A real-world example of why field verification matters
Nobody knew there was a problem.
The chillers were running. The plant was operating. The BAS was online. Nothing was screaming for attention.
But during a baseline inspection, one extra reading changed the operating story of the plant.
A normal chiller log can tell you what the machine was doing at that moment.
A stronger workflow asks a better question:
Do the readings agree with the rest of the plant?
At one large central plant, the answer was no.
As part of the inspection workflow, independent ambient readings were taken near the cooling towers and cooling tower approach was calculated across the operating chillers.
The tower approach readings were not lining up the same across the plant.
Something did not look right.
That led to verifying the BAS ambient inputs against independent field measurements.
At the time of the field check, the BAS showed an outdoor-air wet-bulb value of 81.7°F. Two independent Testo probes both showed a wet-bulb temperature of 78.2°F.
That is a difference of about 3.5°F.
The BAS outdoor-air temperature and humidity values also did not agree with the independent field readings.
That matters because the BAS was using that environmental input to influence cooling tower control.
The control system appeared to be responding to the input it was being given.
The problem was that the BAS input did not agree with the independently measured field conditions.
One Extra Reading Changed the Story
This was not a failed chiller.
It was not a major alarm.
It was not an obvious mechanical problem.
It was a believable BAS value that did not match the independent field measurements.
That is how problems like this can hide.
The number is online. The value looks reasonable. The plant keeps running.
But believable is not the same thing as verified.
That distinction matters whenever a control decision depends on a sensor or calculated value. A precise-looking number can still be weak evidence if the source has not been checked.
That is the same issue covered in A Clean Number Is Not Always Believable .
When a BAS input is wrong, the plant can make poor control decisions quietly. The operator may not see a failure. The chiller may not trip. The BAS may not alarm.
In this case, the effect showed up through tower behavior, condenser water temperature, refrigerant lift, and chiller kW.
The Control Chain
The important part was not just that a sensor value was off.
The important part was what that value was connected to.
Bad ambient and humidity data affected the wet-bulb calculation.
Bad wet-bulb information affected tower control.
Tower control affected condenser water temperature.
Condenser water temperature affected chiller lift.
Lift affected chiller kW.
That is the chain.
Sometimes the most important chiller-related finding is not on the chiller itself.
It can be upstream from the machine, inside a control input that everybody assumes is close enough.
Why This Can Be Missed
A normal log captures readings.
That still matters.
But the value increases when those readings are checked against the rest of the system.
Do the readings agree with each other?
Do they agree with the BAS?
Do they agree with the field?
Do they explain the way the plant is operating?
Without the independent ambient reading, the BAS value may have continued to look believable.
Without calculating tower approach, the inconsistency across the plant may not have stood out.
Without preserving the readings for comparison, the issue may have stayed buried inside normal-looking plant operation.
The Result
After the control issue was identified, the condenser-water setpoint was reduced by 3°F.
During comparable operating conditions reviewed afterward, plant efficiency improved from approximately 0.560 kW/ton to approximately 0.506 kW/ton, including the effect of increased tower fan energy.
That is a reduction of approximately 0.054 kW/ton.
Using the original value as the baseline:
(0.560 - 0.506) ÷ 0.560 = 9.6%
That supports describing the observed change as roughly a 9–10% reduction in plant kW/ton during the comparable conditions reviewed.
Based on the available operating history and the savings model, the estimated annual savings range was approximately $50,000–$60,000 per year.
That is an estimate, not a guaranteed savings figure. Actual savings depend on runtime, load profile, weather, electric rate, and how consistently the improved operating condition is maintained.
The important point is that this was not a capital project.
It was a field-verified operational correction.
Why This Matters Beyond One Site
A BAS ambient-input error at one large central plant was enough to influence tower control and the conditions the chillers were operating against.
That should raise a larger question for owners and facility leaders:
Is this isolated, or is this happening somewhere else too?
Most large facilities already have plenty of data.
The harder problem is knowing which data can be trusted, which readings need to be verified, and which changes are actually affecting performance.
Those changes are easier to see when there is a useful baseline and comparable history to measure against.
The Real Lesson
This was not found because a chiller failed.
It was found because the inspection looked beyond the machine and checked whether the plant inputs made sense.
That is the useful progression:
What are the readings?
Do they agree with the field?
What changed?
What does the evidence support?
What should be checked or corrected next?
The value of the record is not just proving that someone was there.
It is preserving enough trustworthy evidence to make the next decision clearer.
Bottom Line
A bad BAS input can make a chiller plant operate less efficiently without creating an obvious alarm.
Because the number can still look believable, the problem can remain hidden inside otherwise normal plant operation.
That is why field verification, tower approach, and baseline comparison matter.
One extra reading exposed a control issue that was quietly affecting plant efficiency.
Sometimes that is all it takes to stop guessing and start seeing what the plant is actually doing.