Predictive Maintenance

Hi ​@TIM,

You can take the following 3 step approach to handel the optimal heat exchanger cleaning.

Quantify efficiency. Build a normalized KPI rather than watching raw outlet temperature. A practical proxy is steam consumption per unit throughput, or product outlet temp normalized by steam-valve position. If you want a real heat duty / U-value, that needs thermodynamics there's a ready CoolProp heat-exchanger example in the Custom Calculations repo (note: Custom Calculations is a LAB feature and needs the Expert user license).

Decide when to clean. Trend that KPI over a wide context bar (1–3 years). The cleaning events show up directly in the trace as step-recoveries — the KPI sawtooths down as fouling builds and jumps back up after each clean — so you can read your current cadence straight off the time series without any extra bookkeeping. Set the "fouled" threshold from where the curve flattens into diminishing returns, then VBS for periods past that threshold with a long duration cut (>3 days) to filter noise.

Detect fouling early. Promote that VBS to a Monitor on an early threshold so you're warned before throughput drops, and/or use Diagnose → Cross-correlations to confirm the fouling driver vs. ambient. One note: if you want a multi-day rolling baseline, Aggregation tags cap at 24 h, anything longer is Custom Calculations territory.

Kind regards
Frederik