Definition
Crude Blend Cross-Functional Cost Index, or CBCCI, is the ratio of downstream processing cost increases attributable to AI-optimized crude blend decisions to the crude procurement cost savings those decisions achieved, measured across a complete crude-processing-to-product cycle. The KPI exists to answer one decision: is the AI blending optimizer''s procurement savings being offset by downstream costs the optimizer''s objective function did not consider, and what is the net economic contribution of the optimizer after the offset?
The Measurement Gap
Ines is the planning and economics manager at a coastal refinery configured for a wide variety of crude slates. The refinery''s planning department uses an AI blending optimizer that ingests current crude availability, prices, product demand, and unit constraints, then produces a recommended crude purchase and blend plan for the next two-week feed window. The optimizer''s output has been a clear win on the crude procurement budget: average procurement savings of about forty-eight cents per barrel against the planning-team baseline, mostly by finding favorable spot opportunities in heavier and more sulfurous crudes that the human planners would have been more conservative about.
The procurement savings showed up in the right column of the right ledger every month. The trouble showed up in three different ledgers nobody was netting against the procurement number. The first ledger was hydrotreater catalyst spend. The hydrotreater team had been replacing catalyst about every twenty-eight months pre-AI. Under the optimized slates, catalyst life had dropped to twenty-one months on average, with one bed coming out at sixteen months after a particularly metals-laden crude package. The replacement cost was carried as maintenance and reliability spend, not as a crude-decision cost. The second ledger was the corrosion inhibitor and neutralizer program. The crude unit''s TAN-handling chemical spend was up about thirty-eight percent year-over-year, driven by occasional high-TAN slates the optimizer had selected for their attractive price spread. The third ledger was the off-spec product downgrade. Twice in the last eight months the FCC had produced gasoline with a sulfur signature that required downgrading a couple of cargoes: once because a low-quality intermediate from a heavy slate was harder to desulfurize than the unit''s model had anticipated, once because the cumulative catalyst poisoning from the prior several months caught up with the unit''s yield slate.
When Ines totaled the incremental cost on those three ledgers and divided by the procurement savings the optimizer had generated, the ratio was 0.94. The optimizer had saved $11.8 million in crude procurement over the year and produced $11.1 million in incremental downstream cost the optimizer''s objective function had not been asked to think about. The program was not a disaster (it was net positive by a thin margin) but the headline procurement savings number that the executive review meeting reported was almost twice the true net contribution of the optimizer to the refinery''s bottom line.
Legacy refinery economics metrics do not catch this. Crude procurement savings are reported by procurement against the procurement budget. Catalyst spend is reported by reliability against the maintenance budget. Chemical spend is reported by operations against the operating budget. Off-spec product is reported by commercial against the production accounting ledger. Each function reports its own number. No standing meeting nets them against each other for AI-driven blend decisions specifically. The optimizer''s objective function operates on a subset of these costs (typically crude prices and a few unit constraints) and the optimization happens against the subset.
What is missing is a metric that runs the subtraction across the cost centers the optimizer''s decisions touch.
Formula
Crude Blend Cross-Functional Cost Index, or CBCCI, is the ratio of downstream processing cost increases attributable to AI-optimized crude blend decisions to the crude procurement cost savings those decisions achieved, measured across a complete crude-processing-to-product cycle. The KPI exists to answer one decision: is the AI blending optimizer''s procurement savings being offset by downstream costs the optimizer''s objective function did not consider, and what is the net economic contribution of the optimizer after the offset?
The structural form is a ratio. The numerator is the incremental downstream cost attributable to the AI-optimized blends versus a historical-average reference blend, summed across the cost categories the optimizer''s decisions actually affect: catalyst life reduction, corrosion-program chemical increase, fouling-driven heat exchanger cleaning, off-spec product downgrade, unplanned outages attributed to feed quality. The denominator is the crude procurement savings the optimizer generated against the same reference. CBCCI below 1.0 means the optimizer is delivering net economic value. CBCCI above 1.0 means the downstream cost has exceeded the procurement savings and the optimizer is destroying value on a fully-loaded basis.
The reference blend matters more than is usually appreciated. The procurement savings number is meaningful only against a counterfactual: what would the refinery have purchased and processed if the AI had not been selecting slates. The reference should be defined explicitly, a historical average over a stable period before the optimizer was deployed, or the planning team''s recommended baseline produced for the same period as a parallel exercise. Without a defended reference, both the numerator and the denominator are arguable, and the resulting CBCCI is a number people can dispute.
Attribution is the harder methodological question. Catalyst replacement decisions, corrosion treatment levels, and off-spec product events all have multiple plausible causes. A catalyst bed that comes out early may have done so because of a metals-loaded crude, because of an upstream operational excursion, because of a regenerator issue, or because of cumulative aging that would have produced replacement around that time regardless. The attribution to specific blend decisions versus other operational causes requires engineering judgment, has to be documented, and benefits from being reviewed by a cross-functional team rather than left to a single analyst. The integrity of CBCCI depends on the discipline of the attribution practice.
The time-lag structure of the metric is the second methodological challenge. Some downstream effects are immediate: distillation yield against an unusual crude shows up in the same operating day. Some effects are days or weeks away: corrosion impact may take a few cycles to manifest as inhibitor demand. Some effects are months away: catalyst life reduction may not be measurable until the bed is taken out. CBCCI should be computed on a rolling window (typically thirty days for chemical and product-quality effects, extending to a full catalyst cycle for catalyst-life effects) and operators should expect the numerator to be back-revised as longer-lag effects mature.
Proposed formula, pending validation pilots.