Dry cooler inspection and maintenance field guide

What does a dry cooler inspection cover?

A dry cooler transfers heat from single-phase liquid inside an outdoor finned coil to fan-driven air outside it. The IEA SHC Task 38 C5 report, December 2010, separates dry sensible heat rejection from evaporative and hybrid arrangements [S6].

In steady dry-only cooling, leaving liquid stays above coil entering-air dry-bulb; check sensors, timing, wet assistance or another cooling path for an apparent below-air result. Adiabatic pads change the entering-air condition. A wetted tower instead references wet-bulb: see cooling-tower operating differences.

Trace the piping diagram to identify the served loop and shared or dedicated rejection; data center cooling system selection explains that heat path. A refrigerant condenser instead rejects heat while refrigerant condenses inside its coil.

Include valves, strainers, pump interfaces and expansion connections. Reconcile bypass/changeover positions with the waterside economizer sequence; service tower and adiabatic systems separately.

How often should dry coolers be inspected and cleaned?

Vertiv's Liebert Drycooler Installer/User Guide, revision 7, March 2026, recommends monthly visual inspection and normal spring-and-fall cleaning for standard drycooler coils [S1]. Carrier's 09VE instructions, EN7536635-09, February 2024, specify annual coil cleaning adjusted to environmental exposure [S3]. Establish the route from the installed unit and coating instructions, then shorten it where inspections show recurring contamination.

Use Vertiv §10.2.1's spring/fall plan alongside the annual minimum in §10.2.3. Record a separate cleaning interval for the installed coating.

Inspect fin depth after cottonwood or dust, airflow after snow or storms, rotation after motor work, and fluid after dilution. Record exposure and next due date.

What should you record before touching the unit?

Save the baseline before changing settings. Record model/serial, coil/coating, controller, fan count, pump arrangement, selected duty, design ambient and fluid/concentration basis. Photograph faces, underside, headers and fans from repeatable positions.

Capture simultaneous fluid inlet/outlet temperatures, measured coil-branch flow, inlet-air temperatures across every active face, remote ambient, fan command, individual feedback, electrical input and liquid pressure loss. Put the flow and temperature measurements on the same side of bypasses or mixing junctions. Pump nameplate flow, a valve command and total header flow do not establish flow through the tested coil; coil pressure loss is also distinct from total pump head [S1, S2, S3, S4].

Compare inlet air with remote ambient for recirculation or hot exhaust; retain individual readings, averaging method, probe locations and wind. Verify rotation and discharge direction after motor work [S1, S2, S3].

Instruments and repeatable operating conditions

Use paired temperature probes, suitable flow/pressure instruments, fan/electrical readings and a fluid-specific refractometer. Record calibration and uncertainty; shield air probes and use verified liquid wells or documented insulated surface probes.

Anvilfield suggests 15 minutes of synchronized one-minute readings after operation settles, holding fan state where approved. Extend unstable tests. This field method is separate from EN 1048 certification.

Isolation before intrusive work

Agree alternate cooling and the work window. For covered U.S. general-industry maintenance, OSHA 29 CFR 1910.147 requires hazardous-energy isolation, control of stored energy and verification before work; its definition excludes control switches as isolation devices [S14]. Under 29 CFR 1910.333(b), exposed electrical work requires the prescribed lockout/tagout procedure and verification of deenergization by a qualified person [S15].

Vertiv revision 7 warns that controller-off leaves components energized; isolate local and remote supplies and verify absence of voltage [S1]. Relieve trapped pressure and control temperature before opening the liquid circuit, following GFH §10.1.2 and Carrier 09VE §13.1 [S2, S3]. Secure fans against unintended movement before reaching through guards.

How do you calculate and normalize dry cooler capacity?

Calculate liquid-side heat rejection by multiplying mass flow by specific heat and the liquid temperature drop. For measured volume flow, first multiply by density; use properties for the actual formulation, concentration and mean fluid temperature from its supplier. Dow's September 2001 engineering guide supplies the property basis used in the worked calculation below [S5].

Eurovent's Heat Exchangers programme identifies EN 1048:2014 SC15 dry-cooler rating conditions as 25°C entering air and 40°C entering, 35°C leaving water [S4]. These are rating conditions, not field alarm limits. Keep rejected heat, entering-fluid driving difference and leaving-fluid approach temperature separate: approach compares outlet liquid with coil inlet air, not remote weather-station air.

Normalize with an OEM performance calculation at the observed liquid inlet, coil entering air, actual fluid, flow, active fans/speed and altitude. Supply inlet temperature and flow so the model predicts outlet temperature and duty. Independently forcing measured outlet temperature and a desired duty into the same calculation conceals the comparison being tested.

Anvilfield compares measured duty with that field prediction, then converts it using the reference-to-field OEM prediction ratio. Retain both outputs, reference conditions and software version.

Method box: heat balance, approach and comparison ratio

Use Q = ρ × V̇ × cp × (Tin − Tout), where Q is kW, ρ is kg/m³, V̇ is m³/s and cp is kJ/(kg·K). Define DT1 = Tin − Tair,in and approach = Tout − Tair,in. The matched-state screening index Kscreen = Q/DT1 has units kW/K and tracks performance only with fluid, flow and fan state held constant [S4, S5, S6].

The reference conversion is Qnormalized,reference = Qmeasured × QOEM,reference / QOEM,field. The comparison ratio is R = Qmeasured / QOEM,same conditions; 0.90 means 90% of the prediction, with the project criterion and measurement uncertainty setting acceptance. This method transfers observed relative performance within the selection model's valid range.

For a calculated 5 K liquid temperature drop, Anvilfield's heat-balance arithmetic gives a 4% duty change for a 0.2 K error in that temperature difference, before flow or fluid-property errors. Two probes each having 0.2 K error do not automatically yield a 0.2 K difference error; combine their uncertainties on the actual stated basis.

Retest when measurement uncertainty crosses the agreed acceptance boundary. A small liquid temperature drop makes probe error a larger fraction of calculated duty.

Worked examples: glycol duty, fan mode and ambient correction

Anvilfield estimates 38.67 kW for DOWFROST solution containing 40% propylene glycol by volume at 1 L/s and 45/35°C liquid temperatures. The chosen operating inputs use Dow's September 2001 Tables 10 and 22 properties at the 40°C mean liquid temperature [S5]. Glycol volume fraction and formulated-product volume fraction remain separate labels.

Method note: multiply 1,026.4 kg/m³ by 0.001 m³/s, the unrounded specific heat 3.768 kJ/(kg·K), and the 10 K liquid drop. Use mean-temperature properties over this span; use an enthalpy difference or temperature-dependent integration where the span makes constant properties inaccurate. Attach the field-condition OEM prediction before assigning a capacity ratio, and recalculate properties after concentration changes.

Published fan modes are different rating states

From Eurovent's LU-VE SAL8T-3112 D record, certificate 00.10.214, accessed November 5, 2026, Anvilfield estimates a low-to-high rated-duty ratio of 82.84% and a fan-input ratio of 67.07% [S7]. The common liquid temperature drop implies different flows between these rating states.

Method note: divide 36.70 by 44.30 for the duty ratio and 550 by 820 for the input ratio. The capacity difference is 7.60 kW; the differing pressure losses reinforce the need to match hydraulic conditions.

The model record also contains a generic water-plus-34%-ethylene-glycol field, while SC15's programme definition specifies water [S4, S7]. Preserve both labels and obtain a fluid-specific OEM selection for acceptance. Infer neither absolute flow nor a fixed-flow fan-speed penalty from these points.

ANSI/AMCA 99-25 section 4 and ASHRAE's 2024 Fans chapter describe fan similarity and fan/system interaction [S8, S12]. The fan laws and affinity relationships do not turn a fan power ratio into a coil heat-rejection ratio. Match actual fan feedback and hydraulic state before attributing a duty change to fouling.

Warmer air changes both duty and outlet temperature

Anvilfield estimates 29.53 kW and 36.67°C leaving liquid when entering air rises to 30°C from the published high-speed reference, holding 40°C liquid inlet, liquid flow, fan mode, fluid properties and effective heat-transfer response fixed.

Method note: scale reference duty by the new-to-reference driving difference. Divide that duty by the fixed liquid heat-capacity rate and subtract from inlet temperature to obtain outlet temperature. Use the OEM model for acceptance; this constant-response estimate explains why a smaller approach can accompany lower duty.

Convert the measured result to a reference condition

For the glycol example, choose a 20°C entering-air reference with unchanged liquid inlet, fluid, flow and fan mode. Using its 38.67 kW duty, a field-prediction input of 42 kW and a reference-prediction input of 50 kW, Anvilfield estimates 46.04 kW at the reference condition and a 92.07% performance ratio. Method note: multiply measured duty by 50/42, then divide measured duty by 42 for the ratio. These prediction inputs are calculation assumptions; enter the actual saved OEM outputs and project acceptance criterion in Table B.

How do you survey finned-coil condition?

Map a V-bank by side and panel, or a flat coil by numbered face sectors. Inspect inlet and discharge sides and the accessible fin depth. Vertiv's revision 7 manual warns that repeated outside-only washing can drive contamination farther into the coil [S1]; a clean outer face is not proof of an open air path.

Record contamination, crushed fins, coating failure, tube/header corrosion, leakage and support defects separately. Anvilfield's survey below translates the OEM maintenance subjects into a cause/action record [S1, S2, S3]. Estimate affected projected face area alongside photographs and geometry, keeping blocked-face percentage separate from capacity loss.

Repeat photographs and matched-state duty after cleaning. Retain metal-loss and fin-damage findings even when the current load is met; thermal images do not prove pressure integrity. Classify each finding as a service hold, repair before peak season, or monitor with a dated photograph and next inspection.

What makes a condition an immediate service hold?

Hold equipment for unsafe fluid release, damaged/loose fans, missing protection or electrical defects under its isolation procedure [S1, S2, S3]. Distinguish surface staining from confirmed leakage or metal loss.

Specify tube/header repairs, test medium and allowable test pressure from the unit documents.

How do you clean coils without damaging fins or coatings?

Choose the cleaning method by unit, coil construction and coating before connecting a washer. Start with accessible loose debris and the approved low-impact method, protect electrical equipment and control runoff. Vertiv's standard-coil instructions prefer inside-out cleaning [S1]; inspect depth and discharge surfaces rather than stopping at the inlet face.

For U.S. general-industry cleaning, OSHA 29 CFR 1910.242(b) requires compressed air below 30 psi with effective chip guarding and personal protective equipment [S13]. OSHA directive STD 01-13-001, October 30, 1978, applies that pressure limit at the nozzle under static conditions, including a blocked outlet [S16]. Select a compliant nozzle and air supply within the equipment ratings; the European manuals' coil-cleaning ceilings do not establish OSHA compliance.

These pressures are construction-specific ceilings, not starting settings. Güntner gives separate limits for fans; use the component's instructions when cleaning motors or blades.

Follow supplied unit/coating documents: Carrier prohibits detergent on specified BLYGOLD, ALTENA and HERESITE fins; Vertiv permits mild liquid detergent in its Heresite procedure [S1, S3]. Resolve overlapping instructions with a written compatible method. Cleaner pH and loop-fluid pH are separate controls.

Identify microchannel versus fin-and-tube construction before choosing tools; these limits do not approve every microchannel coil. Clean screens and accessible internals, rinse, dry and refit guards as required. Record method, cleaner/dilution and photographs.

What glycol protection does the outdoor loop need?

Dow's September 2001 DOWFROST engineering guide calls for a solution freezing point at least 5°F, about 3°C, below the lowest anticipated ambient when freeze protection is required [S5]. Select for the coldest exposure during normal service and power loss, with the exact product and concentration basis identified.

Vertiv Table 5.6 states 36% propylene glycol by volume at 0°F, while Dow Table 4 states 37.7% DOWFROST product by volume for freeze protection at that temperature [S1, S5]. Glycol volume percent, glycol mass percent and formulated-product volume percent are different quantities. Use the supplier's conversion rather than averaging these values.

Freeze protection, burst protection and power loss

Freeze protection applies where liquid must circulate at the minimum temperature or ice cannot be accommodated. Burst protection applies to inactive systems with sufficient expansion accommodation and does not establish pumpability [S5]. For a 5°F site minimum, choosing Dow's 0°F row supplies its stated margin; the burst-only concentration in that row serves a different objective.

Carrier's 09VE maintenance instructions, EN7536635-09, February 2024, state that standard coils cannot be completely drained by gravity and prescribe draining, clearing residual water with air and filling with antifreeze for freeze protection [S3]. Fan shutdown, a closed drain valve or a normally energized pump does not protect trapped water during a prolonged cold outage. LU-VE's legacy catalogue also addresses residual-water freezing [S11].

Sampling and chemistry decisions

Sample after circulation and record location, temperature, top-up history and instrument scale. ASHRAE's 2016 Hydronic Heating and Cooling chapter calls for annual refractometer checks and fluid-specific inhibitor maintenance [S9]. TPI's Model 395 specification lists separate PG/EG scales and a −60 to 20°F freeze-point range [S10]; match the scale and supplier conversion to the actual fluid.

Concentration does not prove inhibitor health. Dow's September 2001 DOWFROST/DOWFROST HD guide specifies a desired in-service pH of 8 to 10 and replacement below pH 7 [S5]. Use the supplier's corrective plan for abnormal chemistry, as described in closed-loop glycol treatment.

Record makeup volume and batch identity, distinguishing water, concentrate and premix. Follow supplier dilution-water and materials-compatibility instructions, and keep unlike inhibitor packages separate unless mixing is approved [S5, S9].

Dow's September 2001 solution-makeup procedure calls for at least 24 hours of circulation before the concentration check [S5]. After correction, vent high points and verify expansion and air separation with cold/hot pressures against the installation design, then retest. Repair leakage behind recurring top-ups; excess glycol also changes viscosity, heat capacity, flow and pump duty, requiring a revised selection.

How do you verify the fan-control sequence?

Test against the approved liquid-temperature sequence and actual sensor location. Record enable conditions, staging, minimum speed, hysteresis, fan command/feedback and active limiters. Outdoor axial fan banks do not automatically use an AHU fan wall's duct-static reset or damper arrangement [S1, S2, S3].

A 100% command is not proof of full RPM. Check each fan's feedback, electrical input, motor faults, supply, night quiet mode and demand limits. Use the shipped motor/controller documents for allowable speed and operating limits; Carrier's February 2024 09VE instructions exclude EC fans from its reverse-rotation cleaning method [S3].

A real staged-control benchmark: two-fan Liebert unit

Vertiv revision 7 Table 9.2, drawing DPN001602 revision 5, gives the two-fan factory fan-off values and 8°F rise-to-close differential in Table G [S1]. Adding that differential produces the stated closing temperatures. The optional GLYCOOL settings require the correct application: the manual identifies GLYCOOL/Dual Cool and specified single-drycooler loops with motorized ball-valve flow control, plus separate CRV minimum settings.

For modulating control, raise the approved test demand, confirm each fan follows the ramp, then lower demand to verify turndown and deadband without hunting. Use the shipped schematic for staging; six/eight-fan grouping and EC modulation require their own sequences.

Schedule fault/restart simulations in the approved work window with alternate cooling. Verify surviving duty at required ambient/fluid conditions; spare fan count alone does not prove N+1. Apply fan-array redundancy principles with this bank's temperature controls.

Why does a dry cooler still run hot after cleaning?

Separate air, hydraulic and control evidence before ordering parts, using Table H. Compare coil pressure loss with the clean baseline at matched fluid and flow; a pressure reading at a different operating state cannot identify coil restriction.

Verify branch flow, strainer differential and actual valve positions through hydronic flow verification. Follow low delta-T diagnosis when system mixing or load changes distort the liquid drop. Trace economizer bypasses and CDU facility-loop commissioning interfaces before assigning a plant-level problem to the outdoor coil.

Cold weather, high ambient and adiabatic attachments

Remove snow and seasonal debris using the unit's maintenance instructions [S1, S3]. Record air binding, charge loss and expansion problems separately from airflow obstruction.

At high ambient, compare required duty with the selection at actual coil entering air before treating full-speed operation as a fault [S4, S6]. Use the selected approach for that fluid, flow and fan state.

Keep dry-mode and adiabatic-mode trends separate. Pads, sprays, wetting control, water quality and seasonal drain-down have their own manufacturer service requirements; IEA's 2010 report distinguishes pre-cooled and hybrid heat rejection [S6]. Wet assistance changes inlet conditions and cannot serve as the dry-only performance record.

How long should dry cooler service take?

Anvilfield estimates 4 to 8 labor-hours for routine service on one accessible one- or two-fan dry cooler. This November 5, 2026 planning allowance covers readings, debris removal, approved coil cleaning, field fluid tests, fan checks and documentation.

Method note: Anvilfield estimates labor by adding these task allowances. Two technicians working three hours use six labor-hours; elapsed time also includes drying, load availability and permitted restart. A fluid recharge needing Dow's mixing period or a laboratory result extends handback beyond routine service.

Estimate larger banks from coil area, fin depth, fan count, access and cleaning method. Scope lifts, runoff containment, coating repairs, replacement fluid volume, laboratory work, fan replacement, temporary cooling and witnessed peak testing separately.

Clean and retest an obstructed coil; replace a failed fan to its approved selection. Compare repair with replacement when metal loss, repeated leakage, unavailable parts or inadequate selected duty prevents dependable service. Retain repair history and the consequence of another outage.

Building a seasonal dry-cooler service route

Build the route around identifiable outdoor heat-rejection banks, site access, actual unit inventory and owner-approved work windows. Visible fan banks establish the survey destination; the paid site scope establishes construction, fluid, defects and capacity. Route managers tracking outdoor heat-rejection banks use PorchRocket Asset Search for scheduled monitoring that reports new, removed or altered assets with dated views.

What proves the dry cooler is ready to return to service?

Join all five records by asset ID, work-order ID, timestamp and operating mode. Retain units, raw readings, manual revision, result, action owner and next due date so another technician can reproduce the decision.

The contractor records measurements and repairs; the owner authorizes handback and restrictions. The project engineer or commissioning authority applies witnessed thermal criteria.

Restart with secure guards and connections, correct valve lineup, stable inventory, verified flow and every required fan responding [S1, S2, S3]. Clear unexplained faults and remove test overrides. Tie the outdoor record to the liquid-loop commissioning records so plant acceptance uses the same operating boundary.

A cool-weather test proves its recorded load and conditions; design-day capacity requires the appropriate selection and agreed verification. Assign deferred tests an owner/due date and retain restrictions when closing routine service.

Field checklist

• Identify model/serial, coil material/coating, controller and fan count.
• Obtain the actual selection, fluid data, shipped schematic and maintenance editions.
• Confirm served load, loop routing, alternate cooling and authorized work window.
• Photograph all faces, fin depth, headers, supports and fan compartments.
• Save synchronized air/fluid temperatures, flow, pressures, fan command/feedback and input.
• Note weather, exposure, operating limits, sensor positions and instrument checks.
• Apply lockout/tagout, verify deenergization and control stored pressure, heat and fan movement before intrusive work.
• Map debris, crushed fins, coating failure, corrosion, leaks and support defects separately.
• Record the approved cleaner/method and its coating-specific conditions.
• Clean through the accessible coil depth; rinse and dry as required.
• Inspect every fan, blade, guard, fixing, connection and fault indication.
• Verify fluid identity, concentration basis, freeze target and chemistry/sample findings.
• Resolve leakage, air binding, low flow, strainer restriction and incorrect valve lineup.
• Refit guards, restore normal valves and restart under the approved procedure.
• Test fan ramp/staging and applicable approved failure/restart modes.
• Remove manual overrides and confirm normal operating mode.
• Repeat the valid matched-condition duty and retain the raw trend.
• Issue acceptance disposition, open repairs and the next dated inspection work order.

Sources, review date and how to cite this guide

Sources checked and reviewed: November 5, 2026. Update log: November 5, 2026, editorial review added U.S. cleaning and isolation requirements, completed the reference-duty calculation and clarified service scope. Historical editions remain identified.

• S1. Vertiv, Liebert Drycooler Installer/User Guide, SL-10080_REV7_03-26, March 2026.
• S2. Güntner, Dry cooler GFH operating instructions, 2022-05.
• S3. Carrier, 09VE installation, operation and maintenance instructions, EN7536635-09, 02.2024, supersedes -08 12.2022.
• S4. Eurovent Certita Certification, HE programme public description, live page, accessed 2026-11-05.
• S5. Dow Chemical, Engineering and Operating Guide for DOWFROST and DOWFROST HD, Form 180-01286-0901 AMS, published September 2001.
• S6. IEA SHC Task 38, C5 Heat rejection, technical report of subtask C work package 5, December 2010.
• S7. Eurovent Certita Certification, LU-VE SAL8T-3112 D model sheet, certificate 00.10.214, live record accessed 2026-11-05.
• S8. ANSI/AMCA Standard 99-25, Standards Handbook, adopted March 14, 2025, ANSI approved March 25, 2025, §4 The Fan Laws.
• S9. ASHRAE Handbook, HVAC Systems and Equipment 2016, chapter 13 Hydronic Heating and Cooling, public S16 IP chapter, §2.11 Antifreeze Solutions.
• S10. TPI Europe, Model 395 Handheld Refractometer, live manufacturer specification, accessed 2026-11-05; no revision stated.
• S11. LU-VE, Dry-Coolers catalogue, public page 13, undated legacy edition.
• S12. ASHRAE Handbook, HVAC Systems and Equipment 2024, chapter 21 Fans, public S24 SI chapter.
• S13. OSHA, 29 CFR 1910.242(b), compressed-air cleaning; current text accessed 2026-11-05.
• S14. OSHA, 29 CFR 1910.147, hazardous-energy control, especially (b) and (d); current text accessed 2026-11-05.
• S15. OSHA, 29 CFR 1910.333(b), electrical work practices; current text accessed 2026-11-05.
• S16. OSHA, STD 01-13-001, compressed-air nozzle pressure and chip guarding, October 30, 1978; accessed 2026-11-05.

How to cite this page: Anvilfield, Dry cooler inspection and maintenance field guide, 2026, https://anvilfield.com/field-guides/hvac/dry-cooler-inspection-maintenance/

Related tools

FAQ

People also ask

• How are data centers cooled?
• What is a waterside economizer?
• What is a fan array in an HVAC air handler?
• What does a hydronic make-up water assembly do?
• How do you balance a hydronic system?