A computational approach for preliminary combustor design and gaseous emissions evaluations using a method for sparse kinetics

Computational methods;Numerical methods;Combustion;Numerical computing;Design;Gas turbine combustion;Algorithms

Siegel, Adam D. ; Mavris, Dimitri Aerospace Engineering Jagoda, Jeff Denney, Russell Nastase, Cristian Cummings, William ; Mavris, Dimitri

University:Georgia Institute of Technology

Department:Aerospace Engineering

关键词: Computational methods;    Numerical methods;    Combustion;    Numerical computing;    Design;    Gas turbine combustion;    Algorithms;   

Others  :  https://smartech.gatech.edu/bitstream/1853/56327/1/SIEGEL-DISSERTATION-2016.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

Despite numerous advances in the fields of numerical computing and experimentation, the development of new gas turbine combustors continues to be an extremely costly endeavor. Contemporary aircraft engine component design involves utilizing knowledge from previous architectures (in the form of trends from data, etc.), which the designer uses to develop an improved product. Burners are no different, as gas turbine combustors are conventionally designed using correlations for initial sizing and precursory emissions evaluation. With pollutants becoming an ever increasing concern in air-breathing propulsion systems, the International Civil Aviation Organization strictly regulates the amount of NOx which can be emitted from a given engine. This makes oxides of nitrogen one of the central drivers of modern combustor design, since NOx production considerations cause major restrictions on the properties of the combustor and the conditions under which it can operate. Future engine cycles are constantly pushed to operate at conditions which tend to increase NOx emissions, and so new combustor concepts and technology must therefore be employed to achieve the lower levels called for by regulations. Utilizing correlations for these future designs however is not possible (since these combustors do not yet exist), and creating these correlations involves very expensive experimentation via computational fluid dynamics (CFD) and rig tests. To counter this cost ramification, many techniques oriented at combustor sizing and NOx prediction have been developed in the past. These approaches still fall short of what is needed, since most are either correlation based (or necessarily involve a calibration step), and all are unable to predict NOx emissions quickly enough and accurately enough for a concept which does not yet exist. This work aims to establish a new sizing technique which will give the combustor designer a starting point in the development of advanced concepts. By introducing a novel approach to indirectly couple chemical kinetics and conservation equations, the MoST (Method of Sparse Thermochemistry) algorithm offers a solution to this problem by utilizing sparse kinetics grids in combination with non-reacting CFD solutions. The two codes are tied together and controlled by an overarching algorithm that indirectly couples the flow field and kinetics in such a way that the flow field is still influenced by the chemical process happening within it. This eliminates the need to solve coupled nonlinear PDEs, and instead solves them as a smaller ODE (or algebraic) system. Research is presented which aids to understand the elements which must be captured to turn the flow field into a chemical kinetics network (CKN), exactly how to orient the reactors to best mock the physical processes within the flow field, and how to converge on a flow field solution while taking into account its response to chemical reactions. Grids with varying cell densities and chemical networks with varying numbers of reactors are run using this method and presented as proof of concept. Good agreement is shown between this approach and emissions data from reacting flow CFD.