Energy efficiency in coal pulverizing systems for utility boilers: a critical review of particle-size control, drying–classification interactions, and coordinated optimization

Coal pulverizing in utility boilers is not simply a milling duty; it is a coupled fuel-preparation, drying, classification, and transport process that governs burner input quality, auxiliary power demand, combustion stability, and emissions. Performance is still often judged by mill motor power or a single fineness index, although the evidence reviewed here shows that plant-level efficiency is controlled by trade-offs among grinding electricity, drying duty, internal recirculation, primary-air power, unburned carbon, NOx, wear, and load-following constraints. This review critically evaluates recent and seminal studies from a coal-preparation-and-utilization perspective. Compared with earlier reviews focused mainly on combustion modeling, coal-mill control, or boiler-side optimization, the present synthesis explicitly connects PSD descriptors, moisture conditioning, classifier sharpness, and pulverizer dynamics with boiler heat-rate and emissions penalties. Quantitative comparisons indicate that reported improvements are highly context-dependent: pulverizing auxiliaries may consume about 0.5–2% of gross generation, circulation-load control can change the yield of −0.09 mm pulverized fuel from 22% to 43% at the same energy input, and advanced boiler optimization studies typically report thermal-efficiency gains below 1% with NOx reductions from several to tens of mg/m3. These magnitudes are industrially meaningful only when safety and fuel-quality constraints are preserved. The review therefore defines economic fineness as a mode-dependent minimum of a total plant penalty function, proposes a constrained objective for coordinated mill-classifier-airflow-boiler optimization, and identifies deployment barriers including hidden PSD states, model drift under wear and coal changes, actuator limits, and operator trust. The most defensible improvement pathway is a layered supervisory architecture supported by reliable state estimation, physics-constrained soft sensing, and wear-aware model maintenance.