Where reliability is actually won
A modern oil-immersed power transformer leaves a serious factory in the best condition it has ever been in: fully assembled, type-tested to IEEE C57.12, routine-tested to the owner's specification, vacuum-dried, oil-filled, sealed, and signed off on a witness package. From that moment, every mile and every degree on the journey to a Midwest substation or a Texas data-center campus is a chance for that condition to drift.
Utilities, IPPs, ITC-grade transmission owners, AI data-center developers and industrial asset managers across the US increasingly agree on a quiet truth the wider industry has known for years: the overwhelming majority of in-service issues in the first two years are not built into the unit at the factory. They are introduced on site — by storage, by handling, by the speed of energization, by a neutral grounding decision taken before the SAT was finished.
Installation support is how a serious supplier closes that gap. In a market where large power transformer lead times now routinely run 80 to 120 weeks, it is also the cheapest insurance line item an asset owner can specify.
The US is a harder operating environment than the spec sheet suggests
Three forces are compounding in the US transformer market in a way they were not five years ago.
LPT lead times. The shortage of large power transformers — driven by domestic core steel constraints, hyperscale AI data-center buildout, and ERCOT-PJM-MISO interconnection queues — means that a unit damaged in year three cannot be replaced in year three. It will be replaced in year four or five. The economic case for getting installation right has never been stronger.
EPC labor compression. The industrial labor market is the tightest it has been in a generation. EPCs and electrical contractors are stretched thin. Site crews are good, but they are not infinite, and they rotate. The discipline of a single manufacturer engineer on a unit from FAT to energization is the cheapest way to standardize execution.
Climate and storm exposure. Texas thunderstorms, Gulf-Coast hurricanes, Midwest tornadoes, California wildfires, Appalachian ice storms — the asset population in service today is being asked to handle conditions that the old spec sheets did not center. The pre-energization baseline is what makes future condition monitoring meaningful when those events happen.
None of this is a criticism of US contractors. It is a description of the actual conditions that a transformer specified to IEEE C57.12 has to survive between leaving the factory gate and being energized onto a 13.8, 34.5, 138, 230 or 345 kV system.
What great installation support looks like on a US site
A good installation supervision package is concrete. Six things should always be visible:
1. Site-readiness review before delivery. A manufacturer engineer surveys the pad: levelness, oil-containment integrity, ground grid continuity per IEEE 80, lightning arrester lead length, fence clearances, and approach for the SPMT or rail offload. Issues raised on day minus 30 are cheap; issues raised on day plus 1 are not.
2. IEEE C57.93-aligned storage protocol. Units stored for more than 30 days need an active nitrogen blanket, a documented inspection cadence, and dew-point logging. Texas humidity and Midwest summers will move moisture into a sealed tank faster than most schedules allow for. C57.93 is the standard; following it is what separates a manageable laydown from a full reprocess.
3. Supervised vacuum oil filling. Treatment-plant calibration, dew-point window verification, filtration to better than 3 micron, vacuum hold-and-rise tests, settle-time-respected DGA baseline — all witnessed and logged. The settle-time discipline matters: a DGA sample drawn before the unit has settled gives a false baseline that follows the asset for years.
4. Neutral grounding, arrester and surge counter commissioning. Correct neutral grounding strategy on Y-Y GSUs verified before energization. Lightning arrester lead length kept inside the IEEE C62 envelope. Surge counters zeroed and recorded. These are the small things that field engineers know to check and busy crews sometimes do not.
5. NETA acceptance testing. Insulation resistance, power factor, winding resistance, turns ratio, exciting current, DGA baseline, partial-discharge SAT where the specification requires it, tap-changer operation through full range. NETA acceptance testing standards are the operating language of US commissioning; the test pack should drop straight into the owner's asset-life file.
6. Owner's-engineer FAT and SAT. A single engineer present at both ends — the factory test bay and the energization sequence. The continuity is what removes the gaps that 80-week lead times cannot afford.
Why this is now a procurement question, not a site question
The US industry has historically treated installation as the EPC's problem and the manufacturer's afterthought. That model is breaking under three pressures at once: LPT lead times, AI-data-center schedule discipline, and the IRA-driven domestic-content emphasis that makes every unit harder to replace.
A 100 MVA GSU that fails internally in year three costs the owner the original capital, the unplanned outage, the emergency-replacement premium (rarely under a year of waiting), and the lost-revenue economics on the asset behind it. For an AI data-center owner, that lost revenue scales linearly with megawatt commitments to hyperscale tenants. The total exposure is rarely under four times the original PO.
Forward-leaning US owners — major IOUs, ITC-grade transmission operators, hyperscale developers, and the larger industrial customers — are starting to specify installation supervision and NETA acceptance testing explicitly in their procurement packages. EPCs and suppliers that bring it as a default rather than a change-order item are winning long-term framework positions.
How ETS partners with US customers from delivery to energization
ETS treats installation as part of the product, not a service line that is sold afterwards. A single ETS engineer is named on the project from FAT through to energization witness, and owns:
- A pre-shipment site-readiness checklist agreed with the asset owner and EPC
- IEEE C57.93-aligned storage and laydown management
- Supervised vacuum oil filling with an independent oil report and DGA baseline
- NETA-format acceptance test pack ready for the owner's asset-life file
- HV commissioning with partial-discharge SAT where the specification requires it
- A 24-month installed-asset support window backed by ETS service engineers covering the US East and growing across the Midwest
- An owner's-engineer service tier for clients who want manufacturer-independent oversight
Deeper detail on the individual services is on the Installation, Testing & Commissioning, and Consulting pages. The Standards & Compliance library covers the IEEE C57.12, C57.93, NETA ATS and utility-specific test packages we work to.
What to specify in your next PO
A short checklist any US procurement engineer can drop into a transformer tender:
1. Installation supervision by a manufacturer engineer, named in the bid
2. IEEE C57.93-aligned storage protocol for any laydown longer than 30 days
3. Supervised vacuum oil filling with dew-point and filtration acceptance limits
4. NETA ATS acceptance test pack, with DGA baseline and PD test for units rated above 69 kV
5. Owner's-engineer FAT and SAT continuity
6. 24-month installed-asset warranty extension contingent on the supervision being used
None of these add meaningful cost on a unit above 5 MVA. All of them measurably reduce the chance of an early-life event in a market where replacement is now an 80-to-120-week proposition.
The takeaway
A transformer is only as reliable as the worst day it spent between the factory and energization. In today's US market, that worst day is rarely about manufacturing — it is about a long laydown, a compressed energization window, a settle time skipped, or a neutral grounding decision taken in a hurry. Installation support is the engineering discipline that closes the gap. At ETS, it is part of how the product is delivered, not an upsell after the fact.
If you have a US project where the energization date is tight, the asset replacement window is unforgiving, or the AI-data-center schedule cannot slip, talk to us early. The smartest engineering decision an owner can make is the one taken before the unit ships.
