Based on insights shared by one of our customers, an unplanned furnace shutdown can last up to ten days, with lost production averaging $1 million per day. According to their experience, this single event can result in a loss of $10–15 million in value. Crude and vacuum heaters are vital for keeping the entire refinery operational, but everyday practices, such as pushing harder when fouling occurs, allowing imbalanced flows, and ignoring weak alarms, gradually compromise their health. This accelerates coke buildup, raises tube-metal temperatures, and ultimately causes premature tube failure, leading to costly outages.

These failures aren’t inevitable. By combining disciplined operating practices with modern analytics and AI-driven control, front-line teams are extending run lengths, reducing energy use, and protecting equipment. The five practices outlined below address the most common life-shortening habits, explain how to avoid them, and highlight emerging technologies that make sustained improvement achievable.

1. Pushing Harder When Fouling Begins

You notice outlet temperature drifting or draft dampers creeping open—early hints of fouling. The instinct is to preserve throughput by cranking up feed or firing rates. Yet that approach floods already compromised tubes with more heat and flow, accelerating coke growth inside the coils and ash deposits on the fireside.

Even a thin film of deposits can cut heat transfer enough to lift tube-metal temperature significantly, pushing you toward the creep limit of the alloy. As heat transfer deteriorates, you burn extra fuel just to hit the target outlet temperature, a feedback loop that raises stack temperature and wastes energy. Some refineries document substantial throughput penalties before a single cleaning pass restores efficiency.

The smarter approach involves maintaining stable rates once fouling indicators emerge, and relying on predictive analytics to set firing limits that protect run length. Trend skin-thermocouple data, schedule pigging or steam-air decoking based on those trends, and consider ceramic coatings that deter coke adhesion.

For sustainable efficiency improvements, Imubit’s Closed Loop AI Optimization solution evaluates fouling propensity in real time, moderating firing rates and suggesting the earliest safe window for cleaning. This keeps barrels on-spec without sacrificing coil life, turning reactive maintenance into proactive optimization.

2. Allowing Uneven Heater Pass Flows

Multi-pass furnaces perform best when each pass receives an equal share of heat, ensuring that the flow is balanced. When one pass is restricted, the others can speed up, leading to concentrated heat, higher tube-metal temperatures, and faster coking.

In some cases, even a small imbalance can cause a tube to rupture quickly, resulting in an emergency shutdown. Investigations into reactor-feed furnaces often link many tube failures to flow imbalances that arise after maintenance or rate changes.

Imbalances may stem from worn orifice plates, mis-calibrated differential-pressure transmitters, uneven manifolding, or deposits that restrict one part of the coil. Although outlet temperatures might appear steady, these issues can remain hidden until significant damage occurs.

To help address this, it can be beneficial to:

  • Re-calibrate orifice plates and flow instruments during turnarounds, while confirming per-pass flows on start-up.
  • Monitor flow across each pass and set alerts for any deviations to maintain consistent performance.
  • Consider conducting hydraulic studies before rate increases to identify any potential adjustments to headers or choke fittings.

Imubit’s technology continuously monitors real-time flow, temperature, and pressure across each pass. The system learns to detect imbalances early, adjusting charge-pump speed or trim valves to restore balance—often before operators notice any drift—helping extend run length and protect tubes.

3. Reacting Too Late to Early Warning Signals

Refinery furnaces rarely fail without early indicators. Slight increases in pressure drop, a one-degree drift in outlet temperature, or a slow rise in stack temperature often appear days—sometimes weeks—before severe fouling sets in. When those signs go unaddressed, coke builds on tube walls, metal temperatures soar, and unplanned shutdowns become inevitable.

The challenge lies in separating meaningful signals from everyday operational noise. Small anomalies hide in normal variability, especially during busy shifts, and traditional trend charts make them easy to miss. Meanwhile, fuel demand creeps upward, equipment efficiency slides, and the first visible sign may be a hot spot captured by thermal imaging. At that stage, damage is already underway.

Effective prevention requires tightening alarm limits and adding rate-of-change logic so slow drifts trigger action before absolute thresholds are breached. Deploy dashboards designed to surface weak signals—sub-percent shifts in fuel efficiency or skin thermocouple readings. 

Trend key indicators daily: heat-duty efficiency, stack temperature, and pressure drops. Combine acoustic monitoring, infrared scans, and periodic convection-section inspections to corroborate sensor data. Most importantly, train operators to treat small, persistent deviations as actionable—not background noise.

Advanced AI systems can learn the heater’s normal operating patterns and flag deviations long before they cross traditional alarms. By evaluating patterns across pressure, temperature, and flow in real time, these solutions distinguish routine variability from genuine early warning signals, then alert the distributed control system (DCS) and moderate firing rates automatically. The result is extended run length without sacrificing throughput—and far fewer unexpected shutdowns.

4. Operating the Furnace in Isolation

A crude or vacuum heater rarely fails on its own. Its thermal stability depends on steady feed properties, balanced downstream draw, and consistent combustion conditions. When a distillation column suddenly cuts rates or a hydrotreater goes off-spec, the heater must absorb rapid swings in flow and firing demand. These abrupt transitions drive tube metal temperatures above safe limits, encourage coking, and strain refractory linings—an efficiency loss quickly visible in rising stack temperature and fuel use.

The root cause is often organizational: operators optimize their own unit without seeing real-time constraints elsewhere. Deploying a plantwide monitoring platform that shares unit limits across control rooms creates the visibility needed for coordinated operations. 

Site-wide linear-program (LP) models should prioritize total margin instead of isolated throughput goals, while clear communication protocols before feed switches, rate changes, or maintenance work prevent unexpected thermal stress.

Cross-training personnel ensures every shift understands how upstream quality shifts or downstream bottlenecks translate to heater stress. When operators see the broader picture, they naturally make decisions that protect equipment while maintaining profitability across the entire system.

Modern optimization solutions can act as refinery-wide coordinators, continually reading live constraints, predicting their impact on heater duty, and writing optimized setpoints back to the distributed control system (DCS) in real-time. By moderating firing when downstream units throttle back, this technology prevents over-firing, extends tube life, and helps every unit operate as part of a single, profit-driven network.

5. Relying Solely on Manual Intuition

Heater operation involves hundreds of interacting variables—feed composition, draft, excess air, upstream constraints—changing minute by minute. Relying on a single operator’s experience to juggle all of them inevitably leaves blind spots. 

Manual trending, even on multiple screens, cannot detect the faint correlations that warn of rising tube-metal temperature or the slow drift toward inefficient firing. Human-only oversight often misses weak signals hidden inside normal noise, allowing fouling, coking, and metal creep to progress unnoticed until alarms trip.

Consistency presents another constraint. Each shift interprets data differently, so setpoints wander and tribal knowledge gets lost when experienced staff retire. Documented handover procedures help, but they cannot match the persistence of a digital memory.

Integrating operator expertise with AI-driven optimization offers a practical path forward. Advanced models translate historian and lab data into recommended setpoints, display confidence levels, and log every adjustment, creating a shared reference for all crews. Operators remain in charge yet gain a data-rich co-pilot that never tires.

This approach builds on operator wisdom by learning from historical actions, updating setpoints within safe limits, and recording outcomes so crews can refine future decisions. The result is steadier operation, fewer surprise shutdowns, and a clear framework for transferring knowledge across generations of staff.

Continuous Optimization: The Smarter Way Forward

You extend heater life when you treat every data point as an early signal, not a post-mortem clue. Traditional run-to-failure cycles—over-fire, decoke, restart—waste fuel and expose tubes to repeated thermal shock. Continuous optimization flips that script by tracking skin temperatures, stack O₂, and per-pass flows in real time, then adjusting firing and feed limits before hot spots emerge.

With this approach, you can predict fouling days in advance, keep pass flows balanced, and coordinate targets across upstream and downstream units. Fewer shutdowns and lower fuel intensity quickly translate to higher margins and longer tube campaigns.

Ready to see similar results? Get a Complimentary Plant AIO Assessment to uncover what continuous optimization could deliver for your heaters. Modern AI optimization solutions learn from your plant data, automate safe control moves, and keep improvements compounding—setting the stage for the next generation of high-reliability refining.