A delayed coker converts low-value vacuum residue into naphtha, gas oils and other products that keep downstream units fed. Every day the coker heater stays online protects that conversion. As crude slates trend heavier and feed quality shifts, many refineries are watching heater run lengths shrink.
Run length is not fixed by equipment design. It responds to operating discipline, to the monitoring strategy and to how well the interacting variables around the heater are controlled. Heater fouling is the dominant constraint, and the causes of premature fouling are well understood.
Why run length defines the coker's economic value
When a coker heater fouls early, the impact reaches well beyond the coking unit. It starves hydrocrackers and catalytic cracking units of feed, forces the crude unit to cut rates or divert residue to lower-value outlets, and limits the refinery's ability to capture heavy-light crude differentials. Every week lost means an earlier decoking outage and less total conversion from the barrels the refinery ran that cycle.
That exposure grows as crude slates get heavier. Heavier barrels add load on conversion capacity for high-residue feeds, and a refinery that can reliably extend coker run length recovers more conversion from every cycle.
What causes heater fouling
Heater fouling is the primary run-length constraint in nearly every delayed coker. As vacuum residue flows through the furnace tubes, rising film temperatures destabilize the asphaltenes in the feed. These heavy molecules precipitate onto the tube walls, dehydrogenate into hard coke and progressively restrict flow.
The relationship between film temperature and fouling rate is exponential. A small increase in tube-wall temperature can accelerate coke formation sharply.
Heavier crudes make this worse. They carry higher concentrations of asphaltenes, metals and Conradson carbon residue, all of which accelerate coke deposition. As deposits insulate the tube, wall temperatures rise further and the cycle becomes self-reinforcing.
Contaminants add to it. Sodium, typically from desalter upsets or crude unit caustic injection, and iron compounds both promote fouling. The result is rising pressure drop across the coils, climbing tube-metal temperatures (TMT) and, eventually, a forced shutdown for cleaning.
Operating factors that shorten the run
- Flow imbalance across heater passes. This is the biggest controllable driver. When one pass receives less flow than its neighbors, its film temperature rises, fouling accelerates, and the whole heater's run length is limited by the weakest coil. A single starved pass can force a shutdown weeks before the other passes approach their limits.
- Excessive firing in response to declining throughput. Pushing film temperatures past the point where fouling becomes self-reinforcing is the most common and most counterproductive response to early fouling signals. It shortens the remaining run instead of extending it. Operators often face real pressure to hold throughput targets, which makes it hard to reduce firing even when the heater data says they should.
- Upstream instability. Upsets in the crude distillation or vacuum unit change feed density, metals loading and asphaltene content from shift to shift. Each upset forces the heater to chase a moving target, and the compounding effect on fouling rate is hard to reverse.
None of these drivers is new. The difficulty is managing all of them at once, across multiple heater passes, shifting crude slates and three shifts of operators. That exceeds what manual oversight can reliably sustain.
Operating practices that extend run length
The following practices extend heater run length with disciplined execution and no new technology.
1. Pass balancing
Pass balancing is the highest-value intervention. Comparing pressure drop across every coil at the start of each shift reveals emerging imbalances before they become emergencies. When a deviation appears, adjusting pass flow control to equalize flow keeps tube-metal temperatures converging instead of diverging.
Balanced flow keeps velocity in every pass high enough to scour early deposits from the tube walls, and uniform heat flux keeps film temperatures from drifting apart. That alone can add days to the run. Automatic ΔP alerts configured in the DCS catch imbalances faster than manual monitoring.
A weekly audit of the tube-metal temperature spread across all passes, and not just the average, confirms whether balancing is holding or whether slower-fouling passes are masking a problem coil.
2. Predictive fouling monitoring
Predictive fouling monitoring turns existing data into an early warning. The key variables are already in most plant data systems:
- pressure drop per pass
- tube-metal temperature trends
- heater firing rate progression
- calculated fouling rate, expressed as degrees of TMT rise per day
Combined into a composite fouling index and tracked shift to shift, they show process engineers how fast the heater is degrading and how much run time remains. A heater fouling at 1 °F per day has different economics from one fouling at 3 °F per day. Without consistent tracking, both look the same until the shutdown alarm. Fouling still surprises plants because the right indicators are not watched together.
3. Upstream coordination
Upstream coordination keeps the coker from absorbing instability it did not create. Holding crude unit cut points, maintaining a consistent vacuum furnace coil outlet temperature and managing feed contaminant levels keep the coker heater's operating window stable. When upstream variables stay flat, heater firing can run lower, tube-metal temperatures drop and fouling slows.
A cross-functional team spanning operations, process engineering and maintenance is what makes this coordination sustained rather than episodic.
Where conventional control reaches its limits
Advanced process control (APC) handles coker heaters reasonably well under stable conditions. When feed quality shifts, or when several variables interact nonlinearly, static linear models lose accuracy. The heater then needs a continuous recalibration that manual adjustment and conventional controllers cannot sustain across a full crude-switch cycle.
What a learned model adds
A model trained on the heater's own operating history learns the relationships between feed properties, firing patterns, steam-to-oil ratios and fouling progression. It does not rest on a single linearized operating point. It reflects how the heater has responded across the range of crude slates and equipment conditions it has actually run. The controller that acts on that model is trained with reinforcement learning across many simulated operating scenarios before it moves a setpoint.
To start, the model needs consistent process data on pass temperatures, coil pressure drops, flow rates and firing. Lab feed quality data and upstream unit conditions strengthen it. The data does not have to be perfectly structured: the model can learn from the existing plant history while the data infrastructure improves.
Advisory mode
Plants usually start in advisory mode. The model highlights emerging fouling trends from the combination of indicators instead of a single screen, which gives engineers a window to intervene before a trend becomes a crisis. Teams can test alternative firing strategies or steam injection rates against the model's predictions before committing to a move, and see the result of a change before they make it.
Because every shift references the same model, variability between shifts in heater management drops. The model runs alongside the existing APC and validates its recommendations against live operation before it is given authority to act.
Closed-loop control
As operators gain confidence, the model writes setpoints to the DCS, adjusting heater duty, steam rates and recycle to keep every pass inside its safe operating window. Continuous adjustment lets the heater run closer to its actual constraints instead of the conservative margins manual operation requires.
Board operators bring pattern recognition from years of operating experience that a model does not reproduce. Managing dozens of interacting variables in real time across an entire shift exceeds what any individual can sustain. Operators set the boundaries, and the model optimizes within them. The aims are longer intervals between decoking outages, steadier feed to downstream units and lower energy intensity per cycle.
How the Imubit Platform applies
Imubit applies Map, Model and Control to the delayed coker as part of its work in refining. Map traces shortened heater runs and lost conversion to the recurring decisions behind them, such as pass flows, firing response and upstream cut points. Model uses Deep Learning Process Control® (DLPC®) to learn the heater's nonlinear fouling behavior from its own history, guided by first principles. Control takes the heater's pass flow, firing and steam decisions to closed loop, writing setpoints to the existing APC and DCS, inside limits set by plant engineers.
