A crude switch on paper looks like a scheduling decision. In a crude distillation unit running near its naphthenic acid corrosion limit, it is closer to a metallurgy decision made by whoever wrote the blend plan, and most blend plans are written without a live corrosion model in the loop. TAN alone does not predict where a unit will corrode — a high-TAN crude at low velocity through a resistant alloy section can run for years, while a moderate-TAN crude hitting a carbon steel elbow at high velocity and 550°F can pit through in months. See how AI-predicted NAC risk changes your next blend decision before the next crude switch finds a weak point.
TAN Alone Doesn't Predict Corrosion. Four Variables Together Do.
Naphthenic acid corrosion depends on TAN, temperature, velocity, and sulfur compound ratio acting together. AI models trained on all four variables show exactly where a crude blend will attack before it's running through the unit.
The Four Variables That Actually Drive NAC
Naphthenic acid corrosion is a function of chemistry and physics working together at each point in the unit, not a single crude assay number applied uniformly across every piece of equipment.
Total Acid Number (TAN)
TAN sets the raw corrosivity potential of the crude, but the same TAN value produces very different corrosion rates depending on the acid species distribution within it.
Temperature
NAC activity is negligible below roughly 425°F, accelerates sharply between 450-750°F, and drops off again above 750°F as acid species begin to decompose.
Flow Velocity
High-velocity zones — elbows, tees, and areas of turbulence — strip away protective sulfide films faster than they can reform, concentrating corrosion at specific geometric points.
Sulfur Compound Ratio
Sulfur compounds in the crude form a protective iron sulfide scale that can offset acid attack; the TAN-to-sulfur ratio is often a better corrosivity predictor than TAN alone.
TAN and Temperature Zones by Corrosion Risk
Corrosion rate does not scale linearly with TAN — the same crude produces very different rates depending on which temperature zone of the column it's passing through.
See Where Your Next Crude Blend Will Attack
iFactory maps TAN, temperature, velocity, and sulfur ratio across your CDU and vacuum column geometry to flag exactly which elbows and transfer lines carry elevated NAC risk for a given blend.
Turning Prediction Into Blending and Metallurgy Decisions
A NAC prediction only earns its keep when it changes a decision — either the blend going into the unit or the metallurgy standing in its path.
Blend Optimization
Blending high-TAN and low-TAN crudes to hit a target corrosivity ceiling for the unit's current metallurgy, rather than relying on a single blended TAN average that hides local hot spots.
Metallurgy Prioritization
Ranking which carbon steel components sit in the highest-predicted-corrosion zones for the next several planned crude runs, focusing upgrade budget on the segments that actually need it.
Inspection Targeting
Directing UT thickness surveys toward the elbows and transfer lines the model flags as highest risk for the current blend, instead of a uniform inspection interval across the whole circuit.
Crude Purchasing Input
Feeding predicted corrosion cost into crude economics evaluation, so a cheaper high-TAN crude's true cost includes the metallurgy and inspection burden it creates.
Assay-Based Estimates vs. AI-Predicted Corrosion Maps
A Blend Change That Would Have Passed Unnoticed
A refinery shifted its crude slate to include a higher proportion of a moderate-TAN, low-sulfur opportunity crude — a blend that looked acceptable against the unit's overall TAN limit on paper. The corrosion model flagged a specific vacuum column transfer line elbow where the combination of local velocity and the reduced sulfide-film protection from the lower sulfur content pushed the predicted corrosion rate well above the segment's historical range, even though the blended TAN stayed within the normal operating envelope. A UT scan confirmed early thinning at exactly that elbow. The unit adjusted the blend ratio and scheduled a metallurgy upgrade for that specific fitting during the next planned turnaround, avoiding what the model projected would have become a leak within the following two crude cycles at the prior blend ratio.
What Changes With Continuous NAC Prediction
Getting a NAC Model Running on Your Unit
Phase 1 — Geometry and metallurgy mapping
Unit piping isometrics, existing metallurgy records, and prior CML/UT history are compiled to build the physical model the corrosion prediction runs against.
Phase 2 — Crude assay integration
TAN, sulfur content, and acid species distribution data from crude assays feed into the model as blend ratios shift, connecting purchasing decisions to corrosion prediction.
Phase 3 — Process data connection
Live temperature and flow data from the unit's control system lets the model recalculate predicted corrosion rates as operating conditions change, not just when a new crude arrives.
Phase 4 — Inspection and blending workflow integration
Flagged high-risk segments route into the inspection planning system and blend decisions get a corrosion-cost input, closing the loop between prediction and action.
Common NAC Management Mistakes
Relying on Blended TAN Alone
A unit-wide average TAN can look acceptable while a specific high-velocity elbow experiences a corrosion rate several times the systemwide average.
Ignoring the Sulfur Ratio
Two crudes with identical TAN can produce very different corrosion rates depending on how much protective sulfide film their sulfur content supports.
Static Metallurgy Assumptions
Metallurgy selected for one crude slate may no longer be adequate once the slate shifts toward higher-TAN opportunity crudes.
Inspection Intervals Set by Calendar, Not Risk
Fixed inspection schedules miss the fact that corrosion risk changes every time the crude blend changes, sometimes within the same turnaround cycle.
Frequently Asked Questions
Does this replace crude assay testing?
No — crude assay data including TAN and sulfur content remains the core input the model uses. What changes is how that data gets applied: instead of a single blended TAN number checked against a unit limit, the model calculates point-by-point predicted corrosion rates across the unit's actual piping geometry. Talk to a specialist about connecting your assay data into the model.
How does the model account for velocity effects?
Piping geometry data including elbow angles, tees, and known turbulence points is mapped against the unit's flow model, so the corrosion prediction weights those high-velocity locations differently than a straight run of pipe at the same temperature and TAN exposure.
Can this help with crude purchasing decisions?
Yes — predicted corrosion cost for a candidate crude or blend ratio can be factored into the broader economic evaluation alongside price and yield, giving purchasing a clearer picture of a discounted high-TAN crude's true cost to the unit's metallurgy and inspection budget.
How often does the corrosion prediction update?
The model recalculates continuously as live temperature and flow data changes and whenever a new crude assay or blend ratio is entered, rather than waiting for a periodic manual review cycle to reflect a slate change.
Which units benefit most from this kind of monitoring?
Crude distillation and vacuum distillation units processing variable or opportunity crude slates see the most value, since blend variability is exactly what makes a static, assay-only corrosion estimate unreliable. Book a demo to see how the model maps to your specific unit configuration.
Know Where Your Next Crude Slate Will Attack
Book a 30-minute assessment. iFactory builds a corrosion risk map of your CDU and vacuum column against your current and candidate crude blends.
