Related Links:

Premixed Flames
Diffusion Flames

Personnel
Publications
Software

CCSE Home
CCSE AMR
CCSE Research
SciDAC APDEC

Burners are used in industry for a wide range of applications including water heaters, power generators, and HVAC systems. Consumer applications include gas-fired home water heaters, stoves, ovens and clothes dryers. In spite of its technological importance, our knowledge of combustion process is surprisingly incomplete. Theoretical combustion science provides us with a foundation for basic flame analysis, but is unable to directly address the complexity of realistic flames. Laboratory measurements are difficult to interpret and are often limited in the level of detail they can provide about a flame. Computation has the potential for closing the gap between theory and experiment and enable dramatic progress in combustion science. However, simulation of practical-scale combustion devices is an immense undertaking. The problem is inherently multi-scale both in time and space, the fuel is often turbulent, and the combustion process may involve hundreds of species and thousands of chemical reactions.

Traditional direct numerical simulation (DNS) approaches based on explicit numerical methods for the compressible flow equations on uniform grids require very fine spatial grids to resolve the local flame structure. In addition, they require small time steps to resolve the acoustic and chemical time scales inherent in the model. DNS is normally reserved in combustion applications for small idealized problems geared at the fundamental nature of turbulence/chemistry interactions. For engineering design applications on the other hand computional models have been developed to approximate underresolved physics. But these models are incomplete, do not have general applicability, and certainly provide no means of exploring fundamental fluid/chemistry interactions.

The research approach taken by CCSE has explicitly targeted both the temporal and spatial multiscale aspects of combustion modeling. First, a low Mach number formulation is used instead of the traditional compressible equations, thereby eliminating the acoustic time step restriction while fully maintaining the compressibility effects due to heat release. Second, adaptive mesh refinement (AMR) is used to focus computational resources in regions of interest without wasting resources in regions requiring less resolution. Third, robust integration methods are employed to allow reasonable solution behavior with a minimum of computational resolution. The combination of AMR and a robust low Mach number implementation for reacting flows has reduced the computational requirements of simulating laboratory-scale low-speed methane combustion by a factor of 10,000 relative to traditional approaches (compressible equations solved on a uniform grid). With these advanced methods, we can simulate time-dependent, laboratory-scale, turbulent premixed combustion experiments in three dimensions, while including detailed chemical mechanisms to describe the combustion process and the differential diffusion of the various chemical species. Our recent calculations are some of the most detailed and complete turbulent methane combustion simulations to date. Current CCSE research projects that utilize the adaptive low Mach number algorithms include our work in turbulent premixed flames, and nuclear flames found in Type 1a supernovae. Earlier CCSE research in this area includes two and three dimensional studies of premixed and nonpremixed flames.

The adaptive projection methods that form the basis of our low Mach number simulation capability are derived from more than ten years of algorithm research at CCSE. The work began with adaptive methods for Godunov-based integration of compressible flows over hierarchical structured grids. An adaptive projection scheme for incompressible flows was developed from much of the same software technology. The algorithms for incompressible flow extended quite naturally to the case of low Mach number flows. Over more recent years, the general low Mach solution scheme was tailored to the special cases of laboratory combustion and nuclear flames. All this simulation capability was built on CCSE's library of data structures for hierarchical structured grids, BoxLib. BoxLib has evolved with the changing needs of the group, and in its present form is a C++ class library for distributed memory parallel hardware. BoxLib is available for download along with some of our adaptive integration algorithms for compressible and incompressible flows. The low Mach combustion and nuclear flames applications are not openly available for download primarily since they contain source code we cannot legally distribute (such as the CHEMKIN-III routines). We are always looking for new applications and collaborators that can make use of our software, and encourage interested researchers to contact the CCSE group leader, John Bell.