Immersion Fluid Temp Creep from Undersized Heat Exchanger

Pull tank temperature trend from the BMS or controller for the last 30-90 days and confirm the climb. A 5 °C+ rise over weeks at constant load, ambient, and pump flow confirms creep — distinguishes from acute pump or fluid-level events. Free, takes five minutes, settles whether you're chasing creep or an acute fault. If trend is flat, you're on the wrong page; check pump and fluid-level pages instead.

Walk to the heat exchanger with a flashlight. Inspect the fin pack from 10 cm — dust mat, pollen, cottonwood fluff, oil film, biological growth, grease deposit between fins all degrade the air-side heat-transfer coefficient. Inspect every fan: visible rotation, tach reading if instrumented, anemometer at fan exit if you have one. Most HX-creep diagnoses end here, with a flashlight and ten minutes.

Clear gross obstructions on the intake side: snow drifts, pollen accumulation, cottonwood fluff, accumulated leaves, cardboard boxes leaned against the intake, parked vehicles, or growth from nearby vegetation. Clearance to maintain typical: 1-2 m around a vertical-discharge dry cooler. Free, fast, and often the entire fix on outdoor installs going through a seasonal change.

Install thermocouples at HX fluid-in, fluid-out, and ambient intake if not already present. K-type probes plus a $50 datalogger gives you every diagnostic on this page going forward. The four-number trendline (fluid-in, fluid-out, air-in, air-out) is the single most valuable diagnostic addition you can make to an immersion loop, and it costs less than one cooked hashboard.

Power off the dry cooler. Fin-comb or brush the air side with a soft-bristle brush from intake side, vacuum from discharge side. For oil / grease film, alkaline coil cleaner spray (HVAC-supply-house product, pH 11-12, foaming) followed by a low-pressure water rinse — never pressure-wash aluminium fin packs at high pressure, you'll flatten fins and destroy airflow worse than the fouling did.

Check every fan on the HX for rotation, RPM, and bearing health. Replace bearings or motors that are grinding, slow, stalled, or running below nameplate RPM. Most axial-fan motors on commercial dry coolers are commodity ECM units, replaceable in 30-60 minutes with hand tools. A stalled fan at 100 % commanded duty contributes zero CFM but reads green on the BMS — find them with a tach or anemometer, not the dashboard.

Verify intake clearance and check for exhaust recirculation. Hand-anemometer or thermometer at intake face — if intake air is more than 5 °C above outdoor ambient, you have a recirculation problem. Indoor and lean-to-enclosure installs are most prone. Re-route exhaust through-the-wall or add a Z-cabinet baffle to break the recirculation path. Plumbing-shop work, not HX work, but high-leverage.

Service the strainer / pre-filter on the fluid side if not done in the last quarter. A restricted strainer drops fluid flow at the HX without changing pump speed, and the symptom is identical to a fouled HX. Pull the strainer cup, clean or replace the element, restart and re-baseline the four temperatures.

Re-baseline the four temperatures after cleaning. Run the loop at design load for 2-4 hours, then log: HX fluid ΔT (target 8-12 °C), HX air ΔT (target 8-15 °C), approach temperature (HX outlet fluid minus inlet air, target 5-15 °C). Tank temp dropping 3-8 °C in 24-48 hours after cleaning confirms air-side fouling was the root cause — schedule recurring service every quarter going forward.

On a plate / brazed-plate HX in a fluid-to-water build, measure water-side flow, water-side inlet temp, water-side outlet temp. Healthy water-side ΔT roughly equals dielectric-side ΔT at the HX (energy balance). Collapsed water-side ΔT = water-side fouling — usually scale on chilled-water plates from untreated building water. Fluid-side fix won't help; problem is on the water side.

Acid-clean the water side of a brazed-plate HX with confirmed scaling. CIP (clean-in-place) chemistry — citric acid 5-10 % solution, circulated for 2-4 hours at 50-60 °C, neutralized with sodium hydroxide, flushed with clean water, and validated with a return-to-baseline ΔT measurement. Restores HX capacity to 90 %+ of new condition on most scaling jobs. Don't attempt without neutralization waste handling and PPE.

Sample the dielectric and send to the fluid vendor's lab. Karl Fischer water content, particulate count, acid number, and dielectric breakdown voltage. If lab returns out-of-spec on water content or particulate, the dielectric-side fouling diagnostic branch becomes live — pull the sample point downstream of the HX to see if particulate is bypassing the strainer.

Add airflow sensing — anemometer or hot-wire probe at fan exit, logged to the BMS or datalogger. Convert fan-duty status into actual CFM. Closes the dashboard-vs-reality gap that lets stalled fans hide. A $50 vane anemometer plus a $30 datalogger covers an 8-fan dry cooler permanently for under $300.

Replace failed fans with OEM-spec or upgraded ECM units. If three of eight fans on a dry cooler have failed in 18 months, the fan family is at end of life — replace all of them on a campaign basis, not one at a time. Plan a 1-3 week parts window on commercial axial fan motors.

Add evaporative pre-cooling for summer capacity. Wetted-pad or mister system upstream of the dry-cooler intake drops air temperature 5-10 °C on dry summer days, 2-5 °C on humid days. $200-$800 parts cost; hours-to-day to install. Single biggest summer-capacity win on Canadian and northern-US outdoor installs facing climate-creep ambient.

Right-size or supplement the HX. If the build is correctly plumbed, fans / fins are clean, water side is clean, and ambient is genuinely above design — the HX is undersized. Options: upsize to next-larger model with +25-50 % heat-reject capacity, install a second HX in parallel, or both. Engineering work; verify against vendor's published heat-rejection curve at your peak summer ambient.

Set up a fluid-temperature alarm at tank temp = chip throttle floor minus 5 °C. For Antminer S19-class hardware running stock firmware, throttle floor is around 60-70 °C chip junction, corresponding roughly to 50-55 °C tank fluid depending on plate / paste / mount thermal stack. Alarm at 45-50 °C tank fluid gives days of warning before throttle, not hours. Wire alarm to PDU shutoff or operator alert.

Stop DIY and book D-Central's immersion service when cleaning + fan replacement + flow verification haven't restored capacity, the build needs structural HX upsize or pre-cooler retrofit, plate-HX water side is scaled and you don't have CIP equipment, multiple top-bank hashboards show thermal-stress signatures, or you've cleaned the same HX three times in 12 months. Engineering review at the bench, not the loop.

Document everything for warranty and insurance: HX fault history, fluid sample lab reports, BMS trend logs, photos of fin / fan condition, the four-temperature trendline, and a written diagnostic chain. Dry-cooler manufacturers and dielectric-fluid suppliers want documentation for warranty claims; insurance carriers want proof of preventative service for any silicon-damage claim.

Recommission the loop end-to-end before walking away. HX cleaned / replaced / upsized, fans validated, four thermocouples logging, alarm thresholds set at 45-50 °C tank fluid and 8 °C tank stratification delta, service interval scheduled (quarterly fin inspection, semi-annual fin cleaning, annual fan-bearing audit). Without the prevention scaffolding you're scheduling the next failure.