The 7 most common CRAC and CRAH failures (and how your operator solves them in 30 minutes without calling the provider)

When a CRAC or CRAH unit fails, the usual reaction is to call the provider and wait. However, 80% of common operational failures can be resolved by your in-house operator in under 30 minutes with basic tools, avoiding 4 to 8 hours of thermal unavailability and risk to your servers.

This article covers the 7 operational failures any trained operator can diagnose and fix on site. For major failures (burned compressors, refrigerant coil leaks, damaged control boards), you will need specialized manufacturer support.

Failure #1: High refrigerant pressure (HP alarm)

Symptom: the high pressure alarm activates on the display, the compressor shuts down automatically, and the unit stops cooling. Most common causes:

• Dirty or blocked condenser. The external heat exchanger (condensation side grille) cannot reject heat to the environment. Solution: clean with pressurized water and soft brush. Time: 20-40 minutes.
• Condenser fan not working. Verify the fan motor spins freely and the start capacitor is in good condition. If the motor is burned, you need a replacement (motor change).
• Excess refrigerant. If you recently recharged gas, it may have been overfilled. Requires a provider with a scale and gauges to recover the excess.
• Very high outdoor temperature (typically >40°C in shade). During heat waves in CDMX or Monterrey, the system can enter high pressure if sized for lower nominal conditions. Temporary solution: shade the condenser or pre-cool outdoor air with water.

Your operator can diagnose by reading the high pressure gauge on the unit. If it reads above the operating range (typically 250-400 psi depending on refrigerant), it’s high pressure. Diagnosing the specific cause is what takes time; resetting the alarm takes only 5 minutes once the root cause is resolved.

Failure #2: Low refrigerant pressure (LP alarm)

Symptom: low pressure alarm, compressor shuts down. Causes:

• Refrigerant leak. The most common. Locate the leak with an electronic leak detector (not soapy, does not detect modern refrigerants like R-454B). If the leak is under 100g, recharge and repair the leak point. If larger, you need a provider with a recharge cylinder.
• Clogged filter drier. It’s a small component in the refrigerant circuit that absorbs moisture. When saturated, it restricts refrigerant flow and trips low pressure. Solution: change the filter drier (cheap refill, 30-45 minute operation with manifold).
• Blocked expansion valve. The valve regulating refrigerant flow to the evaporator can stick closed. Requires a provider experienced to free it or replace it.

If your operator has an electronic leak detector and a small backup refrigerant cylinder, they can fix minor leaks and recharges. Large leaks or persistent leaks require a provider visit with gas recovery equipment.

Failure #3: Open or shorted thermistor

Symptom: the display shows unrealistic temperature (-50°C or +200°C), or the unit enters temperature alarm. The typical cause is a damaged thermistor (temperature sensor) or cut cable. Solution:

• Identify which thermistor is signaling the error (check the unit alarm log). CRACs have between 2 and 6 thermistors: return air temperature, supply air temperature, evaporator temperature, outdoor temperature.
• Verify thermistor cable continuity with a multimeter. Typical NTC thermistors have 10K ohms at 25°C. If it reads open (infinite resistance) or shorted (zero resistance), the cable is cut.
• Replace the thermistor (costs between $200 and $800 MXN). If your operator has spare thermistors, the operation takes 30 minutes.

Thermistors are normal wear components. Having 2-3 spares for the models you have installed is basic inventory. If you don’t have them on hand, the downtime will cost much more than the part.

Failure #4: Evaporator fan failure

Symptom: reduced airflow alarm, supply air temperature rises above target. Causes:

• Burned motor. Verify voltage to the motor (must receive 208V or 220V three-phase depending on model). If it receives voltage but doesn’t spin, the motor is burned. Requires motor change (expensive part, typically $15,000-40,000 MXN).
• Damaged start capacitor. Fan motors use a capacitor for startup. If the capacitor is swollen or leaking, the motor won’t start. Solution: replace the capacitor ($300-1,200 MXN, 20 minute operation).
• Broken belt (in belt-driven motors). Verify belt tension. If broken, replace it ($500-2,000 MXN, 30 minutes).
• Mechanical obstruction. The blade may be blocked by an object. Inspect visually before assuming electrical failure.

Your operator can distinguish between electrical failure (motor not receiving voltage) and mechanical failure (motor receives voltage but doesn’t spin) with a multimeter. That determines whether you need a capacitor (cheap and fast) or a motor (expensive and slow).

Failure #5: High or low humidity alarm

Symptom: the display shows relative humidity outside range (typically 40-55% per ASHRAE). Causes and solutions:

• Miscalibrated humidity sensor. Compare the display reading with a reference hygrometer. If difference is greater than 5% RH, recalibrate or replace the sensor.
• Persistent high humidity. May be humidifier failure (if you have one) or humid outdoor air infiltration. Check door seals and outdoor air filter condition.
• Persistent low humidity. Humidifier failure or humidity control system failure. Verify water supply to the humidifier (if electrode or steam type).

Humidity outside range is less urgent than temperature outside range (servers tolerate 20-80% RH for short periods). But sustained out-of-range conditions damage components and generate ESD (electrostatic discharge).

Failure #6: Clogged condensate drain

Symptom: “condensate pump failure” alarm or visible water leaking from the unit. Typical cause: algae and sediment buildup in the drain or pump. Solution:

• Check the condensate pump. Activate manually and observe if it expels water. If it doesn’t expel, the pump is burned or clogged.
• Clean the drain siphon with water and chlorine solution (1:10). Algae grow in the siphon if light enters. Operator can do this every 3-6 months as preventive maintenance.
• If the pump is burned, replace the pump ($2,000-5,000 MXN, 30 minute operation).

A clogged drain is the #1 cause of secondary water damage in data centers. If not addressed quickly, water rises from the drain, overflows the condensate pan, falls to the false floor, and contacts power cables. Electrical arc scenario.

Failure #7: Erratic temperature sensor reading

Symptom: display temperature fluctuates erratically (rises 5°C in 30 seconds, then drops, then rises). Compressor starts and stops repeatedly. Typical cause: thermistor with poor contact in the connector or cable damaged by flexing.

Solution:

• Check the thermistor connector on the control board. Re-insert to ensure clean contact.
• Inspect the thermistor cable along its entire path. If bent or crushed, the cable is internally damaged even if it looks fine externally.
• Replace the thermistor. It’s the most reliable solution when in doubt.

Erratic thermistor problems are particularly damaging because they trigger short cycling (compressor starting and stopping every 1-3 minutes). The compressor suffers more in short cycling than running continuously. If you see the compressor start and stop in under 5 minutes, first suspect an erratic thermistor, not a compressor failure.

The diagnostic kit your operator should always have

To resolve the 7 failures without waiting for the provider, the operator needs:

1. True RMS multimeter (Fluke 87V or equivalent).
2. Refrigerant manifold with hoses (to measure high and low pressure in CRAC).
3. Electronic refrigerant leak detector.
4. Calibrated thermometers (dry/wet bulb thermometer for humidity).
5. Thermistor spare kit (1-2 per installed model).
6. Capacitor spare kit (1-2 per installed model).
7. Small refrigerant gas cylinder (for minor recharges).
8. Corporate camera or smartphone (to document).

The kit typically costs between $30,000 and $80,000 MXN per operator. The response time savings (versus waiting for the provider) pays for the kit in 2-3 incidents resolved in-house. If your operator doesn’t have this kit, they depend 100% on the provider for failures they could resolve in 30 minutes.

When to DO call the provider

There are failures your operator should NOT attempt to resolve and that require the provider:

1. Burned compressor (verified by winding resistance measurement). Compressor change requires welding equipment, gas recovery, and certification.
2. Dead control board. If the board doesn’t respond to reset or correct voltage, it needs board replacement (expensive part, model-specific programming).
3. Coil or heat exchanger leak. Repairing a copper coil requires specialized welding and pressure testing.
4. Recharge over 2 kg of refrigerant gas. Indicates a large leak requiring professional leak location with detector, and possibly coil repair.

The operator must know how to diagnose AND recognize the limits of their expertise. Trying to repair what they don’t know ends in greater equipment damage and provider relationship damage. If in doubt, document the problem with photos and call the provider. 30 minutes of provider response is still better than worsening the failure.

Sources

1. ASHRAE TC 9.9 — Thermal Guidelines for Data Processing Environments (5th edition) — https://ashrae.org/technical-resources/ai-data-center-framework/energy-and-thermal-efficiency
2. AHRI — Air-Conditioning, Heating, and Refrigeration Institute: Standard 540 for CRAC performance — https://www.ahrinet.org/standards
3. Vertiv — Liebert CRV Service Manual: Alarm codes and troubleshooting — https://www.vertiv.com/en-us/products/brands/liebert/
4. Schneider Electric — EcoStruxure for Data Centers: HVAC Operations Manual — https://www.se.com/mx/es/product-category/data-centers/
5. Stulz — CyberAir Service Documentation: Common alarm codes and field troubleshooting — https://www.stulz.com/en-us/products/detail/cyberaircwanddx/