INSUS learns each exchanger's clean-operation performance from existing temperature, flow and crude property data, and flags fouling onset when actual heat transfer drifts from that baseline. Process engineers decide when to clean.
01
Fouling flagged against each exchanger's clean baseline.
02
Complements ε-NTU monitoring. Does not replace it.
03
Existing temperature, flow and crude data. No new sensors.
The problem
A crude distillation unit preheat train recovers heat from hot product streams before the crude reaches the furnace. As exchangers foul, heat recovery falls and the furnace compensates with additional fuel. Heat transfer also moves with inlet temperatures, flow rates and the crude slate, so a performance drop is not automatically fouling. In a train of exchangers in series, fouling in one unit may not be isolated until its effect on the whole train is already significant.
The shift
ε-NTU monitoring can stay in place, and cleaning decisions stay with engineers. What changes is that each exchanger's expected clean performance is learned from its own data, so developing fouling shows up as a deviation earlier.
Today
With INSUS
How it works
This fits OptiStudio as process-monitoring decision support above the DCS and historian. Access is read-only. It never triggers cleaning. It complements ε-NTU monitoring rather than replacing it.
Connect
Connect read-only to DCS and historian data for inlet and outlet temperatures and mass flow rates on each exchanger, specific heat capacity, crude oil property data from LIMS, preheat train configuration, and cleaning and maintenance history.
Understand
Using a period of confirmed clean operation, such as the weeks after a cleaning, the model learns how each exchanger's heat-transfer performance relates to its inlet conditions and the crude being processed.
Predict
Predict the heat-transfer performance each exchanger should achieve under current inlet conditions if it were still clean, using a model trained on the plant's own clean-operation history.
Detect
When actual performance deviates from the predicted clean-operation performance in a way consistent with fouling, raise a flag on that exchanger and add cleaning prioritisation guidance. Process engineers review each flag before acting on it.
Verify
Compare detection flags with confirmed cleaning and maintenance findings, and compare detection timing with existing ε-NTU-based monitoring. Each exchanger's model is refined against its own operating history.
Getting started
For crude preheat trains where fouling raises furnace fuel before the fouling unit is isolated. Delivered through OptiStudio as process-monitoring decision support. Expert Services can help with data assessment and validation. Advisory and read-only. Complements ε-NTU; does not replace it.
Related for industrial operations
Heat exchanger fouling detection sits beside compressor predictive maintenance and Seveso III compliance on the same industrial-operations portfolio.
Preheat-train fouling raises furnace fuel, but train-level monitoring often isolates the fouling exchanger late. This use case flags fouling against each exchanger's learned clean baseline for process engineer review.
Historical inlet and outlet temperature, flow rate and crude property data, ideally including a confirmed clean-operation period, together with cleaning and maintenance history. No new sensors when a modern preheat train already generates that data.
No. It complements ε-NTU monitoring with an earlier, data-driven signal. The process engineering team reviews every flag. It does not trigger cleaning or maintenance.
Attribution to a specific exchanger depends on that exchanger having its own temperature and flow measurements. Where those exist, the system compares each unit against its own clean baseline.
Through OptiStudio as process-monitoring decision support. Expert Services can support data assessment, integration and validation with your process team.
Peer-reviewed work has shown that data-driven models can be more sensitive to fouling onset than ε-NTU on studied preheat trains. Those findings belong to those plants, not to every train, and not to us. Value is measured against your own cleaning findings and fuel baseline.
It needs a usable clean-operation baseline that covers the crudes the unit runs. Models are calibrated per exchanger and do not transfer unchanged. Some model families extrapolate poorly to unfamiliar conditions.
Industrial Operations · Heat recovery
A flag view on one exchanger or train segment, run beside your ε-NTU monitoring, will show whether a learned clean baseline spots fouling before the furnace has been making up lost heat recovery.
