m ■ ■ ■ ■ I ■ ■ ■ ■ ■ 1 1 I ft*! J * I H ■ ■ ^H ■'-•* J ... ^ •••»• J ^v ** v \ •••• /■ . \ *-"" , '.< s o > * v ** 'Wot' v^ : ^m^» ,0 -^V°^^*ak-.%. y--aK&^. y,aat-X y.-sfc.^ ^ •^o< "bv* • ; \o^ v'-^\^ %^r^\^ V-^*\/ V^ f, %°° V-- v •:• y.i^-V .c°*..^i:->o .^\v^/V • <*. \/ *^ .V^M\ - - ^^v ^o 11 '* »J^L'» : «^^ J .^i^/ ^ y ^ ; ..' ( OTl\¥; ^v '.»#• .^-v. *bV° ^ ^c ^ ••«• V -A- V* ••«& W ^ v ^~ * *♦ ' r \- jP-v v^»: .«5°* a°« a< ? ^ ».vw * v '* 3^ ♦ AV ^a • .G v o, *• . » * A <. >,• .^ \ '-W^ ***** /J§&'- ^ ** <> *-TVV' ,6* i * «? #■, »*( O, »o , t * A y ^ J? ^ ^> ° 4V V*. v * 5 • 4 aV <* 4 -7^«' ,G r *o. 'o.T* * A y *7* •4°*, - -•\/ %v^*/ \-^^V V^^'^° " •^ :*W ^/ :#fe %/ .-ate- \/ .MC^ ^\^.:.,^ ./ .-^sfei-. \^ 'v^C,- . , , • if c <* S • • / .": tS* :'^^°: "?>* ""Sill- **o< :^»>*- «bv* »'^&'- ■"+*„■* -*^^.-. -u.* 1 • A V *^. - IC 9144 Bureau of Mines Information Circular/1987 Spontaneous Combustion Fire Detection for Deep Metal Mines By William H. Pomroy UNITED STATES DEPARTMENT OF THE INTERIOR Information Circular 9144 Spontaneous Combustion Fire Detection for Deep Metal Mines By William H. Pomroy UNITED STATES DEPARTMENT OF THE INTERIOR Donald Paul Hodel, Secretary BUREAU OF MINES Robert C. Horton, Director Library of Congress Cataloging in Publication Data: Pomroy, William H. Spontaneous combustion fire detection for deep metal mines. (Information circular; 9144) Supt. of Docs, no.: I 28.27: 9144. 1. Mine fires. 2. Combustion. Spontaneous. 3. Fire detectors. 4. Metal sulphides. 5. Mines and mineral resources. I. Title. II. Series: Information circular (United States. Bureau of Mines); 9144. ^TN295JJ4- * [TN315] 622 s [622'.8] 86-600294 CONTENTS Page Abstract 1 Introduction • First-generation spontaneous combustion fire detection system 2 Characterization of spontaneous combustion fires in metal mines 3 Identification of composition of spontaneous combustion fuel 3 Identification and quantification of products of combustion resulting from spontaneous combustion 4 Organic vapor evolution 6 Analysis of test results * 6 System design 7 Laboratory tests 10 In-mine test 11 Second-generation spontaneous combustion fire detection system 16 Design modifications 17 Laboratory tests 19 In-mine tests 19 Summary and conclusions 25 ILLUSTRATIONS 1. Determination of fire hazard of various compositions of spontaneous combustion fuel 4 2. Combustion products for sawdust mixtures 5 3. Surface telemetry and recorder assembly 11 4. Laboratory testing of spontaneous combustion fire detection system 12 5. Results of laboratory testing 12 6. Immediate vicinity of in-mine test 13 7 . Mine level on which in-mine test was conducted 13 8. Fire detection enclosures 14 9. Induced spontaneous combustion test fire underground 16 10. Results of in-mine induced spontaneous combustion testing 17 11. Surface telemetry and control unit for second-generation spontaneous combustion system 18 12. Layout of major components of second-generation spontaneous combustion fire detection system 19 13. Section through test mine showing locations of sensor assemblies 20 14. Sensor assembly B, prewired and mounted on panel 21 15. Layout for in-mine fire test of second-generation spontaneous combustion fire detection system 22 16. Igniting test fire 23 17. Test fire burning 23 18. CO and C0 2 measured during fire test 23 19. Smoke measured during fire test 24 TABLES 1. Fuel compositions and chamber temperature set points for tests to identify and quantify products of combustion resulting from spontaneous combustion 6 2. Comparison of fire detection systems using pneumatic tube bundle and electronic telemetry 8 3. Comparison of CO and C0 2 concentrations measured by detection system versus laboratory analysis 24 UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT °F degree Fahrenheit mA milliampere ft foot mCi raillicurie ft 3 cubic foot min minute ft 3 /min cubic foot per minute oz ounce gal gallon pet percent h hour pct/ft percent per foot Hz hertz ppm part per million in inch s second in/h inch per hour scf/h standard cubic foot per hour kHz kilohertz V volt lb pound vol/h volume per hour lb/ft 3 pound per cubic : foot yr year SPONTANEOUS COMBUSTION FIRE DETECTION FOR DEEP METAL MINES By William H. Pomroy 1 ABSTRACT Spontaneous combustion fires involving high-sulf ide-content ores are a relatively infrequent yet serious safety hazard in mining. They are also the cause of lengthy mine shutdowns because they typically occur in areas of a mine where abundant fuel material is present but which are inaccessible for fire fighting. This Bureau of Mines report describes research to design, fabricate, and test in the laboratory and field a system that warns of spontaneous combustion fires in metal mines. Over- all performance of the detection system was found to be satisfactory, in that the system was capable of reliably detecting low levels of com- bustion products believed to indicate the preflaming stage of sponta- neous combustion in metal mines. Installation of similar systems in mines with a high risk of spontaneous combustion is recommended. Prin- cipal operating problems and recommended corrective actions are also discussed. Group supervisor, Twin Cities Research Center, Bureau of Mines, Minneapolis, MN. INTRODUCTION Exothermic oxidation reactions in metal sulfide ores can occur in underground mines. If the heat generated by these reactions is not dissipated, temperatures sufficient for rapid oxidation and com- bustion of both the sulfides and the adjacent timber and other mine combusti- bles may be produced. Although spontane- ous combustion fires are relatively in- frequent, accounting for only about 2 pet of all underground noncoal mine fires, they are generally quite disruptive to mine operations and represent a signifi- cant safety hazard to miners. Spontane- ous combustion fires often start in abandoned, backfilled, and/or caved mine areas where access for fire-fighting operations is difficult or impossible. Compounding the problem of accessibility is the large amount of fuel that is gen- erally available to a spontaneous com- bustion fire. Fires on discrete pieces of equipment are generally of short dura- tion because they self-extinguish when the available fuel is consumed. However, the large quantity of support timber present in many older mines can provide fuel sufficient for fires of many months' duration, and spontaneous combustion fires typically involve such support tim- ber. Since 1950, about 57 pet of noncoal underground mine fires lasting longer than 24 h were caused by spontaneous combustion. Though research is now under way, the precise chemical and physical mechanisms giving rise to sulfide oxidation and spontaneous combustion in mines are still not well understood. Hence, pre- vention of spontaneous combution fires is generally limited to sealing abandoned, backfilled, and caved areas known to be susceptible to self -heating, in the hope that the denial of oxygen will retard the oxidation processes. Sealing is, however, an imperfect solu- tion, and despite efforts to locate and seal "hot spots," high-risk mines are still vulnerable to spontaneous combus- tion events. Oxygen can leak into sealed areas as a result of ground movement, caving, natural fracturing, and faulty bulkhead construction. And spontaneous combustion fires can occur outside sealed areas. As an added precaution to fire prevention efforts, systematic fire de- tection is practiced at most high-risk mines. Fire detection can include atmos- phere monitoring behind bulkheads, as well as temperature and air quality checks in working areas. In 1978, the Bureau of Mines embarked on a research program to upgrade the technology for detecting spontaneous com- bustion fires, to fabricate a prototype system for spontaneous combustion fire detection, and to perform in-mine valida- tion tests of this system. The work was performed under contract; preliminary re- search findings, covering the development and testing of a first-generation sponta- neous combustion fire detection system, are contained in the contractor's final report. 2 In addition to summarizing those find- ings, this Bureau report describes the design and long-term, in-mine testing of an improved, second-generation spontane- ous combustion fire detection system. FIRST-GENERATION SPONTANEOUS COMBUSTION FIRE DETECTION SYSTEM Development of a first-generation spon- taneous combustion fire detection sys- tem was accomplished in four steps: (1) characterization of spontaneous com- bustion fires in metal mines, (2) design of a spontaneous combustion fire detec- tion system, (3) laboratory testing, and (4) field testing in an operating under- ground mine. ^Stevens, R. B. Improved Spontane- ous Combustion Protection for Under- ground Metal Mines (contract HO282002, FMC Corp.). BuMines OFR 79-80, 1979, 262 pp.; NTIS PB 80-210461. CHARACTERIZATION OF SPONTANEOUS COMBUSTION FIRES IN METAL MINES Accelerated spontaneous combustion tests were performed in the laboratory to identify and quantify the products of combustion that could be expected to re- sult from an actual spontaneous combus- tion event in a mine. These data were then used to guide the selection of de- tection instruments to be incorporated into the prototype system for spontaneous combustion fire detection. Several test series were conducted. Significant results are summarized in the following sections. Identification of Composition of Spontaneous Combustion Fuel Samples consisting of sawdust and saw- dust with sulfide and sulfur additives were evaluated in a specially designed chamber under mine conditions (95° F, 95 pet relative humidity (RH)) and induced spontaneous combustion conditions (300° to 400° F) to determine which fuel com- positions represent the most serious spontaneous combustion hazard. The fuel compositions shown to exhibit the great- est propensity toward self-heating were used in follow-on tests to identify and quantify the products of combustion in- dicative of spontaneous combustion. An 8-ft '-capacity chamber with variable temperature capability from ambient to 400° F and variable RH capability from 20 to 95 pet was used. The chamber was equipped for sample temperature monitor- ing, variable chamber ventilation, and gas sampling. The ventilation flow rates were 0, 2, 4, and 8 scf/h (correspond- ing to 0-, 0.25-, 0.5-, and 1.0-vol/h exchange rates in the chamber). In the initial test sequence, eight approximate- ly 1-ft 3 samples were tested. The eight samples included 10- and 20-lb/ft 3 pack- ing densities of the following composi- tions: pure sawdust, sawdust plus 5 pet FeS, sawdust plus 1 pet S, and sawdust plus 5 pet FeS and 1 pet S. These compo- sitions corresponded to samples collected at several underground metal mines and were believed to represent medium- to high-risk, spontaneous combustion fuel ma- terials. After 144 h at 95° F, 95 pet RH, chamber conditions were altered to 120° F, 95 pet RH, and held for 186 h. Finally, the temperature was increased to 145° F, 95 pet RH, and held for an addi- tional 144 h (for a total of 19 days, 18 h). Under these conditions, none of the samples showed signs of spontaneous heat- ing. Five of the samples, all containing FeS and/or sulfur, were then heated to higher temperatures (180°, 210°, 250°, 300° F) in an effort to induce spontane- ous combustion under conditions that might be encountered if combustibles were located near a self-heating ore body. Although some decomposition of the sam- ples occurred under these conditions, producing high CO and CO2 levels, no evi- dence of self-heating was observed. At 338° F, all samples underwent combustion. In similar tests where sawdust with 5 pet FeS and 1 pet S was aged for 123 h at 95° F, 95 pet RH, spontaneous combustion was Induced by elevating the chamber tem- perature to as low as 293° F. However, when sawdust alone (no FeS or sulfur ad- ditives) was pretreated at elevated tem- peratures and humidities for extended pe- riods, no self-heating was observed, even at chamber temperatures of 350° F. The appearance of the residue indicated that wood modified by long-term treatment at elevated temperature and humidity in the presence of FeS and/or sulfur burned more completely than unmodified wood. In addition, the combustion temperatures of the aged, FeS- and/or sulfur-treated sam- ples, ranging from 295° to 340° F, are considerably lower than the literature value of 392° F for untreated wood, indi- cating reduced ignition points for woods in the presence of FeS and/or sulfur contamination. Based on these test results , further experiments were conducted in the chamber to delineate the Ignition potential of five high-risk, fuel compositions when heated for prolonged periods at higher pretreatment temperatures: (1) aged saw- dust plus 5 pet FeS and 1 pet S, (2) aged sawdust plus 5 pet FeS, (3) aged sawdust plus 1 pet rotted wood, (4) aged sawdust with no additives, and (5) virgin saw- dust. After 118 h at 250° F, the chamber was shut off for thermocouple repairs. The chamber was restarted at 302° F. Both samples containing FeS and sulfur then began to display visible signs of smoldering, including smoke generation and pile discoloration. However, the three samples containing virgin and aged sawdust and rotted wood, but no FeS or sulfur additives, did not undergo combus- tion. These results, summarized in fig- ure 1, confirm results of earlier tests at lower pretreatment temperatures. These brief laboratory studies cannot absolutely rule out the possibility that wood alone may be responsible for sponta- neous combustion in mines. However, the evidence strongly suggests that mine tim- bers alone do not exhibit self -heating tendencies when exposed to conditions that prevail in most underground metal mines. Rather, the principal source of spontaneous combustion appears to be 425 400 375 350 - 325 - 300 275 - 250 1 1 1 1 1 1 KEY i Aged sawdust +5 pel FeS *l pel S Aged sawdust +5 pet FeS Virgin sawdust; aged sawdust (no additives); aged sawdust (1 pet rotted wood) Oven set temperature / - - / / / / / / ^ / ^* • /> ,' // ,- yU — " yv S*' //'' /// -XV2" ^O-' _ ^£VS-' — — — "i^^"^ ' ^^^ ! r / '// ~ '/ - r 1 118 119 120 121 122 123 124 125 126 HOURS INTO TEST FIGURE 1.— Determination of fire hazard of various com- positions of spontaneous combustion fuel. high-sulf ide-content ores that ignite secondary combustibles, especially timber modified by long-term exposure to high temperatures and humidities, and contami- nated by sulfides and/or sulfur. Identification and Quantification of Products of Combustion Resulting From Spontaneous Combustion Twenty-eight fixed-temperature pyro- lytic decomposition tests of several high-risk compositions of spontaneous combustion fuel were performed in a 128- ft -capacity chamber. The objective was to identify and quantify the mix of com- bustion products that evolve during the preflaming stage of a spontaneous combus- tion fire. This stage may last from 1 day or less to several weeks or more. The goal of spontaneous combustion fire detection is to detect this characteris- tic mix of combustion products during the preflaming stage, so that appropriate emergency procedures (evacuation, fire fighting, etc.) can be initiated before the fire reaches the flaming combustion stage and threatens rapid growth and con- taminant spread. The chamber was instru- mented for monitoring smoke obscuration, CO2, CO, hydrocarbons, oxygen, and SO 2. The samples were contained in a 6.3-in- diam by 3.0-in-high cylindrical pan. Eight thermocouples were arrayed in the sample to measure sample temperatures throughout each test. The compositions of fuels and chamber temperature set points tested are shown in table 1. Heat was applied to the samples through the chamber floor. All tests were limited to 60 min. Figure 2 depicts the most significant results of this test series. Sawdust plus 1 pet rotted wood (rotted wood col- lected from several sulfide mines) heated to 410° F (fig. 24) produced moderate levels of CO and CO2 after 50 min. The 5-ppm-S02 level is believed to result from long-term sulfide exposure of the rotted wood samples in the mine. Sawdust plus 5 pet FeS heated at 410° F (fig. 25) produced high levels of CO and CO2 after 45 min, as well as moderate levels of SO2, smoke, and oxygen depletion after 21 19 50 40 - 17 - e 30 15 o 20 13 - 10 - 2lr 50 lO.OOO^zzax 19 - 40 -17 - £30 15 - g 20 13 - I l - 21 10 - 50 19- 40 - I7h u a. N ° 15 6 30 ° 20 13- 10 - 21 50 19- 40 r. 17- 15- E 30 - a 6,000 - a. O f> 20 13- 10- I I 1 - O 1 - 8,000 - 5 4,000 2,000 - 30 40 TIME, min FIGURE 2.— Combustion products for sawdust mixtures. A, Sawdust plus 1 pet rotted wood heated to 41 ° F; B, sawdust plus 5 pet FeS heated to 41 ° F; C, sawdust plus 1 pet S and 5 pet FeS heated to 410° F; D, sawdust plus 5 pet CuS heated to 410° F; £, sawdust plus 5 pet ZnS heated to 410° F; F, sawdust plus 1 pet S and 5 pet FeS heated to 520° F; G, sawdust plus 1 pet S and 5 pet FeS heated to 800° F. 20 30 40 TIME, mihr 50 60 TABLE 1. - Fuel compositions and chamber temperature set points for tests to identify and quantify products of combustion resulting from spontaneous combustion (X indicates testing) Fuel compositions tested 390° F 410° F 465° F 520° F 590° F 800° F Virgin sawdust: X X X X X X X X X X X X X X X X X X X X X X X X X x 1 pet S, 5 pet FeS.... X X X 5 Dct PbS Rotted wood (100 pet)... 50 min. Sawdust plus 1 pet S and 5 pet FeS heated at 410° F (fig. 2c ) produced very high levels of S0 2 and oxygen de- pletion after 30 min. CO and C0 2 also reached very high levels after 35 min; however, smoke levels were only moderate after 60 min. The samples of sawdust plus 5 pet CuS (fig. 2D) and sawdust plus 5 pet ZnS (fig. IE) heated at 410° F followed much the same pattern as the sawdust plus 5 pet FeS, producing high levels of CO and C0 2 after 45 min and moderate levels of S0 2 , smoke, and oxygen depletion after 50 min. At higher temperatures (figs. 2F-2G) , evolution of combustion pro- ducts, especial-ly S0 2 , occurred more rapidly (closed-cup flashpoint of sulfur is 405° F, autoignition temperature is 450° F). Organic Vapor Evolution The presence of a sweet odor, or "sweet gas. is frequently reported by miners prior to outbreaks of spontaneous combus- tion fires. This sweet gas is believed to be a mix of organic vapors produced during the preflaming stage of a sponta- neous combustion fire. Thus, organic va- por detection was considered for possible inclusion in the prototype spontaneous combustion fire detection system. Dif- ferential scanning calorimetry evolved- gas analysis was performed in order to determine whether the quantity of organic vapor production would be sufficient for detection. Methane, propane, formaldehyde, metha- nol, acetaldehyde, and acetone were the primary species detected. Trace levels were measured at temperatures as low as 212° F, with moderate levels observed in most species between 392° and 482° F. Peak concentrations occurred in all spe- cies at 662° F. Analysis of Test Results Of the four gases examined in detail in the pyrolytic decomposition studies, oxy- gen and CO respond sooner and at higher levels than C0 2 and S0 2 . However, the high levels of CO and C0 2 measured dur- ing the low-temperature (302° to 662° F) studies in the 8-ft 3 chamber suggest the value of both of these gases as early in- dicators. In addition, S0 2 detection may warn of a sulfide heating event that does not take place in the presence of second- ary combustibles. The failure of smoke obscuration levels to exceed the standard 2 pct/ft in the majority of the pyrolytic decomposition tests (below 482° F, the levels ranged from 0.5 to 2.0 pct/ft) im- plies that smoke detection may not offer an advantage over gas detection. Organic vapors, though produced in sufficient quantities at higher temperatures, were not proven reliable indicators at temper- atures below 392° F. Although air temperature sensing was not specifically addressed in these studies, it is practiced at most mines with recognized spontaneous combustion problems. A temperature rise may be the first indication of a self -heating event, especially if secondary combustibles are not immediately adjacent to the hot spot. Thus, the recommended mix of detection instruments to provide early warning of a spontaneous combustion fire in an under- ground metal mine includes those for oxy- gen, SO2, CO, CO2, and temperature. The recommended sampling ranges of the in- struments, as indicated by the test pro- gram, are as follows: O2, to 21 pet; S0 2 , to 30,000 ppm; CO, to 50 ppm (0 to 500 ppm if sampling behind bulkhead); CO2, to 30,000 ppm; and temperature, 0° to 150° F. SYSTEM DESIGN Two basic detection system configura- tions were considered: a pneumatic "tube bundle" approach and a fully electronic telemetry approach. Pneumatic tube bun- dle fire detection involves sampling the mine atmosphere at various locations through a network of plastic tubes to which a vacuum is applied. The airflows from each tube are sequentially directed to a central analytic station equipped for gas monitoring. The fully electronic telemetry approach involves the placement of detection instruments at each under- ground site to be monitored. Detector outputs are transmitted to a central con- trol point over electronic telemetry lines. Some advantages and disadvantages of each approach are outlined in table 2. The selection of a fully electronic te- lemetry configuration for the prototype spontaneous combustion fire detection system was based on two factors: First, only one sampling point was anticipated for this experimental system; thus, cost tended to favor the telemetry approach. This cost advantage could be magnified greatly if existing telemetry lines at the in-mine test site could be used (this indeed was the case at the test mine). Second, air temperature was identified as an important parameter to be measured, and air temperature cannot be detected by tube bundles. The selection of detection instruments was based on their field-proven resist- ance to harsh environmental exposures (heat, cold, humidity, dust, blast fumes, diesel exhaust, input voltage fluctua- tions, transients, and electromagnetic interference). The oxygen detector was a diffusion electrochemical cell type. The microfuel cell consumes oxygen from the atmosphere and generates a current proportional to the concentration. Life of the cell is approximately 12 months at 25 pet O2. No routine maintenance is required except periodic recalibration. Since use under the environmental exposures anticipated underground (heat and humidity primarily) was nonstandard, a determination of re- calibration frequency would be made as part of the in-mine test. SO2 detection was provided by a nondis- persive infrared absorption-type detec- tor. The unit, including the analyzer cell and power supply, are housed in a rugged fiberglass enclosure designed for underground installation. A dust filter is fitted to the sample plenum. No pumps or other auxiliary sample delivery or pretreatment are required, and the unit contains no moving parts. The analyzer detects the attenuation of radiation due to molecular absorption by the sample gas. A Nichrome^ filament pulsed at a specified frequency radiates broadband energy. This energy passes through the sample gas in a reflective optical chamber, through a spectral fil- ter, and is measured by a pyroelectric cell photo detector. The electrical sig- nal output is inversely proportional to the gas concentration and varies loga- rithmically with concentration. Selec- tivity to the sample gas is determined by the band-pass spectral filter. The de- tector's output signal does not indicate ■^Reference to specific products does not imply endorsement by the Bureau of Mines. TABLE 2. - Comparison of fire detection systems using pneumatic tube bundle and electronic telemetry Pneumatic tube bundle Electronic telemetry MAINTENANCE Simple for analyzers, pumps, controls, etc. , because all electronics and mechanical components can be located on surface. Plugs or breaks in sample tubes can be difficult to locate and repair. Access to some underground locations, for maintenance, calibration, and/or repair of equipment can be a problem. Trouble- shooting telemetry lines (opens, faults, etc. ) is generally easier than trouble- shooting sample tubes (plugs breaks). AREA OF COVERAGE In large, spread-out mines, the time re- quired for a sample to travel the entire length of a sample tube can be a limit- ing factor. E.g. , a 3/8-in sample tube 10,000 ft in length will result in a tube travel time of about 17 min. Very long transmission distances (5 to 15 miles) may require special telemetry provisions. COST In general, the larger the system (i.e., number of sampling points), the lower the cost per sampling point because only one analytic station is required. Like- wise, however, small systems have a rel- atively high cost per sample point. Since each sampling point requires a full compliment of detection instruments, system costs increase in proportion to the number of sampling points. Teleme- try line acquisition and installation costs are generally lower than for tube bundles servicing the same number of sampling points, especially if multiplex data transmission is employed. MEASUREMENT PRECISION AND RELIABILITY Because the analytic station can be on surface in a relatively clean environ- ment, high precision detection instru- ments can be utilized. Also, in that location, detector maintenance and calibration is likely to be frequent. Measurement precision and reliability is directly proportional to the adequacy of detector maintenance and calibration. Where accessibility is difficult, mea- surement precision and reliability may be low. NEED FOR ELECTRIC POWER Since all electronic equipment can be on surface, electric power is not required at each sampling point — a definite ad- vantage if underground power is lost during a mine emergency. In general, electric power is required at each sampling point; however, some detectors can be powered through the telemetry lines, (i.e., from surface). TABLE 2. - Comparison of fire detection systems using pneumatic tube bundle and electronic telemetry — Continued Pneumatic tube bundle Electronic telemetry PRODUCTS OF COMBUSTION THAT CAN BE DETECTED Most gases, including oxygen, CO, and C0 2 , can be accurately measured. Smoke particles tend to diffuse into the tube walls; thus, depending on the tube length, diameter, sample air velocity and other factors, smoke detection may or may not be possible for a given application. Some gases, such as N0 X and S0 2 , react with the tube or the wa- ter that may collect in the tubes, producing erroneous readings. Measure- ments of air velocity, direction, and temperature at the sampling point are impossible. Any gas, particulate, or condition (air velocity, direction, temperature, etc.) for which a detection instrument exists can be measured. ENVIRONMENTAL EXPOSURE FACTORS Analytic station can be located in a clean environment. Tubes are subject to plugging from dust accumulation and condensation of humid sample air. Freezing conditions further exacerbate humidity and condensation problems. Broken lines may result from abrasion, cuts, roof falls, etc. Ruggedness and resistance to harsh envi- ronmental effects (heat, humidity, cold, dust, diesel exhaust, blast fumes, etc.) are essential for detection instruments. Both detection instruments and telemetry lines may be affected by electromagnetic interference from nearby power lines and other sources. Broken telemetry lines may result from roof falls, etc.; how- ever, redundant telemetry lines, telem- etry loop configurations, etc., can be employed to improve telemetry reliabil- ity. Input voltage fluctuations and electrical transients (voltage spikes) can also be expected underground. 10 accurate gas concentrations without ref- erence to calibration curves, but normal gas levels are easily distinguishable from excursions that could result from a spontaneous heating event. Input power is 110 V ac, 50 to 60 Hz; output is to 1 V dc analog; and the range is to 30,000 ppm S0 2 . An identical infrared analyzer, fitted with a different spectral filter, was used for CO2 detection. The output range of the CO2 analyzer is also to 30,000 ppm. The CO detector selected was an elec- trochemical fuel cell type developed for underground mine use. The unit is housed in a rugged fiberglass enclosure. A mechanical pump is provided to draw sam- ple gas into the cell. CO is oxidized to CO2 at a potentially controlled elec- trode, and the current produced is pro- portional to the partial pressure of CO in the atmosphere. An electronic circuit then amplifies the signal to the 0- to 1-V-dc output level. The unit's output range is to 50 ppm CO; however, the de- tector can be retrofitted with new cells at nominal cost to alter the range (0 to 10, to 500, to 2,000 ppm, etc.). The life of the cell may vary from 6 to 12 months; however, monthly zero and span checks are recommended. Past experience also indicated that the sample pump re- quired occasional checks. Temperature detection was provided by an electrical resistance-type sensor. A 1-mA current is passed through an elec- trical circuit whose resistance varies with temperature. The signal is lin- earized and amplified, yielding a 0- to 1-V-dc output proportional to the 0° to 150° F range of the instrument. Selection of a telemetry system was based on several factors, including proven resistance to harsh environmental exposures; compatibility with existing mine telephone lines, chart recorders, and detection instruments; capability to transmit both analog and digital signals over distances up to 30,000 ft; minimum capacity of six outstations (each with five detection instruments); low cost; and minimal maintenance requirements. Fifteen candidate systems were evaluated with respect to these selection criteria. The system chosen is manufactured in the Republic of South Africa and is used primarily in deep South African gold and coal mines for fire detection. It em- ploys low-frequency division multiplex telemetry to transmit up to 48 signals over 1 twisted-wire pair. The system spans a frequency band of 0.3 to 10 kHz. Telephone conversation over the twisted pair does not interfere with data trans- mission. Operating on a balanced-line principle and incorporating special line filters and protection networks, the sys- tem is designed to be noise immune and interference free. Both the transmitters (underground) and receivers (surface) are housed in rugged fiberglass enclosures and are intended for mine installation. Since the focus of this research pro- gram was detection system design and function, the cost of an elaborate system control and alarm master panel could not be justified. A simple means for record- ing long-terra trends in the sensed param- eters and distinguishing shorter terra excursions from these baseline values (which might result from a spontaneous heating event) was sufficient. To sim- plify output display and data recording, a multipoint chart recorder was selected. The recorder uses colored ink ribbons and a "chopper bar" printer to display six channels of data. A chart speed of 1.25 in/h was selected to provide 30 days be- tween paper changes. Data are recorded at approximately 2-s intervals. Figure 3 shows the surface telemetry and recorder assembly. LABORATORY TESTS The spontaneous combustion fire warn- ing system was laboratory tested to ver- ify the compatibility of the various system elements; to identify potential in-mine installation, operation, mainte- nance, and troubleshooting problems; and to evaluate system performance under controlled conditions. Laboratory test- ing was focused in four areas: (1) long- terra calibration, zero drift, and 11 FIGURE 3.— Surface telemetry and recorder assembly. stability, (2) telemetry system perform- ance, (3) effects of harsh environmental exposures, and (4) observation and docu- mentation of induced system malfunctions. Simulated conditions included power-line voltage fluctuations, long-terra black- outs, high and low temperature and humid- ity extremes, and electromagnetic noise interference. The system was finally subjected to an induced spontaneous com- bustion fire to assess overall response to abnormal contaminant levels. Figure 4 represents the physical layout of the in- duced spontaneous combustion testing. After a "burn-in" period sufficient to achieve stable readings from the detec- tors, the system was functionally tested by exposure to combustion products from an induced spontaneous combustion event in the 128-ft 3 chamber. A fuel sample, consisting of sawdust plus 5 pet FeS and 1 pet S treated for 2 weeks at high tem- perature and humidity, was heated to ap- proximately 480° F. Figure 5 is a re- cording of the detector outputs. All system elements functioned proper- ly, with good agreement between the de- tection system and the laboratory analyz- ers. A peak CO2 level of 1,400 ppm was recorded 8 min after the heat source was removed from the sample. A peak SO 2 lev- el of 60 ppm occurred 5 min after the heat source was removed. CO levels in excess of the 50-ppm range of the CO de- tector were recorded within 12 min of the start of the test. The oxygen content of the atmosphere inside the chamber dropped to a low of 19 pet at 8 min into the test. The temperature peak occurred at 35 min. IN-MINE TEST Selection of a site for in-mine testing was based on four criteria: (1) mine fire history, spontaneous combustion oc- currences, and/or known heating condi- tions, (2) a high level of management in- terest and commitment to support system installation, operation, and maintenance, 12 Laboratory analyzers Spontaneous combustion fire detection system CO, 2£ Temp Pan Opt tempi IT / \ HHSELJ Receiver Surface receiver and \ recorder assembly Recorder - Underground telemetry assembly l ' Optical path C= -Thermocouple and anemometer — — Exhaust ('0 OTJT \ -Fuel pan Test chamber FIGURE 4.— Laboratory testing of spontaneous combustion fire detection system. (3) ease of access for project personnel, and (4) the availability of power and dedicated telemetry lines in the area designated for equipment installation. Three mines satisfying all the selection criteria were identified. Unfavorable economic conditions, howev- er, resulted in a cessation of operations at two of the candidate mines. An agree- ment was negotiated with the third mine, a deep, hot underground copper mine in Arizona, to permit the required tests. The mine operated three shifts per day using the underhand cut-and-fill mining method. Discussions with mine personnel led to the decision to install the detection system at a point 3,600 ft underground CO, ppm C0 2 , ppm S0 2 , ppm 2 , pel Temp, "F 15 20 25 30 35 40 45 50 25 10,000 1,700 760 290 10,000 1,700 760 290 3 5 10 15 20 30 45 60 75 90 105 120 135 150 FIGURE 5.— Results of laboratory testing. and approximately 3,100 ft horizontally from the shaft used for personnel and supply access. The system was installed in an inactive shop carrying exhaust air from several mine production and worked- out areas. Figures 6 and 7 depict the immediate vicinity of the testsite and the mine level on which it was located. Airflow through the area was approxi- mately 20,000 ft 3 /min, and temperature and relative humidity were both in the mid-90' s. The detection instruments and under- ground telemetry modules were installed on mounting panels, prewired, and en- closed in fiberglass housings for ease of transport and installation (fig. 8). Mine installation and an initial operat- ing checkout were accomplished over a 1-week period. Telephone communication was maintained between the underground testsite and the surface telemetry and recorder site during zero purge and span 13 ■Underground telemetry assembly und sensor assemt FIGURE 6.— Immediate vicinity of in-mine test. N6000 N4000 Main shaft (downcast) 9 shaft 400 Scale, ft 400 FIGURE 7.— Mine level on which in-mine test was conducted. i4 FIGURE 8.— Fire detection enclosures. calibration of the sensors to assure that all signals were aligned and functioning properly. Performance of the detection instru- ments over the 5-month in-mine test was generally good except for early problems with the CO and oxygen detectors. The CO detector experienced sample pump and dc circuit card failures during the first month of operation. The pump was re- placed, and the circuit card failure was traced to a grounding incompatibility between the negatively grounded battery backup power supply on the detector and the positively grounded telemetry system. The detector was operated on ac power for the remainder of the test period, and no further detector failures occurred. While the CO sensor was under repair, a second CO sensor was installed as a temporary replacement. This replacement unit was a catalytic semiconductor type that responds preferentially to CO but is also sensitive to other oxidizable gases. When the original CO sensor was returned, both units were operated simultaneously for the remainder of the test period, with the replacement sensor connected to 15 the telemetry channel previously assigned to the SO2 analyzer. The oxygen detector was removed from the system after 30 days, owing to a steady decline in oxygen readings. The vendor discovered a faulty microfuel cell. Although the vendor's failure analysis was not conclusive, possible causes were an imperfection in the cell cathode, an inadvertent application of 115-V-ac power to the cell during instal- lation, or a drop from a workbench to the floor when the unit was being packed for shipment to the test mine. Upon replace- ment of the faulty cell, the detector was returned to the system and operated with- out failure for the remainder of the test. During the 141-day test period, the temperature detector was the least active and the most stable. No adjustments or calibrations were required after the ini- tial installation and telemetry align- ment. Following its repair, the oxygen analyzer operated continuously for 75 days with a stable and accurate trace. The original CO detector operated for 11 days with an accurate and stable trace, and following vendor repairs, for the fi- nal 72 days of the test. Blasts were easily distinguishable on the chart recordings, showing peaks of 30 to 39 ppm CO shortly after scheduled blast times, compared with background levels of to 6 ppm CO. During one 30- day period, 99 pet of the blasts that oc- curred were identified and confirmed. The replacement CO sensor tracked closely with the original unit when both were op- erating simultaneously. The replacement unit indicated abnormally high levels of CO during the peaks, however, with values ranging from 40 to 250 ppm. The SO2 sensor was very stable during the test period, with the only measurable recorder deflections attributed to slight telemetry alignment error. This perform- ance confirmed laboratory findings that SO2 production occurs only at higher tem- peratures and prompted the decision to terminate SO2 measurements after 64 days so that its telemetry channel could be used for continued operation of the replacement CO sensor. The CO2 analyzer provided consistent, stable readings during the test period. Peak CO 2 mea- surements of 800 to 1,100 ppm correlated closely with blast patterns reported. Air samples were collected periodically during the period and analyzed in the laboratory. Close correlation between these measurements and the detection sys- tem was achieved. The five underground transmitters, five surface receivers, and power supply for the telemetry system performed well throughout the test. After initial in- stallation, the alignment was checked and adjusted twice. The purpose of checking and adjusting was more for familiariza- tion with the instruments than for cor- rection of drift. Surface receiver volt- ages were consistently identical with the input voltages from the underground transmitters. After the 5-month test period, a final functional test of the system was per- formed to evaluate system response to ab- normal levels of the sensed parameters. An induced spontaneous combustion event was staged in the vicinity of the detec- tion instruments to provide the required contaminants. A fuel source, consisting of 5 lb of pulverized mine timber mixed with 5 pet FeS and 1 pet S, was heated in an oven at about 390° F for approximately 23 h. The oven was then opened and the fuel pile inspected. The pile center was heavily charred and flames ignited spon- taneously several times (figure 9). Chart paper traces of this test are shown in figure 10. Slightly elevated CO and CO2 levels were noticeable over the entire test period until about 16 h after heating began when both CO detectors in- dicated sharp increases. When the oven door was opened after 23 h, the CO level increased to about 27 ppm and the CO 2 level showed a rise of about 250 ppm above ambient. Oxygen levels remained stable throughout the test at about 20.4 pet, and no temperature change was observed. 16 FIGURE 9.— Induced spontaneous combustion test fire underground. Although this test was too short to al- low observation of the slowly increas- ing trend in gas and heat levels that are typical of a spontaneous heating event (the fast rate of contaminant rise ob- served when the oven door was opened is more indicative of a fire than incipient spontaneous combustion), it is signifi- cant that elevated CO and CO2 concentra- tions were detected and could be distin- guished from normal background levels of these gases. The failure of the SO2, oxygen, and temperature detectors to respond during the 5-month test or during the induced spontaneous combustion test indicates that these parameters, although poten- tially reliable indicators of fires in more advanced stages, may not be reliable indicators of incipient spontaneous com- bustion as it occurs in mines. SECOND-GENERATION SPONTANEOUS COMBUSTION FIRE DETECTION SYSTEM Operation of the first-generation spon- taneous combustion fire detection system was successful in that a capability for detecting the products of combustion an- ticipated during a spontaneous heating event was demonstrated. However, the 5-month in-mine field test and final functional test also highlighted certain aspects of the system where improvements were needed. These needed improvements are summarized below: 1. Actual oxygen and CO detector main- tenance frequency (monthly) was some- what more than anticipated. Although not out of line with manufacturer recom- mendations, it was in excess of the de- sign goal of 6-month maintenance-free operation. 2. Analysis of detector output data was time consuming and confusing. Close, manual inspection of chart recordings was necessary to identify slowly developing 17 23 — i 1 1 1 1 1 1 r CO Temperature C0 2 S0 2 Oven door_qpen_ ~21"i -f Recorder off CO, ppm C0 2 , ppm S0 2 , ppm 2 ,pct Temp, "F 10 15 20 25 30 35 40 45 50 10,000 1,70 760 290 10,000 1,7 00 760 290 5 10 15 20 25 15 30 45 60 75 90 105 120 135 150 FIGURE 10.— Results of in-mine induced spontaneous com- bustion testing. problems (over several months, possibly). Confusion resulted because some traces indicated a positive deflection for in- creasing contaminant levels and others produced a negative deflection for in- creasing contaminant levels. Also, as airflows past the detection instruments changed, dilution of the sensed gases also changed, causing the detectors to report erroneously increasing or decreas- ing rates of contaminant generation. 3. Three of the detectors, SO2, oxy- gen, and temperature, showed very little movement during both the 5-month test and the induced spontaneous combustion test. DESIGN MODIFICATIONS The second-generation spontaneous com- bustion fire detection system included three significant design modifications to Improve system performance. 1. The detection instruments that re- quired the most maintenance (oxygen and CO) were either modified or eliminated from the system. Since oxygen did not prove to be a reliable spontaneous com- bustion indicator, oxygen detection was not included in the second-generation system. The CO detector's principal maintenance problems were the sample pump seizing and a grounding incompatability between a dc power circuit and the telemetry system. These problems were corrected by replac- ing the failed pump with a newer model featuring a modified mechanical linkage, and by bypassing the backup dc power operating mode. Following these modifi- cations, the detector operated without failure for the remainder of the test pe- riod. This sensor, as modified, was in- corporated into the second-generation de- tection system, along with two other electrochemical CO detectors that utilize diffusion-type cells. Use of diffusion- type cells eliminated the need for sample pumps. 2. A more elaborate output display and system control was provided to simplify and enhance data analysis. The six-chan- nel chart recorder was considered a suf- ficient display for the 5-month test of the first-generation system. However, the difficulty and occasional confusion encountered in reading the charts indi- cated the need for several changes in output display, especially if longer time scales were anticipated. The new output display and system control provides audi- ble and visual alarm annunciation at in- dividually adjustable output levels for each detector, and more uniform, linear chart recorder traces with positive de- flections indicating increasing contami- nant levels. This control unit (fig. 11) 18 FIGURE 11.— Surface telemetry and control unit for second- generation spontaneous combustion system. was supplied by the manufacturer of the telemetry equipment and is intended for mine use. 3. The detection instruments that showed little or no activity during the previous test were eliminated, and alter- native detectors were added to the sys- tem. As noted above, the oxygen detector failed during its first month of opera- tion; even when restored to proper func- tion, it showed little responsiveness for the remainder of the test (including the induced spontaneous combustion test). Its elimination from the system thus ne- gates a potential maintenance problem without adversely affecting system per- formance. The SO2 and temperature de- tectors were the most stable of all de- tectors tested, requiring almost no maintenance while maintaining proper cal- ibration over the entire 5-month test pe- riod. However, analysis of the results of the induced spontaneous combustion test indicated that measurement of the low level of SO2 and slight temperature rise expected from low-temperature incip- ient spontaneous combustion is not prac- tical from detectors placed in moderate to high airflows. Temperature and SO2 would be better measured nearer to poten- tial hot spots and where ventilation is lower (i.e., in stopes, behind bulkheads, etc. ). Submicrometer particulate, or smoke, detection was included in the second- generation detection system. Although the laboratory pyrolytic decomposition studies indicated that smoke levels were less than the standard 2 pct/ft in the majority of tests, levels of 0.3 to 1.8 pet were consistantly achieved, and lev- els exceeding 1.5 pet were measured in over half the tests. Using specially de- signed analog-output smoke detectors, these levels can be distinguished from normal background; thus, smoke may be considered a potential early indicator of spontaneous combustion. Two submicrometer particulate detectors were included in the system. The primary unit was supplied by the same vendor that manufactured the CO2 and SO2 analyzers and the telemetry and control system. The unit is a single-chamber, analog-out- put, ionization-type combustion particle detector. The ionizing source is 5 mCi of Kr-85 gas contained in a glass vial. The radioactive emitter ionizes the air in a chamber between two electrodes. Current is produced from production and transport of positive and negative ions to the opposite poles of the plates. A decrease in current, relative to clean air, is obtained when combustion products enter the chamber because the ionized 19 combustion particles are larger and heav- ier than the air molecules and move more slowly toward the end of the chamber. An electronic circuit detects the drop in current, which is proportional to the concentration of particles in the incom- ing air stream. The unit is intended for mine installation and is designed for long-term exposure to the harsh mine en- vironment. Previous experience with this detector under harsh mine conditions in- dicates excellent calibration stability with little maintenance required. The second submicrometer particulate detector, also an ionization type, is an experimental prototype model designed and fabricated by the Bureau. An air-velocity transducer was added to the system to permit more meaningful in- terpretation of reported contaminant lev- els. Changes in airflow that result in higher or lower smoke and gas concentra- tions can be factored into the data anal- ysis, thus avoiding erroneous conclusions 3- pen chart recorders (3), surface console, annunciotor panel, event printer, telemetry receivers with alarm outputs (6) 110 V — Combustion particle (Bureau experimental design) i no v — CO (diffusion — type) Combustion particle Telemetry transmitters (3) 110 V- C0 2 Telemetry transmitter CO (diffusion type) Combustion particle Spare Telemetry transmitters (3) .2 wires with ground Telemetry .. CO transmitters (3) i- Spare (pump type) Combustion particle Telemetry transmitter -110 V — CO, Telemetry Telemetry transmitters (3) CO (pump type) -110 V | Air velocity J Combustion particle Telemetry transmitter CO? FIGURE 12.— Layout of major components of second- generation spontaneous combustion fire detection system. regarding the source of contaminants. This unit is also intended for mine use and is designed for prolonged exposure to high temperatures, dust levels, and humidity. In addition to the design modifications of the system, the number of sampling locations was increased from one to four. This system expansion provided an oppor- tunity to evaluate a greater variety of detector types in various combinations and to test a larger capacity, more real- istically configured telemetry-control- alarm-display system. Layout of the second-generation spontaneous combustion fire detection system is shown in figure 12. LABORATORY TESTS The entire system was thoroughly tested under controlled laboratory conditions. The primary objectives of the laboratory testing were to verify the compatibility of the various system elements and to identify potential in-mine installation, operation, maintenance, and troubleshoot- ing problems. Prior experience was con- sidered sufficient to justify omitting rigorous environmental exposure testing of individual components. No unusual problems were discovered during the labo- ratory testing, and in-mine installation and testing proceeded on schedule. IN-MINE TESTS In-mine testing was performed at the same Arizona copper mine where first- generation system testing occurred. Be- cause of project funding constraints, the installation was split into two phases. Sensor assemblies A and B, plus the nec- essary surface telemetry and control equipment, were installed first, followed by sensor assemblies C and D, 7 months later. Three of the four sensor assem- blies were located near the mine's No. 6 exhaust shaft — each on a separate mining level. Sensor assembly A was located on the 3600 level, sensor assembly B on the 2550 level, and sensor assembly D on the 3400 level. Sensor assembly C was 20 located at the No. 4 exhaust shaft on the 500 level. Figure 13 shows the locations of the sensor assemblies on a vertical section map of the mine. The sensors and telemetry modules for each underground assembly were mounted on a panel and prewired to simplify instal- lation (fig. 14). Minor difficulties were encountered during startup of assem- blies A and B, owing to telemetry mis- alignment resulting from transportation and handling of the control console. Several solid-state electronic components in the control console were also damaged during installation from transients on the transmission lines. Repairs were made by replacement with spare compo- nents, and normal system operation was achieved. Performance of sensor assemblies A and B during the initial 7-month test period was generally good. Day shift, afternoon shift, and night shift blasts were easily distinguishable on the chart recordings. Following blasts, typical CO levels were 5 to 10 ppm and typical CO2 levels were 450 to 650 ppm. The diffusion-type CO detector on the 3600 level tracked close- ly with the sample pump unit on the 2550 level, with peaks occurring at corre- sponding times on the chart recordings. However, the levels of CO indicated by the diffusion-type sensor were a factor of 2 to 3 times higher than those of the pump-type unit. It is not clear whether these differences were real (actual CO levels that varied from area to area) or the result of instrumentation or teleme- try errors. Blasts were also clearly delineated on the smoke detector chart recordings; however, the actual particulate levels (percent obscuration, particles per cubic meter, etc. ) were uncertain because No. 7 No. 5 No. 8 No. 3 No. 4 shaft shaft shaft shaft shaft No. 6 shaft Sensor assembly B S .^Sprisnr Sensor—^ ^Sensor assembly A assembly D 1,600 Horizontal scale, ft FIGURE 13.— Section through test mine showing locations of sensor assemblies. 21 FIGURE 14.— Sensor assembly B, prewired and mounted on panel. calibration curves relating detector out- put to particulate levels were not avail- able from the detector's manufacturer. Maintenance frequency for the CO de- tectors was higher than desired. Each unit required zero adjustment and filter cleaning weekly, and the mechanical sam- ple pump was oiled about every 2 months. However the span calibration was per- formed only once during the 7-month peri- od, and that was during the week of in- stallation. Maintenance frequency for the CO2 and smoke detectors was lower than the design goal of twice per year. In fact, no maintenace, repair, or cali- bration was performed during the entire 7-month initial test period. Intermittent telemetry failures and false alarms occurred as a result of electromagnetic interference with signal transmission. The telemetry lines used for transmission of detector outputs (ex- isting mine wiring was used for this purpose) were not shielded and were occa- sionally run in close proximity to high- voltage power cables. The interference was traced to radiation from these lines. Rerouting the telemetry cables a few feet away from the power lines corrected the problem, but its intermittent nature made troubleshooting a difficult and time-con- suming process. At the midpoint of the initial 7-month test period, a fire test was conducted to evaluate the performance of sensor assem- blies A and B. The site selected was on the 3600 level, about 650 ft upstream of sensor assembly A. Airflow through the 22 fire area was 30,400 ft 3 /min. This air combined with a 33, 100-f t 3 /min split up- stream of sensor assembly A. Total flow past sensor assembly B, consisting of the 63,500 ft 3 /min from the 3600 level and 75,500 ft 3 /min from the 3400, 3200, 3000, and 2550 levels, was 139,000 ft 3 /min. Combustion products from the fire were thus carried by the mine's ventilation past both sensor assemblies. Figure 15 depicts the layout for the fire test. Permission to conduct the test was obtained from the U.S. Mine Safety and Health Administration, a rescue team was assembled, fire extinguishers and water were supplied to the test area, and as an added precaution, no personnel were per- mitted downstream of the fire. The fire was built in the bottom of a 55-gal drum that had been cut in half. The drum had holes along the bottom edge to draw air, and a screen was laid over the top as a spark arrestor. The primary combustible material consisted of 30 lb of 1-in scrap lumber. The wood was doused with char- coal lighter fluid and ignited (figs. 16- 17). After 25 min, a piece of flexoid and 18 in of 1-in water hose was added. Ten minutes later, 6 oz of rock drill oil was added. After 65 min, all materials had been consumed and the residue was doused with water. Figures 18 and 19 show the chart traces recorded during the fire test. Table 3 shows a comparison between CO and CO 2 levels measured by the detection system, and results of a laboratory analysis of air samples collected near sensor assem- bly A on the 3600 level. Values at the 2550 level are adjusted in the figures FIGURE 15.— Layout for In-mine fire test of second-generation spontaneous combustion fire detection system. 23 and table to reflect airflow differences (i.e., dilution) between sensor assem- blies A and B. All three traces show that the detec- tors on the 3600 level responded 3 to 5 min earlier than the detectors on the 2550 level (corresponding to the 1,100-ft separation between the sensor assem- blies). The measured gas concentrations show a slight variance from the labora- tory determinations; however, absolute agreement would not be expected under these test conditions (because of the possibilities of slight telemetry mis- alignment, imperfect mixing of air in the drifts, "non-plugged" flow, minor re- corder paper misalignment, ventilation uncertainty between the 3600 level and 2550 level, etc.) FIGURE 17.— Test fire burning. 40 30 20 "i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — r A, CO Rubber hose added to fire 3600 J I I I I I I I I I I I I I I I L FIGURE 16. Igniting test fire. 600 I I I I I I I I I I I I I I I I ' ' -25-20 -I5-I0 -5 5 10 15 20 25 30 35 40 45 50 55 60 65 TIME, min FIGURE 18.— CO and C0 2 measured during fire test. 24 80 i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — r Rubber hose/ added to fire/ 20 J I I I I I I L J I I I I I L -25 -20 -15 -10 5 5 10 15 20 25 30 35 40 45 50 55 60 65 TIME, min FIGURE 19.— Smoke measured during fire test. The two smoke detector traces track closely, with the detector on the 3600 level recording higher readings than the more distant detector on the 2550 level. As noted earlier, the diffusion-type CO sensor on the 3600 level recorded higher readings than its pump-type counterpart on the 2550 level. However, it is note- worthy that both the CO and C0 2 sensor pairs responded in similar proportions over the test period. E.g. , between 2 and 15 min after the fire was ignited, both CO sensors increased by about 80 pet whereas both C0 2 sensors increased by 17 pet. It is also noteworthy that the addi- tion of flexoid and hose to the fire are clearly distinguishable on the CO and smoke recordings. As these materials burn with a high emission of CO and smoke, but little C0 2 , a positive deflec- tion of the C0 2 trace was not expected. Following the initial 7-month test, sensor assemblies C and D plus the re- maining telemetry and control equipment were installed. Several faulty circuit boards were discovered in the new teleme- try panels when the system was energized, and the faulty boards were returned to the laboratory for repair. Before the circuit board repairs were completed, however, mine production was halted at the test mine due to depressed economic conditions and the low price of copper. The shutdown was indefinite, pending an improvement in the copper market, and it resulted in an almost immediate suspen- sion of activity related to the operation and maintenance of the detection system. Although project personnel eventually succeeded in completing the repairs to the system and achieving proper system operation, further testing of the system was impractical. Mine personnel re- quested that the equipment be retained at the mine site to supplement manual fire bossing; however, adequate staff to main- tain the system was not available. Project personnel visited the mine about 5 months after the shutdown to as- sess system operation under circumstances of almost no maintenance or calibra- tion. System operation was found to be TABLE 3. - Comparison of CO and C0 2 concentrations measured by detection system versus laboratory analysis Gas concentration, ppm Change in 2 min after 15 min after concentration, 1 ignition ignition pet CO analysis: 4 7.2 80 12 22 83 11 31 180 C0 2 analysis: 650 760 17 655 770 18 600 700 17 1 Between 2 and 15 min after ignition. marginal. The detection instruments at the one sensor assembly inspected (D on the 3,400 level) were only slightly out 25 of calibration, but telemetry problems resulted in inaccurate readings on the surface. SUMMARY AND CONCLUSIONS The potential seriousness of spontane- ous combustion fires in deep metal mines prompted the Bureau to perform research to upgrade technology for detecting such fires. The research was performed in three parts, beginning wi.th a laboratory study of induced spontaneous combustion events to identify and quantify reliable indicators of a spontaneous combustion fire in a metal mine. The recommended spontaneous combustion fire detection system included CO, CO2, oxygen, SO2, and temperature detection. The second stage of the program in- volved the assembly and evaluation of a prototype detection system. Tests were performed in the laboratory and in a deep metal mine with a history of spontaneous combustion problems. After 5 months un- derground, the system was functionally tested to evaluate response to abnormal contaminant levels. An induced spontane- ous combustion event was staged near the detection instruments to provide the re- quired contaminants. Of the five parame- ters monitored, only CO and CO2 were de- tected above background levels. The final stage of the program involved the design and long-term testing of a second- generation spontaneous combustion detec- tion system. Based on the previous stud- ies, CO and CO2 detection were included in the second-generation system; however, because of their relative inactivity dur- ing in-mine tests, the temperature, oxy- gen, and SO2 detectors were omitted. Submicrometer particulate (smoke) detec- tion was added, as the laboratory studies indicated that measurable smoke may be generated during a low-temperature spon- taneous combustion event. In-mine test- ing of the second-generation system in- cluded a staged fire test 3-1/2 months after installation. However, the test mine was forced to shut down, owing to economic conditions, before long-term endurance testing was completed. Two of the system's four sensor assemblies oper- ated for about 1 yr. The other two as- semblies received only a 3-week trial be- fore testing was halted. Overall, second-generation system per- formance was satisfactory, and installa- tion of similar systems in mines with a high risk of spontaneous combustion is recommended. The system demonstrated a capability to detect low levels of com- bustion products believed to be reliable indicators of spontanous combustion in metal mines. Principal problems and rec- ommended actions include (in descending order of significance) — 1. The method of recording detector outputs (chart recorder) resulted in the generation of a large amount of data that needed to be manually analyzed for slow- ly developing trends. Computerized data storage and analysis is recommended to reduce labor requirements and possible data analysis errors. 2. Both in-mine tests highlighted the need for controlling interference on te- lemetry lines. Shielded telemetry cables and placement away from sources of elec- tromagnetic interference are recommended. 3. Although CO2 gas detection is rec- ommended, the nonlinear output of the CO 2 detector used in this test program made data analysis difficult. Linearization of the CO2 output is recommended. If a computerized data storage and analysis system is used to record and analyze detector outputs, linearization of the CO2 unit's output could be incorporated into this system. 4. The outputs of the smoke detectors used in this test program could not be correlated with commonly used quantita- tive measures of smoke exposure (obscura- tion, particles per unit volume, etc.). Standardized testing to develop an appro- priate calibration curve for this detec- tor is recommended. U.S. GOVERNMENT PRINTING OFFICE: 1 987 605-01 7 '60044 INT.-BU.OF MINES,PGH.,PA. 28509 244 U.S. Department of the Interior Bureau of Mines— Prod, and Distr. Cochrans Mill Road P.O. Box 18070 Pittsburgh. Pa. 15236 OFFICIAL BUSINESS PENALTY FOB PRIVATE USE. MOO "2 Do not wi sh to receive thi s material, please remove from your mailing list* ]] Address change. Please correct as indicated* AN EQUAL OPPORTUNITY EMPLOYER ■Ox-v V* ^ ^ V c ° " ° « "* A°«. ** ne ? i: %^ •»&• X,^ ^M^ V\ « • • ., .4 0* ^ X/^^fc-S.^ *$£k*\ X-^fe-X y*JB/&*\ ^-^fe-% V * • „ «y V* » ►* v % °JpP« ; «/\. vSk** ♦**% \v&$> : ,^f \ "-s^ : ^ ^ ^ /,^v>o A^-V y..^;% o y,.^A .V«* .«* ^ %•« ^o^ °*S^®~'. ^o>* v ^ u % ->a ALT ^ve. - A.iak.V »* o ?Vt' .^ •^^ ♦"♦, °wwiw: • 4 1 * ,0 r /«6* ^°A v" . *? +< \*^/ \' % ^\^ V* 7 ^** 00 ?x.' ,0 /.^^•X ^*..3iai&.V. ,//^i-.% .'°*,^ ^ .^^L% % b ^,^v^ :- ^o« • ^-o^ .' ^ ,^°- ^ V\^ V\^ ■ ■ ■ I At ■ ■ ' f. ■ ,* # , ■ ■ I m LIBRARY OF CONGRESS 002 951 01 2A i '$i ■ \s ■ H