FCC zeolite catalyst — today's makeup rate is decided with an analysis from three days ago.
FCC catalyst activity moves with every crude change and every metals spike — and the E-cat analysis arrives when the catalyst has already been working several shifts on last week's makeup rate. iLEAN Brain cross-references E-cat, feed metals and chromatograph yields to propose the fresh-to-equilibrium makeup rate before the next analysis arrives. The process engineer signs — the fresh catalyst silo never actuates on its own.
The FCC catalyst has a memory — and the system decides without it.
The FCC zeolite catalyst does not perform on its own: it performs through the combination of its history (how much it has aged in the regenerator, how much metal it has deposited, which rare earths it has left) and what it sees today (a heavier crude, a severity change, a vanadium spike). In most FCC units that combination is rebuilt by hand with data that arrives out of step:
- E-cat MAT/ACE analysis — external lab, 3-7 days of delay. The activity of the equilibrium catalyst is known a week after the fact.
- ICP analysis of feed metals — plant lab, a delay of hours to a full shift. A vanadium spike can pass through before the result arrives.
- Per-component yield from the online chromatograph — comes in real time, but raw, without context, not cross-referenced with makeup or with metals.
- Fresh-to-equilibrium makeup rate — decided by the process engineer with the best available combination of stale data and gut feel.
The process engineer knows it, but cannot be in four places at once. The consequence is an average activity loss worth a lot of money every year — and, in the worst case, an irreversible deactivation from a vanadium spike not defended in time.
iLEAN Brain does not replace the DCS — it seals the crack between the lab, the chromatograph and the decision.
The FCC yield problem is not a lack of data: it is data living on islands, cross-referenced out of step. The DCS runs the unit. The LIMS stores the analyses. The chromatograph spits out yields every few minutes. And the vital decision — how much fresh catalyst do I add today — is made with the best possible combination of that data whenever there is a gap in the schedule. iLEAN acts as the putty that fills the dead zone between the three, without asking you to change the DCS, the LIMS or anything in the lab.
Brain maintains the living model of the FCC. Connect captures the lab analyses at second zero. The agent cross-references, anticipates the drift and proposes the makeup rate. The engineer signs — never the other way around.
The iLEAN pieces applied to FCC catalyst yield:
- Connect — captures the E-cat MAT/ACE analyses from the external lab (PDF by email), the ICP feed metals analyses from the internal LIMS (or from the Excel sheet the lab actually uses), and the yields from the online chromatograph via the DCS historian. At second zero, without anyone forwarding anything.
- Brain (Agents + Central) — maintains a living model of the FCC: E-cat history, feed, severity, fresh makeup applied and resulting yields. It learns the unit's normal signature and detects early deviations. When the chromatograph shows a signature incompatible with the latest metals analysis, the agent proposes stepping up the fresh makeup gradually and notifies the process engineer on whatever channel they use. The proposal carries the specific data point — what moved the decision and by how much.
- Edge — optional, at sampling: vision over the automatic E-cat sampler to verify that the sample going out to the lab is the one from the declared shift, and OCR of the seals on the fresh catalyst bags on receipt from the supplier.
Makeup dosing on stale data vs. anticipated dosing with iLEAN
| Aspect | Spreadsheet + gut feel | With iLEAN Brain + Connect |
|---|---|---|
| Fresh-to-equilibrium makeup decision | Weekly, with a MAT/ACE analysis 3-7 days old | Per shift, with a living model of the FCC |
| Metals spike detection | When the ICP analysis arrives | When the dry gas chromatographic signature moves |
| Crude change | Manual reaction by the engineer | Automatic proposal with data and reasoning |
| Rare earth conservation | Quarterly estimate | Cross-referenced with E-cat and actual makeup |
| Vanadium poisoning warning | After the damage | Before the second critical shift |
| Traceability for audit / vendor management | Rebuild by hand | Monthly dossier with every cross-reference |
Impact estimate for your plant — to be validated with your numbers.
The block below is an estimate to be validated with the specific data of your unit. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.
- Refinery with a mid-capacity FCC unit, variable diet (heavy crudes with metals), conventional regenerator, integration with existing DCS and LIMS.
- iLEAN Brain deployment with living model + Connect for capture from the external lab (MAT/ACE) and the internal lab (ICP). First value expected within a few weeks: first makeup proposals cross-referenced with the chromatograph signature.
- Indicative payback between 4 and 9 months, depending on the annual cost of fresh catalyst and the cost of a lost point of activity or gasoline yield.
- Expected reduction in fresh catalyst consumption per point of activity protected ≥ 30% on metals deviations caught in time.
- The hard lever is a single irreversible deactivation avoided from an undefended vanadium spike — and that is worth millions.
And the process engineer's reasonable doubt
“What if the AI proposes the wrong makeup rate and unbalances the unit?” — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI merely cross-references MAT/ACE analyses with the metals history and the chromatograph signature, the best models brought error below 1.5% [1]. And even so, what is critical is never decided alone: iLEAN proposes and the engineer signs the makeup rate. The three safety rings exist precisely for this — the fresh catalyst silo is never actuated from the outer ring, only from the inner one, and with a human signature.
[1] OpenAI paper “Why Language Models Hallucinate”, 2025 — on the reliability of AI in anchored tasks.
What people ask about FCC zeolite catalyst yield
What drives the yield of an FCC zeolite catalyst in refining?
FCC catalyst activity depends on the Y zeolite (Si/Al ratio, rare earths), the matrix, hydrothermal aging in the regenerator, poisoning by crude metals (nickel, vanadium, sodium), residual coke and the fresh-to-equilibrium makeup rate (the ratio between fresh catalyst going in and E-cat coming out). Yields of gasoline, LPG and olefins shift with every crude change and every regenerator shift — and the decision of how much fresh catalyst to add today is made with data from two days ago.
Why does the fresh-to-equilibrium makeup decision come too late?
Because the critical data each live on their own island: the E-cat MAT/ACE analysis is delivered by the external lab in 3-7 days, the feed metals analysis arrives from the plant lab with its own delay, per-component yields come from the online chromatograph but raw and without context, and the operating setpoints live in the DCS. By the time the process engineer cross-references everything by hand, the catalyst has already been working several shifts on last week's makeup rate — and a lost point of activity is worth a lot of money.
How does iLEAN Brain cross-reference all those sources to anticipate the makeup rate?
An iLEAN agent maintains a living model of the FCC with three inputs: the E-cat analysis history (MAT, surface area, retained rare earths, deposited metals), feed metals and feed property analyses, and gasoline, LPG and olefin yields from the online chromatograph. Cross-referenced with the fresh makeup applied each shift, the agent proposes the fresh-to-equilibrium makeup rate for the coming shifts before the next external analysis arrives. The process engineer validates and signs — the makeup to the silo is never changed on its own.
What happens with a metals spike in the feed (nickel/vanadium)?
A nickel spike multiplies dehydrogenation (more dry gas and coke); a vanadium spike destroys the Y zeolite irreversibly. The decision to increase fresh catalyst to defend activity comes too late if you wait for the weekly analysis. iLEAN cross-references the plant lab's ICP analysis with the chromatographic signature of the dry gas — if the signature moves before the analysis arrives, the agent raises an alert to the engineer and proposes stepping up the fresh makeup gradually. The person validates and signs; the silo never actuates on its own.
What return does a Brain FCC optimization project deliver in a refinery?
In an FCC unit, one point of activity recovered or one point of gasoline yield protected is worth several hundred thousand euros a year in a mid-sized unit. The pilot covers the living model, integration with the DCS and the laboratory systems (LIMS), and training for the process engineer. The hard lever is a single irreversible deactivation avoided from a vanadium spike not caught in time. We ask for your data and send you the estimated ROI in 48h.
Tell us your case and in 48h we'll send you the estimated ROI of this AI project for your FCC unit.
We work on your refinery's real data, not ours. Diagnostic with no commitment.
Request estimated ROI in 48h See specialty chemicals