A cracker furnace starts losing run length as soon as it returns to hydrocarbon service. Coke begins to build on the radiant tube walls at once, and operators have to judge how fast each furnace is moving toward its real limit. Furnace efficiency degrades nonlinearly through each run, so the timing of the next decoke is a recurring economic decision, made furnace by furnace.
Decoking timing depends on how feed, furnace condition and the tightening constraint evolve through each run. A fixed calendar rarely matches that.
How coke buildup forces the decoking decision
Coke buildup tightens several constraints on different schedules: tube metal temperature (TMT), coil pressure drop and transfer line exchanger (TLE) fouling. Each begins to move as soon as hydrocarbon feed enters the radiant coil.
Two coking mechanisms are at work. Catalytic coking forms early on bare tube metal. Pyrolytic coking dominates later, on the growing coke layer. Heavy aromatics also condense in the TLE, so the binding limit is not always in the radiant coil.
The problem reinforces itself:
- As coke insulates the tube wall, operators adjust setpoints to hold cracking temperature. That shifts yield trade-offs and demands more firing, which accelerates coking further.
- Pressure drop rises as coke narrows the effective tube diameter. Higher hydrocarbon partial pressure then speeds deposition.
- TMT rises through the run even when firebox temperature stays constant, and it does not rise linearly. Degradation is slower at the start and faster near the end.
That last point is one reason fixed schedules often miss the most important part of the curve.
Why TLE fouling can become the binding limit
TLE coke deposition can accelerate faster than radiant coil deposition as temperature rises. A team focused only on radiant coil performance may miss the limit that actually decides when the furnace has to come down.
What a steam-air decoking cycle costs
Steam-air decoking is the traditional method, and each cycle costs production hours, tube life and emissions. Feed flow stops, steam purges the coil, and a controlled steam-air mixture oxidizes the coke to CO₂ until the CO₂ concentration confirms the coil is clean.
Depending on coil configuration, the full cycle from isolation to return to service typically takes 1 to 3 days. It can stretch longer when coke removal is slow or when downstream conditions delay the return to hydrocarbon service.
Every day a furnace is offline removes production capacity, and repeated decokes add up to structural downtime across the year. Each cycle also consumes tube life through thermal fatigue, with the hottest coil sections taking the most damage from aggressive decoking conditions, and each one generates CO and particulate emissions.
Cycle duration as a capacity lever
Cycle duration varies more than the textbook description suggests. Coil layout, decoking medium ratios, hold times at temperature setpoints and the experience of the team running the procedure all change how long a cycle takes. Two furnaces of the same design at the same plant can run cycles that differ by several hours, and the same furnace can run faster or slower from one decoke to the next.
In a furnace-limited plant, where the cracker section caps ethylene output, decoking duration matters almost as much as run length. A typical olefins plant runs 6 to 12 furnaces, with at least one in decoke and one in hot standby at any time, so downtime on a single furnace is a meaningful share of available capacity. Cutting hours from each cycle recovers capacity that would otherwise stay unavailable. That lever often gets less attention than run length extension, although its economic effect can be similar. Documenting the variation in cycle time, then narrowing it, is one of the more underrated improvements in furnace-limited operations.
What sets run length
Beyond the physical constraints inside the coil, feed type and severity drive most of the variation in run length from one furnace to the next. Naphtha furnaces typically run 30 to 60 days between decokes. Ethane furnaces can run longer under favorable conditions, often 60 to 90 days. The range reflects how feed quality, coil material, severity and design choices interact.
Even in naphtha-only operation, batch-to-batch variation in aromatic content changes run length at similar severity settings.
Mixed feeds and the furnace fleet
Mixed-feed crackers face a harder scheduling problem. Different feeds coke at different rates, and several furnaces usually cannot be decoked at the same time without unacceptable production loss. Run length, feed assignment and decoking sequence have to be managed together across the fleet, which makes decoking a plantwide problem more than a single-furnace one.
Feed supply and maintenance conditions also change continuously. A decoking schedule works better when it follows current conditions, both the coil limits and the yield trade-offs across severity, feed quality and available furnace time.
Forecasting when each furnace reaches its limit
Coking rate responds to many interacting parameters, which is why observed run lengths vary so widely. The list includes:
- tube metal temperature
- dilution steam ratio
- coil pressure ratio and coil outlet pressure
- feed composition and feed rate
- firing rate and excess oxygen
- burner configuration and venturi ratio
No single-variable rule captures how these interact.
Where conventional control drifts
Advanced process control (APC) delivers value in furnace stabilization. Its accuracy can drift as coke accumulates, because the gap between the controller's design basis and actual conditions widens through the run. Accuracy falls off near the decoking limit, which is the period when better predictions matter most.
What a learned model adds
A model trained on the plant's own data can learn how that wider set of parameters drives the TMT trajectory. That gives operations a forward view of when each furnace is likely to reach its threshold under current conditions, and how a severity adjustment would move that date.
The data that matters most tracks whichever constraint is tightening first: TMT, coil pressure drop, TLE outlet temperature, feed composition, severity, dilution steam ratio and firing conditions. Looking at one variable in isolation usually misses how coking develops through the run.
The model does not need a clean, complete data foundation to start. Most plants already hold years of historian data at the resolution needed to learn from past runs. Data quality usually improves as the model surfaces gaps that the team then closes, which is why sound historian practice matters as much as the modeling layer above it.
Manual decision support can combine feed quality, severity and coil condition in one view. Doing that across eight furnaces with shifting feed slates and staggered run phases stretches what a team can hold in mind. Decades at the board build pattern recognition no model reproduces, but coordination at that scale is hard to optimize in real time without one.
One view for planning, operations and maintenance
When planning, operations and maintenance share one model of plant behavior, decoking becomes a coordinated decision instead of a negotiation between functions. The three groups can compare run-end risk, feed assignment, decoking sequence and outage timing against the same picture of furnace condition. Planning can set LP targets from actual furnace condition, and operations and maintenance can agree on which furnace to pull next. That kind of coordination is what lets optimization extend beyond a single unit, as described in self-optimizing petrochemical plants.
How trust builds in advisory mode
Plants usually start in advisory mode, where the model recommends decoking timing and severity adjustments and operators decide whether to act. The model compares decoking timing against current furnace condition, shows how feed quality and severity shift run-end risk, and gives every shift a consistent view of which constraint is tightening first.
Typical advisory situations include:
- a feed switch that changes coking rate mid-run
- a severity decision when a downstream unit constraint emerges
- a shift handover where the outgoing crew has been managing a tightening constraint that the incoming crew may not yet see
In each case the team works from the same picture, which shortens the time to a decision and reduces the chance that conservative operating margins accumulate across shifts.
This is where human and AI collaboration matters. Experienced operators test the recommendations against live furnace behavior and their own judgment, learn where the model adds value, and identify where it needs refinement. Newer operators get a clearer view of how furnace condition, feed quality and run-end risk fit together. As that confidence builds, plants move toward closed-loop control, with operators setting the limits and able to take over from the board.
How the Imubit Platform applies
Imubit applies Map, Model and Control to cracker furnaces and the wider olefins plant as part of its work in chemicals and polymers. Map traces lost run length and decoking capacity to the recurring decisions behind them, such as severity targets, feed assignment and decoke timing. Model uses Deep Learning Process Control® (DLPC®) to learn each furnace's nonlinear behavior through the run from its own history, guided by first principles. Control takes the severity decisions to closed loop, writing setpoints to the existing APC and DCS, inside limits set by plant engineers. Decoke timing and feed assignment stay with planning and operations, who schedule them against the model's run-end forecast.
