Using Bayesian analysis to quantify uncertainties in the H+O<inf>2</inf>→OH+O reaction

Abstract

A stochastic Bayesian approach is applied to investigate the uncertainty in the rate coefficient of H+O2→OH+O (k1) using the latest shock-tube experimental data. We simultaneously calibrate all random variables using a recently developed stochastic simulation algorithm which allows for efficient sampling in the high-dimensional parameter space. We introduce the idea of " irreducible" uncertainty when considering other parameters in the system. Nine stochastic models are constructed depending on the choice of uncertainties, hydrogen concentration, gas temperature, pressure, and rate coefficients of other reactions. The sensitivity analysis of uncertainty in k1 on these uncertainty parameters is performed. It is shown that the introduction of " irreducible" uncertainty does not always increase the uncertainty of k1. In addition, we observe the high sensitivity of uncertainty in k1 to the uncertainty in the measured time-shift. Our results show the strong temperature dependence of the uncertainty in the rate coefficient. © 2013 The Combustion Institute.