Uncertainty analysis of steady state incident heat flux measurements in hydrocarbon fuel fires.

Abstract

The objective of this report is to develop uncertainty estimates for three heat flux measurement techniques used for the measurement of incident heat flux in a combined radiative and convective environment. This is related to the measurement of heat flux to objects placed inside hydrocarbon fuel (diesel, JP-8 jet fuel) fires, which is very difficult to make accurately (e.g., less than 10%). Three methods will be discussed: a Schmidt-Boelter heat flux gage; a calorimeter and inverse heat conduction method; and a thin plate and energy balance method. Steady state uncertainties were estimated for two types of fires (i.e., calm wind and high winds) at three times (early in the fire, late in the fire, and at an intermediate time). Results showed a large uncertainty for all three methods. Typical uncertainties for a Schmidt-Boelter gage ranged from ±23% for high wind fires to ±39% for low wind fires. For the calorimeter/inverse method the uncertainties were ±25% to ±40%. The thin plate/energy balance method the uncertainties ranged from ±21% to ±42%. The 23-39% uncertainties for the Schmidt-Boelter gage are much larger than the quoted uncertainty for a radiative only environment (i.e., ±3%). This large difference is due to the convective contribution and because the gage sensitivities to radiative and convective environments are not equal. All these values are larger than desired, which suggests the need for improvements in heat flux measurements in fires.

Validation experiments to determine radiation partitioning of heat flux to an object in a fully turbulent fire.

“…Figueroa et al [2005] estimated the maximum uncertainty in heat flux using the inverse heat conduction program IHCP1D (Beck Engineering & Associates) to be about ±20% for a cylindrical calorimeter near, but not in, a fire. Nakos [2005] showed that the incident flux uncertainty could be as large as ±40% in situations broadly similar to those of the present experiments. An uncertainty of ±40% will be assumed for the present experiments both in Penlight and in FRH.…”

“…In the absence of a known calibration constant for the convective heat flux of the gauges used in the present experiments, the calibration constant of the convective fraction will be assumed to have the same relationship to the radiative calibration constant as the gauge calibrated by Diller. The overall uncertainty for Schmidt-Boelter total heat flux gauges has been estimated to be ±23% for the net absorbed heat flux and ±39% for the incident heat flux under conditions similar to those of the present experiments [Nakos, 2005].…”

“…The higher uncertainties in abnormal thermal environments are caused by increased errors due to the effects of imperfect TC attachment to the test item. Any systematic bias errors that can be modeled, such as those due to TC attachment imperfection, are algebraically subtracted from the TC reading, and any modeling uncertainty about the nominal mean estimate of the bias is included in the RSS calculation [Romero et al, 2004;2005]. Table 5-1 provides a summary of the uncertainty sources present in a typical TC measurement.…”

“…(5.1) [Nakos, 2005]. In this equation, q net is the net heat flux into the gauge, mV is the output signal from the gauge, K refers to a calibration constant and F refers to a fraction of the total heat flux.…”

“…An analysis similar to that performed by Nakos [2005] for Schmidt-Boelter total heat flux gauges can be performed on a Schmidt-Boelter radiometer with the assumption that there is no convective heat transfer to the radiometer. If the gauge emissivity is taken to be that of Pyromark 2500, the gauge temperature is taken to be the temperature of the cooling water with an uncertainty of ±10%, and the uncertainty in the net radiative flux is ±5%, the total uncertainty in the incident radiative flux is ±10% when the net flux is 5 kW/m 2 and ±11% when the net flux is 30 kW/m 2 .…”