The Open Loop: Why Fault Detection Alone Can't Fix Your Building

Fault detection software can be highly effective at identifying issues inside your building, but detection without diagnosis, prioritization, and correction simply produces a long list of work orders, not operational results.

In a 2023 Lawrence Berkeley National Laboratory (LBNL) study analyzing over 65,000 pieces of commercial HVAC equipment–all monitored by enterprise fault detection software–40% of air handling units reported an active fault on any given day.  

Think about what that actually means. Every building in that study had already bought the exact solution the industry recommends to help improve operational performance and energy efficiency. The fault detection software was installed, configured, and running. The faults stayed anyway.  

40% of Equipment is Faulty Right Now

The data comes from a massive study led by LBNL researchers across 317 commercial buildings, tracking 3,660 air handlers, 53,865 terminal units, and 7,974 rooftop units.  

The key findings:

• Daily Failure Rates: On any given day, 40% of air handlers and 30% of terminal units had at least one active, reported fault.  

• Alert Fatigue: The average building generated 245 fault alerts every single month.  

• Chronic Drift: 18 distinct air-handler fault types persisted for more than 20% of the entire year.  

  
  

It's likely that a small percentage of these alerts are false positives or configuration noise, but the overarching pattern is undeniable: HVAC faults are not isolated glitches; they are a permanent state of operation.  

The Dashboard Fallacy: Why Alerts Don't Equal Results

Why do faults sit active for months in buildings that already pay for fault detection software?  

Because detection produces work, and facilities teams are already drowning.  

In the LBNL dataset, a single office building logged 98,234 daily fault records in a single year. A hospital in the same study logged 443,308. No engineering team on earth has the headcount to manually triage, investigate, and fix that volume of alerts.  So what happens? The alerts age, the software dashboard becomes background noise, and the building keeps wasting energy.  

This isn't a failure of the operators and it isn't a failure of the detection software. The operators are focused on the most pressing priorities. The detection tool did its job. The problem is structural: Detection was sold as a complete product, when it's actually just step 1 of a 4-step loop. These vendors handed operators a list of problems and left the remaining three steps as "an exercise for the customer."  

The Cost of the Open Loop

Leaving the loop open bleeds cash:

1. The Human Bottleneck: The LBNL Smart Energy Analytics Campaign showed that traditional fault detection tools yield a median energy saving of 9%. That’s decent, but it means the average building is still leaving massive operational savings on the table simply because human maintenance teams can't clear the queue.  

   
   

2. Operational Decay: Things don't stay fixed. Texas A&M research (Toole & Claridge) proves that manual tune-up savings decay at 6% to 8% per year due to operational drift. A setpoint fixed manually to start the year has drifted back into waste by fall.  

   
   

What Closing the Loop Actually Looks Like

Real, permanent building optimization requires four continuous steps:  

1. Detect: Monitor raw BAS data continuously at fine intervals. (The industry commoditized this step—now we move past it.)  

2. Diagnose & Prioritize: Separate root causes from symptoms, kill false positives, and rank real issues by dollar impact. If an engineer only has two hours this week, they need to know which two faults cost the most money.  

3. Correct: Write the fix directly back to the Building Automation System (BAS). Most reported faults aren't broken hardware—they are setpoint overrides, fighting loops, and schedule errors. The fix is software parameters, not replacement parts.  

4. Verify & Hold: Prove the savings against a weather-normalized baseline, and continuously apply correction logic to stop drift forever.  

   
   

This is the fundamental difference between Analytics (looking at charts) and Optimization (fixing the building).  

When you run continuous, closed-loop correction, whole-building energy reductions don't stop at 9%—they jump to 13%–21%. (See: Real World Outcomes).

As owners and operators face increasing pressure to improve the operational and energy performance of their buildings, more and more vendors are presenting their solution as the answer. It's more important than ever that operators are clear on what these solutions can and can't deliver, and whether or not they're providing a net positive return to the building.

Frequently Asked Questions

What is the difference between Fault Detection (FDD) and BTune?

Traditional FDD alerts you when something goes wrong and adds a work order to your queue. BTune's automated FDD goes far beyond, reading your BAS data, prioritizing faults, and streamlining the correction process for contractors or automatically writing corrective logic back to your BMS to fix control-layer issues in real time, without adding work to your facilities team.  

Does automatic fault correction risk tenant comfort or safety?

No. BTune operates within strict, operator-approved boundaries and standard BAS interfaces. Operator overrides always take precedence, every action is logged in an audit trail, and building teams can pause or modify strategies at any time.  

Are most reported HVAC faults physical equipment failures?

No. The vast majority are software and control issues: overridden schedules, fighting heating/cooling loops, and drifted setpoints. They waste massive amounts of energy while the physical equipment passes a visual inspection.