Per-MW pricing, regional variance, and cost drivers for owners scoping hyperscale & AI builds.
Salary benchmarks across the 14 mission-critical disciplines.
If you get staffing wrong by phase, cooling commissioning can slip the whole data center turnover. In this article, I break down five cooling systems and six commissioning roles across preconstruction, construction, startup, IST, and turnover so you can see who should lead, when risk peaks, and where handoff gaps show up first.
Here’s the short version:
A few numbers make the stakes clear:
What I like about this piece is that it doesn’t just describe equipment. It shows how the work shifts by phase:
Data Center Cooling Commissioning: 5 Systems × 6 Roles by Phase
The core point is simple: I wouldn’t staff all five systems the same way, and I wouldn’t keep the same team shape from start to finish. The article shows where each system needs deeper BAS skill, TAB support, hydronics knowledge, OEM help, or water chemistry oversight so you can line up people before the risk window hits.
The chiller plant makes chilled water, sends it to the CRAH units, and completes the loop through condenser water and heat rejection equipment. The commissioning scope usually covers water-cooled chillers, pumps, piping, valves, controls, and heat rejection equipment. Every link in that chain needs to be ready before the plant can be fully validated.
Redundancy is where chiller plant commissioning stops being routine and starts getting serious. In many U.S. hyperscale and colocation projects, the target is N+1. That means if one chiller goes down, the remaining units still have to carry 100% of the design load [2]. Some Tier IV facilities push that to 2N. So the job is not just checking that the equipment is in place. The team has to prove that the installed setup does what the spec says it should do.
This phase focuses on design review and testability. Before construction starts, the CxA and mechanical Cx lead should review the Basis of Design, often in coordination with hiring data center construction project managers who oversee these early-stage reviews, P&IDs, and the sequence of operations (SOO). The main issue is simple: can the sequence be proved in the field, or does it only look good on paper?
That means checking a few things early. Does the SOO handle N+1 failover the right way? Are test ports, flow meters, and differential pressure taps called for? Can any chiller or pump be isolated and taken out of service without shutting down the whole plant? If those details are missed at this stage, testing later gets messy fast.
The field commissioning engineer and TAB coordinator drive this phase. The big issue here is hydraulic balance. Bad pipe sizing, missing balancing valves, or a poor primary/secondary setup can lead to low delta-T syndrome. And when that happens, chiller loading and coil performance both take a hit.
This is why TAB ownership matters so much. It helps catch those water-side problems before functional testing starts. Controls integration is another common trouble spot. If BMS point mapping is incomplete, or alarms are missing, the plant may not react the way it should during a failure.
The Cx engineer leads FPT with support from the chiller OEMs, the mechanical contractor, and the BMS integrator. This is the phase that proves failover behavior, not just normal day-to-day operation.
Testing should confirm:
Redundancy failover tests matter here. They should not be treated like a nice extra. The team needs to intentionally trip a chiller or pump and confirm the rest of the system responds the right way. BMS trending under real or simulated IT load shows whether staging and load response stay stable when the plant is under pressure.
Turnover is about operator readiness and clear, documented failover steps. The Cx lead should hand over full FPT records, SOPs for common failure cases, and the BMS trend setups. Operations staff need to be able to run a manual failover on their own, without calling the commissioning team for help.
Change control matters here too. Once the site is live, the approved sequence needs to stay intact. At that point, attention moves from central cooling generation to room-level cooling delivery.
Once the chiller plant is proven, the next dependency is CRAH unit response and control.
CRAH units turn chilled-water capacity into actual room cooling at the cold aisle. Each unit pulls in warm return air, runs it across a chilled-water coil, and sends cooled air back through a raised floor or overhead ductwork. The mechanical scope includes chilled-water piping and valves, condensate drainage, fans and variable-speed drives, filters, and BAS integration. Once the chiller plant is proven, the next issue is simple: does chilled water make it to the data hall and stay under control? What matters here isn't just whether the unit starts. It's whether it works as part of the live thermal system.
Start with the SOO and the sensor plan. The CxA and mechanical Cx lead should check that the SOO covers chilled-water loss, high return-air temperature, fan failure, valve faults, restart behavior, and interlocks with the chiller plant and pumps. Sensor types and locations also need to be set early, including supply and return air temperature, humidity, filter differential pressure, and underfloor plenum pressure.
ASHRAE guidelines recommend keeping IT inlet air temperatures between 64.4°F and 80.6°F for most enterprise equipment [3][5]. That means design setpoints should be checked against those limits before construction begins. Coil selection also needs a close review against the chilled-water supply range the plant is meant to deliver, which is typically 45°F to 54°F [5][3].
At this stage, installation quality is the big issue. The field Cx engineer should verify unit anchoring, vibration isolation, valve orientation, drain-line slope and trap depth, and piping insulation before any functional work starts. Raised-floor tiles and the supply and return paths should match the design layout exactly.
Airflow paths deserve an early check too. Short-circuiting between supply and return is a common miss, and it can undermine the whole setup. Pre-functional checks should include fan rotation, valve stroke testing, and sensor checks against a calibrated reference. If the install is off, functional testing gets messy fast, and control stability usually becomes the first problem.
This is where CRAH issues tend to surface. A unit can pass mechanical startup and still fall apart under part load or during transients if the control loop isn't stable. Valve hunting can cause room-temperature swings and shaky control.
Functional testing should confirm a few things:
Sensor calibration needs extra care. A return-air sensor that's out of calibration can make a unit look like it's holding setpoint when the room is actually drifting. BMS trending helps catch sensor drift and loop instability before they become bigger operating problems.
Once IT load comes online, there's very little room for control error. Cooling-outage tests show that room temperature can move past limits within minutes of CRAH power loss and recover in about 17 minutes [4]. That's why restart behavior and chilled-water response matter during live turnover. They're not just testing notes buried in a startup report.
The handover package should include calibration records, trend setups, alarm checks, startup sheets, and corrective-action logs. Operations staff need to understand how fast the room drifts, how the unit restarts, and what actions to take if one unit goes down.
Once room-level cooling is stable, the next commissioning risk shifts outdoors to cooling towers and heat rejection.
Cooling towers reject heat through an open evaporative condenser-water system. They’re usually set up in N+1 or 2N, so one cell can be taken offline without losing capacity. That matters because tower issues can snowball fast. If condenser-water performance drops, the rest of the cooling system can feel it almost right away.
At this stage, the CxA and mechanical commissioning lead should make sure tower selections are CTI-certified and sized for the design range and approach at local design wet-bulb conditions, such as 95°F entering / 85°F leaving condenser water. The specs should also spell out commissioning needs early, including CTI ATC-105 field acceptance testing and a performance tolerance of ±1°F on condenser water supply temperature.
The specs should also require an ASHRAE 188-compliant Water Management Plan. That plan needs to cover basin cleaning, drift control, biocide dosing, and microbiological testing [6][8][12][13]. This is a life-safety and compliance item, not just a paperwork step. BMS point lists and sequences of operation for cell staging, basin level control, blowdown, and freeze protection should be finalized before construction starts.
During construction, the mechanical commissioning lead and field commissioning engineer—who should possess essential construction management skills—should inspect each cell and verify the structure, basin assembly, piping, electrical terminations, and interlocks before fill. Pipe connections - makeup, overflow, blowdown, and condenser supply and return - should be checked for correct sizing, valve orientation, and strain relief.
Before final fill, flushing, chemical cleaning, and passivation of condenser and tower piping should be completed. This work should be led by the commissioning engineer with the water treatment specialist involved. It’s the kind of step people can treat like a box to check, but skipping it can come back to bite later.
This phase comes down to cell-by-cell testing. The CxA or commissioning engineer verifies basin level setpoints, makeup valve operation, low-level cut-out, overflow function, and basin heater response. Nozzle distribution should be inspected to confirm even water coverage across the fill media before thermal performance testing starts.
The team should also document fan rotation, blade condition, vibration readings, and response to BMS commands and high-vibration or high-amperage trips. In plain terms, you want proof that each cell responds the way it should before the system is under pressure.
Thermal performance testing shows whether the tower is meeting design. Achieved range and approach are compared against design values, and condenser water flow rates should be verified within ±5% of design [9][10]. Trending fan speed, basin level, and condenser water temperatures over several hours and across different load conditions can help spot control hunting before it turns into an operations headache.
Turnover should leave operators ready to handle cell staging, freeze protection, and loss-of-flow alarms without needing commissioning support on standby. The handover package should include:
Operations stakeholders also need to understand the cell staging logic, freeze protection sequences, and how the BMS reacts to a lost cell. That’s not a small detail. A basin-level failure or loss of condenser water flow can trip chillers and bring down IT load fast [7][8][9].
That handoff leads to the heat-rejection loops that move water between towers and chillers.
The heat rejection loop moves heat out of chilled-water or condenser-water circuits and sends it to a cooling tower, dry cooler, fluid cooler, or heat exchanger. That includes pumps, piping, isolation and balancing valves, strainers, air separators, VFDs, sensors, and the controls that connect the whole loop. In many new AI data center designs, facility water loops are set at about 114°F, which supports warm-water heat rejection and can allow heat reuse before final rejection [17].
This loop has little room for error. Cooling failures - many tied to heat rejection loop problems - make up about 13–19% of impactful data center outages [14][15][16]. So if role ownership gets fuzzy at any stage, turnover gets messy fast. The first big test happens before construction even starts, when the team sets topology, controls, and redundancy.
The CxA sets the Owner's Project Requirements (OPR) for the loop. That means target availability tier, redundancy targets, the planned water temperature class, and water use limits. The mechanical lead then locks down the loop topology - primary/secondary or variable primary - so the system doesn't run into low-delta-T issues or unstable flow later on.
At the same time, the controls team finishes the sequence of operations, and the TAB specialist checks that the system can actually be tested before the build begins. Redundancy has to be defined at every layer: pump, header, loop, and plant. A single nonredundant header can wipe out upstream N+1 capacity [19][20].
During construction, the mechanical commissioning lead checks that the installed pumps, valves, heat exchangers, and towers match the design selections for flow, range, and approach. On standard condenser loops, that means about ~3 gpm/ton [9][18].
Pressure testing should be done at 1.5 times working pressure before insulation and concealment [21]. That's a simple checkpoint, but missing it can turn into a painful problem later.
Instrumentation also has to be in place early - temperature sensors, pressure transducers, flow meters, and valve actuators. Those devices need to be installed, calibrated, and trend-logging from day one. If sensor placement is off, the controls can start hunting right out of startup. Once the hardware is in, the next question is pretty direct: can the loop stay stable as load comes on in stages?
Functional performance testing has to show that the loop can run across the full operating range:
The commissioning engineer proves staging and alarms, while the TAB specialist confirms design flow. Flow stability is the clearest sign that the loop was commissioned well. If differential pressure swings too much, pumps cavitate, or valves hunt during staging changes, thermal instability can follow - and that can roll straight into the IT load.
At live turnover, the commissioning lead moves from proving the install to proving the team is ready to run it. Operations staff need to know the normal control sequences, alarm priorities, temporary bypass conditions, water treatment needs, and escalation paths before commissioning support steps back.
The handover package should include as-built sequences, trend logs from functional testing, water treatment records, and maintenance access details. By this point, the heat rejection loop is already carrying production cooling loads, so any open punch-list item comes with real risk.
The next dependency is water treatment, which protects the loop from scaling, corrosion, and biological growth.
That heat-rejection loop stays dependable only when water chemistry stays in range. Water treatment keeps condenser water, chilled water, and other hydronic loops within spec so corrosion, scale, and biological growth don't eat into uptime or shorten equipment life.
This work covers open condenser water systems, closed chilled water loops, and ancillary closed hydronic loops like heat recovery circuits. Open systems need corrosion, scale, and biological control. Closed systems are more about corrosion protection, oxygen control, and keeping chemical balance steady.
Before installation starts, the CxA and mechanical lead need to lock down the water quality limits. That includes conductivity, hardness, silica, chlorides, and dissolved oxygen for both makeup water and circulating water. Cycles of concentration should be set early. High-efficiency plants often aim for 6–10, with RO pretreatment when feedwater TDS or hardness is too high [11][22][24][13].
A written water management plan, aligned with ASHRAE 188-aligned guidance, should be a contract requirement. That plan needs to spell out sampling frequency, biocide schedules, and corrective-action triggers before any field work begins [13][9]. The CxA should also check that the mechanical design and sequence of operations include side-stream filtration, chemical dosing skids, sampling ports, and online monitoring sensors.
During construction, the mechanical commissioning lead checks that chemical feed pumps, storage tanks, injection quills, and controllers are installed per manufacturer requirements. Materials of construction need to be right, and operations staff need safe access to the equipment.
Before startup, the team should confirm secondary containment, eyewash stations, and proper ventilation so the installation lines up with U.S. safety code requirements [27][28]. After that, the mechanical lead and water treatment vendor carry out the flushing and passivation workflow: flush at 6–8 ft/s [25][26][28], clean strainers, and then follow the written sanitization and passivation sequence before startup.
At this stage, the commissioning engineer, controls/BMS specialist, and water treatment vendor need to show that chemistry control works under actual operating conditions. Start by recording baseline chemistry for both open and closed loops. Then prove automatic blowdown, alarm points, and operator actions [23][29][31].
Common closed-loop targets include:
BAS integration testing should verify that out-of-range conditions like high conductivity, low inhibitor residual, or high bacteria counts trigger alarms and the defined operator response [9][13][23]. Once baseline chemistry and alarm response are proven, turnover shifts into an operator handoff.
Turnover should give the operations team the data and procedures they need to keep chemistry steady without leaning on the commissioning team. The commissioning lead should hand over a full water treatment package with as-built chemical schematics, baseline lab results, operating procedures, biocide schedules, and emergency response protocols.
From there, operations owns the sampling schedule, corrective-action thresholds, and the escalation path when chemistry drifts. Operators should be trained to track that drift back to makeup-water changes, leaks, or added heat load.
Once the system comparisons are done, the hiring question gets pretty simple: who should own each phase?
That matters because each mechanical system needs a different kind of operator, engineer, or specialist.
Chiller plants usually need the deepest BAS/BMS and OEM controller knowledge. CRAH units lean hard on TAB and airflow diagnosis. Cooling towers bring a different set of demands, including outdoor work, water safety, and freeze protection. Heat rejection loops are often a better fit for engineers who can connect hydronics with controls integration. And water treatment is one area teams often underrate in hiring. That can be a mistake, because it calls for chemistry depth and BAS alarm integration work that goes well beyond commissioning.
Phase ownership shifts as the job moves forward.
Preconstruction usually sits with the CxA and the mechanical lead. During construction, the commissioning engineer, TAB team, and controls team step in. Startup and FPT need more specialized support, especially around chiller plant sequencing, CRAH and TAB coordination, and cooling tower and water treatment systems. IST then calls for senior-level coordination across mechanical, electrical, and BAS scopes. At turnover, ownership moves to operations.
By that point, the biggest issue is no longer just the technical setup. It’s whether the operations team can run the system without outside help. iRecruit.co supports mission-critical hiring for commissioning, MEP, and field roles with data center cooling experience.
The table below turns those tradeoffs into a staffing map.
In practice, the toughest hiring gap usually isn’t technical knowledge by itself. It’s finding people who combine OEM or water chemistry depth with hands-on IST and BAS integration work. That mix is hard to find, and it’s often where hiring timelines start to slip.
Across all five systems, the commissioning load tracks the risk in each phase. Mechanical commissioning doesn’t shift much by system. It shifts by phase.
Some patterns stand out fast. Chiller plants and heat rejection loops carry most of the load in preconstruction and IST. CRAH units hit their high point during construction and functional testing. Water treatment climbs at turnover.
That’s why staffing can’t stay flat from start to finish. The work changes, so the team should change with it. A phase-based staffing plan puts the right people on-site before each system hits its highest-risk moment.
The payoff is lower commissioning risk. On one 32 MW hyperscale project, full cooling-unit testing reduced commissioning risk and protected the ready-for-IT-load date. [32] When that timing lines up, test windows hold and turnover stays on schedule. On mission-critical construction projects, the right specialist in the right phase is a schedule safeguard, not a staffing luxury.
Commissioning roles change as the project moves from planning to startup and handoff.
In Level 0, during design and planning, the Commissioning Authority (CxA) plays the biggest role. This is the stage where the project team sets the direction, reviews the basis of design, and makes sure commissioning is built into the job from the start.
By Levels 1 and 2, the focus moves into the field. Here, field engineers and startup technicians check equipment integrity and confirm that major components are installed and ready for the next step.
Level 3 puts more weight on controls specialists, TAB technicians, and startup technicians. Their work supports safe system energization, which is a big moment on any project. If Level 0 is the blueprint, Level 3 is when the building starts to come alive.
In Level 4, commissioning engineers and discipline leads take the lead. They manage test scripts, guide system validation, and make sure each system performs the way it should under planned test conditions.
Level 5 depends on senior Cx managers and mission-critical project managers. At this stage, the work centers on integrated systems testing and owner handoff, where separate systems must perform together as one.
The chiller plant is the highest-risk cooling system because it sits at the center of the facility’s thermal management. If sequencing breaks down, hydronic delivery slips, or a key component underperforms, the entire data center’s uptime can be on the line.
It also relies on complex controls. Lead/lag staging and pump interlocks have to work without a hitch to hold critical rack temperatures and meet load and redundancy requirements.
Before mechanical turnover, operators should get a complete turnover package that has been built up during commissioning, not thrown together at the last minute.
That package should include as-builts, equipment warranties, training records, final O&M manuals that match the systems as installed, spare parts lists, CMMS asset data, and final commissioning reports, including integrated systems testing results and certificates of completion.
