Low-Manifold Biofuel Fast Combustion Simulation

Predictive combustion simulation is an essential but computationally demanding part of contemporary design. One must consider over a hundred chemical species undergoing thousands of reactions to conduct a thorough simulation. Those reactions span a broad range of timescales, ranging from nanoseconds to seconds. This scaling problem, called stiffness, makes the detailed simulation prohibitive, e.g., the direct numerical simulation (DNS) of a high-Reynolds-number jet flame with a reduced mechanism of order twenty species may take millions of CPU hours on a supercomputer [1]. To overcome this, it is imperative to discover a low-dimensional manifold that can efficiently approximate the composition space. A promising development in such modeling derives from our previous research findings indicating that such a manifold can accurately simulate a flame [2]. The algorithm utilized principal components (PCs) parameterizing the composition space [2]. However, a comparable method simulating combustion in other schemes remains lacking, e.g., large eddy simulations (LES) in real-scale conditions. Therefore, there is a critical need to design a method that can efficiently simulate a more comprehensive range of flame models. Through the present work, the proof of the concept within the proposed algorithm is laid out.