Fouled Cooling Coils Across a University Campus

How continuous heat balance monitoring detected coil fouling in 12 of 34 AHUs, saving an estimated $180k/year in wasted energy.

The problem

A 34-AHU university campus was experiencing inconsistent cooling complaints. Some buildings couldn’t maintain setpoint during peak afternoons. The facilities team had been chasing complaints reactively, replacing coils only after total failure.

What we did

We deployed edge devices on the campus BACnet network and connected to all 34 AHUs. Over a 3-month pilot:

  1. Built the plant model. Ingested the campus P&IDs, Metasys point export (2,400+ points), and equipment schedules. Our tooling produced a typed asset graph with every AHU, its zones, its coils, and their connections.

  2. Ran continuous heat balance. For each AHU, we computed the heat balance residual: the difference between the energy entering the coil (chilled water side) and the energy leaving (air side), accounting for steady-state conditions. A clean coil has a near-zero residual. A fouled coil shows a persistent positive residual — the water side is delivering energy the air side isn’t picking up, because the heat transfer surface is degraded.

  3. Identified fouled units. 12 of 34 AHUs showed heat balance residuals exceeding the threshold during steady-state cooling operation. The residuals correlated with coil age and service history.

Findings

  • 12 AHUs with measurable coil fouling (35% of fleet).
  • 4 severe cases where fouling reduced effective cooling capacity by an estimated 20–30%.
  • Estimated energy waste: $180k/year across the fouled units (excess chiller energy from reduced heat transfer efficiency + excess fan energy from increased air-side pressure drop).
  • Prioritized remediation list ranked by energy penalty, with payback period estimates for each coil cleaning or replacement.

Outcome

The facilities team cleaned the 4 worst coils within 6 weeks. Post-cleaning heat balance residuals dropped to baseline. The university approved campus-wide continuous monitoring on a retainer basis.

Technical notes

  • Steady-state detection was critical. Heat balance is only meaningful when airflow and temperatures have stabilized. We used a variance-ratio method with a 15-minute qualification window.
  • BACnet polling at 1-minute intervals was sufficient. Higher frequency added noise without diagnostic value for coil fouling.
  • The P&ID ingestion took 2 days of effort. Subsequent campuses with similar Metasys setups took under 1 day, because the asset library had learned the naming patterns.