Freeze Protection: The Winter BAS Checklist That Prevents Cracked Coils
- Alex Khachaturian
- Oct 27
- 7 min read
Promise: A field-tested, entertaining crash course on freeze protection, what it is, why it fails, where to look, and exactly how to test sequences (AHUs, coils, hydronic, snowmelt).
TL;DR
Freeze protection is a chain: sensors → sequences → hardware fail-safes → trending/alerts. Verify every link.
Test low-limit trips and recirc/valve logic before the first cold snap; confirm glycol mix and slab/supply sensors for snowmelt.
Trend mixed air, coil face, OA temp, slab temp, and valve/relay states. Catch drift and “always-on” snowmelt waste early.
Key Takeaways
Most winter failures start as assumptions: mislabeled sensors, wrong fail positions, or wishful sequences never fully tested cold.
Layered protection wins: BAS logic + physical freeze stats + mechanical fail-safe positions.
Snowmelt should be demand-driven (slab temp, ambient/wind, moisture) with run-down logic, not a seasonal “ON.”
Write and run repeatable tests you can rerun in 10 minutes all season.
Document setpoints/logic in the BAS, future you (or the next tech) will thank you.
Quick Links
Next Read
Story: Aspen, a Sultan’s Hotel, and My First Snowmelt
I grew up a California tech. “Winter” was a hoodie. Then I landed my first hotel job in Aspen, Colorado, the kind of place where foreign sultans and heirs hide from the world and expect things to “just work.” Nothing exploded. But it was my first real exposure to snowmelt and freeze protection.
My peers walked me through why coils crack, why capillary freeze stats must snake across the face, and why “we’ll test it later” is a sentence that creates $30,000 problems.
Now I’m on the East Coast, where freeze protection is standard, and still easy to miss. This post is the checklist I wish I had day one in Aspen, so none of us get stuck holding the bill for a failed sensor and a frozen coil.
Quick Win: 20-Minute Pre-Freeze Validation
Confirm points exist: OA temp, MA temp, coil face/approx proxy, preheat valve %, supply temp, fan status, freezestat trip, economizer position, hydronic pump status, snowmelt slab temp (if present).
Simulate cold: Temporarily bias OA temp down 5–10°F (software override) and/or close OA damper to test mixed air clamp.
Trip test: Momentarily trip the freeze stat (with permission) to prove fan off, OA full close, valve full open, pump run, alarm.
Trend sanity: Start 2-hour trend for OA/MA, coil valve, fan status, freezestat, slab temp (if present).
Snapshot: Export a one-page “Freeze Baseline” screenshot, store in the job’s winter folder.
This alone fixes 50% of “we thought it worked” failures.
Playbook: Test & Implement Freeze Protection (All Systems)
1) Air-Handling Units (Preheat/Coil Freeze Protection)
What to verify (logic):
Low-limit trip sequence: On trip → Supply fan OFF, OA damper CLOSED, return damper OPEN, preheat valve 100%, hydronic pump ON, alarm.
Mixed air low-limit clamp: If MA temp < setpoint (e.g., 38–42°F) → OA damper modulates closed regardless of economizer call.
Start-up warm-up: On morning warmup in sub-freezing OA, preheat opens and OA remains minimum/closed until MA safe.
Hardware & wiring:
Freeze stat type/location: Capillary serpentine across coil face, downstream of OA, upstream of heating coil. Avoid dead corners.
Fail positions: Heating valve fail-open; OA damper fail-closed; ensure actuator springs tested annually.
Prove points: Wire freeze stat auxiliary to BAS for trip trending; don’t hide a hardwired trip from history.
Test it:
Command OA damper closed, bias MA setpoint to provoke heat.
Trip freezestat briefly (mechanically or via test switch) and verify every action above happens within <3 seconds.
Reset, then verify auto-recovery sequence (manual reset often required by design, confirm signage).
Setpoints that work:
MA clamp: 40°F typical (tight buildings), 45°F for leaky OA paths.
Freeze stat trip: 36°F to 38°F; match manufacturer guidance and local risk.
2) Hydronic Coils & Loops (Glycol, Pumps, Bypass)
What to verify:
Glycol concentration (propylene glycol): Target burst protection well below historical lows; typical 25–40%, but follow coil/pump specs.
Recirculation on trip: When any zone/coil freeze risk, enable loop pump and drive 3-way to bypass/coil protection.
Minimum flow: Ensure pump VFDs have low-temp lockouts to maintain min GPM in severe cold.
Hardware & fail-safe:
Valve position: Heat coil fail-open; confirm spring return every fall.
Strainers: Clean; clogged strainers mimic freeze when flow collapses.
Differential pressure: Trend DP across coil at design and low-flow; alarms if DP collapses in cold hours.
Test it:
Simulate outdoor low limit; verify pump start, bypass/coil open, alarm.
Pull a glycol sample and test with refractometer; record % in BAS notes.
3) Economizer / Mixed-Air Boxes
What to verify:
Damper leakage: With OA commanded 0%, check blade seals; smoke test or measure MA vs RA in bitter cold.
Sensor calibration: OA/MA/RA sensors within ±0.5–1.0°F; bad MA sensors defeat clamps.
Priority: Freeze clamp overrides economizer, not the other way around.
Test it:
Command 100% OA, then drop OA setpoint below clamp, verify the BAS forces OA closed until MA safe.
4) Heat Trace, Unit Heaters, Vestibules
What to verify:
Proof of heat trace: CTs or contact inputs; alarm on command without proof.
Vestibule heaters: Link vestibule temp to OA lockout for nearby AHUs; cold air dumps can trip AHU freeze logic.
5) Snowmelt Systems
(Driveways, Ramps, Walks)
Control goals: Keep slab just warm enough to prevent bonded ice, not to heat the neighborhood.
What to verify:
Sensors: At least slab temperature sensor; ideally moisture sensor too. Ambient alone = wasted energy.
Enable logic: Enable when ambient < X°F AND slab wet/precip OR snow forecast, then hold ΔT above freezing.
Run-down: After snowfall stops, tail-off (e.g., 30–90 min) to clear residual moisture, then lockout.
Setpoints that work (starting points):
Slab target: 38–42°F when wet & <32°F ambient; adjust for wind.
Ambient enable: 34–36°F with wet present; higher cut-in for high wind sites.
Test it:
Force “wet” input (or test mode), drop ambient below cut-in, verify valves/boilers/pumps start and slab warms.
Confirm lockout when ambient rises and slab dry, no seasonal “ON.”
Implementation if missing:
Add a slab sensor and a simple logic block: (Ambient < Cut-in) AND (Moisture = WET) → Enable with ΔT hold and post-run timer.
6) Trending & Alarming (Your Early-Warning Radar)
Trend these (5-min interval, winter season): OA, MA, RA temps; preheat valve %; freezestat state; pump status; coil DP; economizer position; slab temp & wet; heat trace proof.
Alarms that matter:
MA < 40°F for >2 min (fan on)
Preheat 100% & Supply < setpoint (possible coil flow issue)
Freeze stat trip (with pre-/post-context screenshot)
Snowmelt ON without wet proof (>30 min)
Heat trace commanded ON without proof
Troubleshooting: Symptom → Cause → Fix
Symptom: “We cracked a coil last night.”
Cause: No/failed freeze stat, wrong valve fail, economizer leakage, low/no flow.
Fix: Add/replace freeze stat (serpentine), set heat valve fail-open, seal OA dampers, set MA clamp 40–45°F, verify pump min-flow, and trend alarms.
Symptom: “We keep false-tripping on freeze.”
Cause: Capillary on coil edge, turbulent eddies, sensor noise.
Fix: Re-route capillary in zig-zag across coil face, add 2–5 sec input filter, and shield from direct discharge.
Symptom: “Economizer caused a midnight freeze scare.”
Cause: MA sensor drift; economizer priority over clamp.
Fix: Calibrate OA/MA sensors, enforce freeze clamp priority over economizer, and test with OA @ 100%, then invoke clamp.
Symptom: “Hydronic loop slushed up.”
Cause: Weak glycol or added plain water during service.
Fix: Test with refractometer, restore %, lock makeup to premix only, and label drums & valves.
Symptom: “Snowmelt never shuts off.”
Cause: Ambient-only control or wet sensor failed “on.”
Fix: Add/repair slab & wet sensors, use ΔT-based control, add max runtime and ambient high lockout.
Symptom: “Heat trace shows ON in BAS but pipe froze.”
Cause: No proof input, breaker tripped, failed run circuit.
Fix: Add current/aux proof, alarm on command without proof, and perform annual megger/visual inspections.
Symptom: “No one can tell if freeze ever tripped.”
Cause: Hardwired trip with no BAS feedback.
Fix: Wire aux contacts to BAS; trend preheat %, MA, fan, and OA positions before/after events.
FAQ
What’s the difference between a low-limit freeze stat and a BAS low-limit?
Hardware freeze stats cut the fan immediately and don’t care about programming. BAS low-limits provide anticipatory control (clamps, preheat) and history. You want both.
Should every heating coil fail open?
For freeze-risk coils, yes, fail open is standard. Confirm with the engineer; verify actuator springs every fall.
What glycol % should I use?
Follow the coil/boiler documentation. Typical propylene glycol is 25–40%. Set the burst margin below site record lows; don’t guess, measure.
Where exactly do I put a capillary freeze stat?
Serpentine across the leaving air face of the coil; avoid dead zones. Secure with stainless clips, not duct tape.
What’s the fastest way to prove snowmelt works?
Force wet, drop ambient in test, watch valves/pumps start, and verify slab temp climbs 2–5°F within 15–30 min (site-dependent).
How do I stop nuisance economizer freeze alarms?
Seal/leak test OA blades, re-zero MA sensor, and enforce freeze clamp overrides economizer logic.
Do I need heat trace proof?
If it can freeze and you can command it, you should prove it. Add CT/aux and alarm on mismatch.
Field Checklist
Trend pack started (OA/MA/RA, preheat %, fan, freezestat, coil DP, econ %, slab temp/wet, heat trace proof).
Freeze stat location & wiring verified; aux to BAS present.
Valve fail positions confirmed (heat fail-open); actuator springs tested.
MA clamp enabled and priority over economizer tested.
Glycol % measured & logged; premix makeup only enforced.
Hydronic min-flow verified at VFD.
Snowmelt has slab + wet sensing; ΔT logic & rundown.
Heat trace proof input present & alarming.
One-page Freeze Baseline screenshot saved to winter folder.
Results & ROI
Avoided coil replacement: $15k–$30k per coil (plus downtime).
Energy savings: Proper snowmelt control commonly saves 10–40% vs ambient-only “ON.”
Fewer midnight calls: Trend + alarms catch drift days before failure.
Documentation dividend: The baseline screenshot ends “it never happened” debates.
Wrap-Up
One action this week: Run the 20-minute pre-freeze validation on your riskiest AHU. Save the baseline screenshot.
Personal note: Aspen taught me that winter protection isn’t dramatic, it’s discipline. Do the boring things early, and winter becomes just another season.
Comments