Home › Guides › Parallel Operation

Parallel Operation of Transformers

Parallel operation of transformers means connecting two or more transformers so that their primary windings share a common supply and their secondary windings feed a common load. Done correctly, it is one of the most useful arrangements in power distribution: capacity can be increased without replacing an existing unit, a transformer can be taken out of service for maintenance while the load stays supplied, and the loss of one unit does not necessarily mean the loss of the whole supply. Done incorrectly, however, paralleling can set up large circulating currents, overload one unit while another idles, and damage equipment that was perfectly healthy on its own.

The rules that govern paralleling are not arbitrary. They follow directly from how a transformer works: two units connected in parallel must produce secondary voltages that agree in magnitude, phase, and sequence, and they must divide the load current in a way that neither unit is overloaded. This guide explains the four standard conditions for parallel operation, how load sharing really works, why circulating currents appear, why tap positions matter, and the practical checks made before two transformers are paralleled. For the tests that confirm a transformer is fit to be paralleled in the first place, see the transformer testing guide.

Why transformers are operated in parallel

The most common reason for paralleling is growth. A site whose load has outgrown a single transformer can add a second unit beside it rather than replacing the first, which is often cheaper and causes far less outage time. Paralleling also buys reliability: with two units sharing a load, one can be isolated for oil processing, testing, or repair while the other carries at least part of the load. In substations and industrial plants, a bus coupler between two transformer incomers is a familiar example of the same idea, allowing load to be transferred or shared between sections.

There is also an efficiency argument, though it needs care. A lightly loaded transformer still draws magnetising current and core loss, so running two units at very low load is not automatically more efficient than running one. Paralleling is justified by capacity, flexibility, and security of supply, not by efficiency alone. The decision should always be based on the actual load profile, the ratings of the units involved, and the protection arrangement, rather than on the assumption that more transformers are always better.

The four conditions for parallel operation

Four conditions are taught as the standard requirements for paralleling transformers. The first three are effectively mandatory; the fourth, concerning impedance, is sometimes described as desirable rather than absolute, but ignoring it leads to poor load sharing, so in practice it should be treated with the same seriousness. Each condition addresses a different way in which two transformers can disagree with each other.

1. Same voltage ratio

The two transformers must have the same voltage ratio, meaning the same rated primary and secondary voltages, and therefore the same no-load secondary voltage. If the ratios differ, the secondary voltages differ too, and because the secondaries are connected together, that voltage difference drives a current that circulates between the two transformers. This circulating current flows even when no external load is connected, heats both windings, and adds to the load current in one unit while opposing it in the other. Small ratio differences caused by tap position are covered separately below, but the rated ratios themselves must match.

2. Same vector group and phase displacement

The vector group describes how the primary and secondary windings are connected and the resulting phase displacement between primary and secondary voltages. Two transformers can only be paralleled if their secondary voltages are in phase with each other. If one unit has a phase displacement that differs from the other, the secondary voltages will not line up, and the resulting voltage difference drives a very large circulating current, far larger than the modest current caused by a small ratio error. Unlike a tap mismatch, a vector group mismatch cannot be fixed by adjusting a tap changer. Transformers with different vector groups must not be paralleled unless the connections are rearranged to make the phase displacement identical, which is a design-level change, not a site adjustment.

3. Similar percentage impedance

Percentage impedance governs how load current divides between paralleled transformers. For load to share in proportion to ratings, the percentage impedances of the units should be similar. When impedances differ significantly, the lower-impedance transformer takes more than its fair share of the load and may reach full load, or overload, while the higher-impedance unit is still lightly loaded. The total capacity of the pair is then less than the sum of the two ratings, because the pair is limited by whichever unit reaches its limit first. Similarity of impedance therefore matters even though it is sometimes listed as a desirable condition rather than a strict one: unequal impedance does not prevent paralleling outright, but it wastes capacity and risks overloading one unit.

4. Same phase sequence

For three-phase transformers, the phase sequence of the secondary voltages must be the same. If the sequence is reversed on one unit, the corresponding phases do not match, and paralleling would effectively short-circuit phases against each other. This is a basic but critical check, and it is verified during commissioning by phase rotation testing before any paralleling is attempted. Phase sequence is straightforward to check and straightforward to correct by swapping connections, but it must be confirmed rather than assumed, especially after cable or busbar work.

How load sharing works in proportion to ratings

When the four conditions are satisfied, load current divides between paralleled transformers in proportion to their ratings, provided their percentage impedances are equal. A simple way to picture this is to think of two units where one is rated at twice the capacity of the other: the larger unit should carry roughly twice the load current of the smaller one, so that both reach the same percentage of their own rating at the same time. That is the ideal outcome, and it is what correct paralleling aims for.

In practice, load sharing is governed by impedance, not by rating alone. The current taken by each transformer is inversely related to its impedance: the lower the impedance, the larger the share of current. When percentage impedances are equal, this works out to sharing in proportion to ratings. When they are not equal, the sharing is skewed toward the lower-impedance unit, and the pair must be loaded only up to the point where the most heavily loaded unit reaches its rating. Engineers therefore check both the ratings and the impedance values on the nameplates before deciding how much combined load a paralleled pair can safely carry.

Load sharing should also be checked after paralleling, not merely calculated. Current measurements on each unit under load show whether the sharing matches expectations. If one unit consistently carries far more than its share, the cause is usually an impedance difference, a tap mismatch, or a difference in voltage ratio that was not noticed at the planning stage. Discovering this under load is far better than discovering it through a protection trip or an overheating alarm.

Circulating currents from ratio and impedance mismatch

A circulating current is a current that flows between paralleled transformers without supplying any external load. It is wasted energy and wasted heating capacity, and it is the main symptom of a paralleling mismatch. Two distinct mechanisms cause it, and it helps to keep them apart because they behave differently.

The first mechanism is a voltage difference, caused by a ratio mismatch or by different tap positions. The difference voltage drives a current around the loop formed by the two secondaries, limited only by the combined impedance of the two units. Because transformer impedance is low, even a modest voltage difference can produce a substantial circulating current. This current exists at no load and persists under load, adding to the current in one transformer and subtracting from the current in the other.

The second mechanism is an impedance difference combined with load. Even when the no-load voltages agree perfectly, unequal impedance causes the load current to divide unevenly. This is not strictly a circulating current in the no-load sense, but it has a similar practical effect: one unit is worked harder than the other, and the pair's usable capacity falls. Both mechanisms point to the same lesson. Paralleling is not simply a matter of connecting two transformers to the same busbars; the units must be electrically compatible, and compatibility must be verified from nameplate data before any switching is done.

Tap position matching

On-load and off-load tap changers adjust the voltage ratio of a transformer in steps, compensating for supply voltage variation or load-related voltage drop. When two transformers operate in parallel, their tap positions must correspond so that their voltage ratios match. If one unit sits on a higher tap than the other, its secondary voltage differs, and the difference drives a circulating current exactly as a rated ratio mismatch would.

Tap matching deserves attention because it is easy to get wrong in operation even when the planning was correct. After maintenance, after a tap changer operation, or after one unit has been returned to service, the taps may no longer agree. Some installations use master-follower tap changer control, where one unit's tap changer leads and the other follows automatically, keeping the positions matched. Where no such scheme exists, tap positions must be checked manually as part of the paralleling procedure and rechecked after any tap operation. A transformer that has been tapped up or down to correct its own voltage while its parallel partner was left untouched is a common, avoidable cause of circulating current.

Practical paralleling procedure and checks

The following procedure describes the general sequence used when paralleling transformers. It is a guide to good practice, not a substitute for the site's switching programme, the manufacturer's instructions, or local safety regulations, all of which take precedence.

1. Confirm compatibility from nameplates. Check rated voltages, vector group, phase sequence, percentage impedance, and ratings for both units. Confirm that the combined load will not overload either unit given the actual impedance values.
2. Check tap positions. Confirm that both tap changers are on corresponding positions, and record the positions. If taps disagree, correct them before paralleling, following the manufacturer's procedure for the tap changer type.
3. Verify protection and earthing arrangements. Confirm that protection settings cover the paralleled arrangement, that earth fault and overcurrent protection will operate correctly with both units in service, and that neutral earthing is arranged as designed.
4. Check the transformers are fit for service. Review recent test results, oil condition where applicable, cooling system status, and any outstanding alarms or defects. A unit with a known defect should not be paralleled to "share the burden" until the defect is understood.
5. Energise the incoming transformer correctly. Energise the second transformer from the primary side first, allowing inrush to settle, and confirm its secondary voltage, phase sequence, and phase rotation against the live busbar before closing the secondary connection.
6. Check voltage and phase agreement. Before closing the paralleling breaker or bus coupler, verify that the voltages on both sides agree in magnitude, phase, and sequence. Phase checking across the open breaker is the standard method, and it must show agreement before the breaker is closed.
7. Close the paralleling breaker and observe. After closing, check the load current on each transformer, confirm the sharing is roughly as expected, and watch for abnormal noise, heating, or protection indications. Investigate any unexpected circulating current rather than accepting it.
8. Record the arrangement. Log the tap positions, load currents, and any observations. Good records make the next paralleling operation, and any later fault investigation, far easier.

Removing a transformer from parallel operation follows the same care in reverse: load is reduced or transferred, the unit is off-loaded as far as practicable, and only then is it isolated, following the site switching programme. Abruptly opening a paralleling breaker under heavy load, or isolating a unit that is carrying more than its share because of an impedance mismatch, can disturb the remaining unit and the connected load.

Safety notes

Paralleling work involves energised switchgear, high fault levels, and switching operations that must only be carried out by qualified and authorised personnel under a proper switching programme. Two transformers in parallel present a higher fault level than either unit alone, which affects arc flash risk, protection operation, and the rating of connected equipment. Never parallel transformers on the basis of a guess about vector group or impedance: if nameplate data is missing or unclear, obtain it from the manufacturer or from test results before proceeding. If phase checking shows disagreement, do not close the breaker and "see what happens" — stop, investigate, and resolve the discrepancy first. As with all electrical work, follow the project specifications, the manufacturer's instructions, and local safety regulations, and treat any unexpected current, noise, or heating after paralleling as a reason to investigate, not to ignore.

Frequently asked questions

Related guides: start with the transformer testing guide for the tests that confirm a unit is fit for service, then read about transformer energisation and inrush, the choice between dry-type and oil-filled transformers, and transformer losses and efficiency.