FORENSIC ENGINEERING ANALYSIS OF ELECTRICAL FIRES DUE TO ARCING

FORENSIC ENGINEERING ANALYSIS OF ELECTRICAL FIRES DUE TO ARCING By Frederick F. (Rick) Franklin, P.E. www.paceforensic.com. ABSTRACT: AFCI s provide a reference for the amounts of energy and associated copper melting which can and cannot cause a short circuit arcing fire. Quantifying the amounts of arcing energy that cannot cause a fire leads to a better analysis as to whether equipment such as certain fuses, certain circuit breakers, GFCI s, motors, consumer electronics, series faults, and other electrical equipment can cause arcing fires. KEY WORDS: Arcing, Energy, Copper Melt, Globules, AFCI, Fuse, GFCI, Vehicle Fire, Latent, Series INTRODUCTION: How much arcing energy does it take to start a fire? A better question for fire investigators is: How much energy can an arc have and NOT cause a fire? The invention of Arcing Fault Circuit Interrupters (AFCI s) and the fact that they have become code has answered this question definitively. AFCI s are circuit breakers with a computer chip inside. The computer chip detects an arc by constantly monitoring the electrical current s sine wave waveform to detect any missing portions of the sine wave. U.L. 1699 to which AFCI s are designed allows the AFCI to take up to two cycles of sine wave (1/30 th second) to trip. Figure 1 shows an AFCI which has just tripped as the author burned through a power cord to create a carbon path arc. Figure 1 also shows the missing portions of sine wave in the arc which the pattern recognition program in the AFCI has just used to decide to trip. In 60 Hertz alternating current circuits, the electrical current and voltage change direction (polarity) from plus to minus and back to plus 120 times per second. Each time the current goes through zero, the arc extinguishes, since there is no electrical current left to support it. The arc often does not re-ignite until the voltage rises substantially above zero to a significant level. These missing quarter cycles, missing half cycles and other variations in the sine wave are detected by the pattern recognition program in the microprocessor in the AFCI to trip it.

FIGURE 1: MISSING PORTIONS OF SINE WAVE WHICH HAVE JUST TRIPPED THE AFCI FIGURE 2: SMALL COPPER MELTS ALLOWED BY AFCI IN 16 GAUGE POWER CORD 2

Figure 2 shows the typical small amount of copper melts which form before the AFCI trips. This is a very important reference point for fire investigators, because if an arc producing this amount of copper melt could cause a fire, AFCI s would not be code. Such small amounts of copper melt at a fire scene can be discounted by fire investigators as the cause of a fire, within a reasonable degree of engineering certainty, in the author s opinion. Rather, they occurred as the result of the ensuing fire, as the fire burned through the insulation to create arcing. Such small amounts of copper at a fire scene can be the result of AFCI s tripping. They can also be caused when the magnetic trip in a common circuit breaker happens to be sensitive enough to trip on the arcing current (many are not 2.) (Editor s Note: The six videos mentioned in this article may be viewed at www.paceforensic.com/videos) ARCING VIDEO: In Video 1, made in 1988, a power cord arcs over 30 times and for two minutes before the common circuit breaker in the author s office building trips. Every one of the 30 arcs in this video melts the power cord completely apart. If the power cord was not melting completely apart, the circuit breaker s thermal element would trip the circuit breaker within a few more seconds. All of these arcs last for ¼ second, or more. Based upon this video and his observations at fire scenes, the author published in 1990 that virtually every arc which causes a fire in 15 and 20 ampere circuits has enough energy to melt the conductors completely apart. 2 20 more years of experience has confirmed this. Thus, any arc in 15 and 20 ampere circuits which did not have enough energy to melt the conductors completely apart at a fire scene probably occurred as a result of the ensuing fire, in the author s opinion. ENERGY FORMULA: The electrical energy, in Joules, in an arc is given by: E = V x I x t Joules = Volts x Amperes x Seconds When an arc occurs in electrical wiring, the protective device cannot control the voltage or the current in the arc. It can only reduce the energy in the arc by opening the circuit more quickly. Reducing the opening time reduces the energy in the arc, and this reduces the probability of the arc causing a fire. It is not necessary to totally prevent the arc from forming. 3

FLYING COPPER GLOBULES: How exactly does the electrical arc cause a fire? First, the arc itself has a temperature of over 5000 F., so if the arc touches a combustible, it can ignite it. Second, the arc throws out numerous melted copper globules, as you can see in Video 1. Since copper melts at 1980 F., these copper globules are hot enough to ignite combustibles where they land (most combustibles ignite between 400 F. and 800 F). Video 2 shows flying copper globules igniting three blankets before a common circuit breaker trips. Not only can these flying copper globules ignite curtains, bedding, carpeting, and upholstery, but it is the author s experience that they can occasionally ignite wood directly. Sometimes the flying copper globules ignite more than one fire, creating multiple origins. Figure 3 is just such a case, in which two areas of origin occurred from one short circuit arc at an electrical outlet. The flying copper globules ignited paper materials on the floor at two separate locations. FIGURE 3: TWO AREAS OF ORIGIN CAUSED BY FLYING COPPER GLOBULES FUSES: We have discussed that the 1/30 th second (.033 second) energy allowed into an arc by an AFCI is too short to cause a fire, so we can use this reference to prove which other things cannot cause a fire. For example, can fuses allow enough energy into an arc to cause a fire? This can be learned from considering their opening times at various electrical currents. The following table is taken from an earlier publication 2, in which the typical opening times are listed for various types of household fuses. 4

OPENING TIMES IN SECONDS TYPE OF FUSE 200 AMPS 250 AMPS ABC-8 GLASS FUSE.005.003 ABC-10 GLASS FUSE.008.006 MDA-15 GLASS FUSE.015.008 ABC-15 GLASS FUSE.015.005 AGC-20 GLASS FUSE.015.021 TL-15 PLUG FUSE.025.012 TL-20 PLUG FUSE.063.016 T-30 PLUG FUSE.375.080 It may be observed that at 200 and 250 amperes, the opening times of 15 ampere household fuses is less than the.033 second allowed by an AFCI. The opening times of 20 ampere fuses are also in this range, and certainly less than.250 second. So 15 and 20 ampere fuses are very good protection from an arcing fire. But a 30 ampere fuse has an opening time of.375 second at 200 amperes, and thus, a 30 ampere fuse is poor protection against a short circuit arc. Video 3 is a demonstration of the relative amounts of energy allowed into a carbon path arc by 15 and 30 ampere fuses, and a 16 ampere European circuit breaker with a 5X magnetic trip: CONSUMER ELECTRONICS: If a 15 ampere fuse is good fire prevention, then fuses with lesser ratings are even better protection, because they pop even more quickly. This includes the glass fuses inside consumer electronics, which typically are rated at 10 amperes or less. In addition, most consumer electronics connect the input 120 volts, A.C. from the power cord directly to a small step-down transformer, which is used to create the lower D.C. voltage for the electronics. This small transformer has a comparatively high internal resistance, which also limits the amount of energy the transformer can supply to any short circuit arc which might develop inside the electronics. Finally, the small diameter wires inside these transformers cannot generate enough energy in an arc to cause a fire inside the transformer. Like a fuse, small conductors melt apart before an adequate amount of energy is generated to cause a fire. This electrical engineer has investigated over 2000 structural fires as a cause and origin expert, and, except for televisions, not once has he ever blamed consumer electronics for the cause of a fire, except for a short circuit arc in their power cords (power cord arcs are almost always caused by user abuse, in the author s experience). This includes 5

radios, stereos, clocks, DVD players, CD players, VCR s, desk top computers, and computer monitors. (Laptop computers and some cell phones reportedly cause fires, because their batteries can overheat.) BATTERY CHARGERS: A similar analysis holds for the small battery chargers which plug into an electrical outlet to charge a cell phone and other electronic devices. The fuse, the internal resistance of the transformer, and the small diameter of the wires inside these transformers greatly limit the amount of energy into any arc which might occur. The author has never blamed one of these devices for causing a fire (except for an arc between the male prongs). CHRISTMAS LIGHTS: A string of Christmas tree lights has also never been blamed as the cause of a fire. These miniature lights generally have two 1 ampere fuses inside their male plugs, and their conductors are very small in diameter, generally 26 gauge. Even if the 1 ampere fuse did not open, the 26 gauge wires would melt apart before they could generate enough arcing energy to cause a fire. Even the old style strings of Christmas lights have a fuse with a rating of much less than 15 amperes, and so they also do not cause short circuit fires, in the author s opinion. Recently an electrical expert testified that a series fault in Christmas lights with a 1 ampere fuse could cause a fire. The impedance matching theorem states that the maximum power into a series fault with a one ampere fuse is 30 watts, and this power only occurs for a very particular value of series fault resistance namely, when the series fault resistance exactly equals the circuit resistance. For a more typical current in the lamp string of 0.25 ampere, the maximum power into a series fault is 7.5 watts, and again, this occurs only when the series fault resistance happens to exactly equal the circuit resistance of 120/.25 = 480 ohms. Such low powers are extremely unlikely to cause a fire, in the author s opinion. SERIES FAULT ARCING: There is controversy in the field of fire investigation about series faults and how they cause fires. The author can only relate his own experience: It is well known that aluminum service entrance cables can overheat at their connections to the circuit breaker panel, where a large amount of energy is available. But in 15 and 20 ampere circuits, the author has only blamed a very few series faults for the cause of a fire. In all of these cases, there was metallic melting at the series fault, indicating that arcing had occurred. This paper has described how short circuit arcs (parallel faults) must leave a metallic melt in order to cause a fire, and that the metallic melt must be substantial. So it just makes sense to this writer that for anything else in 120 volts, A.C. electrical wiring to cause a fire, the same amounts of energy must first be present. In 6

the author s opinion, if one does not find a metallic melt, it is very unlikely that a series fault was the cause of the fire. SMALL CONDUCTORS: To prove that the tiny wires in electric blankets cannot cause an arcing fire, the author measured the opening time of a short circuit between such wires using a storage oscilloscope. Video 4 demonstrates that the energy in this arc was less than that allowed by an AFCI. However, it is well known that when electric blankets are balled up by their user into a pillow, etc., the trapped heat can ignite the blanket to cause a fire. CONDUIT: Knowing that short circuit arcs in 15 and 20 ampere circuits can last for 1/4 th second (or a few multiples of 1/4 th second), it becomes apparent why metallic conduit is so effective at preventing fires. The arc almost never lasts long enough to melt through the metallic conduit. This is also true for flexible metallic conduit (Bx and Greenfield). Once AFCI s become ubiquitous, conduit may no longer be necessary for fire prevention, in the author s opinion. In higher current circuits, the arc does sometimes have enough energy to melt through conduit and sputter molten metal onto surrounding combustibles. But even in higher current circuits, if there is no hole in the conduit, it is extremely unlikely that any arcing inside the conduit lasted long enough to cause a fire. ELECTRIC MOTORS: The metallic casings of electric motors act much the same as conduit. Again, for any type of short circuit, there is insufficient arcing energy to melt through the casing surrounding the motor. In addition, every motor has a high temperature limit thermostat which turns off electrical current to the motor to prevent a fire from any other type of overheating in the motor. This electrical engineer has never blamed an electrical motor for the cause of a fire, except on one occasion. That was a 3000 horsepower motor in a concrete manufacturing plant. Figure 5 shows the remains of a ¼ horsepower soft drink dispensing pump motor which an electrical expert blamed for the cause of a fire in a tavern. 7

8 FIGURE 4: ELECTRIC MOTOR NEAR AREA OF ORIGIN IN A TAVERN IN LOUISVILLE, KY. The motor did appear to be near the origin of this fire, but there was also a plastic trash can at this origin. It is well known that tavern fires can occur hours after an employee dumps ashtrays into a plastic trash can when they leave for the night. Disassembly of this motor revealed no metallic melts inside the motor. GFCI S: When the author founded his forensic engineering firm in 1974, he investigated an accidental electrocution from 120 volts, A.C., every three to six months. That is about the time that ground fault circuit interrupters (GFCI s) were invented. Their sole purpose was to prevent electrocutions, and once they were phased into new construction, GFCI s have been phenomenal at preventing accidental electrocutions. GFCI s are preventing most 120 volts, A.C., electrocutions today. When all circuit breakers are eventually changed to AFCI s, it is believed AFCI s will eventually prevent most short circuit fires. Electronics in a GFCI constantly monitor the electrical current being supplied to the circuit through the Hot conductor, and the current returning through the Neutral conductor. Whenever these currents differ by more than 0.005 ampere, the GFCI assumes that the difference current is going to ground through a human body, and it opens the circuit (pops it) within 0.005 second to prevent an electrocution. Some years ago, the author conducted an experiment with a GFCI. He slowly burned through a two wire power cord fed by the GFCI to create a carbon path arc. The arcs in these experiments never popped the GFCI. The reason is that even in the range of 150

ampere arcing currents or more, the detection circuitry in the GFCI never saw a difference between the current in the Hot conductor and the Neutral conductor. However, three-wire power cords, extension cords, and NM cables are different, as proved during recent experiments. To begin, if the arc first occurs between the Hot conductor and the Grounding conductor, the GFCI immediately trips within one cycle (1/60 th second) to prevent a fire. From the physical makeup of power cords and NM cables (except for bare gounds), it should be clear that a defect in the insulation will occur between the Hot conductor and the Grounding conductor 50% of the time and between the Hot conductor and the Neutral conductor 50% of the time, on average. This is because there is no physical difference between the Neutral conductor and the Grounding conductor (bare grounds excepted), except for their colors. For three wire cords and cables (not two wire), when the short circuit arc first occurs between the Hot conductor and the Grounding conductor, the GFCI will trip immediately. But even when the arc begins between the Hot and the Neutral conductor in three wire cords and cables, research shows that the arc occasionally extends to the adjacent Grounding conductor (in 2 of 9 tests). As soon as the arc first touches the Grounding conductor, the GFCI pops immediately. The bottom line is that household GFCI s have somewhat greater than a 50% probability of preventing a short circuit arcing fire when they feed three conductor cables and cords, in the author s opinion. SURGES: When a parallel arc occurs between the service conductors feeding a house, it generates very high voltages inside the house. These high voltages are due to the surging electrical currents in the parallel arc and the inductance of the windings in the power transformer producing those currents. The author has had a number of cases where surge voltages from service cable arcing outside the house broke down (pierced) the electrical insulation in the branch circuits inside the house. This resulted in one or more arcing fires in the branch circuits inside the house. VEHICLE FIRES: Vehicle short circuit fires are much different from household arcing, because in vehicle fires, it is a glowing arc, not a sputtering arc, which causes the fire. The vehicle voltage is 10 times less and the arcing currents are roughly 10 times less, as illustrated in Video 5. This results in 100 times less power in the arc. But the total arcing energy is similar, because the arcing can go on for a minute or more. In Video 5, you may see that as the torch slowly burns through the vehicle conductors to produce a carbon path arc, the arcing currents last for over a minute and finally melt 9

both conductors completely apart without ever drawing over 20 amperes of electrical current. So this arc would not have popped a 20 ampere vehicle fuse. The arc also does not last long enough to pop a 15 ampere fuse. But the arc does draw enough current to pop a 10 ampere vehicle fuse quickly. It has been the author s opinion for 20 years that the automotive industry should use more fuses with a 10 ampere rating, or smaller, to prevent short circuit arcing fires 3. Figure 5 is a typical logarithmic time-current curve for one brand of vehicle fuses. In Figure 6, these time-current curves have been used to draw energy curves on conventional scales. It may be observed that the let-thru energies drop dramatically at a certain threshold, much like the magnetic trips in residential circuit breakers. FIGURE 5: TYPICAL LOGARITHMIC TIME-CURRENT CURVES FOR VEHICLE FUSES 10

FIGURE 6: ENERGY ALLOWED INTO THE ARC BY 5, 10 AND 20 AMPERE FUSES A thorough discussion of this subject is given in the article Vehicle Fires and Their Prevention, 3 in the author s CV on his website. VEHICLE AFCI S: An AFCI-like device was invented for automobiles in 20056, but unfortunately, it is not in use. Many years ago, the author sent the information in this presentation and his video tapes to Boeing Aviation. AFCI s have now been flight tested for airliners. Unfortunately, there has been no such success with the shipping and boating industries, or the U.S. Coast Guard. 11

FIGURE 7: ARTIST S DRAWING ILLUSTRATING A LATENT SHORT CIRCUIT DEFECT LATENT SHORT CIRCUIT DEFECTS: Video 6 discusses how a power cord can suddenly arc when no one is around to move it or disturb it in any manner, and how such latent short circuit arcing can occur months or years after a power cord or cable is damaged. CONCLUSION: In summary, videos on the author s website illustrate the typical energy in short circuit arcs which cause fires in 15 and 20 ampere circuits. It is pointed out that the energy allowed into an arc by an AFCI and the copper melts allowed by an AFCI may be used as a reference for amounts of arcing energy too low to cause a fire, within a reasonable degree of engineering certainty. This is useful to understand which fuses will allow an arcing fire, which conductors are too small, whether small transformers can allow a fire, and to make comparisons with various other types of electrical and electronic equipment. 12

CASE STUDY: VEHICLE SERIES FAULT: Deposition on August 1, 2005: FIGURE 8: THE ONLY MELTED COPPER WAS AT THIS SERIES CONNECTION The fire in this 1991 automobile s engine compartment destroyed two vehicles and a house. The fire was caused by series arcing at the alternator s push-on connector for the B+ battery cables. This connection was the only location in the entire engine compartment where a copper melt had occurred. After diligent research, my client attorney discovered that this alternator connector has an acknowledged overheating problem. After the author gave a 3 ½ hour deposition, the case settled well. CASE STUDY: SURGE VOLTAGES CAUSE TWO ARCING FIRES: FIGURE 9: TWO TOTALLY SEPARATE ORIGINS, IN BASEMENT AND UPSTAIRS 13

This arcing fire in the wiring below the vent fan in the basement and the arcing fire in a power cord in the upstairs bedroom occurred just after the homeowner and another witness saw and heard arcing in the service cable outside the house. They went to the basement and spent 20 minutes dealing with the fire there, without realizing that another fire had begun in the first level bedroom. That fire then extended to other areas of the house. This case was settled without a deposition. CASE STUDY; VEHICLE ARCING: Case No. 08-116875-NP; 3rd Judicial Court, Wayne County, Michigan: FIGURE10: SUV FIRE In this case, a brand new SUV was stolen from the dealership. It was found abandoned on a street about 10 days later and towed into a building owned by a wrecker service. Some hours after that, the firemen arrived to find fire coming only from the SUV. At trial, the author testified that with a cold engine, the only source of heating power in the SUV was a short circuit arc. The opposing expert claimed this was an arson fire, in spite of the security system in the building which never tripped to indicate that anyone was in the room, and in spite of any other lack of evidence, except his claimed burn patterns. Significant amounts of copper melting were found in the left, rear corner of the SUV. To explain how vehicle short circuit arcing occurs, the author showed Video 5 to the jury. He also stated that greater use of 10 ampere fuses and AFCI s might have prevented this fire. The jury ended up agreeing that this was a short circuit fire, but they decided that 14

the wiring in the SUV could have been damaged by the thief, and found against the wrecker service. CASE STUDY; SERIES ARCING: Case No. 06 CV 66720, Court of Common Pleas, Warren County, Ohio: FIGURE11: HOT TUB REMAINS In this fire, after the plaintiff s electrical expert and the author examined the hot tub remains at the plaintiff expert s facility, the plaintiff expert testified that this hot tub was ignited by a series fault. He claimed the series fault occurred where a 30 ampere conductor was found separated from its connector, even though there was no copper melt on either end of the separation. However, when the author examined his earlier photographs from the fire scene, he found that this conductor was not even disconnected before the hot tub was removed from the scene. It had come apart during transport. The homeowner eventually admitted that she had dumped ashtrays into the plastic garbage can near the hot tub prior to the fire. The case settled well for the author s client. CASE STUDY: ELECTRIC MATTRESS PAD: USDC Case No.: 1:04-0231: This homeowner awoke in her bed to find her legs on fire. The author found a box of cigarettes and lighters next to the bed. Furthermore, the lady eventually admitted that she had ignited her clothing on a previous occasion while smoking in bed, that she had been drinking and that she had taken sleeping pills just prior to this fire. 15

The plaintiff sued the manufacturer of this lady s electric mattress pad, which has the same type of tiny heater element wires as an electric blanket. The plaintiff s electrical expert admitted that we could find no copper melt anywhere in the electric mattress pad, but that it was still the cause. To prove how little energy a short circuit in the heater wires could produce, the author made Video 4 for this case. His client attorney stated that this video saved a substantial amount in the eventual settlement. BIBLIOGRAPHY: 1. Frederick F. Franklin, A Survey of Electrical Fires, The Fire and Arson Investigator (IAAI), Dec. 1981: 35-37. Reprinted in Fire Journal (NFPA), Volume 78, No. 2, March, 1984, pages 41-44 2. Frederick F. Franklin, Circuit Breakers: The Myth of Safety. Professional Safety (ASSE), June 1990; 28-31. Reprinted in The Fire and Arson Investigator (IAAI), June 1991 3. Frederick F. Franklin, Vehicle Short Circuit Fires and Their Prevention, Fire and Arson Investigator (IAAI), Sept. 1992. Reprinted in Professional Safety (ASSE), August 1993 4. Frederick F. Franklin, Latent Short Circuit Defects, Fire and Arson Investigator (IAAI), December 1991. Reprinted in Professional Safety (ASSE), Sept. 1992 5. Underwriters Laboratories, Inc., File E87837, Project 92ME51901, Fact-Finding Report on An Evaluation of Branch-Circuit Circuit-Breaker Instantaneous Trip Levels for Electronic Industries Association, Washington, D.C., October 1993 6. Malakondaiah Naidu, Thomas J. Schoepf and Suresh Gopalakrishnan, Arc Fault Detection Schemes for an Automotive 42 V Wire Harness, Society of Automotive Engineers (SAE) 2005-01-1742, April 11-14, 2005 16