Smart Stove Fire Detection Technologies

This page may contain links to non-U.S. government websites. What this means to you »

Development and Testing of Temperature Sensors for Preventing Cooking Fires on Glass Ceramic Electric Ranges

Executive Summary

The overall objective of this project was to identify sensors that could be used with a smooth-top electric range to sense pending ignition in a cooking vessel and intervene to prevent range-top fires. The Consumer Product Safety Commission (CPSC) prepared a report in 1999 that documented 85,000 fires that occurred annually involving range tops and ovens that were attended by fire departments. Deaths averaged 250 annually along with 4,080 injuries and a loss of $295.6 million in property damage during the four-year period covered in the report. This report clearly shows that unattended operation is a common factor in most of the fires. Further data collected for the 1994 to 1998 period show that 47,200 fires originated from food preparation on the cooktop as opposed to the oven. These fires alone resulted in 80 deaths with an additional 2,440 injuries and $134.6 million in property loss.

An off-the-shelf smooth-top electric range was procured for use in this project. The range was modified to incorporate thermocouples and other instrumentation to allow the temperatures of the glass, the heating element, ambient conditions under the glass and the cooking pan to be monitored during various cooking scenarios. A series of baseline laboratory tests were conducted to determine the thermal environment within the heated zone of the range during normal cooking operations such as boiling water and blackening meats. The testing resulted in two significant findings; (1) the emissivity1 of the cooking pan strongly impacts the operating temperature of the glass, and (2) the thermal time response of the glass cooktop was significantly slower than the response of a typical cooking pot or pan.

The original concept for a fire-prevention control was to radiantly couple a thermal sensor located below the glass to the bottom of a pot. While this was successfully accomplished, it became apparent that, due to wide differences in pan emissivity, the response of the thermal sensor varied widely. In fact, observations taken during a normal cooking operation such as boiling water with a low-emissivity pan (stainless steel, for example) fell within the range of readings taken from a high emissivity pan (cast iron) during an event that concluded with ignition. Thus, a radiant sensor was excluded from further consideration.

Simply measuring the bottom of the glass with a contact temperature measuring device such as a thermister or thermocouple showed promise in the early testing. Because the glass temperature rise significantly lags the cooking pan temperature rise, the response of the control was dependent upon the previous operations conducted on the range and on the existing thermal conditions of the range components at the time that an ignition event is likely to occur. For example, if the range was used to boil water for several minutes, the glass, heating element and the air under the range top would be preheated to a safe but high level. Then, if a pan of oil or other flammable food material was heated to a temperature approaching ignition, simply measuring the temperature of the glass bottom would be sensitive enough to detect pending ignition and interrupt the heat input.

If, on the other hand, the range was initially at room temperature and a pan of flammable food material was heated until ignition, the glass temperature reading would indicate a value well within the normal, safe limits. This was due to the thermal lag of the glass and is a result of the optical/thermal characteristics of the glass itself. The glass is formulated to be fairly transparent to the heat source wavelengths. Thus, most of the radiant heat from the heating element is radiated directly to the pan. The glass is heated from a combination of convection from the air in the heating element space, conduction from the pan itself and absorbed radiation passing through the glass from the heating element and as re-radiation from the pan. Because of the strong radiant coupling between the heater and the pan, pan temperature rises faster than the glass temperature. Once the system has been preheated, glass temperature follows pan temperature fairly closely.

Several attempts were made to create a thermal environment wherein the glass could closely track pan temperature. For example, an effective insulation system was devised to isolate the glass sensor from the environment under the glass. This system consisted of three concentric rings of low-emissivity material (stainless steel) sandwiched with thin ceramic fiber insulation. Further, the glass was machined with a diamond hole saw such that a circular "moat" was formed around the region of the glass that was being sensed. Even with the high thermal isolation of the glass via the moat, glass temperature did not respond rapidly enough. Moreover, the machining operations effectively weakened the glass to the point that it would have failed the Underwriters laboratories (UL) drop test.

All data collected during testing was logged using LabVIEW software. The resulting data files were analyzed for trends in temperature response. After detailed analysis, there appeared to be a solution to the glass response time in the form of a control algorithm that responded to the timerate-of-change of glass temperature; the first derivative with respect to time. By incorporating a term in the algorithm that was derivative-dependent and adding it to the measured glass temperature, the resulting value closely matched actual pan temperature under all conditions and with all pan sizes and types.

The resulting control system consists of a simple thermal sensor held in contact with the glass bottom and isolated from the heating element with insulation. No machining or other modifications to the glass are necessary. The control simply monitors actual glass temperature, modifies the reading based on the instantaneous derivative, and interrupts power to the range top when the resulting value reaches a dangerous level.