TL;DR
This paper introduces a novel "dual-table" approach for the Flamelet-Generated Manifold (FGM) combustion model to significantly speed up and reduce the memory footprint of simulating dual-fuel internal combustion engines. By separating premixed and non-premixed combustion regimes into two distinct lookup tables, the method achieves comparable accuracy to single-table approaches but with substantial computational gains, making it more practical for industrial applications.
What Problem Does This Paper Address?
Simulating the complex combustion processes in dual-fuel internal combustion (IC) engines is crucial for developing cleaner and more efficient engines, especially for heavy-duty applications where electrification is challenging. These engines burn a mixture of a gaseous fuel (like natural gas) and a pilot diesel fuel, which ignites the mixture. This dual-fuel combustion involves both premixed (gaseous fuel) and non-premixed (diesel spray) combustion regimes.
Traditional detailed chemical kinetics simulations are computationally prohibitive due to the vast number of species and reactions involved. Flamelet-Generated Manifold (FGM) tabulation offers a way to pre-compute complex chemistry into lookup tables, drastically reducing simulation time. However, a major challenge with FGM for dual-fuel engines is that representing both premixed and non-premixed regimes within a single FGM lookup table leads to an explosion in table size and generation time, making it impractical for industrial use. This paper aims to overcome this by proposing a more efficient FGM approach.
Methods
The core innovation is a dual-table FGM approach. Instead of a single, massive lookup table encompassing both premixed and non-premixed combustion, the authors create two separate tables: one for the premixed regime and one for the non-premixed regime. During the CFD simulation, a "combustion regime indicator" is calculated in real-time. Based on this indicator, the simulation interpolates the necessary thermochemical data from the appropriate table. This effectively decouples the two combustion regimes, leading to smaller tables and faster generation.
To further enhance the simulation workflow's cost-effectiveness and accuracy, the dual-table FGM model is coupled with the Partially Averaged Navier-Stokes (PANS) turbulence model. PANS is a hybrid approach that aims to capture essential large-scale turbulent fluctuations while modeling smaller scales, offering a better balance between accuracy and computational cost compared to traditional RANS or computationally expensive LES. The specific PANS variant used is based on the k-ζ-f model.
The FGM tables were generated using the AVL TABKINTM tool, and the CFD simulations were performed using the AVL FIRETM code.
Datasets
The paper utilizes experimental data from a large, single-cylinder research engine at the Large Engine Competence Center in Graz for validation. Specifically, in-cylinder pressure traces and rate of heat release data were used for three distinct operating points, varying the pilot diesel injection timing.
Key Insights and Ideas
- Decoupling for Efficiency: The fundamental insight is that separating the premixed and non-premixed combustion regimes into distinct FGM tables is key to managing computational complexity. This "divide and conquer" strategy significantly reduces memory and generation time without sacrificing accuracy.
- PANS as a Practical Turbulence Model: The PANS turbulence model proves to be a valuable tool, providing more detailed flow field information than RANS on coarser meshes, which are often more practical for industrial simulations than the fine meshes required for LES.
- Dual-Fuel Chemistry is Crucial: The study highlights that accurately capturing the thermochemical effects of both fuels (methane and diesel) is essential for predicting dual-fuel combustion, particularly ignition delay and emissions. The FGM approach, even with tabulation, can capture these effects.
- Cycle-to-Cycle Variation Prediction: The proposed methodology shows promise in predicting cycle-to-cycle variations, a critical aspect for engine performance and emissions, with significantly lower computational cost than detailed chemistry simulations.
Key Results
- Reduced Table Generation Time and Memory: The dual-table FGM approach demonstrated significant reductions in table generation time and memory requirements compared to a single-table approach that attempts to capture both regimes simultaneously. For example, memory sizes were reduced from 44 GB (PV&Z) to 8.7 GB (Dual Table) for mechanisms without NOx.
- Accurate Combustion Prediction: 3D CFD simulations using the dual-table FGM and PANS turbulence model showed good agreement with experimental data for in-cylinder pressure traces and rate of heat release across three different engine operating points.
- Improved Accuracy over Simplified Models: Simulations considering only premixed (PV2) or only non-premixed (Z2) regimes individually led to over- or under-prediction of pressure and heat release, respectively. The dual-table approach (Z2 & PV2) provided the most accurate results.
- Computational Cost Savings: The FGM-based approach drastically reduces computational time. A single high-pressure phase combustion cycle simulation took 77 days with detailed chemistry, whereas the FGM PANS approach required only 23 hours per case, and 28 days for 30 cycles.
Limitations and Future Work
- Chemical Mechanism Dependence: The accuracy of the FGM model is inherently tied to the quality and completeness of the underlying chemical kinetics mechanisms used to generate the lookup tables.
- Experimental Data Limitations: While validated against experimental data, the study focused on a specific single-cylinder engine. Further validation across a wider range of engine types and operating conditions would be beneficial.
- Future Work: The authors suggest that this methodology can be extended to other dual-fuel applications and that further optimization of the PANS model and FGM coupling could lead to even greater computational efficiency and accuracy. The potential for using this approach for detailed design and optimization of dual-fuel engines is highlighted.
Summary updated