Data center commissioning operations and process field guide

Commissioning as a process, not a test at the end

Commissioning is the structured, documented process that verifies a building's systems are installed and perform to the owner's project requirements. On a data center it is run by an independent commissioning authority, the CxA, and it covers the whole job, from design review through the integrated test that proves the plant before any IT load lands. It is a program, with its own plan, its own documents, and its own acceptance criteria. The test at the end is the last gate, not the whole of it.

The mistake people make is treating commissioning as a punch walk with a meter. It is not a final inspection and it is not the contractor's quality control. By the time the integrated systems test runs, commissioning has already confirmed the design captured what the owner needed, the gear that got bought matched the design, it was installed and terminated right, and every system did what its sequence says on its own. The end test only exists to prove the seams between systems, and it only means something because the work ahead of it was verified and recorded.

This guide is the process and program view: how commissioning is organized, who runs it, what documents govern it, and how the record gets built and handed over. The technical tests on the power chain live in the data center electrical commissioning guide, and the airside and waterside verification lives in the cooling guide. The subject here is the operating system that holds all of that together. The plan, the scripts, the deficiency log, the witnessing, and the owner report.

The OPR and the BOD: the yardstick everything is measured against

The owner's project requirements, the OPR, is the document that states what the owner needs the building to do, in measurable terms, and it is the yardstick every commissioning test is measured against. Not the drawings, not the submittals, the OPR. It captures the redundancy the owner is paying for, the temperature and humidity envelope the floor has to hold, the concurrent-maintenance and fault-tolerance expectations, the acceptable downtime, and the operating and maintenance assumptions. ASHRAE Guideline 0, the commissioning process guideline, puts the OPR at the front of the whole process for a reason. If the OPR is vague, every acceptance decision downstream is an argument.

The basis of design, the BOD, is the design team's answer to the OPR. It records the assumptions and the engineering decisions the engineer of record made to meet what the owner asked for: the topology chosen, the design loads, the codes and standards applied, the redundancy scheme on the one-line. Commissioning checks the design against the OPR, then checks the build against the BOD. Two documents, one chain of intent from what the owner wanted to what got installed.

On a real job the OPR and BOD are often thin or stale by the time commissioning shows up, because they were written early and never maintained. That is a problem to flag at the first commissioning meeting, not at the integrated test. A test script has to trace back to a requirement. When the requirement does not exist in writing, the CxA is left inventing the acceptance criteria, and the owner is left with no basis to reject a result they do not like. LEED commissioning credits, where a project is chasing them, require the OPR and BOD precisely because the rest of the process is built on top of them.

Who is on the commissioning team?

The commissioning team is the CxA, the owner, the engineer of record and the MEP designers, the general contractor or construction manager, the trade contractors who built it, and the equipment vendors and integrators. Each has a fixed role, and the value of the whole process rests on the CxA being independent of the people whose work is being checked.

The CxA runs the process. On a data center the CxA is almost always a third party hired directly by the owner, not a subconsultant to the GC, because the entire point is a check that is not grading its own homework. The CxA writes or owns the commissioning plan, builds the test scripts, schedules and witnesses the tests, keeps the deficiency log, and assembles the turnover package. As a rule the CxA witnesses and accepts tests rather than performing them. The owner sits at the top and inherits everything at the end. The engineer of record owns the design intent and answers when a test result does not match the sequence of operations. The GC holds the schedule and coordinates access and energization. The trade contractors execute their own pre-functional checks and support the functional and integrated tests, and the vendors run startup on their gear and stand behind the warranty.

When the roles blur, the process quietly loses its worth. The most common failure is the commissioning agent reporting to the general contractor instead of the owner. Now the party being checked controls the checker's schedule, scope, and paycheck, and the deficiencies that threaten the GC's milestone are the ones that get softened. If you read one thing on a contract, read who the CxA works for.

The commissioning levels, L1 through L5

Most data center programs organize commissioning into levels, commonly numbered Level 1 through Level 5, that move from the factory floor to the fully integrated plant. The levels are gates. You do not start one until the level below it is signed off, because skipping forward is how a deficiency hides until the worst possible moment. The framework below is the common shape, but the exact boundary between levels shifts from one program to the next, and the project commissioning plan is the authority for what each level includes and who witnesses it.

The variance is real, not a technicality. Some programs split static installation checks and pre-functional energization into separate levels, some fold them together, and some add a Level 0 for design review and commissioning planning and a Level 6 for post-occupancy and seasonal testing. The tag-color shorthand some sites use, yellow tag for installation, green for startup, blue for functional, also varies by program. Do not assume the numbers or the colors mean the same thing across two jobs. The level boundaries and the witnessing for the electrical scope are mapped in detail in the data center electrical commissioning guide; here the point is that the levels are a sequence of witnessed checkpoints, not a filing system.

Level 5, the integrated systems test, is the keystone, and it is the level schedule pressure attacks first because it needs the whole plant, full load banks, and a long block of uninterrupted time. Everything before it proves a piece. Level 5 proves the pieces work together under a simulated failure.

The commissioning plan and the Division 01 specification

The commissioning plan is the operating manual for the whole program. It names the systems in scope, the levels and what each one includes, the team and who witnesses what, the hold points, the schedule, the deficiency process, and the format of the turnover package. A good plan is specific enough that a new CxA technician could walk onto the job and know what test runs next and who has to be in the room. A plan that is a generic template with the project name swapped in is a tell that the program was never really thought through.

The contract requirement for commissioning lives in the specifications, typically in Division 01, the general requirements division, under the commissioning sections commonly numbered 01 91 00, with general commissioning requirements often at 01 91 13. That spec is what gives commissioning teeth: it obligates the trade contractors to support the process, to submit their startup reports, to provide labor for functional testing, and to correct deficiencies. The exact section numbers follow the CSI MasterFormat the project adopted, so confirm them against the actual project manual rather than citing a number from memory.

When the commissioning requirement is buried in a discipline spec instead of called out in Division 01, or when it is missing the obligation for trades to provide test support, you find out at functional testing, when the contractor who never priced the labor refuses to staff it. Read the commissioning spec at buyout, not at startup. The plan describes how commissioning will run. The spec is what makes the rest of the team show up.

Scripts and checklists: PFCs, FPTs, and the IST

Commissioning runs on three kinds of documents, and knowing which one you are using matters because they prove different things. Pre-functional checklists, FPTs, and the integrated test script form a ladder, and each rung is a precondition for the next.

The pre-functional checklist, the PFC, is the static, de-energized verification that a system is installed correctly and ready to energize. Connections landed and torqued, labels in place, settings staged, no shipping splits left open. The trade contractor usually fills the PFC and the CxA spot-checks and accepts it. A signed PFC is the permission slip to energize.

The functional performance test, the FPT, is the energized demonstration that a single system does what its sequence of operations says under normal and fault conditions. The CxA writes the FPT straight from the sequence, with an explicit expected result and a pass or fail box for every step, and witnesses it run. The discipline that separates a real FPT from a checkbox is that the witness signs each step, not the page. A script watched once in a hurry and checked off as a block proves almost nothing.

The integrated systems test, the IST, is the level-5 script that exercises all the systems together through a simulated failure at design load. It is built from the sequence of operations across the whole plant, branch by branch: normal, single utility loss, a generator failure during the event, a UPS module loss, the return to normal. Each branch has an expected outcome and a timing window, and you record what the plant actually did against what the sequence said it should. Write the scripts so a stranger could run them and so the results stand on their own when the owner reads them two years later.

What is the commissioning deficiency log and why does it run the job?

The deficiency log, also called the issues log, is the running list of every problem commissioning finds, tracked from the moment it is identified to the moment it is verified fixed and closed. It is the single most important operational artifact of the whole program. Tests prove the plant in the moment. The log is what carries the findings until they are actually resolved, and it is the document that tells you, on any given day, how far from done the building really is.

Every finding moves through a fixed lifecycle, and shortcutting any step is where the value leaks out. Identify it and write it down with enough detail and a photo that someone who was not in the room understands it. Classify it by severity, because a cosmetic label and a sequence-breaking fault do not get the same urgency. Assign it to the responsible party with a due date. Resolve it, meaning the responsible party makes the fix. Verify it, meaning the CxA re-tests the change under the conditions that found it. Then close it, with the evidence attached. Identify, classify, assign, resolve, verify, close.

The verify step is the one that gets skipped, and skipping it is the failure that defines a bad commissioning program. A deficiency closed on paper is a deficiency still in the building. A relay setting noted to be corrected but never re-read in the gear, a failed transfer marked resolved on the strength of a controls change nobody re-ran, a leak fixed and never re-pressurized. The log looks clean and the problem is still there, waiting for the night the utility actually fails. Closing on paper is the quiet failure mode that every other failure traces back to.

Keeping that log honest across a long project with many parties is hard, because the finding, the photo, the assignment, the fix evidence, and the signoff usually live in different places and different inboxes. This is the kind of field tracking the tradeos workflow is built to carry, so the log, the photos, and the signoff stay attached to the same record instead of scattering between the installer, the test agency, and the CxA.

Hold points and witnessing

A hold point is a step the work cannot pass until the required witness has seen it and signed. Pouring a critical slab, energizing switchgear for the first time, running a functional test, dropping the utility for the integrated test. The commissioning plan lists the hold points and names who has to be present: the CxA always, often the owner, sometimes the engineer of record, the equipment vendor, and on a Tier job the Uptime Institute witness. Miss the witness and the test does not count, no matter how well it went.

Witnessing only works on notification. The plan should set a notice period, often several days, so the witnesses can travel and the test does not run with the wrong people in the room. The classic and expensive failure is the contractor running a witnessed test without proper notice to make a schedule date, then being told it has to run again because the CxA or the owner was not there to see it. That is rework with no defect, pure schedule and cost burned because the notification step got skipped.

The other half is documenting what was witnessed at the moment it happens. A witness signature against a script step, dated, beats a recollection a month later when the question is whether the generator actually accepted block load or just started. Sign at the test, not at the closeout meeting.

Why commissioning gets compressed, and how to protect it

Commissioning is the last activity before turnover, so it absorbs every slip ahead of it. When the gear ships late, when the install runs over, when the controls integration drags, the date the owner wants the building does not move, and commissioning gets squeezed into whatever window is left. That is the structural reason a program that was planned for twelve weeks gets eight, and the eight get eaten by rework nobody planned for.

The honest driver of the schedule is not the number of tests. It is how much rework the testing uncovers. A clean plant flows. A plant full of deficiencies turns every test into find, fix, re-test, and that loop is where the weeks disappear. You cannot schedule your way out of a building that was not ready to be tested.

Protecting commissioning is a scheduling decision made at the start, not a heroic recovery at the end. Build the commissioning sequence into the master schedule as its own track, tied to construction milestones, with the integrated test window fenced off and defended. Front-load the levels that can run early so the only thing left at the end is the integrated test that genuinely needs the whole plant. When commissioning gets compressed, the integrated systems test is the first thing to go: shortened, run at partial load, or quietly broken into pieces that never get strung back together. The schedule you protect is the IST window, because that is the one you cannot fake.

The integrated systems test, the keystone of the program

The integrated systems test is the full-plant test that simulates a real failure and proves every system reacts together to keep the critical load up. It is the keystone of data center commissioning because it is the only test that puts the whole sequence under a clock and a load at the same time. Every system can pass its own functional test and the plant can still fail the IST, because the failure lives in the handoff, not in any one box.

A real IST loads the plant to a defined percentage of design with load banks, because there are no servers yet to provide the load. Then it drops the utility. The UPS and any battery energy storage carry the load through the gap, the generators start and parallel and take block load within the time the sequence allows, the transfer scheme moves the load without dropping the critical bus, and then the utility is restored and the plant transfers back, also without a drop. Along the way you fail individual components on purpose, a generator, a UPS module, a feeder, to prove the redundancy the one-line claims is real. The mechanics of the electrical side, the failure scenarios, and how the load banks are sized and staged are covered in the data center electrical commissioning guide and the load bank test acceptance criteria guide.

From the program side, the rule is simple and it is the one to hold the line on. A plant that never had a complete, continuous integrated test has never actually been proven. An IST that hit a wall at hour three, got broken into pieces to save the night, and never got strung back into one run is not an IST. That is the result to refuse to sign, no matter what the schedule wants.

Turnover and the systems manual

Turnover is where the project hands the building to the people who will run it, and the deliverable is a record, not a handshake. The commissioning portion of turnover is the commissioning record: the plan, the test reports at every level, the signed scripts, the integrated test results, and the deficiency log closed out with evidence. Around it sits the rest of the closeout, and commissioning is usually the team that verifies the rest is actually present and usable.

The systems manual is the operations-facing document that ASHRAE Guideline 0 and the broader commissioning guidance call for, and it is more than the equipment cut sheets. It pulls together the OPR and BOD so the operator knows what the building was meant to do, the final sequences of operations, the as-left settings, the operation and maintenance information, the recommended maintenance and testing intervals, and the training records. The as-builts have to reflect what was actually installed, including every field change, because the operations team will trust those drawings the day something fails at 2 a.m.

Two parts of the turnover carry the most weight later. Training is the one that gets rushed and the one the operators remember, because a building handed to a team that was never trained on its sequences runs on guesswork until something breaks. And the warranty clock starts somewhere defined in the contract, often at substantial completion or beneficial occupancy, so the dated commissioning records become the proof of when the gear was accepted and in what condition. A turnover package missing the as-lefts, the closed log, or the training records is missing its spine, and the owner inherits a building they cannot operate against a baseline.

What is recommissioning and ongoing commissioning?

Recommissioning is repeating the commissioning process on a building that was already commissioned, to confirm it still performs to the requirements after time, modifications, and drift. Ongoing commissioning, sometimes called monitoring-based or continuous commissioning, does the same job continuously instead of as a one-time event, using the building's own metering and trends to catch performance falling off before it becomes a failure. Retro-commissioning is the related case for a building that was never formally commissioned the first time.

Data centers need this more than most buildings because they never stop changing. Racks get added, loads grow, cooling setpoints get nudged, firmware gets updated, and the redundancy that was proven at turnover slowly stops matching the building as it actually runs. The ASHRAE guidance distinguishes the new-construction commissioning process from the process for existing systems and assemblies, and the existing-building track is what recommissioning follows. The Building Commissioning Association best practices similarly separate new-construction commissioning from existing-building and ongoing commissioning.

The owner handoff is the hinge between the two. A commissioning program that ends at turnover and never sets up the ongoing checks hands the owner a building that was perfect for exactly one day. The strong programs use the commissioning record as the baseline the operations team re-tests against on a defined interval, often tied to the maintenance the systems manual recommends, so the building is held to the OPR for its life and not just at acceptance.

The commissioning report and the owner record

The commissioning report is how the whole program rolls up into something the owner can actually use. It is not a marketing summary. It states what was commissioned, what was tested at each level, what was found, what was fixed, what remains open, and the CxA's professional opinion on whether the building meets the owner's project requirements. The good ones are blunt about the open items, because the owner needs to know what they are accepting, not be told everything is fine.

The hard part of the report is not writing it. It is that the report is only as good as the field records behind it, and on most jobs those records are scattered across checklists, photos, emails, and three different spreadsheets by the time anyone tries to assemble the closeout. The report gets stitched together by hand in the last two weeks, which is exactly when nobody has time, and the gaps in the field record become gaps in the owner's permanent baseline.

The way out is to build the report from the field records as the work happens, not after. When every pre-functional checklist, every signed functional step, every deficiency with its photo and its closure evidence, and every witness signature is captured against the system it belongs to, the owner-ready report assembles itself from the record instead of being reconstructed from memory. That is the role the field-record layer plays in this process: the turnover document is a view of the field data, so the owner inherits a baseline that is complete because it was captured, not because someone had a heroic month at the end.

Commissioning metrics and the readiness gate

Commissioning is measured by a small set of numbers that tell the owner how close the building really is, and the most honest one is open versus closed deficiencies. Total findings, how many closed with re-test evidence, how many open, and of the open ones how many are severity-one items that affect operation or safety. A program reporting ninety percent of scripts complete while sitting on a stack of open critical deficiencies is not ninety percent done. It is gated on the items that can still hurt someone.

Percent complete by level is the other working metric: how many systems have passed each level, so you can see the bow wave of work moving toward the integrated test and spot the discipline that is behind before it stalls the whole IST. A daily or weekly commissioning huddle that puts these numbers in front of the team, the open critical items, the witnessed tests due, the systems blocking the next level, is how a large program stays coordinated across many parties. The huddleboard-dc tool exists to run that commissioning readiness board so the open items and the next hold points are visible to everyone instead of living in the CxA's head.

The number that decides turnover is the readiness gate. The plant is ready when the integrated test has run complete and continuous at design load and every severity-one deficiency it produced is closed with re-test evidence. Not when the schedule says, and not when the percentage looks good. The gate is the open critical list at zero against a real IST result, and that is the one metric worth refusing to round up.

What goes wrong in data center commissioning?

The failures in a commissioning program are predictable, and they cluster, because they all come from the same pressure: the building is late, the owner wants it, and commissioning is the last thing standing between the GC and the milestone. Knowing the pattern is how you catch it before it becomes the owner's problem.

Schedule compression is the root. Commissioning absorbs the slip ahead of it, the window shrinks, and the work that needs the whole plant gets squeezed. From that, the specific failures follow. Deficiencies get closed on paper without a re-test, so the log looks clean and the problems stay in the building. The integrated test gets shortened, run at partial load, or broken into pieces that never get strung together, so the plant is never actually proven under a real failure at load. Witnessing gets skipped or run without notice, so tests either do not count or have to run again. And underneath all of it, the deficiency log lives in someone's personal spreadsheet, so when that person rolls off the job the institutional memory of what was actually fixed leaves with them.

There is one more that does not come from schedule: the independence of the CxA gets compromised, usually by the agent reporting to the GC instead of the owner. When the checked party controls the checker, the findings that threaten the milestone are the ones that get softened, and the whole point of an independent process is gone. Every item in the common-mistakes list below is one of these wearing different clothes.

Field checklist

What to document

The record is judged by whether someone two years out can reconstruct what was proven and operate the building against it. Capture the requirements, the plan, the results at every level, the deficiency log closed with evidence, and the turnover package. The table is the spine of a commissioning record set.

Common mistakes

• Treating commissioning as a final inspection instead of a process that runs across the whole job.
• Starting without a current OPR and BOD, leaving the CxA to invent acceptance criteria nobody can defend.
• Letting the CxA report to the general contractor instead of the owner, which erases the independence of the check.
• Compressing commissioning to recover schedule slip, so the integrated test gets shortened or run at partial load.
• Closing deficiencies on paper without re-testing the change that was supposed to fix them.
• Running each system's functional test but never stringing them into one complete, continuous integrated test.
• Running a witnessed hold point without notice, so it does not count and has to run again.
• Keeping the deficiency log in one person's spreadsheet, so the record walks off the job when they do.
• Reporting percent complete while sitting on open severity-one deficiencies and calling the plant nearly done.
• Turning over a record set missing the as-lefts, the closed log, or the training, so the owner has no baseline to operate against.

Standards and references

The commissioning process framework comes from ASHRAE Guideline 0, the commissioning process, developed with the National Institute of Building Sciences, which describes the role of the commissioning authority and the framework for the OPR, the BOD, the commissioning plan, the specifications, the procedures, and the reports. ASHRAE Standard 202, the commissioning process for buildings and systems, carries the same process into an ANSI standard. The HVAC technical application sits in ASHRAE Guideline 1.1 for new HVAC and refrigeration systems, and the existing-systems and ongoing-commissioning processes follow the related ASHRAE guidance, with ASHRAE Guideline 0.2 commonly cited for existing systems and assemblies. Confirm the current edition of any of these against the project documents, because the numbers and titles shift between cycles.

Best-practice guidance for running the process comes from the Building Commissioning Association, the BCxA, which publishes best practices separating new-construction, existing-building, and ongoing commissioning. The contract requirement for commissioning lives in the project specifications, typically in Division 01 under the commissioning sections commonly numbered 01 91 00, and that spec plus the commissioning plan control the actual scope. Where a project pursues LEED, the commissioning credits for fundamental and enhanced commissioning impose their own requirements on the OPR, BOD, and process by topic.

For data centers chasing an Uptime Institute Tier certification, the constructed-facility certification is a witnessed demonstration that the built plant meets its claimed Tier, so the integrated systems test and the redundancy demonstrations do double duty as both commissioning and the Tier witness. The discipline standards that govern the actual tests, NETA acceptance testing for the electrical scope, NFPA 110 for standby power, the IEEE protection and grounding references, and the ASHRAE thermal guidelines for the cooling scope, are covered in the discipline guides. The commissioning plan and the contract documents control scope. The standards give the framework.

Terms and acronyms

Commissioning carries its own vocabulary, and the same word can read differently across a commissioning plan, a spec, and a Tier document. The terms below are the ones that travel across the whole process.

Cx / CxA

Commissioning, and the commissioning authority or agent who plans, witnesses, and signs off the process, independent of the installer

OPR

Owner's project requirements, the measurable statement of what the owner needs the building to do, the yardstick for every test

BOD

Basis of design, the design team's documented assumptions and decisions for how the design meets the OPR

PFC

Pre-functional checklist, the static de-energized verification that a system is installed correctly and ready to energize

FPT

Functional performance test, the energized demonstration that a single system performs its sequence of operations

IST

Integrated systems test, the full-plant level-5 test that proves all systems work together through a simulated failure at load

Deficiency / issues log

The tracked list of every finding from identify through verify and close, the program's most important operational artifact

L1 to L5

The commissioning levels, from factory test through receiving, installation, functional, to integrated test, with boundaries set by the plan

Hold point

A step the work cannot pass until the required witness has seen it and signed

Systems manual

The operations-facing turnover document pulling together the OPR, BOD, sequences, settings, O&M, and training

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• How long does data center electrical commissioning take?
• What is bypass airflow in a data center?
• What is data center commissioning?
• What is an integrated systems test?
• What is the difference between a PDU and an RPP in a data center?