Field-Test-Based Heat-Loss and Resolved Exergy-Loss Assessment of a 660 MW Ultra-Supercritical Boiler Under Variable Load

Flexible operation alters the thermodynamic performance of coal-fired boilers, but conventional heat-loss indicators do not fully describe the quality of energy discharged from the system. This study evaluates a 660 MW ultra-supercritical boiler and its regenerative gas–air heater (GAH) using thermal-performance-test data at 264, 330, 396, and 660 MW. A data-traceable framework is established to distinguish conventional lower-heating-value-based heat-loss indicators from independently resolved exergy-loss terms. The second-law assessment focuses on the physical and chemical exergies of the dry flue gas and the chemical exergy associated with unburned carbon in solid residues, all of which can be independently reconstructed from the archived field-test records. The LHV-based boiler efficiency increases from 93.75% at 264 MW to 94.32% at 660 MW. The dry-stack chemical-exergy-loss fraction ranges from 0.0294 to 0.0312 and exceeds the corresponding physical-exergy-loss fraction of 0.0060–0.0069 at all four loads, while the chemical exergy associated with unburned carbon remains below 8.7×10−4 of the fuel chemical exergy. The total explicitly resolved exergy-loss fraction increases from approximately 0.0360 at 264 MW to 0.0389 at 660 MW. Heat-loss-based proxies are numerically higher than the explicitly calculated dry-stack exergy-loss fractions at all four loads and cannot be substituted directly for second-law exergy-loss quantities; the numerical difference reflects distinct thermodynamic definitions rather than an additional physical loss. Because sufficiently complete air-side mass-flow and leakage data are unavailable, the GAH is characterized using directly traceable temperature-change indicators rather than a quantitative exergetic efficiency. The results demonstrate that first-law heat-loss indicators and explicitly resolved second-law exergy losses provide complementary but non-interchangeable diagnostics of wide-load boiler operation, while the proposed framework provides a traceable basis for thermodynamic assessment using archived industrial field-test data when complete boundary information is unavailable.