Abstract
<jats:p>Specific fuel oil consumption (SFOC) deviation in service is routinely attributed, without formal separation, to some combination of bunker fuel quality, engine mechanical condition, ambient operating conditions, and load profile. Existing correction practice — net calorific value (NCV) normalisation to an ISO reference and ambient correction under ISO 3046-1/15550 — removes the deterministic component of fuel-quality effects but leaves the combustion-quality component (incomplete combustion, ignition delay, afterburning associated with poor ignition characteristics) unseparated from genuine engine degradation. This paper proposes a regression-based framework that reconstructs bunker-batch boundaries from tank remaining-on-board (ROB) records, applies deterministic NCV correction, and then estimates a fixed-effects model in which each bunker batch enters as a categorical variable alongside engine load and a continuous time-drift term representing mechanical degradation. The batch coefficients recovered by this model are interpreted as the combustion-quality penalty of each bunker batch, net of energy content, load, and engine wear, with confidence intervals derived from heteroskedasticity- and autocorrelation-consistent standard errors. Three identification threats — confounding with maintenance events, confounding with load profile, and confounding with ambient conditions — are formalised as automatic diagnostic checks, and a placebo test using randomised batch boundaries is proposed to guard against spurious attribution. The framework is demonstrated on a synthetic twelve-month operating dataset for a representative handysize bulk carrier with injected ground-truth effects; the batch with the largest injected effect is recovered within its estimated confidence interval, while smaller and sequentially later batches show attenuated identification consistent with a formally diagnosed collinearity between batch timing and the continuous degradation term — a limitation the paper reports explicitly rather than obscures. The paper situates this contribution against the existing literature, which addresses fuel ignition quality (CCAI, CII, EFN) and in-service SFOC monitoring as separate bodies of work; to the authors' knowledge, no published methodology jointly estimates bunker-batch-specific combustion penalties from routine in-service reporting data under an explicit identification framework. Implications for bunker procurement, fuel quality dispute resolution, and vessel performance monitoring practice are discussed, along with the method's data and instrumentation requirements and its limitations.</jats:p>