Buying a power transformer in 2026 comes down to five numbers: the kVA you actually need, the voltage class, the no-load and load losses in watts, the standards your inspector will enforce, and the delivery date you can live with. Nail those five and everything else is paperwork. Miss one and you’ll pay for it twice — once on the invoice, once on the power bill.
This guide walks you through all five, in the order a real project actually needs them. It includes the 2026 lead-time picture, the US efficiency compliance date most buyers still haven’t written into their specs, and the sizing math that stops you from overbuying and underbuying at the same time.
The 60-second version
If you only skim one section, skim this one.
- Size from measured demand, not from nameplates. Add up connected kW, apply a diversity factor, divide by power factor, add 15–25% growth, then round up to a standard rating. Non-standard ratings cost 15–30% more and sit in the queue longer.
- Pick the insulation based on where it sits, not on habit. Outdoor and high-kVA almost always means liquid-filled. Indoor, occupied, or fire-rated spaces mean dry-type or ester-filled.
- Losses are the real price tag. Over a 30-year life, lost energy routinely costs more than the transformer itself. Buy in watts, not in “high efficiency.”
- US compliance date is April 23, 2029 — and it applies to the manufacture date, not your PO date. If you order in 2026 on a two-year lead time, this is your problem now.
- Budget 30–50 weeks for distribution-class, 75–110 weeks for substation-class, and 100+ weeks for GSU units. Pre-2020, these were 8–12 weeks.
- Impedance is a two-sided constraint. Too low and your switchgear can’t interrupt the fault current. Too high and your voltage sags on motor starts.
- A 60 Hz transformer on a 50 Hz system will overheat. The reverse is usually fine. Most guides get this backwards.
- Ask for type test reports, not “complies with IEC 60076.” Compliance without a report is a marketing claim.
- Send every supplier the same RFQ package. If the quotes aren’t line-by-line comparable, you didn’t send a spec — you sent a wish.
Who this is for
This is written for engineers, facility managers, and procurement leads specifying a three-phase transformer above roughly 500 kVA for an industrial plant, commercial campus, data center, renewable interconnection, or utility substation.
If you’re buying a 25 kVA single-phase pole-mount, you can stop at Step 2. If you’re buying a 300 MVA autotransformer for a transmission intertie, this guide will be too light — you already have a consultant.
One more note on vocabulary before we start.
Step 0: Get the word straight, because “power transformer” means three different things
Here’s the thing nobody says out loud: “power transformer” is not a precise product category. It gets used three ways, and suppliers exploit the ambiguity.
| Usage | What it actually means | Typical range |
|---|---|---|
| Loose / marketing | Any transformer that isn’t tiny | Anything |
| IEC / IEEE technical | Transmission and sub-transmission class, usually with OLTC | ~10 MVA to 1,000+ MVA |
| US legal (DOE) | A distribution transformer: input ≤ 34.5 kV, output ≤ 600 V, 60 Hz, 10–2,500 kVA liquid-immersed or 15–2,500 kVA dry-type | 10–2,500 kVA |
That third row matters enormously, and it’s per the US Department of Energy’s own definition at 10 CFR 431.192, quoted on DOE’s distribution transformer standards page (accessed September 2026). If your equipment falls outside it — a rectifier transformer, a grounding transformer, a welding transformer, an autotransformer, a unit with a tap range of 20% or more — the federal efficiency standard doesn’t apply to you at all. Buyers waste real money specifying 2029 compliance on equipment that was never covered.
So before you write a single line of the spec: decide which of the three you mean, and write it down. Then send that to every supplier so they’re quoting the same animal.
Step 1: Size it from load, not from nameplates
This is where most budgets die. Two failure modes: overbuy because someone added up every nameplate on site, or underbuy because someone used last year’s average demand.
The formula
Demand kVA = (Connected kW × Diversity factor) ÷ Power factor
Design kVA = Demand kVA × (1 + growth margin)
Rating = next standard IEC 60076-1 rating above Design kVA
Worked example. A food plant with 800 kW of connected load:
- Diversity factor 0.8 (typical for mixed industrial — not everything runs at once)
- Power factor 0.85
- Demand kVA = 800 × 0.8 ÷ 0.85 = 753 kVA
- Growth margin 20% → Design kVA = 904 kVA
- Next standard rating → 1,000 kVA
Standard three-phase ratings per IEC 60076-1 run: 100, 160, 250, 315, 400, 500, 630, 800, 1,000, 1,250, 1,600, 2,000, 2,500, 3,150 kVA — and 3,750 and 5,000 kVA are standard in North American pad-mount practice.
Don’t skip the motor starting check
Capacity and starting current are different constraints, and on motor-heavy sites, starting wins.
Take that same plant: 400 kVA of running load, and a 90 kW motor that starts across the line.
- Motor running: 90 kW ÷ 0.9 pf = 100 kVA
- Motor starting (≈6× current at ≈0.2 pf): ≈600 kVA for a few seconds
- Momentary demand: 400 − 100 + 600 = ≈900 kVA
A 630 kVA unit would see a momentary 143% load. Transformers tolerate short-time overload — but the question that actually matters is voltage dip, and for that you need a study, not a rule of thumb. Across-the-line starts above roughly 20–25% of transformer kVA are where engineers start reaching for soft starters or VFDs.
Two sizing traps
The “we’ll just size it big” trap. Transformer efficiency peaks somewhere in the 40–80% load band. An oversized unit runs cooler and lasts longer, yes — but no-load loss is charged 24/7/365 regardless of load. Buy headroom on purpose, not by accident.
The “custom is better” trap. Ordering 3,200 kVA instead of the standard 3,750 kVA doesn’t save you money. Non-standard ratings typically add 15–30% to the price and lengthen the queue, because cores and tanks aren’t tooled for them.
Step 2: Choose the insulation — oil, dry, or ester
The choice is really about fire code and location, and everything else follows from that.
| Liquid-immersed (mineral oil) | Dry-type (cast resin / VPI) | Liquid-immersed (ester fluid) | |
|---|---|---|---|
| Typical range | 500 kVA – 800 MVA | 15 kVA – 30 MVA | Similar to mineral oil |
| Where it goes | Outdoor substations, yards | Indoor rooms, hospitals, data centers, tunnels | Fire-sensitive or environmentally sensitive outdoor sites |
| Fire behavior | Combustible; containment usually required | Self-extinguishing, no liquid | Fire point typically >300 °C, far above mineral oil |
| Upfront cost | Baseline (1.0×) | Roughly 1.4–1.8× | Roughly 1.2–1.5× |
| Maintenance | Oil sampling, DGA, gasket management | Visual + cleaning, minimal | Similar to oil, but moisture-tolerant |
| Heat handling | Best overload capability | Limited; AF fans help but add noise | Comparable to mineral oil |
| Watch out for | SPCC containment (40 CFR Part 112), leak liability | Higher no-load loss at same rating; derate at high altitude | Retrofit compatibility with existing gaskets and paints |
Dry-type costs more per kVA, but it deletes the entire oil-handling program — containment, sampling, disposal, spill liability. In an occupied building, the premium usually pays for itself in avoided fire-rated vault construction.
Step 3: Lock the electrical specs that actually get mis-specified
Here’s the short list of parameters where I see quotes come back wrong most often.
| Parameter | What to specify | Why it bites |
|---|---|---|
| Primary / secondary voltage | Exact kV or V on both sides, plus system frequency (50 or 60 Hz) | A 34.5 kV unit is not a 33 kV unit. Tolerance matters. |
| BIL (Basic Insulation Level) | e.g. 170 kV BIL for 33 kV class | Under-specifying here is how transformers die in lightning season |
| Vector group | Dyn11, YNd11, Yyn0, etc. | Mandatory for paralleling. Get it wrong and you can’t parallel, ever |
| Impedance (Z%) | 4–6% distribution; 6–12%+ industrial/furnace | Low Z = better regulation but higher fault current. Check your switchgear’s interrupting rating first |
| Tap changer | DETC (±2 × 2.5%) or OLTC (e.g. ±8 × 1.25%) | Rural feeders with big swings need OLTC. Urban feeders usually don’t |
| Temperature rise / insulation class | e.g. 65 K rise, Class F or H | Drives life expectancy and overload headroom |
| Altitude & ambient | Site elevation in meters, max ambient °C | Thin air cools worse. Above 1,000 m you derate or redesign |
| Noise (LWA) | Guaranteed sound power level in dB(A) | The #1 complaint source on commercial sites |
| Enclosure / IP or NEMA | IP20–IP54, NEMA 1/2/3R, C4/C5 corrosion class | Coastal sites need C5-M. This is not optional near salt air |
Two things most guides get wrong
60 Hz vs 50 Hz is not symmetric. Core flux is proportional to voltage ÷ frequency. Run a transformer designed for 60 Hz on a 50 Hz system and flux rises 20% — the core saturates, magnetizing current spikes, and it cooks. Running a 50 Hz design on 60 Hz is generally fine. So the risky direction is the one people assume is safe.
You probably don’t need a K-rated transformer. K-factor derating was designed for an era of 6-pulse drives and legacy rectifiers. Modern IGBT-based UPS and VFD front ends routinely land under 5% THD. Specify K-13 because the one-line has “VFD” written on it and you’ve just added cost, size, and heat for nothing. Specifying it where harmonics genuinely are high — large DC drives, older 6-pulse rectifiers, big electrolysis loads — is correct.
Step 4: Read the loss numbers like a CFO
“High efficiency” is not a spec. Watts are.
There are two numbers, and they behave completely differently:
- No-load loss (core / iron loss) — measured in watts or kW. Charged 8,760 hours a year, whether you’re running load or not. Never switches off.
- Load loss (copper / winding loss) — also in watts, but it scales with the square of load. At 50% load you pay 25% of the rated load loss.
Worked example — 1,500 kVA unit, 60% average load factor, $0.12/kWh:
| Standard design | Low-loss design | |
|---|---|---|
| No-load loss | 2.0 kW | 1.4 kW |
| Load loss at 75 °C | 14.0 kW | 13.0 kW |
| No-load energy/yr | 17,520 kWh | 12,264 kWh |
| Load loss energy/yr (0.6² × kW × 8,760) | 44,150 kWh | 40,997 kWh |
| Total losses/yr | 61,670 kWh | 53,261 kWh |
| Annual cost | $7,400 | $6,391 |
| 30-year undiscounted cost | $222,000 | $191,700 |
That low-loss design saves about $1,009 a year, or roughly $30,000 over 30 years. If it costs $8,000 more, buy it. If it costs $40,000 more, don’t.
Utilities have a formal version of this called loss capitalization — commonly cited evaluation ranges are $2,000–$6,000 per kW of no-load loss avoided and $400–$1,200 per kW of load loss avoided (figures widely used in North American utility procurement; confirm the current values your own rate case uses). Note the spread: no-load loss is worth 3–5× more per kW. That’s because it never stops.
One counterintuitive note on amorphous cores
Amorphous metal cores can cut no-load loss by 40–70% versus grain-oriented silicon steel. Genuinely excellent — but they’re typically a few dB louder, and they’re physically larger. If the unit sits near a classroom, a hospital ward, or a bedroom wall, an SCB13-style silicon steel design plus anti-vibration mounts and acoustic treatment will usually hit the noise limit more cheaply than an amorphous unit will. Combine both and you’ve spent a lot to solve a problem you created.
Step 5: Compliance in 2026 — and the date almost everyone gets wrong
This is the single most misunderstood part of US transformer buying right now. Let me be precise.
The facts (US DOE):
- DOE published its final rule on distribution transformer efficiency standards in the Federal Register on April 22, 2024 (89 FR 29834).
- The rule’s effective date was July 8, 2024.
- Compliance with the amended standards is required on and after April 23, 2029.
- The standards live at 10 CFR 431.196; the test procedure is at 10 CFR 431.193 and Appendix A.
- DOE has required efficiency standards for distribution transformers since 2007 — this isn’t new, it’s an update.
- DOE stated in its April 2024 analysis that roughly 75% of the market can comply using grain-oriented electrical steel (GOES), avoiding a wholesale shift to amorphous supply.
So: there is no “DOE 2026 efficiency standard.” If a vendor or a blog told you the 2026 update is live, that’s wrong. The date is April 23, 2029.
Why a 2029 date is a 2026 problem
The compliance obligation attaches to the date of manufacture or import, not the date you signed the PO. On a 30–50 week distribution lead time, and up to 100+ weeks for substation class, orders placed in 2026 and 2027 are the first ones that can land on the wrong side of the line. If you take delivery of a non-compliant unit in 2029, you own a very expensive paperweight.
Minimum efficiency, before and after April 23, 2029
Three-phase liquid-immersed, certified at 50% per-unit load (industry summary of 10 CFR 431.196 — verify against the current CFR tables before using these in a bid):
| Rating | Manufactured before 4/23/2029 | Manufactured on or after 4/23/2029 |
|---|---|---|
| 500 kVA | 99.35% | 99.38% |
| 1,000 kVA | 99.43% | 99.46% |
| 1,500 kVA | 99.48% | 99.51% |
| 2,500 kVA | 99.53% | 99.55% |
| 3,750 kVA | Not covered | 99.54% |
| 5,000 kVA | Not covered | 99.53% |
Two things worth noticing. First, the increments look tiny — but at 500 kVA, 0.51% loss versus 0.41% loss is a ~20% reduction in losses, and roughly 2,200 kWh/year per unit at 50% load. Second, coverage extends to 3,750 and 5,000 kVA in 2029; those sizes simply have no federal minimum today.
The test-point trap
Liquid-immersed efficiency is certified at 50% load; low-voltage dry-type is certified at 35% load. And the reference temperatures differ — 20 °C for no-load loss, 55 °C for load loss on liquid units. Which means:
A dry-type efficiency number and a liquid-filled efficiency number are not directly comparable, and neither describes a unit held near nameplate.
If your site runs at 70% load, both certification points flatter the equipment. Ask for the loss values in watts, then do your own arithmetic at your load factor. That’s Step 4 again, and it’s where the real money is.
Other US requirements worth writing into the spec
- NEC Article 450 (NFPA 70) — installation, overcurrent protection, ventilation, and clearances for transformers over 600 V.
- IEEE C57.12.00 (liquid-immersed), C57.12.01 (dry-type), C57.12.90 (test code), C57.12.20 (pad-mounted), C57.91 (loading guide).
- NERC CIP-014 — physical security for transmission substations, if you’re utility-side.
- EPA SPCC, 40 CFR Part 112 — oil containment above threshold volumes.
- IEC 60076 series — the reference outside North America: -1 general, -2 temperature rise, -3 insulation levels, -5 short-circuit withstand, -7 loading guide, -10/-10-1 sound levels, -11 dry-type.
Step 6: Total cost of ownership — the 10-year math
Indicative FOB price ranges reported across 2025–2026 supplier and trade sources. Treat these as scoping numbers, not quotes:
| Rating | FOB China (indicative) | EU / global brand (indicative) |
|---|---|---|
| 1,000 kVA distribution | $6,000 – $14,000 | $15,000 – $28,000 |
| 10 MVA, 33/11 kV power transformer | $90,000 – $180,000 | $180,000 – $320,000 |
Freight, insurance, duties, and rigging typically add 8–15% on top of FOB. And price momentum is real: US Bureau of Labor Statistics producer price index for power transformers (series WPU117409) ran roughly 80% above its flat 2014–2021 level through 2025, driven by GOES and copper costs, 2029-driven redesigns, and relentless demand.
Now put that next to the losses. Using the 1,500 kVA example from Step 4:
| Cost line | Year 0 | 10 years |
|---|---|---|
| Purchase (1,000–1,500 kVA class, indicative) | $25,000 | $25,000 |
| Freight, duties, installation | $4,000 | $4,000 |
| Lost energy at $0.12/kWh | — | $74,000 |
| Maintenance (oil sampling, thermography, cleaning) | — | $6,000 |
| Total | $29,000 | $109,000 |
| Purchase price as share of 10-year cost | — | 27% |
The sticker price is roughly a quarter of what you’ll pay. Compare quotes on 10-year cost, not on invoice.
Step 7: Lead times — plan backwards from your energization date
This is the part that breaks schedules. Pre-2020, a distribution transformer was an 8–12 week line item. It is not anymore.
| Class | 2026 lead time | Pre-2020 norm |
|---|---|---|
| Dry-type, under ~3.75 MVA | 8–16 weeks | 4–8 weeks |
| Distribution, liquid, under 500 kVA | 12–20 weeks | 4–8 weeks |
| Distribution / pad-mount, 500 kVA–2.5 MVA | 16–65 weeks | 8–12 weeks |
| Distribution, 2.5–10 MVA | 24–36 weeks | 12–16 weeks |
| Substation / power, 5–50 MVA | 75–110 weeks | 20–30 weeks |
| Generator step-up, >50 MVA | 100–150+ weeks | 40–52 weeks |
Ranges are wide because published figures disagree — some trackers report distribution units improving to 26–40 weeks, others report 30–50, and pad-mount three-phase is moving the wrong direction. Get a firm slot date in writing from your actual supplier; don’t plan off an industry average.
Underneath all of it: grain-oriented electrical steel. GOES is made by a handful of mills globally, and the same coils feed distribution transformers, large power transformers, GSUs, and the data-center buildout simultaneously. Domestic mills supply an estimated minority of US GOES demand. That’s the real bottleneck — not factory floor space.
What actually works:
- Freeze the spec early. Every late change resets your place in the queue.
- Order 18–24 months ahead of energization for large units; 30–36 months for specialized ones.
- Consider remanufactured for schedule-critical replacements. Fully rebuilt and ANSI-tested units are quoted at roughly 50–70% of new cost at 4–6 weeks, versus 30–50 weeks.
- Hold buffer stock of your most common kVA/voltage combinations if you run a fleet.
- Put delivery penalties in the contract, and require monthly production updates.
Step 8: Vetting a supplier — 12 questions that separate factories from brochures
Most “manufacturers” online are traders. That’s not automatically bad, but you should know which one you’re dealing with and price accordingly.
Ask these:
- Can I see the type test reports for this design, with the lab name and date? (Not a certificate — the actual report.)
- Is the factory ISO 9001 certified? Can I see the certificate number?
- Will you do a factory acceptance test (FAT) I can attend or watch by video?
- Where is the core wound — in-house or bought in? (This is the real “how much is actually yours” question.)
- What’s your current GOES and copper stock position? (No stock = no schedule certainty.)
- Who does the routine tests, and can I have the raw readings?
- What’s in the warranty, and what’s the response time?
- Can you provide three references from projects of similar voltage class in my region?
- Do you hold UL listing for the US market where it’s needed? (Answer “available on request” should be followed up on.)
- What exactly is excluded from your quote — delivery, accessories, spares, commissioning, training?
- Are spare parts stocked for 10+ years?
- What happens to my delivery date if I change the vector group in week 6?
Red flags: vague “complies with all international standards,” no test lab on site, pressure to skip the FAT, unwillingness to give references in your market, quotes with no loss values stated, and delivery promises shorter than every competitor’s by a wide margin.
Step 9: The RFQ package
Send this to every supplier and the quotes come back comparable. Skip it and you’ll spend two weeks doing spreadsheet archaeology.
Required in every inquiry:
- Application and site description
- Rated power (kVA/MVA), phases, frequency
- Primary and secondary voltage, and BIL
- Vector group and neutral grounding arrangement
- Impedance (Z%) and tolerance
- Cooling class (ONAN / ONAF / OFAF / AN / AF)
- Guaranteed no-load loss in W and load loss in W at 75 °C — maximum values, not typical
- Tap changer type and range
- Insulation class and temperature rise
- Ambient temperature, altitude, and installation (indoor/outdoor)
- Enclosure rating (IP / NEMA) and corrosion class (C4 / C5-M)
- Guaranteed sound power level LWA in dB(A)
- Applicable standards (IEC 60076 series and/or IEEE C57.12 series)
- Required tests: routine, type, and special
- Documentation: drawings, test reports, manuals, spares list
- Quantity, target delivery date, and delivery terms (Incoterms)
- Warranty terms and after-sales expectations
That’s 20 lines. It will save you more money than any negotiation.
Five mistakes we see over and over
- Buying to nameplate sum. You’ll overpay by one rating step and then pay no-load losses on the excess for 30 years.
- Comparing efficiency percentages instead of watts. Different test points, different reference temperatures, no common basis.
- Ignoring the 2029 manufacture date. Units ordered in 2026 arrive in 2028–2029. Read Step 5 again.
- Specifying impedance in isolation. It has to be coordinated with your switchgear’s interrupting rating and your voltage-dip tolerance.
- Treating noise as an afterthought. Noise complaints arrive after commissioning, and retrofitting acoustic treatment costs far more than specifying LWA up front.
Bottom line: the nine steps, in one loop
Buying a power transformer in 2026 is a sequence, not a checklist you can do out of order. Get the vocabulary straight so everyone’s quoting the same product. Size from measured load with real diversity and growth factors. Choose insulation from fire code and location, not habit. Lock voltage, BIL, vector group, impedance, and taps before anything goes to bid. Buy losses in watts and run the 30-year arithmetic. Write the April 23, 2029 compliance date into the spec if you’re in DOE scope. Compare quotes on 10-year cost, not invoice. Plan backwards from energization using real slot dates. And send everyone the same 20-line RFQ.
Do those nine things and the transformer becomes what it’s supposed to be: the piece of equipment you never think about again.
Frequently asked questions
What size power transformer do I need? Take connected kW, apply a diversity factor (0.6–0.85 is typical for industrial loads), divide by power factor, add 15–25% for growth, and round up to the next standard IEC 60076-1 rating. For 800 kW at 0.8 diversity and 0.85 power factor, that’s 753 kVA → 904 kVA with growth → a 1,000 kVA unit. Then check motor starting separately, because starting current can be the binding constraint.
How much does a power transformer cost in 2026? Indicative FOB ranges from 2025–2026 trade sources: about $6,000–$14,000 for a 1,000 kVA distribution unit from a Chinese supplier and $15,000–$28,000 from a European or global brand; roughly $90,000–$180,000 versus $180,000–$320,000 for a 10 MVA 33/11 kV power transformer. Freight, duties, and installation typically add 8–15%. Prices have run roughly 80% above 2014–2021 levels on the US producer price index.
How long does it take to get a power transformer in 2026? Dry-type under about 3.75 MVA: 8–16 weeks. Distribution liquid-filled: 16–65 weeks depending on size and configuration. Substation-class 5–50 MVA: 75–110 weeks. Generator step-up above 50 MVA: 100–150+ weeks. Pre-2020, these were 8–12 weeks.
Should I buy oil-filled or dry-type? It’s mostly a fire-code and location decision. Liquid-filled wins outdoors and at high ratings — better cooling, better overload capability, lower cost per kVA. Dry-type wins indoors, in occupied buildings, hospitals, tunnels, and data centers, where it removes fire load and the entire oil-handling program at a premium of roughly 40–80%. Ester-filled liquid units are the middle path where you want outdoor performance with a much higher fire point.
What’s the difference between a power transformer and a distribution transformer? Technically, “power transformer” usually refers to transmission and sub-transmission class units above roughly 10 MVA, often with on-load tap changers. “Distribution transformer” refers to the final step-down to utilization voltage. Legally in the US, DOE defines a distribution transformer as having input ≤ 34.5 kV, output ≤ 600 V, 60 Hz operation, and a rating of 10–2,500 kVA (liquid) or 15–2,500 kVA (dry-type). In practice, suppliers use the terms loosely, so always pin down the actual ratings.
Does the DOE 2029 efficiency standard affect a purchase I’m making in 2026? Very likely yes. Compliance is required for units manufactured or imported on or after April 23, 2029, and that attaches to the manufacture date, not your purchase order. With distribution lead times of 30–50 weeks and substation lead times of 75–110 weeks, a 2026 order can easily be built on the wrong side of the deadline. Write the requirement into the spec now.
Which standards should a power transformer comply with? Outside North America, the IEC 60076 series: -1 general, -2 temperature rise, -3 insulation levels, -5 short-circuit withstand, -7 loading guide, -10 sound levels, -11 dry-type. In the US and Canada, IEEE C57.12.00, C57.12.01, C57.12.90, C57.12.20, and C57.91, plus NEC Article 450 for installation. US efficiency standards sit at 10 CFR 431.196.
Can I run a 50 Hz transformer on a 60 Hz system, or the other way around? A 50 Hz design on 60 Hz is generally fine. The reverse is dangerous: core flux is proportional to voltage divided by frequency, so a 60 Hz design on 50 Hz sees about 20% more flux, driving the core into saturation, spiking magnetizing current, and overheating. If you must, derate the applied voltage. This is the direction most people get backwards.
Is a K-rated transformer necessary? Usually not anymore. K-factor derating was developed for 6-pulse drives and legacy rectifiers. Modern IGBT-based UPS and VFD front ends often produce under 5% current THD, so a standard transformer is fine. Specify K-rated where harmonics genuinely run high — large DC drives, old rectifiers, electrolysis loads — and not merely because a VFD appears on the one-line.
How loud will it be, and how do I specify that? Specify the guaranteed sound power level LWA in dB(A), measured per IEC 60076-10. As a rough field conversion: in free-field conditions, sound pressure at 1 m is approximately LWA − 11 dB, and it drops another 6 dB each time you double the distance. Hard-walled transformer rooms reflect sound and push levels back up, and forced-air (AF) cooling adds several dB the moment the fans start. Anti-vibration mounts are usually the cheapest decibel you can buy, because structure-borne noise dominates.