On the uncertainty of temperature estimation in a rapid compression machine

Abstract

Rapid compression machines (RCMs) have been widely used in the combustion literature to study the low-to-intermediate temperature ignition of many fuels. In a typical RCM, the pressure during and after the compression stroke is measured. However, measurement of the temperature history in the RCM reaction chamber is challenging. Thus, the temperature is generally calculated by the isentropic relations between pressure and temperature, assuming that the adiabatic core hypothesis holds. To estimate the uncertainty in the calculated temperature, an uncertainty propagation analysis must be carried out. Our previous analyses assumed that the uncertainties of the parameters in the equation to calculate the temperature were normally distributed and independent, but these assumptions do not hold for typical RCM operating procedures. In this work, a Monte Carlo method is developed to estimate the uncertainty in the calculated temperature, while taking into account the correlation between parameters and the possibility of non-normal probability distributions. In addition, the Monte Carlo method is compared to an analysis that assumes normally distributed, independent parameters. Both analysis methods show that the magnitude of the initial pressure and the uncertainty of the initial temperature have strong influences on the magnitude of the uncertainty. Finally, the uncertainty estimation methods studied here provide a reference value for the uncertainty of the reference temperature in an RCM and can be generalized to other similar facilities.

Advances in rapid compression machine studies of low- and intermediate-temperature autoignition phenomena

“…The importance of temperature specification has motivated recent efforts to rigorously quantify experimental uncertainties associated with the evaluation of this parameter [482], and these analyses should be extended to account for, not only random measurement errors, but also systematic errors associated with experimental configurations and operating protocol, e.g., [483,484]. Related to this, as discussed in Section 4, recent efforts have been undertaken to quantify the extent of uncertainties associated with various approaches used to model the experiments, and the level of detail necessary to minimize these for various reacting systems, though open questions still remain.…”

Broadband dual-frequency comb spectroscopy in a rapid compression machine

“…This equation relies on the so-called adiabatic core assumption, namely that the compression occurs on such a short timescale that there is negligible heat transfer from the gas. This assumption will lose validity due to aerodynamic mixing during the compression as well as due to any chemical reactions that may occur [3][4]. After a compression the pistons reach their final position, and the gas has additional time for heat transfer with the RCM walls, further deviating from the idealized process.…”

Autoignition study of tetralin in a rapid compression machine at elevated pressures and low-to-intermediate temperatures

“…The error bars for the ignition delay represent an uncertainty of ±10% of the ignition delay, corresponding to the maximum variation in the various experiments at a given condition. The error bars for the compressed temperature were estimated using the detailed uncertainty analysis method of Weber et al [14]. In this study, we used the Monte Carlo method detailed by Weber et al [14] and considered the uncertainty in measurements of the initial pressure, compressed pressure, initial temperature, ambient temperature, mixing tank volume, injected fuel mass, and proportions of oxygen and nitrogen.…”

“…The error bars for the compressed temperature were estimated using the detailed uncertainty analysis method of Weber et al [14]. In this study, we used the Monte Carlo method detailed by Weber et al [14] and considered the uncertainty in measurements of the initial pressure, compressed pressure, initial temperature, ambient temperature, mixing tank volume, injected fuel mass, and proportions of oxygen and nitrogen. From this analysis, the error in T C was estimated to be less than ±0.5% of the deduced compressed temperature of a given experiment, and hence error bars of ±0.5% are drawn for the inverse temperatures (1000/T C ) in Fig.…”