'£...♦ < V v^*^ ,v^^-/ \-^\/ -ojm--/ \^'/ v^v \'^> :- K^ X v''> 0^ r^ ^A **'% '^<,' v .•iq .^^ ,.^'% ^-^^.i^ \> .^^ V '^•b J'^'^ V ^ .^^ •^^o< ^■^..z ^^ A: j."^ '^^.v^ o .0; .^^ " <^ ^^■ ' M ' ,0 ^%0. ^ d^ ts7 Library of Congress Cataloging in Publication Data; Yenchek, M. R. (Michael R.) Evaluation of sensitive ground fault interrupters for coal mines. (Information circular / United States Department of the Interior, Bureau of Mines ; 9057) Bibliography: p. 15. Supt. of Docs, no.: I 28.27: 9057. 1. Coal mines and mining— Electric equipment. 2. Electric cir- cuit-breakers. I. Ackerman, Melvin N, II, Title. III. Series: Infor- mation circular (United States. Bureau of Mines) ; 9057. TN295.U4 [TN343] 622s [629'. 8] 85-600192 ■A "j^ CONTENTS Page Q Abstract. 1 "13 Introduction 2 N^ Ground fault protection in U.S. mines 2 ^^Description of evaluated ground fault interrupters 3 \. General Electric ground break relay 3 V) GBS Harrison ground fault detector. 4 Mindel ground fault circuit interrupter 4 Test rationales and results,... 5 Proper design and construction , 5 Applicable military standards 5 Mineworthiness 6 Size limitations 6 Electrocution prevention 7 Tests at 60 Hz 7 Power harmonics 8 Transient immunity 10 Voltage surges 10 Common mode transients 12 Current withstand 13 Reliability 13 Quality assurance 13 Safe failure modes 14 Summary and conclusions 14 References 15 ILLUSTRATIONS 1 . Zero-sequence relaying 2 2. General Electric type MC ground break relay 3 3. GBS Harrison type GF2B ground fault detector 4 4. Mindel model 21-7000 Shok-Blok ground fault circuit interrupter (older version) 5 5. Mindel model 21-7000 Shok-Blok ground fault circuit interrupter (newer version) 6 6. GFI dimensions 7 7. Ventricular fibrillation threshold at 60 Hz 8 8. Test setup to determine tripping characteristics at 60 Hz 8 9 . GFI response at 60 Hz 8 10. GFI frequency response 10 11. Frequency-response test circuit 10 12. Basic impulse generator circuit 12 13. Relay control circuit for impulse generator 12 14. Current withstand test circuit 13 >S TABLES .C^J^l. General Electric: Tripping characteristics at 60 Hz 9 jvN2. GBS Harrison: Tripping characteristics at 60 Hz 9 3. Older Mindel: Tripping characteristics at 60 Hz 9 4. Newer Mindel: Tripping characteristics at 60 Hz 9 5. General Electric: Frequency response data 11 ii TABLES — Continued 6. GBS Harrison: Frequency response data. 7. Older Mindel: Frequency response data. 8. Newer Mindel: Frequency response data. 9. Winding resistance 10. Current ratio tests 11. Summary of results Page 11 11 11 13 13 14 UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT A ampere ms millisecond °C degree Celsius ys microsecond Hz hertz mV millivolt h hour fi ohm in inch s second kHz kilohertz V volt kfi kilohm VA volt ampere kV lb kilovolt pound V ac volt, alternating current mA milliampere W watt PF microfarad yr year UH microhenry EVALUATION OF SENSITIVE GROUND FAULT INTERRUPTERS FOR COAL MINES By Michael R. Yenchek and Melvin N. Ackerman ABSTRACT Contacts with energized conductors are a major cause of electrocutions in underground coal mines. Sensitive ground fault interrupters (GFI's) installed on in-mine three-phase ac utilization circuits would probably prevent the majority of these deaths. A sensitive GFI is a protective device that detects and interrupts small deadly ground currents in the milliampere range before those currents can cause ventricular fibrilla- tion in humans. Commercially available three-phase sensitive GFI's have not been specifically designed for application in coal mines. The Bu- reau of Mines therefore tested three commercial GFI models to determine their worthiness for mine power systems. GFI design and construction, transient immunity, reliability, and time-current characteristics were evaluated in laboratory tests. No commercial device was found suitable for mine use without design modifications. The tests results will serve as a basis for the development of a mineworthy sensitive GFI in ongoing Bureau research. Electrical engineer. ^Electrical engineering technician (retired). Pittsburgh Research Center, Bureau of Mines, Pittsburgh, PA. INTRODUCTION In a study completed in 1981, 307 sepa- rate accidents in the coal industry dur- ing a 3-year period were attributed to electric shock, from contacts with ener- gized conductors (J^).-^ These accidents resulted in seven fatalities and the loss of over 5,000 person-days from work due to nonfatal injuries. The majority of the nonfatal injuries and nearly all the electrocutions could have been eliminated if ground fault protection, designed to protect people, had been installed on the coal mines' resistance-grounded systems. The grounded phase protective devices presently used on these power systems are inadequate from a shock-prevention stand- point. Typical relay current pickup or response levels are in the ampere range, considerably in excess of the electrocu- tion threshold. Increasing the sensitiv- ity of these electromechanical devices results in undesirable nuisance tripping and unscheduled interruptions of coal production. What is needed is a sensi- tive GFI that identifies and inter- rupts the small deadly ground currents that can electrocute people, yet ignores spurious signals such as those from motor startups. Criteria have been established for the use of sensitive GFI's on low-voltage ac utilization circuits in U.S. mines (2). These practical guidelines include specific recommendations concerning GFI design and construction, transient immu- nity, reliability, and time-current char- acteristics. This report documents tests conducted by the Bureau of Mines in ac- cordance with these criteria using com- mercially available three-phase sensitive GFI's. GROUND FAULT PROTECTION IN U.S. MINES The majority of U.S. mine power systems employ radial distribution, wherein the supply power branches out through switch- houses and terminates at utilization points throughout the mine. The utili- zation system includes power centers, rectifiers, cables, motors, and the asso- ciated protective devices. It is the most troublesome part of the power sys- tem in terms of safety and reliability due to its temporary nature. As mining advances, the utilization system is stretched to its limit and then reposi- tioned. Thus, the circuit protective de- vices must adapt to constantly changing conditions . Typical ground fault protection in min- ing consists of high-resistance grounding and ground fault protective relaying. The resistance Inserted between the sys- tem neutral and ground limits the fault current and energy dissipation. The re- lay monitors the circuit and removes — > ■ — — — -•Underlined numbers in parentheses re- fer to items in the list of references at the end of this report. power upon indication of a hazardous current flow. In present U.S. mining systems, the current permitted, and the current required to operate electro- mechanical relays , can create a personnel hazard before power is removed. Zero-sequence or balanced-flux relay- ing (fig. 1) is the most reliable and most common method employed for ground fault relaying. As shown in figure 1, the phase conductors pass through the Phase Current transformer B C- Zero- sequence relay Ground wire FIGURE 1. - Zero-sequence relaying. current transformer (CT) window. The sum of the three phase currents is the CT primary current and is proportional to the zero-sequence current O) . In an unfaulted balanced system, there is lit- tle or no zero-sequence current, and the CT secondary current is approximately zero. However, when a ground fault occurs , the resultant secondary current is used to trip a relay. Zero-sequence relaying is unaffected by phase voltage fluctuations, and, since only the ground leakage current is monitored, the relay can be made very sensitive. All of the commercial GFI's evaluated in this paper were the zero-sequence type. DESCRIPTION OF EVALUATED GROUND FAULT INTERRUPTERS Tests were conducted using three com- mercial sensitive GFI's identified under Bureau contract JO 113009 (2^) as having potential applicability to mining. GENERAL ELECTRIC GROUND BREAK RELAY The General Electric Co. (GE) type MC ground break relay (fig. 2) was I I I I I I I I I I I I I I I FIGURE 2. - General Electric type MC ground break relay. specifically designed to protect mo- tor circuits from ground faults. The TMCGS200GT current sensor provides a min- imum tap setting or pickup of 100 mA. Its rated response time is 300 ms at 150% of trip level. The relay is equipped with a button for resetting following clearance of a ground fault and a button to test relay operation, GBS HARRISON GROUND FAULT DETECTOR The GBS Harrison Ltd. type GF2B ground fault detector (fig. 3) was developed by the National Coal Board for coal mines in the United Kingdom, It has a rated sen- sitivity of 90 mA ±20% at 20° C and a rated trip speed of less than 100 ms at 150% of trip current. A small light- emitting diode (LED) indicates the pres- ence of control power to the unit. A re- mote pushbutton can be added for periodic testing. MINDEL GROUND FAULT CIRCUIT INTERRUPTER The Mindel Corp.'s model 21-7000 Shok- Blok ground fault circuit interrupter (figs. 4-5) was originally designed by Thomas Gross, a consulting engineer and holder of patents for several ground fault detection techniques. The Mindel relay has been used by the irrigation industry, but only recently has been mar- keted for use in coal mines. The first units tested (fig. 4) were housed in plastic boxes with a test button on the front. Some featured a knob to adjust sensitivity from 20 to 100 mA. Following failure of two original units, a newer version of the model 21-7000 was also tested (fig. 5). It consisted of a plug- in module with test and reset buttons. It had a rated sensitivity of 60 mA with a rated maximum response time of 3 s. l^^-A.,4. GBS HARRISON LTD GROUND FAULT DETECTOR TYPE Gf2B POWER (SfNSlTIVITV 80-tOOM*) M*Of IN tUGLAND (120V AC i _^_ _^ fCOMTAC-S JSa VAC SAMP) FIGURE 3. - GBS Harrison type GF2B ground fault detector. lOOO ftOO*' e»»***" FIGURE 4. - Mindel model 21-7000 Shok-Blok ground fault circuit interrupter (older version). TEST RATIONALES AND RESULTS PROPER DESIGN AND CONSTRUCTION Applicable Military Standards Rationale Proper design and construction will re- duce the amount of downtime caused by GFI failures and thereby help bring about ac- ceptance of the GFI as a useful safety item. Electronic instruments designed and constructed for military use must comply with Military Standard 454 ( 4_) . The following two summarized portions of that standard can be applied to GFI's used in underground mining: Safety Hazard The design shall incorporate methods to protect personnel from accidental contact with voltages in excess of 30 V RMS or dc during normal operation. All external surfaces shall be at ground potential during normal operation. All terminals shall be corrosion-resistant. Sharp ex- ternal projections shall be avoided. Accessibility Suitable access shall be provided for adjustments, testing, and routine main- tenance. No unsoldering shall be nec- essary to remove the front cover for troubleshooting. Findings All of the relays were housed in plas- tic cases, and internal adjustments were unnecessary in any of them. Only the older Mindel relay did not have exposed terminals energized at 120 V ac. Access is required to change a fuse in this relay, but the front case is easily removable without unsoldering. The relay FIGURE 5. - Mindel model 21-7000 Shok-Blok ground fault circuit interrupter (newer version). components of the newer Mindel relay were mounted in a convenient plug-in module. The terminals on both the GE and Harri- son relays corroded badly when coated with acetic acid for 24 h. Mineworthiness Rationale Underground, GFI's are located inside metal-clad load centers, so both the re- lay and CT should have metal mounting lugs. Terminal strips should be sized for No. 12 AWG wire. In addition, the relay case should be moisture- and dust- resistant. Findings Only the Harrison GFI was equipped with metal lugs on both the CT and relay. However, it was also the only model with undersize wire terminals. All relays were housed in moisture- and dust-resistant cases. Size Limitations Rationale Space is limited in typical mine power equipment. Since several GFI's may be used in a single power center, they must not be much larger than present ground fault relays. Thus, the relay components should be mounted in a compact enclosure not exceeding 3 by 6 by 6 in. To minimize leakage flux, the CT window should only be large enough to accommo- date the encircled power conductors. Trailing cable size is limited to No. 4/0 AWG to facilitate handling underground. The outside diameter of a 4/0 single con- ductor cable is 0.807 in O ) . Three such cables fit snugly through a 1 .750-in-diam windows. For ease of installation of cables with terminals, this value should be increased to 2.100 in. Present ground fault CT's in use under- ground have outside diameters smaller than 4 in. Since they are placed between the molded-case circuit breaker and the load-center coupler, they are no more than 3 in wide. Findings The dimensions of the CT's and relays are given in figure 6. All of the CT's have undersized windows, and only the outside diameter of the GE CT did not ex- ceed 4 in. All relay enclosures except the older Mindel were appropriately sized. CURRENT TRANSFORMER RELAY DO /a U|5/8-^' 3'/-;' General Electric \-^x^ n ■ — X^A 1^/4 h Harrison =0^ VA © 6 /4 1 1 1 1 I M 1 1 43/4" -4 'A" Mindel (older version) J nn n h-2VH Mindel (newer version) FIGURE 6. - GFI dimensions. ELECTROCUTION PREVENTION Tests at 60 Hz Rationale The primary reason for employing sen- sitive GFI' 8 in mining is to prevent ac- cidental electrocutions. The resultant cause of death in these instances is ven- tricular fibrillation. In this condi- tion, the normal heartbeat stops, and the ventricles twitch irregularly. The 60-Hz threshold has been statistically defined as the current through the chest that will produce ventricular fibrillation in 1 out of 200 people. For 110-lb individ- uals, this threshold can be expressed as I = 116//t (6^), where I is the current in milliamperes and t is the time in seconds. This re- lationship is shown graphically in figure 7. The safe area, to the left of the plotted line (in figure 7) , is the desired region for GFI operation. It should be noted that the equation given above is only valid for shock durations of less than 5 s. Procedure A variable voltage source in series with a 50 ft, 225-W fixed resistance was used to inject a 60-Hz current through each CT primary as shown in figure 8. A double-pole single-throw switch initiated the test and triggered the storage oscil- loscope. Test currents were varied from to 1,000 mA. Results The test data are listed by manufac- turer in tables 1-4. The data were aver- aged and plotted with the ventricular fibrillation threshold superimposed, as shown in figure 9. Statistically, cur- rents less than 50 mA should not cause ventricular fibrillation. In light of this, and since ground currents on a typ- ical resistance-grounded system protected by a sensitive GFI would be limited to E III 50 100 FIGURE 7. at 60 Hz. 500 1,000 5,000 CURRENT, mA Ventricular fibrillation threshold 50 100 500 1,000 5,000 CURRENT, mA FIGURE 9. - GFI response at 60 Hz. models lacked the sensitivity necessary for protection against 50-mA. faults. The curves depict relay operation time only. To obtain the total clearing time, about 32 ms should be added, to account for opening of the molded-case circuit breaker. Resistor -^WV — ^To oscilloscope >K ^ trigger C r/\ Voltmeter Current I transformer •-Current probe to oscilloscope channel 2 I'll To channel 1 Relay FIGURE 8. - Test setup to determine tripping characteristics at 60 Hz. 500 mA, the area of interest lies between 50 and 500 mA. Close examination revealed that only the Mindel GFI's provided complete pro- tection against electrocution at 60 Hz. However, the older Mindel, with a pick- up of 8 mA, was judged to be overly sensitive. Both the GE and Harrison Power Harmonics Rationale The filtering for GFI's must be de- signed so as to preclude false tripping by any harmonics superimposed on the power conductors. However, attenuation of these higher frequency currents must not be so severe that hazardous currents are permitted to flow. The ventricular fibrillation threshold for humans as a function of frequency has been extrapo- lated from experiments with animals (7). As shown in figure 10, this threshold is at a minimum at about 60 Hz and increases exponentially with frequency. Procedure An audio oscillator and power amplifier provided high-frequency test currents from 60 Hz to 10 kHz, as shown in fig- ure 11. For each frequency, the voltage TABLE 1. - General Electric: Tripping characteristics at 60 Hz I, mA Time , ms Unit 1 Unit 2 Unit 3 Unit 4 Unit 5 Unit 6 ND ND ND ND ND ND ND 7,200 5,400 5,700 6,000 6,700 4,700 900 1,000 1,100 900 1,200 1,000 800 350 500 350 750 100 180 120 100 200 250 70 40 40 40 50 50 30 25 23 30 40 25 20 10 15 25 30 15 20 10 13 25 25 20 100^., no.., 125.., 150.., 200.., 300.., 500.., 800.., 1,000. I Current. ND No data; relay did not activate. ^ Rated sensitivity of General Electric relay. TABLE 2. - GBS Harrison: Tripping characteristics at 60 Hz I, mA Time , ms Unit 1 Unit 2 Unit 3 Unit 4 Unit 5 Unit 6 90 540 50 50 50 50 50 50 50 50 50 50 50 ND 310 88 50 56 50 50 50 50 50 50 50 760 390 50 50 50 50 50 50 50 50 37 40 1,100 720 500 50 43 50 50 43 40 40 43 30 600 280 63 45 55 50 50 45 45 40 40 35 120 95 80 100 60 120 55 160 50 200 45 250 45 300 400 40 40 600 35 800 35 1,000 35 I Current . ND No data; relay c id not a LCtivate. TABLE 3. - Older Mindel: Tripping characteristics at 60 Hz I, mA Time , ms I, mA Time , ms Unit 1 Unit 2 Unit 3 Unit 4I Unit 1 Unit 2 Unit 3 Unit 4I 10 ND ND 900 75 100 50 65 45 45 15 540 560 300 35 300 40 50 35 35 25 220 180 100 25 600 40 50 35 35 50 75 95 65 20 1,000... 40 50 35 35 I Current. ND No data; relay did not activate. ^Pickup adjusted to maximum setting. NOTE. — Units 5 and 6 failed prior to this test. TABLE 4. - Newer Mindel: Tripping characteristics at 60 Hz 50., 60.. 75.. 100. 150. I , mA Time , ms ND 2,500 1,100 500 250 I, mA 200... 300... 500.., 800... 1,000. Ti me , ms 150 80 45 35 33 I Current. ND No data; relay did not activate. 10 0.1 I FREQUENCY, kHz FIGURE 10. - GFI frequency response. Current transformer Frequency counter Frequency generator \®H=^ FIGURE 1 Light Ammeter ^ Frequency-response test circuit. attenuated higher frequency currents to the extent that ventricular fibrillation would be possible. The GE model had no filtering and yielded a flat response. The frequency characteristics of new Min- del units lie close to, but always on the safe side of, the fibrillation threshold. TRANSIENT IMMUNITY was slowly increased until the relay activated. Results The test data are tabulated in tables 5-8 for each manufacturer, A plot of the averaged values for each relay is shown superimposed on the allowable attenuation curve in figure 10, The results indicate that the filtering in the Harrison and older Mindel models Voltage Surges Rationale Mine power systems frequently excper- ience voltage surges when circuit break- ers and switches are opened or closed. Although the duration of these tran- sients is quite short — a few 60-Hz cycles — past research indicates their magnitude can reach 5 per unit crest voltage at utilization levels (8), These 11 TABLE 5. - General Electric: Frequency response data Freq, Hz 60...., 100 500 1,000., 3,000., 5,000., 8,000., 10,000, Unit 1 112 100 80 80 80 82 80 100 Current, mA Unit 2 105 100 95 90 92 95 100 105 Unit 3 105 100 90 90 90 90 95 95 Unit 4 105 100 90 90 90 90 95 95 Unit 5 100 95 85 85 85 85 90 90 Unit 6 100 95 85 85 85 90 95 95 TABLE 6. - GBS Harrison: Frequency response data Freq, Hz 60 100.., 200.., 500.., 800.., 1,000. Current , mA Unit 1 90 180 400 1,000 1,650 1,950 Unit 2 95 180 390 1,050 1,600 2,000 Unit 3 90 170 350 1,000 1,700 2,000 Unit 4 85 160 320 950 1,500 1,900 Unit 5 90 180 380 1,000 1,550 2,000 Unit 6 90 180 380 1,050 1,500 2,000 TABLE 7 . - Older Mindel: Frequency respons e data Freq, Hz Current , mA Freq, Hz Current , mA Unit 1 Unit 2 Unit 3 Unit 4I Unit 1 Unit 2 Unit 3 Unit 4I 60 100 200 14 13 33 12 15 40 9.5 10 28 10 13 31 500 1,000... 3,000... 130 420 1,300 200 700 1,350 110 500 1,350 100 450 1,200 ^Pickup adjusted to maximum setting. NOTE. — Units 5 and 6 failed prior to this test. TABLE 8. - Newer Mindel: Frequency response data Freq, Hz I, mA Freq, Hz 60 55 100 55 200 68 I Current. I, mA 500 160 1,000 340 3,000 1 ,400 surges , when present on the power conduc- tors encircled by the GFI CT, should not falsely activate the relay. In addition, they should not damage the relay control circuitry. Procedure The impulse tester used to generate voltage transients was constructed in accordance with Section 19A of Under- writers' Laboratories' Standard 943, "Ground Fault Circuit Interrupters." Consisting of a relay switch and resonant circuit, the tester simulates transient overvoltages as they would occur on resi- dential and industrial power systems. Schematics are shown in figures 12 and 13. The generated waveform exhibited the following characteristics under no load: (1) initial rinse time of 0.5 us between 10 and 90% of peak amplitude, (2) lO-ps period of following oscillatory wave. 12 0-9 kVdc -WV- 1-2 _nTn_ CR, -O I o- _mTi — ,1 12) rfn . ^ S-; h Relay control j ITT! Cathode roy oscilloscope trigger output ; Device i 1 under test ! 3 J Neutral ground Ci Capacitor 0.025^F; 10 kV Cg Copacitor 0.01 ,iF, |0 kV Cj Capacitor 4 jlF, 400 v L, Coil l5;iH (32 turns. No 23 wire, 0.7-in-diam air core) L2 Coil 70,iH (28 turns. No 23 wire, 0.6-in-diam air corel Tf Transformer l:l, 500 V-A S, Switch Ri Resistor 22 fl^ I W Rg Resistor 12 a. IW Rj Resistor 1.3 Mfl(l2xll0kfl, '/2W) R4 Resistor 4 7 kQ (10 x 47 kfl, '/2 W) /?5 Resistor 200 Q, I/2 W CRi Relay 2 normally open poles in series GE CR2790 E 100 A2 F, Fuse 3 A FIGURE 12. - Basic impulse generator circuit. Cathode ray oscilloscope gate input.^ SCRi rm KEY /?/ Resistor 10 kft, I W R2, R3 Resistor I kil, I/2 W Ci Capacitor 32 ^F, 250 V SCR, CR, Thyristor GE CI22B Relay GE CR2790 E 100 A2 Transformer Triad N4S X Di, Dg Diode IN5060 FIGURE 13. - Relay control circuit for impulse generator. and (3) amplitude of each successive peak 60% of the preceding peak. The amplitude of the first peak was fully adjustable from to 8,000 V. In the first part of the test, ten successive 5-kV surges were imposed on the power conductors encircled by the CT while the relay contacts were observed. Next, ten 1-kV impulses were applied in parallel with the 120-V ac control voltage and at random with re- spect to its phase. Afterward, relay op- eration at 60 Hz was checked to confirm possible circuit damage. Results Application of the 5-kV surge did not affect the GE and Harrison relays. How- ever, the Mindel relays were activated momentarily following each impulse. This may be attributed to their lower pickup. All relays operated satisfactorily fol- lowing the 1-kV surge to the control circuits. Common Mode Transients Rationale Sensitive GFI's used on coal mine power systems must be unaffected by the large transient currents common to all phases of ac utilization circuits. Such cur- rents may briefly exceed six times full motor rating during startup or heavy in- termittent loading. The maximum short- circuit settings listed in 30 CFR 75.601 effectively limit balanced three-phase loading to 2,500 A. Nevertheless, bal- anced currents up to 2,500 A should be tolerated for up to 5 s without actuation of the relay. Procedure The motor test station in the Bureau's Pittsburgh, PA, Mine Electrical Labora- tory was used to variably supply balanced three-phase currents through a trailing cable encircled by the GFI CT. The load consisted of three 0.3 J2 resistors con- nected in a delta configuration. The supply voltage was gradually increased until the relay activated or the 2,500-A ceiling was reached. Tripping thresholds were confirmed through repeated tests. Results Only the Harrison relay was immune to common mode transients. All the GE units tripped at between 200 and 250 A, while the older Mindel GFI's required only 125 to 200 A to operate. The newer Mindel relay did not actuate until 1,700 A was reached. 13 Current Withstand Rationale The molded-case circuit breakers used on low-voltage ac mine power circuits typically have an interrupting rating of 30,000 A. Currents near this magnitude are quite possible for three-phase faults. Since the GFI CT is a part of the power system, it too must withstand up to 30,000 A for the time it takes the breaker to clear (a few cycles) . High- current source To storage oscilloscope 1 Ar Shunt Mine duty circuit breaker Current \ transformer Secondary shorted FIGURE 14. - Current withstand test circuit. Procedure The withstand test was conducted using a high-current circuit breaker tester, as shown in figure 14. The tester was equipped with an initiate switch that could be jogged to reasonably control the test duration. Current magnitudes were recorded on a storage oscilloscope con- nected across a 400-A, 100-mV shunt. The CT secondaries were shorted to preclude high secondary voltages. Each was sub- jected to 30,000 A for approximately 4 cycles. The 60-Hz current ratios and winding resistances were measured before and after the withstand test to detect any degradation of the CT. Results The winding resistance and current ratios listed in tables 9 and 10 did not change after the withstand test. This indicates that all the CT's safely tolerated 30,000 A for the time it took the breaker to clear. RELIABILITY Quality Assurance Rationale For dependability underground, all devices made by the same manufac- turer should operate in the same manner and have a reasonable service life. In TABLE 9. - Winding resistance, 9. Unit GE GBS Harrison Older Mindel Newer Mindel 1 0.5 .6 .5 .5 .5 .5 1.3 1.3 1.4 1.4 1.4 1.3 1.2 1.3 1.2 1.3 1.2 1.7 1.3 2 3 4 5 NAp Nap NAp NAp NAp 6 NAp Not applicable. TABLE 10. - Current ratio tests (secondary shorted), mA Manufacturer Primary Secon dary Unit 1 Unit 2 Unit 3 Unit 4 Unit 5 Unit 6 General Electric 50 500 0.13 3.68 0.38 4.02 0.38 4.03 0.38 4.07 0.38 4.06 0.39 4.07 GBS Harrison 50 500 .13 1.63 .18 1.64 .13 1.63 .13 1.64 .13 1.64 .13 1.64 Mindel (older version) 50 .71 .68 .68 .71 .71 .70 500 6.91 6.92 6.92 6.92 6.92 6.91 Mindel (newer version) 50 .35 ND ND ND ND ND 500 5.02 ND ND ND ND ND ND No data. 14 addition, each GFI should be equipped with a means to test its operation. Results There was considerable conformity in the frequency response data for all units. However, the time versus current results correlated well only above 150% of pickup. The older Mindel relays were judged to be unreliable, as two of the original six units purchased failed dur- ing the test program. Quality assurance judgments of the newer version could not be made since only one unit was tested. All of the relays evaluated featured test circuits to simulate ground faults. Safe Failure Modes Rationale In the event of failure of the GFl's internal circuitry, the unit should react to remove power, to prevent a false sense of security. Two common failure modes are loss of 120-V ac control power to the GFI and opening of the CT winding. Results Only the GE relays failed to operate when control power was removed or the CT winding opened. SUMMARY AND CONCLUSIONS Results of the sensitive GFI tests are summarized in table 11. They show that none of the commercial three-phase GFI's evaluated is suitable for underground mining without design modifications. Overall, the newer Mindel GFI came closest to compliance with the estab- lished criteria. Increasing the number of secondary turns on the Mindel CT would probably eliminate any false tripping during motor startups (common mode) . False tripping due to voltage surges on the power conductors may be eliminated by (1) increasing the CT burden impedance or (2) changing the rating of the back-to- back diode shunt that protects against transients coupled through the CT's. The unreliability of the older Mindel unit TABLE 11, - Summary of results General Electric GBS Harrison Older Mindel Newer Mindel Military standard com- Not corrosion- Not corrosion- Exposed at Passed. pliance (terminals). resistant. resistant. 120 V ac. Mineworthiness Lacked metal Undersize wire Lacked metal Lacked metal mounting lugs. terminals. mounting lugs. mounting lugs. Proper dimensions CT ID too small; CT inadequate; CT OD and re- CT OD too relay OK. relay OK. lay oversized. large; relay OK. Passed. Electrocution preven- Pickup too high. Pickup too Too sensitive.. tion at 60 Hz. high. Power harmonics May nuisance Too much Too much Do. trip. attenuation. attenuation. Voltage surges Passed, but pickup high. Passed, but pickup high. False tripped. . False tripped. Common mode Failed J o Failed; too sensitive. Failed- Current withstand Passed Passed ...QO.. ....... Passed. ........ Prl-^ " , , . « o,^ °> * " » ' aO ^- -^ v•••-^^^^^,.,V•'-^/ \/^-\/ v^-/ \/^\/ v^*/ ^~ ..^''*^^. .^^ ^^^ \^ ■3^r *'" > y -^i'^ ^°^:^^^"- ./^•^i>- ^°^:^:^'°- /' '°"° • ^^^ >^ o_ ^°^ ^^•^t. ^0" % ^ 0!,*°' ;>■ -> • o « » :^ "•'° aV' y -^^ ■^ iOv\ v' ..^l^Lr. "ci r.«* ,0^ .»* A <> *^.. .* ,0 .^^ -^„ * V 4.* .oHo^ '^^ r:^" .t<., "^o "' <'^'^ c"""* ^"**' r^'^" .t"« "^o J^" oo.l"* ■>> 0' t ^^-n^^ iOv\ C, vP M *•* .^" ..... ^-^^ '•' -^' -"- ^^^ ■*• 'bV" -^ C vP OHO' .O*^ V'^'^'\^^'' '\'*^'*aO'^ v-^^ ^^.-iv^ oV^^^Pi^^ -'^.^^- :m^_r^. ^o.\ <»t..-j^ ^o\y '^0^ «5°^ - 'bV" '>0^ HECKMAN BINDERY INC. ^^APR 86 W=^ N. MANCHESTER, ^^'^^ INDIANA 46962 V ^-p/ C, iP '>'^.-J^' oV'^mil^l'" "^^r^" ^'SM^^r^. ^r.-^" oV^^^'- '-'^^^^ " "O. A'' vP V • A c " " " * <^ •I ^oV liaii