Forensic Engineering:
Analysis of Copper/Brass Contacts
by Roger L. Owens, P.E. (NAFE 412F)

A great many articles have been written about the phenomena of high resistance electrical connections. These articles both suggest and acknowledge that in the presence of combustible materials, inferior electrical connections can cause a fire. In an attempt to explain the mechanics of a failure at an electrical connection, the authors discuss loose connections, dissimilar metals at the connection, oxidation at the contact points and a host of other variables. A thorough review of forensic publications provides very little information about copper to brass failures even though failures involving connections between other conductive metals are discussed in great detail. Copper and brass have a very high conductivity, are ductile, resist oxidation, and are generally more resistive to progressive failure as compared to other conductors when the mechanical connection is inferior. Further, the characteristics of copper and brass make it difficult to reproduce connection failures in the laboratory. This paper presents data from laboratory experimentation where localized temperatures at brass to copper connections escalated above the ignition temperature of light weight combustibles. Copper to brass connections can fail and the result can result in a fire. This article is intended to document laboratory testing of electrical connections involving good conductors, in this case copper to brass, where some of the mechanical connections were such that the given operating conditions caused some connections to reach temperatures sufficient to ignite light weight combustibles. Copper to brass connections were tested because copper to brass failures were alleged to be issues in the cases cited below and because no specific data was found in the forensic literature which specifically and empirically addressed copper to brass failure scenarios. It is generally accepted by knowledgeable people within the forensic community that improper electrical connections can cause a fire and this paper specifically documents that copper to brass connections are no exception.

Case Discussion
A manufacturer had experienced an unusually high failure rate at a terminal connection involving stranded copper to brass crimped connections at a switch located within an appliance. The failures were such that a potential for fire causation existed and the design of the connection had to be modified. A recall occurred and the manufacturer of the appliance sought to recover the cost associated with the recall from the vendor who manufactured and supplied the crimped connections. The manufacturer of the appliance and the manufacturer of the crimped connections disagreed on the specific cause of the overheat condition at the connection. An experimental project was undertaken by the author in order to determine the most probable failure mode at the connection. Those findings and test results are described in this paper. Further, the test results demonstrated that the failures were a result of inadequate crimping pressure during the manufacturing process for the connectors. Subsequent to the actual test project for the appliance manufacturer but prior to publication of the test results, a fire occurred in a residence involving two fatalities where it was alleged that the fire occurred as a result of a high resistance electrical connection between copper and brass in a receptacle. The attorneys for the manufacturer of the receptacle argued that the lack of published material involving copper to brass failures confirmed that it is improbable for copper to brass connections to initiate a fire. The test results obtained in the earlier experimental project was utilized to show that inadequate electrical connections between copper and brass can cause a fire just as poor electrical connections of other conductive materials can cause a fire. The results of the testing described in this paper assisted in the resolution and settlement of the case involving the receptacle by demonstrating that under the given laboratory conditions temperatures at a poor copper to brass connection reached ignition levels for light weight combustible material.

Experimental Project
Failures at the specific connections being studied were a known fire cause and this project was undertaken to determine the mode of failure. The actual connection was between an untreated stranded copper conductor and a crimped brass terminal lug (see Figure 1 and sketch below). The crimp connection was formed by an automated machine with the crimp height being the only variable. Since the conductor being crimped was stranded #12 awg, the crimp had to be of sufficient pressure and contact area to restrain movement of individual strands under normal thermal cyclic conditions. Thermal cycling or expansion and contraction of strands can cause movement of the strands. During the cool down transition, individual strands have the tendency to move to lower levels of both thermal and mechanical stress if not mechanically restrained. In those situations where the crimp is insufficient to inhibit movement, long term cycling can result in a voltage drop at the crimp which causes localized heating. Over time, a crimp joint of this type can reach an equilibrium safe state or can progress toward thermal failure. In certain situations, minor electrical arcs can occur causing a permanent conductive bridge across the boundary. This situation could be classified as self-healing and may account for the equilibrium thermal levels found at higher crimp height levels during these experiments. Further testing would be appropriate where such factors are a concern.

Figure 1

A series of fundamental tests were set up using some of the recalled connectors in an attempt to define the parameters associated with the connector failures. The height of the crimp was measured since overall contact pressure was a function of the height and shape. The connectors were then separated into lots in accordance with the crimp height and resistance and voltage drop measurements were performed. The data collected from all of those laboratory tests were inconsistent since the resistance and voltage drop was erratic due to the sensitivity of the test equipment. No realistic hypotheses or conclusions could be reached based on the data collected with instruments.

A test stand in an environmentally controlled chamber was fabricated in order to duplicate the conditions in the appliance where the connectors were utilized and to control the current flow through the connectors. A series loop of 32 new connectors were used with the only variable being the crimp height which is a function of contact pressure. The height was set by varying the pressure of the automated crimping machine and the connectors were then measured and separated into lots based on their crimp height (figure 1). Based on the measurements taken of the recalled connectors a range of height was selected starting at .050" and ending at .085". The increments were .005" with a tolerance of .001". Four connectors of each height were used with two crimped lugs per brass terminal (figure 2). Each terminal was fitted with a thermocouple and was calibrated to feed to a real time data recorder. An electric current of 12 amps was passed through the system ten minutes on and ten minutes off which simulated the normal current distribution in the appliance. The tests were continued for a total of 120.6 hours in 16 periods between four and nine hours. See Figure 3 which documents the overall test set up. It should be noted that the average ambient temperature in the chamber was controlled and measured at 130° F and that the temperature data collected was a rise in temperature above ambient. For example, a 300° F reading was actually equivalent to 430° F.

Figure 2

At the first three levels of crimp height (.050", .055" and .060") no appreciable rise in temperature was observed and the connection proved to be reliable for the given conditions. At the .065" level, elevated temperatures were observed but as the test continued, the temperatures stabilized. At the .070" and .075" positions, temperatures continued to escalate during the test period and reached temperatures above 425°F (temperature sufficient to ignite light weight combustible materials) at which time the tests were terminated. At the .080" and .85" crimp height, positions temperatures crept up, stabilized and remained below the temperatures recorded at .070" and .075" during the duration of the test.

The chamber utilized in the test maintained an average temperature of 130°F with a humidity at start up as high as 90% with a decrease to about 30% at the end of the cycle. Water reservoirs were never depleted even though the humidity decreased during the test period. See Figure 3 for overall test set up. The test data shows temperatures at 320°F rise or the equivalent of 450° during the test at the .070" test lugs. One of the .075" test lugs reached a similar temperature. The attached graph (Figure 4) demonstrates the rise in temperatures of the various crimp heights.

The test results presented demonstrate what is common knowledge among those schooled in failure analyses. Namely, that under certain conditions, the connection between good electrical conductors can fail. Specifically, this paper demonstrates that faulty copper to brass connections can fail in a manner such that temperatures can be generated sufficient to ignite combustible materials. For the first six levels of crimp height the failure mode appears to be functionally related to contact pressure since the temperatures recorded increase with increasing crimp height. Further an aging process or an oxidizing of the conductor materials was observed to increase consistently with increasing crimp height. The test data also indicates that, for the given geometry and current, the temperatures stabilized at the .080 and.085 inch crimp heights. This data suggests that at those particular crimp heights bridging occurred and the connections reached equilibrium. Additional testing including metallurgical analyses can be undertaken in order to further investigate the failure mode at those levels.

Figure 3

Figure 4

The author wishes to acknowledge that Beryl Gamse, Ph.D., P.E. and Joe Stainton, BSEE, P.E. were of tremendous assistance in setting up and monitoring the experimental project