Multiphase combustion modeling of ammonia–diesel RCCI engines using RANS, DES, and LES turbulence models

This work quantifies how spray atomization and vaporization govern ammonia–diesel RCCI combustion by comparing RANS (RNG k –ε), DES (SST k –ω), and LES (dynamic Smagorinsky). Beyond validation against in-cylinder pressure, HRR, CA50, and ITE, and pollutant emissions, we analyze DV10/DV50/DV90, SMD, parcel count, and penetration to expose turbulence–spray–chemistry couplings. LES Captures fine-scale dynamics, including rapid DV50 reduction (≈28 to ≈11.5 µm), enhanced secondary breakup, and core instabilities, generating >160,000 parcels and accelerating vaporization. These resolved phenomena translate into superior combustion fidelity, closely matching experiment peak pressure (84.2 vs 85 bar), CA50 (10.24° CA), ITE (37.218%), IMEP (≈9.1 bar), while accurately predicting emission (UHC: 0.48 g/kWh vs 0.46 g/kWh, NO: 2.15 g/kWh vs 2.20 g/kWh). DES offers intermediate fidelity, whereas RANS retains oversized droplets, slow vaporization, and underprediction of combustion intensity and The total computational expenditure amounted to approximately 410 CPU-h for RANS, 1240 CPU-h for DES, and 3240 CPU-h for LES, using meshes ranging from 200,000 to 8,000,000 cells.