CFD & AMP Center
Department of Mechanical & Aerospace Engineering
West Virginia University

Computational Strategies for Faster Combustion Simulations with Detailed Chemistry

Computational Strategies for Faster Combustion Simulations with Detailed Chemistry
National Energy Technology Laboratory (NETL)
Dr. Ismail Celik
Jose Escobar

Executive Summary

Pollutants released to the atmosphere from combustion units represent a risk for human health and the environment. In order to reduce the negative impact of combustion units the government has imposed controls for the regulated pollutants. Environmental controls have lead research to find innovative strategies to reduce pollution. Nowadays numerical simulations have become an important tool in research and design of combustion systems. The main advantages of numerical simulations are: relatively low cost of development, fast (compared to build and perform experiments in combustion rigs), and relatively ease with which parametric studies can be conducted. Despite all the advantages simulation of combustion phenomena is far from accurate and detailed predictions. The difficulty of combustion simulation is due to the several physical processes that occur simultaneously such as turbulence, chemical reactions, and heat transfer.

Log time integration method (LTIM)

An explicit, second order integration method is proposed for integration of stiff equations. This novel method was developed by transforming the time variable into the logarithmic space. Stiff equations are characterized by large slopes (∆φ/∆t, being φ any variable). The transformation of the time variable into the logarithmic space could decrease the slope value thus making the integration more treatable. Being an explicit scheme, this method will allow easy implementation in problems with large number of species without the requirement of inverting any Jacobian.

Composition predictions calculated in closed reactors using the LTIM were obtained faster than using traditional integration methods (Euler and 5th order Runge-Kutta Method). Predicted compositions are within reasonable accuracy.

Chemical Reactor Network (CRN)

CFD simulations of combustors using detailed mechanisms for practical problems are virtually impossible to perform due the size of the chemical model (e.g. Methane-Air combustion GRI 3.0 mechanism has 53 species and 325 reactions). For this reason simplified chemical mechanisms are employed but the assumptions used in the development of these mechanisms affect the accuracy of the predictions, especially for the chemical species whose concentration is in the parts per million (ppm) level.

CRN calculations required the input of CFD in order to build a network of simplified reactor that represents the combustor. These CFD simulations are obtained using relatively inexpensive models (e.g. RANS, one step global mechanism). A Computer program wad created in order to create a reactor network using the results form the CFD predictions. Detailed chemistry simulations were performed using the GRI 3.0 detailed mechanism.

Different reactor network configurations were simulated and the results compared against the CFD and experimental measurements. CRN calculations were obtained faster than the CFD calculations and they also provide extra information about the composition of the gas inside the combustor.