Home › Guides › Losses & Efficiency

Transformer Losses and Efficiency Explained

No transformer transfers all of its input power to its output. Some energy is always lost as heat in the core, in the windings, and in nearby metal parts. Those losses matter far beyond a textbook definition: they raise operating temperature, they shorten insulation life when cooling is poor, they add to energy bills over decades of service, and they shape which transformer is the right choice for a given load. Understanding transformer losses and efficiency is therefore a practical field skill, not just theory.

This guide breaks losses down into their main families, explains how no-load and load losses behave differently, shows why the efficiency curve has a peak, and explains how energy-efficient designs and total cost of ownership thinking fit together. For the tests used to measure and verify losses in practice, see the transformer testing guide.

What counts as a transformer loss?

In simple terms, efficiency is output power divided by input power. The difference between the two is total loss. For a power transformer, total loss is usually described as the sum of no-load losses and load losses, with stray losses counted as part of the load-loss family in many specifications. Keeping those families separate matters, because they respond to different things: voltage and frequency on one side, load current on the other.

A useful mental model is to think of a transformer as having a standing cost and a working cost. The standing cost is paid whenever the unit is energised, whether it is supplying a full load, a light load, or nothing at all. The working cost grows with the current the transformer actually carries. A lightly loaded transformer can therefore look inefficient in a different way from a heavily loaded one, even though the absolute losses are smaller. Both effects need to be understood before judging a design or a duty cycle.

Core losses: the no-load loss

Core losses, also called iron losses or no-load losses, occur in the magnetic core whenever alternating flux is present. Because the core is magnetised as long as the transformer is energised at its rated voltage and frequency, core losses are essentially constant with load. They do not disappear when the secondary circuit is open. That is why a transformer left energised but unloaded still consumes energy and still produces heat.

Core losses have two main components: hysteresis loss and eddy-current loss. Both turn magnetic energy into heat inside the core steel, but they arise from different physical mechanisms and are reduced in different ways.

Hysteresis loss

Hysteresis loss comes from the repeated magnetisation and demagnetisation of the core material during each alternating-current cycle. The magnetic domains inside the steel must realign twice per cycle, and that realignment takes energy. The energy lost per cycle depends on the material and on how strongly the core is magnetised, which is why core flux density is a key design choice.

Materials with a narrow hysteresis loop lose less energy per cycle than materials with a wide loop, because less work is needed to reverse their magnetisation. Grain-oriented electrical steel, in which the crystal structure is aligned to favour magnetisation in the rolling direction, is widely used for this reason. Amorphous core material, discussed later, takes this idea further: its disordered atomic structure gives it very low hysteresis loss, at the cost of different mechanical and manufacturing characteristics.

Eddy-current loss

Eddy-current loss comes from currents induced inside the core material itself. A changing magnetic flux induces circulating currents in any conductor it passes through, and the core is no exception. Those circulating currents flow through the resistance of the steel and dissipate energy as heat, exactly like a short-circuited winding that was never meant to exist.

The classic defence against eddy currents is lamination. Instead of a solid block of steel, the core is built from thin insulated sheets stacked together. Each lamination breaks up the path available to circulating currents, so the induced currents stay small and confined. Thinner laminations reduce eddy-current loss further, but they also increase manufacturing effort and cost, and they reduce the proportion of the core cross-section that is actually steel. Core design is therefore always a balance between loss, cost, size, and ease of manufacture.

How core materials and lamination reduce losses

Putting the two mechanisms together, low-loss core design works on three fronts at once. First, choose a material with low hysteresis loss for the operating flux density and frequency. Second, laminate it thinly, with good insulation between laminations, so eddy currents cannot circulate freely. Third, keep the flux density and the core geometry within the range the material handles efficiently, rather than pushing the core hard to save steel at the expense of loss. Good jointing and clamping matter too: damaged insulation between laminations, burrs at cut edges, or poor core earthing can create unintended current paths and raise losses, which is one reason core condition is checked during testing and maintenance.

Copper losses: the load loss

Copper losses, also called winding losses or load losses, are the resistive heating losses in the primary and secondary windings. They follow the familiar relationship that loss in a resistance rises with the square of the current through it. Double the load current, and the copper loss roughly quadruples, assuming winding resistance stays the same. In practice resistance rises slightly with temperature, so copper loss at a hot winding is somewhat higher than the same current would produce in a cold winding.

Because copper loss depends on load, it is zero at no load and largest at full load. Winding design controls it: larger conductor cross-section lowers resistance and therefore loss, but uses more copper, adds weight and cost, and takes up more window space in the core. Shorter winding paths and good cooling also help, because a cooler winding has lower resistance and because removing heat promptly protects insulation. When a specification quotes a load loss, it is normally quoted at a stated reference temperature and load, so like-for-like comparison needs those conditions to match.

Copper loss is the reason heavily loaded transformers run hot and why sustained overload deserves care. Heat is the main enemy of insulation life: a winding that runs hotter than its design limit ages faster, even if it does not fail immediately. Safety note: never assume an overloaded transformer is safe because it has not tripped. Follow the manufacturer rating, the project specification, and local regulations, and let only qualified personnel assess loading, cooling, and protection settings.

Stray losses

Stray losses are the losses that do not fit neatly into the core-loss or copper-loss boxes. Leakage flux, the part of the magnetic field that does not link both windings, passes through structural steel, tank walls, clamping parts, and other conductive components near the windings. That flux induces eddy currents in those parts, and those currents dissipate energy as heat.

Stray losses also grow with load, because leakage flux grows with load current, so they are normally grouped with load losses in specifications and tests. They are harder to calculate exactly than winding resistance loss, which is why load-loss measurement by test remains important. Design measures such as magnetic shunts, non-magnetic inserts, and careful tank and clamping design are used to guide leakage flux away from vulnerable steel or to shield it, reducing local heating. Hot spots on a tank or structural part during a load test can be a clue that stray loss is concentrated where it should not be, and such findings should be investigated by qualified personnel rather than ignored.

No-load vs load loss behaviour

The contrast between no-load and load losses explains a great deal about how transformers behave in service. No-load loss is present from the moment of energisation and stays roughly constant as load changes, provided voltage and frequency stay near rated values. Load loss starts at zero and climbs steeply with current. Total loss is the sum of the two, so at light load the total is dominated by the constant core loss, while at heavy load the rising copper and stray losses take over.

This has a direct consequence for energy use. A transformer that spends most of its life lightly loaded wastes most of its energy through its core, even though its copper losses look small. A transformer that spends most of its life heavily loaded wastes most of its energy through its windings. Two transformers with the same total loss at full load can therefore have very different lifetime energy costs if their load profiles differ. Any honest efficiency comparison has to ask how the transformer will actually be loaded, for how many hours, and at what voltage.

Voltage matters on the no-load side as well. Raising voltage above rated value raises flux density and therefore core loss, and it can push the core towards saturation, where loss and magnetising current rise sharply. Operating a transformer at the wrong tap or on a system voltage above its rating is not a harmless way to gain a little extra output. Follow the rating plate, the tap setting instructions, and the system design, and treat abnormal humming, heating, or vibration as a reason to investigate, not to continue.

Efficiency curves and the load point of maximum efficiency

Efficiency at any load can be written as output power divided by the sum of output power and total losses at that load. At very light load, output is small while core loss is still present in full, so efficiency is low. As load rises, output grows faster than total loss at first, and efficiency climbs. Eventually copper loss, growing with the square of current, catches up and then dominates, so efficiency peaks and then falls as load approaches and passes full load.

The peak of that curve, the load point of maximum efficiency, occurs where variable losses equal constant losses: where copper and stray load losses equal core losses. This is a standard result, and it is worth pausing on what it does and does not mean. It does not mean a transformer should always be run at that exact load. It means the designer can place the peak where the expected load profile will benefit most. A distribution transformer that spends long hours at moderate load may be designed so its peak sits at a moderate load, while a unit expected to run near full load most of the time may be designed with its peak higher up the curve.

For an engineer comparing options, the curve matters more than a single efficiency figure. Ask for losses at more than one load point where possible, and weight them by the real duty cycle: hours at each load level, seasonal variation, and any planned growth. A single full-load efficiency number can hide a high no-load loss that dominates a lightly loaded installation, just as a good no-load figure can hide heavy copper loss in a heavily loaded one.

Energy-efficient designs

Energy-efficient transformer design is not one invention but a set of choices, each trading loss against cost, size, weight, or complexity. The most widely discussed is the amorphous core transformer. Amorphous metal has a non-crystalline structure that gives it much lower hysteresis loss than conventional grain-oriented steel, so no-load loss falls substantially. The trade-offs include a material that is harder and more brittle to handle, a core that can be larger for the same rating, and manufacturing processes that need care to avoid degrading the material. Whether those trade-offs pay depends, again, on load profile and energy cost.

Low-loss designs using improved grain-oriented steel, thinner laminations, better core joints, and optimised winding geometry offer a more incremental path: lower core loss, lower copper loss, or both, without the handling changes of amorphous material. Other efficiency measures are about the system around the transformer rather than the core itself: correct sizing so the unit is not grossly oversized for its load, good cooling so winding resistance stays low and insulation stays within limits, and sound tap and voltage management so the core is not driven harder than intended.

Safety note: efficiency upgrades never justify improvised modifications. Replacing a core, rewinding, changing cooling, or altering protection must follow manufacturer guidance and be carried out by qualified personnel, with the unit de-energised, isolated, earthed where required, and tested before return to service. Efficiency is a lifetime benefit; a shortcut that damages insulation or voids a warranty is not a saving.

Total cost of ownership thinking

Total cost of ownership, often shortened to TCO, compares transformers on lifetime cost rather than purchase price alone. The purchase price is only the first line. Over a service life measured in decades, the cost of losses, meaning the energy wasted as heat over every operating hour, can rival or exceed the purchase price, especially for units that run continuously. Maintenance, cooling energy, and eventual replacement or disposal also belong in the comparison, where they can be estimated honestly.

A practical TCO comparison starts with the load profile: how many hours per year at each load level, and how that profile may change. It then applies the no-load loss for all energised hours and the load loss at each load level, using energy prices that reflect the real tariff, including any demand or loss-related charges where applicable. Two designs can then be compared on equal terms: the extra purchase cost of the low-loss option against the energy saving it delivers over its life, discounted sensibly rather than assumed. Where energy prices are uncertain, testing a range of prices shows whether the conclusion is robust or fragile.

TCO thinking also guards against false economy in the other direction. Oversizing a transformer to cut copper loss can raise core loss, because a larger core is energised all the time. Undersizing to save purchase cost can push copper loss and temperature up and shorten life. The cheapest transformer to buy is rarely the cheapest to own, and the most efficient transformer on paper is not automatically the best value if its premium never pays back under the actual duty cycle. The right question is always: for this load, for these hours, at these energy prices, over this life, which design costs least in total?

Practical takeaways for the field

For engineers in the field, a few habits make loss and efficiency thinking concrete. First, know the rating plate and the loss figures for the units you work on, and note whether they are quoted at a reference temperature. Second, record load and temperature trends rather than single snapshots, because loss behaviour only shows itself over a duty cycle. Third, treat unexplained heating, rising no-load current, or changing sound as information: they can point to core, winding, cooling, or connection problems that testing should investigate. Fourth, when a replacement or upgrade is proposed, ask for the loss figures and the assumed load profile behind any saving claim, and check that the claim survives a realistic profile rather than an optimistic one.

Finally, keep efficiency in proportion. Losses are real and costly, but they are one part of a safe, reliable installation. Protection, cooling, insulation condition, oil or insulation system health where applicable, and correct installation and earthing all come first. A transformer that is efficient but poorly protected, poorly cooled, or operated outside its rating is not a good transformer. Measure losses, compare designs honestly, and let the lifetime numbers, not the brochure headline, decide.

Frequently asked questions

Related guides: start with the transformer testing guide for how losses are measured in practice, then read about transformer energisation and inrush, transformer parallel operation, and dry-type vs oil transformers to place efficiency choices in their wider operating context.