atel y 2,350 feet o u t from a n d 600 feet i n from t h e ju nction o f west mains. Twenty-one s moldering fires were found extendi ng 1 ,350 feet from the origin of the ex plosi on . The force of the ex plosion radi ated from the fi re area and traversed distances of a bout 1 ,300 feet out from and 600 feet in from the area. Thi rteen concrete block s toppi ngs were blown out; fou r other stoppi ngs were d ispl aced. Man d oors i n un­ damaged stoppi ngs were either bl own off or bl own open. F i ve overcasts were de­ stroyed and one was damaged. The force of the ex plosion also reached the su rface and one of the explosi on doors at the fan was bl own open but explosive forces d i ssi pated rapidly as they travel ed towa rd the drift openi ngs. D iscon necting switches were not i nstal led in the de trolley and trolley feeder l i nes. I m p roper or inadequ ate short-circuit protection in the de power system per­ mitted the i n tense a rc. 37 

M I N ES AND B U N K E R S No empl oyee was stati oned on the su rface to receive and dispatch emergency cal ls origi nating in the m i ne. No attempt was made to restrict the vel ocity of the a i r over the fire a n d reduce the amount o f smoke a n d gases brought b y the a i r current i nto the i nner mine wo rkings, and n o attempt was made to short c i rcuit the con­ tam inated a i r from the fire i nto the retu rn. The main vent i l ating fan was stopped i ntentionally fo r 1 5 to 20 m i nutes and then restarted without proper consi deration of consequences - a d isastrous explosion resu lted . The location of the bodies indicated that the trapped men h ad tried to reach safety; arrows and cha l k marki ngs made by the victims on stoppi ngs and track ra i l s were fou nd i n several l ocati ons. S i x men died from asphyx iation fol l owing th e m i n e fire and explosion; a foreman found unconscious 22 hou rs l ater presu mably died o f carbon monox i de poisoning wh i l e enrou te t o a hospital . Only 2 of the 1 3 sel f­ rescuers fou nd in the possessi on of the 7 victi ms showed evidence of attempts to use them. It appears that th is fire occu rred when a continuous m i ner being trammed in the mine came in con tact with the energ i zed trol ley and/or trol ley feeder wires ; the resu lting short circu it caused i ntense arcing and fl ame that ignited the coal roof and ribs and hydrau l ic rubber belti ng used as insul ation. When the main fan was stopped, explosive gases accumul ated i n and arou nd the fi re zone; when the fan was restarted, some of the oxygen-en riched gases were moved i n to the fire area where they were ign ited. Analysis of the scenario suggests that : 1 . When moving bu l ky equ i pment along m i ne routes where clearances a re smal l , special precauti ons shou l d be ta ken t o insu re that : ( a ) n o men are worki ng near the area; (b) power wi res are deenergized along the travel route ; (c) a l l employees connected with the tramming operation are th orough ly tra ined concerning procedu res, the location of cutout switches, and the location and use of fi refighti ng equi pment; and (d) a surface d ispatch er is posted to relay eme rgency calls. 2. All organ ic materials used i n un dergrou nd operations shou l d be reevalu ated for fire safety at reasonable intervals to keep up with advancing tech nology. These materials i nclude conductor insu l ation, tempora ry i nsulation, hydrau l ic fluid and hoses, tires, conveyor belt material sealants, and brattice cloth materi a l . 3. Con tro l o f venti l ation, both su rface i nduced or modified u ndergrou nd, shou l d be d irected by the best ava i l able technical authori ty ; decisions shou ld b e based on the latest known information concerning the su bsu rface situati o n (especi al l y t h e l ocation o f m e n u nde rgroun d in case o f fire or imm inent fire/ex plosion situ atio n ) . 4. The safety o f a l l me n undergrou nd shou ld be i mproved by : (a) train i ng the m t o u se se l f-rescuers a n d t o carry such u n i ts at a l l ti mes, ( b) tra i n i ng them i n the procedu res to be fol l owed i n case of fire or ex plosion, (c) tra i n i ng them 38 

F I R E D Y N AM I CS A N D SC ENAR I OS to erect barricades i n an emergency, and (d) training them in firefighti ng procedures. 5. Underground electrical equ i pment shou l d be provided with the best protec­ tion technology can offer, i ncl uding adequate short-c i rcu it protection for de power systems and disconnecting switches in trol ley and tro l l ey feed wires at adequate i ntervals and near the beginning of each branch l ine. 3.2. 3.9 Conveyor Belt and Timber Fire Caused by Electric Arc (Phill ips, 1 969) Summary The fire was discovered at a metal overcast along a bel t conveyor a bout 1 80 feet from the be lt conveyor drive. The fire was i nitiated when electrical arcing ign ited com bu st i ble materials. The arcing was caused when a roof fal l disl odged a metal overcast onto the 4 80-volt ac power wi res. Because it was the mi ners' vacati on period, only two men were i n another part of the m i ne when the fi re occurred. There were no injuries, and property damage was confi ned to a 25-foot section of the belt conveyor, a porti on of the 480-volt ac power wi res, and several wooden ti m be rs. The fire apparently did not spread rapid· ly because the belt entry and surrounding territory were clean and wel l covered with rock d ust. The fi re was ex tingu ished the same day, and the area was kept u nder su rveil l ance unti l it was completely cool ed by the appl ication of water. Pr• Fire Condition The m i ne was opened by three drifts and nine curcu lar shafts into the h igh­ volati l e coal bed, wh ich averaged 84 i nches in thickness. The coal dust was ex­ plosive. A tota l of 204 men were employed, 1 64 u ndergrou nd, and producti on averaged 4, 500 tons of coal per day. Because it was the m i ners' vacati on period, the mine was idle and on l y one section foreman and a pum pe r were in the m ine when the fi re occu rred. The i m mediate roof in the fire area was com posed of 1 0 inches of coal , which was left to hel p su pport a fragi l e shale about 2 feet in th ickness. The m ine was c l assified as gassy. Vent i l ation was induced by th ree ax ial-flow exhaust fans. The belt entry, the supply-track-haulage road, and a parallel entry were u sed as intake a i rways . Al l of the entries i n the area of the fi re were adeq uately rock-dusted. Suitable f i refighting equi pment and materials were provided throughout the m ine. Shuttl e cars transported coal from the face regions to the dumping poi nt on each section. Wel l instal led, flame-resistant bel t conveyors took the coal to the su rface p reparation plant. E lectric power at 1 2,000 vol ts ac was conducted undergrou nd through two boreholes and reduced to 7, 200, 4, 1 60, 440, 220, and 1 1 0 volts. The h igh -voltage c i rcuits were protected agai nst overloads by oil circuit brea kers equ i pped with 39 

M I N ES A N D B U N K E R S i nstantaneous a n d time overcu rrent rel ays a n d grou nd fault tri p rel ays. The u nder­ ground be l t drive motors were operated with 440 volts ac. Description A resident near the mine fan sme l led smo ke and informed a company official. The general superi nte ndent entered the m i ne and other company officials fol lowed shortly. A search of the entire m i ne was made and the fire was discovered about 2 hours l ater by the general superi ntendent. He i m m edi ately deenergi zed the power. The fi re was exti ngu ished with the contents from two fire extingu ishers and fou r bags of rock dust, which we re obtai ned from the nearby bel t conveyor drive. Although the flame was completely extingu ished, additional wate r was applied for several hours to coo l the fire area. The fire area then was patrol l ed u nti l the i nvestigati ng commi ttee arrived. D i fficu lty i n locating the fire was due to the smoke and fumes bei ng diverted directly i nto the retu rn ai rways through the opening created when the overcast was d isl odged. As the fire area was be i ng rehabi l i tated, it was determ ined that the overcast was no longer needed for ventilation pu rposes; therefore , block stoppings were erected separati ng the i ntake and retu rn ai rways. Analysis When the roof fa l l disl odged the metal overcast onto the 4 80-volt ac power conductors, a fault occu rred between two phases of the secondary circu it and the resistance of th e fau l t was too h igh to permit sufficient fau lt cu rrent to flow and open the fuses in the pri mary circuit. The resultant el ectrical arcing ign ited the wooden timbers and bel t conveyor. Analysis of this scenario suggests that grou nding transformers, a cu rrent-l i m iting resistor or an equ ivalent grounding device, and an autom ati c circuit breaker e­ q u i pped with a grou nd-trip circuit sh ou l d be i nsta l led in the secondary c i rcu its from the de lta connected transformers to provide grou nd fau lt protection for the electric c i rcu its and equ i p ment. 3.3 Fire Dynamics F i re is one of the m ajor hazards of underground m i n i ng. The inherent danger of f i re is greatly exacerbated by the n atu re of mi nes - i .e., restri cted a reas, l i m ited accessi­ bil ity and egress routes, forced ventilation, the presence of bu l ky eq uipment with high-powe r electri c compo nents, roof support problems, a potential for explosive concentrations of dust and gas i n the a i r. A fire situati on i n a m i ne is fu rther complicated by the often complex l ayout of the m i ne and the associated airflows and velocities i n various sections of the mine. G iven th is situation, it is that the natu re and behavior of fi re i n a m i ne environ­ ment be well understood. Exte nsive wo rk has been done in th is area by many nations and a body of information developed on the bas is of experience and post- 40 

F I R E DYNAM I CS A N D SCENAR IOS fire investigations. Government and industria l regu l ati ons based on this knowledge h ave greatly improved fi re prevention and control. New and mo re sophisticated approaches a re needed, however, to fu rther reduce or e l i m i n ate the hazards of mine fires. R esearch efforts d i rected toward the in· depth understand i ng of the h igh ly complex processes associated with m ine fi res shou l d be conti nued and expanded . Of particu lar importance are wel l designed, medium- and l arge-scale ex peri ments and the col lection and correl ation of the great variety of data n eeded to deve l op mean i ngful theories and models. The material presented bel ow is paraphrased or qu oted from a report by G reuer ( 1 973). I t is intended to summari ze the state of fire dynam ics knowledge related to mine fires. Pl ease note that on ly the qu oted words are G rau er's and some judicious deletions have been made without altering the mean ing. 3.3. 1 "Properties of Mine F ires" 3.3. 1 . 1 "Modes of F ire Propagation" Mine fires propagate : " 1 ) through l ocali zed heat feedback from the flames; 2) through all over heat feed back from the fu mes . . . " " F i res of the fi rst type a re control led by the same mechanisms as u nconfi ned fires in the open . " (i .e., • . rad iation and convection from the fl ames and hot gases heat the com busti ble mate­ rial in the i mmediate vici nity of the fire before they m ix w ith the general ai rstream . The l atter does not become hot enough t o ignite o r t o generate gaseous fuel from the com bust i ble materia l ) . Because com bustion occurs on ly i n the i mmed i ate vicini­ ty of the com bust i bl e materia l , considerable quantities of ox ygen can pass through the fire with out be ing consumed. F i res of th is type are therefore frequently cal led u nconfined or oxygen-rich fires. The second type of f i re occurs where the genera l air stream becomes hot enough to generate gaseous fuel from the com busti ble materi al, along wh ich it passes . Th is type of fire grows until all ava i l a ble oxygen is consumed, l i m iting the heat devel op­ ment and frequently is termed "confined" or "fuel-rich" because the high tem pera­ tures req u i red for th is ty pe of fi re propagation are , outside of m ines, reached only i n confined passages. Although most accidental m i ne fi res bei ng started by rel atively smal l ign ition s ou rces, develop i nto oxygen-rich f i res and stay oxygen -rich , fuel- rich fi res have been m uch more thorou{tl l y stu died. Figure 1 is a schematic represen tation of a fue l -rich fire. The fi re h as al ready passed th rough the cooling zone and only heat transfer processes due to forced convection occur (i.e., the wal ls of the ai rway are cooled and the air is heated ) . I n the charcoal zone, t h e carboni zed res idue o f the fue l which sti l l is hot enough to react with the ox ygen in the air is burned. The heati ng of the ventilation air continues, and i ts oxygen content is red uced. In the combustion section of the py ro l ysis zone, the volati l es produced by the decom position of the combust i ble material are bu rned in the venti l ating air. The gas 41 

M INES AND BUNKERS I I I I :���w�:::m�h�=�IJ���� � �i t" *# I TTTI 1 I I r1 I IT 1 I F Coo l i n g Zone Ch arcoal ombust i o E x cess Preheatin g Zone Zone Fuel Pyrolysis Zone Figure 1. Zones dellfi/O(Jtld by fuel·rich fires (Greuer, 1973). temperatu res rise to a max i m u m and the oxygen content is reduced to zero. I n the excess fue l section of the py rol ysis zone, the fumes are su fficiently hot to cause pyrolysis of the com busti ble material . The a bsence of oxygen w i l l , however, pre­ vent com bustion and the pyrolys i s products remain as excess fuel in the fumes. The heat consumption of the pyrolysis ca uses the tem peratu re of the fu mes to drop. F inal coo l i n g of the fumes occu rs in the preheating zone where the ai rway section in front of the fire is preheated and dried. Heat is transferred from the fumes to the airway ma i n l y by forced convection, but radi ation also may be i n ­ volved cl ose t o the pyrolysis zone. F igures 1 and 2 i l l ustrate the zones of a fuel -rich and an oxygen-rich fire respec­ tively. Because vol ati l es evolve only in the i m mediate vicinity of flames, no py roly­ sis zone exists in the l atter. The decrease in the oxygen conten t and the increase i n the carbon content o f the a i rstrea m a s wel l a s its temperature i ncrease i n the combustion zone wi l l be less for oxygen-rich than for fuel-rich fires. 3.3. 1 .2 Controll ing Mechanisms and Equilibrium States In an u nconfined fire with one-di mens i onal fl ame spread ( F igure 3 ) , the rel atio n between the rate of fl ame advance, V, and the width o f the fire, L , will b e a s shown by curve 1 in F igure 4. For L < Lo , no propagation of the fi re takes place due to i nadequate heat transfer ahead of the fl ames . For L < Lo. V wi l l increase with L as the emissivity and height of the fl ames increase until the extension of the fire, L, has reached such a magnitude that V is no longer influenced by L. If D is the depth of the fuel bed consu med by the fi re, a its specific weight and B the vel ocity with wh ich the fire penetrates the fuel bed, the rate of fuel added to the fire (per u n it width of flame front) is V * D * a and the rate of fuel consu mption is L * B * . I n a fu l l y developed fi re, V and L are co nstant with respect to ti me and V * D * = L * B * hence 42 

F I R E DYNAMICS A N D SC ENAR IOS Cooling Zone Preheating Zone Figure 2 Zones delltlloped by oxygen-rich fires (Greuer, 1973). 8 t--- L ---1 Figure .l Model of one dimensionel spread of fire. V = ( 8 /0 ) L (as indicated by cu rve 2 i n F igure 4 ) . Poss i bl e values of L and V are i n d i cated by the two intersections of cu rves 1 and 2. Of these, LV, is, however, n ot a stable conditi on (si nce for L < L1 cu rve 1 shows V < (8/0 ) L, wh ich indicates a decreasing fi re, and for L > L1 it shows that V > (8/0) L, wh ich indicates an i nc reasing fire) . If a fire is i nitiated with L < L 1 , it wi l l die out, and if it is i n itiated with L � L1 , it wi l l adjust itse l f to the condition L = L2 • The rate of fuel added to a fire in a m i ne roadway or duct is best expressed by the dimensionless parameter: V+ = C*V* O *a * Pf/(Va * a * A) 43 .. 

M I N ES A N D B U N K E R S Figure 4. Relationship between V and L in unconfined fires. where C = the mass of a i r requi red for com plete com bustion of unit mass of fuel , P = the perimeter of th e roadway, f = the fracti on of perimeter that is covered with com bustible mater ials, V = air ve locity, aa = the specific weight of the air, and A = a cross section of the roadway. Analogously, the rate of fuel consumption is expressed by : L� = C * L * B * a * P*f/( V a * aa * A ) D u ring its early stages a f i re in a m ine roadway o r duct is control led b y local heat transfe r effects close to the fue l surface and therefore, it beh aves l i ke an u nconfined fire ( F igu re 5). As the fire increases, however, additional heat transfer from the fumes wi l l occur and prov ide an additi onal i ncrease of V� w ith L� which becomes very large when the tem peratu re of the fumes exceeds the th reshold beyond which pyrol ysis of the fuel becomes very rapid. A pea k is reached when L� = 1 when a l l the oxygen i n the a i r supply i s consumed. A further increase i n L + results in a decrease V � si nce excess fuel causes the te mperature of the fumes to drop. Si nce for fuel rich fires the pyrol ysis zone has to have a certa i n length and considerable cool ing of the fu mes due to heat transfer to the walls occurs and since a consi derable temperatu re difference between combusti ble material and air m ust exist to faci l i tate heat transfer, the heat quantities needed to ign ite a fuel rich fi re are most probabl y much h igher than predicted . A burn ing rate of 40 l b/m in of m i neral oil ove r 1 0 m i n utes i n a venti l ation current of 22,000 ft 3 /m i n was about the m i n i m u m to start a fu.e l rich t i m ber fire. 44 

F I R E DYNAMICS A N D SC ENAR IOS Figure 5. Relstionship between V • end L + in confiMd timber fires. 3.3. 1 .3 Observations i n Accidental and Experimental M ine F i res T i m be r h as pl ayed a promi nent role in most l arge m i ne fires and almost a l l p u b l i shed observati ons o n the properties o f accidental fi res deal with ti m ber fires; hence, systematic experimental investigati ons have concentrated on them . "How­ ever, the fire control l i ng mechani sms and the rel ati onsh i p of the dom inating para­ meters do not depend on the ty pe of com busti ble material so that the insights ga i ned from ti m be r fi res can be used for other fi res, too." The great number of fire ex peri ments conducted routinely by experimental m i nes a l l over the wo rl d usually aim at testi ng the inflammabi l ity or fire resistance of materials and equi pment used underground or at measu ring the efficiency of fire e x t i nguish ing devices . Si nce the fi res have usua l l y no opportu nity to deve lop fu l l y to an equ i l i bri u m state, the results obtai ned from these tests have l ittl e general val idity. ( 1 ) Composition of Combustion Products The composition of combustion products can be calcu l ated when the fuel com ­ position and the air/fuel ratio is known . For industri a l fuels . . . charts exist which relate C02 , CO and 0 2 concentrat ions i n the fu mes with the fue l /a i r ratio. I t is concluded that most acci dental m i ne fi res are oxygen rich fi res since they are started by relatively sma l l ignition sources. (a) Fumes and Dispers ion 45 I 

M I N E S AND B U N K E R S The fu mes h ave a tendency to form a layer al ong the roof. In u nventi l ated or l a m i nar venti l ated ai rways this layer is dispersed by mol ecu lar motion only, which is a very slow process. Tu rbu lent dispersion requ i res that the kinetic energy of the turbul ent particles, formed by the venti l ating air cu rrent is h igh en ough to over· come the buoyancy forces. Since th i s ki netic energy is proportional to the air vel ocity, a certa in minimum velocity is requ i red for the turbu lent d ispersi on of such layers. I n l i ttle-venti lated airways the fumes wi l l therefore mainly travel and the fire w i l l propagate al ong the roof. I n i ncl i ned descension a l l y venti lated airways air can flow i nto the fire down-h i l l along the floor whereas the fu mes are trave l l i ng and the fire is spreading u ph i l l al ong the roof. The velocities with wh ich mi ne fi res spread in u nventi l ated ai rways or with wh ich they spread i n ventil ated airways upwind aga inst the air curren t are smal l com pared with those in oth er directi ons . The i r propagation is easier to control, too. Backing of smo ke can be fought by i ncreasing local air vel ocity and using transverse brattice or shields to block the l ower cross secti on of the airway . No systematic investigations on fire propagation velocities against air cu rrents and few on the extension of backed smo ke layers are known al though many of the insights gai ned from the studies of gas l ayers wou l d be transferable to these problems. F i re propagation as wel l as the backing of smo ke aga inst the airfl ow h as so far always been negligible i n West German coa l m i nes. The reason is most probably the h i gh air velocities associ ated with l ongwa l l mining. Only a few cases are known where the backed smoke reached extensions of u p to 1 00 ft. F i res propagati ng a l ong the roof against the vent i l ating air current cou ld always be extingu ished easi ly. I n the experi mental coa l m i ne of the United States Bureau of M i nes th e length of the backed smo ke layer was 1 00 ft . with an air ve locity of 1 20 ft./m in. 50 ft. with 1 80 ft./m i n . , and 1 0 ft. with 230 ft. /m i n . In the best known instance of backed smoke which wa s in 1 91 0 at Wh itehaven Col l iery in G reat Britain with 372 yards aga i nst an i ntake vent i l ation speed of some 325 ft./m in . ; 86 men were lost. In this case, however, th e venti lation must have been descensional and one can suspect that intermittent a i rflow reversals occurred. In f i re ex perime nts in timbered tunnels the fire ve locity aga inst the airflow d id not exceed 20 ft. /hr. H owever, in h igh volati le coa ls the rate of propagation ou t· ward can be both rapid and extensive whereas its rate of travel i nward is fai rly slow or negl igible. On occasion the rate of spread can be 1 50 ft. /hr. aga i nst a ven ti l ation speed of 1 50 ft. /m i n. ( b) Heat of Reactions The heat of reacti on fo r com pl ete com bustion is equal to the l ower cal ori fic value of the fue l. For incomplete com bustion it m ust be corrected by su btracti n g the heat o f com bustion o f a n y u n burnt f u e l d u e t o l ack o f oxyge n or d issociatio n . 46 

F I R E DYNAMICS A N D SC ENAR I OS One h as to consider, fu rthermore, that the products of combusti on can incl u de sol i ds as wel l as gases . The ca lculat ion of the adiabatic flame temperatu re is relative ly cum bersome, si nce, in addition to the specific heat vari ations of the com bustion p roducts, the i r composition ch anges with tem peratu re, due to dissociation , m ust b e considered . Besides trial and error methods the use of charts is therefore advisable . Cal culations of ad iabatic flame tem peratu res are based on the assum ption of a perfectly mi xed gas stream pass ing th rough the fire . This assu m pti on h olds for fuel-rich fires but not for oxygen-rich f i res. Therefore , one can , for the latter, e xpect a wide range of flame tem pe ratures bei ng present in one and the same fire. I t is possi bl e to esti mate the highest tem pe ratu re the fumes can theoretica l l y reach after the combustion products have been mi xed w ith the excess a i r . I n oxy­ ° gen -rich ti m be r fi res a h igher tem perature th an 1 ,692 F has never been observed . T o determ i n e the te mperatu re o f the fumes req u i res the same ca lcu lations which a re necessary for the calcul ation of the ad iabatic flame temperatu res. An addition al complication is that ma ny py rolysis products enter the fumes between the flam e­ z one and the gas samp l i ng point. Experience shows that the tem peratu res of the fumes beh ind oxygen-rich ti m ber ° fires are considerably l ower than 1 , 700 F. If they were th is h igh , a pyrolysis zone wou ld devel op and the fire wou l d become fuel- rich . Beh ind fuel-rich fi res the tem peratu res are much lower than the ad iabatic flame tem pe ratu res. The reasons for th is are the large heat transfer from the fl a mes to the wal l s of the airway . caused by the h i gh flame temperatu res and the addition of gaseou s pyrolysis products at tem peratu res lower th an those of the gas stream. ( 2) Coal F ires Coa l f i res, especia l ly coa l dust fi res, are more frequent than ti m ber fires. How­ e ver, less attention is given them in the l i terature and very few results on system atic f i re ex periments have been pu bl ished. The reason may be th at they are less danger­ o us th an ti m ber fires. Al though coal dust is a m ateri a l w ith one of the lowest i gn ition temperatu res encou ntered u ndergrou nd, the rate of growth of a fi re in coa l d ust is low. It is about 3 1 /2 i nches/hr., droppi ng to 2 i n . /hr. when the coa l d ust is m i x ed with stone dust. Coa l dust fi res, therefore, do not pose any serious th reat to l i fe or production. The danger of coal dust is i nstead as an ign iter and fuse to other more flam m a bl e materials. " Lump or sol i d coal fi res are usua l l y not too serious a r isk to l ife either, espec ial ly i n the i r early stages." Coal, has approx imately 5 per­ cent volatiles in anth rac ite and 40 percent vol ati l es in h igh volatile bitum inous coal . H owever, th is is a much lower volat i l e content than wood which h as 75-85 percent ° vol ati l es. Su bstantial disti l l ati on of vol ati l e m atter on l y begins at about 600 F and ° becomes rapid at about 1 ,300 F whereas the corres ponding tem peratu res fo r wood a re approximatel y 400 and 550° F . Also the coa l ignition temperatu res are h igher, a l th ough the latte r are not constants but depend on oxygen concentrations, ex­ pos u re times, m i neral contents, etc. 47 

M I N ES A N D B U N K E R S Due to its economi c i mportance a s one o f the pri ncipal energy sources, the com busti on of coal in i ndustri al processes has been the su bject of a vast number of i nvestigations. Th is d oes not however appl y to the accidental com bustion of coal i n u nderground mi nes, where, except for fire exti ngu i sh i ng equ i pment, no systematic work seems to have been done. Due to the complexity of fire, it wou ld be risky to try to devel op a mode l for such fi res. Since coal is a fue l that is d i fficu l t to ignite one can expect coal fi res, at least for higher ran king coals, to be oxyge n-rich fi res. Fuel· rich fi res w i l l mainly occu r as a temporary phenomenon after a reduction in venti lation . Few d ata have been pu bl ished o n the composition of com busti on products of open u ndergrou nd coal fi res. Al l deal wi th oxygen-rich fi res, proba bly due to the l i m i ted extensi on of the fuel beds provi ded. No data h ave been pu bl ished so far on fuel consumption, extension and vel ocity of coal fires. 3.3.2 Temperature of Fumes Behind the F i re Zone All major forces exe rted by mine fi res on the ven ti lation can be considered as thermal forces. For their determi nation the k nowledge of the tem peratu re changes, caused by the fi res, is indispensa ble. Th is appl ies less to flame· or pyrolysis zones, wh ich are comparatively short, than to the m uch longer airway sections downwi n d of the fire wh ich c a n experience considerable tem perature changes, too. The temperatu re of the fu mes leaving a flame- or pyrol ysis-zone is changed by m i x ing wi th other air currents and by heat exchange with the airway walls. The effects of m i x i ng are easy to descri be. Except for the entropy, each individual property of the m ixtu re is equal to the mass-weighted arithmetic mea n of the sam e properties of the constituents. No fu rther detailed discussion of m i x i ng processes is therefore necessary. Less simple to descri be are the heat exchange processes . 3.3.2.1 Steady State Heat Exchange with A irway Walls The assumption of an infinite heat capacity of the rock surrou nding an ai rway l eads to constant rock temperatu res and to a steady heat exchange process between air and rock. If the rock, before exposu re to the fu mes with the temperatu re t, h ad assu med t h e average tem peratu re t o f the venti l ating air, the heat transfer d from the fu mes to the rock can be descri bed by : whe re a = heat transfer coefficient p = peri meter of ai rway L = length of a i rway Th is w i l l cause a change of tem peratu re in the fumes by : 48 

F I R E DYNAMICS A N D SCENAR IOS dq dtf = - -­ G * cp whe re G = mass flow rate of fu mes Cp = constant pressure specific heat of fumes W i th tfo = temperatu re of fumes at the begi nning of the airway ( L=O ) one obtains a p L tf = ta + (t fo - ta l exp G * cp The mean temperature of the fumes in the airway, wh i ch is frequently used for the determ i n ati on of therma l forces, i s : aP * L 1 - exp G * cp 3.3.2.2. Non-Steady-State Heat Exchange with Airway Wal ls A m ore accu rate treatment has to take i nto account the l i m ited heat capacity of the rock su rrounding the airways. Whenever a heat exchange between the air and the wal l s occurs, a gradua l l y thickeni ng l ayer of rock wi th te mperatu res between the original rock tempe rature and the air bu i l ds up. This layer form s an insul ation and lets the heat exch ange dec l i ne with ti me. Due to the i mportance of accu rate tem perature precalcu l ations in deeper m ines with cl imatic difficul ties, many attempts at the calcu l ation of the non-steady state heat exch ange between ai r and rock have been made. Within the scope of th is report it is imposs i ble to quote every single paper wh ich ex ists on th is topic. On ly those, wh i ch became better known because they suggested an i m provement of existing sol uti ons and whose res u l ts are appl icable to m i ne fires shal l be discussed. The non-steady state heat exchange between rock and air is a very com plex problem since it comprises ( 1 ) the heatflow by conducti on with in the rock, (2) the heat transfer by convection and rad iati on from the rock to the air or vice versa, and (3) the heat transfer associated with mass flow. Heat f l ow by conduction can be determ i ned with the hel p of F ou rier's equati on of heat conduction : at 2 1 = a V t+ -- ' W aT c*y 49 

M I N E S AND B U N K E R S where t = rock temperatu re T = ti me a k = therma l d iffusivity of rock , cy k = heat conductivity of rock c = specific heat of rock ' y = spec ific we ight of rock W = heat ge neration per u n i t vol u me inside rock In applying th is equation on the heat exchange in m i ne a i rways, the fo l l ow i n g assu mptions are usua l l y made : the roc k is homogeneous a n d isotropica l , t h e tem · peratu re o f the rock su rrou nd i ng the ai rway is un iform a t the begi n n i ng o f the heat exchange, th e airway has a circular cross section, the heatflow para l l e l to the ai rw ay is neg l i gi ble, no heat sou rces or s i n ks exist in the rock , no change of phase of the rock h u m i d ity occurs. Fourier's equation of heat conduction can then be expressed i n cy l i ndrical co­ ord i nates i n the fol lowi ng form : 2 at a t 1 at -= a - + -* ­ 2 aT ar r ar Under the add itional assumptions : the origi nal rock tem perature remains p re· served at a sufficient distance fro m the airway ; the wal l temperature equals th e a i r tem peratu re ( heat transfer coeff. o < = oo ) ; the a i r tem peratu re along the airway does not change with ti me. ( 1 ) Mass Transfer In the preheating zo ne of a fire, where sign ificant pyrolysis or other chem ical reations no l onger take p l ace the ma i n sou rces for heat transfer by mass tran sfer a re the condensation and evaporation of water. Principa l l y , heat transfer by mass trans­ fer cannot be cons idered separate ly from heat transfer by convection at the su rface and from heat flow by con duction in the interior of a wal l , since the mass tra nsfer changes the therma l properties of both the su rface and i nterior. Due to its tech nical importance a vast amount of research has been done on th is problem ; th is report has to l i m i t i tse l f to such work done on mi ne airways. (2) Velocity of Fire Propagation F i re propagati on along a roadway is due to heat transfer by radiation and con ­ vection. T h e l atter ta kes place a s natura l convection, caused b y air movement a s t h e resu lt o f buoyancy forces of t h e hot fumes and a s forced convection, caused b y the 50 

F I R E DYNAM ICS A N D SCENAR IOS ventil ating a i r current. Since the air movements creating the convection provide the fire at the same time with fresh oxygen , heat transfer by convection has a larger influence on th e propagati on of fires than rad iation. And si nce the a i r velocities as the result of buoyancy forces are qu ite l ow, forced convection in the d i rection of the venti l ating a i r current is usual l y considerably larger than natura l convection . A f i re i n a n unventi lated o r l i ttle venti l ated airway w i l l therefore spread in both d i recti ons, u pwind and downwi nd. Since the oxygen supply u pwind is better, the propagation ve locity in th is d i rection may even be higher than downwind. The h igh er the velocity of th e venti l ating air cu rrent becomes, however, the stronger the tenden cy of a fire to spread down wi nd. ( 3) Rad iation Nonpo lar sy mmetric mo lecu les such as 0 2 , N 2 and H 2 are rel ative ly transparent to thermal radiation. Heat transfe r radi ation is, therefore , neglected u nder ord i nary venti l ation conditions with few exceptions by venti lation engi neers. Polar, un· sym metric molecu l es, such as C0 2 , H 2 0, CO, S0 2 and many hyd rocarbons can enter i n to th ermal rad iation exch ange appreciably at the tem pe ratu res encountered in m i n e fires. The i r contri bution to the heat exchange between fum es and ai rway wal ls should therefore be assessed and if necessary taken i nto accou nt. The order of magnitude of heat transfer by rad i ation from a gas to a black surrou nding can be ca lcu l ated by : 2 whe re = constant = 0. 1 7 23 Btu /ft. 0 R 4 hr. = area of black su rrounding emi ssiv ity of gas at temperature Tg a bsorptiv ity of gas at tem pe rature Tg for rad iation from a black body at tem peratu re T w I f the su rrou nd i ng is not black but gray with an em iss ivity t: gr • the net heat tran sfer can be fou nd by considering success ive absorpti ons and refl ections . H ottel and Egbert ( 1 942) performed a ca lcul ation for t: gr = 0.8 - 0. 9, the range m ost freq uentl y encountered. They suggest the use of a factor t: eff to take i n to accou nt the dev iation of a gray from a black surrou n ding and find that this factor can be approx i mated by : s gr + 1 t: eff = --- 2 51 

M I N ES AN D B U N K E R S The emissivity e: g is a fu ncti on o f the gas concentration, t h e th ickness of the gas body and the temperature. Because concentrati on and th ickness com plement each other in the i r effects on the radiation, it is possi ble to consider the em issiv ity as a f unction of the i r product. It has become customary to express the concentrati on by the pa rti a l pressure Pc of the radi ati ng gas measured i n atmospheres and the thick­ ness by the average length of paths for the radiant beams, measu red in feet. Jakob ( 1 959) shows as an example a diagram for the emissivity e: of C0 2 as a function of the product Pc L and the temperature Tg · 3.3.3 Forces Developed by Fumes Temperature changes of the vent i l ating air, caused by mine fires, h ave two m ajor effects on the venti l ation : • a thrott l i ng effect, • a natural d raft effect. Before both effects are n u merical l y assessed it seems advisable to define the energy sca les i n wh ich they are measured. Energies i n m i ne venti lation a re either expressed per unit weight of a i r (ft.-l b./l b.) a n d then called heads ( h ) , or per u n i t vol u me o f air ( ft. -l b./ft. 3 = 2 l b./ft. ) and then cal l ed pressu res ( p ) . Since the re l ationsh i p between unit weigh ts and volumes is the specific weight a ( l b. /ft. 3 ), heads and pressu res are rel ated by the speci fic wei ght, too: h = p/a. With i n a venti lation system a can u ndergo considera bl e ch anges. This has the consequence that equal e nergy qu antities must someti mes be expressed by consider­ a bl y d i fferent pressu res, a fact wh ich comp l i cates the appl ication of energy bal­ ances. To overcome this difficulty but sti l l to mai ntain the uc;e of the fam il iar units of pressu res, heads are frequently ex pressed as pressures by m u l tiplying them w ith a consta nt conversion factor, based on a standard specific weigh t of as 0.07 5 = 2 l b./ft. 3 and the factor 1 /5. 1 94 i nches WG/( I b. /ft. ) : 0.075 h (ft. ) * -- = h (in. WG ) 5.1 94 The th rottl i ng effect arises from the summation of heat and pressu re losses whi ch res u l t from temperature changes i n a m i ne fire. I t w i l l occur mainly in the immed i ate v i ci n ity of the fire. A certain additional thrott l i ng effect is caused by the changes in k i netic energy of the preva i l i ng air current. ( For a sophisticated m athe­ matical treatment of these rel ationsh i ps and effects see G reuer's R eport pp . 68-7 1 ). The great n u m be r of mislead i ng statements fou nd in the l iteratu re concerni ng the natural draft ma kes it advisable to first discuss its defi n i ti on and the m ore popular methods for i ts determ i nation. 52 

F I R E DYNAM ICS AND SC ENAR IOS N atural draft is caused by the conversion of heat i n to mechanical energy, which is then avai lable to propel the air and to overcome friction losses. For such a conversion u nder steady state conditions cycl ic processes are necessary , which are prov ided by every loop of the venti lation network. The amount of heat converted in a cyclic process into mechanical work is indicated by the area encl osed by th is process i n a �v d i agram. Th is statement can easily be proved by applying the fi rst law of thermodynamics, in venti l ation frequently cal led the energy equati on , in the form : Z = elevation for a i rways without fans to a loop, which results i n : f dh< is the sum of all friction l osses experienced by the air, flowing th rough th is l oop, f v dp is the energy necessary, to balance these l osses. I f, in the ai rways of the l oop, mechanical energy dw is exchanged between a i r a n d su rrou nding (fans, d ropping water, etc. ), one obtains - ; v d p + ; dw = ; dh L I n th is case both, the heat energy f v dp and the mechanica l energy f dw have to bal a nce the l osses. The natu ral draft expressed as energy per unit weight of air is cal led the natural venti l ation head h N and its magn i tude is consequently : N atu ral draft is always tied to a cycl ic process, to a l oop . State ments on natural draft without specifying the loop where it is devel oped are not too meani ngfu l . a a dZ + dp = 0 F or the l oop formed by the raise and the hose one obtains : - t ( o f - o a l dZ = f dp and si nce - f ( o f - o a l d Z = P N and ; dp is th e reading of the m anometer, th e latter i ndicates p N · 53 

M I N ES AND B U N K E R S 3.3.4 Qual itative Prediction of Ventilation Disturbances Caused by Fires Thrott l i n g and natural draft effects can cause cons iderable cha nges i n the quan­ tity of venti l ati ng air cu rre nts and someti mes even reverse thei r direction . These changes are not l i m i ted to the ai rway at a fire but can occur in neigh bori ng ai rways as wel l . T h e dangers ca used b y air qu antity reductions ca n i n nongassy m ines be neg­ l ected. I n gassy mi nes they ca n, however, lead to formation of explosive m ixtu res with a l l the i r perti nent dangers, wh ich a re especi a l l y obvious wh en the explosive m i x tu re travels through the firezone. Airfl ow reversals can be the cause of even more severe hazards. CO- l aden air can e nter the i ntake ai rways and poison large sections or all of the m ine. In gassy m i nes explosive mi xtu res can be formed since every ai rflow reversal is preceded by a period of ai rflow reduction. Even in non-gassy m i nes explosions can be caused by explosive fumes, which after a reversal of a i rfl ow reach the fi rezone agai n . Ventil ation disturbances i n the form o f smoke laye rs have bee n discussed . Si nce they a re an easy to survey local phenomenon and can always be fought by l ocal air vel oci ty i ncreases, no fu rthe r comments seem to be necessary . 3.3.4.1 Horizontal Airways Open fi res in h orizontal airways , with only negl igible temperatu re ch anges i n fol l owi ng non-hori zontal ai rways, have a th rottl ing bu t no natural draft effect. The resu lt is an airflow decrease in the ai rway on the fire and a l l airways in series with it. D ue to sma l ler friction losses in these airways , the venti l ating pressure of the airway on fi re wi l l i ncrease and cou nteract the th rottl i ng effect. Air quanti ty de­ creases of up to 30 percent were observed, however, rockfal l m ay h ave been at least partial l y responsi ble. A reve rsa l of airflow i n the airway on fi re and the perti nent airway in series cannot occu r. It is, however, possible in diagonal ai rways, wh ich are connections between paral lel airways, whose airflow directi on is determ ined by th e resistance ratios of the parallel airways . 3.3.4.2 Ascensionally Ventilated Airways Open fi res in asce nsiona l l y ventil ated ai rways also cau se a th rottl i ng as a natu ral draft effect. If the temperatures or the elevation ch anges beh i nd the fire are not too sma l l , or the a i r qu antities are not too large, the natu ral d raft w i l l usual ly be stronger than the throttl i ng effect and i ncrease the ai rfl ow. If enough com bustible material is present, the i ncreased oxygen supply wi l l then i ntensify the fire so that considera ble natural d rafts are fina l l y deve l oped. The in crease in a i rflow in the a i rway at fire is accompan ied by a decrease i n parallel airways. I f the ori ginal venti l ating pressu res for t h e parallel airways are sma l l , even airflow standsti l ls and reversa ls with a l l the i r dangerous consequences 54 

F I R E DYNAM I CS A N D SC EN AR IOS can occur. Remed ies for stabi l i zing the ai rflow i n para l lel ai rways are an i ncrease of the resistance of the ai rway on fi re, wh ich wou l d at the same time red uce the oxygen supply and f i re i ntensity, and an increase i n the venti l ating pressu res . The latter aim ca n be accom plished by i ncreasing the fan pressu re or by l oweri ng the resistance in the intake and return ai rways to the paral l el airways. 3.3.4.3 Descensionally Ventilated Airways Open fires in descensional l y venti l ated a i rways ca use , besides the ever present th rottl ing effect, a natural draft, wh ich is opposed to the origi nal venti l ati ng pres­ sure and has a tendency to decrease or eve n reve rse the airflow. A decrease i n a i rflow usual ly decreases t h e intensity of the fi re, too. The reduced natu ral draft, agai n permits a l a rge r a i r supply, wh ich in turn increases f i re i ntensity and draft. Except for the case of l i mi ted natural drafts due to smal l elevation ch anges beh ind the fi re and a lack of com busti ble material or for the case of very h igh origi nal venti lating pressu res, one can ex pect a violent fl uctuati on of the airflow i n desce n­ sional l y venti l ated ai rways on fire . Whether a perma nent airflow reversal takes pl ace depends on seve ral factors . There have to be sufficient elevation changes for the fu mes on both sides of the fire. A fi re at the bottom of a shaft or raise w i l l not develop enough n atu ral draft to initi ate a reversal, a fi re at the top not enough to mai ntain it. The oxygen content of the fumes is usually consi derably lower th an th at of the fresh a i r. I t i s especia l l y low behi n d fi res of h i gh i n tensities. If these h igh i ntensities effect an ai rflow reve rsa l , the fi re is at fi rst ventilated with ox ygen poor a i r, which wi l l reduce the fire i ntensity. If the plug of oxygen-poor fu mes, trave l l i ng back through the fi re, is long enough, no permanent reversal w i l l be possi ble. When decrease of a i rflow, stan dsti l l or reversal ta ke pl ace as early as possible permanent reversa l m ay occu r. Th is ca n be su pported by a l ow origi nal ve nti l ating pressu re acting on the ai rwa y, or by a fi re developi n g fast to a h i gh i ntensity . Permanent reversal after a longer fi re duration can occu r, when the fire is sup­ p l i e d with oxygen from damaged compressed air l i nes or from other ai rways, joi n­ i ng the retu rn a i rway of the fi re. While in ascensional ve nti lation a i rways pa ra l l e l to the airway fi re are endangered by a i rflow standst i l l s and reversals, in descensional ventil ation the airway on f i re i ts el f is endangered most. The decrease i n a i r qua nti ty i n the descensiona l l y ventil at­ e d a i rway ca uses, l i ke the throttl i ng effect in hori zontal a i rways, an increase in the ventil ating pressure of pa ral lel ai rways. The means to stabi l i ze the ai rflow in the descensionally venti l ated airway is to i nc rease the venti l ating pressu re acting on this ai rway. This can be done by i n ­ creasing t h e f a n pressure and b y throttling para l l e l a i rw ays. Si nce the th rottl i n g effect as wel l as the n atural draft have a tendency to reduce the a i rflow, fi res i n descensiona l l y venti l ated airways usua l l y don't reach the i n­ ten s i ty of fi res in ascensional venti lation and propagate with a consi dera bly sl ower velocity. 55 

M I N ES AND B U N K E RS 3.3.4.4 Examples of Airflow R evenals F or each of the th ree poss i bi l i ties discussed above, fi res in h ori zontal, ascension· a l l y and descensiona l l y ventilated ai rways, one exa mple of an airflow reve rsa l i n a mine fi re is given below. ( 1 ) Horizontal Airway At the D u kl a m i ne ( CS R ) two parallel ventilation spl its were connected by a d i agon al airway . The th rottl ing effect of a fi re (Ju ly 7, 1 96 1 ) i n one of the splits reversed the ai rfl ow i n the d i agonal ai rway and a l l owed fu mes to fl ow i nto the i nta ke airways of the other spl i t. As a res u l t 1 08 m i ners were k i l led . (2) Ascensionally Ventilated Airway I n the Roche-l a-Moliere mine on June 30, 1 928, a fi re in the raise 3-4 close to j unction 4 occu rred. The deve l oped natu ral draft at fi rst caused an a i rflow reversal in ai rways 8- 7- 5- 2, 7-6-3 an d 6- 5. Here , 48 m i ners were k i l l ed . Later, after th e f i re h ad moved down the ra ise away from 4 towards 3, the a i rflow norm a l i zed aga i n i n these ai rways but a short reversal i n ai rway 4· 1 occu rred. (3) Descentionally Ventilated Airway Woropajew ( 1 957) descri bes a fi re in a R ussian coal m i ne, work ing a di pping seam. The fire started i n the descensional l y venti l ated raise 2-3-4 at poi nt 3. The n atural draft reversed the a i rflow i n the ra ise and fi nal l y became so strong th at even i n the inta ke ai rway a reversal occurred . 3.3.4.5 Su itable Ventilation Plans Suitable ventilation plans are of great hel p for every type of venti l ation planning a nd, therefore, for the pred iction of vent i l ation distu rbances caused by m ine fires, too. They shou l d conta i n the main features of the venti l ation system , without confus i ng deta i l s and the sign i ficant venti l ation data. The former shou l d com prise the ai rways, location of fans, venti l ation doors, regul ators, seals and dams, ventila­ tion cu rtains and ducts, crossings, explosion barriers, produ ction wo rki ngs, trol ley and D iesel haul age roads. The latter sh ould i ncl ude d i rection and magn i tude of a i r currents, concentrations o f hazardous gases, measuring stations, elevations o f a i r­ ways, fan heads or pressu res and m i n e ventilating heads or pressures for the indiv id­ ual ai rways. I f it is fou nd too d ifficu l t to enter these data into the plans, they sho u l d be kept in u p to date reference files, wh ich ca n be used in conjunction with the venti l ation plans. That such plans are possi ble is proved by the fact th at they a re req u i red by law in several countries. F or th e planning of fire fighting measu res additional plans shou ld be provided, wh ich conta in informat ion on the water and compressed air pipeline systems, stored fi re fighting and su rvival equ i pment, telephone l i nes, etc. They shou ld be pl otted in the same sca le and P!lrspective as the venti l ation plans to fac i l i tate 56 

F I R E DYNAM I CS A N D SC ENAR IOS simu ltaneous use. ( 1 ) Plans in Use The ventilation plans actually used vary widely . The sim plest type is based on plan maps, which i n many cases suffice, especial l y when all m ine worki ngs are more or less situ ated i n one plane. For more complex l ayouts of openi ngs, as resu lts from m ining more than one coa l seam or ore deposits with a l a rger vertical extension, plan m aps becom e too confusing and perspective plans are preferred. The type of projection, if not pre­ scri bed by l aw, is usua l l y a compromi se between cla rity of the map and the ease with which hori zontal and vertica l d istances can be read from the map. Occasionally, but not too frequently, mode l s are used as a three d i m ensional image of a venti l ation system. The D utch coal m i nes favored these, bu ilt from wires of different colors to indicate the fu nction of airways. The work i nvolved in chang­ i ng the models and the prohi bitive costs of keeping records for certa i n time periods and providing copies requ i red that the models be used only in addition to other venti lation plans. Qu i te frequently simpl i fied vent i l ation pl ans are derived which show only the more i m portant a i rways of a vent i l ation system. Th i s i s especial ly the case when ventilation network calcu lation are performed and one tries to keep the n u m ber of a i rways going i nto the calcu lation as sma l l as possi bl e . Another reason is to gain a better understanding of the mutual interaction of the airways com prisi ng the sys­ tem. Several types of such si mpl ified plans are in use. When ventilation netwo rk ca lcu l ations were sti l l mainly perform ed m anual l y , with many ventilation engineers i t became popu lar t o represent venti l ation pl ans i n a n a bstract form resembling electric wi ri ng d iagrams. I n these plans the configu ra­ tion of the network is more conspicuous than in map plans or perspective d rawings. Airways in series or paral lel can be grou ped together and replaced by equivalent resi stors. In contro l led splitting the number and location of the necessary regul ators to enforce the wanted a i rfl ow distri bution is mo re eas i l y found. In natu ral spl itting the d i agonal ai rways, which cause difficulties in network calcu l ati ons, are more e as i l y detected. An eve n clearer pictu re of the network configu ration is obta i ned when in sche­ matic ventil ation plans the crossing of ai rways or the ove rl apping of l oops are as far as possible avoided. Si nce plans of this type a re most widely used in Pol and (where they are requi red by law for every m i ne ) they are usua l l y known by the i r Pol ish name as "canonical plans". They are, however, gai n ing i ncreas i ng popu larity in other cou ntries as the basis of emergency plans, too, si nce they are especially suited to detect possi ble i nstabi l i ties in venti lation systems. To judge the influence of several pressu re sou rces i n a venti l ation network on the stabi l ity of the a i rfl ow i n se lected ai rways qual itatively and, as fa r as possi ble, quantitati ve l y the Po l ish engi neer B udryk su ggested the use of a so-ca l l ed "closed 57 

M I N ES A N D B U N K E R S schematic plan" nowadays mo re frequently cal l ed a "Budry k plan". Characte ristic for th is plan is that the ai rway, whose stabi l i ty is to be judged, forms the bou ndary between those pa rts of the network wh ich are domi n ated by one of the pressu re sou rces. Alth ough quantitative predictions about the stabi l i ty of an ai rway are on l y possible for comparative ly simple networks, the Budry k plan a l l ows va l u able con­ cl usi ons as how to increase the stabil ity of th is airway . The name "closed p l a n " originiates from t h e fact that every a i r current is th ought t o be short circu ited through the atmosphere (which in fact it i s ) . Every ai rway o r group o f ai rways, whose stabi l ity m ust be i nvestigated and every new l ocation of a pressu re sou rce leads theo retica l l y to a d ifferent B udry k plan. I n the practice o f fire emergency plans vent i l ation engineers usu a l l y l i m it the ai rways to escape routes or stabil ity bou ndaries, beyond which no ai rflow reversal shou l d occur a n d t h e location of the pressure sources t o airways where natu ra l drafts can be created by fi res. Even i f n o actual Budry k p l ans h ave been plotted, fam i l i arity with the pri nci ples on wh ich they are based can be of considerable help in selecti ng the right measu res to stabi l i ze the ai rflow i n critical ai rways. Definition of Ventilating Heads a nd Pressures I t h as been mentioned that the data contai ned in vent i l ation plans shou l d i n ­ cl ude m i n e venti lating heads or pressu res. Th ese are d i fferences in t h e ene rgy con­ tent of the air between two poi nts of the netwo rk, or, when the two poi n ts are the begi nn ing and end of an a i rway, changes in the ene rgy content of th e air when moving th rough this ai rway. Si nce they indicate the stabil ity and economy of ex isti ng a i rflow as wel l as the di rection of potenti al airflow, thei r use h as become very popu lar with venti lation engi neers. According to the above def i n i tion of heads and pressu res, m i ne venti l ating heads h M v are energy d i ffe rences per u n i t weight and mine venti l ation pressu res P M V per u n i t vol ume of air. Being di fferences i n energy contents they can be dete rm ined from the ene rgy equation in the fol l owi ng form : 2 dV a v dp + d Z + -- + dh M V = 0 2g A comparison with the energy equation i n the form : 2 dV a v dp + dZ + -- + dh L - dh F = 0 hF = fan heads 2g 58 

F I R E DYNAM I CS A N D SC ENAR I OS PF = fa n pressu res 3.3.5 Quantitative Predictions of Ventilation D isturbances Ca used by Fires If the natural and th rott l i n g effects are known or can be calcu lated from other data, the i nfl uence of a fi re on the ai rflow distri bution in a venti l ation system can be determ i ned in a venti l ation network calcul ation. If sufficient data on the net­ work as the basis of the network calcu l ation and a compute r for the execution of the calcu l ation are available, th is is no great effort. Where th is is not the case and the calcul ations have to be done manual l y , ventilation engi neers i nvest igating t h e infl uence o f a fire must qu ite frequently be content with abridged and simpl ified network calcu l ations for the i mmediate v ici n i­ ty of the fire. Since it is usua l l y here that the fire has the greatest influence on the venti lation, these cal cul ations can be very usefu l . The accuracy expected from the resu lts determi nes the extent of permiss i ble simpl i ficati ons . To keep the m anual work i n tolerable l imits, vent i l ation engineers, moreover, l i m it the network calcula­ tions freq uently to the i nvestigation of especial ly critical states, l i ke the criteria for ai rfl ow standst i l l s and reversals. Network calcul ations for the vicin ity of the fire are usua l l y based on p ressu res, si nce the specific wei ght of the air in a l i m i ted area, except for the changes caused by the fi re itself, rema i ns fai rl y constant. As pointed out above, it m a kes no d i fference in principle if venti lation calcul ations are based on energies per unit weight (heads ) or energies per u ni t volu me ( pressu res) and ventil ation enginee rs traditi onal ly prefer the pressure approach. 3.4 Conclusions and R ecommendations Conclusion: The fire propagation process over the su rface of sol id polymeric m ateri als is not com pletely understood. The process is viewed primarily from re­ s u l ts of ex peri mental ev idence and is on l y part i a l l y supported by sem i-empirical theories. Recommendation: Theories fu l l y descri bi ng the fi re propagation process s h o u l d be devel oped and shou ld be based on extensive experimentation at various s ca l es accompan ied by accu mu l ation and correlation of a l arge volume of data. Conclusion: Fuel-rich fi res are fai rly we l l descri bed in the l i teratu re, but no schematic re presentation of oxygen-rich fires (except for ti m ber) h ave been pu b­ l ished in the past. Recommendation : Oxygen-rich fires shou l d be thorough ly i nves­ tigated, docu me nted and descri bed, and a schematic model for th is process sh ou ld be developed. Conclusion: M i ne fi res, seldom rema i n at steady state because of the physical confinement. Thus, a n originally oxyge n- rich fi re becomes a fuel- rich fire afte r the avai lable oxygen is exha usted . (This situation also can reverse if venti l ation is ap- 59 

M I N ES A N D B U N K E R S pl ied subsequently, as i n the incident descri bed i n section 3.2.3 .8.9 above ) . The n on-steady-state f i re propagation process is not fu l l y u nderstood. Recommenda­ tion : A m athematical model shou ld be deve loped to descri be and predict the non­ steady-state burning process. Conclusion: Most early mine fire ex periments were di rected toward testi ng the flammabi l ity (or fire resistance) of materials and equ ipment. Si nce such fires have usually no opportun i ty to reach an equ i l i bri u m state, the resu lts h ave l i ttle general validity. Recommendation: Design and conduct ex peri ments i n which steady-state con ditions are reached for meaningful data generation. Conclusion: F i re propagation vel ocities agai nst air currents (frequently encou n­ tered i n real m i ne fires) have not been i nvestigated even though the results of such studies could be d i rectly appl ied to understand i ng the phenomena. Recommenda­ tion : Experi me nts shou l d be conducted to stu dy in deta i l the development and velocities of smo ke (gas) l aye rs and fire spread u nde r a wide range of venti l aton conditions and duct i ncl i nation angles. Conclusion: Coal dust fi res, although mo re frequent and h av i ng l ower ign ition temperatu res than ti mber fires, have not been stu died, proba bly because of the i r l ow propagation rate. These fires, h owever, can act as ign i tion sources to da ngerous secondary fuels, especially the synthetic polymers that are gai n ing increased uti l i za­ tion in m ines. Even worse is the potential of coal dust fi res to become coal dust explosions. Recommendation: Systematic fi re ex periments shou ld be conducted o n coal dusts o f various origin, composition, a n d size d istri bution corresponding t o rea l m i ne envi ron ments. Conclusion: There has been su rprisingly l i ttle work done on the com position of the combustion products of fuel -rich (open, undergrou nd) coal fi res, whereas the oxygen-rich coal f i re has been studied extensively. Specifical l y , information on the process of ignition, composition of com bu stion products, and propagation vel ocity of fuel-rich coal fi res is l acking. Recommendation : The im portance of medium- and l a rge-scale coal mine fire experi ments has been recogn i zed and such experi ments a re n ow being conducted. It is obvious that these experi ments are very expensive and difficu l t to perform especially with the added requirement of protecting the envi ­ ron ment. It i s o f utmost i mportance t o continue a n d expand these experimental programs to generate the l acking information. However, because of the high cost, ful l advantage shoul d be taken of these ex peri ments to extract all possi ble fi re­ related i n formation. They have to be wel l designed , wel l planned, and fu l l y i nstru­ mented not on l y for the pri mary information sought but also for al l possi bl e side effects and i n formation. Conclusion: Theoretical approxi mation of m i ne fire dynam ics h as been based on one- or two-dimensional representations in the past. These appear to be i nadequate for the comp lete understandi ng, descri ption, and pred iction of airfl ow phenome n a i n mine fi res. Recommendation : Deve lop a th ree-di mensional model i n order to c l osely approxi mate buoyancy phenomena and associated flow reversals that devel - 60 

F I R E DYNAM I CS A N D SC ENAR IOS op i n m i ne fi res. This may requ i re extensive large-scale ex peri mentation and a complex mathematica l approach . Of particu l ar interest is the study of the i nter­ acti on of natura l draft, i n du ced venti l ation and throttl i ng effect of the fi re i tself. Conclusion: Venti l ation plans empl oyed to date ge nera l l y a re suitable for most con ti ngencies. They are overly s i m p l i fied, however, for reasons of conve nience and economy. Better plans cou l d s u bstanti a l l y i m prove the efficiency of fire su pression. Recommendation: After sufficient i n formation and u nderstanding of the essenti al elements of fire dynamics become avai l a ble, a highly soph isti cated, general com­ puter ventil ation mode l shou ld be e mployed. The basic m ode l cou ld be pro­ grammed to individual m i nes and u pdated frequently as mine deve l opment pro­ gresses. Such a model it wou l d provide rapid, opti m i zed sol utions and ava i l able options in case of fire incidents. 3.5 References J, I, Dobis et al., Report of Colli Mine Fi�W, Mars No. 2 Mine, Oinchfield Coal Co., Wi/Hnburg, W. V., Oct 16, 1965, u.s. BuMinas, Washington , DC 1 965. P. V. Fanok & M . W. McMan us, Report of Coal Mine FitW, Joanne MlneE•tem Alloc. Cos/ Corp., Rachel, W. V., May 15, 1970 u.s. B uMines, Wash ington , DC, 1 970. J . F reeman, Report of Coiii·Mine FitW, Geneva Mine U.S. Sttltll Corp., Horse Canyon, Utah, Feb. 10, 1969. T. W. Gay, BuMines, Report of Colli Mine FitW, No. 1 Mine Kermit Cos/ Co., Kermit, W.V., Aug, 7, 1 970, u.s. Bu Mi nas, Washi ngto n , DC, 1 970. A. E. G reuer, The Influence of Mine Fi�Ws on the Ventilation of Underground Mines, prepared u nder u .s. BuMinas Contract SO 1 22095, MPIS No, 225834, Michigan Tech nological Uni­ versitY, Houghton , Michigan , 1 973. Hottel and Egbert, "Radiant Heat Transmission from Water Vapor." Trans, American Chem ­ istry Engineers 39, 53 1 -568, 1 942, Jakob, M., " Heet Transfer." John Wiley and Sons, I nc., New York - London 1 959, H. A. Jarvis, Report of Coal-Mine FitW, No. 10 Mine, Finley Cos/ Cos/ Co., Hyden, Ky., May 1 0, 1967, U .S. Bu Minas, Washington, DC, 1 967. J , Matekovic, Report of Coal-Mine Fi�W, Somerset Mine, U. S. Sttltll Corp., Somerset, Col., No v. 1 5- 1 6, 1971, U.S. Bu Minas, Wash i ngton, DC 1 97 1 . A . O'Rourke, Report of Coiii·Mine FitW, Nemacolin Mine, Buckeye Cos/ Co., Nemacoling, Ps., March 26, 1971, U.S. BuMinas, Washi ngton, DC, 1 971 . J. P, Ph illips, Report of Colli Mine FitW, Compess No. 2 Mine, Oinch field Cos/ Co., Dels, W. V., July 7, 1969, U.S. B u Mi nas, Washi n gto n , DC, 1 969. H. Sorrel and H. J. Lyon, Report of Coal-Mine Fire, Island Creek Coal Co., Madisonville, K Y., Apr. 1 7, 1973, U.S. BuMines , Wash ington , DC., 1 973. Woropajew, A. F., Discussion of a Self-Produced Airflow Reversal When a Mine Fire Starts. Ugol, 32, (3) pp. 27- 30, 1 957. 61