Data-driven and intelligent optimisation of coal-fired utility boilers under flexible low-load operation: a review

Abstract

Abstract

Flexible operation subjects coal-fired utility boilers to distinct but interacting combustion, emissions and equipment-safety constraints at low load and during rapid load changes. This review synthesises 203 publications on sensing, prediction, dynamic modelling, optimisation and system integration, linking operating problems to measurements, model resolution and validation requirements. Low-load operating points are better documented than plant responses during ramps. Low-load assessment centres on ignition, burnout and heating-surface conditions; ramping also requires attention to fuel delivery, steam/water dynamics and selective catalytic reduction. Low-order dynamic models support load-tracking analysis, whereas spatially resolved or coupled models address local heat transfer and tube-wall constraints. Validation must distinguish steady regression, transient forecasting and control, and test transfer across relevant fuels and operating conditions. Predicted optimisation gains require independent evaluation before they support claims of plant benefit. For plant application, selected settings must be linked to operating constraints and measured responses, with benefit attribution accounting for fuel and load differences and uncertainty. System integration, physical implementation and repeated automatic feedback remain distinct. Future work should combine complementary measurements with task-appropriate models and independently evaluate the resulting operating actions.