.*'\ JC! 8865 Bureau of Mines Information Circular/1981 ^^&^^^-90=> ^^a-ofW Underground Metal and Nonmetal Mine Fire Protection Proceedings: Bureau of Mines Technology Transfer Seminars, Denver, Colo., Nov. 3, 1981, and St. Louis, Mo., Nov. 6, 1981 By Staff, Bureau of Mines UNITED STATES DEPARTMENT OF THE INTERIOR ^1 ^UJ:^i^ y^^2^. /Oc<-'U^>c<^ ^ ^^^c.!^ , Information Circular 8865 Underground Metal and Nonmetal Mine Fire Protection Proceedings: Bureau of Mines Technology Transfer Seminars, Denver, Colo., Nov. 3, 1981, and St. Louis, Mo., Nov. 6, 1981 By Staff, Bureau of Mines UNITED STATES DEPARTMENT OF THE INTERIOR James G. Watt, Secretary BUREAU OF MINES Robert C. Horton, Director (^^ <4^ PREFACE This Information Circular summarizes recent Bureau of Mines results in the area of fire and explosion prevention. The papers are only a sample of the Bureau's total effort to improve mine health and safety, but they repre- sent the major research effort in the area of underground metal and nonmetal mine fire prevention. Those desiring more information on the Bureau's Min- erals Health and Safety Technology Program in general, or information on specific research, should feel free to contact the Bureau of Mines, Division of Minerals Health and Safety Technology, 2401 E Street, NW, Washington, D.C. 20241, or the appropriate author listed in the following proceedings. ill CONTENTS Page Preface 1 Abstract 1 Introduction, by John F. Papp 2 Shaft fire protection, by Guy A. Johnson 4 Spontaneous combustion fire warning systems for noncoal mines, by Guy A. Johnson 17 Product-of-corabustion fire detection in mines, by Charles D. Litton 28 Improved stench fire warning system, by Willian H. Pomroy 49 Computer-aided ventilation modeling, by John C, Edwards 78 An experimental investigation of the fire hazards associated with timber sets in mines, by Archibald Tewarson and J. S, Newman 86 Underground fueling area fire protection systems, by Guy A. Johnson 104 Fire doors for noncoal mines, by Kenneth L. Bickel 115 Automatic fire protection systems for mobile underground mining equipment, by Guy A, Johnson 133 UNDERGROUND METAL AND NONMETAL MINE FIRE PROTECTION Proceedings: Bureau of Mines Technology Transfer Seminars, Denver, Colo., Nov. 3, 1981, and St. Louis, Mo., Nov. 6, 1981 by Staff, Bureau of Mines ABSTRACT These proceedings consist of an overview of the underground metal and nonmetal mine fire protection research currently being conducted by the Bureau of Mines. The following papers address the areas of prompt, reliable fire detection, fire planning and warning, fire suppression, and personnel pro- tection. Selected topics are included here that cover fire protection sys- tems for shafts, underground fueling areas, and mobile underground mining equipment; spontaneous combustion, improved stench, and product-of-combustion fire warning systems; computer-aided ventilation modeling; fire doors; and an investigation of fire hazards in mine timbers. The projects described provide a current documentation of problems being addressed. INTRODUCTION by John F, Papp^ This report contains nine papers presented at the Bureau of Mines Tech- nology Transfer Seminars on Metal and Nonmetal Mine Fire Protection held at Denver, Colo., and St. Louis, Mo., in November 1981. The seminars are a part of the Bureau's program to apprise the raining industry of the results of Bureau of Mines research. It is hoped that the prompt dissemination of this information will help transfer research results to make a safer mining envi- ronment and more productive mining operation. Fire prevention is a goal of our program and represents a major part of our effort. We recognize that when a fire does occur, prompt detection, alarm, and suppression are critical to protecting the safety of every miner. For a fire to occur, three things are required: fuel, oxygen, and heat. If any one of these is missing, a fire will not occur. If any one can be removed from a fire, it will be extinguished. All of the elements necessary for a fire are found in mining operations. Some common fuels are methane gas, tim- ber, hydraulic fluids, and sometimes the ore itself. Some common ignition sources are sparks, spontaneous heating, hot surfaces, and faulty electrical equipment. Oxygen is supplied by the air. A recent survey of metal and nonmetal mine fires'^ conducted for the Bureau of Mines has shown that most fires are electrical in origin, that elec- trical fires cause the greatest number of injuries, and that fires resulting from welding operations and spontaneous combustion cause the greatest number of fatalities. Two contributing causes to most fires were found to be poor maintenance and the ready availability of fuel. Some of these fire problems are addressed by the papers in this Information Circular. Others will be the subject of future Bureau research work. The institution of better housekeep- ing practices and more frequent maintenance could greatly reduce the fire threat in underground mines. Manager, Fire and Explosion Prevention, Bureau of Mines, Washington, D.C, ^Baker, R.M. , J. Nagy, L. B. McDonald, and J. Wishmyer. An Annotated Bibli- ography of Metal and Nonmetal Mine Fire Reports — Volumes I-III. BuMines Open File Rept. 68(l-3)-81, Dec. 5, 1980, 721 pp.; available for reference at Bureau of Mines facilities in Tuscaloosa, Ala., Denver, Colo., Avondale, Md. , Twin Cities, Minn., Rolla, Mo,, Boulder City and Reno, Nev. , Pitts- burgh, Pa., Salt Lake City, Utah, and Spokane Wash.; National Mine Health and Safety Academy, Beckley, W. Va. ; and National Library of Natural Resources, U.S. Dept. of the Interior, Washington, D.C. Copies available from National Technical Information Service, Springfield, Va. , PB 81-223 711 (Complete 3-volume set), PB 81-223 729 (V. I), PB 81-223 739 (V. II), PB 81-223 745 (appendix). Developing better ways of detecting, warning of, and suppressing fires that cannot be, or are not prevented, is an objective of the Fire and Explo- sion Prevention Program. Work is underway to eliminate ignition sources, to achieve early detection of fires or spontaneous combustion conditions, to remove or replace combustible materials wherever possible, to improve warning systems, and to suppress fires. The papers presented at this seminar address the areas of prompt, reliable fire detection; fire planning and warning; fire suppression; and personnel protection. Prompt, reliable fire detection is addressed in the papers on spontaneous fire detection, mobile equipment fire protection, shaft fire protection, fueling area fire protection, product of combustion detection, and timber set fire hazards. Fire planning and warning are addressed in the papers on stench warning system, ventilation modeling, vehicle fire protection, timber set fire hazards, shaft fire protection, fuel- ing area fire protection, and spontaneous combustion warning. Fire suppres- sion and personnel protection are addressed in the papers on fire doors, vehicle fire protection, fueling area fire protection, and shaft fire protec- tion. Technological progress has been made in these areas through Bureau in-house and contract research projects to advance fire safety in metal and nonmetal mining operations. Reference to trade names and specific makes or models of equipment in this report is made for identification only and does not imply endorsement by the Bureau of Mines. SHAFT FIRE PROTECTION by Guy A. Johnson^ INTRODUCTION As metal and nonmetal mines become deeper, more complex, and more mechan- ized, the fire danger increases because of the increased use of combustibles and the increasing restrictions on miner egress. This fact, unfortunately, was illustrated in May 1972 by the Sunshine Mine fire at Wallace, Idaho, in which 91 miners died. To help solve this growing problem, the Bureau of Mines initiated a program to develop mine shaft fire and smoke protection technology. The methodology used for this project followed that used previously in the Bureau's successful effort to develop automatic fire protection systems for large, mobile mining equipment; that is, the ruggedizing of existing tech- nology to solve a specific mining safety problem. The project's basic meth- odology was to evaluate mine fire and smoke problems, then develop and demonstrate a reliable mine shaft fire and smoke control prototype system to protect against such hazards. This hardware was to be flexible in design so it would be adaptable for use in the majority of metal and nonmetal mine shafts and adjoining shaft stations. SYSTEM DESIGN Electrical shorts, welding, and torch cutting are the chief causes of underground noncoal mine fires. Because these are high energy fires, early detection so the fire will be easier to extinguish is an important design requirement. Also, because more underground miners die from smoke than from burns, smoke control is another important requirement for underground fire protection. As a result of the problem analysis, a design concept for a system was developed. The system used thermal, CO, and ionlzed-particle (smoke) detec- tors because no one fire sensor has proven itself reliable enough in the mine environment. Remotely controlled smoke doors and sprinklers were used to pro- tect both the shaft and shaft station areas of a mine. The system's surface control unit receives underground fire warning signals via multiplexed elec- tronics sent through two separate twisted-pair routings. The system's under- ground control units can be activated from the surface control unit to warn the miners in the shaft station area of fire danger. One underground control unit is needed in each shaft level protected by the system. Surface personnel also would activate a stench system to warn miners of danger. The sprinklers and smoke control doors can be opened or closed by either local controls or rerao'te actuation from the surface control unit. Automatic sprinkler actuation ^Supervisory mining engineer. Twin Cities Research Center, Bureau of Mines, Minneapolis, Minn. is not a current feature of the system, but such a design alternative could easily be added if a user so desired. IN-MINE FIRE TEST During the winter of 1975, a shaft-to-shaft station mockup (fig. 1) was built for component testing. Laboratory testing was highlighted on March 20, 1975, when the system's hardware was demonstrated to Bureau, MSHA, and indus- try representatives by an actual fire test in the raockup. After these preliminary evaluations, the prototype hardware was installed at the 3000 level of the Silver Summit shaft near Wallace, Idaho (fig. 2). The system was successfully demonstrated to Bureau, MSHA, and mine personnel by actual in-mine fire testing in April 1975 (figs. 3-8). The fire was lit in a 4- by 4- by 2-foot steel box, burned for about 5 minutes, then V7as sensed and extinguished, by remote control, from the surface. For this test, the CO sensors were set to detect 200 ppm and the smoke sensors 2 pet per foot obscuration. Followup work was conducted in the laboratory to modify and reinstall the mine shaft fire and smoke protection prototype system at a second location for long-term testing. This took place in summer 1975 at Union Carbide's Pine Creek mine in Bishop, Calif, (fig. 9). The second-generation hardware was subjected to a second actual in-mine fire test in summer 1976 (figs. 10-12) after 12 months of use in the mine. The long-term test demonstrated that the hardware was rugged enough for reliable operation in an underground environment. Concurrent work by the Bureau to in-mine test the reliability of low-cost, designed underground fire sensor packages was initiated and is ongoing at the Lakeshore mine in Casa Grande, Ariz. Figure 13 shows a sensor package cur- rently undergoing long-term evaluation. Both the Pine Creek and Lakeshore testing showed that the charged-particle smoke sensor produced in South Africa (The Becon Mk II) was the most reliable. The CO monitors and optical flame detectors need to be made more rugged in order to be applicable underground. Simulated drift area -Smoke density meter Smol-^ FIRE TEST ZONE 11071 A -LEVEL 100 300 500 700 RAISE : •caved 900 •caved 1300 1100 1500 L ■caved caved -caved -caved* 11 ^f^c^O TO SURFACE 250 FIRE TEST ZONE TO SURFACE 9200 9000 8800 8500 _8300 8100 TO SURFACE AND MILL FIGURE 9. - Schematic of the second-generation system. 14 I 15 1 10 VAC power Phone line 110 VAC power FIGURE 13. - Mine fire sensor package tested at Hecla's Lakeshore Mine, Cosa Grande, Ariz. 16 ADVANTAGES AND COST OF THE SHAFT SYSTEM The mine shaft fire and smoke protection technology gives a mine quick response capabilities in underground fire situations. Also, having remotely controlled sprinklers in the shaft and shaft station areas is much more effec- tive in extinguishing fires than the presently used shaft-collar water rings. The cost of the shaft fire system is estimated to be from $50,000 to $100,000, depending upon the sophistication desired in the system and the extent of the area to be protected. This is a substantial investment, but compared with the millions of dollars a shaft costs, and the increased protection provided to the miner, the system is becoming an important contribution to improved under- ground fire safety. SUMMARY The development of improved shaft fire protection hardware for noncoal mines involved the delineation of the mine fire problem and the design, laboratory testing, and in-mine demonstration of technology that better pro- tects miners than the shaft-collar water rings now commonly used. The shaft fire system used thermal, CO, and smoke detectors, remotely controlled smoke doors at the shaft stations, and remotely controlled sprinklers. It was suc- cessfully demonstrated by actual fire tests in the Silver Summit shaft near Wallace, Idaho, and in Union Carbide's Pine Creek mine in Bishop, Calif. Long-term, in-mine testing showed the system to be a reliable improvement in mining health and safety. More information about this technology's develop- ment and in-mine testing is available from the Bureau's 9-min Technology Transfer film "Mine Shaft Fire and Smoke Protection System."^ Three companies now commercially make systems based on the Bureau's work: FMC Corp., Santa Clara, Calif, (the system's original developer under Bureau contract); Allison Controls Co., Fairfield, N.J. (which designed a system for the Diamond Crystal salt mine near New Iberia, La.); and Aquatrol Corp., St. Paul, Minn, (which is Installing a fire system in FMC's Green River trona mine. Green River, Wyo.). It should be noted that the National Fire Protection Association (writer of the Life Safety Code and the National Electric Code) is now developing, within its new Mining Facilities technical committee, consensus standards for improved mine fire protection. The Mine Safety and Health Administration's monthly magazine Mine Safety and Health published a summary article on "Deep Mine Fires" in the March-April 1981 issue (pp. 10-17). ^Available by writing to Motion Pictures, Bureau of Mines, 4800 Forbes Ave, Pittsburgh, Pa. 15213. 17 SPONTANEOUS COMBUSTION FIRE WARNING SYSTEMS FOR NONCOAL MINES by Guy A, Johnson^ INTRODUCTION Spontaneous combustion fire protection is a new component of the Bureau of Mines' underground metal mine fire protection program, A number of major fires, which have resulted in numerous deaths, have been blamed on spontaneous combustion. This problem will get worse because as mines get deeper, hotter, and more complex, ventilation to cool them is harder to control, and miners will have a much longer evacuation time in case of a fire. SYSTEMS DESIGN To help solve this problem, the Bureau of Mines first analyzed the key components of spontaneous combustion (fig. 1). The analysis was directed towards determining precursors and the relationship between heating of rocks and occurrence of fires in wood sets, trash, etc. To understand this problem, the Bureau studied the relationship between the spontaneous heating and the different minerals in the mines, such as sulfides. This study was conducted to determine the general nature of spontaneous combustion, because it is difficult, very expensive, and really unnecessary to understand the detailed chemistry of the spontaneous combustion process. The need is to determine only the general behavior of the combustion process, since each mine will have its own conditions in terms of what gases have to be sensed, and in what quantities, etc. The Bureau has an ongoing program looking at spontaneous combustion of coal, but the problem in noncoal mines is quite different. The Bureau's objective was to design a versatile system that could handle the general case. The raining companies then could take this as a first step and adapt the system for their spontaneous combustion protection needs. Laboratory testing showed that the gases which must be sensed are CO, CO2, and O2, as well as SO2 (fig. 2), because sulfur is a key ingredient in the spontaneous combustion reaction. The criteria for designing the system, as shown below, were similar to the criteria used for the shaft fire protec- tion system. It is best to extinguish the fire when it is small — that is, when it is easiest to put out. It is essential to be able to give a timely alarm, to get the personnel out safely, and to accomplish this with rugged, cost-effective, system components. ^Supervisory mining engineer. Twin Cities Research Center, Bureau of Mines, Minneapolis, Minn. FIGURE 1. - Spontaneous combustion situation. Surface control and warning unit Multiplex signaling.^ |- ^ rf r Underground fire warning unit (CO, CO2 SO2 O2, iX^ and temperature) FIGURE 2. - Spontaneous combustion fire warning system for metal and nonmetal mines- schematic. 19 Specific criteria for the spontaneous combustion fire warning system for metal and nonmetal mines are as follows: • Components of the system: 1. CO sensors 2. CO 2 sensors 3. SO2 sensors 4. Temperature sensors 5. O2 sensors. 6. Multiplex signaling • Fire warning given as a result of long-term trend analysis of a heating event. • In-mining test of prototype hardware by Magma Copper Co., Superior, Ariz. • Total cost (installed): $20,000 to $40,000. There were two typical tradeoffs that were considered in the system's basic design. The first involved a manual spontaneous combustion fire protec- tion approach. In many mines, people are sent to different parts of hot mines to test the air and to record the temperature. If there is a problem, a man- ual warning is given. The Bureau felt that this option exposed the miners to danger and also took too long to alarm the mine. A second option considered was a tube bundle system. Because of the reliability of the air sampling, this option would be acceptable in single-level coal mines, but not in the multilevel metal mines. To check this assumption, the Bureau engineers visited a coal mine that has used a tube-bundle system for a number of years. This visit confirmed the problems with reliability. Because of unknown breaks in the sampling tubes, it could not be determined exactly where the sample air was coming from. It could be coming from 2,000 feet down the drift, or it could be coming from a location 2 miles away. Pressure drop instrumentation can solve part of this problem, but it is not rugged enough to be used in a mine environment. As a result, the basic spontaneous combustion fire and warning system was designed with an an underground sensing unit that senses CO, CO2, SO2, O2, and temperature. This information is then multiplexed to a surface control unit. The surface unit records and observes long-term trends and warns per- sonnel of abnormal conditions that could lead to a fire or a heating event. The hardware, which costs from $2,000 to $4,000 per area protected, is avail- able from several underground fire protection groups such as ESD of Santa Clara, Calif, (formerly FMC Corp.), The Ansul Co., Marinette, Wis., and Kidde, Belleville, N.J. 20 IN-MINE TEST OF THE SYSTEM Once the system was bench tested, first-generation hardware was installed at an underground location in Magma's Superior mine at Superior, Ariz., in 1979 (fig. 3), After 12 months of in-mine use (fig. 4), Bureau engineers jointly with MSHA officials conducted an underground test in 1980. An ore-wood sample was heated in an oven over a period of days (figs. 5-7). During the third day, the wood went into a high energy flame state. However, starting with the second day the system sensed abnormal amounts of CO and CO2, showing that a spontaneous combustion event was occurring. More recently, the prototype system sensed a heating event at the mine when there were only a few workers underground to discover the event and sound an alarm. The system showed surface personnel that there was a problem developing underground. This occurrence gave us a lot of confidence that this is indeed the proper approach to spontaneous combustion fire protection. The Bureau is now working with Magma to add second-generation hardware to the basic system. It must be remembered, when studying spontaneous combustion fire pro- tection, that the applicable technology is very sophisticated and that the requirements are very mine specific. Typically, mines are not all alike, and also ambient conditions in one area of a mine may be totally different than in another area in the same mine. These conditions sometimes are not taken into account. For these reasons, the developed system uses technology that is flexible and components that are as low in cost as possible. 21 FIGURE 3. - Magma Copper Co. Mine, Superior, Ariz. 22 23 24 25 26 n 27 CURRENT WORK The Bureau is currently conducting long-term, in-mine tests with the system installed in several different locations (fig. 8). The system can be installed behind bulkheads to protect abandoned areas, or at known hot spots. A silver mine in the Coeur d'Alene District, because of the favorable price of silver, is planning expansion that may develop some such hot areas in the mine. The mine operators want to work with us to in-mine test an alternate design of the spontaneous combustion fire protection system. I hope to see the results within the next 1 to 2 years on this and other variants of the spontaneous combustion protection methods. More information is available in IC 8775-^ and a Bureau contract report.^ ^Ninteman, D. J. Spontaneous Oxidation and Combustion of Sulfide Ores in Underground Mines. A Literature Survey. BuMines IC 8775, 1978, 36 pp. ^Stevens, R. B. Improved Spontaneous Combustion Protection for Underground Metal Mines. BuMines Open File Rept. 79-80, November 1979, 262 pp.; avail- able for reference at Bureau of Mines facilities in Tuscaloosa, Ala. , Den- ver, Colo., Avondale, Md. , Twin Cities, Minn., Rolla, Mo., Boulder City and Reno, Nev. , Albany, Oreg. , Pittsburgh, Pa., Salt Lake City, Utah, and Spokane, Wash.; National Mine Health and Safety Academy, Beckley, W. Va. ; and National Library of Natural Resources, U.S. Dept. of the Interior, Washington, D.C. Available from National Technical Information Service, Springfield, Va. , PB 80-210 461, contract H0282002, FMC Corp. 28 PRODUCT-OF-COMBUSTION FIRE DETECTION IN MINES by Charles D. Litton ^ ABSTRACT In order for fire detection systems to provide a specified level of pro- tection for an underground mine, it is necessary to develop guidelines for sensor distribution that are directly related to the fire hazard. This paper discusses the relationships that exist between the developing fire, the prod- ucts of combustion that are liberated, and the sensitivity levels of various detectors; and how these factors can be utilized for determining the optimum distribution of candidate sensors. Data are presented for the production of CO and CO^ and submicrometer particles (SMP's),^ assuming wood to be the primary combustible. The result- ing analysis indicates that both CO and CO2 detection are limited by their characteristic production rates and that SMP detection may be a more funda- mentally sound approach to fire detection. Three promising SMP fire detectors designed for underground usage are briefly discussed. INTRODUCTION A developing fire poses a significant hazard, and it is the purpose of a fire detection system to provide a warning of this developing hazard in a time frame sufficient to successfully initiate evacuation and control mea- sures. To achieve this purpose, certain information relative to the develop- ing fire hazard is necessary, and this information must subsequently be used for the design of fire detection systems. This information forms the vital link between the fire hazard and the detection system. For a fire alarm to occur, two events must take place. First, the fire must produce some detectable quantity (heat, light, or combustion products). Secondly, there must be a means (that is, the sensor) for detecting the pres- ence of some fire-produced quantity. The time necessary for a sensor to detect the presence of a developing fire and issue an alarm is a function of many parameters: 1. Type, quantity, and distribution of combustible, 2. Fire size, 3. Detector sensitivity, ^Supervisory physical scientist, Pittsburgh Research Center, Bureau of Mines, Pittsburgh, Pa. ^Particles smaller than 1 \m in diameter (SMP's). 29 4. Ventilation velocity, 5. Entry dimensions, and 6. Spacing between sensors. In the analysis that follows, the interdependence between these various parameters will be defined in a manner that allows for the determination of the optimum sensor configuration. This analysis places emphasis upon the developing fire hazard and defines the criterion for a detection system in terms of the time available for detection and subsequent action. Any detec- tion system (thermal, optical, product-of-combustion, or otherwise) must satisfy the criterion of detection and alarm within a specified period of time that is directly related to the developing fire hazard. Within the framework of this approach, it is possible to evaluate any candidate fire detection system. However, product-of-combustion sensing sys- tems offer considerable potential for earlier and more reliable detection than do thermal or optical sensing systems. Further, thermal and optical systems require a high density of sensors for large area coverage — such as long, hori- zontal passageways — and are generally not recommended for such applications. Such sensors are more suitable for protection of localized, high-risk areas or for protection of underground equipment. For these reasons, the following analysis is restricted to product-of- combustion sensors and sensing systems, and the area to be protected is limited to long, horizontal passageways. FIRE GROWTH The growth and development of a fire can be characterized by three dis- tinct stages, shown graphically in figure 1. In the first stage, the tempera- ture of the combustible increases, producing sufficient volatiles in the gas phase to support flaming combustion. When this condition is reached, and the surface temperature of the combustible is high enough, flaming ignition occurs and the fire makes the transition to a second stage of growth. During this second stage, owing to the presence of a flame, the fire grows more rapidly, the rate of growth depending upon the type and amount of combustible and its distribution within the entry. If sufficient quantities of combustibles are available in the vicinity of the flaming fire so that it continues to increase in size, it will eventually make the transition to a third stage. During this third stage, the fire begins to propagate down an entry, consuming combustibles at an ever-increasing rate. If the transition to this third stage is allowed to occur, then the potential for escape and control is reduced significantly, and eventually the fire will propagate throughout large portions of the mine with devastating consequences. 30 Mf=Mi flaming ignition Mf = Merit] Pre flaming region (Stage I) Fire growth region (Stage II ) (Stage HI) -•—Flame spread- region TIME— FIGURE 1. - The three stages of fire development. These three stages of development represent an ever-increasing degree of hazard to underground personnel, but it is the attainment of the third stage that poses the most severe threat. For this reason, it must be the purpose of any fire detection system to provide a warning of the developing fire in a time frame sufficient to initiate evacuation and control measures prior to the transition to this third stage. Clearly, then, detection and alarm must occur during either stage 1 or stage 2 in order for a detection system to achieve its intended function. Further, the time of development of the first stage is highly dependent upon the initial source of heat. Because of this, the duration of the first stage is usually unknown. It could develop over a period of a few days to a few weeks if, for instance, it were due to spontaneous heating; or it could develop in a matter of seconds if it were due to an electrical arc or some other equally intense ignition source. Because of these uncertainties, it is often impossible to design detection systems that can reliably provide an alarm prior to flaming ignition. Only in certain, well-defined situations. 31 such as areas prone to spontaneous heating, should such systems be expected to perforin. The onset of flaming ignition, which signals the transition to the second stage, provides a reference point in time, which, generally can be used much more effectively for the design of detection systems. Following flaming igni- tion, the fire will begin to increase in size at a rate that depends upon the following factors: 1. Type and amount of combustible, 2. Distribution of the combustible, and 3. Ventilation velocity. In many cases, in fact, in most cases, the rate of growth during this second stage can be measured for a variety of possible scenarios. These growth rates, once quantified, can then be used to specify the time available for detection and alarm, which, in turn, can be used to define sensor spac- ings, required sensitivity levels of sensors, and required time responses of detection systems. In developing this hazard-oriented approach to fire detection systems, two additional parameters must, of necessity, be included. The first param- eter is the fire size at the instant of flaming ignition; and the second, the critical fire size at the transition point to the third stage of fire develop- ment. These two parameters form the two boundaries, in time, of the second stage, and together with the defined fire growth rates provide the information necessary for designing and Implementing fire detection systems with a defined level of protection, which is directly related to the developing fire hazard. PRODUCT-OF-COMBUSTION FUNDAMENTALS In order to expand upon this type of approach to fire detection, it is necessary to develop relationships between the fire hazard and the detection system. Assume that, for a given fire scenario, the fire growth rate can be expressed as a function of the type and amount of combustible, f, the ventila- tion velocity, vf, and the time, t, measured from the instant of flaming igni- tion; that is. mf = Af (f, Vf, t) (1) where m^ represents the fire growth rate in terms of the mass of combustible consumed per unit time. For a given combustible, it is possible to define some average charac- teristic production parameter, K^, for combustion product, X, as the amount of product produced per mass of combustible consumed. The production rate, X, of product, X, can then be defined as the product of K^ and the mass loss rate, 32 Correspondingly, the resultant increase in the bulk average concentra- I tion, [X][3, is the production rate, X, divided by the average volumetric ven- tilating flow within an entry. But this flow is the product of the average ventilation velocity, v^, and the entry cross-sectional area. A, Then the resultant concentration increase for product, X, becomes, [y^^^ ^h^^A±JLL^±JLjl (2) In order for the detection of some product to be realized, and an alarm given, a sensor must be available to provide this function. The sensor is assumed to have two characteristic parameters: 1. Sensitivity — defined as the minimum reliable concentration that can be measured and denoted by S.^ for product X. 2. Time response — defined as the time necessary for a sensor (or sens- ing system) to issue an alarm once an alarm concentration has been reached, and denoted by Tr. For illustrative purposes, an alarm concentration will be defined as that concentration which is 10 times the detector sensitivity. Consequently, at alarm, the bulk average concentration, [X]|3, must satisfy the condition, [X]b > 10 S^ (3) Substitution of equation 3 into equation 2 yields 10 s < ^x • '^f (^> ^f» ^^ (4) Equation 4 is an explicit relationship between the fire and the hazard. It represents a "missing link" that is crucial to the design of detection systems. Before discussing the manner in which equation 4 can be used in the design of detection systems, it is instructive to determine, from equation 4, the minimum detectable fire size for a given combustible using detectors with fixed sensitivities. Assuming the combustible to be wood, with a heat of combustion of approx- imately 4300 cal/g, the minimum detectable fire size, Q^^, expressed in kilo- watts, using product-of-combustion detector, S^ , becomes Qx(KW) = 180 q^ • ^ (5) 33 where q^ =.v^ • A = volumetric flow in cubic centimeter^ per second within an entry and Q^ = 18.0 m^ with m^ in grams per second and 0^ i^i kilowatts. For product-of-combustion fire sensing, three products have been identi- fied as the most common: submicrometer particles (SMP's), CO, and CO2. For wood, their respective average production parameters have been measured^ to be KsMP = 2.0 X 1013 particles S'^P gram molecules Iqo = 2.5 X 1021 Kco- = 1.5 X 1022 gram molecules 2 gram In order, now, to determine the minimum detectable fire size, it is nec- essary to specify the sensitivity levels of the candidate sensors. For illus- trative purposes, the following values will be assumed: c ,^4 particles (~0.50 mg) 3SMP - i> X 10-* r -— cm-' m^ Sco = 2.69 X 1013 "^"^^^^^^^ (1.0 ppm) cm- Sco. = 6.73 X 1014 '""^^^^^^^ (25.0 ppm) 2 cm-' For SMP's, this sensitivity is about average for this class of sensor. For CO and CO2, the sensitivity levels represent about the optimum values available for these types of fire sensors. Substituting the respective values for K^ and S^ into equation 5, the minimum detectable fire size, Q^, for each sensor can now be determined. The results are illustrated graphically in figure 2 where 0^ is plotted versus the volumetric ventilation flow, q^. This type of information is very useful in that it provides a convenient means of assessing the relative effectiveness of various sensors. Similar information can be obtained for other combustible materials and other combus- tion products, provided that data is available for the respective production parameters, K,,. ^Lee, C. K., R. F. Chaiken, J. M. Singer, and M. E. Harris. Behavior of Wood Fires in Model Tunnels Under Forced Ventilation Flow. Tests With Untreated Wood. BuMines RI 8450, 1980, 58 pp. 34 1,000 FLOW RATE, lO^cfm 50 75 100 125 150 FIGURE 2. 2 3 4 5 6 7 qv, lO'^cmVsec The minimum detectable size of wood fire for three candidate sensors as a function of the volumetric ventilation flov/. 1 il 35 For less sensitive detectors (that is, greater S^ values) the minimum detectable fire size will increase and vice versa. Also, from figure 2, the minimum detectable fire size increases with increasing ventilation flow, owing to the additional dilution of incoming air with the combustion products. Based upon the comparison of figure 2, it is immediately evident that SMP detectors represent a significant improvement over either CO or CO^ detectors. This result is a reflection of the differences in the processes responsible for the production of SMP's as opposed to those responsible for the production of product gases. This important aspect of the overall detection problem is discussed in more detail in a subsequent section. DETECTOR SPACINGS In order to define more explicitly a product-of-combustion detection sys- tem, it is necessary to determine the optimum spacings for the sensors rela- tive to the fire hazard. Products of combustion, upon leaving a fire, are convected downstream at a velocity approximately equal to the average ventila- tion velocity. Assuming that the maximum distance that a product must tra- verse is one detector spacing, then the spacing must equal the product of the ventilation velocity and the time available for transport, Tf Therefore, it is essential that four characteristic times be defined and quantified. 1. Tcrit — This is the total time that it takes for a fire to increase from some initial size to its expected critical size. 2. Tj^ — This is the time necessary for a fire to reach a size suffi- cient to produce an alarm concentration of product, X, when using a sensor of sensitivity, S^. 3. tr — This is the time necessary for a sensor, or sensing system, to issue an alarm when exposed to alarm concentration levels. 4. T^ — This is the time necessary to successfully initiate evacua- tion and control measures once a fire alarm has occurred. The effective time for product transport becomes "Tf = Tcrit - (t^x + "Tr + ^c) (6) for sensor, S^, is £x = VfT^. (7) All of these times can have a significant effect upon the spacing for product- of-combustion sensors. The fire growth rates and the critical fire sizes determine x^^-^^. The values of t^ are determined by the size of a fire, the combustible involved, and the detector sensitivity. The time response of a sensor is dependent primarily upon the sensor and its particular characteris- tics. The evacuation and control time is dependent upon the area to be pro- tected, the proximity of personnel to that area, and the availability of fire extinguishant and suppression equipment. 36 Results of recent full-scale fire tests'* have shown that the critical times are in the 20- to 60-rain range, depending upon the ventilation velocity and the value taken for the critical fire size. These tests have also indi- cated that both the initial fire size, ra j , and the critical fire size, ra^., are linear functions of the ventilation velocity. In order to illustrate the effects of fire growth rates, detector sensitivities, and evacuation and con- trol times on the detector spacings, certain characteristic parameters will be assumed and the resulting detector spacings determined. These characteristic parameters are summarized in table 1. For the haz- ard parameters, both the initial and critical fire sizes are assumed to be linear functions of the ventilation velocity, in agreement with the results of the full-scale fire tests. Assuming wood as the primary combustible, the critical fire size is 100 W"/ at a ventilation velocity of 50 cm/sec (~100 f pm) . The production and sensor parameters are those previously discussed in defin- ing minimum detectable fire sizes. For convenience, the response time of each detector is the same and has a value of 60 sec, A 600-sec (10-min) time per- iod is assumed for the evacuation and control parameter. Although the initial and critical fire sizes are independent of fire growth rate, it is important that the fire growth rate be quantified in order to determine the detector spacings. To demonstrate the crucial importance of fire growths, three different rates, linear, quadratic, and exponential, are defined in table 1, along with respective critical times and alarm concentra- tion appearance times for each growth rate. Each of the three growth rates are normalized so that at a ventilation velocity of 50 cm/sec, the critical time is 1,800 sec (30 min) to reach a fire size of 100 KW, The critical fire sizes and times then scale with the venti- lation velocities according to the relationships defined in table 1, ^Litton, C, D., M. Hertzberg, and A. L. Furno. The Growth, Structure and Detectability of Fires in Mines and Tunnels. Proc. 18th Internat. Symp. on Combustion, Waterloo, Ontario, Canada, Aug, 17-25, 1980, The Combustion Institute, Pittsburgh, Pa,, 1981, pp. 633-639, 37 TABLE 1 . - Design parameters for products-of-combustlon fire protection systems Hazard parameters: m; = 0.011 Vf mc = 0.11 Vf a = m j/mc = 0. 10 Production parameters: KgMP = 2 X lO'^ particles/gram Kqq = 2.5 X 10^^ molecules/gram Kqq = 1.5 X 10^2 molecules /gram Sensor parameters: Sgj^p = 5 X 104 particles/cm^ Sco = 2.69 X 10 1^ molecules/cm^ (1 ppm) Sc02 = ^•'73 X 10^4 molecules/cm^ (25 ppm) Tr = constant = 60 sec Evaluation and control parameter: T^ = constant = 600 sec A = 1.5 X 105 cm2 (8 by 20 feet) Linear fire growth: iTif = ra^ [1 X 10"5(vf)t + a] T^pit = (9.0 X 104)/(vf) T^ = [(9 X 106)(S^A/K^) - 104](vf)-1 Quadratic fire growth: rkf = m^ [5.6 x 10-9(vf)t2 + a] T^^,^ = 1.27 x 104 (l/vf)l/2 T^ = [(1.62 X 1010)(S^A/K^) - (1.8 X 107)(vf)-M^/2 Exponential fire growth: mf = m^ [0.01(vf)e5.6 x 10 4t_^ ^ ^^ T^r-it = 1,800 in [86.2/(vf) + 1] T^ = 1,800 in [(8.62 x 105)SxA/Kx - 9.6](vf-1) + 1 Detector spacings: ^X = Vf [Tcrit - (Tx + Tr + Tc)] 38 With these parameters and growth rates defined, it is now possible to determine the detector spacings and the influence of various quantities on these spacings. Figure 3 is a plot of the detector spacings for SMP detectors as a function of the ventilation velocity for the three growth rates of table 1, Figure 3 vividly illustrates that the spacings are critical functions of the growth rate, and for this reason, growth rates need to be determined before detection systems can be properly implemented with specified levels of protection. Figure 4 is a plot of the detector spacings for the three candidate detec- tors as a function of the ventilation velocity. The much-reduced spacings for CO and CO 2 sensors are a reflection not only of the detector sensitivity, but also of the intrinsic production rates of these two products for wood. For SMP's, alarm concentrations are produced at the instant of flaming so that TsMP - 0> while for CO and CO2, alarm concentrations are produced at some fire size larger than the initial fire size so that Tqq > and Tqq > 0, thus reducing their effective spacings. For the SMP detector, the spacings are limited only by the detector response time and evacuation and control time. Clearly, the indication is that more sensitive CO and CO^ detectors are warranted. However, some caution should be exercised in this regard. The alarm levels are functions not only of the intrinsic signal-to-noise ratio of a detector, as has been assumed here, but also of the ambient background levels present in any real situation. For instance, if the average background level for CO is 5 ppm ± 5 ppm, it would be impractical to use a CO detector with a sensitivity of 0.5 ppm and set the alarm threshold at 10 times this value, or 5 ppm. Even with a sensitivity of 1.0 ppm, the alarm threshold signal may be only twice this background level. 1 / / - / / t / / - - - 1 / ! / I / ^ \ / 1 \ 1 - b) \ i o \ \ Quadrat V 1 1 1 1 1 1 » I 8 8 i O O SJ3i9UJ 1 1 1/ 1 1 1 / y 1 / §1 b> / lb. / y y 1 / I / y y ^; / /^ 1 1 1 Linear- growth 1 1 sjd;auj 39 U - D — r- 5CtO equ Sim xpoi ^ _C (U 0) — . y o ^, c (U >-*- _0 Q o a >-_Q >_ 8) -a ■ - D o V ^ J .i 1 [J O (U _„ — u > I' - s^ s ^^ CD O •- >" tors as a function of FIGURE 4. - Spacings for thr the three firegrowth tion of ventiloti growth rate def spacings exist growth rotes. o U »_ I — " - Spacings for Si ventilation vel rates defined i ro LU Q:: ID o u_ 40 By setting t^ equal to T^^i-t-, it is also possible to determine the limit- ing sensitivity for each detector. These limiting sensitivities are defined as those necessary to reduce the spacing to zero when both tr and t< zero. The respective limiting sensitivities are 'SMP = 1.48 X 10 5 particles Sco = 1.B5 X 1014 "'^l^^^l^^ (6.88 ppm) Spo, = 1.11 X 10^^ "'°^^^^^^" (41.1 ppm) If sensors with these sensitivities were used, then they would have to be essentially continuous throughout the entry, and even then, detection would occur at the critical fire size, which leaves no additional time to initiate evacuation and control measures. Figure 5 is a plot of the SMP detector spacing as a function of the evacu- ation and control time, t,,* at three different ventilation velocities. As t^ increases, the detector spacing decreases dramatically. Further, figure 5 indicates that the evacuation and control time is also a function of the ventilation velocity. As the velocity increases, the maximum allowable time (at i^sivip = 0) for evacuation and control decreases. These results indicate the need for rapid and effective evacuation and control times. 41 1,200 1,000 800 600- 400- 200- 1 1 Quadratic growth KEY Vf = 75 cm/sec Vf = lOOcm/sec Vf = 50 cm/sec ,200 1,600 Tc, sec FIGURE 5. - Spacings for SMP detectors as a function of the evacuation and control time at three ventilation velocities for the quadratic fire growth rate, defined in table 1. 42 CRITICAL FIRE SIZES The detector spacing at any given ventilation velocity is a function of the defined critical fire size. Further, the time available for evacuation and control can increase or decrease depending upon whether the defined criti- cal fire size is larger or smaller. The critical fire size can be defined in one of two ways: 1. It can be defined as the size of fire beyond which localized control measures are ineffective; or 2. It can be defined as the size of fire beyond which escape is either impossible or marginal. Recent data from full-scale mine timber fires ^ have shown that the size of fire (in kilowatts) necessary to insure propagation down a heavily loaded (80 pet wood loading, or greater) entry can be defined by the approximate relationship QP (KW) = 5.5 Vf (8) where the subscript i denotes the limiting fire size for propagation (superscript, p). For lower combustible loadings (~20 pet), the limiting fire size neces- sary for propagation increases by about a factor of 3.0. Then, depending upon the wood loading, the critical size necessary for propagation down an entry is in the approximate range 5.5 Vf < QP (KW) < 16.5 Vf (9) Equation 9, then, defines the critical fire size beyond which localized con- trol measures are essentially impossible. Alternatively, in terms of life safety and escape, the critical size can be defined in terms of the limiting concentrations for CO or SMP's, above which escape is either impossible or marginal. For CO, the limiting concen- tration is 2,750 ppm,^ and for SMP's the limiting concentration is defined as the concentration necessary to produce an 84 pet, or greater, reduction in light transmission over a path length of 366 cm (12 ft), which, in terms of particle concentration, is approximately 1.5 x lO'' particles/cm^. ^Tewarson, A., J. L. Lee, and R. F. Pion. Fuel Parameters for Evaluation of the Fire Hazard of Red Oak. Factory Mutual Research Corp., Norwood, Mass. Tech. Rept. RC 79-T-68, J.I. OC6N2.RC, December 1979, HI pp. ^Lee, C. K. , P. A. Croce, and J. S. Newman. Investigation of the Fire Haz- ards Associated With Timber Sets in Mines. Main Volume. Factory Mutual Research Corp., Norwood, Mass., Tech. Rept. RC80-T-27, J.I. OEONI.RA, March 1981, 108 pp. 43 Inserting these values into equation 4, the limiting fire sizes for escape, expressed in kilowatts, become for CO and SMP's, respectively qCO (kw) > 79 v^ (10) qSMP (KW) > 2.2 Vf (11) These sizes represent the maximum sizes beyond which escape becomes marginal owing to the extremely high levels of either CO or SMP's. By comparing equations 9, 10, and 11, it is evident that the smallest, limiting fire size is that defined by equation 11 for the production of SMP's. Consequently, then, the critical fire size should be that size of fire defined by equation 11. This is, in fact, the approximate critical fire size defined in table 1. Based upon these results, the evacuation and control time, x^,, should be expressed primarily in terms of the time needed for evacuation and escape from the developing fire hazard. SUBMICROMETER PARTICLE DETECTION From the examples presented in figures 2-5 of the previous sections, there is the clear implication that SMP detectors offer significant improve- ment over either CO or CO2 detectors. This aspect of the detection problem is important and worthy of some consideration. The processes responsible for the production of SMP's are physical in nature, rather than chemical. Prior to flaming, particles are produced primarily from thermal processes occurring at the surface of the combustible. At surface temperatures on the order of 150° to 250° C, SMP's are produced in significant quantities. As the combustible transitions to a flaming stage, particles are produced via the physical processes of condensation and nuclea- tion as well as via the thermal surface emission process. These production processes occur for practically all combustibles, and it is not surprising that the production parameters, Ks^p, are essentially constant, independent of the type of combustible and its chemical structure. This production process is in sharp contrast to that for product gases, such as CO or CO2. For a gas to be produced, a chemical reaction must take place between the pyrolysis fuel vapors and the oxygen in the surrounding environment. Because the production of product gases is chemical in nature, it is not surprising that their production parameters tend to vary markedly with the type of combustible. As an example, the CO production parameter for Pittsburgh seam coal has a value of roughly 5.8 x 10^^ . For wood, ^ ^ ^ gram ' the value previously quoted was 2.5 x 10^1 . Variations should be expected even for different types of wood, and for plastics and other synthetics rather severe reductions in the CO production parameter should be expected owing to a decreased carbon content. 44 rpri I The Bureau has, for many years, realized the potential advantages offered by SMP detectors in the area of early and reliable fire detection.^ Extensive research efforts by the Bureau to develop and evaluate the performance of such detectors is actively continuing. Much of this work is now culminating in long-term underground tests of two sensors that offer considerable promise. The first of these is the Becon Mk II, developed by Anglo-American Labo- ratories for use in South African gold mines. ^ This detector, shown in fig- ure 6, operates on the same principle as residential ionization-type smoke detectors. A radioactive source is used to ionize the air space between two electrodes. These ions produce a current, and when SMP's enter the air space between the electrodes, this current decreases, owing to attachment of ions to the particles. An analog signal is available to continuously map the current reductions, and once background levels are determined, alarm thresholds can be set. This detector is designed for mounting within mine entries and contains no moving parts. It is baffled sufficiently so that it is insensitive to ven- tilation flows. Also, the radioactive source is a B-emitter, krypton 85 gas, which renders the device insensitive to ambient dust and humidity. The second promising sensor is a Bureau-developed prototype, shown schematically in figiare 7.*^ In this device, a radioactive source is used to generate a region of unipolar ions (ions of one sign). A small internal pump provides a continuous flow of ambient air through this unipolar ion region. SMP's within the flow are charged by the ions and impact on a third electrode downstream of the ion region. The charged particles produce a current at this electrode which is proportional to the SMP concentration within the flow. The device is insensitive to humidity and temperature extremes, and the constant- flow pump renders the device insensitive to ventilation flow. To protect against ambient dusts a small cyclone is used on the inlet flow. The Bureau is also actively pursuing the development and testing of a fire detector that can be used in areas with high diesel backgrounds. This detector, developed under contract with Environment One Corp.,^^ uses a novel repyrolyzation chamber to precondition a fraction of the incoming air. The ^Hertzberg, M. , C. D. Litton, and R. Garloff. Studies of Incipient Combus- tion and Its Detection. BuMines RI 8206, 1977, 16 pp. ^van der Walt, N. J., B. J. Bout, 0. S. Anderson, and T. J. Newington. An All-Analogue Fire Detection System for South African Gold Mines. J. Mine Ventilation Soc. , S. Africa, v. 33, No. 1, January 1980, pp. 1-12. ^Litton, C. D., L. Graybeal, and M. Hertzberg. A Submicrometer Particle Detector and Size Analyser. Rev. Sci. Instru. , v. 50, No. 7, July 1979, pp. 817-823. ^ ^Under Bureau of Mines contract H0387025. The final report on this research will be available for inspection as soon as patentable information in the report has been cleared for release. 45 FIGURE 6. - The Becon Mk II SMP detector. u a. S-^ M- O 47 Sample in Filter Pyrolyzer 750 °F Cloud chamber B Detector Cloud chamber A Vacuum pump L£ Detector Ratio circuit \.5-\ FIGURE Alarm m The selective SMP detector showing the two mea- surement chambers and the pyrolyzer unit. unit, shown schematically in figure 8, consists of two cloud chambers, one measur- ing the total SHP concentra- tion without passing through the pyrolysis chamber, and the second measuring the SMP concentration after passage through the pyrolysis cham- ber. The diesel particles are unaffected by the pyrol- ysis chamber, and, as a result, both cloud chambers measure the same concentra- tion. When fire-produced particles are present, they repyrolyze, increasing their concentration dramatically, and thus increasing the total concentration as mea- sured by the second chamber. When the concentration ratio between the second and first chamber meets or exceeds some preset value, an alarm is given. This detector is presently being evaluated in underground mines that have significant diesel operation. CONCLUSIONS These three detectors represent significant improvements in the state- of-the-art for underground fire detectors. As these detectors are gradually accepted for use by the mining industry, there will exist a need for deter- mining how such detectors can best be used to provide for improved and well- defined fire detection systems. This need, and the manner in which it can be satisfied, formed the basis for the first sections of this paper. In this analysis, the detection system is defined in terms of a develop- ing fire hazard. If such systems are to be used to provide protection at some specified level, then it is essential that the system be directly related to the hazard. Certainly, the fire growth rates discussed here represent only a fraction of those that might be encountered in underground mines. Under cer- tain conditions, fires may grow more slowly, or they may grow more rapidly. Critical fire sizes may be larger or smaller, depending upon the particular area to be protected. Other factors can confuse the problem, such as high backgrounds from diesels, or ventilating flows that intersect and mix, reduc- ing combustion product concentrations. 48 It is felt, however, that the analysis provides a realistic first step towards the development of evaluation methodologies for fire detection sys- tems. The examples clearly illustrate the utility of this approach toward evaluating and optimizing fire detection system performance. The continuing development of this hazard-oriented evaluation methodology, coupled with the ongoing SMP detector research efforts can be expected to pro- vide the raining industry with both information and instrumentation necessary to insure optimum use of fire detection systems. 4P IMPROVED STENCH FIRE WARNING SYSTEM by William H, Pomroy ^ ABSTRACT Through an integrated program of laboratory research and field tests, the Bureau of Mines has developed and demonstrated an improved stench warning system for underground noncoal mines. The improved system employs a stench agent with chemical and physical properties superior to those of agents now in common use. The system also meters the release of stench fluid to provide closely controlled stench levels in the mine. In addition, the improved stench injectors are more reliable and easier to operate than most systems now used in mines. One complete system consisting of one injector for com- pressed air and one injector for each of two downcast ventilation shafts was installed at Kerr-McGee's Church Rock No. 1 uranium mine near Gallup, N. Mex. The injectors were activated on two different shifts during the course of a week as a part of the mines regularly scheduled semiannual fire drill. Sub- stantial improvements in warning times over those of the existing stench system were achieved for large areas of the mine. Analysis of air samples collected underground during the drills showed stench concentrations that closely matched design specifications, and no miners complained of excessive odor intensity. INTRODUCTION The stench system is widely used in underground noncoal mines to warn of a fire or other emergency. In the event of a fire underground, surface personnel are notified by the fastest means possible to activate the stench system. In a typical stench system, ethyl mercaptan, a highly odoriferous organic compound, mixed with Freon to reduce the liquid's flash point, is injected into the mine's compressed and/or ventilation air supply. The liquid is quickly vaporized, and the stench is carried by the airstreams to the work- ing areas underground. Miners, upon smelling the stench, evacuate the mine according to an emergency preplan. Although the stench system has been used successfully for over 60 years, present systems suffer several serious short- comings. Ethyl mercaptan is highly toxic and sometimes its use causes debil- ity in miners. It is also reactive with iron oxide, resulting in unreliable warning because the odor may fade when the agent must travel long distances in steel pipe. Present injectors are typically crude homemade devices that are unreliable and provide no control over the rate at which agent is released into the airstream. The result of the uncontrolled release of agent is that some work areas may receive unbearably high stench concentrations and other areas may be missed altogether. These devices do not provide for visual ^Supervisory mining engineer, Twin Cities Research Center, Bureau of Mines, Minneapolis, Minn, 50 indication of system status, valve positions, or of proper system operations. Finally, transit times for stench in compressed air lines can be unacceptably long where compressed air usage is minimal. In 1980, the Bureau of Mines embarked on a research program to upgrade the stench warning system. The objectives of this program were to improve the safety, reliability, and effectiveness of stench systems. The work was accom- plished through a research and development contract with Foster-Miller Associ- ates, Inc., Waltham, Mass. EXISTING WARNING SYSTEMS There are presently three types of fire warning system used in under- ground metal and nommetal mines: electrical systems, messenger systems, and air-carried systems. Electrical Systems Electrical warning systems are used where little compressed-air equipment is used and where ventilation velocities are low. Bells, flashing lights, gongs, and horns have been tried and have achieved limited success. These systems are not effective when miners are working outside the visible or audi- ble range of the alarm or when mine power fails. I«Jhere electrical equipment is used extensively in the mining operation, interruption of the power supply has been used to signal an emergency. This system Is also unreliable, how- ever, as power interruptions are common ira many mines, and electrical equip- ment may not be in use during the emergency. Messenger Systems Messenger systems are used in many mines as either the primary warning system or as a backup to the primary system. It is particularly common in small mines where personnel are not spread out over a large area. The dis- advantage of the messenger system is the time required for the signal to be passed to each worker, Air-Carried Systems The air-carried stench system is the most popular means of emergency warning in noncoal mines. It has been used extensively since before 1920. There are two methods of stench dissemination systems now in use: The vial breaking method and the pressurized gas canister method. Both systems use an ethyl mercaptan-Freon mix as the stench agent, and both systems can be used to inject stench into compressed air lines. The compressed-gas canister is also used at some mines to inject stench into the ventilation air stream. As a backup means of warning at some mines, a bottle of stench fluid can be thrown down the downcast ventilation shaft to provide stench in ventilation air. 51 Vial Breaking Method In the vial breaking method, a glass vial containing stench fluid is placed inside an airtight steel cylinder (fig. 1). The cylinder is connected in parallel to the mine's compressed-air line through two hoses fitted with globe valves. The system is activated by opening the globe valves to allow air to pass over the vial. Then a steel plunger is screwed into the cylinder to break the vial, releasing the stench fluid into the airstream. The practice of throwing a bottle of stench fluid down the ventilation shaft is simply a modified form of the vial breaking method. Pressurized Canister Method In the pressurized canister method, the stench fluid is contained in a canister pressurized to 400 psi with a fluorinated hydrocarbon propellant (fig, 2). The canister is connected to the compressed-air line or emptied into the ventilation stream through a short length of tubing. Activation of the system is accomplished by opening the valve on the canister and a second valve at the compressed-air line. The higher pressure in the canister forces the stench fluid into the compressed or ventilation air stream. Flexible hose Valve Stench fluid Vial breaking plunger Screen FIGURE 1. - Vial breaking stench injector. 52 Compressed air line Stench canister Valve Feed line Ball valve FIGURE 2. - Pressurized canister stench injector. 53 Deficiencies of Existing Stench Warning Systems Currently used stench warning systems, while fairly simple and relatively inexpensive, do suffer from several major deficiencies: a. The stench agent, ethyl mercaptan, is highly toxic. The short-term exposure limit recommended by NIOSH is 2 ppm (the exposure allowed for 15 min), b. The odor of ethyl mercaptan becomes Increasingly strong with increas- ing concentrations, with the result that some miners become nauseated. c. Ethyl mercaptan is highly corrosive and readily reacts with iron oxides, d. The Injection methods release the stench in a totally uncontrolled fashion. This uncontrolled release results in areas of the mine near the injection point being overwhelmed by the stench odor and the exposure of miners to potentially toxic concentrations of ethyl mercaptan. e. The injection methods have been unreliable. This problem was highlighted during two attempted, but unsuccessful, fire drills observed by project team members during one mine visit. The system in use at the mine was a vial breaking system injecting into compressed air. During the first attempted evacuation, one valve in the system was frozen in the closed posi- tion. With no visual indication that the valve was in fact closed, the mine employee assumed the valve was open. The result was that no stench was injected into the compressed air. In both attempted tests, the plunger screw was successful only in breaking off the narrow neck of the bottle. Insuffi- cient stench vapor was carried in the airstream during the second test to be detected by miners underground. At the another mine visited during the study, a similar system had never even been tested. Although a vial-breaking ethyl mercaptan stench system was the primary means of emergency warning, fire drills were accomplished by manually pouring banana oil into compressed-air lines underground. f. None of the observed injection systems had visual indications of sys- tem status. g. None of the systems had visual indications of system valve positions. h. None had visual indications of proper system operation. 1. Current systems occasionally miss areas of the mine. Remote areas, areas not drilling with air equipment, haulageways in fresh air, and dead-end headings were highlighted by mine personnel as potential problems. j. Transport times for stench in compressed air (particularly when com- pressed air usage underground is minimal) can be unacceptably long. 54 DEVELOPMENT OF IMPROVED STENCH SYSTEM The development of an improved stench system capable of safely, effec- tively, and reliably warning underground personnel of an emergency was accom- plished in two steps. First, a stench agent was selected, and second, an improved injector was developed. Selection of Stench Agent As noted above, ethyl mercaptan is the most common stench agent now used in mines. It is readily available, inexpensive, has a low odor threshold, and has a long history of use in the industry. Ethyl mercaptan does, however, suffer several weaknesses that can reduce its effectiveness as a stench agent. 1. Ethyl mercaptan is highly reactive with iron oxide. Therefore, it is subject to odor fade when it is transported in steel pipe. For this reason, ethyl mercaptan is no longer used as an odorant in the natural gas industry. 2. Ethyl mercaptan is highly toxic. The 8-hour time-weighted average concentration limit established by NIOSH and ACGIH is 0.5 ppm. A limit of 2 ppm has been set by ACGIH for exposures of 15 rain or less. In low concen- trations, ethyl mercaptan can cause headaches and nausea. At higher levels, it can irritate the skin and eyes, affect liver function, affect amino acid levels in the blood, and retard redox processes. 3. The odor intensity of ethyl mercaptan tends to increase steadily with increasing stench concentration. Where personnel are exposed to high concen- trations of stench, the odor can be overwhelming. 4. Ethyl mercaptan is highly corrosive to injection equipment and air lines. A survey of industrial gas odorizing agents was undertaken to identify a substitute for ethyl mercaptan. Compounds considered included isopropyl mer- captan, amyl acetate (banana oil), dimethyl sulfide, tertiary butyl mercaptan, and thiophane. Each compound was evaluated against ethyl mercaptan. Factors considered were boiling point, freezing point, vapor pressure, flash point, flammability limits, reactivity, solubility in water, toxicity, odor thresh- old, availability, cost, and past industrial uses. Table 1 summarizes the results of the survey. Thiophane emerged as the most desirable agent for use in the stench system. It is widely used in Europe as a natural gas odorant and has been used successfully on a limited basis in this country for that application. Thiophane is in the same chemical family as ethyl mercaptan, so its odor and odor threshold are about the same as ethyl mercaptan. Thiophane is not reactive with iron oxide so it is not subject to odor fade. The odor intensity of thiophane tends to stabilize at a moderate yet easily recognizable level, and it is much less corrosive than ethyl mercaptan. Ii a. 90.20. t-Butanethiol (TBM). 0.830 at 20° C (68° F). NA. -90° C (194° F). 1° C (34° F). -6.8 psia. -12.2° C (10° F). NA. NA. May oxidize in presence of iron oxides. HS; SO2. 0.3 wt-pct at 20° C (68° F). 0.5 ppm. 1,500 mg/kg. 4,020 ppm. None reported. >1 ppb. Readily available. Natural Gas Odorizing Co. (NGO), Tex.; Penn- walt Corp., Pa.; Phil- lips Petroleum, Tex.; NA. Widely used as natural gas odorant; major component in blends with isopropyl mercap- tan; high freezing point precludes use in pure form. a: a. X u 4-1 lA 1 7 141. - z z X j Nonreactive HS; SO2 Negligible Not established 2,450 mg/kg 44,200 ppm None reported 1.3 ppb Readily available Natural Gas Odorizing Co. (NGO), Tex.; Penn- walt Corp., Pa. ; Phil- lips Petroleum, Tex.; Eastman Kodak, N.Y. $1.82/lb (NGO) Widely used as natural ral gas odorant; odor dissipates promptly; has moderate odor strength plateau (that is, odor will not become unbearably strong). ^1 x: ^ ^ Ethanethiol; EtSH 0.841 at 15.5° C (60° F). 1 ? 1_ J °^ t J Oxidizes readily in presence of iron oxides. ^ It \ CO^ c 1 t c c CNS affected as low as 4 ppm; may induce headache, nausea. 0.6 ppb Readily available Natural Gas Odorizing Co. (NGO), Tex.; Pennwalt Corp., Pa. ; Phillips Petroleum, Tex. ; Eastman Kodak, N.Y. $1.02/lb (NGO) Widely used as mine stench-warning agent and liquid petroleum gas odorant. .-H X s S 5 X 1 CD ^OJ 1 . t °7 : c X 1 1 1 S 1 X 2 wt-pct at 20° C (68° F). Not established OC c c c z X C Readily available.. Natural Gas Odoriz- ing Co. (NGO), Tex. ; Pennwalt Corp., Pa.; Phil- lips Petroleum, Tex. Widely used in nat- ural gas; odorant blends with mer- captans; rarely used in pure form. O X I Pentylester; banana oil; pear oil. 0.876 at 20° C (68° F). 4.S X air I -- z OC c z i 1 z Slightly soluble... 100 to 125 ppm ' z .2 " •? 1- i available. Ashland Chemicals, Mass. and Ohio. $0.75/lb (Ashland) Limited use as mine stench-warning agent. 1 ^^ rH X >N "^ a 1 to fc < z z 'I CO z X £ & 1° ^ X t J X 1 i § < X 1 Natural Gas Odorizing Co. (NGO), Tex.; Penn- walt Corp., Pa.; Phil- lips Petroleum, Tex, NA gas; odorant blends with tertiary butyl raercaptan (acts as antifreeze); rarely used in pure form. s X 1 1! 1 •H -c ^ It I a- 1 'il > .^1 >N LXl >- ii CO ,- \ : J I 1 X c "• .X If 0. ^ N 4-1 X X J s> CO c c ^C 1 X c X 1 ^ i tS 56 In addition, available data suggest thiophane may be far less toxic than ethyl mercaptan. Although ACGIH and NIOSH have not established a threshold limit value (TLV) for thiophane, the lethal concentration (LC5Q) for inhaled vapors of ethyl mercaptan is 4,420 ppm for rats. The same LC5Q for thiophane is 44,200 ppm. Based on this toxicity study, MSHA has approved the use of thiophane in mine stench systems. Thiophane 's other pertinent chemical and physical properties were also favorable to its use for this application. Like ethyl mercaptan, thiophane is flammable and needs to be mixed with an inerting agent to avoid possible explosion hazards. Laboratory flammability tests indicate a 10 to 1 mix of Freon 113 and thiophane is totally incombustible. Stench Injector Development Prior to hardware development, numerous mining companies and mine safety specialists were contacted to help set performance specifications for the improved stench warning system. The specifications thus developed cov- ered two categories: performance range specifications and general design specifications. Performance Range Specifications The following subsections define the range of operating conditions for an improved stench fire warning system. Ventilation Airflow The range of U.S. hardrock mine ventilation airflows was established based on the results of the field survey and through discussions with knowl- edgeable personnel in operating mining companies, the Bureau of Mines, and MSHA. It must be emphasized that the specified range of ventilation airflows of from 50,000 to 2 million cfm applies to individual intake systems and not to total mine ventilation flows. It is at these individual intakes that the stench system must be applied. The minimum significant ventilation airflow specified is 50,000 cfm. While individual ventilation airflows of less than 50,000 cfm do occasionally exist — through natural ventilation entering worked-out areas, for example — such flows are not considered as part of the mine ventilation plan. There- fore, 50,000 cfm is specified as the minimum airflow applicable to the warning system. The maximum single ventilation airflow specified is 2 million cfm. This is considered to be a realistic maximum based on current trends. The vast majority of mines do not have individual intake airflows in excess of 300,000 cfm. Exceptions, however, do exist. One mine in Colorado, for exam- ple, has an intake ventilation shaft passing airflows approaching 1.3 million cfm. 57 Compressed Airflow A compressed airflow range from 2,000 to 20,000 cfm has been specified. An absolute practical minimum flow required by the smallest of mining opera- tions is considered to be 2,000 cfm. Such a flow would typically support no more than 10 operating machines. A practical maximum flow derived from a sin- gle compressor plant and delivered through a single supply line is specified as 20,000 cfm. Where flows in excess of 20,000 cfm are required, an addi- tional compressor plant and pipe delivery system would be installed. Compressed Air Pressure An absolute minimum pressure of 70 psi is required by air-operated machines underground. To satisfy this requirement and to handle multiple machine operation and delivery system leakage, a range just downstream of the compressor plant of 80 to 150 psi has been specified. While 150 psi is con- sidered to be a maximum pressure associated with minimum machine usage, 80 psi covers the high machine utilization case. Stench Gas Concentration Concentrations were specified for both stench agents, ethyl mercaptan, and thiophane. The specified concentration range for ethyl mercaptan is 0.5 to 2.0 ppm. This range is for concentrations exiting the air line; injection concentrations cannot be specified, owing to a total lack of data on ethyl mercaptan reactivity. The lower limit of the concentration range (0.5 ppm) is 50 times higher than the odor threshold. The upper limit of the concentration range (2.0 ppm) is the maximum short-term exposure limit recom- mended by NIOSH. Owing to a lack of established TLV's for thiophane, the same concentra- tions were specified. This range is based on the assumption that thiophane is no less toxic than ethyl mercaptan, even though the literature data suggest it may be only one-tenth as toxic. These concentrations are specified for point of injection in compressed air and allow for expected dilution. The point of injection concentrations should approximate the the concentrations at the point of exit from the com- pressed air lines since thiophane does not exhibit the reactivity with iron oxides exhibited by ethyl mercaptan. A stench dilution factor is not required for ventilation air. Thus the lower limit of the concentration range in ventilation air is 0.1 ppm. Equipment Operating Temperature Range An equipment operating temperature range of -30° to 125° F was speci- fied based on the range of climatic conditions that can be anticipated in U.S. hardrock mines. This range does not include solar radiation loading. 58 Ventilation Air Temperature Range Intake air temperatures correspond to the range of system operating tem- peratures discussed above (-30° to 125° F). Compressed Air Temperature Range A compressed air temperature range of -30° to 200° F was specified. Air from the compressor is typically cooled and stored in a receiver vessel. Under low-use conditions the air temperature in the receiver may approach ambient conditions. This establishes the low limit of -30° F, Conversely, under high-use conditions the air temperature may be significantly above ambient. Under these conditions, the maximum compressed air temperature was estimated to be 200° F, Equipment Operating Humidity Range A range of from 20 to 100 pet humidity was specified based on worst-case climatic conditions. Applied Vibration The equipment should be located away from vibration sources such as the compressor plant and main ventilation fan equipment. It should be rigidly mounted to a secure and stable structure to insure that it is not subjected to applied vibrations. Environmental Corrosion Equipment installed in a mining environment must tolerate a wide range of applied corrosives. Corrosives are present in mine water and in mine atmos- pheres. Depending on the mining applications, these corrosives will range from acidic to caustic. A range of pH values from 4,0 to 9,0 was specified. General Design Specifications The following subsections describe the general characteristics required of an improved stench warning system. System Reliability System reliability is considered to be the most critical feature of an improved system. The system, therefore, must be simple, requiring minimal maintenance, and be able to be activated months after installation with a minimal possibility of failure. System Life The basic injection system should last indefinitely requiring only recharging and routine maintenance after each activation. 59 Ease of Operation The activation of the system must be simple. Ideally it would require only the pushing of a button or the turning of a simple valve. A simple sys- tem would require minimal training of mine personnel responsible for system activation during emergencies and would eliminate or minimize any possible confusion during emergency conditions. The system must be capable of manual initiation, and the basic design should also be capable of remote operation. Manual operation must be simple and obvious and must involve a single operation. An example of this would be the movement of a lever arm from horizontal to vertical. Each position of the arm would be clearly marked, as would the required direction of travel. Although remote operation was considered to be beyond the scope of this pro- gram, system design must be compatible with the needs of remote operation. The system should with modification, respond to an electrical initiation sig- nal. This would be typified by the operation of a normally closed solenoid- operated valve. System Status Indication The system should present the following (preferably visual) indication of its operational status: a. Ready to operate. b. System is operating correctly. c. System has operated correctly. Y The system status must be inspected at regular intervals. The easier the system is to inspect, typically, the more often it will be inspected. Visual inspection once a shift is recommended. The inspection time required should therefore be minimal. Inspection and system status indication should take the form of positive go-no go features to provide positive inspection at a glance. Following initiation of the system, a positive indication of system operation should be available. The equipment must show, ideally through visual means, that stench fluid is flowing. This is of particular importance when injecting stench into compressed air where the resultant smell is not apparent at the injection point unless an additional valve is available. Following a complete operation cycle, the equipment should show posi- tively that it has operated and that it must be reset prior to further operation. System Hardware Cost A maximum system hardware cost of $4,900 was established. This figure is the median figure obtained as a result of the field survey. In reality, a total system cost significantly less than this is more realistic. 60 System Installation Labor requirement to install the system should not exceed 8 hours. System Labor Maintenance Cost A maximum labor maintenance cost equivalent to 0.5 worker-hours per month was established. Once again, this figure is based on the median figure obtained during the field survey. This figure represents the time required to check the status of the system, but does not include the time or cost to recharge the system after an evacuation or evacuation drill. The improved stench warning system specifications are summarized in table 2. TABLE 2, - Stench fire warning system specifications Performance range specifications: Ventilation airflow ft^,. 50,000 to 2,000,000, Compressed airflow ft^/rain,, 2,000 to 20,000. Compressed air pressure psig,. 80 to 150, Stench gas concentration, ppm: In compressed-air line: Ethyl mercaptan 0,5 to 2,0, Thiophane 0,5 to 2,0. In ventilation air: Ethyl mercaptan 0.1 to 2.0. Thiophane 0,1 to 2,0, Minimum stench injection time period, min: Compressed air 30. Ventilation air 10. Equipment operating temperature..." F.. -30 to 125, Ventilation air temperature ° F,, -30 to 125, Compressed air temperature ° F,, -30 to 200. Equipment operating humidity pet.. 20 to 100. Vibration (applied) None. Environmental corrosion pH.. 4.0 to 9,0, General design specifications: System reliability Extremely high, the system should be simple with a minimum number of moving parts. System life Indefinite with periodic maintenance. Ease of operation Simple — requiring minimal training. System hardware cost $4,900 maximum. System installation cost 8 worker-hours. System labor maintenance cost (max.). 0.5 man-hours /month maximum 61 The specifications presented above define a system that must introduce a liquid odorant into compressed or ventilation air at a relatively constant rate over an extended period of time. The system must perform this function despite variations in airflow, ambient temperature, and airflow pressure. Above all, the system must require little maintenance, be simple to use and extremely reliable. Design Concept Development Three design concepts were developed and evaluated relative to the per- formance specifications: Pressurized canister with metered orifice, pressure balanced with metered orifice, and variable rate injection pump. Each concept is described below. Pressurized Canister With Metered Orifice This concept is a logical extension of the pressurized canister injection method now in common use. An injector of this type could be fabricated by simply installing a metering orifice in the stench delivery tube (fig. 3). Manual valve Pressurized canister Pressure indicator Filter Solenoid valve Metering orifice FIGURE 3. - Pressurized canister with metered orifice concept. 62 Although this design is simple and in theory could satisfy the system per- formance specifications, injector reliability would be extremely low. The low reliability would result because an extremely small orifice is required to meter the contents of the canister (pressurized to 400 psi) into the airstream over the desired 10-min (for ventilation) or 30-min (for compressed air) time period. Internal canister pressurization to at least 400 psi is necessary to assure complete discharge of the canister contents. Orifices sized from 0.001 to 0.013 inch would be required to achieve the desired stench flow rates. Particles as small as 25 ym could plug an orifice of this size, and the probability of encountering particles of 25 ym diameter, even in a clean environment, is very high. Since nothing approaching clean-room conditions exists at a typical mine, the chances of a plugged orifice (and, therefore, an interrupted stench discharge) are also very high. In addition, a wide varia- tion in stench flow rate would result owing to canister propellant pressure decay. The flow rate could be expected to vary as much as 50 pet over the period of the discharge. Pressure Balanced With Metered Orifice The pressure balanced concept was developed to take advantage of the pos- itive features of the previous concept while eliminating its undesirable fea- tures. The small orifice sizes of the previous example can be replaced with much larger sizes by eliminating the high pressure differentials across the orifice. This can be accomplished by introducing line pressure on both sides of the orifice via a pressure balancing line (fig. 4). Orifice sizes could range from 0.015 to 0.071 inch (versus 0.001 to 0.013) and still achieve the desired stench flow rates. Since high-pressure propellants are not required to expel the canister contents, the problem of variable flow rates of stench fluid can be sharply reduced or eliminated. J 63 Pressure balancing line Stench container Driving liquid Stench liquid Filter Stop valve Metering nozzle FIGURE 4. - Pressure balanced with metered orifice concept. 64 Variable-Rate Injection Pump The variable rate Injection pump concept is the most precise method for stench injection. A precision diaphragm metering pump placed between the stench canister and the airstream can effectively control stench fluid flow rates to extremely close tolerances (fig. 5). The primary disadvantage of this system is its complexity and reliance on electric power. These factors could lead to very low system reliability. In addition, the cost of such a system would be quite high compared to that of the other two concepts. Table 3 shows a performance-comparison tradeoff. TABLE 3. Concept selection matrix Parameter Concept 1 — pressure canister with meter- ing orifice Concept 2 — pressure balanced metering orifice Concept 3 — variable-rate injection pump Rating Value Rating Value Rating Value Stench concentra- tion range Complexity Filtration level required Acceptable Low High 5 Good Good $300 $100 15 Remote only Yes High Average High Low High Medium 3 3 1 2 3 3 3 2 1 3 3 1 2 3 3 3 2 Acceptable Low Medium 5 Good Good $500 $50 1 Remote only Yes Low High High Medium Medium High 3 3 2 2 3 3 2 3 3 3 3 3 3 3 2 2 3 Acceptable Medium High 2 Poor Good $1,200 $100 15 Yes Yes Low-high Low Low High Low Low 3 2 1 Number of sizes required Status indication. Ease of manual operation Estimated initial 3 1 3 1 Estimated operat- ing cost 2 Number of signifi- cant leak points. Electric power required. ........ 1 1 Remote capability. Temperature sensitivity Reliability Maintainability... Training requirements Life 3 2 1 1 1 1 Safety 1 Total 41 46 28 Rating value 3 2 1 Best Second best Worst Maximum rating value total = 51. 65 Solenoid valve Volume metering pump Pressurized canister Pressure indicator Filter Metering nozzle Manual valve Electric motor 4I Back pressure valve FIGURE 5. - Variable rate injection pump concept. Hardware Development In light of the above analysis, the pressure balanced with metered ori- fice concept was chosen for further development. Additional investigation of this concept, however, revealed that while the pressure-balancing part of the concept was attractive, the use of a second fluid to provide the driving head was not straightforward. After considering various alternatives, the decision was made to meter the stench mixture using an orifice and a gravity head. The principle that allows the canister to empty at a constant rate is illustrated in figure 6. 66 Canister body Standpipe Mixture surfoce Orifice Pressure balancing loop Airflow (very slow) Stench mixture stream (enters either compressed air line or ventilation system) where and FIGURE 6. - Mixture metering principle. For an orifice, Q = CpA /2ih Q = flow rate through the orifice, Cj) = orifice discharge coefficient, A = orifice area, g = gravitational acceleration, h = head across orifice. 67 Referring to figure 6, the pressure difference between point 1 (just below orifice, entry into standpipe) and point 2 (base of standpipe, where air enters stench mixture) is negligibly small, so that the orifice sees a con- stant head (h) [not the varying head (H)]. Thus, since all of the other terms in the orifice equation are constant, the mixture flow rate is constant. During startup, before the standpipe is emptied of stench mixture, the flow rate is greater than the steady-state value. This does not last very long and is easily minimized by making the standpipe cross sectional area as small as possible. When the stench mixture surface drops below the tip of the standpipe (fig, 6, point 2) near the end of the run, the flow rate begins to decrease, since effective head decreases, eventually dropping to zero. This effect can be lessened by minimizing the internal area of the canister below the tip of the standpipe. Both of these considerations were incorporated into the canister design. Three models of stench injectors were designed, two for compressed air application and one for ventilation air use. The three models are essentially identical, and differ only in body length, orifice size, and orifice head. All models are manually operated by opening two quarter-turn ball valves. All injectors meter the stench mixture at a constant rate over most of their run- ning times, which are 30 min for the compressed air-models and 10 rain for the ventilation air model. Because air pressures across the orifices are bal- anced, the stench mixture flow rates are immune to air pressure variations, 2 The general arrangement of the stench injector is shown in figure 7, The injector is assembled from standard and modified commercially available hard- ware, a length of machined schedule 40 pipe and two machined end caps. The injector assembly is of all stainless steel construction to minimize the internal accumulation of corrosion particles. The housing, end caps, end cap fittings, and standpipe form a welded assembly. The pressure-tight screw- on arrangement between the housing assembly and lower ball valve traps a com- mercially available 200-mesh strainer and orifice plate within the body of the injector. These units are sealed by an 0-ring and copper gasket arrangement. Body lengths for each of the three models are sized to provide storage for sufficient stench mixture commensurate with the flow rate and duration required. Detailed dimensions and capacities for each of the three models are listed in table 4, 68 TABLE 4. - Stench injector operating data Items Model 1, compressed air Model 2, compressed air Model 3, ventilation air Airflow range scf m. Metering time min. Mix flow rate cm^. Mix volume cm^. Stench volume cm^. Body height in. Orifice size in. Orifice head (h) in. 2,000-6,500 30 4.21 129 11.7 3.2 0.015 1.28 6,500-20,000 30 13.7 414 37.6 8.4 0.027 1.29 40,000-650,000 10 84.1 845 76.8 16.3 0.061 1.87 Standpipe Stench fluid Driving head Ball valve Pressure balance line To compressed or ventilation air FIGURE 7. - Improved mine stench injector. 69 Laboratory Testing The system was thoroughly laboratory tested following fabrication. Tests involved discharges of nonodorized liquids as well as the thiophane-Freon 113 mixture selected for stench warning use. Tests in both ventilation and com- pressed air were performed. Parameters measured Included fluid flow rates, discharge times, and stench concentrations in the airstrearas. All tests indi- cated conformance with the system performance specifications. Field Testing The improved stench warning system was field tested at the Church Rock No. 1 uranium mine of the Kerr-McGee Corp. The system was activated on two different shifts as a part of the mine's regularly scheduled twice-annual fire drill. Substantial improvements in warning times over those of the existing system were achieved for large areas of the mine. Mine management was also Impressed by the simplicity and ease of operation of the systems. A detailed description of the tests follows. Mine Layout and Ventilation The test mine has two main levels, designated 1-4 and 1-5, the 1-5 being 300 feet below the 1-4. Raises are driven from these levels to the ore, which is 50 to 100 feet above the level in each case. The ventilation scheme has intake air downcasting to the haulage levels. The air travels along the haulage drifts, up the raises, through the stopes, and finally through a network of exhaust drifts in the ore horizon and out the exhaust boreholes. Exhausting fans are located on the surface at the exhaust boreholes. For both ventilation purposes and administrative reasons, the 1-4 level of the test mine is divided into two segments. The area to the east of the two air doors has its own hoisting plant, compressed air supply, and ventila- tion system. The two doors form an airlock and are normally closed; however, there is a minor leakage flow from the east side to the west. The compressed- air systems are connected, and there is normally a small net flow also from east to west. Thus, the stench tests were conducted in the west side of the mine without disturbing the east side. Existing Stench System The existing stench warning system at the test mine consisted of two vial-breaking canisters installed on the main compressed air line just down- stream of the receiver. Each canister contains one 500-gram glass vial of a 15 pet ethyl raercaptan in Freon mixture, A similar system is Installed on the air line in the east shaft. Standard procedure is the breaking of both vials simultaneously for each fire drill. 70 The emergency plan at the test mine is different from most plans in that it does not call for immediate evacuation. Standard procedure requires that miners take an air hose into a dead-end drift or stope and barricade. The standard procedure for fire drills (the miners are always informed of drills beforehand) is that the miners note the time at which they smell the stench and proceed to a dead-end drift with an air hose and barricade mate- rials. Barricades are not actually built. Environmental samplers visit each stope and fill out a form showing the location of each miner; the time he smelled the stench, and the strength of the stench; the action taken by the miner; and the miner's comments. Injector Selection and Installation The two intake airways at the test mine have flow rates of 221,000 and 45,000 cfm. Therefore, one ventilation air injector (Model 3) was installed in each. The peak output of the air compressors is about 9,000 cfm, so a Model 2 injector was selected for the compressed-air system. One ventilation air injector was installed in a crawl space located just below ground level near the top of the main shaft. Because of the high rate of activity near the man and supply hoist, this location was chosen so as to minimize the possibility of accidental system activation. Approximately 10 feet of copper tubing was suspended from the injector down into the shaft to carry the stench agent. The other ventilation air injector was installed on a leg of an emergency escape hoist headframe positioned over venthole 6, and a similar length of copper tubing was suspended in the shaft. This injector was mounted approxi- mately 7 feet above ground level, but it could easily be reached by a caged ladder attached to the headframe leg. The compressed-air injector was mounted on the main receiver for the compressed-air system and a piece of copper tubing approximately 2 feet long joined the injector to the main compressed-air line through a shutoff valve (fig. 8). This valve was installed to allow for complete removal of the injector system without interfering with the compressed air system. For each installation, two 2-inch pipe clamps with rods were used to grip the canisters. The clamp bases were welded to a steel plate which in turn was welded to the main support member. This created a solid support unit but still allowed for clamp rod adjustment and locking. The clamps provided a means of quickly removing and reinstalling the canisters during refilling operations. 71 I FIGURE 8. - Improved stench injector mounted on compressed air receiver tank. 72 Test Schedule The test mine normally conducts fire drills on all three shifts during a 1-week period. Tests on both of the new injectors as well as the mine's existing systems were carried out during a single week according to the fol- lowing schedule: a. New system: Tuesday, 6:00 p.m. (second shift), b. Exist- ing system: Wednesday, 6:00 a.m. (third shift), c. New system: Thursday, 10:00 a.m. (first shift). This schedule provided an excellent opportunity to compare the perform- ance of the new and existing systems under constant mine conditions. Stench Release and Sampling Procedure For the two tests of the new system, the injectors were activated simul- taneously at the three injection points at the times indicated in the above schedule. In the test of the existing system, both vials were broken at the indicated times. During the second test of the new system, bistable gas samplers were used to take air samples at some of the workplaces at 5-min intervals beginning with the arrival of the stench. These were later analyzed to provide quanti- tative stench concentrations. Stench Test Results Stench Transit Times The main criteria for the evaluation of a stench warning system are the degree of coverage of the mine and the elapsed time between the activation of the injectors and the detection of the odor at various locations. To give an overall view of the performance of both the new system and the existing sys- tem, transit times to various areas of the mine were plotted on maps of the haulage levels. These transit times are based upon stench arrival times as recorded by the mine's environmental samplers during their survey of the work- ing places during the drills. Figure 9 shows the results of the first and second tests of the new sys- tem, respectively. Figure 10 shows the results of the test of the existing system during the same week. Since this test was run on the third shift, there were fewer active workplaces for which arrival times could be recorded. Therefore, the results of two earlier tests of the existing system, as con- structed from the mine's records, are included in figure 10 in order to pro- vide a better comparison of the old system with the new system. 73 A. Test of September 9,1980, 6.00 PM 1-4 Level IINI5 20-^ /9i Ventik 10 No smell " w — II- V EY Ventilation air injection point L.j^ Stench transit time, r min in stope Stench transit time. 115 min. on haulage level B. Test of September 1 1, 1980, 10:00 AM FIGURE 9. - Stench transit times for improved stench warning system. 74 1-5 Level TK |'/2V/l9 '0 20 '5\ ), 1-4 Level ,^|Jl20 16 18 18 ' 15 >4. Test of December II, 1979, 10:00 AM ■W- 1 fo^JTTi^ 1-4 Level J ktgo smell 1 ^=r pN^^r^^rp Hl- 13 No smell B. Test of December 1 1, 1979, 600 PM ^^ 1-4 Level 1 I ■ ^\\ \l_3 Hh— II- KEY (g) Ventilation air injection point L|Q Stench transit time, r min in stope . .J. Stench transit time. mm. on haulage level C. Test of September 10,1980,6:00 AM FIGURE 10. - Stench transit times for existing stench warning system. 75 Stench Concentrations Table 5 shows the analyses of air samples collected during the second test of the new system. The locations of the sampling points are shown in figure 11. TABLE 5. - Analyses of air samples Sampling point Minutes after activation Thiophane concentrations, ppm 1 7 12 17 22 16 7 12 17 22 27 1.6 1 1 1 3.3 1.6 2.3 2 1.1 3 — haulage level 3 — haulage level 3 — stope. ............... .1 .8 .2 3— stope 3 — haulage level .1 .2 The objective of the new system design was the production of stench con- centrations in the range of 0.1 to 2,0 ppm. The concentrations obtained closely matched the design goal. Miner and Management Comments Of the 23 miners who were interviewed during the second test of the new system, only two rated the stench as reaching an "annoying" level, and none rated it as reaching a "sickening" level. During the first test of the new system, one environmental sampler stated that the miners thought the new stench was not as sickening as the old. Mine management was pleased with the new system not only because of the improvement in warning times, but also because of the ease and simplicity of operation of the new injectors. The turning of the valve stem, which breaks the vial in the existing system, is very difficult. In addition, it is necej sary to open two globe valves, whose positions are not readily apparent. The new system, which requires only the manipulation, in any order, two easy-to-turn, 90° valves, was regarded as far superior. of 76 o) E Q. 0) 3 '^ ■F. £ ■? ^ c .i= c t t -^ en •K c _ o o Q. Q. O a. -:= ^ "Q-'q. Q- c .E £ e E o o o o > Q. (/) W en — ro 77 COMPARISON OF THE PERFORMANCE OF THE NEW SYSTEM AND THE EXISTING SYSTEM The overall average stench transit time for all areas was reduced from 19.6 min with the old system to 10.5 min with the new system. The most dramatic improvement in transit times occurred at the western end of the 1-5 level. No doubt this is due to the injection of stench into intake ven- tilation air at venthole 6. Substantial improvement also occured at the western end of the 1-4 level. This is probably due to the injection of stench into the intake air at the main shaft. It should be noted that, because fires can have a profound influence on ventilation flows, stench injected into downcast shafts may not always result in reliable fire warning. Where compressed air is used, therefore, stench injected into the ventilation stream should be supplemented with stench injected into the compressed air. Both systems have some trouble providing consistent coverage of the area near the eastern boundary of the 1-4 level. This area is at the extreme end of both the ventilation system and the compressed-air system. Changes in the ventilation system to increase airflow to these areas would likely improve coverage of these areas. Recall also that for the tests of the upgraded system, release of the stench was synchronized so as to occur simultaneously at the main shaft and at venthole 6. In an actual emergency, a person would have to drive over a dirt road from the main shaft to venthole 6 in order to activate the injector. This would take 10 min under ideal conditions and may be impossible during inclement weather. Thus, the warning time advantage gained by the injection of stench at venthole 6 would be partially negated by the delayed activation time. I CURRENT AND FUTURE RESEARCH Current research efforts are aimed at providing remote stench injector activation capability and in developing a stench odor neutralizer. Future plans call for demonstrating a complete system in a very large, old, spread- out multilevel metal mine. SUMMARY AND CONCLUSIONS An improved stench warning system for underground noncoal mines has been developed and successfully in-mine tested. The system utilizes a superior stench agent and employs an injector that meters a controlled volume of stench fluid into either compressed or ventilation air. The system is simple, has no moving parts, is easy to use, requires little maintenance, is simple to recharge, and is designed for high reliability. The system is commercially available for about $500 per injector. 78 COMPUTER-AIDED VENTILATION MODELING by John C. Edwards^ ABSTRACT The successful planning of miner escape and rescue routes requires ade- quate information regarding the ventilation, contaminant concentration, and temperature throughout a mine associated with a hazardous event such as a fire. A computer program, was developed by Michigan Technological University (MTU) for the Bureau of Mines under contract J0285002, predicts in a quasi- steady-state approximation the ventilation in the airways of a multilevel mine, as well as contaminant concentrations and temperatures, when a fire is present. The program also predicts the steady-state concentration distribu- tion in a mine network due to a specified gaseous contaminant source. A recent development in the program is the capability to predict the real-time contaminant concentration in the airways produced by a fire in its early stage of development before it affects the flow ventilation. These programs should be useful in the implementation of ventilation to avoid excessive methane con- centrations in gassy mines, metal and nonmetal, and in the development of strategies for miner escape and rescue following a hazardous event. INTRODUCTION In order to establish a safe working environment for miners, it is bene- ficial for mine ventilation planning to have a predictive capability for mine ventilation airflow following the occurrence of a hazardous event, as well as under normal mining operations. This capability is an important component for the identification of escapeways and rescue routes for miners following a haz- ardous event. A number of computer programs,^ among others, have been devel- oped that calculate flow in a mine network under normal mining conditions at ambient temperature. These programs, however, do not account for hazardous ' events such as a fire or methane emission in a mine network. i 'physicist. Fires and Explosions, Pittsburgh Research Center, Bureau of Mines, Pittsburgh, Pa. ^Hashimoto, B. Analysis of Mine Ventilation Distribution Networks by Digital Computers. Bull. Sci. Eng. Lab., Waseda Univ., Japan, No. 17, 1961, pp. 18-29. McPherson, M. J. Mine Ventilation Network Problems. Colliery Guardian, V. 209, 1964, pp. 253-259. Wang, Y. J., and H. L. Hartman. Computer Solution of Three Dimensional Mine Ventilation Networks With Multiple Fans and Natural Ventilation. Internat. J. Rock Mech. Min. Sci., v. 4, 1967, pp. 129-154. Wang, Y. J., and L. W. Saperstein. Computer-Aided Solution of Complex Venti- lation Networks. Trans. SME-AIME, v. 247, 1970, pp. 239-250. I 79 A computer program was developed by Michigan Technological University^ under sponsorship of the Bureau of Mines' Pittsburgh Research Center that predicts ventilation flow under normal and hazardous conditions. Hazardous conditions Include normal methane emission and fires. This program Is distin- guished from other programs^ in that natural ventilation is calculated from the temperature distribution that exists in the mine or is produced in the airways by a fire. A quasi-steady temperature calculation in the airways is used to determine a steady state ventilation flow and contaminant distribu- tion. The program also calculates the methane concentration distribution associated with a given ventilation flow for a gassy mine. The capability to predict airways that would be critical to a miner's safety in planning escape routes is greatly enhanced by utilization of this program to simulate a given mine subjected to various hazardous events. This program is applicable to both multilevel coal and hardrock mines. Details of the program are docu- mented.^ A current modification to the program by Michigan Technological Uni- versity under sponsorship of the Bureau of Mines' Pittsburgh Research Center is the capability to compute the real-time spread of contaminants throughout a mine network. MODEL DESCRIPTION The MTU program represents a mine by a network of N: junctions (cross cuts and intersections) and N^ branches (airways). For computational purposes the mine network is represented by N,„ = Nj, -N: + 1 meshes. The ventilation in the network is determined by simultaneously solving N^^ equations for conserva- tion of energy in a mesh and N- equations for the conservation of mass at each junction. The conservation of energy in a mesh is a restatement of the first law of thermodynamics as applied to energy losses due to friction, fans, and ^Greuer, R. E. Study of Mine Fires and Mine Ventilation. Part 1. Computer Simulation of Ventilation Systems Under the Influence of Mine Fires. BuMlnes Open File Rept. 115(l)-78, October 1977, 165 pp.; available for consultation at Bureau of Mines facilities in Denver, Colo., Twin Cities, Minn., Bruceton and Pittsburgh, Pa., and Spokane, Wash.; the National Mine Health and Safety Academy, Beckley, W. Va.; and the National Library of Natural Resources, U.S. Dept of the Interior, Washington, D.C.; con- tract SO241032, Michigan Tech. Univ. Available from National Technical Information Service, Springfield, Va., PB 288 231/AS. ^Third and fourth works cited in footnote 2. ^Computer Sciences Corp. Computer Simulation of Ventilation Systems Under the Influence of Mine Fires. Program Users Manual and Program Main- tenance Manual. 1980; prepared for the Bureau of Mines under GSA contract GS-045-22715. Greuer, R. E. Influence of Mine Fires on the Ventilation of Underground Mines. BuMlnes Open File Rept. 74-73, July 1973, 173 pp.; available from National Technical Information Service, Springfield, Va. , PB 225 834/AS. (See footnote 3 for open file Inspection facilities.) . Precalculation of the Effect of Fires on Ventilation Systems of Mines. Rept. to U.S. Bureau of Mines on contract J0285002, 3 v., available for consultation at the Bureau's Pittsburgh (Pa.) Research Center. Work cited in footnote 3. 80 natural ventilation. By use of temperature dependent resistances, that is, resistance corrected for temperature induced air density changes, the airflow rates are maintained as volumetric flow rates based upon standard density. Thermal losses to the airway wall is modelled in a quasi-steady state approxi- mation using a "coefficient of age".^ Temperature changes in inclined shafts produces natural ventilation, while thermal effects in horizontal passageways produce a throttling of the airflow. The program user describes the mine network through an assignment of identification numbers to the airways and junctions. Each airway is categor- ized in one of three ways: (1) a regular airway in which the ventilation adjusts to the pressure change along the airway; (2) a fixed quantity airway that maintains a constant flow rate through the use of a regulator; or (3) an airway with a fan. The airway resistance for a regular airway is specified as part of the input data or is computed in the program using Atkinson's formula^ from the dimensions and friction factor specified for the airway in the input data. The thermal conductivity and diffusivity of the airway wall, as well as the reference density and temperature of the air are specified in the input data. The elevation, temperature, and methane concentration are specified for each junction as required. One or more fans may be included in the mine network description. Each fan is represented by a fan characteristic curve that relates pressure drop to volumetric flow rate. Either a Lagrange Interpolation or a linear least squares fit can be used to represent the fan data. The natural ventilation pressure loss for a mesh is evaluated from a sub- stitution of T^ ^ y Tdz for £ -vdp in the determination of the heat converted to work per unit weight of air, where p is the pressure, V the volume, T the temperature, T^ is the average absolute temperature in the mesh under consid- eration, and z is the elevation. This substitution shows that the natural ventilation is induced by thetrmal variation along ascensional airways. Sources of contamination, such as methane emission from a longwall or heat and smoke production from a fire, are entered as input data to the pro- gram. A methane source can be specified either as a volumetric production rate in a passageway or as a measured concentration at junctions from which the program will compute the production rate associated with the airway. In the network computation a perfect mixing of airflows from different airways is assumed to occur at each junction. Similarly, a perfect mixing of the enthalpy is assumed to occur at each junction. A fire in a mine network is simulated in the program in one of three ways: (1) a specified heat production and volumetric flow rate of contam- inant; (2) an oxygen-rich fire with a specified oxygen concentration associ- ated with the fumes that leave the fire zone and a heat generation rate of ^Pages 38-42 of work cited in footnote 3. Pages 41-45 of second work cited in footnote 5. ^Hartman, H. L. Mine Ventilation and Air Conditioning. John Wiley and Sons, New York, 1961, 398 pp. i 'I 81 437 Btu/ft-' of consumed oxygen; or (3) a fuel-rich fire with a specified con- taminant and heat production per cubic foot of oxygen delivered. The capability to include thermal sources and contaminants in the ventilation model as found in the MTU program is a significant advance in simulating ventilation in a mine. Although Wang and Saperstein^ include natu- ral ventilation in their program, it is included as an imposed condition, whereas the MTU program derives the natural ventilation from the temperature changes in an inclined airway, A ventilation program, TVENT, developed by Duerre^ although only applicable to flow at ambient temperature, does include the capability to incorporate transient pressure disturbances introduced at a surface entry. Anderson and Dvorkln ^ ^ performed a network calculation for a mine with 30 airways and three fans using the program of Wang and Saper- stein. ' ^ Computions were made at the Bureau's Pittsburgh Research Center with the MTU program and TVENT ^ 2 f^^ ^he same network. The ventilation flow and pressure changes in the airways as calculated by the three programs was in very good agreement. The MTU program computes thermally generated flow changes in a quasi- steady manner through an iterative procedure. The nonsteady thermal heat flow into an airway wall is treated as a steady process by introducing a dimension- less scale for the thickness of the insulating rock layer, a "coefficient of age." This method, as described by Grever,^^ assumes that steady-state con- vective heat transfer between the air temperature and a rock reservoir tem- perature balances the thermal diffusion into the airway wall from the wall surface. The reservoir rock temperature occurs at a depth in the wall that is inversely proportional to the "coefficient of age." The temperature along the airway follows an exponential distribution between the junctions, which is used in the calculation of the natural ventilation. A convergent iterative procedure solves for the ventilation flow that is coupled with the airway tem- peratures. Once the ventilation is established for a given temperature dis- tribution, the associated steady-state contaminant distribution is established. The program output supplies the user with airflow rates in the airways and pressure changes along each airway. At each junction the temperature as well as the smoke and/or methane concentration is specified. For use of the program in developing strategies for miner rescue, the program identifies those airwaythat are critical (unsafe) for miner survival. For example an airway that satisfies at least one of the following criteria be identified as a critical airway: A temperature greater than 95° F; a smoke concentra- tion greater than 0.05 pet; a methane concentration greater than 1 pet; or a ^Fourth work cited in footnote 2. ^Duerre, K. H. , R. W. Andrae, and W. S. Gregory. TVENT — A Computer Program for Analysis of Tornado-Induced Transients in Ventilation Systems. Los Alamos, N. Mex. , LA-7397-M, 1978. ^^Anderson, T. C, and D. Dvorkin. Mine Ventilation and Digital Simulation and Analysis Capabilities of MESA's Denver Technical Support Center. ^ ^Fourth work cited in footnote 2, ^ ^Work cited in footnote 9. ^■^Pages 41-46 of second work cited in footnote 5. 82 pressure loss less than 0.01 inch WG. The condition for a small pressure change relates to a poorly ventilated airway, while an excessive methane con- centration could result in a partially confined explosion. The user estab- lishes as part of the input data the critical values for each of these variables. APPLICATIONS Small-scale fire tests have been performed by Lee^^ at the Bureau's Pittsburgh Research Center. The ventilation air was observed to be throttled by the fuel-rich fire with a reduction in flow velocity and an increase in the pressure drop across the fire zone. The MTU program was used to simulate the fire experiments. Resistances for the horizontal ducts were determined from Lee's cold flow data. ^ ^ Because of the relatively short lengths of the ducts, less than 3 m, shock losses at bends in the duct were included through at effective length in Atkinson's resistance formula.^'' Modifications were made to the program to account for heat loss from the metal ducts that connect to the wood lined fire zone, and for heat loss to a heat exchanger. Lee ^ ^ depicts a change in ventilation flow from 210 g sec" ^ with no fire to 175 g sec" ^ with a fire and a 50 pet fan setting. In the computer simulation, the fire reduced the flow rate from 207 g sec" ^ to 125 g sec"^. Quantitative agreement most likely would be improved if more accurate information was available regarding the resistance of the gate in the vertical shaft and the heat exchanger, as well as the heat loss to the duct wall and the heat exchanger. A potential hazard in ventilation of a mine is a fan failure. It is important for ventilation and rescue planning to predict the change in methane concentration and pressure drops in airways as a result of a fan failure. A small multilevel, metal or nonmetal, mine with single entries and longwall faces used by Greuer ^ ^ for illustrating applications of the program, including the one presented here, is shown in figure 1. The mine is considered to be gassy with methane. Fans are located in airways 6 and 51; airway 43 is a fixed quantity airway. Methane emission is specified as a volume production rate in airways 9, 17, 33, 34, and 36. At junctions 5 through 16, 19, 21, 22, and 24-28 the methane concentration is specified. T/Jhen the methane production rate is not specified, the program determines the methane production rate in an airway from the methane concentration specified at the airway junctions. This latter quantity is readily accessible to measurement. There is a small amount of natural ventilation in the network due to temperature variations l^Lee, C. K. , R. F, Chaiken, and J, M, Singer. Interaction Between Duct Fires and Ventilation Flow. An Experimental Study. Combustion Sci. and Technol., v. 20, 1979, pp. 59-72, Lee, C. K. , R. F, Chaiken, J, M, Singer, and M, E. Harris. Behavior of Wood Fires in Model Tunnels Under Forced Ventilation Flow. Tests With Untreated Wood. BuMines RI 8450, 1980, 58 pp. ^ ^Table 4 of first work cited in footnote 14. ^^Pages 99-100 of work cited in footnote 7. ^ ^Figure 5 of first work cited in footnote 14. ^^Work cited in footnote 3. 83 KEY Vertical shaft Airway Junctbn Surface junction Longwall face Airways with airflow reversal Methane concentration in junctions 0.8 before ] main fan failure i.o after i FIGURE 1. Concentration changes at junctions and airflow reversal in a multilevel mine resulting from a fan failure. Data are in percent. 84 at the junctions. The fan in airway 51 exerts major control over the ventila- tion, while the booster fan in airway 6 provides minor control over the venti- lation. Figure 1 shows the methane concentrations at junctions before and 15 minutes after the failure of the main fan in airway 51, as well as those 20 airways in which flow reversal occurs as a result of the fan failure. The flow reversal in airways 2 and 24 result in a recirculation of contaminated air throughout the mine via airway 5. The methane concentration generally increases at the junctions after the event. An additional calculation for the case of a failure in both fans resulted in a flow reversal in airways 5, 20, and 24 and in a substantial change in the methane concentration at several of the junctions. In particular, the concentration at junction 7 increased to 5.8 pet and at junction 14 increased to 3.1 pet due primarily to flow reversal in airway 20. With the failure of the main fan only, airways 12, 14, 15 were critical airways with concentrations greater than 1 percent. However, with the failure of both fans these airways were noncritical. This illustrates the complex interaction that can occur in mine network as a result of forced and natural ventilation. The failure of the main fan, with continual operation of the booster fan, produced a flow reversal in the vertical shaft designated airway 22. Prior to the failure of the main fan, the pressure drop along airway 22 was 0.87 inch WG with an ascensional flow of 4,175 cfm. After the fan failure the pressure drop along airway 22 is 0.001 inch WG with a decensional flow of 159 cfm. Compared to the intake flow of 50,000 cfm in the vertical shaft designated airway no. 1, a flow rate of 159 cfm is a relatively small quantity. When pressure differences along an airway are less than 0.01 inch WG, the resultant program computation must be viewed with caution, since a small change in air- way temperature or ventilation elsewhere in the network can readily alter the flow in the airway. With respect to airway 22, there is a positive tempera- ture gradient in the vertical upward direction that favors a stable vertical stratification. The reversed flow that occurs in the shaft after the main fan failure has the effect of drawing warm air vertically downward which is opposed by the buoyant tendency of warm to rise. This opposes the stability of the air in the shaft. The result of the small descentional velocity is that convergence of the iterative process can appear to require an excessive number of iterations, whereas an actual increase in the number of iterations produces no significant changes in the resultant methane concentrations and ventilation flows. To demonstrate this, the maximum number of iterations per- mitted was increased from 10^ to 10"* in the ventilation calculation following the main fan failure. At no junction did the methane concentration change by more than 10" ^ percent from the result with 10 iterations. The most severe change in ventilation occurred in airway 22, where a ventilation flow of 145 cfm resulted from 10^ iterations compared with 159 cfm for 10 iterations. This demonstrates that, although the solution can be further refined, there is no advantage to do so from the practical limitations involved in determining methane concentrations and ventilation flows. In fact, those airways with airflow rates of at least 10^ cfm exhibited no more than a 1-pct change in flow rate when the maximum flow rate was increased to 10^. It is important to recognize that airways, in particular vertical shafts, along which the pressure change is less than 0.01 inch WG are critical with respect to ventilation. 85 Current developments in the NTU computer program under contract J0285002 include a real-time calculation of contaminant concentrations in the network for a steady state ventilation flow. In the early stage of fire development, when the heat output is small, it is quite possible that the flow will remain steady while the contaminants are transported by the ventilation currents throughout the mine network. The program user can predict with a real-time analysis transitions in airways from safe to unsafe, as well as the reverse transition. The model accounts for the mixing of contaminated air at a junc- tion with uncontaminated air, or with contaminated air of equal, less than, or greater concentration. The real time progression of the contaminants is mod- elled through a selection of control volumes of homogeneous contaminant con- centrations and simultaneously advancing each control volume in a fixed time increment. At each junction control volumes of different concentrations merge into a control volume of a new concentration. The real-time analysis can be extended over a period of time until a steady-state concentration has developed. DISCUSSION The MTU mine ventilation computer program has the predictive capability to determine ventilation flow with thermally generated natural convection. Hazardous events, such as normal methane emission, a fuel-rich or fuel-lean fire, introduction of contaminants into an airway, or a failure of one or more fans, can be simulated with the program. The program user can identify those airways in which critically high methane or smoke concentrations, high temper- atures, or small pressure changes exist. This information should be useful in the development of strategies for miner escape and rescue as well as ventila- tion planning. A recent extension of the program to include a real-time cal- culation of the contaminant distribution will enable the user to predict both the immediate contaminant concentration as well as the cumulative exposure at any location within the mine. The computer program is available from the U.S. Bureau's Pittsburgh Research Center, 86 AN EXPERIMENTAL INVESTIGATION OF THE FIRE HAZARDS ASSOCIATED WITH TIMBER SETS IN MINES by Archibald Tewarson^ and J. S. Newman ^ ABSTRACT Fire hazards associated with treated and untreated timber sets were evaluated, using a large-scale fire test gallery. Criteria are presented for (1) "fire propagation" and "no-fire propagation" for the timber sets used for structural support of mines and (2) fire endurance of plywood bulkheads. Hazards due to generation of "smoke" and toxic products in timber set fires are discussed. A comparison of data for laboratory and large-scale fire tests is presented for examining the reliability of the laboratory-scale test method for the fire hazard evaluation of mine materials. INTRODUCTION The hazards associated with a fire are due to generation of heat, toxic products, and "smoke" (reduction in visibility), which, in turn, depend on the extent of fire propagation, generation of fuel vapors, and types of fuel vapors, fuel loading, ventilation, ignition source strength, fuel treatment, fire size and location, etc. In order to investigate the fire hazards associ- ated with timber sets in metal and nonmetal mines, an experimental program was undertaken at the Factory Mutual Research Corp. (FMRC) in Norwood, Mass. , for the Bureau of Mines. Emphasis was placed on determining the maximum loading of timber sets in a passageway that would not propagate a fire. A large-scale fire test gallery was constructed at FMRC's Test Center at West Glocester, R.I., in 1976. •^ The gallery walls consisted of an outer cement-block facing, to which an insulated blanket was attached. The inner side of the insulation blanket material was faced with refractory block. The roof was fabricated from a castable refractory material. Figures 1 and 2 show an aerial view and a schematic of the mine gallery. The gallery is a T-shaped structure with two passageways (drifts) each about 47 m long and about 5.9 m^ in cross-sectional area. The north and east drifts are horizontal, and the west drift is partially sloped at 12.5°. Ventilation air is normally directed into the north drift by a fan at the north portal. The west portal is generally closed so that airflow is exhausted only through the east drift. ^Factory Mutual Research Corp., Norwood, Mass. ^Buckley, J. L. , D. B. Heard, P. A. Groce, and B. G. Vincent. Design, Con- struction, and Operation of Mine Fire Test Gallery. Factory Mutual Research Corp., Norwood, Mass., Tech. Rept. RC76-T-79, Serial 22501, July, 1977, 105 pp. 87 FIGURE 1. - Aerial view of the simulated mine gallery (looking south from the north portal). East portal ^ 1 jgstdntt^-^^^^ p Onr''^ Timber sets Source fire Bulktiead North portal Om Fan section ^a Inlet orifice FIGURE 2. - Schematic of the simulated mine gallery, showing location of data stations and source fire. 88 To photo 1—^ cell -^ 0.76 m ► stainless steel sampling duct-|^ (0.10 m ID) ^ To light source To electronic manometer To temperature and product sampling instruments Exhaust collection cone (aluminum foil) Radiant heaters FIGURE 3. - Factory Mutual's small-scale combustibility apparatus. In the program, fire tests have been performed in the north drift. There were nine measurement stations in the gallery (N5, N9, Nil, N13, N15, N17, N19, E-1 and bulkhead, fig. 2). A total of about 140 data channels were dis- tributed among these stations for measuring concentrations of CO, CO 2> H2> gaseous hydrocarbons. oxygen, temperature, pressure, velocity, and heat flux. Timber sets (Douglas-fir), about 0.15 m X 0.15 m and 0.31 x 0.31 m in dimension, were used on the walls and roof to provide the fuel load- ing in the tests. The standard "source fire" used for igniting the timber sets consisted of 4 to 44 vertical wood boards (each board was about 0.05 X 0.20 X 2.4 m long) and two pre- mixed gas burners positioned in the north drift, as shown in fig- ure 2. Heptane pool fires were used to calibrate the standard "source fire." II 89 Ventilation to the gallery was provided by an adjustable-speed fan, located inside a housing, attached to the gallery at the north portal (fig. 2). Other details of the gallery and test procedures are described at length. Buckley,^ Croce,'^ and Lee. ^ In addition to the large-scale fire tests in the gallery, laboratory- scale fire tests were also performed in the Factory Mutual (FM) small-scale combustibility apparatus (fig. 3),^ In these tests, treated and untreated, dry red oak and Douglas fir specimens, about 0.01 to 0,02 m thick and about 0.006 m^ in area, were used. The specimens were exposed to external heat flux in the range of 31 to 71 KW/m^, and with ventilation air velocity in the range of about 0.01 to 0,68 m/s. In the tests, measurements were made for ignition energy, fuel mass generation rate, heat generation rate, product generation rates (CO2, CO, gaseous hydrocarbons, and "smoke"), oxygen depletion rate, and light obscuration. The objective of the laboratory-scale fire tests to develop a reliable small-scale test for the fire hazard evaluation of mine minerals. EXPERIMENTAL RESULTS Large-scale fire tests were performed in the gallery for the untreated, NCX-treated and intumescent-paint coated timber sets and about 15.9-mm-thick plywood bulkheads. Laboratory-scale tests, in the FM small-scale combustibil- ity apparatus, were performed for untreated, NCX-treated and intumescent paint coated red oak and Douglas-fir specimens. (In addition Fyrepruf treated and precharred specimens were also examined. ) In both large-scale and laboratory- scale tests, measurements were made for defining fire propagation, generation ^Work cited in footnote 2. '*Croce, P. A., J. L. Buckley, B. G. Vincent, and D. B. Heard. Full-Scale Investigation of the Fuel-Load Hazard of Timber Sets in Mines. Factory Mutual Research Corp., Norwood, Mass., Tech. Rept. RC78-T-6, Serial 22501, October 1978, 30 pp. BuMines Open File Rept. 85-79, October 1978, 39 pp.; available for inspection at Bureau of Mines facilities in Denver, Colo. , Twin Cities, Minn., Pittsburgh, Pa., and Spokane, Wash., and the National Library for Natural Resources, U.S. Dept. of the Interior, Washington, D.C.; available from National Technical Information Service, Springfield, Va., PB 299 948/AS; contract H0252085. Croce, P. A., C. K. Lee, and J. S, Newman, An Experimental Investigation of the Fire Hazards Associated With Timber Sets in Mines, Factory Mutual Research Corp., Norwood, Mass., Tech. Rept. 22501, RC80-T-64, July 1980, 134 pp.; 3d ann rept, to U,S, Bureau of Mines on contract H0252085, ^Lee, C, K, , P. A, Croce, and J, S, Newman, An Experimental Investigation of the Fire Hazards Associated With Timber Sets in Mines, Factory Mutual Research Corp,, Norwood, Mass,, Tech, Rept, RC80-T-27 , J, I, Oeoni, RA, March 1981, 108 pp, ^Tewarson, A, Fire Hazard Evaluation of Mine Materials, Factory Mutual Research Corp, Norwood, Mass,, Tech, Rept, RC80-T-77, J, I, 0E0N7,RC, October 1980, Tewarson, A., J, L, Lee, and R, F, Pion, Fuel Parameters for Evaluation of the Fire Hazard of Red Oak, Factory Mutual Research Corp, , Norwood, Mass, , Tech, Rept, RC79-T-68, J, I, 0C3R1,RC, December 1979, HI pp. 90 of fuel vapors, generation of heat, generation of "smoke" and other products (carbon monoxide, carbon dioxide, gaseous hydrocarbon), light obscuration, and oxygen depletion. In the large-scale fire tests, measurements were also made to establish the fire endurance criteria for the bulkheads. Fire Propagation In Timber Sets In the large-scale fire tests for timber sets, "fire propagation" is defined^ as flaming combustion, spreading from the ignition zone to the end of the north drift (that is, to the Intersection of the north and east drift in fig. 2), so that all the timber sets are involved. "No fire propagation" is defined^ as flaming combustion, unable to spread from the Ignition zone to the end of the north drift. "Fire propagation" in the large-scale fire tests of timber sets, in the gallery, was found^ to depend on (1) timber loading density (a), defined as the ratio of exposed timber surface area to the area of the gallery walls and celling; (2) timber size; (3) "source fire" heat output, defined in terms of the number of wood boards (n,^^)' C'^) ventilation air velocity (Va); and (5) timber treatment. Figure 4 shows test data for 20, 40, and 80 pet timber loading densities, where the number of wood boards, n^^^ (that is, "source fire" heat output) and ventilation air velocity, Vg, are plotted to separate the "fire propagation" and "no fire propagation" regions. For each timber loading density, a "boundary" separates the two regions of "fire propagation" and "no fire propagation." This "boundary" is replotted in figure 5A which can be used, conservatively, to determine the probability of fire propagation in the untreated timbers, as used for structural support in mines (that is, setting a maximum timber loading density, a maximum "source fire" strength, or a minimum ventilation flow that will reduce the probability of fire propagation).^*^ The "boundary" criteria for NCX-treated and untreated timber are shown in figure 5B. A more Intense "source fire" is required for "fire propagation" for NCX-treated timber sets than for untreated timber sets, or NCX-treatment is effective in reducing the probability of fire propagation in the timber sets. From the laboratory-scale fire tests, in the FM small-scale combusti- bility apparatus, it was found that the energy required to produce the com- bustible vapor-air mixture near the surface for NCX-treated Douglas-fir was about three times the value for untreated fir; ^ ^ that is, more intense heating is required for fire propagation for NCX-treated Douglas-fir than for the untreated fir, in agreement with the large-scale fire test data. 'Second work cited in footnote 4. ^Second work cited in footnote 4. ^Second work cited in footnote 4; work cited in footnote 5. ^ ^Second work cited in footnote 4. ^ ^Second work cited in footnote 6. 91 Vertical board array • Lagged ignition zone WZZ2^ Fire propagation TD Boundary zone 1^3 No fire propagation a Timber loading density 0.5 1.0 1.5 2.0 2.5 AIR VELOCITY (V^), m/s FIGURE 4. - Fire protection criteria for timber sets-large-scale fire tests. A, 20% timber loading; B, 40% timber loading; C, 80% timber loading. 92 1.0 1.5 2.0 2.5 AIR VELOCITY (Vg), m/s 1.0 1.5 2.0 2.5 AIR VELOCITY (Vo), m/s FIGURE 5. - Boundary criteria. A, at three timber loading densities; B, for untreated treated timber at 80% timber loading density. NCX- KEY > B C Standard source fire (wood boards and burners) Heptane pool fire Burnthrougti of bulkhead Charring of bulkhead ^ ^ B 23^ - Ah 1.49pr^6 Burnthrough / C / + 28 No burnthrough + 44 1.49D + 36 + 28 _ / + 20 / 0.84 a + 20 - b/ ^16 + 20 + 16 + 8 t T "wb "wb 1 1 \ "wb 0,5 1,0 1,5 2,0 2 5 AIR VELOCITY (Va). m/s FIGURE 6. - Criteria for burnthrough of wood bulkhead as a func- tion of fire heat output and ventilation airvelocity. 93 » Bulkhead Fire Endurance Plywood bulkheads, 15.9 mm thick, were tested for fire endurance under convective and radiative heating. Bulkheads were approximately 30 m from the source fire (fig. 2), Figure 6, adapted from Lee,^-^ shows the data for the fire endurance of the bulkheads. For each value of ventilation air velocity, Vg, there is a critical maximum heat output, Qg ^^^ beyond which a plywood bulkhead will be burned through. For example, for Vg = 1.5 ra/s, Qg ^lax ^^ about 4.5 MW. The information in figure 6 is useful in specifying the posi- tion of bulkheads relative to a potential source fire in a mine.^^ EVALUATION OF HAZARD ASSOCIATED WITH TIMBER SET FIRES Fire hazard for humans is associated with the critical limits of heat, toxic products, and light obscuration for human escape. Table 1 lists some "tentative" critical values for human escape; that is, values where human physiological responses are not altered to a significant extent or where human exposure to fire products is tolerable for a short time. TABLE 1. - Tentative critical values for human escape from mines ^ Value Fire product: HCN ppm. . 30 HCl ppm. . 50-100 Benzene ppm. . 1 , 500-4 , 000 CO ppm. . 1 , 500-4,000 CO 2 ppm. . 40 , 000-80 , 000 CO2 vol-pct. . 4-8 O2. ppm. . 60,000-100,000 O2 vol-pct. . 6-10 Temperature ° C. 120 Do ° F.. 250 Smoke m' K , 0. 218 Do ft-1.. 0.066 'Tewarson, A. The Effects of Fire Exposed Elec- trical Wiring Systems on Escape Potential From Buildings. Part I. A Literature Review of Pyrolysis/Combustion Products and Toxicities — PVC. Factory Mutual Research Corp. , Norwood, Mass., Tech. Rept. RC75-T-47, Serial 22491, December 1975. . Fire Toxicology — A Literature Review for Polyvinyl Chloride. Factory Mutual Research Corp., Norwood, Mass., Tech. Rept. RC79-T-41, J.I. 0C1R9, August 1979. '^Work cited in footnote 5. ^ ^Work cited in footnote 5. 94 By measuring or calculating the mass generation rate of fuel vapors, heat, "smoke" and toxic products, and light obscuration and using the "tenta- tive" critical values, the hazard of the fire environment can be estimated for human escape for defined fire scenarios in mines. In the large-scale fire tests for timber and in the laboratory-scale fire tests, measurements were made for fuel mass generation rates, generation rates j of heat and products, and light obscuration (only in the laboratory-scale tests) in order to evaluate the hazards associated with treated and untreated timber sets. Fuel Mass Generation Rate Figure 7 shows the data for fuel mass generation rate per unit fuel area plotted against the total heat flux received by the fuel. Good agreement between the data from the large-scale and laboratory-scale fire tests suggests that the FM small-scale combustibility apparatus is useful for mine material evaluation. Data in figure 7 also suggest that the higher the value of total heat flux, the higher is the fuel mass generation rate. Variations in total heat flux are due to a combination of factors such as timber loading density, "source fire" heat output, ventilation air velocity, and timber treatment or 20.0 15.0 10.0 5.0 KEY ADO Large-scale galley Small-scale tunnel • Laboratory scale 140 TOTAL HEAT FLUX, KW/m^ FIGURE 7. - Fuel mass generation rate versus total heat flux. 95 imposed external heat flux. Figure 8 shows an example of the effectiveness of the NCX treatment and intumescent paint coating of the timber sets in reduc- ing the fuel mass generation rate in the large-scale timber set fire tests. Laboratory-scale fire tests also show the effectiveness of the treatments in reducing the fuel mass generation rate as shown in figure 9. A comparison of the data in figures 8 and 9 indicates the usefulness of the FM small-scale combustibility apparatus as a test for mine materials. Generating Rates of Heat, "Smoke," and Toxic Products The generation rates of heat, "smoke," and toxic products depend on the fuel mass generation rate and also on the amount of air available for com- bustion; a reduction in the fuel mass generation rate is expected to be ben- eficial in reducing the fire hazard. Figure 10 is an example showing the variation of the mass ratio of air to fuel, that is, mass ventilation airflow rate to fuel mass generation rate with time, for untreated timber sets in the large-scale fire tests. A ratio >6 Indicates that combustion of the timber sets is in excess air (fuel lean). All timber sets in this study had fuel- lean mass ratios, A ratio <6 would indicate combustion of the timber sets in deficient air (fuel rich). Vg = 1 5m/s n^t,=20 cr=80% KEY o Untreated o Intumescent ■ pant coated NCX-treated 5 10 15 20 25 3 FIGURE 8. - Fuel time, showi and intume scale tests. TIME (t). mm mass generation rate versus ng effects of NCX treatment scent-paint coating, large- 20.0 15.0 10.0 5.0 KEY Red oak o Untreated o Fyrepruf treated • Intunriescent-paint coated Douglas-fir '^ Untreated _L_ 10.0 I 20.0 NCX treated J I 30.0 40.0 50.0 60.0 EXTERNAL HEAT FLUX, kW/m^ 70.0 80.0 FIGURE 9. - Fuel mass generation rate versu; external heat flux, laboratory-scale tests, for red oak and Douglas-fir. 96 I I A a =20% 1 1 1 V3 m s g — -^^:»— 5 R ~ _0. 1 5 i / - ..0-- 30 ■" ^ - ; p- #- <9 - -sC^ ll - I 1 1 1 1 1 1 B a =40% Va m s _ - -C- - 1 5 ■••Q- 30 1 1 1 /7- 1 1 1/ , , ' 160 140 120 •2 100 o ■™ 80 O < 60 CO ^ 40 20 1 1 c 1 ! 1 n^b a = 80"n 1 — A 5 8 (0 30 m'l - -0-15 16 (0 30 m') ■ •O-. 3 26 10 30 m') . ...■•.. 30 20 (0 15 m'l ~ \ '> 1 00 ;>100 '■TBurners off _ ■'Tor TS 123? - \y\J'' ^.'■■■* " 'f^n / **• \*' /' / \ P"^. J I- -^ y \ 1 1 I 1 cr 1 1 1 D a= 80% 1 1 _ .05 20INCX treated] _ ...Q.. 1 5 36(NCX treated) ........ 1 5 20 (Untreated) - - P _ p-d** _ 5- .A - 10 15 20 TIME(t), min 25 30 10 15 20 25 30 TIME (t), min FIGURE 10. - Mass ratio of air to fuel as a function of time for the large-scale fires of timber sets. Heat Generation Rate Variations in the ratio of heat release rate to fuel mass generation rate with the mass air-to-fuel ratio are shown for wood in figure 11. For a mass air-to-fuel ratio <6, that is, fuel-rich combustion, the ratio of heat gener- ation rate to fuel mass generation rate decreases. Data in figure 11 show that combustion in most of the large-scale fire tests, for untreated timber sets, was fuel-lean. In figure 11, a reasonable agreement can be noted for the large-scale and laboratory-scale fire test data. f 20.0 15.0 10.0 5.0 15 30 45 MASS RATIO, air to fuel 60 97 ' ' 1 ' 1 1 1 I 1 1 1 1 1 1 - - - - D - A A O G* -o / ••- ' ° A - - < 200 j,^ V3 = 1.5m/s - I 1 I 1 - n^l^=20 o- = 80% . KEY rs Untreated — o- i-»h p/ \ Intumescent- y-*-*^\\ / V paint coated --^- i-ah / \ NCX-treated — n— (') / f^ ^^^^^■-^ o 1/ y- -°— -o— -o— o. \ / ^2.8 MW from source fire. ^'^"^^-a 1 1 1 1 1 1 FIGURE 13. 35 -30 25 20 15 10 10 15 20 TIME (t), min 25 30 35 Average gas temperature and heat generation rate versus time, showing benefici effects of intumescent paint and NCX. al Generation Rates of Fire Products Variations in the ratio of mass generation rates of CO2 and CO to fuel mass generation rate with the mass air-to-fuel ratio are shown in figure 14, For the mass air-to-fuel ratio less than 6, that is, fuel-rich combustion, the ratio of mass generation rate of CO to fuel mass generation rate increases for CO but decreases for CO2. A reasonable agreement between large-scale and laboratory-scale fire test data can be noted. 100 2.0 "1 ^ T- Carbon dioxide o a o o KEY o A a Large-scale galley © Small-scale tunnel • A ■ Laboratory scale i!R..^^^»ww&ogPo°o°P°° 30 45 MASS RATIO, air to fuel FIGURE 14. - Variations in the ratio of mass generation rates of CO and CO2 to fuel mass generation rate as a function of mass air-to-fuel ratio. 75 Figure 15 shows some large-scale fire test data for untreated timber sets, where human physiological responses to CO are also included. The data in figure 15 show that, near the measurement section, the environment was haz- ardous for human escape. The hazard can be reduced by fresh air mine venti- lation downstream of the fire and by a combination of factors such as timber loading density, "source fire" heat output, ventilation air velocity, timber treatment, etc. Figure 16 shows an example of the beneficial effects of NCX treatment and intumescent paint coating of timber sets in reducing the ratio of generation rate of CO to fuel mass generation rate. Note that fuel mass generation rate is reduced substantially by the treatments (fig, 8); thus, the generation rate of CO is also reduced substantially. This beneficial effect of the treatment is also found in the laboratory-scale fire tests, as listed in table 2, columns 3 and 4, 101 UJ < < Z LL Qo < 15.0 10 5,0 KEY Air velocity. Timber loading, pc m/s 20 40 80 0,5 ■ • A 1.5 D A 3.0 Q © A AA A A Marked poisoning ©GO • lew minutes tolerance 00 O O «P A O O A • •• O -£i A n_£^ _„ . ± A 03 ^ o ^aDG»0aDc5^ aye* ^ a^^ ^^ FIGURE 15. - Critical value of CO for human escape as a func- tion of mass air-to-fuel ratio, large-scale gal- lery fire test data for tim- ber. Critical valueof CO for human escape is 2,750 ppm. 12 24 36 48 60 MASS RATIO, air to fuel z O CC t5 < Q. O ^ is o| LL O COS CO < 5 Va = 1.5m/s 'wb 20 a =80% KEY o Untreated A Intumescent-paint coated, a NCX-treated 15 20 TIME (t), min 35 FIGURE 16. - Mass fraction of CO versus time, showing beneficial effects of intumescent paint and NCX. 102 S2 Q. -1 < O ^ LU < ^ > 3 _l I < CC y O b LL O < < O 5.0|-»-r 4.0 3.0 2.0 1.0 o o Air velocity m s 0.5 1.5 3.0 KEY Timber loading, pet 20 40 80 Human escape possible Light Obscuration by "Smoke" Although light obscuration was not measured in the larger scale fire tests of untreated timber sets, obscuration data from the laboratory-scale fire tests, conducted in the FM small-scale combustibility appa- ratus,^^ were used to estimate light obscuration in the large- scale fire tests on timber sets. If the light obscuration is less than the critical value required for visibility, human escape from a fire is possible. Figure 17 shows the estimated data for the larger scale fire tests on untreated timber sets, in terms of the fraction of cri- tical value of light obscuration for human escape. Near the mea- surement section, in some tests, the obscuration was less than the critical value, and in some tests it was greater than the critical value. The reduction in "smoke" formation, for exam- ple, by timber treatment or decrease in light obscuration by increased mine ventilation may be helpful. Table 2, column 5, shows that wood treatment is helpful in reduc- ing "smoke" formation. Generation Rate of Other Toxic Species In the large-scale fire tests, HCN was measured for the NCX-treated tim- ber sets.^^ Hydrogen cyanide (HCN) is a toxic species that can influence the human physiological response, affecting human excape from fires. Column 6 of table 2 lists data for the maximum amount of HCN that can be expected per unit amount of fuel mass generated from the treated and untreated wood specimens. ^^ It appears that, although the treatments of the timber sets are effective in reducing the fire propagation hazard and hazard due to CO, light obscuration, and heat; the hazard due to HCN may be present depending on the fire condi- tions and extent of treatment. 18 36 MASS RATIO, air to fuel FIGURE 17. - Light obscuration and human escape-large- scalefires. Critical optical density per unit path length for human escape is 0.218 m" \ ^ ^Second work cited in footnote 6. ^ ^Work cited in footnote 5. ^^First work cited in footnote 6. 103 SUMMARY A highly instrumented, large-scale fire test gallery has been used to establish the fire hazards of timber sets in mines. "Fire propagation" cri- teria for the timber sets have been proposed, in terms of timber loading den- sity, ventilation (or air velocity), source fire heat output, and timber treatment (fig. 4). A criterion has also been proposed for the fire endur- ance of bulkheads (fig. 6). From the large-scale and laboratory-scale fire tests, the effectiveness of the NCX-treatment and intumescent paint coating of the timber sets has been shown in terms of reduced fire propagation and decrease in the generation of fuel, heat, CO, and "smoke," The influence of other species, such as HCN on the environment, however, has not been established. The usefulness of the FM small-scale combustibility appratus for small- scale fire testing of mine materials is suggested. From the large-scale fire test data, some preliminary estimates have been made for the human escape potential from specific locations and fire scenarios in the gallery tests. The large-scale fire test gallery has been calibrated and tested success- fully for repeatibility of data. The gallery, thus, can be used for solving numerous fire problems that may be present in coal as well as metal and non- metal mines. The FM small-scale combustibility apparatus has also been calibrated and tested successfully for large-scale fire conditions. A simplified version of the apparatus and test procedures, thus, can be used for testing the fire haz- ard of mine materials. In addition, the large-scale fire test gallery, in combination with the FM small-scale combustibility apparatus, offers experimental facilities for establishing criteria for human escape from mine fires. 104 UNDERGROUND FUELITO AREA FIRE PROTECTION SYSTEMS by Guy A, Johnson^ INTRODUCTION Because of the increasing amount of mechanisation in today's mines, and because this trend will continue as the industry strives for more productiv- ity, increasing amounts of combustibles are required underground. Fuel, lub- ricants, and solvents are a few examples. Because miners have taken many precautions, industry has until now had no major fires in a fueling area des- pite the tremendous danger that these combustibles present. But because even a moment of carelessness can be extremely dangerous, the Bureau has sought to develop a system to protect against an underground fueling area fire before such a major fire occurs, rather than wait to react to such a disaster. To develop this program, the Bureau worked with numerous different com- panies that currently make surface-fueling rack fire protection systems. The idea was not to "rediscover the wheel," but to investigate available technol- ogy and see if it could be ruggedized to make it applicable to the underground fire problem. And, indeed it can. SYSTEM DESIGN Mines have recognized the fueling area fire problem and have installed numerous "do it yourself" systems, most of which are manually activated. The Bureau's goal was to develop a system that would be automatic and could be modified to fit most mine fire situations. So that the results will be more attractive to industry, the Bureau places a high priority on keeping the cost of its fire protection systems as low as practical. The Bureau strongly recommends that major fuel storage be on the surface and that only smaller amounts be stored underground. This is not always pos- sible, but it is the first step to good underground fueling area fire protec- tion. The second step involves two typical situations underground. One is a situation in which fuel is stored behind bulkheads. The Bureau has developed systems for this situation. The second situation, which is possibly more dan- gerous, is the fuel transfer area where the mine vehicles come to refuel. This location has the heat of diesel engines as an ignition source. There- fore, there is a need for a very quick response to sense a fire and extinguish it. Working with the Ansul Corp. (under contract HO262023), which has been building very fast response fire systems for many years. Bureau researchers ^Supervisory mining engineer. Twin Cities Research Center, Bureau of Mines, Minneapolis, Minn. 105 chose optical detectors for the system. These detectors are blind to most kinds of light other than fire and are rugged enough for mine conditions. The fire suppressant used is a combination of agents used at airports in case of a plane crash. The suppressant contains dry chemical, which has throwing power and knockdown power, and aqueous film forming foam (AFFF) agents that cool and smother the fire. These two agents are more expensive than water sprin- klers, but they are very efficient in quickly extinguishing a fire, and that is what is really needed. This technology is currently available, and the cost of the total system can be kept to a minimum. The Bureau built a mockup of an engine area of a load-haul-dump (LHD) at the contractor's fire field in northern Wisconsin and set many test fires to measure the type of response needed to control a fire in a refueling area. Although the heat, thus the energy, is tremendous, the dry chemical and AFFF are equal to the task. IN-MINE FIRE TESTIMG Following our normal technique for developing prototype hardware, the Bureau worked with Union Carbide's Pine Creek tungsten mine near Bishop, Calif., to install the prototype hardware in one of the lower levels of the mine (fig. 1). The system's control panel and backup battery power panel (fig. 2) were installed in 1977, as well as the optical detectors (fig. 3) and suppressant piping (fig. 4). Note that the fire detectors were "crossed zone," This means that if someone lights a match near the fire area, even though nobody should do this because of the high fire danger, the match would have to be lit for at least 10 sec. , and both sensors would have to see the fire for the system to automatically activate. This greatly increases the system's reliability and reduces misfires to the point where we have not had a misfire with the system in 3 years of testing. After leaving the system in the mine for about 1-1/2 years, we set pan fires in the fueling area to test the system. The system's optics sensed the fire and warned the miners of a fire, and the dry chemical and AFFF extin- guished the fire within 10 sees, A more difficult fire situation was then tested, A fire was set under the LHD mockup vehicle (figs, 5-7). This is a very hard fire to get to and put out. The sensors again detected the fire, and the tremendous throwing power of the dry chemical got under the vehicle and put out the fire. The AFFF agent provided an aqueous-film-forming foam to prevent the reignition of the fire. 106 2 |g ? 1 ' ■ ~' -5 15 i ^ £ s ..-'" ' |l ^-^ / II ^ 1 / / 1 / i 1 1 / a.--».^ / c """^ --^1 1 1 1 1 ' 1 ^ X i^ / 1 i 1 1 1 / ^ \ / \ V. / ^ /^ /' / / \ / \ 1 E / \ ^ 1 / \ 0«5 ,1 / / 4 e / ii ■ 1 « n 3 S C 3 ; «'L#»;,i., l< \ ■ / 107 'i ^ 108 109 mx 1» K. no Ill 112 113 FIGURE 9. - Underground fueling area fire protection system demonstration unit at the Bureau's Twin Cities (Minn.) Research Center. 114 CONCLUSION The hardware is currently availahle from the Ansul Corp. , the Walter Kldde Corp., and MSA. It ranges in price from $12,000 to $20,000. The Bureau is currently conducting follow-on, long-term testing of alternative system designs (fig. 8) at the Pine Creek Mine and at the Boss Mine in Mis- souri, where we are studying a large fuel storage area fire protection system. More information can be obtained in a Bureau contract report. 2 The Bureau has a working hardware demonstration unit at its Twin Cities Research Center (fig. 9). If you would like to see the system function, feel free to visit the Center and evaluate its applicability to your situation. 1 ^Christensen, B. C, and G. R. Reid (The Ansul Corp.). Improved Fire Protec- tion System for Underground Fueling Areas. BuMines Open File Rept. 120-78, 1977, 325 pp.; available for consultation at Bureau facilities in Denver, Colo,, Bruceton and Pittsburgh, Pa., Twin Cities, Minn., and Spokane, Wash.; at the National Mine Health and Safety Academy, Beckley, W, Va.; and at the National Library for Natural Resources, U.S. Dept. of the Interior, Wash- ington, D.C.; available from National Technical Information Service, Springfield, Va., PB 288 298/AS; contract H0262023. 115 FIRE DOORS FOR NONCOAL MINES by Kenneth L. Bickel ^ ABSTRACT During an underground mine fire, controlling the spread of toxic gas and smoke is essential. If a fire were to occur in an area such as a shaft sta- tion or shop, controlling the spread of the fire itself may determine if the miners will have sufficient time to evacuate the mine. In 1978, the Bureau initiated a contract with Unidynamics, Inc., St. Louis, Mo., to evaluate cur- rent mine air and fire door technology and develop a fire door for mine use. The door was to be designed for use in large openings. This paper describes the types of ventilation and fire control doors evaluated, and the design, fabrication, and fire testing of an improved fire door designed specifically for use in large openings in underground noncoal mines, INTRODUCTION In 1978, the Mine Safety and Health Administration (MSHA) was consider- ing a regulation calling for the application of air and fire doors in certain locations of underground noncoal mines where they had not previously been required. With few suppliers making doors for use in large mine openings (cross sectional areas on the order of 120 square feet or larger), and with few mines having experience in designing or fabricating fire doors, it appeared that a lack of appropriate technology might make it Impossible for mines to comply with a regulation requiring installation of fire doors. Therefore, the Bureau of Mines awarded a contract to Unidynamics, Inc., St. Louis, Mo., 2 to evaluate current mine air and fire door technology and to design and test an improved fire door for use in large mine openings, TYPES OF DOORS The study was initiated by determining the types of doors In use in mines. The different types of doors encountered fall into the five follow- ing categories: (A) roll doors, (B) swing doors (panels opening in the same direction), (C) canton doors (swing doors with panels opening in opposite directions), (D) telescoping doors, and (E) sliding doors. Most air doors were power operated. Usually, one or two air cylinders were used to open and close the canton and swing doors. Electric power was used to raise and lower ^Mining engineer, Twin Cities Research Center, Bureau of Mines, Minneapolis, Minn. ^Contract H0282003, Unidynamics, Inc. Evaluation of the Construction of Fire Doors, Air Doors, and Bulkheads in Non-Coal Mines. 116 the roll doors and telescoping doors. The sliding doors were used primarily as manually operated service doors or as gravity closed fire doors (normally in the open position) in storage or maintenance areas. The fire doors encountered were either the roll or sliding type and were normally open. In either case the door was held open by a fusible link that would release the door during a fire, allowing gravity closing of the door. The roll door consists of a series of horizontal metal slats interlocked to allow them to roll around a metal shaft when the door is opened (raised vertically), and yet allow little air leakage between the slats when the door is closed. The door is electrically operated, controlled by pushbutton or lanyard. The swing door has two panels that open in the same direction. Gener- ally, one air cylinder operates one panel. The canton door consists of two panels opening in opposite directions. The panels are connected by a linkage and are operated by one air cylinder. The doors were controlled by a lanyard located away from the door or by the use of photoelectric cells that are trig- gered by a passing vehicle. The telescoping door consists of a series of horizontal, channel-shaped sheet metal sections that nest upon one another. As the lowest section is lifted by two cables being wound on a drum, it moves up into the next section directly above it and lifts it also. This process continues until all sec- tions have been lifted and the door is open. The door is electrically oper- ated, controlled by pushbutton or lanyard. The sliding door has a door panel mounted on a track. The door is gen- erally opened and closed manually. However, the door can be set up on an inclined track so that it can close itself after it is released. Table 1 gives a comparison of various doors, their sizes, type of opera- tion, type of construction material, fire rating, and design pressure ratings that were commercially available at the time the evaluation of the available air and fire doors was conducted. 117 TABLE 1 . - Steel mine door comparison Type of door | Clear opening size (W X H), ft Design-pressure, per square foot AIR CYLINDER OPERATION Swing — 2-panel, opposite opening 13 X 9.5, 12 X 10, 15 X 12 12 X 10, 15 X 12 12 X 10, 15 X 12 ELECTRIC MOTOR OPERATION Roll: Nonlabel. Labeled Titan (roll, heavy duty) Telescoping 12 X 10, 15 X 12 12 X 10, 15 X 12 12 X 10, 15 X 12 12 X 10, 15 X 12 12 X 10, 15 X 12 12 X 10, 15 X 12 12 X 10, 15 X 12 12 X 10, 15 X 12 20 30 20 30 40 20 30 70 '3-hour fire rating; other doors had no fire rating. COMPARISON OF DIFFERENT TYPES OF DOORS In order to determine the best type of door design to use In constructing the fire door, an analysis of each type of door was conducted. The following is a listing of advantages and disadvantages of each type of door design: A. Roll doors Advantages a. Raises and lowers vertically, so is not susceptible to damage from debris on the roadway. b. Channels on either side of door provide good air seal. c. Complete door and operating mechanism can be mounted on the bulkhead. Disadvantages a. Space required above door to house operating mechanism and door. b. Constructed of sheet metal that is susceptible to damage from equipment and blasting. c. Maintenance and lubrication required so that corrosion does not render door inoperable. d. In large openings or high differential pressures, the door may deflect excessively, rendering it Inoperable. 118 B. Swing doors 1. Advantages a. Very rugged. b. Opening and closing speed can be set as desired. 2. Disadvantages a. Debris on the floor may prevent door from closing. b. Sealing around door perimeter is difficvilt. c. Two air cylinders are generally required to operate the door. Where it is possible to mount on air cylinder on the mine roof, one air cylinder can be used with a rod connected to the top of each panel. C. Canton doors 1. Advantages a. Because the door panels open in opposite directions, the air pressure in the drift has little effect on the power required to open and close the doors. b. Very rugged. c. The doors can be set at the desired opening and closing speed. 2. Disadvantages a. Debris on the floor may prevent the door panels from closing properly. b. A floor trough might be required to accommodate a bar link- age at the bottom of the door. D. Telescoping doors 1. Advantages (same as for roll doors) a. Raises and lowers vertically, so is not susceptible to damage from debris on the roadway. b. Channels on either side of door provide good air seal. c. Complete door and operating mechanism can be mounted on the bulkhead. 119 Disadvantages a. Space required above door to store the raised door sections and house the operating mechanism, b. Susceptible to damage from equipment, c. Contains many separate pieces, all of which must fall into place for door to operate. E. Sliding doors 1, Advantages a. The door panel or panels can be set up on an inclined track to provide self closing, b. Air pressure has little effect on the opening or closing force required, 2, Disadvantages a. An area along side each opening large enough to accommodate the door is required, FIRE DOOR DESIGN AND TESTING After analysis of the data available for each type of door design, improved fire door design criteria were established: a. Type: The canton style door was selected because the panels open in opposite directions, balancing pressure on the door, and because it can be ruggedly built, b. Size: 12 feet by 10 feet is the maximum size door that will fit into the fire door test fixture at Underwriter's Laboratories (UL), c. Long Life: The door was designed to last for over 200,000 cycles, or for at least 5 years, d. Fire Rating: A fire door rating of 1-1/2 hours was proposed in the MSHA definition of a fire door, e. Operation: Compressed air, f. Air Pressure Rating: 6 inches WG, 120 First-Generation Fire Door The first fire door consisted of 11-gage sheet steel welded to a steel frame, with steel strip horizontally welded to the face of the door for rein- forcement. Neoprene door seals were used for sealing around the perimeter of the door. The door assembly included a 2 ft 6 in by 6 ft 8 in, airlock 1-person door. Mockup testing was conducted at Unidynamics facilities. The door assem- bly was mounted in a test fixture (figs. 1 and 2). The test fixture consisted of a frame for mounting the door in, an exhaust fan pumping air behind the door, and an air bag on the downwind side of the door for determining leakage through the entire door assembly. After leakage tests, the plastic bag was removed and door deflection measurements taken at 6 inches X-JG. It was also demonstrated that one person, by pushing on one door panel, could easily open the door despite the pressure differential. This is possible because the door panels open in opposite directions to balance the pressure. A series of cycling tests at varying pressures were also conducted. After successful mockup testing the door was sent to Underwriters Laboratories (UL), Chicago, 111., for the 1-1/2-hour fire rating test. The door was built into a 16-inch masonry wall contained within a test frame (fig. 3) and sealed using the neo- prene seals. After the masonry wall had seasoned, the fire test was conducted in accordance with UL Standard lOB, "Standard for Fire Tests of Door Assemblies." The test frame was attached to a test fixture consisting of gas ports to pro- vide fuel for the fire and thermocouples to measure temperatures on the fire side of the door (fig. 4). Throughout the fire test, observations were made to note the character of the fire and its control, the condition of the exposed and unexposed faces, and all developments pertinent to the doors as a fire barrier with special reference to stability and flame passage (fig. 5). At the conclusion of the fire test it was determined that the door failed the 1-1/2-hour fire rating test because of the following: 1. The 3-in deflection occurred along the meeting edge of the two pan- els. This deflection exceeded the thickness of the doors and allowed the pas- sage of flame through the door. 2. An opening occurred between the door and frame along the hinge edges because the neoprene seal burned away. Second-Generation Fire Door The second generation fire door design was very similar to the first door, with changes made to reinforce the door and to improve the door seal. The second door consisted of 14-gage sheet steel over a steel frame, with a 5-in-wide steel channel welded to the skin both horizontally and diagonally for reinforcement. An asbestos material specifically designed for fire doors was used for sealing purposes. After successful mockup testing, the second door was sent to UL for fire testing (fig. 6). This door successfully passed the UL Standard lOB 1-1/2- hour fire rating test (fig. 7). 121 122 < i 123 124 i n FIGURE 4. - Fire door test fixture. 125 •nv V *\ .L^1 FIGURE 5. - Technician taking door deflection measurements during Underwriters Laboratories fire test. 126 127 m£^ u o '* -^ 'P iMK^ (1) ^- / 3^«^ 1 iSm ^ I ■ •»* % 0) i .. E o \ > \ o \ E \ 0) c (U V _Q \ ' O O 128 Third-Generation Fire Door After testing the second-generation fire door, the following mandatory regulations were adopted: 1. "To prevent the spread of smoke or gas in the event of a fire, venti- lation doors shall be installed at or near shaft stations of intake shafts and at any shaft designated as an escapeway under standard 57.11-53 or at other location which provide equivalent protection"... Doors constructed by this standard shall be... "Constructed according to the specifications within the definition of 'fire door' in section 57.2, if located in a timbered area, in an area where the exposed rock is combustible, or in an area where a signifi- cant fire hazard is present." (U.S. Code of Federal Regulations — Title 30, sections 57.4-61A). 2. "To confine or prevent the spread of toxic gases from a fire origi- nating in an underground shop, the mine operator shall install in each opening to the shop a fire door or bulkhead constructed in accordance with the defi- nition of 'fire door' contained in section 57.2." (U.S. Code of Federal Regu- lations — Title 30, sections 57.4-61B). The definition of fire door contained in section 57.2 stated: "'Fire door' means an openable closure for a passageway, shaft, or other mine opening to serve as barrier to fire, the effects of fire, and air leakage. A fire door shall be constructed of materials and assembled so as to be equivalent to a door having a fire resistance rating of one and one half hours or greater, and on exposure to fire on one side for 30 minutes, the temperature on the unexposed side shall not exceed 250° F, as determined by a nationally recog- nized testing agency in accordance with 'Standard Method of Fire Tests of Door Assemblies,' National Fire Protection Association (NFPA) Code No. 252, 1972 or equivalent. The framework assembly of a fire door and the surrounding bulk- head, if any, shall be at least equivalent to the fire door in fire and air- leakage resistance, and in physical strength. NFPA Code No. 252 is hereby incorporated by reference and made a part hereof." The second-generation door met the new definition of fire door except the 250° F surface temperature requirement. A third-generation fire door was designed to meet all requirements of the new definition. (It should be noted that MSHA allows doors meeting all fire door requirements except the 250° F requirement to be used if an automatic sprinkler system is installed, main- tained, and used to retard the passage of fire on both sides of the door. They can only be used where current fire door technology precludes the use of fully approved doors in certain large openings, and the preclusion is based upon the inability to meet the 250° F requirement on the unexposed side of the door (MSHA Policy Memorandums 81-9MM and 81-lOMM). The third-generation door consists of 14-gage sheet seel welded to a steel frame, reinforced with a 5-in-wide channel. This door has two sheet steel skins, with 5 inches of insulation between the two skins (figs. 8-10). It will be fire tested at UL facilities to determine if it meets the 1-1/2- hour rating test, as well as the 250° F surface temperature requirement. 129 m' ^7:^m^^^^smMi 130 131 ^'GUf^E 10. - Thin •generation fne door f^ounted in test fix ^ure. 132 Another third-generation door has been installed at a Missouri lead mine for endurance testing. SUMMARY AND CONCLUSIONS During an underground mine fire, controlling the spread of toxic gas and smoke is essential. If a fire were to occur in an area such as a shaft sta- tion or shop, controlling the spread of the fire itself may determine if the miners will have sufficient time to evacuate the mine. The Bureau of Mines has developed a 1-1/2-hour rated fire door for use in large openings. An additional door, built to meet MSHA requirements, has been fabricated and will be fire tested at Underwriters Laboratories. Another door of the same con- struction is currently undergoing in mine tests at a Missouri lead mine. ij 133 AUTOMATIC FIRE PROTECTION SYSTEMS FOR MOBILE UNDERGROUND MINING EQUIPMENT by Guy A. Johnson^ INTRODUCTION Because of the Increased use of mechanized equipment underground in order to improve productivity, the fire danger associated with this equipment has increased. The Bureau has worked in the area of surface vehicle fire protec- tion for several years and has finished the basic research and development work. The Bureau is now working with MSHA and the National Fire Protection Association to develop standards with manufacturers and insurers to commer- cialize the technology. Current efforts within the Bureau involve expansion of the fire protection research into the area of underground vehicles. Ini- tiative came in 1976 when the mining industry had two fatalities because of a fire on a load-haul-dump (LHD), Because of this growing problem, a solution is needed now to prevent fatalities in the future. The basic technology is at hand, but it needs to be modified, debugged, and applied to the underground mining hazard. SYSTEM DESIGN The Bureau realizes that there are manually activated, fixed piping, dry-chemical fire protection systems now on some underground vehicles. This equipment has had some success, but it had problems too. Using these systems as a starting point, the Bureau proceeded to automate fire protection so that a fire would be sensed by a system and suppressed automatically, if the opera- tor did not take positive action. This is important because the industry has found that when fires occurred, many machine operators abandoned their equip- ment. As a result, a system was developed that uses thermal fire sensors for fast, reliable sensing of a fire and dry-chemical fire suppressants. These suppressants are commonly used underground, so the mines are comfortable with them. The system (fig. 1) would have options for the underground vehicles that would automatically shut down the engine and set the brakes. The hardware (fig. 2) developed by the Bureau is very versatile (fig. 3) because no one system will be able to solve every type of fire hazard. Start- ing in 1977-78, the Bureau did a number of fire tests in the laboratory inves- tigating the timing sequence to sense a typical vehicle fire in a simulated underground situation (fig. 4). We then moved outside and, with a mockup of an engine area of a LHD, lit a number of fires to find out the time it took to extinguish the fire (fig. 5). Some very large fires were tested on the 'Supervisory mining engineer. Twin Cities Research Center, Bureau of Mines j Minneapolis, Minn. 134 135 1 ) \ O) ;d CO o o N. I 136 II c E T3 5 S-^ CO 15 '" ? E S 137 138 ^^^ 139 IN-MINE FIRE TESTING After this lab testing in 1977, the Bureau went to the Lakeshore mine, then operated by Hecla Mining Co., Casa Grande, Ariz., and installed a first- generation prototype system (fig. 6). This system had a control box (fig. 7) that would warn the driver of a fire, fire sensors in the vehicle's engine and transmission area, and dry-chemical fire suppressant distribution piping. Fire pans (fig. 8) that contained heated hydraulic oil were installed, and the fire pans were set on fire. The system sensed the test fire, warned the dri- ver, and the dry chemical efficiently extinguished the fire (figs. 9-11). This underground fire test was conducted in close cooperation with the mine and MSHA personnel. IB PI 19 1 fc«.:^^H' ^-^ ^^H p "'Si-- " 1 m 8 H FIGURE 6, - Installing the first-generation system at the Lakeshore Mine, Arizona, 1977. 140 141 142 ^r\^^^^:, ''1 i 1 ■QMV 143 144 145 CURRENT WORK In 1979 and 1980, numerous design alternatives were developed and mines tested to demonstrate to the industry the many available options (figs. 12- 15). These systems were developed for different sizes of load -haul -dumps, underground trucks, fire warning only systems, etc. More information is available in a Bureau contract report. ^ A system demonstration unit (fig. 16) is available for inspection at the Bureau's Twin Cities Research Center at your convenience. As a parallel program, the Bureau investigated the toxicity of dry chemi- cal fire suppressants when used underground. This work was done in Missouri at the Fletcher mine. Bureau engineers went underground, installed several different off-the-shelf, manually activated, dry-chemical systems, then lit a number of fires. Again, working closely with MSHA to get the proper variance approvals, we conducted toxicity studies. These studies showed that the use of dry chemicals underground is very safe. Automatic systems are currently available from the Ansul Co., Lease-AFEX, and Kidde, and several other manufacturers are getting into the market. The systems are currently finishing long-term testing in Missouri, in the Coeur d'Alene area of Idaho, and in Arizona. The systems range in price from about $1,000 to about $3,000. The National Fire Protection Association is expanding its consensus standards work to address the problem of underground vehicle fires. -Reid, G. R. , D. L. Stockwell, and R. J. Plog (The Ansul Corp.). Automatic Fire Protection System for Mobile Underground Metal Mining Equipment. Phase II Report. BuMines Open File Rept. 81-76, 1975, 151 pp.; available for consultation at BuMines facilities in Denver, Colo., Twin Cities, Minn., Bruceton and Pittsburgh, Pa., and Spokane, Wash.; at U.S. Dept of Energy facilities in Morgantown, W. Va.; and at the National Library for Natural Resources, U.S. Dept. of the Interior, Washington, D.C.; available from National Technical Information Service, Springfield, Va. , PB 254 851/AS: contract HO 252038. 146 Schematic of automatic fire protection system on Eimco, model 91 2, load haul dump (retrofit of existing manual system) „ Detection cable (thermal) 1 2 VDC: Control console Dashboard actuator Check valves (2) 30-lb dry chemical tank Electric actuation device FVz nozzles (2) Air cylinder (fuel shutoff ) FIGURE 12. - Alternate system design for Eimco, model 912 LHD. 147 Air cylinder (fueishutoff) Control valve (set brakes) Schematic of automatic fire protection system on Sien Brut, model 59D, diesel tractor (retrofit of existing system) FVi nozzles (4) Detection cable (thermal) Control console 1 2VDC Dashboard actuator Check valves (2) 30-lb dry chemical tank Electric actuation device FIGURE 13. - Alternate system design for Sien Brut, model 59D, diesel tractor. 148 Schematic of automatic fire protection system on Wagner,ST-8, load haul dump (retrofit of existing manual system) Check valves (2) i 30-lb dry chemical tanks FIGURE 14. Detection cable (thermal) Fy2 nozzles (2) Electric actuation device Control console Alternate system design for Wagner ST-8 LHD Air cylinder (fuelshutoff) 149 150 FIGURE^16. - Automatic fire protection system demonstration unit at the Bureau's B D ** ^ ' Twin Cities (Minn.) Research Center ■CiU.S GOVERNMENT PRINTING OFFICE: 1981-703-002/88 IT.-BU.OF MIN ES,P GH.,P A. 25753