Digital modelling tools for designing and optimizing MILD biomass combustion
Please login to view abstract download link
With VEOLIA, a 176 MWt wall-fired boiler (430 t/h steam) was selected for Moderate or Intense Low-oxygen Dilution (MILD) conversion. Baseline simulations showed uneven temperatures and residual oxygen, and set fixed wall boundary conditions. The revised design keeps the convection pass but remodels the furnace chamber. Fuel and oxidizer inlets with flue gas recirculation allow adjustable angles; the oxidizer is split into three rows (60–90 m/s) for better mixing. The primary oxidizer consists of dried, desulfurized flue gas blended with fuel moisture. The selected baseline model configuration will serve as a reference for systematically varying key operating parameters—specifically, the excess oxygen ratio, the flue gas recirculation rate, and the oxidizer distribution. By adjusting these variables, a comprehensive dataset will be generated to support the development of a reduced-order model suitable for subsequent process optimization. In this phase, the relevant operating parameter space was explored using a simplified three‑dimensional numerical model. On top of this framework, a supervised AI/ML methodology was applied to learn the underlying relationships between input conditions and system behavior, enabling accurate prediction of the outlet conditions based on the simulated operating scenarios.
