A straight answer to a question the industry should have solved a long time ago
Written by Travis Riley, chiller mechanic and co-founder of Chiller Trend.
Every chiller log already has the ingredients: temperatures, pressures, amps, volts, setpoints, flow estimates, and operating notes.
The formulas are not new either. Tons, kW per ton, approach, lift, heat balance, pressure drop to flow, and basic electrical calculations have been around for a long time.
So why do most chiller logs still stop at readings instead of turning those readings into a clear diagnostic picture?
Because the industry never really built the workflow layer.
That is the answer.
The math was there. The machines were there. The need was there.
But from what I have seen in the industry, applying the basic calculation layer is treated like an optional step.
That is the part that never made sense to me.
If the readings are already being taken, and the formulas already exist, why are we still leaving so much of the interpretation inside one experienced mechanic’s head?
That gap is where the industry has been stuck.
The Formulas Were Not the Missing Piece
The formulas give us the numbers.
Judgment comes from knowing what to compare those numbers to.
That is where the workflow matters.
A formula by itself does not create a defensible conclusion. It only gives you a calculated value.
The real work is knowing whether the input was trustworthy, whether the condition was comparable, whether the number changed from baseline, and whether the rest of the machine supports the same story.
Routine readings can support a full calculation layer. The hard part is not the math. The hard part is applying it consistently and knowing what the result should be compared against.
That is the part the industry kept leaving inside senior mechanics’ heads.
So the industry did what it always does when the system is weak.
It leaned on experienced people.
Senior mechanics became the database.
Intuition became the analytics engine.
Old notes became the history.
Memory became the comparison layer.
That works better than it should.
Until it does not.
The Industry Built Around Getting the Work Done
The HVAC industry got very good at service execution.
Get the machine running. Solve the immediate problem. Get through the visit. Call the chiller guy when it gets weird.
That is not the same thing as structured diagnostic intelligence.
Structured diagnostic intelligence means:
- standardize the inputs
- verify the measurements
- compare against a real reference
- preserve history
- make the next decision easier to defend
The industry got strong at the first one.
It never really built the second one.
Logging Got Treated Like Paperwork
This is a huge piece of it.
Logging chillers should be performance verification.
Instead, a lot of the time it got treated like:
- paperwork
- proof the visit happened
- a few readings on a sheet
- maybe a spreadsheet if somebody cared enough
That is not a diagnostic workflow.
A real workflow has to do more than collect numbers. It has to:
- measure
- verify
- interpret
- compare
- support judgment
That is the layer that stayed weak.
The Data Is Messy, and Most People Backed Away From That Problem
The math was never the hardest part.
The hard part was dealing with real field data.
Chiller data is messy. Inputs are inconsistent. Sensors drift. Flow is hard to pin down. Design data is missing. Operating conditions change. Instruments vary. Readings are partial. Different mechanics bring different levels of confidence.
The industry looked at that mess and mostly landed in the same place:
the data is too noisy to trust
So instead of building systems that manage uncertainty, score confidence, compare trends, and apply guardrails, most of the industry stayed safer with raw readings, charts, and human interpretation.
That is a huge reason this stayed open.
Most Performance Data Was Not Really Comparable
This is one of the biggest problems in the whole industry.
Even when people report kW per ton, the readings vary, the methods vary, the assumptions vary, and the calculations vary.
So the output looks precise, but it is not truly comparable.
A lot of the field is not comparing performance.
It is comparing methods.
That is a completely different thing.
Once the method changes, the comparison gets weak fast.
That is why controlled inputs matter so much.
If the inputs are not controlled, the conclusion is weaker than it looks.
The OEM Path Went Somewhere Else
The major OEM path was not built around the normal field-service reality.
It was built around connected equipment, controls, and live monitoring:
- permanent sensors
- permanent instrumentation
- cloud monitoring
- BAS integration
- continuous data streams
That works in higher-budget, highly integrated places.
But it skipped a huge part of the real market where chillers are still inspected periodically, BAS access is inconsistent, instrumentation quality varies, and no permanent monitoring exists.
That middle layer never got a clean diagnostic workflow.
The Trade Got Comfortable With Experienced Guessing
This part is real, whether people like it or not.
For years, the trade normalized smart people making good calls without being able to fully quantify or trend what they were seeing.
Especially around:
- flow
- load
- tower performance
- sensor drift
- expected efficiency
- what feels normal and what feels off
A senior mechanic could know something was wrong without being able to prove it cleanly on paper.
That became normal.
The problem is, once that judgment mostly lives inside a few experienced people, it gets hard to scale, hard to teach, hard to trend, and hard to defend to a customer.
So the industry did not really solve continuity.
It just leaned harder on experience.
Nobody Really Owned the Missing Layer
This may be the biggest reason of all.
Historically:
- BAS companies owned controls
- OEMs owned machines
- contractors owned service
- TAB firms owned verification
- commissioning firms owned optimization
But nobody really owned this layer:
routine field diagnostic interpretation
That part got left with senior mechanics, spreadsheets, notebooks, and memory.
One of the most valuable layers in the whole workflow stayed mostly human and mostly unstructured.
That is why it never scaled well.
The Workflow Was Not Rewarded
This part is uncomfortable, but it is real.
A lot of the industry was structured around:
- breakdowns
- quoted repairs
- emergency calls
- reactive service
Not around:
- standardized diagnostics
- repeatable trending
- quantified early detection
- clean continuity
So there was less pressure than people think to build a workflow that catches problems early, standardizes reasoning, and makes judgment more repeatable.
That does not mean people wanted bad outcomes.
It means this layer was never rewarded hard enough to get built properly.
The Software People Usually Missed the Field Reality
The people building software usually did not understand the chiller nuance.
The people who understood the chiller nuance usually were not building software.
That gap mattered a lot.
Most software teams did not really understand:
- why flow uncertainty matters
- why sensor drift matters
- why operating conditions can wreck a comparison
- why one clean-looking number can still be weak
- why field reality has to be managed, not ignored
Meanwhile, the mechanics who did understand all that were busy keeping machines running.
So the split stayed in place.
The software side often lacked field context.
The field side usually lacked a usable software layer.
The Missing Piece Was the Packaging
This is the piece people miss.
The formulas were not missing.
kW per ton existed.
Approach temperatures existed.
Flow relationships existed.
Heat balance existed.
Lift existed.
None of that was the breakthrough.
What was missing was the packaging.
The repeatable workflow.
The guardrails.
The comparison logic.
The confidence handling.
The operational memory.
The usable output.
That is why this stayed stuck for so long.
Not because the engineering did not exist.
Because nobody successfully turned that engineering into something the field could use over and over without pretending certainty.
Why It Matters More Now
Now the pressure is different.
The industry has:
- shrinking senior talent
- higher energy costs
- more demand for proof
- more complex equipment
- weaker continuity across teams
- more pressure to explain recommendations clearly
Now the old system is showing its cracks.
The model of “just trust the experienced guy” gets harder to hold together when the experienced guy is overloaded, retiring, or spread too thin.
That is why this problem feels a lot more urgent now than it did ten or fifteen years ago.
Bottom Line
I am not saying the industry does not have smart people.
It does.
That is actually the point.
Too much of the reasoning still lives inside individuals instead of a repeatable workflow.
The industry did not fail because it lacked the math.
It failed because it never successfully packaged the math into a scalable field workflow.
That is why so much chiller diagnostic reasoning still lives in senior mechanics’ heads.
And that is why the gap matters now more than ever.
Once experienced people become the system, the whole system gets harder to scale, harder to teach, and harder to trust consistently.
The formulas were already there. What was missing was a workflow the field could actually use.