TY - JOUR

T1 - Effects of fuel composition variability on high temperature combustion properties: A statistical analysis

AU - Khandavilli, Muralikrishna

AU - Yalamanchi, Kiran K.

AU - Huser, Raphaël

AU - Sarathy, Mani

N1 - KAUST Repository Item: Exported on 2021-02-02
Acknowledged KAUST grant number(s): OSR-2019-CRG7-4077
Acknowledgements: This work was supported by King Abdullah University of Science and Technology (KAUST) Office of Sponsored Research under the award number OSR-2019-CRG7-4077, and the KAUST Clean Fuels Consortium (KCFC) and its member companies.

PY - 2020/12

Y1 - 2020/12

N2 - Real fuels used in combustion devices are complex mixtures of hundreds to thousands of compounds. Understanding the effect of fuel composition variability on important high temperature combustion properties such as pyrolysis product mole fractions and ignition delay times is important for the design of practical devices. In recent works, the effects of variations in fuel compositions on combustion properties were studied using Monte Carlo simulations. It was found that combustion properties follow a Gaussian-like distribution with decreasing variation (2σ/µ) as the fuel palette size increases. The present work attempts to investigate this behavior from the viewpoint of statistical fundamentals. The basis to our investigation is premised on the ability to express combustion properties of blends as the weighted harmonic and arithmetical means of pure component combustion properties in the palette, with the weights being mole fractions. Real fuel compositions were analyzed to justify our selection of a probability distribution for generating random mole fractions. Four different palettes from the literature, comprising 18, 22, 32, and 58 components, respectively, were selected to test our approach. For random compositions of each palette, both mean and standard deviation of the combustion properties from proposed statistical formulae were found to be within fifteen percent of Monte Carlo simulations. Furthermore, we utilize our statistical methodology to further understand the role of fuel composition variability on high temperature combustion properties. The aim of this study is to provide statistical inference to the findings of prior literature, while presenting more validated sets and simple quantitative formulae that serve as an initial screening prior to computationally expensive Monte Carlo simulations.

AB - Real fuels used in combustion devices are complex mixtures of hundreds to thousands of compounds. Understanding the effect of fuel composition variability on important high temperature combustion properties such as pyrolysis product mole fractions and ignition delay times is important for the design of practical devices. In recent works, the effects of variations in fuel compositions on combustion properties were studied using Monte Carlo simulations. It was found that combustion properties follow a Gaussian-like distribution with decreasing variation (2σ/µ) as the fuel palette size increases. The present work attempts to investigate this behavior from the viewpoint of statistical fundamentals. The basis to our investigation is premised on the ability to express combustion properties of blends as the weighted harmonic and arithmetical means of pure component combustion properties in the palette, with the weights being mole fractions. Real fuel compositions were analyzed to justify our selection of a probability distribution for generating random mole fractions. Four different palettes from the literature, comprising 18, 22, 32, and 58 components, respectively, were selected to test our approach. For random compositions of each palette, both mean and standard deviation of the combustion properties from proposed statistical formulae were found to be within fifteen percent of Monte Carlo simulations. Furthermore, we utilize our statistical methodology to further understand the role of fuel composition variability on high temperature combustion properties. The aim of this study is to provide statistical inference to the findings of prior literature, while presenting more validated sets and simple quantitative formulae that serve as an initial screening prior to computationally expensive Monte Carlo simulations.

UR - http://hdl.handle.net/10754/667128

UR - https://linkinghub.elsevier.com/retrieve/pii/S2666352X20300121

U2 - 10.1016/j.jaecs.2020.100012

DO - 10.1016/j.jaecs.2020.100012

M3 - Article

VL - 1-4

SP - 100012

JO - Applications in Energy and Combustion Science

JF - Applications in Energy and Combustion Science

SN - 2666-352X

ER -