Inside the furnace, fuel energy becomes a dynamic radiative field — composed of flame luminosity, particulate radiation, and infrared-active combustion gases. The effectiveness of that radiative coupling determines how efficiently the boiler absorbs useful heat. Combustion is controlled. Heat transfer is not. Existing systems control combustion inputs. We analyse the resulting radiative-transfer state. A boiler can be perfectly controlled — and still not be thermodynamically aligned. Current systems control fuel, control air, and manage constraints. They do not explicitly control: - effective flame emissivity - radiative coupling to furnace surfaces - real-time heat-transfer efficiency Result: the plant can be "well controlled" but still not maximising heat transfer per unit of fuel. YBG ControlAlign™ reconstructs operational indicators of the furnace heat-transfer state from plant operating history, and identifies the conditions associated with stronger radiative coupling and lower fuel intensity. The question then becomes whether those conditions are being consistently sustained. Boiler performance is governed by heat transfer. We make it controllable. --- Radiative field — operational contributors The furnace radiative field is not one thing. It is a superposition of several contributors, each with a distinct operational signature. We decompose the radiative environment into its dominant operational contributors — not a spectral inversion. 1. Luminous particulate radiation (often dominant) Sources: soot, char, ash particulates. Broadband, visually luminous, typically the largest practical contributor to wall heat absorption in coal flames. Driven by air/fuel ratio, burner mixing, residence time, flame stability. 2. H₂O vapour infrared emission A powerful infrared-active radiative gas. Emits strongly in specific IR bands; redistributes thermal radiation through the gas volume and influences furnace optical thickness. 3. CO₂ infrared emission Wavelength-selective IR emitter/absorber. Not visually luminous but thermally important — modifies radiative transport and the effective emissivity of the combustion environment. 4. Flame radicals / chemiluminescence (OH*, CH*, C₂*) Diagnostically valuable for flame structure and combustion quality. Small role in total furnace heat transfer. 5. Refractory / wall re-radiation Walls absorb, heat, and re-radiate energy back into the furnace volume — part of the total radiative field. Effective furnace emissivity is therefore not a property of one object. It is the collective radiative effectiveness of the entire combustion environment — a dynamic system property. The field drifts with burner balance, excess O₂, secondary-air distribution, coal fineness, moisture, volatile release, flame attachment, soot loading, turbulence, and furnace stoichiometry. Scope note: day-to-day analysis uses routinely available DCS signals as an operational proxy for this radiative state. The engine works in band-integrated, operationally-proxied quantities (E_f, RCI, ΔR, BBI, p·L terms) — not wavelength-resolved spectra. Spectral character is referenced only to justify why gas-phase and particulate contributors are carried separately; we do not claim spectral inversion or spectroscopic reconstruction from DCS alone. What stronger radiative coupling does — and does not — do: Drum boilers. Furnace water-walls are evaporators operating in nucleate boiling at the saturation temperature set by drum pressure. Increasing radiative coupling raises the heat absorbed in the furnace, which raises steam generation and lowers furnace exit gas temperature (FEGT). It does not, by itself, raise water-wall metal temperature in any meaningful way — the saturation temperature is fixed by pressure, and the tube metal sits only tens of °C above it as long as nucleate boiling is maintained. Nucleate boiling is the design regime: vapour bubbles nucleate, detach, and are continuously replaced by liquid, holding the inside-wall heat-transfer coefficient very high. The failure mode is departure from nucleate boiling (DNB) into film boiling, where a vapour blanket insulates the wall and metal temperature rises sharply — that is a loss-of-cooling event, not a "too much radiation" event. Benson (once-through) boilers. There is no drum and no fixed saturation plateau. Feedwater is pushed through the water-walls at full pressure, and the evaporation end-point — the wet-to-superheat transition — sits somewhere along the tube circuit. As long as the end-point stays inside its design window, stronger radiative coupling shifts the end-point earlier in the circuit and lifts enthalpy at the water-wall outlet, which is exactly what the unit is designed to do. Feedwater flow is trimmed against firing to hold the end-point in place, so metal temperatures do not rise by themselves. The failure mode is the same in physical terms: loss of cooling at the end-point (insufficient mass flux for the local heat flux), not radiation per se. Downstream metal. In both designs, stronger furnace radiative coupling typically eases superheater, reheater, and economiser metal temperatures, because less of the total duty has to be picked up convectively from hotter gas — FEGT falls and downstream gas temperature drops with it. Surface temperatures only run away when the cooling regime itself fails — DNB or dryout in drum units, end-point excursion in Benson units, waterside scale or deposits raising the tube-side resistance, or sustained flame impingement on a localised patch. The operating risk on the radiative axis is therefore not "hotter tubes from more radiation" in aggregate; it is flux distribution, waterside cleanliness, and (on once-through units) end-point control. This is why the two radiative-state metrics are framed envelope-relative rather than as quantities to be maximised without bound: RCI — Radiative Coupling Index. The share of released combustion energy that is actually being absorbed in the furnace as evaporative duty, proxied as (Steam_Flow × Δh_evap) / (Coal_Flow × HHV) and corrected for the furnace temperature gradient (FEGT vs. saturation temperature derived from main-steam pressure). It is a state variable, not a target. A high RCI in a given load band means the flame envelope, the gas-phase emitters (H₂O, CO₂), the particulate field, and the wall condition are jointly delivering strong radiative transfer at that operating point. A persistently low RCI at the same load points at a degraded radiative state — emissivity loss, flame lift or skew, slagging, or fouling — even when combustion KPIs (O₂, CO, FEGT) look acceptable. ΔR — Drift from Radiative Optimum. The gap between the current RCI and the best RCI the same plant has historically demonstrated in the same load band: ΔR(t) = R*_b − RCI(t), with R*_b = max(RCI_b) over the reference window. ΔR is recoverable by construction — the plant has already achieved R*_b under its own constraints, fuel, and hardware, so ΔR represents performance the unit has shown it can deliver, not a theoretical ceiling. ΔR ≈ 0 means the unit is operating at its own demonstrated radiative optimum for that load; ΔR > 0 is the size of the recoverable radiative gap and is what the advisory layer attributes across the operator-side levers (O₂ trim, CO breakthrough, FEGT, sootblowing cadence, burner balance). Together, RCI describes where the radiative state is, and ΔR describes how far it is from where this same plant has already shown it can be. --- Extension of best-performance output gaps to radiative-state gaps, and specific levers to regulate radiative transfer. E_f (the effective flame emissivity proxy — how strongly the flame envelope is actually radiating) and ΔR (the drift from the radiative optimum — the gap between current radiative coupling and the best the plant has demonstrated at this load) do more than describe the radiative state. They convert the engine's residual — the unexplained portion of the load-normalised gap — into a physically attributable signal rather than a statistical one. Once the residual is attributable, it can be decomposed across the operator-side levers that actually move it: - excess O₂ → FD fan bias / windbox damper - CO breakthrough → burner tilt / secondary air staging - FEGT → burner tilt / OFA bias - sootblowing cadence → wall-blower sequence interval The advisory logic is bounded: ΔR(t) = R*_b − RCI(t) → attribute ΔR across { O2, CO, FEGT, sootblowing } → clamp each lever target to the historian-allowed range → emit ranked guidance, decay to zero as ΔR closes Every recommendation sits inside operating points the plant has already demonstrated at the current load band. Guidance is auditable and advisory — not prescriptive control. See: /control-guidance (internal sandbox).