AFCI - Important Update from a Certified Fire Investigator

Margaret Simonson, PhD
Science Advisory Committee
National Association of State Fire Marshal’s
1319 F Street
Suite 301
Washington, D.C.

June 11, 2002


First let me thank you for providing me with the opportunity to address such a prestigious organization. The list of members on the Science Advisory Committee is impressive and includes several individuals whom I know personally and whose opinions I respect. I have been aware of the position of the NASFM as your support for the AFCI requirement has been used several times as a blanket rejection of the issues I raise against the AFCI.

The remarks attributed to me likely resulted from one or more of several demonstrations that I have given to show the effectiveness of the AFCI. These demonstrations have been attended by representatives of CPSC, UL, Cutler Hammer, the NEC code making panel, and other organizations familiar with the AFCI, its history and its technical performance. I have yet to receive a single comment from any person who has stated that what I am saying and showing is technically incorrect.

Your letter states that “...NASFM continues to hear reports that the devices may not perform in the ways claimed by AFCI manufacturers.” I have no information that will show that the AFCI will not perform as advertised by the manufacturers. I make it very clear in each of my demonstrations that the AFCI will do exactly what the manufacturers claim it will do. I follow this remark with the comment that the real issue is that the AFCI manufacturers do not clearly state what are the areas that the AFCI will not address.

To attempt to reduce a doctoral thesis to a few sheets of paper is difficult, and I continue to accumulate information on a daily basis, so please consider the following as not totally complete or fully researched, this is a work in progress.

Safety has a cost. If an AFCI cost the same or less than a normal circuit breaker, but improved safety .001% then it would be a desirable feature and it is probable that no one would have any disagreement. Regrettably, the AFCI costs more than a normal circuit breaker and that cost is being passed on to the consumer. A higher level of safety is being provided by installing the AFCI, but how much higher is that safety level and does that unknown level of increase justify the cost to the consumer? At a June 7,2002 meeting with the National Association of Home Builders, a representative of Cutler Hammer indicated that an AFCI will sell for approximately seven times the cost of a typical circuit breaker.

The main objections (random order) to the mandatory requirement for an AFCI are as follows:
1- The main cause of electrical fires is resistance heating, an area not addressed by the AFCI
2- Figures lie and liars figure. Valid statistical data to support or deny an AFCI is not available through any of the currently used fire incident data bases. Most fire investigators will state that electrical wiring fires normally start at a termination point (wire nut, junction box, etc). Failures at these points result in resistance heating or series arcing, areas not covered by the AFCI.
3- Many of the faults detected by an AFCI will also be detected by a normal circuit breaker and/or a GFCI. It is recognized that the AFCI will detect them faster, with a lower release of total energy.
4- Electrical arcing is not recognized as a common cause of electrical fires. An electrical arc will not normally ignite a wooden structural member.
5- The costs associated with mandating the AFCI should include the almost certain expansion of the requirement and the need for a larger sized load center to accommodate the devices.
6- The problems currently being investigated with the high failure rate of GFCI may also apply to the AFCI.
7- The requirement for the AFCI was adopted with minimal supporting data on a non-UL listed product. The consumer is now being used as a mandated test subject.
8- The original and present concern of the CPSC was the high percentage of fires being reported in homes 40 years old, not new homes.

I am attaching to this letter a slightly disorganized expansion of the above points with supporting statements in some instances. This attachment is also an evolving and incomplete work and not really ready to be released, but because of the apparent time restraints under which you are operating, I will enclose it.

As you are no doubt aware, there are counter arguments to many of the points that I raise. For your convenience, I am attaching a copy of a letter from Eaton Corporation which raises some counter issues. I am also enclosing a copy of an article that I recently had published in the “Fire & Arson Investigator” I invite you to consider both sides of the argument and will gladly make myself available to participate in any discussion/presentation. I also have copies of most of the reports, letters, tests, etc. and will make these available for review if needed.

There is little doubt that the AFCI is an improvement over the standard circuit breaker. The question that needs to be resolved is - Is the AFCI so much better that the consumer must be forced to install it.

Thank you once again for taking the time to review this matter. Your organization is to be commended for having the courage to re-evaluate its position on this matter.


Bernard A. Schwartz, P.E., CFI

Response to NASFM ltr. of 6/5/02

The available fire data is not adequate to give an adequate confidence level to determine how many fires could be prevented by a properly functioning AFCI. The following statements on INJURY DATA will perhaps give you an understanding of the foundation data upon which the justification of the need for this device is based.

1- NFPA, FEMA, and CPSC electrical fire data are all collected by computerized data sheets filled out by fire fighters, not fire investigators and virtually never by persons with specialized skill in the investigation of electrical fires.

2- The National Electric Code mandated the requirement for AFCI protection with only generalized statements of the electrical fire problem for supporting data.

3- Articles supporting the need for AFCI do not adequately address the number of fires caused by arcing nor do they address what percentage of fires will be prevented by use of an AFCI. Regrettably, there is no currently available injury data system that will provide adequate detail to support or refute the belief that a significant number (or percentage) of fires are started by arcing.

4- FEMA data codes which are used to fill out the computerized data forms used by the fire departments do not include resistance heating.
See Form of Heat of Ignition - National Fire Incident Reporting System Handbook - Version 4.1
Although fires caused by electrical arcs constitute only a small percentage of the fires, eight of the ten possible data codes for electrical fires use the word arcing. The remaining two are light ballasts and heat from overloaded equipment. Although resistance heating is thought of as the leading cause of electrical fires, it is not a choice on the computerized form.

5- The computerized data sheets are filled out by the senior fire fighter on the call and are filled out immediately upon return to the fire house. It is very unusual for a fire investigator to fill out the form or for the form to be completed after a detailed fire investigation is complete.

6- The fire data is based on all fires. An incident is reported each time the fire truck leaves the fire house with the siren going. This includes false alarms, smell of smoke and thousands of small fires where no or minimal damage is done.

CONCLUSION - Because of the minimal skill level of the person filling out the report and inadequate data codes, it is impossible to collect accurate data to support or deny the correct percentage of fires that are or are not started by electrical arcing.


1- The original UL report to CPSC outlined 14 tests that were needed if ALL electrical fires were to be prevented. The present UL standard only contains 4 of those 14 tests.
See article by David Dini of Underwriters Laboratories Inc. in IAEI News - September/October 2001.
CPSC study “Technology for Detecting and Monitoring Conditions That Could Cause Electrical Wiring System Fires” (September 1995). The recommendations section states “It was determined that no single product or technology in the examined state of development would provide protection against all electrical ignition scenarios likely to be encountered in residential wiring systems.”

2- The most common cause of electrical fires is the overheated connection such as a loose wire nut, back wired receptacle, or corroded terminal screw. This type of heating is often referred to as resistance heating or “Joule heating”.
In an article by David Dini of Underwriters Laboratories Inc. in IAEI News - September/October 2001, he states For example, It should not be expected that those ignition scenarios representing Joule (I2R) heating would necessarily be prevented by an AFCI.

An article by Dr. George Gregory of Square D Co. “Using Arc-Fault Circuit Interrupters to Reduce Residential Fires” states “An AFCI will not detect a glowing connection or hot spot, unless arcing is present.”

An article in Fire Technology (vol 36, No.3) - co-authored by William King of CPSC states “Existing AFCI’s do not detect another cause of electrical fires - the glowing connection - unless an arc or ground fault is also present.”

3- Some allegations are being made that when resistance heating occurs, an arc will eventually result and the AFCI will operate. This is a true statement. As the fire spreads, it is extremely likely that the spreading fire will contact energized electrical wires and cause a fault. This may or may not occur before the fire spreads beyond the junction box. UL has proposed two tests, the Overheating Conductor: Hot plug; and the Overheating Conductor: Glowing connection (“Technology for Detecting and Monitoring Conditions that could Cause Electrical Wiring System Fires”, September 1995 pages 103 and 108). If indeed the AFCI will prevent resistance heating faults, they should be capable of passing these tests.

4- In addition to an arcing condition, operation of an AFCI requires a minimum peak arcing current. For a Branch/feeder type AFCI, this current is approximately 50 amperes. See article by Dr. Joseph Engel, February/March 2002, NEC Digest which states “The electronic instantaneous trips for an AFCI for arcing faults are set to trip with PEAK arcing currents as small as 50 amperes.” UL standard 1699, table 50.2 notes that the branch feeder AFCI is not tested to the tests in section 56.4 which includes a 5 and a 10 ampere current load arc test. The branch feeder is tested to section 56.3.3, which uses a 75 ampere and a 100 ampere current.

5- Operation of an AFCI does not mean that it detected a condition that would not have been detected by an ordinary circuit breaker. Every AFCI includes all the functions of a normal circuit breaker as well as those of a GFCI (with the trip level set at 30 ma, not 6 ma.). Thus when an AFCI operates, it cannot be determined which function caused the operation or if the fault condition was detected by the AFCI, would the circuit breaker have still shut down the system in sufficient time to prevent a fire. An article in Fire Technology (vol 36, No.3) - co-authored by William King of CPSC states “Parallel arcing faults usually result in peak currents above the handle rating of the conventional circuit breaker. This may trip the circuit breaker magnetically, if the impedance of the fault is low and the available fault current is sufficient.”

6- NEC approved the AFCI requirement before the UL standard was even published. Thus the decision was made with a minimum of test data or real life field experience and the American consumer is now being used to gain the needed experience.
Cutler Hammer files for patent on AFCI Sept. 26, 1991

“Technology for Detecting and Monitoring Conditions That Could Cause Electrical Wiring System Fires” (September 1995).
December 1996 NEMA task force completes draft AFCI standard
Cutler Hammer passes UL DRAFT test in Nov/Dec 1996
1997 - Fall - NEMA transfers standard to UL
Cutler Hammer AFCI commercially available Sept. 30, 1997
May 1998 - NFPA code making panel (CMP-2) approves and ratifies new section 210-12
1999 - February - First edition of UL 1699 is published
AFCI mandatory for bedrooms January 1, 2002

7- CPSC started the original work to address the problems faced in older homes. CPSC continues to be concerned with the need for AFCI protection in older homes as evidenced by the last code change when the CPSC proposal for 2002 code asked for AFCI protection in older homes.

NEC proposal 2-119 by Wm. King of CPSC states “The objective of the study was to safely permit the occupancy of dwellings (many of which in this country are 40 or more years old with electrical wiring system elements remaining in service beyond their original design life) without manifestation of electrical symptoms that can cause fires.”

8- Although homes built today will someday be older homes, it must be remembered that older homes do not have many safety features required by today’s codes. One of the more obvious improvements in the code is that 3 wire Romex is now mandatory.

9- The GFCI is now being investigated. In certain areas of the country failure rates as high as 50% have been reported. The exact cause of the failures has not been defined, but one possible cause is lightning affecting the internal circuitry. In spite of the high failure rate, CPSC has not elected to re-call these NEC mandated safety devices. It is possible that the same problem may also present itself with the AFCI.

10- Presently there is no independent test device to determine if an AFCI is functioning properly. Pushing the test button is the only choice. As noted with the GFCI problem noted above, pushing the test button does not provide an adequate level of assurance of safe operation. An independent, certified test device is needed for inspection authorities to verify correct operation.

11- The AFCI is not a totally passive device. It is recommended that the AFCI be tested at least monthly. It is unreasonable to believe that the typical consumer is ever going to test the device.



1- The AFCI requires a full breaker space. This means that it is not a one to one replacement. Many load distribution panels (circuit breaker panels) accommodate “Piggy Back” breakers which allow two breakers to fit into the space normally required for one breaker. Installation of several AFCI’s could require the installation of a physically larger panel, adding to the overall cost.

2- Like the GFCI, the use of an AFCI is sure to be expanded. This was evidenced by the submission of proposed code changes during the most recent code cycle. The increased use will further increase the expense.

In a letter dated May 17, 2002 from Eaton Corporation they admit that installation of a full panel AFCI will add “...about $3.50 to the monthly mortgage payment.” For a 25 year mortgage this means an added cost of $1,050.00 to the consumer.

3- When the CPSC mandates a standard, a cost/benefit analysis is required. By introducing a mandatory requirement into an NFPA code, no such requirement exists. No cost/benefit analysis has been preformed and with the inherent weaknesses in the existing data, the accuracy of such a study would be subject to question.

AFCI vs Circuit Breaker

1- An AFCI will not address “SERIES” arcing until it becomes a parallel arc.
An article by Dr. Joseph Engel in the February/March issue of NEC digest he states “Also, series faults are mitigated as they tend to escalate either into a parallel arcing fault or a ground leakage fault.”

2- An AFCI will address “PARALLEL” arcing, but in most instances so will an ordinary circuit breaker. An AFCI will respond quicker than a standard circuit breaker and for some limited cases the arcing may be intermittent and/or below the instantaneous trip level of a circuit breaker.

3- The panel mounted (branch/feeder) AFCI is designed to protect the wire from the panel to the receptacle and is not tested to detect faults in the devices plugged into the receptacle. It may detect such device failures, but it is not required to with the UL test.
In an article by David Dini of Underwriters Laboratories Inc. in IAEI News - September/October 2001, he states “For example, a branch/feeder type AFCI is intended to be installed at the origin of the branch circuit to protect the branch circuit wiring against the unwanted effects of arcing faults.”

UL standard 1699, table 50.2 notes that branch feeder AFCI is not tested to the tests in section 56.4 which includes a 5 and a 10 ampere current load arc test. The branch feeder is tested to section 56.3.3, which uses a 75 ampere and a 100 ampere current.

Mike Holt's Comment: If you have any comments or feedback, please let me know,

Copyright © 2002 Mike Holt Enterprises,Inc.
1-888-NEC-CODE (1-888-632-2633)