A quadrature closure for the reaction-source term in conditional-moment closure
Introduction
As the modeling of mixing and reactions in non-premixed turbulent combustion steadily becomes more advanced, the diversity of physical phenomena that can be described accurately is naturally broadened. The advancement from mixing-limited chemistry to equilibrium chemistry allowed the ability to model the chemical thermodynamics. Steady laminar flamelets offered the ability to model the overlap of mixing and reaction timescales, and consequently flame extinction could be modeled in a qualitative way. Unsteady flamelets, conditional-moment closure (CMC) and the Lagrangian probability-density-function (PDF) approach have offered, in different ways, the ability to model the overlap of multiple reaction timescales with the mixing timescale(s). Yet, the accurate prediction of homogeneous extinction and re-ignition are phenomena that are challenging for even the most advanced combustion models. Stated simply, extinction is the consequence of strong mixing overbalancing the chemical reactions. Physically, homogeneous re-ignition is the consequence of un-extinguished reaction zones interacting with extinguished zones. Capturing this interaction is required to not only describe re-ignition, but also to accurately describe the dynamics of extinction. As Bilger et al. [1] stated, “In flames with significant local extinction and re-ignition it has been found that [the] first-order closure is not sufficiently accurate.” It is the opinion of the authors, of the current paper, that higher-order CMC offers the greatest potential to accurately model the dynamics of extinction and re-ignition for a reasonable computational cost. However, the higher-order CMC methods do require refinement. The purpose of this paper is to propose an adjustment to one aspect of the higher-order CMC model. More specifically, we suggest that the use of the Taylor series as a closure for the conditional reaction-source term is unsuitable, and that the source-term integral can be more accurately approximated by an appropriate Gaussian-quadrature formula.
Section snippets
Theory
The homogeneous, one-point, constant-density joint PDF, fY,ξ, of mixture fraction, ξ, and reaction-progress variable, Y, evolves in time, t, according to [2], [3]where y and ζ are the sample-space variables for the reaction-progress variable and the mixture fraction, respectively, SY is the chemical-reaction source term for the reaction-progress variable, and the rates of dissipation by diffusion with
Direct numerical simulation data
In order to analyze the validity of CMC models to describe such phenomena, we chose to a priori test the models using the direct numerical simulations (DNS) of Sripakagorn et al. [4]. The DNS included the transport of two scalars: a passive mixture fraction and a reaction-progress variable. The two scalars were injected into isotropic decaying turbulence. The mixture fraction was initialized such that it was nearly segregated with a mean near 〈ξ〉 = 0.5, and the reaction-progress variable was
Modeling
In the process of analyzing the “standard” higher-order CMC models, we were surprised by how large the error was in the pivotal conditional reaction-source term. In this paper, we discuss the error of the closure for this term as presented by Klimenko and Bilger [2]. Additionally, we present a new closure based on Gaussian-quadrature integration.
Taylor-series approximation
Figure 3 shows the second- and third-order Taylor-series approximations to the reaction-source term using the DNS at a time near extinction. One can see that for a complex chemical-source term, the lower-order Taylor-series approximations, although accurate in the region near the conditional mean, do not approximate the source term well across the realizable region. This weakness is accentuated by the fact that the corresponding conditional PDF, shown in Fig. 4, is largest in regions where the
Conclusions
Based on the a priori analysis using DNS data for a relatively complex kinetic expression, we can conclude that the quadrature approximation for the conditional chemical-source term offers a powerful approach for modeling the reaction-source term in higher-order CMC. In practice, use of the quadrature approximation will require the numerical simulation of transport equations for the conditional moments 〈Yα∣ζ〉, where the maximum value of α depends on the order of the approximation. For the case
Acknowledgments
This research was partially supported by Grants (CTS-0336435, CTS-0403864) from the US National Science Foundation. The authors thank Drs. Sripakagorn, Kosály and Riley for providing us access to their DNS data.
References (9)
- et al.
Energy Combust. Sci.
(1999) - et al.
Combust. Flame
(2004) - et al.
Proc. Combust. Inst.
(2004) Computational Models for Turbulent Reacting Flows
(2003)
Cited by (5)
A quadrature-based conditional moment closure for mixing-sensitive reactions
2020, Chemical Engineering ScienceCitation Excerpt :Here, our goal is to reduce this cost by tracking a small set of composition moments, which are used to find the conditional moments via Gaussian quadrature. One of the first study dealing with the application of quadrature-based moment methods (QBMM) to CMC (Smith et al., 2007) focused on the quadrature approximation to the conditional reaction-source term. Earlier, Fox (2003) proposed an extension of the multi-environment PDF (ME-PDF) method, called multi-environment conditional PDF model (MEC-PDF) by using conditional scalar means that was subsequently solved by the direct quadrature method of moments (Fox and Raman, 2004).
Effect of the conditional scalar dissipation rate in the conditional moment closure
2020, Physics of FluidsA Universal Concept Based on Cellular Neural Networks for Ultrafast and Flexible Solving of Differential Equations
2015, IEEE Transactions on Neural Networks and Learning SystemsQuadrature-based moment methods for polydisperse multiphase flows
2014, CISM International Centre for Mechanical Sciences, Courses and Lectures