Three-phase generator load balancing has no single limit. It has two, they apply to different equipment, and they are measured in different quantities.
That’s why the numbers you find contradict each other. One source says 10%. Another says 1%. A third says 10 to 15%.
All three are real. Reading any of them against the wrong reading on your meter is how a set gets blamed for a problem it didn’t cause.
Aquí está la versión corta. The alternator is limited on current. The motor is limited on voltage. No standard sets an on-load voltage unbalance limit for a generating set at all, which is the part almost nobody publishes.
This article takes the limit side of the question. For what imbalance physically is and how to distribute circuits across a board, our generador diésel trifásico guide covers it. What follows answers the narrower question an engineer actually asks: I have a percentage. Is it a problem?
We build three-phase sets from 5 kW to 3,000 kW and size them against real load lists, so we see which numbers sites are actually held to.
Running an unbalanced set and unsure whether it matters? Send us the three-phase currents and the alternator rating, and we will tell you where you stand. Habla con nuestro equipo de ingeniería →
Puntos Clave
- The alternator’s limit is on negative-sequence current, not voltage: 8% of rated current under IEC 60034-1 for an indirectly cooled machine, and 10% under NEMA MG1 Part 32 and ISO 8528-3.
- The 1% figure that circulates is a motor voltage limit (NEMA MG1) and a no-load set limit (ISO 8528-3). It is not an on-load limit on your generator’s output.
- No standard sets a maximum on-load voltage unbalance for a generating set. Obligations are written as current limits because unbalanced current is what heats the rotor.
- IEC and NEMA limits are continuous ratings, not survival limits. Cummins states its damper windings tolerate roughly twice the 8% minimum, which is where the “10 to 15%” you have read comes from.
- The bigger the set, the smaller the share of its rating you can take as single-phase power. The instinct to solve an unbalanced site with a larger machine works against you.
Three Phase Generator Load Balancing: One Question, Two Limits
Most guides tell you to balance the load and stop there.
The useful question is what you’re balancing a, and it splits the moment you ask which device is at risk.
The Limit on the Alternator Is a Current
When the three-line currents differ, the difference doesn’t just disappear. It appears in the alternator as a negative-sequence component, a rotating field that turns against the rotor instead of with it.
That component is what the standards limit, and they limit it as a corriente: a percentage of the alternator’s rated current. Not a voltage, and not a simple percentage spread across your three clamp readings.
The Limit on the Motor Is a Voltage
The motors on the load side are limited differently, on voltaje unbalance. NEMA MG1 puts that at 1%, a far tighter figure than the alternator’s, and it is the number most people have read somewhere.
So a site can hold two readings at once: an alternator sitting comfortably inside its current limit and a motor that is outside its voltage one. They are two measurements of two different things, and neither substitutes for the other.
Why does a three-phase generator need load balancing at all? Because both limits are real, both are thermal, and both are reached faster than most operators expect.
What the Standards Actually Set
This is the part the SERP does not give you. Three standards govern unbalanced loading, and they are specific.
| Limitar | Cantidad medida | Se aplica a | Estado del producto | Estándar |
|---|---|---|---|---|
| 8% | Negative-sequence current (I₂) | Alternator, indirectly cooled | Continuous, on load | IEC 60034-1, Table 2 |
| 10% | Negative-sequence current (I₂) | Alternator with interconnected damper winding | Continuous, on load | NEMA MG1 Part 32, §32.14 |
| 10% | Negative-sequence current | Alternador | Continuous, on load | ISO 8528-3, §10.1 |
| 1% | desequilibrio de tensión | Motor | Correr | NEMA MG1 |
| 1% | Voltage unbalance (δU₂₀) | Generating set | Sin carga, classes G1 to G4 | ISO 8528-3, §8.11 |
Read the last two rows again. Both are 1%, and neither is a limit on the set’s output while it’s carrying load.
IEC 60034-1 and the 8% Figure
IEC 60034-1 carries a table of continuous negative-sequence current limits by machine type. For an indirectly cooled machine, the class that includes the standard open-ventilated IC01 alternator sitting on most diesel gensets, the figure is 8% de la corriente nominal.
The same table gives different figures for other machine types: 5% for a directly cooled salient-pole machine, 1% for a synchronous compensator, and 10% for a motor. The 8% is not universal, and any article quoting it as a blanket number is overstating it. It is the figure for the class of machine you are most likely running.
NEMA MG1 Part 32 and ISO 8528-3 at 10%
NEMA MG1 Part 32, §32.14 permits a negative-sequence current of 10% for a generator with an interconnected amortisseur, or damper, winding.
ISO 8528-3 §10.1 says the same thing in a slightly different voice: the alternator shall be capable of operating continuously with a negative phase sequence current up to and including 10% de la corriente nominal.
So there are two defensible numbers for the alternator, 8% and 10%, and they come from different standards bodies answering the same question. That discrepancy is real, it’s documented, and manufacturers resolve it by designing above both.
Why There Is No On-Load Voltage Limit for a Generating Set
Here is the fact that resolves the contradiction.
No standard stipulates a maximum on-load phase-to-phase voltage unbalance for a generating set. Cummins Generator Technologies states this plainly while walking through the standards that do apply. The standards express the on-load obligation as a corriente limit instead, because it is unbalanced corriente that thermally stresses an alternator.
The 1% figures you have read are something else entirely. One is a rule about what a MECÁNICA can tolerate. The other is a sin carga requirement on the set’s own output, tied to ISO 8528 performance classes.
If you’ve been comparing a voltage reading against 1%, or a current reading against a figure meant for voltage, you’ve been comparing the wrong quantity to the wrong device. That single mix-up explains most of the contradictory advice on this subject.
Why the Limit Is Lower Than the Machine Can Survive
An 8% limit sounds arbitrary until you look at what the current does inside the machine.
The negative-sequence field rotates at twice synchronous speed relative to the rotor. In a two-pole machine turning at 3,000 rpm, that field sweeps past the rotor at 6,000 rpm equivalent. It induces current in the damper winding, the cage of bars that exists to damp oscillation.
Those bars are not sized for continuous induced current. They are sized for transients.
So the standard figure is a resistencia térmica limit, not a performance one. Exceeding it doesn’t degrade your output gracefully. It overheats the damper cage and the stator winding, and it does so quietly, because the set’s own instruments may show nothing unusual at the total load level.
Gita ran into exactly that. A small fabrication unit on a 200 kVA set had one phase carrying a large single-phase welder while the other two carried lighting and small tools. On total load, the set looked half empty. The alternator ran hot enough to discolour its paint, and two separate technicians recommended a bigger machine.
The fix was moving the welder to a different phase and splitting the lighting. Total capacity didn’t change. The temperature did.
Both limits are thermal, so a hot machine always has a cause worth finding. Phase imbalance is one of them, and our guide to causas de sobrecalentamiento del generador covers the rest of the list.
The 10% and the 15% You Have Read About
This is where the SERP contradicts itself most openly, so it is worth resolving properly.
Some sources quote a 10% tolerance. Others quote 10 to 15%. Both figures are real, and neither is a voltage tolerance.
- The 10% is the standard. NEMA MG1 Part 32 and ISO 8528-3 both set the continuous negative-sequence current limit at 10% of rated current.
- The 15% is the design margin. In the same application guidance, Cummins Generator Technologies states that its damper windings are designed to tolerate roughly dos veces the required 8%, which lands near 15%. It then advises that protection be set at the “textbook” 8% rather than at the survival figure.
So the range isn’t a range of opinions. It’s one standard limit and one engineering margin, sitting 7 points apart, and the gap is deliberate. A manufacturer who sized the damper cage to the standard would have no margin at all for the transients the winding exists to absorb.
La lectura práctica: design and set your protection against the standard figure. Know that the machine will take more, and treat that headroom as insurance rather than as capacity you are entitled to use.
From Clamp Meter to a Number You Can Compare
You can measure this with a clamp meter, and the arithmetic is well known. What matters more is what you do with the answer.
What the Field Formula Gives You
The common field method takes the three line currents and expresses the largest deviation from the average as a percentage of that average. Our guide to three phase generators walks through the calculation and a worked example.
Eso te da un current imbalance percentage, which is a serviceable proxy and a useful trend line. It isn’t the same quantity as the negative-sequence current the standards are written in, and treating the two as interchangeable is where the wrong-comparison error starts.
What the Standards Are Actually Written In
The standards limit Yo, the negative-sequence component. If you want the exact figure rather than a proxy, it comes from the sequence components of the three currents. A power quality analyser, or the alternator manufacturer’s own data, will give you that.
For most sites the practical path is simpler. Compare against the alternator’s rating and the 8 to 10% basis, and keep clear of the lower number when the machine runs continuously.
Two things to check before you accept your own reading:
- Take it under representative load. Imbalance usually grows as load grows, and a reading taken on a lightly loaded set tells you little.
- Check whether the load you are measuring has a rated single-phase share. Equipment that is single-phase by design belongs on one phase, and the question becomes whether the set can accept it. That’s the next section.
Yusuf spent a fortnight chasing a phantom fault. He had measured 7% current imbalance on a 400 kVA set, found the 1% figure in a motor article, and concluded the alternator was failing. It wasn’t.
His 7% was a current reading, his 1% was a voltage limit applying to the motors, and the set was running inside the standard that actually governed it. Two weeks of downtime came from comparing two different units.
How Much Single-Phase Load a Three-Phase Set Can Take
This is the most commercially useful number in the article, and it runs against intuition.
Single-phase loads have to go somewhere, and on a three-phase set they land on one phase. Manufacturers cap how much of that a set should carry. Cummins’ application guidance puts single-phase unbalance at no more than 10% of the set’s rating.
Then comes the part most buyers have backwards. As Cummins states the rule: the larger the generator set, the lower the percentage of single-phase power that can be taken. On the smaller sets in its range, the allowable single-phase share runs above half the three-phase rating. As the rating climbs, that share falls.
The reason is structural. A single-phase load on one phase of a large set produces a proportionally larger negative-sequence current in a rotor that isn’t built proportionally bigger for it. Scaling the machine up doesn’t scale the imbalance problem away.
Which means the answer to a single-phase-heavy site is usually distribution, not capacity.
Hendrik learned that the expensive way. A print shop moved from a 100 kVA set to a 400 kVA unit to handle a growing single-phase load of presses and dryers. The new machine was four times the capacity and had less allowable single-phase headroom, proportionally, than the one it replaced. The presses still had to be spread across phases.
Had the phases been balanced on the original set, the upgrade wouldn’t have been needed for that reason at all.
If you’re specifying against a load list with a heavy single-phase component, get this number right at the enquiry stage rather than after delivery. It’s settled by the load list, not by the machine.
The Neutral Under Imbalance
There’s a persistent claim that imbalance makes the neutral carry more current than the phase conductors. It’s worth being precise, because the claim is half right and the half that’s wrong sends people looking in the wrong place.
From imbalance alone, the neutral cannot exceed the largest phase current. The neutral carries the phasor sum of the three line currents. Load one phase fully and the other two not at all, and the neutral carries exactly that phase current, not more.
The neutral-exceeds-phase effect is a harmonic phenomenon, not an imbalance one: triplen harmonic currents add in the neutral instead of cancelling, and that is how a neutral ends up carrying more than a phase conductor.
The two get conflated because both appear on the same four-wire system. They are separate problems with separate remedies, and our guide to generator alternator specifications covers how alternators are built to handle the harmonic side.
For load balancing specifically, the working rule is the one in this article: keep the phase currents inside the alternator’s current limit. Treat the neutral as a conductor with its own sizing question, not as a symptom of imbalance.
Preguntas frecuentes
How much load imbalance is acceptable on a three-phase generator?
Against the alternator, the standard limit is 8% negative-sequence current under IEC 60034-1, or 10% under NEMA MG1 Part 32 and ISO 8528-3, as a share of rated current. In field terms a simple current imbalance reading should stay well inside those figures, and the lower one applies where the set runs continuously rather than as standby.
Can one phase be overloaded on a three-phase generator?
Yes, and it’s the most common way a set is quietly overworked. One phase can sit at its current limit while the total load across all three reads well below the set’s rating, because the rating applies to the balanced condition. The machine heats on the loaded phase regardless of what the total says.
What happens if the phases are unbalanced?
The negative-sequence current heats the rotor damper winding and the stator. Short term, the set runs hot and voltage shifts as loads switch. Long term, insulation life shortens, and the set can fail at what looks like a fraction of its rated load. If the symptoms have already started, our Solución de problemas del generador guide covers the diagnosis order.
Does balancing the load increase usable generator capacity?
It restores capacity you already paid for. An unbalanced set cannot deliver its rating because one phase reaches its limit first, so the available output is set by the most heavily loaded phase. Redistributing single-phase circuits across the phases raises the usable total without changing the machine. If you want the background on how that capacity is calculated in current terms, our guide to amperios del generador trifásico lo cubre.
Is a 10% imbalance reading a problem?
Depends which reading it is. A 10% corriente imbalance sits at the NEMA and ISO continuous limit for the alternator, which is acceptable by the standard but leaves no margin. A 10% voltaje unbalance is a serious fault condition for connected motors, well outside the 1% they are rated for. Same number, different quantity, different conclusion.
How do I know if my equipment is three phase?
Look at the machine rather than the supply. Our guide to what equipment needs a three phase generator covers the field method for reading a load’s own requirement off the machine, including what to do when the plate is missing.
Conclusión
Three-phase generator load balancing comes down to knowing which limit you are measuring against.
The alternator is limited on negative-sequence current: 8% of rated current under IEC 60034-1 for an indirectly cooled machine, 10% under NEMA MG1 Part 32 and ISO 8528-3. The motor is limited on voltage at 1% under NEMA MG1. No standard sets an on-load voltage unbalance limit for a generating set, so any figure you read needs to be checked against the device it was written for.
De ello se derivan tres consecuencias prácticas:
- Compare current readings against current limits, and never against the 1% voltage figure.
- Treat the 10 to 15% range as one standard limit plus a design margin, not as a choice of tolerances.
- Expect the allowable single-phase share to fall as set rating rises, and fix an unbalanced site by redistributing circuits before buying capacity.
If you’ve measured the phases and aren’t sure whether the result is a problem, that’s a five-minute question for someone who sizes these machines daily. Send us the three currents and the alternator rating → and we’ll tell you where you stand, and whether the answer is a distribution change or a different set.
