Why Inrush Currents Destroy Your UPS: How to Calculate Them Before Installation

The first time a 40 kVA UPS refuses to start after a new installation, the technician checks the battery first. Second, the internal fuse. Third, the input breaker. Fourth, with the UPS already in the shop, is the inrush current of the downstream transformer that nobody measured. The UPS didn’t fail from age or a dead battery. It failed because the protection coordination calculation didn’t account for the fact that during the first half-cycle after breaker closing, the current can be 8 to 25 times the transformer’s nominal current. This is what follows.

Inrush current is not unique to transformers. It shows up in any inductive equipment with a magnetic core: motors, transformers, ballasts, reactors, filter inductors. But in a data center context, the most common and destructive case is the isolation transformer or the UPS input transformer, because its inrush travels straight into the UPS power electronics with no intermediate attenuation.

What exactly is inrush current

When you energize a transformer for the first time (or after a de-energization), the iron core goes from zero magnetic flux to nominal flux. That change requires an enormous but brief current, which decays over a few cycles (typically 0.1 to 1 second) until it settles at the no-load current. The peak inrush current can reach:

• 8 to 12 times the full-load current (FLA). This is the typical range cited by IEEE 141-1993 for distribution transformers with no prior residual flux.

• Up to 25 times FLA in the first half-cycle (8-10 ms). This is the asymptotic peak that protections must tolerate. Some transformers, especially if re-energized with residual flux in the core (a transformer de-energized with capacitive load), can hit 30 times in the first quarter-cycle.

• A significant DC component. The inrush current has a DC component that saturates the core even more, which in turn deepens the saturation. It’s a self-perpetuating loop until the transformer inductance starts to recover.

The decay time depends on transformer size. Small transformers (15-50 kVA) drop to nominal current in less than 1 second. Large transformers (500 kVA – 2.5 MVA) can take 1 to 5 seconds to stabilize. During that window, the upstream UPS is seeing a current that no nameplate specification anticipates.

How to calculate inrush current

The simplified formula in IEEE C57.12.00-2021 is:

I_inrush_peak = K × √2 × I_FLA

Where K is a factor that depends on transformer type and residual flux in the core, typically between 6 and 25. For new transformers at first energization with no residual flux, K = 8-12. For re-energization with trapped residual flux, K can be 15-25.

For a 150 kVA, 480V three-phase transformer:

• I_FLA = 150,000 / (480 × √3 × 0.9) ≈ 200 A

• I_inrush_peak = 12 × √2 × 200 = 3,394 A

• Peak duration: 0.1 to 0.5 seconds.

That means that for half a second, that transformer is drawing 3,394 A from the grid or from the upstream UPS. If the upstream UPS is 40 kVA (60 A nominal), the obvious question is: what does an instantaneous load of 3,394 A do to the UPS? The short answer: nothing good.

Why the UPS suffers from inrush (and why it sometimes dies)

A double-conversion online UPS has three stages: rectifier (AC → DC), battery bank, inverter (DC → AC). The rectifier is the first component that sees the inrush current, and it’s the most vulnerable. Three failure modes:

Rectifier IGBT/SCR failure

The rectifier semiconductors have a non-repetitive peak forward current limit (IFSM). For a 100 A continuous IGBT, the IFSM is typically between 500 and 1,200 A peak for 1 ms. If the inrush current exceeds that value, the IGBT is destroyed by localized overheating. The UPS shuts down with a rectifier fault code. In the best case, you replace the power module. In the worst case, you replace the entire UPS.

Input breaker trip

The UPS input breaker is sized for the nominal current plus a margin. For a 40 kVA, 480V UPS, the input breaker may be 80 A. When an inrush current of 3,000+ A appears, the breaker trips instantly on its magnetic (instantaneous, no intentional delay) protection curve. The UPS loses AC input and switches to batteries. If the batteries are new and no operator is present, the UPS shuts down from battery depletion in 5-15 minutes, taking the critical load with it.

Downstream breaker trip

The opposite case but equally problematic: the transformer-side breaker trips (because it’s calibrated to the transformer’s FLA) and the transformer fails to energize. But the operator expects to see the transformer live and resets the breaker, which trips again, and so on. The upstream UPS gets hit with an inrush-pause-inrush-pause sequence that degrades the semiconductors with each cycle.

The real case: 40 kVA UPS + 150 kVA transformer

It’s a common scenario in small and mid-sized data centers: a 40 kVA UPS is installed to feed the critical IT load, and a 150 kVA isolation transformer is used to step 480V service to 480V or to galvanically isolate the UPS from the grid. The intent is good; the sizing is not.

The problem: the 150 kVA transformer has an FLA of 200 A and an inrush peak of 3,400 A. The 40 kVA UPS delivers at most 60 A continuous and has an 80 A input breaker. The UPS protection is sized for 60 A; the inrush demands 3,400 A. The mismatch is 56 to 1. Any energization is going to trip the UPS protection.

The right answer is not sizing the UPS for the inrush (it would cost 8 times more UPS, with no operational justification), but avoiding the inrush in the first place. The techniques are in the next section.

Techniques to mitigate inrush

Four techniques used in the field, in order of complexity:

1. Pre-energization with pre-insertion resistor. Some medium-voltage breakers come with pre-insertion resistors that limit the current during the first 100-200 ms, before the main contact fully closes. It’s the most used technique in industrial substations. For low voltage, there are contactors with resistive pre-insertion (Schneider, ABB, Eaton manufacture them for specific applications).

1. Transformers with inrush control. Manufacturers such as Schneider (Trihal), ABB (DryType), and Eaton (DS3) offer models with CRGO (cold-rolled grain-oriented) core and pre-magnetization techniques that cut typical inrush to 3-5 times FLA instead of 8-12 times. They cost 10-25% more than a standard transformer.

1. Controlled energization sequence. Before energizing the transformer downstream of the UPS, set the UPS to static bypass (if the model allows it). Static bypass is more tolerant of current peaks. Energize the transformer, wait 2-3 seconds for it to stabilize, then return the UPS to inverter mode. This procedure requires the grid quality to be acceptable (regulation, frequency), because during bypass the load is exposed to the raw AC grid with no filtering.

1. Soft-start with variable frequency drive (VFD). If the transformer feeds an inductive load (large motor), install a VFD that ramps up the motor with a voltage and frequency ramp. The VFD limits the motor inrush to 1.5-2 times FLA. It doesn’t apply if the transformer load is purely electronic (servers, switches).

How to correctly size the downstream breaker of the UPS

If the UPS is going to feed a downstream transformer (a common scenario), the transformer secondary breaker should be sized like this:

I_breaker ≥ I_inrush / k_coordination

Where k_coordination is the breaker’s ability to tolerate peaks without tripping. For standard low-voltage thermal-magnetic breakers, k_coordination ≈ 5-7. For electronic breakers with adjustable curves, k_coordination can reach 10-12. What this means in practice:

• If your 150 kVA transformer has an inrush of 3,400 A, and you use a standard breaker with k_coordination = 5, you need a breaker of at least 680 A. That’s 3.4 times the transformer’s FLA, which looks irrational from an overload protection standpoint but is exactly what’s needed to tolerate the inrush.

• To coordinate it with the upstream UPS, the transformer breaker must have a slower trip curve than the UPS breaker. If not, the UPS breaker trips first and the transformer breaker never gets exercised.

The importance of the protection coordination study

In serious data center projects, the protection coordination study is done by a registered electrical engineer and delivered before first energization. The study includes:

• Short-circuit current calculation at every point of the system (transformer, UPS, switchboards, PDUs).

• Inrush current calculation for every transformer in the system.

• Breaker selection with a trip curve that discriminates between short-circuit (must trip instantly) and inrush (must not trip) at each level.

• Verification that the trip times are coordinated so that the breaker closest to the fault trips first (selectivity).

If your project doesn’t have that study, the odds of failure at first startup are high. I’ve seen data centers where the first energization took three days because each attempt tripped the UPS breaker. The cost of three days of downtime, plus the engineering hours, far exceeds the cost of the upfront study.

Practical recommendation

If you’re about to install a UPS under 100 kVA with a downstream transformer, ask the UPS manufacturer for the rectifier peak current tolerance curve. That curve tells you for how many milliseconds and at what current level the rectifier holds up without damage. With that curve in hand, size the transformer and breaker so the inrush peak × duration falls inside the rectifier’s safe zone.

If the rectifier tolerance curve is unknown or the manufacturer won’t release it, assume conservatively that it holds 8 times FLA of the UPS for 100 ms. Any inrush that exceeds that limit will degrade the semiconductor with each energization, even if it doesn’t fail it immediately.

And if you’re going to buy a transformer to install downstream of a UPS, ask explicitly for the inrush control option. The 10-25% cost premium pays for itself in reliability at first energization and at every subsequent re-energization.

Sources

• IEEE Std C57.12.00-2021 — General Requirements for Liquid-Immersed Distribution, Power, and Regulating Transformers — https://standards.ieee.org/ieee/C57.12.00/6962/

• IEEE Std 141-1993 (Red Book) — Recommended Practice for Electric Power Distribution for Industrial Plants — https://standards.ieee.org/ieee/141/1395/

• Schneider Electric — Trihal Dry-Type Transformer technical guide (CRGO core, inrush control) — https://www.se.com/us/en/product-range/63498-trihal/

• ABB — Distribution Transformers Handbook (inrush mitigation techniques) — https://new.abb.com/docs/librariesprovider55/abb-review/transformers/abb-distribution-transformers.pdf

• CalcPanel — Transformer Inrush Current: Causes, Typical Multiples and Mitigation — https://calcpanel.com/guides/transformer-inrush-current