A graph neural network based chemical mechanism reduction method for combustion applications

Manuru Nithin Padiyar; Priyabrat Dash; Konduri Aditya

cs.LGMar 20, 2026

A graph neural network based chemical mechanism reduction method for combustion applications

Authors: Manuru Nithin Padiyar, Priyabrat Dash, Konduri Aditya

Organizations: Department of Computational and Data Sciences, Indian Institute of Science, Bengaluru, 560012, KA, India

Abstract

Direct numerical simulations of turbulent reacting flows involving millions of grid points and detailed chemical mechanisms with hundreds of species and thousands of reactions are computationally prohibitive. To address this challenge, we present two data-driven chemical mechanism reduction formulations based on graph neural networks (GNNs) with message-passing transformer layers that learn nonlinear dependencies among species and reactions. The first formulation, GNN-SM, employs a pre-trained surrogate model to guide reduction across a broad range of reactor conditions. The second formulation, GNN-AE, uses an autoencoder formulation to obtain highly compact mechanisms that remain accurate within the thermochemical regimes used during training. The approaches are demonstrated on detailed mechanisms for methane (53 species, 325 reactions), ethylene (96 species, 1054 reactions), and iso-octane (1034 species, 8453 reactions). GNN-SM achieves reductions comparable to the established graph-based method DRGEP while maintaining accuracy across a wide range of thermochemical states. In contrast, GNN-AE achieves up to 95% reduction in species and reactions and outperforms DRGEP within its target conditions. Overall, the proposed framework provides an automated, machine-learning-based pathway for chemical mechanism reduction that can complement traditional expert-guided analytical approaches.

Figures & tables

Fig. 1: A flowchart representation of the (a) GNN-SM implementation, (b) GNN-AE implementation along with graphs of the (c) input and (d) output mechanisms.

GNN-SM
Mechanism	$\mathbf{n_{s}/n_{r}}$	Avg. IDT Error (%)
GRI-Mech 3.0	53 / 325	—
DRGEP	23 / 123	1.53
Pass 1	35 / 189	—
Pass 2	29 / 177	—
Pass 3	27 / 157	1.45

Table 1: Comparison of chemical mechanisms for methane oxidation (1000K to 2000K).

Panel of Figure 3 — Fig. 2: Temporal evolution of temperature and heat release rate at initial temperatures of (a) 1500K and (b) 1000K for methane oxidation.

Fig. 3: Temporal evolution of scaled species mole fractions at initial temperature of 1500K for methane oxidation.

Fig. 4: Validation of ignition delay times for the reduced methane mechanisms over a range of initial temperatures and pressures.

GNN-SM
Mechanism	$\mathbf{n_{s}/n_{r}}$	Avg. IDT Error (%)
FFCM2	96 / 1054	—
DRGEP	38 / 288	11.56
Pass 1	77/814	—
Pass 3	44/397	—
Pass 5	37/302	—

Table 2: Comparison of mechanisms for ethylene oxidation (1000K to 2000K).

Panel of Figure 7 — Fig. 5: Temporal evolution of temperature and heat release rate at initial temperatures of (a) 1500K and (b) 1200K for ethylene oxidation.

Fig. 6: Temporal evolution of species mole fractions at initial temperature of 1500K for ethylene oxidation.

Fig. 7: Validation of ignition delay times for reduced ethylene mechanisms across conditions. The dotted cyan and green lines indicate the regions around which the two GNN-AE-G mechanisms were reduced.

Fig. 8: Validation of ignition delay times for reduced iso-octane mechanisms across conditions.

GNN-SM
Mechanism	$\mathbf{n_{s}/n_{r}}$	Avg. IDT Error (%)
Curran et al.	1034 / 8453	—
DRGEP	63 / 637	8.81
Pass 7	247 / 2817	9.19
GNN-AE
GNN-AE-G	55 / 553	16.69

Table 3: Comparison of mechanisms for iso-octane oxidation (1000K to 2000K).

Explore similar work

Machine Learning-Driven Chemical Reactor Network Modeling of the Sandia-D Flame
Jun 1, 2026Nicolas J. Tricard, Benjamin C. Koenig, Sili DengSurrogate Modeling Computational Fluid Dynamics
Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
Entropy-Constrained Machine Learning with Residual Data Augmentation for Modeling Chemical Kinetics
Jul 10, 2026Okezzi Ukorigho, Opeoluwa OwoyeleSurrogate Modeling PDE Surrogate Modeling
Department of Mechanical and Industrial Engineering Louisiana State University, Baton Rouge, LA 70810, USA
Modeling subgrid scale production rates on complex meshes using graph neural networks
Mar 20, 2026Priyabrat Dash, Mathis Bode, Konduri AdityaComputational Fluid Dynamics Graph Neural Networks
Department of Computational and Data Sciences, Indian Institute of Science, Bengaluru, 560012, KA, India · Jülich Supercomputing Centre, Forschungszentrum Jülich GmbH, Jülich, 52425, NRW, Germany

Jun 1, 2026J/K move · Enter open · S save

Machine Learning-Driven Chemical Reactor Network Modeling of the Sandia-D Flame

Nicolas J. Tricard, Benjamin C. Koenig, Sili Deng

Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA

Surrogate Modeling Computational Fluid Dynamics

Open paper AskPDF arXiv

Turbulent combustion simulations are crucial for many scientific and engineering systems. However, the high cost to fully resolve the complex multiscale and multiphysics behavior makes direct simulation typically infeasible. The equivalent reactor network (ERN) approach attempts to improve computational efficiency by replacing a multidimensional turbulent simulation with a series of much cheaper 0-D and 1-D chemical reactors, providing a surrogate model that retains detailed chemistry at the cost of simplified flow physics. However, their development remains a challenge, often requiring either expert analysis, or automated approaches that sacrifice accuracy. In this work, we develop an automated machine-learning-assisted framework for constructing ERNs of the Sandia-D turbulent methane/air flame. Principal component analysis is first used to reduce high-dimensional thermochemical computational fluid dynamics (CFD) data to a low-dimensional latent space, where k-means clustering identifies physically interpretable flame regions used to initialize a reactor-network graph. This initialization is then refined using finite-difference gradient descent wrapped around non-differentiable Cantera reactor simulations. Across 30 RANS simulations spanning a range of pilot temperatures and inlet methane compositions, the optimized 7-reactor ERN achieves a maximum-temperature

R^2

score of 0.7945 while preserving a

\sim6000\times

speedup over the CFD solver. Outlet CO prediction remains more challenging, with a final

R^2

score of

-0.4183

, but improves substantially from the unoptimized clustering initialization. These results show that unsupervised thermochemical feature extraction can provide effective physics-informed initializations for ERN construction, while gradient-based refinement can significantly improve predictive accuracy without manual reactor-network design.

Abstract

Explore similar work

Machine Learning-Driven Chemical Reactor Network Modeling of the Sandia-D Flame

Entropy-Constrained Machine Learning with Residual Data Augmentation for Modeling Chemical Kinetics

Modeling subgrid scale production rates on complex meshes using graph neural networks