The NFPA's National Electrical Code (NEC1) may be the most technically detailed code - on any topic - that exists in U.S. practice. It has served well for the 110 years of its existence, especially in the area of electric shock hazard. While electrocutions still do occur, they usually require "a chain of human errors," something that a code generally cannot eliminate.
The other role for NEC is in fire safety. Here, the safety record has certainly improved over the decades, but there remains room for further improvement. A few years ago, this author had the opportunity to comprehensively examine the state of the art of the knowledge underlying electrical fires. This first-ever systematic review was published in 2003 as part of the Ignition Handbook.2 The review indicated that while some topics had received considerable study, many other aspects of electrical fires are still not well enough understood. It is the purpose of this article to consider some of the salient gaps. In addition, some existing, but not well-publicized, research has already identified some safety concerns that should be addressed by appropriate actions.
The system of obtaining fire statistics in the United States, based on NFPA 9013 and the U.S. Fire Administration's NFIRS,4 generally works well for identifying the broad problem areas. But it is very limited in the quality and accuracy of information that it can produce when it comes to fires caused by electrical or mechanical equipment.
To see why, it is necessary to consider how these statistics are obtained. NFIRS data are supplied by individual fire departments. Once the fire is extinguished, the fire department fills out a form describing the details of the event, and most fire departments use the coding system contained in NFPA 901 and NFIRS.
It is important to understand who fills out the form. The delegated individual can either be a fire company officer or the department's fire investigator. Except for the largest cities, it will be rare indeed that this individual would have any detailed knowledge of electricity or electrical fires. The national statistics2 (Table 1) reported that around 33% of electrical distribution fires are attributed to "fixed wiring."7 This is by far the largest single cause of electrical fires, with the next-largest category, "light fixture, lampholder or sign" accounting for only about 16%.
These statistics are almost certainly biased and incorrect. Fixed wiring, while it has various potential failure modes, is generally robust, in comparison to connections in various devices (e.g., outlets) or cords used by occupants.
It is easy to see why such statistics would be reported. If the individuals delegated with the task of providing the NFIRS information do not have good knowledge of both electricity and the modes of faults and failures, one can hardly expect them to provide detailed, reliable information on the topic.
This a not a criticism of fire departments in this regard. In most jurisdictions, the main investigation task for the fire department is to determine if arson has been committed. If it was not, their responsibility typically ends. Since it is exceptionally rare for arson to be committed by tampering with electrical systems, fire department investigatory expertise should not be expected in electrical matters.
Progress could be made on this issue, without requiring a major change in fire department training or investigation procedures. In actuality, fire investigation in the United States is typically a three-tier system. The initial investigation is conducted by the local fire department. In all except trivial losses, the next investigation is normally performed by a private fire investigator sent out by the insurance company that has insured the property. This investigator typically has more extensive training and better resources for conducting the investigation.
In addition, it is not uncommon for a private investigator to devote several days to investigating a fire that the fire department investigated in a few hours. If the loss is large, then generally the third tier comes into play, whereby electrical engineers, metallurgists or similar professionals examine the evidence in great detail. The third tier is invariably associated with lawsuits, and it cannot readily be expected that these detailed results would be disseminated. The NFIRS system could readily be changed to make use of the second-tier information. A system could be set up whereby insurance companies that have sent out fire investigators to do a detailed investigation feed the information obtained back to the fire department, which, in turn, uses this as the basis for the NFIRS report, instead of the fire-cause information derived during the fire department's own, much briefer investigation.
RESEARCH ON MECHANISMS
In order to best make progress, it will be helpful to have a good knowledge of the underlying physical mechanisms that lead to fire, rather than solely categorizing fires by the type of device that has failed. Published statistics on this point are, again, dubious (Table 2). The eight categories listed in Table 2 are the only categories into which the national NFIRS fire data system recognizes as electrical "Factors contributing to ignition."
Most specialists consider that short circuits are not a major cause of fires (Table 3). Instead, the primary mechanisms for causing electrical fires are poor connections and arc tracking, with overloads being a much-lower third cause. Poor connections and arc tracking are, in fact, completely missing from NFIRS. In addition, contributory factors that must be considered are mechanical damage and aging. "Overloads" are to be construed broadly and are meant to include not just exceeding the nominal ampacity of a circuit by excessive load but also problems that cause the effective ampacity to be decreased, e.g., interfering with cooling by excessive thermal insulation.
The review undertaken in the Ignition Handbook2 indicated the available technical knowledge on electrical fire mechanisms is spotty. In some areas, for example, the aluminum wiring problem or "last-strand" failure of power cords, extensive research has been done and the failure modes have become understood quite well. But in other areas, research has been so limited that good guidance cannot be given. This article will examine some salient areas that are either notably short on research or where research failed to lead to needed actions: arc tracking, mechanical damage, aging, and reduced ampacity.
Poor connections (Table 1) is one of the areas where basic physics has been identified,2 due largely to research conducted in Japan and by the Consumer Product Safety Commission (CPSC). The physics involved is complex, and CPSC does not have any sustained project to advance the knowledge further. More recent progress has been made by research6 conducted at the Eaton Corp., but public-sector or university research is notably scant.
ARCING ACROSS A CARBONIZED PATH
Electrical current is intended to flow only through conductors, e.g., copper wires. Wires are to be separated from each other by insulators, which must not conduct electricity. The bulk of modern electrical insulators are plastic, and these are susceptible to a failure mode whereby they stop being a good insulator and become a semiconducting material. This process can take place when the material is excessively heated, or carbonized.
The carbonization can occur due to external thermal heating, or due to current flow itself. The latter becomes possible if a wet, polluted insulator surface allows small amounts of current to start flowing along the surface of the insulator; the latter phenomenon is more specifically identified as "wet-tracking." The fact that fires can occur due to arc tracking has been known since the pre-World War II era, yet research available on the topic is limited.
For house wiring and wiring in appliances and equipment, the most common plastic is PVC (polyvinylchloride). Japanese research2, 7 identified a unique failure mechanism: wet-tracking induced under originally dry conditions (Figure 2). Most electrical-wiring grades of PVC contain a calcium carbonate filler. When initially dry, but subjected to sustained, yet only modestly elevated temperatures (around 120 C), PVC begins to decompose and reactions with calcium carbonate cause a hygroscopic reaction to occur along the surface. If continued, this can then lead to wet-tracking and possibly to an electrical fire. Thus, PVC is a unique material that can fail by wet-tracking even when it is located in a dry environment.
Additional Japanese research showed that PVC wire insulation is very susceptible to wet-tracking even at the Japanese supply voltage of 100 Volts, which is slightly lower than the 120 V used in the United States. This is of serious concern because standard test procedures used for assessing wet-tracking (IEC 601128 and UL 746A9) give misleading results for PVC, incorrectly indicating that it can withstand much greater voltages without tracking failure. This appears to be due to the physical configuration of the test method, which does not resemble construction of actual cables. Tests conducted on actual cables demonstrated that PVC-insulated cables have very poor resistance to wet-tracking.7 The IEC Technical Committee in charge of IEC 60112 has been notified of this problem, but has not taken any action.
It is no surprise that mechanical damage can lead to failures and fires with electrical components. One of the few U.S. studies where more detailed statistics on electrical failures were obtained identified that "mechanical damage" was the foremost cause leading to fires originating in both branch-circuit wiring and in cords and plugs.10 Yet, such statistics do not tell how these fires occur. Mechanical damage can cover a large number of different phenomena, including both manufacturing defects and user abuse. A few have been studied in detail,2 primarily by Japanese researchers. These include parallel arcing across a damaged line cord, "last strand" failure of power cords, and severing of wire conductors by a penetrated staple. Yet others are notably lacking in research. This includes factors such as progressive creep and nicked or gouged insulation.
This situation is especially of interest since the thickness of electrical insulation in 120 V circuits is determined by serviceability considerations, not by electrical breakdown considerations. Minuscule insulation thickness would suffice at these voltages, were it not for the need to provide durability of the insulation over time. Yet, studies on this point are unavailable.
A study by CPSC11 identified "deterioration due to aging" as being the factor leading to electrical distribution fires in 13.8% of the cases. These determinations were conducted by CPSC field-office personnel, who presumably have some electrical knowledge and skills.
But, are such determinations realistic or correct? In fact, there are no guidelines that would allow an individual - even a highly experienced electrical engineer - to correctly categorize a fire as being due to this cause. A detailed review2 indicated that research on aging of electrical wiring has been nearly non-existent. Recently, NFPA's Fire Protection Research Foundation (FPRF) completed a research project12 on the subject of "Aging of Residential Wiring" and a related study,13 "National Electrical Grounding Research Project." This research is an important first step, but the data on aging of wiring collected by FPRF were primarily better statistics and the work did not encompass physics and chemistry that would allow the failure mechanisms to be more fully understood.
Even the very concept of aging of wiring is fuzzy and poorly founded. Aging of any device implies that, after a certain length of time, its service life will be expired and it must be replaced if functionality is to be maintained or a hazardous condition avoided. Thus, the concept would require that, if electrical wiring or an electrical device is over X years old, it must be removed and replaced.
Manufacturers of smoke alarms have in recent years taken the posture that smoke alarms over 10 years old should be replaced. This type of strategy is generally unknown for other types of electrical devices. And, even for smoke alarms, there has been no technical basis for the 10-year period. For wiring, this would require that an endpoint be rationally established and wiring exceeding this limit be mandated to be replaced. But at the present, since research is lacking, there would be no basis to make such a determination.
For electrical wiring, inextricably intertwined with aging is the concept of temperature classes. Any physical material has a maximum temperature which, if exceeded, can result in precipitous failure. The electrotechnical community, specifically IEEE and UL, have established a methodology for classifying insulation materials into temperature classes; 14, 15, 16, 17, 18 internationally, IEC uses different standards, but with the same philosophy.
A wide range of these classes exists, but the common ones are 60C, 75C, 90C and 105C. There are a number of problems with the scheme, but a major problem is that endpoint failures have not been rationally studied. In most cases, a number of tests are used to establish the temperature class (in some cases, however, classification is based solely on "experience"), but none of them determine ignition or even gross electrical failure under conditions reasonably representative of in-use conditions. Some of the tests relied upon are very indirect, i.e., mechanical testing for the ultimate elongation at the breaking point; it is by no means clear that their results can be correlated to the initiation of fire.
Correct determination of the maximum temperature to which an insulation is permitted to be subjected is of major importance for both safety and functionality. Thus, it is very troubling that research validating the IEEE/UL/IEC methodology is lacking. The basic scheme originated in a 1948 paper by Dakin.19 It represented the best-available research for 1948; however, much has changed in the last 60 years. Even PVC as the insulating material for building wiring did not come into use until after 1948.
In these intervening decades, there have not been any significant studies completed that would verify the soundness of this methodology. In 1974, Ontario Hydro conducted tests 20 on a variety of commercial PVC-insulated cables. The cables were rated at 90C or 105C, but in each case the results were that after roughly one month of aging at 71 - 77C, the cables were deemed to have functionally failed. Consequently, the report concluded that these 90C and 105C- rated cables must not be used at temperatures that exceed 71C.
Thus, the only available study that attempted to validate the temperature-class scheme not only did not confirm Dakin's methodology, but found that serious errors on the unconservative side were being made. It is further troubling that these unconservative results emerged after only a month's worth of aging. In actual use, wiring must safely function for decades, not month-long periods. Thus, true long-term aging may well reveal a more problematic situation than showed up in the Ontario Hydro testing. Of equal concern is a more recent study21 where the authors examined NM cables subjected to 90C temperature and found failure at 4,020 hours, or about half a year. Equally problematic is that in the IEC version of the temperature class rating procedures,22 it is explicitly stated that the lifetime only needs to be 20,000 hours, which is only 2.28 years.
The ampacity of wire or cable is the current that it is allowed to safely carry. Calculating ampacity requires a detailed engineering effort, because it is affected not only by the construction details of the wire or cable but also by the environment. The ampacity ratings published by NEC are reliable, but only if the usage conditions correspond to the calculation premises. Often, this may not be true.
The temperature attained by a conductor is governed by the interplay between the heat produced (due to flow of current) and the heat lost (due to convective cooling). The latter, however, depends not only on details of the cable but also on anything that affects convective cooling, e.g., presence of other adjacent conductors and presence of thermal insulation.
Most are aware that it is unsafe to stick electric cords under carpets, mattresses or similar environments since they may overheat. Electricians who install NM-type house-wiring cables generally have no basis to know that they may be creating an overt fire hazard. This hazard normally arises because thermal insulation is used in a wall or ceiling cavity where the cabling is located (Table 3).
Goodson et al23 became concerned upon finding houses under construction, where NM cables suffered charring damage. They then conducted a study where they measured the short-term temperatures obtained on cables in wall cavity spaces. They noted that the standards for circuit breaker performance do not require tripping if the sustained load is less than 120% of the rating. Thus, they considered that the appropriate current flow to consider for establishing the thermal hazard is 120% of the breaker rating.
Using 90C-rated NM cables, a single run of cable in an uninsulated cavity space registered 114C, which is well above the 90C limit (even setting aside Stricker's finding that a 90-rated cable may fail at 71C). As insulation was added to the cavity spaces and as additional runs of NM cable were placed adjacent to each other - as fully permissible in the NEC - the situation got progressively worse. With three parallel runs of cable and fiberglass thermal insulation, for example, a temperature of 211C was reached.
Goodson et al were not the first researchers to be alerted to this problem. Already in 1980, Evans24 at NBS studied the temperatures developed on cables covered by fiberglass thermal insulation. His study was less extensive and involved only single runs of cable, but he too reported that actual temperatures exceeded the rated 90C value.
Vytenis Babrauskas is with Fire Science and Technology Inc.
- NFPA 70, National Electrical Code, National Fire Protection Assn., Quincy, MA, 2008.
- Babrauskas, V., Ignition Handbook, Fire Science Publishers and Society of Fire Protection Engineers, 2003.
- NFPA 910, Standard Classifications for Incident Reporting and Fire Protection Data, National Fire Protection Association, Quincy, MA, 2006.
- "The National Fire Incident Reporting System (NFIRS)," FA-290, U.S. Fire Administration, Washington, DC, 2007.
- Babrauskas, V., "Research on Electrical Fires: The State of the Art," Fire Safety Science-Proc. 9th Intl. Symp., Intl. Assn. for Fire Safety Science, London, 2009.
- Shea, J. J., "Glowing Contact Physics," 52nd IEEE Holm Conference on Electrical Contacts, IEEE, New York, 2006.
- Babrauskas, V., "Mechanisms and Modes for Ignition of Low-voltage, PVC-insulated Electrotechnical Products," Fire & Materials 30, 150-174, 2006.
- IEC 60112, Method for the Determination of the Proof Tracking and Comparative Tracking Indices of Solid Insulating Materials, International Electrotechnical Commission, Geneva, 2003.
- UL 746A, "Polymeric Materials - Short Term Property Evaluations," Underwriters Laboratories Inc., Northbrook, IL, 2000.
- Hall, J., Bukowski, R. and Gomberg, A., "Analysis of Electrical Fire Investigations in Ten Cities," NBSIR 83-2803, National Bureau of Standards, Gaithersburg, MD, 1983.
- Smith, L. and McCoskrie, D., "What Causes Wiring Fires in Residences," Fire Journal 84:1, 19-24, 69 Jan/Feb 1990.
- Dini, D., "Residential Electrical System Aging Research Project," Fire Protection Research Foundation, Quincy, MA, 2008.
- Lindsay, T., "National Electrical Grounding Research Project," Fire Protection Research Foundation, Quincy, MA, 2007.
- IEEE 1, General Principles for Temperature Limits in the Rating of Electrical Equipment and for the Evaluation of Electrical Insulation, IEEE, New York, 2000.
- IEEE 98, Standard for the Preparation of Test Procedures for the Thermal Evaluation of Solid Electrical Insulating Materials, IEEE, New York, 2002.
- IEEE 99, Recommended Practice for the Preparation of Test Procedures for the Thermal Evaluation of Insulation Systems for Electrical Equipment, IEEE, New York, 2007.
- IEEE 101, Guide for the Statistical Analysis of Thermal Life Test Data, IEEE, New York, 1987.
- UL 1581, Reference Standard for Electrical Wires, Cables, and Flexible Cords, Underwriters Laboratories Inc., Northbrook, IL, 2001.
- Dakin, T., "Electrical Insulation Breakdown Treated as a Chemical Rate Phenomenon," Trans. AIEE 67, 113-122, 1948.
- Stricker, S., "Thermal Design of PVC-Insulated Heating Cable," Report 74-26-K, Ontario Hydro, Toronto, 1974.
- Morsy, M. and Shwehdi, M., "A Supportive Approach into Life Testing and Characterization of PVC and XLPE-Insulated Cable Materials," 2005 Annual Report, Conf. on Electrical Insulation and Dielectric Phenomena, IEEE, New York, 2005.
- IEC 60216-1, Electrical Insulating Materials - Properties of Thermal Endurance - Part 1: Aging Procedures and Evaluation of Test Results, International Electrotechnical Commission, Geneva, 2007.
- Goodson, M., Perryman, T., and Colwell, K., "Effects of Polyurethane Foam Systems on Wiring Ampacity," Fire & Arson Investigator 52:4, 47-50, July 2002.
- Evans, D., "Calculated Operating Temperatures of Thermally Insulated Electric Cables," NBS TN 1133, National Bureau of Standards, Gaithersburg, MD, 1981.