SciNet Users Group Meeting University of Toronto Toronto, Ontario February 9, 2011 Computational Combustion: Toward the use of Sustainable and Alternative Fuels Dr. Seth B. Dworkin Mechanical and Industrial Engineering University of Toronto Toronto, Ontario Collaborators and Acknowledgements Prof. Murray Thomson Dr. Qingan Zhang Dr. Victor Chernov Meghdad Saffaripour Armin Veshkini Parham Zabeti Nick Eaves SciNet (Present Company) The Demand for Energy • 75% of power generation and transportation energy comes from combustion 3 Global Climate Change • CO2 and soot (black carbon particulate) are the #1 and #2 contributors to global climate change • Addressing CO2 • Replace fossil fuels with biofuels that take in CO2 as they grow • Addressing Soot • Fundamental soot studies • Apply this knowledge to engine and power plant technology 4 Research Streams • Use fundamental thermofluid sciences and computation 5 Research Methodology • Modelling/experiment understanding 1. 2. fundamental 3. 4. • Models help us understand properties and trends that apply to practical combustion devices 6 Research Topics 1. Bio-Jet Fuel 2. Soot formation in flames 3. Premixing fuel studies 4. High pressure combustion 7 Chemically-Reacting Flow Models • Conservation equations for mass, momentum, energy, and species conservation are simultaneously solved in a computational domain. • A library of chemical reactions is used, which includes the prediction of particle precursor molecules • Conservation equations for a statistical representation of particle size and shape are solved • The particle equations are coupled to the gas phase: • Inception rates from collision theory • Oxidation via O2 and OH • Growth via surface chemistry • Enthalpy changes 8 1. Bio-Jet Fuel Background • Airplanes need liquid fuel • Biomass To Liquid (BTL) has been proposed as an alternative jet fuel • Considerations: – – – – Production & Distribution Economic Viability Airplane corrosion and degradation Emissions • Goal is to compare BTL combustion to standard jet fuel via simulation 9 Computational Results • After 3 months of compute time on 192 4.7-GHz CPUs (43 CPU-years!) S1: n-decane S2: n-propylbenzene S3: n-propylcyclohexane 10 Validation (Species) • Comparisons are made for major species along the centreline. • Agreement between model and experiment is excellent. 11 Validation (Soot) • Comparisons are made for soot volume fraction along the centreline and along radial cuts at three axial heights. • Data is well reproduced at the lower axial locations. • Soot is again underpredicted along the centerline • Current efforts are on computing a BTL flame and on studying soot formation 12 2. Soot Formation in Flames 5. There are ongoing processes of coagulation, fragmentation, oxidation, surface condensation, and surface reaction 4. Spheres coagulate into soot molecules 3. Structures grow to spherules via surface reactions 2. Aromatics combine to form 3D structures 1. Fuel gets decomposed into small molecules which form aromatics 13 Problem Statement • Soot formation in flames remains poorly understood – most models are semi-empirical • Predicting soot inception is done empirically with approximations Goals: • Develop and validate a model for particle inception • Build the new model into flame computations • Model a well-understood flame – Use a steady ethylene-fueled system – Wealth of experimental data • Perform engineering analysis on the results 14 Chemical Model Validation • With collaborators at DLR – new mechanism for PAH growth with over 700 reactions 15 Chemical Model Implementation • The new chemical model is coupled to a full CFD and soot model and computations are performed in parallel. 16 Model Validation 17 Data Interpretation • The data can be processed to track a fluid parcel • This technique can relate its exposure to temperature, concentrations, etc. 18 Data Interpretation 2000 Temperature (K) 1600 Appel et al. 7.0E-07 Pyrene Concentration 1800 8.0E-07 New Mech. 1400 1200 1000 800 600 400 New Mech. 5.0E-07 4.0E-07 3.0E-07 2.0E-07 1.0E-07 200 0 0.00 6.0E-07 Appel et al. 0.05 0.10 0.15 0.0E+00 0.00 time (s) 0.05 0.10 0.15 time (s) • Lagrangian temperature historesis are similar • Vast differences in aromatic exposure S. B. Dworkin, Q. Zhang, M. J. Thomson, N. A. Slavinskaya, U. Reidel, Accepted to Combust. Flame (2011) in press. 19 Summary • Using parallel computation we can study combustion systems • The models contains few assumptions – Solve the full set of conservation equations for fluid mechanics and thermodynamics – Consider detailed chemical reaction libraries – Contain a particle formation and dynamics model • We can use experiment and computation to study and compare proposed alternative fuels • We can predict the formation tendencies of harmful pollutants • We can use data processing to increase our fundamental understanding of flames and soot dynamics 20 Questions?