Modeling Commercial Combustion Systems

Nitrogen oxide emission reduction systems are essential in helping industry meet the environmental requirements mandated by the 1995 Clear Air Act. The industrial participant in this project, Nalco Fuel Tech, is an industry leader in designing these systems. The key technology to the efficient design and operation of their system is accurate computational modeling of the combustion process. We have identified several components of this modeling process that could be significantly enhanced by combining the expertise of Nalco Fuel Tech and ANL. In particular, our goal is to increase computational throughput with parallel processing, take advantage of effective three-dimensional data visualization techniques, and increase flexibility by including the ability to model complex, multiscale geometries.

Nalco Fuel Tech uses computational fluid dynamics and chemical kinetic models to improve the effectiveness of their nitrogen oxide reduction system in commercial combustion systems. There are three key technical issues to be addressed through this collaboration. First, current project turnaround time is not adequate to handle rapidly growing customer demands, and the large number of high-performance computers at Nalco Fuel Tech have not been effectively used in parallel. Second, the ability of engineers to efficiently interpret computation results has been limited by unavailability of advanced visualization methods. Finally, the current system cannot adequately model complex, multiscale geometries, severely limiting its applicability.

The following approach will be used to address these key technical issues:

• Parallelization of the Computational Kernels: We will develop a new, parallel version of the production code used by Nalco Fuel Tech. To take advantage of the multiple processors available, a task-parallel paradigm with a master node controlling several worker processes will be used. That is, each of the chemical process calculations may be treated as a separate problem with communication required between processors every several iterations to transfer global information. To ensure portability, we will use the MPI communication standard. This approach will allow the engineers of Nalco Fuel Tech to make the best use of their computing environment as well as giving them the flexibility to use massively parallel computers such as the IBM SP with little additional effort.

  
  

• Advanced Visualization Procedures: Essential to the understanding of the results of a computational model is the ability of field engineers to effectively visualize the data obtained from the process simulation. We will develop the visualization tools necessary to expand current capabilities to include real-time three-dimensional visualization. These capabilities will include real-time viewing of the development of chemical species distributions, temperature distributions, and dynamic tracers for examining fluid flow. These techniques will use proven commercial packages such as AVS for workstations so that immediate benefits to Nalco Fuel Tech are realized in a general-purpose, cost effective manner. In addition, we will explore the utility of virtual reality environment in the MCS CAVE to immerse the engineers in a dynamic, three-dimensional simulation domain.

  
  

• Unstructured Mesh/Adaptive Mesh Strategies : The problem of modeling complex, multi-scale features can be solved by replacing the regular computational meshes currently used with unstructured meshes. In addition, rapidly changing solutions can be much more accurately modeled with adaptive mesh techniques. Examples of such solution regions include areas of high concentrations of chemical species, modeling high temperature gradients, or examining small, critical features in boiler domains. At Argonne, we have developed the parallel algorithms and software necessary for the construction of these meshes. A final component of this work will be a reformulation of the finite-elements required by the governing equations.