Prediction of emissions from combustion systems using 0D and 1D reacting flow models Prediction of emissions from combustion sys- tems using 0D and 1D reacting flow models Chemical Reactor Network modeling B. Rosati Technische Universiteit Delft Prediction of emissions from combustion systems using 0D and 1D reacting flow models Chemical Reactor Network modeling by B. Rosati to obtain the degree of Master of Science at the Delft University of Technology, to be defended publicly on Monday December 14, 2015 at 10:00 AM. Student number: 4259343 Project duration: November 15, 2014 – December 14, 2015 Thesis committee: Prof. dr. D.J.E.M. Roekaerts, TU Delft, supervisor Dr. A. Gangoli Rao, TU Delft, co-supervisor Dr. A. Bhat, Honeywell, co-supervisor Dr. H. B. Eral, TU Delft Faculty of Mechanical, Maritime and Materials Engineering (3mE). Process & Energy Department. Delft University of Technology An electronic version of this thesis is available at http://repository.tudelft.nl/. Abstract Emission prediction is a complex problem involving the coupling between the flow field and chemistry. Most of the time CFD is the preferred modeling approach, yielding predictions with varying degrees of accuracy. But because of a high computational cost, CFD investigations are often limited to the use of reduced chemical mechanisms. In this work the specific features of chemical reactor networks are exploited to build a fast and reliable emission estimator. The main advantage of this modeling approach is a much lower computational cost than CFD, hence offering the potential for relatively fast predictions while allowing the use of detailed chemistry. This methodology has been applied to three different combustion systems, with mixed results. It may not be the most suitable modeling technique to obtain emissions from a lifted jet flame, but a successful estimator has been designed for flameless furnaces. It is based solely on analytical sub– models, giving it the potential to predict the emissions from any type of flameless furnace installation. For three different experimental setups, the correct trends were reproduced as well as the right order of magnitude for 푁푂 and 퐶푂 emissions, if not within experimental measurements uncertainty. Finally the emissions from a lean-premixed gas turbine combustor burning cryogenic fuel have been successfully modeled and this investigation has brought out the major sensitivities of this system. Lastly, despite some promising results, several developments have been suggested to improve the accuracy and stability of the flameless furnace estimator. The combustor estimator, for one, can be used as basis to investigate the behavior of the more comprehensive hybrid combustion system it has originally been designed for: the dual combustion chamber of the AHEAD hybrid engine (Advanced Hybrid Engines for Aircraft Development). Keywords: Emissions, Reactor networks, Flameless combustion, Strong–Jet / Weak–Jet, Lean- premixed combustion. iii Contents List of Figures ix List of Tables xiii 1 Introduction 1 1.1 Energy & Environment . 1 1.2 Design & Modeling . 2 1.3 Thesis objectives . 2 1.4 Thesis outline . 3 2 Literature : scientific background 5 2.1 Emissions . 5 2.1.1 Carbon monoxide (퐶푂) . 5 2.1.2 Nitrogen oxides (푁푂) . 6 2.1.3 Quantification of emissions . 8 2.2 Chemical mechanisms . 9 2.2.1 GRI-Mech 3.0. 10 2.2.2 퐶2_푁푂 Mechanism . 10 2.3 Flameless combustion . 10 2.3.1 One technology, different names . 10 2.3.2 Theory and characteristics . 11 2.3.3 Desirable consequences of flameless combustion . 14 2.3.4 Operation and stability domain. 14 2.3.5 Burner and furnace technology . 15 2.4 Jet modeling . 17 2.4.1 Free jet structure . 17 2.4.2 Entrainment models . 17 3 Emission modeling of a lifted jet flame 21 3.1 Introduction . 21 3.2 Additional literature . 23 3.2.1 Turbulent jet flame emission modeling : a 2–reactor model . 23 3.2.2 Diffusion flame stability. 24 3.3 Reactor network . 26 3.3.1 Architecture . 26 3.3.2 Model inputs . 28 3.3.3 Simulation cases . 30 3.3.4 Calibration of the reactor network . 30 3.3.5 Conclusion . 30 3.4 Results . 32 3.5 Conclusion . 34 4 Emission modeling of flameless furnaces 35 4.1 Introduction . 35 4.2 Additional literature . 36 4.3 The ”Bending Model” . 37 4.3.1 Setup and equations . 37 4.3.2 The 3 jets problem . 38 4.3.3 Implementation of the dual–hole configuration . 39 4.3.4 Validation against the reported performance . 40 4.3.5 Validation against the SJ/WJ model predictions . 41 v vi Contents 4.3.6 Final validation: the IFRF furnace . 45 4.3.7 Final validation: the University of Mons furnace . 45 4.3.8 Final validation: the University of Adelaïde furnace . 45 4.3.9 Conclusion . 48 4.4 Jet entrainment modeling in a SJ/WJ configuration. 49 4.4.1 Mutual ”shielding” of the jets . 49 4.4.2 Shielding factor . 49 4.4.3 Conclusion . 51 4.5 CRN modeling of a flameless furnace . 52 4.5.1 General architecture . 52 4.5.2 Role of the analytical submodels . 53 4.5.3 Parameters of particular interest. 54 4.5.4 Conclusion: model parameters classification . 55 4.6 Prediction of emissions from the IFRF furnace . 56 4.6.1 Inputs . 56 4.6.2 Simulation cases . 57 4.6.3 Base case results. 57 4.6.4 Parametric study . 62 4.6.5 Conclusion . 63 4.7 Prediction of emissions from the Mons furnace . 64 4.7.1 Inputs . 64 4.7.2 Simulation cases . 66 4.7.3 Results . 66 4.7.4 Conclusion . 68 4.8 Prediction of emissions from the TU Delft furnace . 68 4.8.1 Furnace characteristics . 68 4.8.2 Necessary approximations . 69 4.8.3 Inputs . 70 4.8.4 Simulation cases . 71 4.8.5 Results . 71 4.9 General conclusion . 73 5 Emission modeling of a jet engine lean–premixed combustor 75 5.1 Introduction . 75 5.2 The cryo–combustor of the AHEAD engine . 76 5.3 Network architecture . 80 5.4 Model inputs . 81 5.5 Calibration . 83 5.5.1 Heat loss calibration . 83 5.5.2 Recirculation intensity calibration . 83 5.6 Results: comparison with experimental data . 84 5.7 Parametric study . 89 5.7.1 Influence of the overall residence time: ̇푚 . ..