COMMON SOURCES OF ARC FLASH HAZARD IN INDUSTRIAL POWER SYSTEMS

COMMON SOURCES OF ARC FLASH HAZARD IN INDUSTRIAL POWER SYSTEMS Joost Vrielink Hans Picard Wilbert Witteman Eaton Eaton SABIC-IP Europalaan 202 7559 SC Hengelo Europalaan 202 7559 SC Hengelo Plasticslaan 1 4600AC Bergen op Zoom Netherlands Netherlands Netherlands Abstract - Electrical design of industrial facilities addresses basic design considerations such as reliability, flexibility, cost and safety. In Europe, arc flash hazards are not always dealt with in great detail beyond reliance on intrinsic equipment safety. Reality is, however, that even with the best equipment design, arc flash accidents still occur, often as a result of human error during work activities. IEEE Standard 1584 has been published as a result of many years of research and empirical testing of the arc flash phenomenon. The standard provides guidance in calculating the arc flash incident energy, taking into account multiple design factors. Incident energy calculations require relatively accurate assessment of the short circuit levels and the characteristics of the protective devices. Such input is conveniently provided by conventional short-circuit and protective device coordination studies. The combination of factors influences incident energy levels rendering some common estimation techniques useless, perhaps even dangerous. For example, a lower short-circuit current can lead to higher incident energy levels because some types of protective devices take longer time to trip for lower current. This paper presents some typical scenarios in industrial facilities that are prone to yield high incident energy levels. The information presented is based on authors practical experience with arc flash hazard assessments. Index Terms Electrical Safety, Arc Flash Hazard Assessment, Incident Energy. I. INTRODUCTION In the United States, electrical safety is covered by the National Fire Protection Association (NFPA) in its Standard for the Electrical Safety in the Workplace (NFPA 70E-2009 [1]). In this standard, arc flash hazard is defined as a dangerous condition associated with the release of energy by an electric arc. The hazard is to be considered when work is performed near energized conductors, because of the likelihood of an electrical arc occurring due to human interaction with the equipment, or condition of the very equipment. An arc flash hazard analysis quantifies the risk of human exposure to the energy of an arc event under worst-case conditions, as well as effectiveness of measures of reducing incident energy and/or additional protective measures. Many multinational corporations today implement a global safety program based on universal set of minimum safety requirements and procedures. When such a program originates from the US, arc flash hazard assessment is often included as a requirement. Increasingly, industrial plants operating in Europe are introduced to arc flash hazard in this fashion. In the United States arc flash has been a recognized electrical hazard since 1991 [2], with the fifth edition of the NFPA 70E published in 1995 specifically addressing arcflash hazard. In this edition the requirements for protective clothing and a definition for the arc-flash boundary are included. Later revisions provided additional guidance for calculation of arc flash incident energy and protective boundary. The first edition of IEEE Standard 1584, Guide for Performing Arc-Flash Hazard Calculations [3] was issued in 2002. Empirical equations were derived from statistical analysis of extensive test data. The standard also provides a procedure for conducting a complete arc flash hazard analysis. In 2006 the IEEE and NFPA agreed to collaborate on a joint research initiative to increase the understanding of arc flash phenomena. Part of the plan is working with the international community for global adoption of a new standard. European codes and standards do not require assessing arc flash hazard to the detail the NFPA 70E does. European Directive 89/391/EEC requires an employer to evaluate risks and take the necessary preventive safety measures. The general principles of the Directive put emphasis on avoiding risks, combating them at the source and taking collective protective measures over individual ones. An arc flash hazard analysis therefore will fit into the European standards framework, while risk analysis and subsequently taken preventive measures will most likely differ from US approach. II. ARC FLASH HAZARD IN EUROPE The goal of an arc flash hazard assessment is to determine the energy released when an arc flash occurs. The calculated energy can be used to select Personal Protective Equipment (PPE) that can protect workers against the hazard. European electrical equipment design and work safety standards seek to provide passive safety against shock hazard with preference of collective over personal safety measures. Although the arc flash hazard is not explicitly addressed in such standards, this approach often does offer protection against arc flash risk by reducing the odds of arc initiation. An arc flash hazard assessment should not serve as a means of substituting these measures with mere personal protection, but rather as a way to protect workers in the case when passive and other measures do not adequately protect against arc flash risk. A. Internal arc proof switchgear IEC equipment guidelines define requirements and test

methods for internal arc in metal clad switchgear and controlgear. These are IEC 62271-200 [Fig. 6] for medium voltage and IEC/TR 61641 [5] for low voltage. Although these are not mandatory requirements and, for low voltage, the guidelines do not have the weight of a standard, the testing methods clearly demonstrate how a particular design limits the risk of injuries to personnel and damage to switchgear in case of an internal arc. Care needs to be exercised in how such tests are interpreted. For example, when the test is performed with all doors closed, the internal arc classification will not apply when doors are opened while work is performed. B. IP codes and forms of separation Enclosures of electrical equipment are designed to prevent access to hazardous parts, and specifically the contact with low-voltage live parts, mechanical parts or approach to high-voltage live parts. The IP code indicates the degree of protection provided by the enclosure. IEC metal-enclosed switchgear equipment standards for low voltage (IEC 61439-1 [6]) and medium voltage (IEC 62271-200 [4]) prescribe minimum degree of protection for enclosures depending on the type of assembly and the environment. For low voltage switchgear different forms of separation are defined where compartments can separate busbar, functional unit and cable connection. These compartments will have a minimum degree of protection expressed by an IP code. It is important to keep in mind that there is no relationship between the degree of mechanical protection and the internal arc flash rating of an enclosure. Although higher degrees of ingress protection may be thought of as reducing the likelihood of an arc initiation from external sources, the safety hazard may not be reduced due to other effects of arc flash such as release of flying shrapnel. C. Arc detection and reduction systems Various systems are available that detect internal arcs in switchgear and act to reduce the arcing duration, either by fast tripping of protective devices or by creating an intentional short-circuit to quench the arc [7]. These systems do not work to prevent an arc flash from occurring, but are often an effective way of reducing the hazard. As these are active means of suppressing the effect of an internal arc, routine periodical testing and maintenance of these systems is necessary. D. Safe work practices For CENELEC countries minimum requirements for electrical safe working are laid out by European Standard EN 50110, Operation of Electrical Installations [8]. The standard is endorsed by national standards and regulations. The standard divides working procedures into three categories: dead working, live working and working in the vicinity of live parts. Dead working is considered to be the base safe working condition where essential requirements are satisfied to ensure that the electrical installation at the work location is safely de-energized. Live working and working in the vicinity of live parts are subject to national regulations, often requiring additional measures as well as training of the workers. Arc flash risk may be present in all three conditions. Some examples include verifying that the installation is de-energized using voltage detectors, connecting of portable meters to the live parts, phase comparison checks, placing of the shorting grounds, replacing fuses that are energized and live fault finding. It is therefore prudent to evaluate each work activity individually and consider how existing measures protect against arc flash hazard and if additional measures can be taken. Performing an arc flash hazard assessment helps to select adequate PPE when the risk is unavoidable.. III. ARC FLASH HAZARD ASSESSMENT WITH IEEE STD 1584-2002 The IEEE Standard 1584-2002 provides techniques in determining arc flash hazard due to an arc event in 3- phase ac systems. Two models are presented for arc flash incident energy and boundary calculation; an empirically derived model for voltages from 208V up to 15kV, and a theoretically derived model for all other voltages (also called the Lee method). All calculations in this paper were carried out with the empirically derived model. Hazards are calculated as arc flash incident energy, or radiated energy density, that is impressed on a worker s skin in joules per cm 2. A researcher named Stoll determined by experiment that it takes 5 joules per cm 2 of heat energy for human skin to reach the threshold between a first degree burn and a second degree burn. Therefore, the value of 5 J/cm 2 has been established as the maximum amount of energy to be allowed to contact the worker s skin during an arc event. Incident energy is always calculated at a certain work distance. IEEE 1584 provides default distances for different types of switchgear and voltages. Incident energy is in indirect proportion to work distance; therefore increasing the distance can be an effective solution to high arc flash hazard. Moving a worker to twice the original work distance reduces the energy level by a factor four. The arc flash boundary is defined as the distance from the arc where incident energy has attenuated to 5 J/cm 2. The boundary is a function of arc flash incident energy and the work distance. It is essentially the minimum working distance for injuries not to exceed first degree burns. Incident energy is often expressed in calories because US standards such as the NFPA 70E did not fully adopt the metric system. The IEEE 1584 calculation methods do provide results in joules, but because testing was conducted with calories, both terms appear in the standard. Clothing and other PPE available in the market today use ratings in cal/cm 2. The widespread use of calories as a measure for incident energy can make it difficult to exclusively use joules when dealing with arcflash, but it should be the unit of choice by virtue of being

a SI unit. The following relations can be used to convert between calories and joules: 1 calorie = 4.184 joules 1 joule = 0.239 calories There are many of factors influencing incident energy levels, adding to the complexity of calculations [9]. Two main factors are the three-phase fault current and the fault clearing time, which are taken from a short-circuit study and a protective device coordination study, respectively. Most protective devices applied in the electrical distribution systems operate based on sensing the overcurrent condition at the point of their placement. A time-current characteristics (TCC) graph of the overcurrent protective device will be used to explain the effect of the fault current and clearing time, see Fig. 1. In a TCC diagram the time it takes the protective device, in this case a low voltage circuit breaker, to interrupt a certain current is plotted on a log-log scale. The scales show time vertically from 0.01s to 1000s and the current horizontally from 500 A to 100 ka. In a system with several voltage levels the current is conveniently shown for all protective devices upon selecting the reference voltage level. LV-I 167.4 J/cm2 104.6 J/cm2 33.5 J/cm2 16.7 J/cm2 5 J/cm2 Ia1 Ia2 Ia3 Ia4 Ia5 Fig. 1: TCC diagram showing constant energy lines Low voltage air circuit breaker (ACB) trip units are commonly equipped with different protective segments; each segment can be independently set to provide protection of equipment and coordination with other protective devices. Although different names are used for these segments among manufacturers, they will generally fall within the following categories: Long Time provides protection against overload currents and mimics a thermal type protection with an I 2 t characteristic. The long time segment has a pickup (LTPU) value ranging from 0.5x to 1.0x of the sensor rating, and a time delay (LTD) defined at some multiple of the pickup value, calibrated in seconds. Short Time This segment is commonly used for backup protection. Shorter tripping time is required for close faults while longer tripping time is used for the backup of the downstream devices. The short time segment has a pickup value (STPU) of about 2x to 10x the sensor (or LTPU) rating and a time delay (STD) of typically 0.1 to 0.5s. Instantaneous the instantaneous stage has no intentional time-delay and it will trip as soon as the current reaches or exceeds the pickup value, ranging from 2x to 16x the sensor rating. This segment provides maximum protection against close faults for both the protective equipment and the circuit breaker itself, but may lead to miscoordination as two instantaneous protection stages in series cannot discriminate unless separated by significant impedance. A short-circuit study yields a so-called bolted fault current for each location; this is the maximum value of the short-circuit current based on a calculation using only the system and source impedance. The arcing current in low voltage systems is always lower than the bolted shortcircuit current because of the finite impedance of the plasma cloud conducting the arc current. This reduction is different at different voltage levels (see Fig. 2) and is significant in low voltage systems. Shown in Fig. 1 are five arcing current levels as Ia1 through Ia5. The variance of the arcing current in a real distribution system is typically not as dramatic as shown in this picture where the current levels are chosen arbitrarily only for the purpose of demonstration. Ia% - arcing current as % of Ibf 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 400V 10kV 0% 0 20 40 60 80 100 Ibf - bolted fault current (ka) Fig. 2: Relationship between bolted fault current and arcing current for two voltage levels Constant energy is shown as diagonal dotted lines, showing the maximum fault clearing times for the entire range of currents. The energy values chosen and the corresponding hazard categories are the reference incident energy levels based on NFPA 70E, for reason of

simplifying PPE selection. This illustration can be useful for understanding the relationship between arcing current and clearing time when determining incident energy. Table I shows the calculated incident energy for each of the arcing current levels. Higher arcing current can result in either lower or higher incident energy, depending on the protective device characteristics. The highest energies in this case occur at the lowest arcing currents. The challenge of an arc flash hazard analysis is to find the worst-case incident energy occurring in the system. As we have seen, this is not simply a case of finding a scenario with the highest (or lowest) short-circuit levels. All possible scenarios, from changing network topology to the minimum and maximum level of short-circuit contribution should be evaluated for incident energy at every location. TABLE I CALCULATED INCIDENT ENERGY FOR ARCING CURRENTS IA1 THROUGH IA5 Ia (ka) T (s) E (J/cm 2 ) Ia1 2 14.5 84 Ia2 3 6.5 58 Ia3 5 0.22 3.4 Ia4 9 0.22 6.5 Ia5 14 0.085 4.0 IV. COMMON SOURCES OF ARC FLASH HAZARD Fig. shows a simple model of an industrial system that will be used to demonstrate some common situations that can lead to high arc flash hazard. Buses LV1 and LV2 are being evaluated, each fed by its own 630 kva transformer. A standby (emergency) generator is connected at Bus LV2. It is only allowed to operate when the normal power from utility is not present. MV 10000V LV1 400V MV-Tx1 Relay Tx1 S 630kVA, 5.00% 10000/400V CBL-Tx1 (3) x185 Copper PVC 10m LV-I1 ACB 1000A LV-F BS88 Fuse 160A UTIL 10000V 3P 250MVA SLG 10MVA LV2 400V Fig. 3: Single Line Diagram MV-Tx2 Relay Tx2 S 630kVA, 5.00% 10000/400V CBL-Tx2 (3) x185 Copper PVC 10m LV-I2 ACB 1000A G 630 kva X"d 0.15 OpenLV-G ACB 1000A In the normal operating condition the tie-breaker and generator breaker are open. Typical values have been assumed for MV level utility contribution and transformer impedance. LV-I1, LV-I2 & LV-G are 1000A air circuit breakers with adjustable trip units with the following settings: LTPU/LTD of 0.9x/2s, STPU/STD of 6x/0.2s, Instantaneous disabled. LV-F is a typical slow acting current limiting BS88 / IEC 269 fuse. Next, a single incident energy is calculated for each case of protective devices settings, based on the assumption that work is performed on the bus side of the protective device, which in this case is the main device for the panels LV1 or LV2. This means that the arcing current is interrupted by the protective device being evaluated, as long as the space that containing the device doesn t include the line side bare conductor whereby the fault would need to be interrupted by an upstream (backup) device. Hence for such assumption to be valid, line side elements would have to be compartmentalized, or separated from the work area with barriers containing the arc flash. A. High pickup settings, high trip time Maximum settings for protective devices lead to high potential arc flash hazard. This is illustrated on the TCC diagrams in Figs. 4 and 5, where the pickup setting of the Short Time segment has been increased from 6x to 10x (typical maximum setting for air circuit breakers). Table II shows the resulting arc flash hazard. The fault clearing time for the second case increased dramatically with incident energy increased by almost a factor of four. This also demonstrates that in the absence of instantaneous protection or fuses, the Short Time segment settings of the air circuit breakers are pivotal in managing the arc flash incident energy levels. When dealing with outgoing (feeder) sections of the switchgear, the calculated bus side incident energy applies. When looking downstream of the switchgear, the feeder protective devices will be responsible of clearing the arc fault currents. As these devices are sized well below the main and bus tie devices, the incident energy and arc flash hazard levels downstream of the branch circuit protective devices decrease in relation to faster fault clearing and current-limiting effects of the cables. TABLE II HAZARD INCREASE DUE TO A CHANGE IN SETTING 6x setting fig. 4 10x setting fig. 5 ACB LT 0.9x, 2s 0.9x, 2s ACB ST 6x, 0.2s 10x, 0.2s Ibf (ka) 16.9 16.9 Ia (ka) 8.85 (100%) 7.55 (85%) t (s) 0.218 1.022 E (J/cm 2 ) 27 105 Next example shows the effect of sizing the low voltage fuses that are for the purpose of illustration assumed to be connected at panel LV1. In table III three fuse sizes are compared, sized at 18%, 44% and 69% of the nominal current (In) of the upstream transformer. Fig. 6 shows their time-current characteristics together with the arcing current at LV1. It is interesting to note a dramatic increase of incident energy for the 630A fuse as this fuse clears the fault by the I 2 t part of the fuse curve and not in the current limiting mode as the smallest fuse in this example. In this example, an arc fault clearing by the 400 A fuse may require a little closer scrutiny. In fact, while the effect of the cable impedances can often be neglected in most short circuit estimates, as it is often considered to be within margin of error, arc flash assessment may demand a second look.

LV-I 167.4 J/cm2 104.6 J/cm2 16.7 J/cm2 33.5 J/cm2 5 J/cm2 LV-I 167.4 J/cm2 104.6 J/cm2 16.7 J/cm2 33.5 J/cm2 5 J/cm2 Fig. 4: ACB with low STPU Fig. 5: ACB with high STPU Fig. 6: Three fuse sizes 160A 400A 630A In the radial system, a difference in the bolted fault current between both ends of the low voltage feeder can be as much as 5 % but more commonly less than 3 %. Even this relatively small difference can cause the fuse, such as the 400 A fuse in our example, to clear the fault in the current limiting mode at the beginning of the feeder, while clearing the fault at the remote end of the same feeder in its I 2 t mode, resulting in higher incident energy at the remote end compared to the beginning of the radial circuit. TABLE III HAZARD FOR DIFFERENT FUSE SIZES Fuse 160A Fuse 400A Fuse 630A % In 18% 44% 69% Ibf (ka) 16.9 16.9 16.9 Ia (ka) 8.85 (100%) 7.55 (85%) 7.55 (85%) t (s) 0.01 0.04 0.70 E (J/cm 2 ) 1.2 4.1 71 B. Low arcing current, high trip time Another, much more dramatic effect of the low fault levels can be seen on an example with the system supplied from the emergency generator. The emergency generators are typically sized at the fraction of the normal system capacity for the economical reason and the fact that not all loads in the system require emergency power supply. The source reactance represented by the typical standby generator can be therefore many times higher than the source reactance largely represented by the utility and the supply transformers. From what was demonstrated earlier, it follows that if, instead of a utility, the main LV2 bus is now supplied by the backup generator, arc flash hazard may change and it needs to be closely investigated. The change will depend on the response of the protective devices to lower bolted fault and therefore arc fault currents. Also demonstrated earlier was the difference in response between air circuit breakers and standard fuses. In this case bus LV2 is evaluated with incoming circuit breakers LV-I1 & LV-I2 open, generator G running and generator circuit breaker LV-G closed. This is the most common emergency scenario employing standby generator system. Paralleling design is sometimes employed only for generator testing and source transitions. In emergency mode, a fault on bus LV2 will only see the generator s contribution. Overcurrent settings of the generator circuit breaker are kept the same as the transformer incoming circuit breakers (Long Time at 0.9x, 2s, Short Time at 6x, 0.2s, Instantaneous disabled). Table IV compares the arc flash hazard for both modes of the system operation. In the emergency mode, the arc fault current at the LV2 bus is less than half of the fault current in normal condition, below the short time pickup of the main circuit breaker. Consequently this leads to very high incident energy, more than eight times higher than in the normal system condition. TABLE IV HAZARD WITH A TRANSFORMER AND GENERATOR CONTRIBUTION Utility Generator Ibf (ka) 16.9 6.05 Ia (ka) 8.85 (100%) 3.3 (85%) t (s) 0.218 5.39 E (J/cm 2 ) 27 224 C. High arcing currents When evaluating arc flash hazard, fault levels from all possible system scenarios should be considered. When multiple sources exist in the system that can be paralleled together such as via system tie breakers, arcing currents will increase. Such situation would occur if two utility sources were to be tied together at the customer site in a ring or Ring Main Unit (RMU) configuration which is relatively rare in industrial systems, or the site has another source of power, such as a generator used for combined heat and power generation (CHP). We will demonstrate the effect of higher fault levels on paralleling of the two transformers in our sample case. The arc flash hazard for both cases is shown in table V.

In our example, increasing the arcing current leads to somewhat moderate increase of incident energy. In this case the 100% increase of fault current leads to a mere 44% increase of incident energy. This is evidently because of the segmentation of the air circuit breaker trip curve discussed earlier and in the given situation, higher current will not cause trip unit to act slower. In fact, it could only result in faster tripping if the instantaneous segment were employed. TABLE V HAZARD WITH SINGLE AND PARALLEL TRANSFORMER CONTRIBUTION Single Parallel Ibf (ka) 16.9 32.3 Ia (ka) 8.85 (100%) 12.76 (100%) t (s) 0.218 0.218 E (J/cm 2 ) 27 39 D. Lack of a main protective device In cases where no incoming protective device is present in the panel or switchgear, or energized parts of the bus are not separated by compartmentalization or barriers, arc fault must be cleared by the protective devices upstream of the panel or switchgear. The incident energy results for the faults at bus side versus line side of the main low voltage switchgear are compared in table VI. In this example, the upstream protective device is a medium voltage relay of definite time characteristic, with a fault clearing time of 1.1 s. The effect of relying on upstream device, in the given example, leads to an increase of incident energy of over four times. TABLE VI HAZARD WITH TRIP BY ACB OR UPSTREAM RELAY Trip by ACB Trip by relay Ibf (ka) 16.9 16.9 Ia (ka) 8.85 (100%) 8.85 (100%) t (s) 0.218 1.1 E (J/cm 2 ) 27 121 V. CONCLUSIONS Arc flash hazard assessment is a relatively new development in Europe, but it fits the general framework of requirements imposed on employers in performing safety risk assessments. Equipment and work safety standards seek to provide passive protection against shock hazard and although the arc flash hazard is not explicitly addressed, existing measures often can reduce the likelihood of an arc initiation. Despite these measures the arc flash risk cannot be completely avoided during some work activities. An arc flash hazard assessment can be used to select PPE that protects against the hazard. The level of arc flash hazard is difficult to predict and depends on many factors. A detailed study following the IEEE Standard 1584 procedures is in many cases required to reliably find the worst-case incident energy in each location. A few scenarios typically found in industrial facilities were shown to yield high arc flash incident energy levels. The high hazards were shown to have multiple sources: low or high fault contribution, protective device characteristics and system properties. Performing an arc flash hazard assessment can readily identify these issues, as well as indicate how these issues may be addressed to reduce arc flash hazard. VI. REFERENCES [1] NFPA 70E-2009, Standard for Electrical Safety Requirements for Employee Workplaces. [2] Ravel F. Ammerman, P. K. Sen, John P. Nelson, Electrical Arcing Phenomena, A historical perspective and comparative study of the standards IEEE 1584 and NFPA 70E, IEEE Industry Applications Magazine, May / June 2009. [3] IEEE Std 1584-2002 IEEE Guide for Performing Arc-Flash Hazard Calculations. [4] IEC 62271-200 AC metal-enclosed switchgear and controlgear for rated voltages above 1 kv and up to and including 52 kv. [5] IEC/TR 61641 Enclosed low-voltage switchgear and controlgear assemblies Guide for testing under conditions of arcing due to internal fault. [6] IEC 61439-1 Low-voltage switchgear and controlgear assemblies Part 1: General rules. [7] L. Kumpulainen and S. Dahl, Minimizing Hazard to Personnel, Damage to Equipment, and Process Outages by Optical Arc-Flash Protection in IEEE PCIC Conference Record, 2010. [8] EN 50110-1 Operation of electrical installations. [9] H. Wallace Tinsley III, M. Hodder, A Practical Approach to Arc Flash Hazard Analysis and Reduction, IEEE Industry Applications Magazine, Volume 41 Issue 1, pp 144-154, Jan/Feb 2005. VII. VITA Joost Vrielink joined Eaton in 2008 and is currently working as a power system engineer. He studied electrical engineering at Twente University in Enschede, The Netherlands. Joost performed arc flash hazard calculations in many industrial installations in Europe, and has developed professional experience in aligning US safety programs to European standards and practices. Hans Picard has worked in engineering, sales and marketing for Emerson Process Management, Holec Algemene Toelevering and Eaton Electric. He is currently working for the Eaton's Electrical Solution & Services Business Unit. He also has professional experience in contract management and asset optimization. He currently has marketing responsibilities for Eaton's Engineering Services in the EMEA region. Wilbert Witteman graduated as BSc in 1997. He worked as electrical engineer for several Engineering Procurement and Construction companies, including seven years with Technip. He joined SABIC Innovative Plastics in March 2005 as electrical maintenance engineer and became responsible for the Bergen op Zoom site HV grid.