A Computational Facility for Reacting Flow Science (CFRFS)

H.N. Najm¹, J. Ray¹, J.C. Lee¹, S. Lefantzi¹, C.A. Kennedy¹, P.P. Pébay¹, M. Valorani², D.A. Goussis³, W. Kollmann⁴, M. Frenklach⁵

¹Sandia National Laboratories, ²University of Rome-Italy, ³ICEHT-Greece, ⁴University of California-Davis, ⁵University of California-Berkeley

Summary

Our focus is the development and demonstration of a modular software toolkit that makes use of the Common Component Architecture for assembling scalable massively parallel adaptive mesh refinement reacting flow computation and chemical analysis codes. This development is crucial for enabling efficient computational studies of reacting flow on terascale hardware, allowing the extraction of physical insights from resulting databases, and enabling automatic chemical model reduction and adaptive chemistry computations. We will use these tools to address a range of chemical/reacting flow questions of interest and of practical importance, improving our understanding of turbulence-chemistry interactions in combustion.

1.  Introduction

This work is motivated by the need for new approaches to the development of reacting flow codes. This need arises because of the challenges inherent in the development and maintenance of parallel reacting flow codes, with adaptive mesh refinement (AMR) and dynamic load balancing, in the context of massively parallel terascale computing. The complexity of large-scale reacting flow databases also requires the development of new analysis techniques for extraction of physical insights, chemical model reduction, and scientific discovery.

Our goal is to develop a Computational Facility for Reacting Flow Science (CFRFS) that has at its core a new approach for assembling massively parallel reacting flow computation and analysis codes using a modular software toolkit. We use the common component architecture (CCA) software framework. We are developing flexible, & reusable “components” of this toolkit for enabling the assembly of different reacting flow codes. CCA provides standardized component interfaces and connectivity. It alters the conventional code development paradigm into one of specialized modular component development. We also use computational singular perturbation (CSP) theory as a basis for data analysis components development, to allow automated chemical data analysis and model reduction, and to enable an adaptive chemistry computational strategy.

2.  Technical Highlights

We have made significant progress toward both the computational and CSP analysis component development and demonstration.

On the computational side, we've built upon our earlier development of components ( mesh, thermochemistry, etc) for reaction-diffusion AMR computations using the parallel AMR library GrACE. We have implemented detailed mixture-averaged transport, operator-split time integration and AMR-recursive stabilized explicit Runge-Kutta-Chebyshev diffusion integration. We implemented high-order spatial discretization, interpolation, and filtering AMR components. This is crucial for ensuring shallow adaptive mesh hierarchies and facilitating dynamic load balancing with structured AMR (SAMR). We demonstrated novel high-order SAMR reaction-diffusion computations of hydrogen-air ignition in 2D, investigated order requirements for interpolation and filtering operations, and identified realizable gains in performance over low order SAMR. We are in the process of integrating an elliptic Poisson equation solver (provided by the SciDAC APDEC ISIC) into our component toolkit. This will enable the assembly of momentum solution components for low Mach number reacting flow, which is work in progress. We have also assessed the utility of existing SAMR dynamic load balancing tools and have identified limitations having to do with the spatial variability of the chemistry time integration CPU load. We have identified potential remedies for this problem, which are the subject of ongoing work. We have also continued to provide support for the SciDAC TSTC project for implementation of a turbulent reacting flow DNS code with GrACE and its ultimate componentization under CCA.

On the chemical analysis/reduction front, we developed a more robust CSP analysis formulation using Singular Value Decomposition. We also implemented the full suite of CSP analysis and model reduction tools into a general CCA-based toolkit. We used this toolkit to analyze a set of reacting flow results, investigating stoichiometric premixed methane-air flame-vortex interactions for a range of values of vortex strength, using the GRI3.0 chemical mechanism. We investigated the dependence of the CSP analysis results on the flow time scales. We found the results to be reasonably robust against flow disturbances in this flame, particularly in post-flame gases. This is encouraging for our adaptive chemistry effort involving PRISM tabulation. Given the robustness of the CSP analysis results, we now plan to tabulate exhausted CSP vectors, rather than the integrated chemical solution. Using this tabulation, we can potentially use explicit time integration of the tabulated chemistry. Preliminary tests have shown that this is feasible, and is potentially highly efficient. Current work is focused on implementing this construction using PRISM tabulation of the exhausted CSP vectors, thereby providing a fast and accurate adaptive chemistry capability for reacting flow computations. With the completion of CSP/PRISM adaptive chemistry tabulation, we will pursue testing against computations with detailed chemical models to evaluate accuracy and efficiency of the tabulation strategy. We will also work with the SciDAC CMCS and SDM ISICs toward an integrated reacting flow analysis toolkit.

3. Publications

Over the past two years, this project has supported 6 archival articles (5 submitted, 1 published), 2 articles in published books, 4 articles in refereed conference proceedings (2 submitted, 2 published), and various conference presentations.

For further information on this subject contact:
Dr. Habib N. Najm
7011 East Ave, MS 9051
Sandia National Laboratories
Livermore, CA 94550
Phone: (925) 294-2054
hnnajm@sandia.gov
http://www.ca.sandia.gov/cfrfs

back to project page