When to Replace Your Precision Air (Not When to Repair It): The 5 Obsolescence Signals Your Provider Will Not Tell You
When a precision air conditioner fails in production, the operator’s immediate decision is to repair it. But there is a point in the useful life of a piece of equipment where repairing costs more than replacing, and your maintenance provider is not going to tell you that because their business depends on keeping you repairing. This article covers the 5 concrete obsolescence signals that indicate it is time to change the equipment, with figures for when repairing stops making sense and when it still does.
What follows applies to CRAC units (with compressor) and CRAH units (with chilled water) from the main manufacturers: Vertiv (Liebert), Schneider Electric (Uniflair/APC), Stulz, Daikin, Climaveneta, and some legacy Emerson models. For high-density in-row units the signals are similar but the thresholds change.
Signal #1: the compressor is already over 10 years old or has more than 80,000 operating hours
The compressor is the most expensive component and the one that defines the practical useful life of a CRAC. In units that operate 24/7, 80,000 hours is approximately 9 years. From that point on, three indicators should be on your radar:
- Operating current 15% above the nameplate rating. This indicates internal mechanical wear (valves, bearings) or low performance from insufficient refrigerant gas.
- Short compressor cycles (on-off every 2-4 minutes instead of running continuously). This signals that the compressor cannot hold suction pressure stable and starts and stops repeatedly, which destroys contactors and relays.
- Abnormal mechanical noise (vibration, knocking, hum different from the nameplate). Vibration analysis above 4.5 mm/s RMS on the compressor body indicates bearing failure within the next 6-12 months.
- Intermittent failures with no clear pattern. The board starts, runs for 10 minutes, fails, starts again. If replacing the board fixes it but it fails again in 2-4 weeks, the problem is not the board, it is the power supply or the environment.
- Display with ghost characters, flickering, or loss of segments. Classic symptom of swollen capacitors on the display board.
- No spare part is available from the manufacturer. If the manufacturer declares the board ‘obsolete’, or the price of a new one exceeds 30% of the equipment value, it is time to replace the complete equipment.
- R-22 (HCFC): out of production since 2010 in developed countries, available only as recovered refrigerant. Current price: $80-150 USD per kilogram. If your provider charges you $300+ USD for a 5 kg recharge, they are selling you second-hand or stolen refrigerant.
- R-123 (HCFC, historical reference): this refrigerant is used mainly in large-tonnage, low-pressure centrifugal chillers (Trane CenTraVac family and similar), NOT typically in perimeter direct-expansion compressors (CRAC/CRAH). For editorial focus, this article centers on R-22 and R-410A as the predominant refrigerants in CRAC/CRAH units; nevertheless, if you had a chiller with R-123, you would face the same progressive phase-out scenario under the Montreal Protocol and Kigali Amendment.
- R-410A (HFC): being reduced on account of GWP. In Mexico there is no ban yet, but the global trend is migrating to R-454B or R-1234ze. If you buy equipment with R-410A today, in 8-10 years you will be in the same situation.
- No new spare parts. Only remanufactured or used parts, with no warranty of provenance.
- The manufacturer’s technical support is “best effort”, not guaranteed. Your local provider is left alone facing the problem.
- The resale value of the equipment drops to zero. If you need to get rid of it, you end up paying someone to take it away, not receiving anything in return.
- Compressor with insufficient gas: it runs longer and through more cycles to deliver the same capacity.
- Evaporator coil with compacted dirt that reduces heat transfer by 20-30%.
- Worn fans: the motor draws more watts to deliver the same airflow.
- Repair cost over 24 months: if you spent more than 40% of the value of a new unit on repairs, the rational decision is to replace.
- Annual extra electricity consumption: multiply extra kW × 8,760 hours × kWh rate. If it exceeds $80,000-120,000 MXN per year per unit, the savings from an efficient unit pay for the replacement in less than 3 years.
- Downtime cost: each hour of failure of a CRAC that supports an IT room costs between $5,000 and $50,000 MXN depending on the room’s load (measured by the opportunity value of the services that depend on it). If in the last 24 months you had more than 2 major failures with downtime >2 hours, the risk is already operationally unacceptable.
- Temporary backup equipment (CRAC rental). Typical cost: $8,000-15,000 MXN per week including freight.
- A planned maintenance window during low-load hours (Sunday early morning, typically).
- A documented rollback plan in case the new unit has factory defects. Keep the old equipment available for 30 days after installation.
- Factory acceptance tests (FAT) of the new unit before shipping. The manufacturer should offer a signed test protocol.
- 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
- EPA — Regulation of HCFC Refrigerants (Section 82.15) — https://www.epa.gov/section608
- UN Environment Programme — Kigali Amendment to the Montreal Protocol — https://ozone.unep.org/kigali-amendment-overview
- AHRI — Air-Conditioning, Heating, and Refrigeration Institute standards for HVAC equipment life cycle — https://ahrinet.org/standards
- Uptime Institute — Data Center Cost Optimization: CapEx vs OpEx for cooling infrastructure — https://datacenter.uptimeinstitute.com/
If you have 2 of these 3 indicators on a compressor with 80,000+ hours, the rational decision is to schedule replacement of the complete equipment, not the compressor alone. Replacing only the compressor on a 12+ year-old unit is putting a new engine in a chassis with a dirty coil, old boards, and cabinets with degraded seals. You end up replacing the rest of the equipment within 2 years.
Signal #2: the electronic control boards are over 7 years old
Electronic boards are the second component with scheduled obsolescence. Electrolytic capacitors have a useful life of 8 to 12 years under normal operating conditions (ambient temperature below 35°C, relative humidity below 60%). Past that point:
Reprogramming the control board of a 12-year-old piece of equipment with modern firmware is expensive and prone to BMS integration problems. If your site monitoring system (DCIM) reports ‘equipment X with no communication’ recurrently, the problem is protocol obsolescence, not the board.
Signal #3: the equipment’s refrigerant is no longer manufactured or available
This is the least visible but most decisive indicator. Refrigerants are being phased out by environmental regulations (Montreal Protocol, Kigali Amendment, EPA rules). When your equipment uses a refrigerant that is no longer manufactured:
If your equipment uses R-22 and you need a recharge of more than 2 kg, the refrigerant cost alone is comparable to the equipment’s depreciation value. Migrating the equipment to a modern refrigerant requires changing the compressor, the expansion valve, and the filters, which is practically building new equipment.
Signal #4: the manufacturer no longer provides technical support or spare parts for your model
Precision equipment manufacturers typically drop models from the active catalog 7-10 years after the last year of sales. After that:
How to verify it: call the manufacturer (Vertiv, Schneider, Stulz, Daikin) with your equipment’s model and serial number and ask whether spare parts are still in their catalog. If the answer is ‘we have to check with engineering’ or ‘that model is discontinued’, start the replacement. The difference between buying the new equipment today and buying it when the old one no longer turns on is typically 30-50% in price (due to urgency, express freight, installer overtime).
Signal #5: the equipment consumes more than 30% extra energy versus the nameplate rating
The clearest economic signal. If your equipment’s nameplate is 30 kW of cooling and by measurement and certification you consume more than 39 kW electrical to deliver that same capacity, efficiency has degraded to less than 77% of the original design. Typical reasons:
The math: if your equipment consumes 9 kW extra × 24 hours × 365 days × $1.8 MXN/kWh = $141,912 MXN per year of wasted electricity. Over 3 years, that is $425,736 MXN. An equivalent new unit costs between $250,000 and $400,000 MXN installed. The replacement pays for itself in 2-3 years on energy alone, without counting the reduction in failure risk.
The final calculation: repair vs replace (with numbers)
To decide objectively, add up the costs of the last 24 months of repairs + maintenance + extra energy consumed + opportunity cost (downtime hours), and compare against the cost of new equipment:
When 2 of these 3 criteria are met, replacement is economically justified. The final decision must also consider whether the replacement is done in planned fashion (without affecting operations) or as an emergency (with downtime). Planned replacement costs 30-50% less than emergency replacement.
How to plan the replacement without affecting operations
If you have N+1 or 2N redundancy in your design, the replacement is straightforward: you shut down one unit, install the new one, validate, and return to operations. But if your design is N (no redundancy), the replacement requires:
If your data center is on N redundancy, seriously consider adding N+1 redundancy BEFORE the critical equipment reaches the end of its useful life. The cost of installing redundancy is much lower in an existing room than during an emergency replacement.
Sources
Want to master this?
Noxtel Academy →
