화학공학소재연구정보센터
Combustion and Flame, Vol.180, 239-249, 2017
The role of correlations in uncertainty quantification of transportation relevant fuel models
Large reaction mechanisms are often used to describe the combustion behavior of transportation relevant fuels like gasoline, where these are typically represented by surrogate blends, e.g., n-heptane/isooctane/toluene. We describe efforts to quantify the uncertainty in the predictions of such mechanisms at realistic engine conditions, seeking to better understand the robustness of the model as well as the important reaction pathways and their impacts on combustion behavior. In this work, we examine the importance of taking into account correlations among reactions that utilize the same rate rules and those with multiple product channels on forward propagation of uncertainty by Monte Carlo simulations. Automated means are developed to generate the uncertainty factor assignment for a detailed chemical kinetic mechanism, by first uniquely identifying each reacting species, then sorting each of the reactions based on the rate rule utilized. Simulation results reveal that in the low temperature combustion regime for iso-octane, the majority of the uncertainty in the model predictions can be attributed to low temperature reactions of the fuel sub-mechanism. The foundational, or small-molecule chemistry (C-0-C-4) only contributes significantly to uncertainties in the predictions at the highest temperatures (Tc=900K). Accounting for correlations between important reactions is shown to produce non-negligible differences in the estimates of uncertainty. Including correlations among reactions that use the same rate rules increases uncertainty in the model predictions, while accounting for correlations among reactions with multiple branches decreases uncertainty in some cases. Significant non-linear response is observed in the model predictions depending on how the probability distributions of the uncertain rate constants are defined. It is concluded that care must be exercised in defining these probability distributions in order to reduce bias, and physically unrealistic estimates in the forward propagation of uncertainty for a range of UQ activities. (C) 2016 The Combustion Institute. Published by Elsevier Inc. All rights reserved.