Refinery margin often slips away in the handoff between units. A crude change at the front end shows up later as tighter hydrogen, weaker conversion or a harder blend. Those losses matter most when product spreads narrow and operating costs rise, and refineries that hold profitability through those periods manage the crude-to-product chain as one connected system.
The difference shows up inside the operating day. Planning sets a margin target, but refinery operations run against real conditions: crude quality drifting from the assay, catalysts aging at different rates across units, and blendstock properties shifting with every cut-point adjustment. Margin is defended by coordinating those interactions during the shift as well as in the LP.
Crude receipt through distillation sets the margin ceiling
Much of a refinery's margin potential is set before crude reaches the atmospheric column. Crude selection is both a procurement decision and a unit loading decision. Sour crudes load the hydrotreaters harder, heavy crudes generate more vacuum residue, and opportunity crudes test desalter limits. When the crude actually run differs from the assay the LP used, yield predictions start to diverge from plant behavior before the first barrel clears the desalter.
Desalting decides whether crude flexibility pays
Salt left in the crude deposits in heaters, forms acids that corrode downstream equipment and poisons catalysts. Chloride carryover can move through atmospheric residue, hydrolyze in the vacuum column and reach diesel hydrotreater feed as hydrochloric acid.
What starts as a desalter problem can surface as a reliability event several units away and weeks later. The link between crude processing cost and equipment damage is often hidden until someone traces the crude batch back.
Where the VGO-to-residue split gets set
The crude distillation unit (CDU) and vacuum distillation unit (VDU) set the refinery's yield structure. Every barrel shifted from vacuum residue to vacuum gas oil (VGO) at the VDU captures a value differential, because VGO feeds the conversion units while vacuum residue routes to the coker. That differential is one of the largest levers in the refinery's yield economics, and it moves with every change in crude quality, column temperature profile and pumparound flow.
Flash zone management controls the split. Higher vaporization recovers more VGO, but it also calls for higher slop draws to keep metals and asphaltenes out of the heavy VGO. CDU steam injection, pumparound flows and side-draw control all affect distillation yield and VDU feed quality at the same time. Optimizing one without the other leaves value behind.
Conversion and treating set the margin split
The fluid catalytic cracker (FCC) and the hydrocracker are the main conversion units in a complex refinery, and they compete for the same VGO. Allocating VGO between them against current product spreads is a central decision in refinery LP planning. The LP sets that allocation on static assumptions, so when crude quality or unit performance shifts mid-cycle, the best split moves away from the plan.
FCC
The FCC converts heavy feed into gasoline-range blendstock, LPG and light cycle oil. Catalyst selection has to match the crude slate, because heavier, more contaminated crudes bring metals that deactivate catalyst and narrow the operating window. Small reductions in FCC bottoms yield can have an outsized effect on total bottoms production, and that effect carries into downstream blending constraints.
The relationship between feed contaminants, catalytic cracking severity and product slate is nonlinear enough that unit-level controllers can miss the wider impact.
Hydrocracker
The hydrocracker converts VGO into middle distillates under high hydrogen pressure. Hydrocracker optimization carries more weight as diesel and jet spreads move. Fractionation scheme design, management of heavy polycyclic aromatics and equipment margin all shape the unit's contribution.
When conversion decisions are not coordinated with downstream blending needs, each unit is optimized within its own boundaries while system-level margin leaks between them.
Hydrogen balance ties the units together
Hydrotreating links the whole refinery. Naphtha hydrotreaters prepare reformer feed, diesel hydrotreaters meet sulfur specifications, and FCC naphtha hydrotreaters manage the sulfur and octane trade-off ahead of gasoline blending. Hydrogen consumption across those units ties back to the catalytic reformer, which is both an octane unit and often the main internal hydrogen source. A reformer outage constrains hydrogen to the hydrotreaters and the hydrocracker at the same time.
Individual unit controllers do not see those trade-offs. The hydrogen balance is one of the most consequential cross-unit interactions in a complex refinery, and it shifts every time feed quality or unit severity changes.
Product blending captures or gives away margin
Blending is where upstream improvements are captured or lost. Specification giveaway at the blend header can erase value created in the process units. At a refinery running 100,000 bbl/day, a small quality giveaway on every batch adds up to a significant annual margin loss. Most discussions of refinery ROI focus on throughput and yield, but blending giveaway is often where recoverable value is easiest to measure.
Every blend carries some built-in giveaway. Blenders hold average quality on the safe side of each specification limit so batches pass consistently, and that safety margin has a cost. The harder constraint is that octane, sulfur, RVP, cetane and distillation targets all move against each other as blendstock qualities change from several upstream units. High-octane blendstocks often carry elevated RVP, which creates trade-offs at the blend header that cannot be resolved one property at a time.
How conservative decisions compound
Blendstock qualities are not static. Every cut-point adjustment at the CDU, every severity change in conversion and every shift in feed quality arrives at the blend header as a changed property in a component stream. Under rundown blending, where unit streams route directly to finished product tanks, the options for correction narrow sharply.
Giveaway rarely comes from operator error. It accumulates through repeated conservative decisions, shift after shift, made to protect product quality when blendstock properties are moving targets. Over weeks and months, those small concessions add up to a measurable gap between what the refinery could have captured and what it shipped.
Cross-unit coordination recovers margin between units
The conventional refinery optimization stack has a gap between its two layers. The LP works on a global horizon, but it produces static plans, usually on weekly or monthly cycles, that cannot adjust as conditions shift. Advanced process control (APC) executes in real time, but each controller typically sees its own local objective. The coordination problem sits between those two layers.
Where static plans and local controllers leave margin
When CDU cut points shift in response to a crude slate change, FCC feed quality changes with them. That alters conversion, yield structure and blendstock availability at the blend header. Unit-scoped APC cannot coordinate that cascade because each controller optimizes inside its own boundaries.
The effects are not proportional either. A modest shift in CDU feed quality can alter VGO composition, shift hydrogen demand and change blendstock properties by more than the change at the front end would suggest.
LP models add another limit. Even a well-maintained LP works from simplified representations of nonlinear reactor behavior, and those simplifications compound when several units interact. When crude quality drifts from the assay, yield predictions diverge from plant behavior, most of all in conversion units with strong feed-quality sensitivity. Operations is left with a planning target it cannot reach with the tools it has.
What models learned from plant data add
Plantwide optimization addresses that gap by learning relationships across unit boundaries from the refinery's own operating data instead of simplified linear models. It can follow interactions that no single operator or unit-scoped controller can track at once.
It complements the LP. Planning keeps setting strategic targets on its own cycle, and the real-time layer adjusts setpoints as conditions change, closing the gap between plan assumptions and how the plant runs.
Advisory mode is usually where teams start to see that coordination. The model recommends setpoint changes that operators evaluate against their own experience. It is useful for what-if analysis when throughput, quality, hydrogen and blending constraints conflict at once, and planning teams can test scenarios against the same plant-specific relationships the operations team relies on during the shift.
Engineering can test whether recurring trade-offs reflect equipment limits, degradation or changing feed behavior. Instead of each group working from separate assumptions, operations, planning and engineering discuss constraints and priorities around one model of the plant. That shared model is the basis of a self-optimizing plant.
From recommendations to closed loop
Once the team has validated the quality of the recommendations, the controller moves to closed loop, writing setpoints to the APC or DCS inside boundaries the operators define. Operators keep authority throughout and can take over from the board.
Blending giveaway and cut points are usually the easiest places to audit the margin effect. Coordination across more units follows as operations, planning and engineering gain confidence through daily use.
A model will not capture every instinct behind an experienced judgment call, and unusual disturbances or novel operating conditions still need operator and engineering judgment. What it does capture is what the data shows about which operating moves produced the best outcomes under similar conditions.
How the Imubit Platform applies
In refining, the Imubit Platform applies Map, Model and Control across the crude-to-product chain. Map traces the margin lost between units to the recurring decisions behind it, such as cut points, conversion severity, hydrogen allocation and blend targets. Model uses Deep Learning Process Control® (DLPC®) to learn a nonlinear dynamic model of the units and how they interact from the refinery's own history, guided by first principles. Control takes the critical decisions to closed loop, writing setpoints to the existing APC and DCS minute by minute, inside limits set by plant engineers. Imubit is trusted by 7 of the top 10 U.S. refiners.
