Fault Characteristics in Electrical Equipment

Transcription

1 1. Introduction Proper design and installation of electrical equipment minimizes the chance of electrical faults. Faults occur when the insulation system is compromised and current is allowed to flow through an unintentional path. IEEE Std 493 Recommended Practice for the Design of Reliable Industrial and Commercial Power Systems 1 publishes data that reports that the vast majority of faults are classified as line-toground, or more simply ground faults. Industry data collected from published reports indicates that many faults are low level [1][2][3][4] [5][6][7][8][9]. Those reports describe equipment meltdown when the faults were not cleared by the overcurrent device in a timely manner. The reason that a low level ground fault can burn down a power distribution board while typically a phase-to-phase fault does not is that, counterintuitively, a low-current fault can release far more energy than can a high current fault. The reason is that for most overcurrent protective devices, lower level faults require more time to clear. As will be shown later, this slower clearing time causes much more energy to be released. This situation is particularly dangerous for ground faults as a ground fault has the potential to be a low-current fault. While ungrounded distribution systems would reduce the likelihood of a ground fault, ungrounded systems are almost never recommended due to the potential for damaging high line-to-ground voltages that can occur during intermittent ground faults [10]. 2. Published IEEE Ground Fault Failure Data Electrical equipment is very reliable. Even when looking at the systems that are not maintained or installed properly, failure rates for devices are typically fractions of 1% 2. However, should a work tool be dropped or an accidental short circuit occur, a fault may result. Faults that involve a ground path back to the source are called ground faults. IEEE Std 493 examined the total number of faults that occurred over a sample of equipment, and compared the quantity of those involving ground with those that didn t and calculated a percentage of each type of fault relative to the total. Table 1. Data Extracted from IEEE Std Table a b c Table 1 is a subset of Table from IEEE Std It shows those types of circuit elements that involve cable conductors. Focusing on the bottom section of this table (labeled Failure Type) for a moment, and using the Cable column as an example, note that 73% (marked with a a in Table 1) of all cable failures involve arcing faults to ground, 1% (marked b) involve arcing faults that don t involve ground, and the remaining 26% (marked c) don t involve an arcing fault 3. Focusing only on the arcing failures that involve a fault, we see that out of the 74% percent (73+1 out of 100) of failures that involve a fault, 99% (73 divided by 74 rounded up) involved ground. The remaining 1% (1 divided by 73%) did not. 1 Also known as the Gold Book. See reference [6] at end of paper. 2 IEEE Std Table 10-2 describes the expected failure rate of all electrical devices surveyed. 3 An example of a failure that does not involve a arcing fault would be a cable break (e.g. continuity failure) where neither of the two ends of the break touched an energized or, in the case of a loose energized conductor, grounded conductor. If no current flows after the failure, it is not considered a fault. Another type of fault would be a wiring error where conductors are accidentally wired as a short circuit. Such a fault does not involve arcing.

2 This IEEE 493 standard is stating that for a cable, it is nearly 100 times more likely (99% divided by 1%) for a cable fault to be a ground fault versus a fault not involving ground. From an analysis of the physical properties of how a cable fails this would seem to make sense. After all, a ground fault only requires one conductor s insulation to fail, while a phase-to-phase fault would require insulation to fail on two separate conductors at the same time. 3. Published Equipment Meltdown Reports Many authors [1][2][3][4][5] have reported electrical equipment being destroyed by persistent arcing faults. In each case, the reason for the extended clearing time was that the amount of current flowing was substantially less than bolted fault levels. Overcurrent protective devices like fuses and circuit breakers are designed to operate quickly at 3-phase bolted fault levels. Due to the inverse time current characteristics of these devices, they do not respond as quickly at reduced current levels. In the case of the melt-down, the fuse or circuit breaker did not malfunction. Rather the fault current was too low to clear the fuses or operate the breaker quickly enough to reduce the I2t energy to a sufficiently low level. It is a common misperception that low current faults release less energy than high current faults. In particular with fuses and to a lesser extent with circuit breakers, the exact opposite is true. So how much current flows during a ground fault? In section of IEEE Std 1584 [12], the authors explain that arcing currents, in general, are very difficult to predict. In lab testing the authors stated that for one test point the calculated current was between 45% and 50% higher than the measured arc current (emphasis added). In other words, the actual current was only ½ what was expected. This was for phase-to-phase fault with known circuit path impedance under controlled laboratory conditions! While not tested by the 1584 team, single-phase ground faults compound the uncertainty since with a ground fault, the current flows through other than phase conductors with known impedance. Ground impedance may include enclosure metal, ground bar 1, conduit or other material, much having a higher impedance than the phase conductors. The problem is that since current limiting fuses have a very steep time current curve, any reduction in the magnitude of current results in an increased clearing time 2, with typically the time increasing proportional to the reduction of current raised to the fourth power (I 4 ). In the example shown in Figure 1, reducing the current by one-half from 9000A, (i.e. 15x rating) to 4500A (7.5x rating) flowing through a 600A RK-1 fuse increases the clearing time by 25 times (0.5 s from 0.02 s). Figure 1 current = 2x decrease causes a time = 25x increase If we change the current by a factor of two, to what exponent must you raise that value to obtain a change of 25? Mathematically: 2 x = 25? To solve this equation for the unknown x, we take the logarithm base 2 of each side of the equation. Since most scientific calculators only include log base 10 and base e (ln), we must convert a logarithm from one base to another: log b (x)= (x) Eq. (1) (b) So starting with 2 x = 25 and taking the log base 2 of each side we have: log 2 (2 x )= log 2 (25), simplifies to x = (25) = = 4.64 (2) In other words, = 25, or the increase in clearing time is inversely proportional to the change in current raised to the 4.64 power. As can be seen, clearing time for a fuse increases rapidly as fault current drops. 1 The National Electrical Code (NEC) does not require that ground conductors be sized as large as phase conductors. 2 For circuit breakers, the current reduction would need to drop below the instantaneous pick up before the clearing time increases. 2

3 White Paper TP E How does this affect energy flowing through the fuse; the i 2 t? Since incident energy is proportional ( ) to current squared times time (i.e. I 2 t), a reduction in current reduces the energy by the square, but increases the time by the 4.64 power. Energy 9000A I 2 t = K (0.02) = 1.62x10 6. K amp 2 -seconds Energy 4500A I 2 t = K (0.5) = 1.01x10 7. K amp 2 -seconds 1.62x10 7. K = x10 6. K The result is that the energy increases by a factor of 6.25 times as the fault current magnitude is halved. What exponent describes this equation? 2 x = 6.25, x =? log 2 (2 x ) = log 2 (6.25) x = (6.25) = = 2.64 (2) Therefore, = 6.25 Eq. (2) Notice that the exponent for the change in energy (2.64) is exactly 2 less than the exponent for the magnitude of the current change (4.64). Why? From the equation E i 2 t it is clear that if the current is reduced, the energy contributed by the current is reduced by the square (i2). Likewise, as we ve just calculated, if decreasing the current increases the clearing time proportional to the 4.64 power, then the net result is the energy decreases by the square while simultaneously increasing by the 4.64 power or ( ) = Current Limiting versus Non-Current Limiting Does this increase only occur with current limiting fuses, and more specifically, only when the current-limiting fuse is operating in the current limiting region of the time overcurrent curve? No. Consider the non-current limiting region in Figure 2 bounded by the two current values of 4500A (7.5 times 600A) and 1000A (1.67 times 600A). For a current ratio change of 4.5 (4500A/1000A), the time changes by 1000x (500s/0.5s). As before, we solve the exponential equation as: 4.5 x = 1000 with x =? Transforming the logarithm base using Eq. (1): log 4.5 (x) = Solve for x: (x) (4.5) log 4.5 (4.5 x ) = log 4.5 (1000) or x = log (1000) 3 10 = (4.5) = 4.59 Calculating the change in energy at the two clearing times: Energy 4500A I 2 t = K (0.5) = 1.01x10 7. K amp 2 -seconds Energy 1000A I 2 t = K (500) = 5x10 8. K amp 2 -seconds When taking the ratio of the two energy values at the two current levels, the energy released at the 1000A current level versus the 4500A level was: 5x10 8. K 1.01x10 7. K = 49.5 times higher incident energy! 4.5 x = 49.5, x =? log 4.5 (2 x ) = log 4.5 (6.25) x = log (49.4) = (4.5) = 2.59 Eq. (3) Therefore, = 49.5 In other words, incident energy increased by a value proportional to the ratio of the current change raised to the 2.59 power even when not in the current limiting region of the fuse! Clearly it doesn t matter whether fuses are operating in the current limiting or non-current limiting region. In both cases reducing current flowing through the fuse increases energy released through the fuse 3. Figure 2 3 As mentioned earlier, this does not occur in circuit breakers until the fault current drops below the instantaneous pickup of the circuit breaker. Therefore, in non-current-limiting breakers, the incident energy reduces as the fault current reduces. In current-limiting circuit breakers, the incident energy remains constant as the fault current reduces. 3

4 4. Supplemental Ground Fault Protection Since these lower current faults take so long to clear, thus releasing increased incident energy, adding ground fault protection might seem to be an appropriate method to help reduce clearing time and therefore lower incident energy. One issue to remember is that ground fault provisions added to a fuse protected system require that the ground fault open a device in series with the fuse. This is usually the disconnect switch immediately upstream from the fuse and packaged as a complete switch and fuse combination. The problem is that the interrupting rating of the switch in series is almost always lower than the interrupting rating of the fuse. This isn t a problem since normally the switch relies on the fuse to clear any fault that exceeds the interrupting rating of the switch. A particular problem occurs if a ground fault relay attempts to open a switch while the current flowing through the switch is above the interrupting rating of the switch and simultaneously the fuse is sized such that it s clearing time is slower than the switch s opening time. Attempting to clear such a fault would likely result in the destruction of the switch, likely with significant collateral damage. Typical solutions to this problem have been either to include a highcurrent inhibit (also called Class II ground fault protection) or to add a time delay to the GF tripping. We examine each of these methods and the problems with both. If the ground fault exceeded 1413 amperes, the GF relay would be inhibited from sending a trip signal and the fuse would have to clear the fault. Referring to Figure 3, a current of 1413 amperes flowing through this 600A RK-1 fuse would result in a clearing time somewhere between 19 and 50 seconds. As shown in Eq. (2) and Eq. (3), incident energy increases by the ratio of current raised to the 2.5 power GF High Current Inhibit Be careful when selecting a Class II ground fault relay, since many are designed to be applied with motor contactors. Motor contactors have very low interrupting ratings, so many Class II relays have very low inhibit levels. As an example Table 2, shows available models from a particular vendor s Class II ground fault relay family. Choosing a GF relay for a Size 5 starter, for example operating at 208V, we see that this relay would not send a ground fault signal for any current that exceeded 1413A. Table 2. Typical Class II GF Relay Family Figure 3 So, if arc flash incident energy calculations are evaluated using higher current values (assume 20,000 amps will flow in a phase fault), then if only 1413 amperes were to flow during a ground fault, then these equations imply that you might well see incident energy values of: ( ) = = 753 times higher! 1413 If we applied a 125 HP, 208V, 3-phase motor to a fused switch protected by this GF relay combination, the motor would consume 342 amperes full load. Using Table from the National Electrical code, this motor could not use a time-delay fuse larger than 175% of this value, or 600A. If the ground fault current did not exceed 1413 amperes, the GF relay would trip the stored energy switch or motor starter instantaneously, clearing the fault quickly and reducing incident energy GF Time Delay To get around the problem of the switch opening before the fuse has enough time to clear a fault, some engineers have added time delay to the ground fault relay. If the delay is enough to allow the fuse to operate faster than the relay for any current value above the switch interrupting rating, the fuse will always operate first. A typical delay is 0.1 seconds (6-cycles). The problem identified was that prior to the end of this 6-cycle delay, an arc may have propagated from a ground fault into a phase fault. At this point, the currents are again balanced and the ground fault relay would believe that the ground fault has disappeared even though an arc flash event continues. In the Malmedal [1] example, the phase-to-phase arcing current was calculated to equal only 29% of available fault current and the installed fuse needed over 2 minutes to clear the fault. 4

5 White Paper TP E Unfortunately, today, the IEEE-1584 equations do not calculate the additional energy released from the single-phase, lower-current fault, prior to the escalation into a 3-phase fault. As pointed out in [8], this can result in a substantial underestimation in incident energy. As a result, Malmedal [1] recommends against adding time delay to ground fault relays protecting fuses. Note that this problem of slow clearing time is not a problem with circuit breakers, since unlike a switch and fuse, the contacts of a breaker are rated for the full interruption rating of the breaker. As a result, these contacts do not require any intentional delay when clearing a ground fault. Consequently, a circuit breaker can be programmed to clear a ground fault faster than a fused switch equipped with ground fault protection. This faster clearing time by the circuit breaker for these lower current faults results in lower incident energy released. Another feature of many ground fault equipped circuit breakers is the inclusion of zone selective interlocking (ZSI). Circuit breakers can be provided with zone selective interlocking (ZSI) to remove the intentionally delay normally added to the upstream clearing time. As shown in Figure 4, when a fault is detected downstream of two selectively coordinated breakers, the breaker closest to the fault should clear the fault first. That means the upstream breaker should delay just enough time to maintain selective coordination with the downstream breaker. What is different in a ZSI scheme is that the downstream breaker sends an inhibit signal to the upstream breaker, confirming to the upstream breaker that the downstream breaker does indeed see the fault and will clear the fault. Figure 4 However, if as shown in Figure 5, the fault occurs between the two breakers, the upstream does not receive a ZSI inhibit signal from the downstream breaker because the downstream breaker does not have fault current flowing through it. Without the ZSI inhibit signal, the upstream breaker reprograms itself to remove all intentional delays, thus clearing the fault faster. ZSI has been shown to be so effective that a new Section has been added to the 2011 version of the National Electrical Code [13] that now requires ZSI or equivalent protection for any circuit breaker that includes intentional tripping delay [14][15]. 5. Statistical Evidence of Low Level Ground Faults Eaton molded case circuit breakers include a built-in method of detecting if a single phase fault exceeds 10 ka. Evidence that a single phase fault exceeded 10 ka is evident should that circuit breaker ever be returned to the factory for testing. Over the 30 plus years this method has been included, next to none of the breakers returned for analysis show this evidence. This would imply that nearly all single phase (i.e. ground) faults are faults of less than 10 ka. Therefore, to better understand the risk your personnel face, it is suggested that a minimum arcing fault current be calculated. Measure resistance between a point on enclosure sheet steel to ground bond. Since bonding impedances may be in the 0.1 ohm range, a line to ground fault on a 277 or 347 line-to-neutral may only result in 2770 or 3470 amperes of fault currents. These low levels of fault current may result in extended clearing times and high levels of arc flash incident energy. 6. Summary Based on data from IEEE Std as well as common sense, ground faults are substantially more common than phase-to-phase faults. Since ground current flows, by definition, through conductors other than expected current carrying conductors, the impedance of those conductors is unknown. Eaton s own statistical evidence from warranty and repair records lends evidence to this conclusion. The problem is that many authors report that low-level ground faults failed to be cleared by conventional overcurrent protection quickly enough to save the equipment from substantial damage. While ground fault protection can be added to some fused switches, problems have been reported due to the required intentional delay that must be added and/or with coordinating Class II ground fault relay to insure that the switch opens only during times when low (or no) current is flowing. Circuit breakers do not require this intentional ground fault delay and therefore can clear a fault more quickly reducing incident energy below the incident energy released when the same circuit is protected by a fuse, whether current limiting or not. Remember, fuses work better at higher current levels. Low level faults cause problems for fuses. As this paper and IEEE [16] has shown, most low voltage fault magnitudes are below the current limiting range of a fuse. Depending on the range of your selected overcurrent device, this may result in extended clearing time and excessive arc flash incident energy. Ground fault sensing with instantaneous clearing and ZSI is recommended to mitigate these high incident energy levels. Figure 5 Since time delay is always required on an upstream breaker, any system with two or more levels of ground fault protection will have mandatory time delay on the upstream breaker. Because of this requirement, it is very important to insure that any ZSI system selected reduces the clearing time of the upstream breaker to sufficient fast clearing times (<80 ms). 5

6 7. References [1] K. Malmedal, P.K. Sen, Arcing fault currents and the criteria for setting ground fault relays in solidly-grounded low voltage systems, Industrial and Commercial Power Systems Technical Conference, 2000, Conference Record, Annual Meeting, Issue 2000, pp [2] H. Bruce Land III, The Behavior of Arcing Faults in Low- Voltage Switchboards, IEEE Transactions on Industry Applications, Vol. 44, No. 2, March-April 2008, pp [3] R. Doughty, T. Neal, T. Macalady, V. Saporita, K. Borgwald, The use of low-voltage current-limiting fuses to reduce arcflash energy, IEEE Transactions on Industry Applications, Vol. 36, Issue 6, Nov-Dec 2000, pp [4] H. Bruce Land III, Determination of the Cause of Arcing Faults in Low-Voltage Switchboards, IEEE Transactions on Industry Applications, Vol. 44, No. 2, March-April 2008, pp [5] J.G.J. Sloot, R.J. Ritsma, Protection against fault arcs in low voltage distribution boards, Eindhoven University of Technology, Netherlands, Protection-against-fault-arcs.pdf. [6] IEEE Recommended Practice for Electric Power Distribution for Industrial Plants, IEEE Std , Approved December 2, [7] H. Stanback, Predicting Damage from 277-V Single Phase to Ground Arcing Faults, IEEE Transactions on Industry Applications, Vol. IA-13, Issue 4, Jul 1977, pp [8] D. G. Loucks, Calculating Incident Energy Released With Varying Ground Fault Magnitudes on Solidly Grounded Systems, IEEE Transactions on Industry Applications, Vol. 46, Issue 2, 14 Jan 2010, pp [9] G. Parise, et. al., Arcing Fault in Sub-Distribution Branch- Circuits, IEEE Transactions on Power Delivery, Vol. 8, No. 2, April 1993, pp [10] D.G. Loucks, Transient Overvoltages on Ungrounded Systems from Intermittent Ground Faults, Eaton Application Note, download from: [11] Design of Reliable Industrial and Commercial Power Systems, IEEE Standard 493, Approved 7 February [12] IEEE , IEEE Guide for Performing Arc-Flash Hazard Calculations [13] NPFA 70, National Electrical Code, 2011, National Fire Protection Association. [14] Eaton Code Breaker 2011, NEC Arc Flash Reduction Requirement, pub/@electrical/documents/content/pa e.pdf [15] Eaton, Arc Flash Energy Reduction Techniques Energy- Reducing Maintenance Switching, ecm/groups/public/@pub/@electrical/documents/content/ tp e.pdf [16] T. Papallo, M. Valdes, G. Roscoe, Predicting Let-Through Arc-Flash Energy for Current Limiting Circuit Breakers, IEEE Transactions on Industry Applications, Vol. 46, Issue 5, 16 September 2010, pp Eaton Corporation Electrical Sector 1000 Cherrington Parkway Moon Township, PA United States 877-ETN-CARE ( ) Eaton.com 2011 Eaton Corporation All Rights Reserved Printed in USA TP E / TN September 2011 Eaton is a registered trademark of Eaton Corporation. All other trademarks are property of their respective owners.