Transformer Sizing and Inrush: A Practical Method

---

Sizing Is a Process, Not a Division

Dividing connected load by a power factor and rounding up gives a number, but it does not give a transformer that is correctly sized. A transformer has to carry the real design load for its whole life, survive the site's temperature and altitude, limit fault current to what the downstream equipment can take, energize without tripping its own protection, and do all of that efficiently. This guide walks through the sizing steps in the order they are normally done, then runs one complete example.

It builds on the Transformer Application Engineering guide, which covers impedance, cooling classes, and vector groups in more detail. Here the focus is the calculation sequence, which the Transformer Sizing and Inrush Calculator automates. Both North American (NEC and IEEE) and IEC practice are covered: each step gives the common method first and marks where the IEC approach differs, and there is a worked example for each system.

Step 1: Start From Design Load, Not Connected Load

The input to transformer sizing is the diversified design load in kVA, not the sum of nameplates. If you add every connected load you will oversize the transformer, pay for unused capacity, and often increase secondary fault current for no benefit. Work the load up with demand and diversity factors first, as described in Demand Load Calculation Basics, and keep the motor and non-linear load portions identified because they matter later for inrush, starting, and harmonics.

• Use kVA, not kW: a transformer's thermal limit is set by current, so the apparent power at the worst-case power factor is what counts
• Separate continuous and intermittent load: continuous load sets the thermal sizing, while large short-duration steps set the voltage dip and protection checks
• Record the source of every number: when the design is reviewed or revised, the basis of the load figure is the first thing people ask about

Step 2: Apply Growth and Redundancy

Two multipliers are applied to the design load before a size is chosen.

Future growth is a project decision, not a standard value. A growth allowance of 10 to 25% is common in industrial and data center work, but the right figure depends on how likely and how soon expansion is. Document it as an explicit line, not a hidden safety factor.

Redundancy changes what "carrying the load" means. If the load must continue after the loss of one transformer, each unit is sized for the post-failure case:

• N+1: if N units are needed to carry the load, install N+1 units each sized for the load divided by N. For a 3,000 kVA load with N = 2, each unit is at least 1,500 kVA, and with three units installed each one runs at about 67% of its rating in normal operation
• 2N: each of two independent paths is sized for 100% of the critical load, so in normal operation each runs at about 50% or less
• Do not count on overload ratings to cover a failure: short-time overload capability exists in the loading guides, but it depends on prior loading, ambient temperature, and insulation condition, so it is better treated as a bonus than as a design basis

Step 3: Derate for Ambient and Altitude

Standard transformer ratings assume a limited operating environment. Typical standard conditions are a maximum ambient of 40 °C with a 24-hour average of 30 °C, and an installation altitude of 1,000 m (3,300 ft) or less. Outside those conditions the usable capacity falls.

• Altitude: thinner air removes heat less effectively. A common manufacturer rule for ventilated dry-type units is to reduce kVA by about 0.3% for every 100 m above 1,000 m. Liquid-filled units have their own factors, so ask the manufacturer for the correct value, and tell them about any site above 1,000 m so insulation levels can be checked as well
• Ambient temperature: a transformer in a hot electrical room, or in an enclosure with poor ventilation, runs hotter than the standard test condition. Either specify a lower-temperature-rise design or derate the capacity. Room ventilation should be sized for the transformer's full-load losses
• Enclosure and cooling method: a unit in a NEMA 3R outdoor enclosure, a sealed vault, or a forced-air stage that may be unavailable should be rated on the conservative stage

IEC reference conditions. IEC 60076-11 for dry-type units defines normal service as cooling air not above 40 °C at any time, 30 °C monthly average of the hottest month, and 20 °C yearly average, at an altitude not exceeding 1,000 m, with a minimum of -25 °C outdoors and -5 °C indoors. IEC 60076-1 uses the same 1,000 m altitude limit and 40 °C maximum. Note that the IEC averaging periods differ from the 24-hour average used in the IEEE conditions above, so a room that is acceptable under one basis should be re-checked against the other. Above 1,000 m, IEC 60076-11 provides a temperature-rise correction rather than a single kVA percentage, so the 0.3% per 100 m figure above is a manufacturer rule of thumb and not a value from either standard. The IEC loading guides (IEC 60076-7 for liquid-immersed and IEC 60076-12 for dry-type units) cover loading beyond the nameplate and ambient conditions outside the normal range.

The derating factor is applied to the required rating:

Required nameplate kVA = Design load kVA × (1 + growth) ÷ derating factor

Step 4: Select a Standard Size

Transformers are built in standard sizes, so the calculated number is always rounded up to the next one. Common North American three-phase kVA ratings include 15, 30, 45, 75, 112.5, 150, 225, 300, 500, 750, 1,000, 1,500, 2,000, and 2,500 kVA, with larger sizes continuing upward for substation work. IEC 60076-1 states that for transformers up to 10 MVA the rated power should preferably be taken from the R10 series of ISO 3, which gives 630, 800, 1,000, 1,250, 1,600, 2,000, 2,500, and 3,150 kVA in the range most used for building and plant distribution. The gaps between neighboring sizes are therefore not the same as in the North American list, so check the size step in whichever system the project follows.

Round up, then check three things before accepting the size:

• Day-one loading: a transformer that starts at 30% loading will run below its best-efficiency point and cost more in no-load losses than a smaller unit, but it also leaves room to grow. Check that the choice is deliberate
• Gap to the next size: if the calculated value sits just above a standard size, consider whether the load estimate is tight enough to justify jumping up, or whether a better load estimate would let the smaller unit work
• Secondary current and bus rating: the transformer's full-load amps drive the secondary main breaker and bus rating, so a larger transformer cascades into larger downstream equipment

Full-load current for a three-phase transformer is:

FLA = kVA ÷ (√3 × kV line-to-line)

Step 5: Check Impedance and Secondary Fault Current

Once the size is set, the nameplate impedance (%Z) sets the maximum fault current the transformer can deliver, so the secondary equipment's interrupting and withstand ratings depend on it. The quick infinite-bus estimate is:

I_sc (secondary) = FLA ÷ (%Z ÷ 100)

This assumes the utility source is infinitely strong, which overestimates the fault current. A better estimate adds the source impedance, expressed on the transformer's kVA base, to the transformer %Z:

%Z(total) = %Z(transformer) + %Z(source), where %Z(source) = transformer MVA ÷ source short-circuit MVA × 100

Motor contribution is added on top for any motors connected on the secondary side. The full method, including the MVA method and motor contribution, is in Short Circuit Current Calculations, and the Short Circuit Calculator will do the arithmetic.

Two practical points about %Z on a real project:

• Tolerance is real: IEEE C57.12.00 allows ±7.5% on the impedance of a two-winding transformer, and IEC 60076-1 allows ±10% when the impedance is below 10% and ±7.5% at 10% and above. When you are checking the maximum fault current, use the low end of the tolerance, because a transformer built at the low end of its range delivers more fault current than the nameplate suggests
• Higher %Z helps fault current and hurts regulation: if the secondary equipment rating is the problem, a higher impedance can solve it, but the voltage drop under load and during motor starting gets worse, so check both

IEC approach. The same infinite-bus and source-impedance arithmetic applies, but IEC 60909 is the reference method for a full study. It applies a voltage factor c to the equivalent voltage source (1.05 for low-voltage systems with a +6% tolerance, 1.10 otherwise, for maximum fault current) and a correction factor to the transformer impedance, so a full IEC 60909 result differs slightly from the quick estimate. For the transformer itself, IEC 60076-5 sets the requirement to withstand external short circuits, both the thermal effect (the duration is normally 2 seconds unless specified otherwise) and the dynamic peak, so the declared impedance and the system fault level must be given to the manufacturer. In European distribution practice the impedance is commonly 4% up to 630 kVA and 6% above that, but confirm the value for the specific unit.

Step 6: Check Voltage Regulation

A first approximation of the voltage regulation at a given load and power factor is:

VR% ≈ (%R × cos φ) + (%X × sin φ), multiplied by the per-unit loading

For a typical distribution transformer, where %X is much larger than %R, a lagging power factor produces a larger voltage drop than a unity power factor. If the downstream load includes large motors, check the starting dip as well, using the method in Motor Starting and Protection Methods. Tap settings of ±2.5% and ±5% are commonly available, but they are fixed adjustments made with the transformer de-energized, not a regulation tool.

Inrush Current and Primary Protection

When a transformer is energized, the core can saturate and draw a large, brief, asymmetric current that has nothing to do with a fault. Its size depends on the residual flux in the core, the point on the voltage wave at which the breaker closes, the core design, and the source impedance. Small transformers typically draw a higher multiple of rated current than large ones, though the large ones draw more in absolute amps and take longer to decay.

A common coordination convention, used when drawing protection curves, is to represent inrush as two points on the time-current plot:

• About 12 times full-load current for 0.1 second
• About 25 times full-load current for 0.01 second

The primary fuse or relay curve should sit to the right of and above both points, while still protecting the transformer against a real through-fault. Where several transformers are energized together, or the source is very stiff, the actual inrush can be higher, so use manufacturer data for a critical project. Because inrush is rich in second harmonic, transformer differential relays use second-harmonic restraint to tell inrush from an internal fault.

NEC Primary Protection for 1,000 V and Below

For transformers rated 1,000 V nominal or less, NEC 450.3(B) sets the maximum overcurrent device rating as a percentage of rated current. When protection is on the primary only, the commonly cited limits are:

• Primary current of 9 A or more: 125% of rated primary current
• Primary current of 2 A to less than 9 A: 167%
• Primary current of less than 2 A: 300%

Where the calculated value does not match a standard device rating, the next higher standard size is permitted. For example, a 500 kVA, 480 V three-phase primary has a full-load current of 601 A. At 125% the limit is 752 A, so an 800 A device is permitted. Alternatives exist that combine primary and secondary protection, and transformers above 1,000 V fall under NEC 450.3(A) with different percentages. Always check the exact table in the edition of the NEC adopted for the project.

IEC Primary and Secondary Protection

IEC has no equivalent of the NEC 450.3 percentage table. On the medium-voltage side the protection is a fuse, a fuse-switch combination, or a circuit-breaker with a relay, and it is selected from the manufacturer's transformer protection tables for the kVA, voltage, and inrush, then confirmed in a coordination study. The same inrush convention shown above (roughly 10 to 12 times rated current for 0.1 second) is widely used for fuse selection, but the fuse manufacturer's data decides. IEC 60076-8 is the application guide that covers inrush and other transformer characteristics.

On the low-voltage side, IEC 60364-4-43 sets the conditions for overload and short-circuit protection of the circuit. In practice that means the main breaker's rated current is chosen at or above the transformer's secondary full-load current (and within the cable and busbar rating), and its breaking capacity must be at least the prospective fault current at the point of installation, as in IEC 60947-2.

Losses and Efficiency

A transformer has two kinds of loss, and they behave differently:

• No-load (core) loss: present whenever the transformer is energized, 8,760 hours a year, regardless of load
• Load (winding) loss: rises with the square of loading, so a transformer at 80% load has about 64% of its full-load winding loss

In the United States, the DOE 2016 efficiency standard (10 CFR Part 431) sets minimum efficiency for most low-voltage dry-type distribution transformers, measured at 35% of nameplate load. Published values for three-phase units include roughly 99.14% at 500 kVA and 99.28% at 1,000 kVA. Verify the exact figures against the regulation for the unit being purchased, because the requirements differ for liquid-filled and medium-voltage units.

In the European Union, Commission Regulation (EU) 548/2014 (amended by 2019/1783) sets Ecodesign limits for small, medium, and large power transformers. Tier 2 has applied to units placed on the market since 1 July 2021. For medium power transformers it limits load and no-load losses by rating, and for larger units it sets a minimum Peak Efficiency Index. Test methods and loss values follow IEC 60076-1 and, for energy efficiency, IEC 60076-20 and the harmonized standard EN 50588-1. The DOE figure above is stated at 35% load, while the EU limits are stated as loss values and peak efficiency, so the two should not be compared directly.

For loads that run continuously near full output, such as data centers and process plants, the loss figures matter more than the minimum standard suggests. Comparing bids on a total owning cost basis, which adds the purchase price to the capitalized value of no-load and load losses, often favors a lower-loss design that costs more up front.

Worked Example A (NEC): 1,500 kVA, 13.8 kV to 480Y/277 V

Given: diversified design load 1,150 kVA at 480Y/277 V, 20% growth allowance, single transformer with no redundancy requirement, ventilated dry-type unit at a site 1,500 m above sea level, utility short-circuit level at the primary of 500 MVA, and a transformer with 5.75% impedance.

• Growth: 1,150 kVA × 1.20 = 1,380 kVA
• Altitude derating: 500 m above 1,000 m, so 5 × 0.3% = 1.5% reduction, giving a factor of 0.985
• Required nameplate: 1,380 ÷ 0.985 = 1,401 kVA
• Standard size: the next standard size up is 1,500 kVA
• Day-one loading: 1,150 ÷ 1,500 = 77% of nameplate, which leaves room for the planned growth
• Secondary full-load current: 1,500 ÷ (√3 × 0.48) = 1,804 A, which sets the main breaker and bus rating
• Fault current, infinite bus: 1,804 ÷ 0.0575 = 31.4 kA
• Fault current with source impedance: source %Z on the transformer base = 1.5 ÷ 500 × 100 = 0.30%, so total %Z = 6.05% and the fault current is 1,804 ÷ 0.0605 = 29.8 kA, before adding any motor contribution
• Primary full-load current: 1,500 ÷ (√3 × 13.8) = 62.8 A
• Inrush check points: 12 × 62.8 = 753 A for 0.1 second and 25 × 62.8 = 1,569 A for 0.01 second, which the primary device curve must clear

The result is a 1,500 kVA unit with secondary equipment rated above roughly 30 kA plus motor contribution. If the next available standard interrupting rating is only just above the calculated value, a transformer with a slightly higher impedance, or a current-limiting arrangement, may be worth evaluating.

Worked Example B (IEC): 1,600 kVA, 11 kV to 415/240 V

Given: diversified design load 1,250 kVA at 415/240 V, 15% growth allowance, single liquid-immersed transformer at a site below 1,000 m with no redundancy requirement, MV short-circuit level of 250 MVA at 11 kV, and a transformer with 6% impedance.

• Growth: 1,250 kVA × 1.15 = 1,437.5 kVA
• Derating: site within the normal service conditions, so the factor is 1.0 and the required rating is 1,437.5 kVA
• Standard size: the next R10 size up is 1,600 kVA (1,250 kVA falls short)
• Day-one loading: 1,250 ÷ 1,600 = 78% of nameplate
• Secondary full-load current: 1,600 ÷ (√3 × 0.415) = 2,226 A, which sets the main breaker and busbar rating
• Fault current, infinite bus: 2,226 ÷ 0.06 = 37.1 kA
• Fault current with source impedance: source %Z on the transformer base = 1.6 ÷ 250 × 100 = 0.64%, so total %Z = 6.64% and the fault current is 2,226 ÷ 0.0664 = 33.5 kA
• Low end of the impedance tolerance: 6% - 10% = 5.4%, so total %Z = 5.4% + 0.64% = 6.04% and the fault current is 2,226 ÷ 0.0604 = 36.9 kA, before adding motor contribution
• Primary full-load current: 1,600 ÷ (√3 × 11) = 84.0 A
• Inrush check points: 12 × 84.0 = 1,008 A for 0.1 second and 25 × 84.0 = about 2,100 A for 0.01 second, which the MV fuse or relay curve must clear

The result is a 1,600 kVA unit with 415 V switchgear and busbars rated for at least about 37 kA, which in practice means selecting the next standard rating above that. The comparison with Example A shows how the system voltage changes the outcome: at 415 V the same kVA gives a higher secondary current than at 480 V, so the fault current and breaker sizes are driven up. A full IEC 60909 study would add the voltage factor c and the transformer correction factor, and would include motor contribution.

Where Standards Diverge

• Reference conditions: IEC 60076-11 uses 40 °C maximum, 30 °C monthly average, and 20 °C yearly average cooling air at up to 1,000 m, while IEEE standards use 40 °C maximum with a 30 °C 24-hour average, so the same room can pass one test and fail the other
• Standard sizes: North American practice uses the ANSI/IEEE kVA series, while IEC 60076-1 recommends the R10 series (630, 800, 1,000, 1,250, 1,600 kVA), so the "next size up" can differ
• Impedance tolerance: ±7.5% under IEEE C57.12.00 for two-winding units. Under IEC 60076-1 it is ±10% below 10% impedance and ±7.5% at 10% and above
• Short-circuit method: NEC projects typically use the IEEE and ANSI method, while IEC projects use IEC 60909 with voltage factor c, and transformer withstand is specified under IEC 60076-5
• Protection rules: NEC 450.3 prescribes percentage limits for overcurrent devices, while IEC practice relies on manufacturer protection tables, coordination studies, and IEC 60364 for device selection, so the rule-based shortcut does not carry over
• Voltage systems and units: North American 480Y/277 V and 60 Hz against IEC 400/415 V and 50 Hz, with kVA ratings the same in both and current, cable size, and busbar rating changing with the voltage
• Efficiency regulation: the US DOE 2016 standard (35% load efficiency) and EU Regulation 548/2014 (loss limits and Peak Efficiency Index) use different tables and test points, so efficiency values should not be compared across them without checking the basis

Practical Checklist

• 1. Start from the diversified design load in kVA, with the basis recorded
• 2. Apply the growth allowance and the redundancy rule as explicit steps
• 3. Derate for altitude and ambient against the project's standard (IEEE or IEC 60076-11 conditions), and confirm the manufacturer's factors
• 4. Round up to a standard size (ANSI/IEEE list or IEC R10 series) and check day-one loading
• 5. Calculate secondary fault current with source impedance, using the low end of the %Z tolerance, and compare it with downstream ratings
• 6. Check voltage regulation at full load and during the largest motor start
• 7. Place the primary protection curve above the inrush points and within the code limits (NEC 450.3) or the manufacturer and IEC 60364 selection rules
• 8. Check harmonic loading and specify a K-factor or derated unit if non-linear loads dominate, as described in Harmonics & Power Quality
• 9. Compare bids on total owning cost, not purchase price alone

Summary

• Transformer sizing is a sequence: design load, growth and redundancy, derating, standard size, fault current, regulation, inrush, and losses
• Required nameplate kVA = design load × (1 + growth) ÷ derating factor, rounded up to the next standard size
• Infinite-bus fault current overestimates the real value, so include source impedance and check the %Z tolerance
• Inrush is normal. Represent it as roughly 12× FLA for 0.1 s and 25× FLA for 0.01 s on the protection curve, and use manufacturer data for critical work
• For 1,000 V and below, NEC 450.3(B) limits primary-only protection to 125% of primary current (167% and 300% for small currents), with the next higher standard size permitted
• Under IEC, there is no percentage rule: protection comes from manufacturer tables and IEC 60364 selection, and the impedance tolerance is ±10% below 10% impedance and ±7.5% above
• For continuously loaded facilities, losses can outweigh purchase price, so evaluate total owning cost, bearing in mind that the DOE and EU Ecodesign rules use different test bases