A motor draws six to eight times its running current while it accelerates, and the set has to supply that surge without dropping the voltage far enough to trip contactors. The starting method changes the surge far more than the set size does, and a starter on one motor is usually cheaper than the next alternator up.
The current a motor asks for at the first instant
A motor standing still is a transformer with its secondary shorted. The rotor is not turning, so there is no back emf to oppose the supply and the only thing limiting the current is the impedance of the windings. That current is the locked rotor current, and on a standard three phase motor it is six to eight times the current the motor draws at full load.
The figure belongs to the motor and not to the machine it drives. A pump starting against a closed valve draws the same current as the same pump starting against an open one; it simply reaches full speed sooner. That is why the motor schedule, not the process description, is what a starting calculation is built from.
| Rotor design | Locked rotor current | Locked rotor torque | Where it is used |
|---|---|---|---|
| Standard three phase induction motor | 6 to 8 times full load | 1.5 to 2.5 times full load | Pumps, fans, air handling and most building plant |
| High starting torque design | 8 to 10 times full load | 2.5 to 3 times full load | Conveyors, crushers and mills that must break away under load |
| Slip ring (wound rotor) motor | 1.5 to 2.5 times full load | 1 to 2.5 times full load | Large mills, where resistance is switched into the rotor circuit |
Two more properties of that instant matter as much as the current does. The first is the power factor: during the start it is poor, typically 0.3 to 0.4, so the starting kVA is large while the in-phase part the engine has to drive is much smaller. The second is the duration. The surge lasts from a fraction of a second on a small motor to several seconds on a large one driving a high inertia load, and the set has to hold its voltage for the whole of it.
Motor starting kVA: from a kW schedule to a kVA step
A load schedule is written in kW, the set is bought in kVA, and a motor start is a kVA event. Three conversions take the schedule from one to the other, and they are worth writing down once because every later number depends on them.
- Running kVA. Divide the shaft kW by the efficiency and by the power factor. At 92 % efficiency and 0.8 power factor that is kW x 1.36, which is the basis used on this page.
- Starting kVA. Multiply the running kVA by the current multiple. Six times the current at the same voltage is six times the kVA.
- Current per phase. kVA x 1000 divided by 1.732 times the line voltage. At 400 V, 1 kVA is 1.44 A, as set out in generator voltage and phases.
| Motor rating | Running kVA | Running current at 400 V | Starting kVA, direct on line | Starting kVA, star-delta |
|---|---|---|---|---|
| 7.5 kW | 10.2 kVA | 14.7 A | 61 kVA | 20 kVA |
| 15 kW | 20.4 kVA | 29.4 A | 122 kVA | 41 kVA |
| 22 kW | 29.9 kVA | 43.1 A | 179 kVA | 60 kVA |
| 30 kW | 40.8 kVA | 58.8 A | 245 kVA | 82 kVA |
| 45 kW | 61.1 kVA | 88.3 A | 367 kVA | 122 kVA |
| 55 kW | 74.7 kVA | 107.9 A | 448 kVA | 150 kVA |
| 75 kW | 101.9 kVA | 147.1 A | 611 kVA | 204 kVA |
| 90 kW | 122.3 kVA | 176.5 A | 734 kVA | 245 kVA |
| 110 kW | 149.5 kVA | 215.7 A | 897 kVA | 299 kVA |
| 132 kW | 179.3 kVA | 258.9 A | 1076 kVA | 359 kVA |
| 160 kW | 217.4 kVA | 313.8 A | 1304 kVA | 435 kVA |
| 200 kW | 271.7 kVA | 392.2 A | 1630 kVA | 544 kVA |
The table is the reason the largest motor is treated separately from the rest of the list. A 90 kW motor is 122 kVA of running load and 734 kVA of starting load. Adding it as 122 kVA and adding it as 734 kVA are two different generator sizes, and the second is the one the site experiences at the moment the contactor closes.
What the alternator does with that current
Starting kVA is not a number the set either has or does not have. It is a current, the current meets an impedance, and the impedance turns it into a voltage dip. The impedance that matters here is the alternator’s subtransient reactance, the figure on the datasheet that describes how the machine behaves in the first few cycles of a fault or a motor start.
Standard alternators sit around 0.10 to 0.18 per unit. A lower figure is a stiffer machine and holds its voltage better on the same starting load, which is why the same kVA from two manufacturers can behave differently in front of the same motor.
As a planning approximation, the depth of the dip in per cent is the subtransient reactance multiplied by the ratio of starting kVA to the set rating:
dip, per cent = subtransient reactance (per unit) x starting kVA / set kVA x 100
| Starting kVA / set kVA | At 0.12 per unit | At 0.15 per unit | At 0.18 per unit | Typical outcome |
|---|---|---|---|---|
| 0.25 | 3 % | 4 % | 5 % | Invisible in the building |
| 0.50 | 6 % | 8 % | 9 % | Inside the 10 to 15 % a well matched set shows |
| 0.75 | 9 % | 11 % | 14 % | Sensitive loads begin to notice |
| 1.00 | 12 % | 15 % | 18 % | The practical limit with other load connected |
| 1.50 | 18 % | 23 % | 27 % | Contactors and drives start to drop out |
| 2.00 | 24 % | 30 % | 36 % | Other motors on the busbar are at risk of stalling |
The approximation is deliberately simple-minded. It ignores saturation, the short term response of the exciter and the fact that the falling voltage reduces the starting current, which in turn reduces the dip. What it does well is separate the cases that are comfortable from the cases that cannot work, which is the decision a first pass has to make. Borderline cases belong with the alternator manufacturer’s transient data, and asking for that data before the order is one of the cheapest things a project can do.
The regulator is not the answer to a deep dip. A modern automatic voltage regulator restores the voltage in a fraction of a second and that recovery is its job, but the depth of the first dip is a property of the machine, and no setting on the regulator changes it.
What the engine does with the same step
The alternator reacts to the starting kVA. The engine reacts to what is left after the reactive part is taken out, and at a starting power factor of 0.3 to 0.4 that is a much smaller number. A 734 kVA start is close to 250 kW of real power at the instant the contactor closes, which is why a set sometimes rides through a start that the voltage dip table calls marginal.
It is not a licence to ignore the engine. Two limits do not appear in the voltage figure at all. The first is the governor: on a sudden load step the speed falls before the fuel rack answers, and the frequency dips with it, typically two to four per cent on a mechanical governor and under two per cent on an electronic one. The second is the turbocharger: a turbocharged engine needs time to build boost, so the torque it can deliver in the first second of a step is less than the steady rating implies. That is why load acceptance is a separate figure on the engine datasheet and not something that can be read off the power rating.
The start is therefore calculated per set rather than once for the site. A single set accepting a 734 kVA step is one calculation; two sets sharing the same motor through a busbar are another, because the step each governor sees depends on how the load divides between them. Where starting duty is what drives the size, paralleling is often chosen for load acceptance rather than for redundancy.
The kVA a set can accept also moves with the conditions on site. A set rated at 40 °C derates at 50 °C, and the derating applies to load acceptance as well as to running load, so the hot afternoon is the moment the margin goes. Those figures are set out in diesel generator performance in a hot and humid climate. The rating itself is the other half of the question: a standby rating allows a different load acceptance step from a prime one, and the gap is not small. The kW and kVA sides of the same machine are explained in kVA and kW in a diesel generator.
For the load schedule the working rule is the one used on how to size a diesel generator: add the running load, then add the largest motor at its starting kVA rather than at its running kVA, and size on whichever of the two totals is larger. The running load and the starting step are two different problems, and a set has to answer both.
Four ways to start a motor
There are four ways an induction motor is started, and the choice between them changes the starting kVA far more than the choice of set does. The multiples below are against the motor’s own full load figures.
| Method | Starting current | Starting torque | What it needs on site |
|---|---|---|---|
| Direct on line | 6 to 8 times full load | 1.5 to 2.5 times full load | A contactor and an overload, and a supply that can take the step |
| Star-delta | 2 to 2.5 times full load | 0.5 to 0.8 times full load | A six lead motor wound for 400 V delta and a changeover contactor set |
| Soft starter | 2 to 3.5 times full load | 0.3 to 0.8 times full load | A thyristor starter, sized on the start and on the number of starts per hour |
| Variable frequency drive | 1 to 1.1 times full load | 1.5 to 2 times full load from zero speed | A drive, its filters and the earthing and cable arrangement it brings |
The torque column is what decides the method. A pump or a fan is a light start, because the torque the load asks for rises with the square of the speed and is close to nothing at the moment of starting, so a method that halves the starting torque costs nothing there. A conveyor, a crusher or a mill needs torque at zero speed, and a method that removes two thirds of the torque removes it exactly where it is needed. That is why star-delta fails on those loads while working perfectly well on a fan.
The drive is the only method that holds the current near the running figure, because it builds voltage and frequency together and the motor never sees the full supply across a stalled rotor. It also brings its own effects: the drive draws a non-sinusoidal current, the harmonics flow in the supply and in the neutral, and the earthing and cable arrangement around the motor change. Those effects on a generator supply are set out in generator cable sizing and voltage drop.
Two of the four methods are worth a closer look, because both are commonly misunderstood. Star-delta is sold as the cheap answer to a starting problem, and the drive is sold as the answer to every starting problem. The next two sections take them in turn.
Star-delta is not a free fix
Star-delta works by starting the motor in star, where each winding sees the line voltage divided by 1.732, which is 58 per cent of it. The current in a winding falls with the voltage, and the line current falls by the same factor twice over, so the starting current is one third of the direct on line figure. Torque falls with the square of the voltage across the winding, which is also one third. The three to one reduction in current and the three to one reduction in torque are the same fact seen from two sides.
That is why the method is described both as a two to three times start and as one third of direct on line. Both statements are true of the same starter, and the difference between them is only which figure the author took as the direct on line base.
The changeover is the part that is left out of the sales description. The starter runs the motor up in star and then switches to delta, and the switch is not instantaneous: it takes tens of milliseconds, during which the motor is disconnected and its flux decays. Reconnecting the delta winding to a supply that is no longer in phase with the decaying flux produces a second current transient, often as large as the first. On a soft supply the changeover can be the worst moment of the whole start rather than the gentlest, which is the opposite of what the starter is bought for.
The method also needs the motor to be built for it. A star-delta starter requires six leads brought out of the terminal box, so that the star point can be made and broken externally, and the winding has to be rated for 400 V in delta on a 400 V supply. A motor wound for 400 V star and 690 V delta, which is common, cannot be started this way without losing a third of its rated torque for the whole of its working life.
Where star-delta fits, it fits well: a fan or a pump of moderate size, a supply stiff enough to make the changeover survivable, and a budget that will not stretch to a drive. Where it does not fit, the failure usually shows itself in the first week of operation rather than in the design office.
What else is running when the big motor starts
The motor start does not happen on an empty busbar. Whatever else the set is carrying is still connected, and the dip the start produces is a dip that every one of those loads sees. A start that a 500 kVA set absorbs comfortably on an empty board can be the start that resets a controller when the board is already at 60 per cent.
Two effects combine. The starting kVA of the motor adds to the running kVA already on the set, so the ratio of total kVA to set kVA is higher than the motor alone would suggest. A 90 kW direct on line start is 734 kVA on its own and 1,034 kVA when 300 kVA of other load is connected, and the second figure is what the alternator’s reactance actually sees.
| Load | At a 10 % dip | At a 20 % dip | At a 30 % dip |
|---|---|---|---|
| Contactors and relays | Hold in | Drop out below 70 to 80 per cent of coil voltage | All drop out |
| Drives and soft starters | Ride through | Undervoltage trip on the smaller drives | Trip on almost every drive |
| UPS and IT load | Ride through, battery untouched | Transfer to battery | Transfer to battery, runtime begins |
| Discharge and LED lighting | No visible change | Visible flicker | Lamps extinguish and must restrike |
| Other induction motors | Unaffected | Slower, drawing more current | Stall and draw locked rotor current |
| PLC and instrumentation | Undisturbed | Some switched mode supplies reset | Controllers reset and the process is lost |
The second effect is that the load already running may itself be a motor. A motor running at part load draws a current that rises as the voltage falls, and if the dip lasts long enough a loaded motor can stall and join the start with its own locked rotor current. This is the mechanism behind the start that passes in a factory acceptance test and fails on site, where the other loads happen to be heavier than the test bench ever was.
Diversity is not a licence here. The load schedule may quite correctly say that not everything runs at once, and the set may still be sized on the running load that happens to be on at the moment of starting, which is not the same number. The safe practice is to name the worst combination, write it down and design for it: the set, or the starting method, or the interlock that prevents the motor starting until another load has been taken off.
Worked example: 90 kW on a 500 kVA set, three ways
Take a 90 kW motor on a 500 kVA set with a subtransient reactance of 0.15 per unit, and start it four different ways. On the planning basis used on this page the motor is 122 kVA at running load and 734 kVA at six times, and the other loads are off.
| Starting method | Starting kVA | Starting kVA / set kVA | Estimated dip |
|---|---|---|---|
| Direct on line | 734 kVA | 1.47 | 22 % |
| Soft starter | 367 kVA | 0.73 | 11 % |
| Star-delta | 245 kVA | 0.49 | 7 % |
| Variable frequency drive | 128 kVA | 0.26 | 4 % |
Direct on line is the one that does not work. A 22 per cent dip will drop contactors and trip drives, and no setting on the regulator changes it. The three other methods all sit inside the 10 to 15 per cent that the voltage page calls normal for a well matched set, and the drive is comfortable even if the alternator turns out to be a softer one than the datasheet promised.
Now put 300 kVA of other load on the same set before the motor starts, which is a realistic afternoon rather than a worst case.
| Starting method | Total kVA at the start | Total kVA / set kVA | Estimated dip |
|---|---|---|---|
| Direct on line | 1,034 kVA | 2.07 | 31 % |
| Soft starter | 667 kVA | 1.33 | 20 % |
| Star-delta | 545 kVA | 1.09 | 16 % |
| Variable frequency drive | 428 kVA | 0.86 | 13 % |
The motor that was comfortable on star-delta on an empty board is only marginal on star-delta with the load connected, and the drive that was comfortable is now at the edge. Two answers are available. One is a starter further down the list, which is the cheaper answer on almost every project. The other is a larger set, and the comparison is worth doing: on an 800 kVA set the direct on line start becomes a 14 per cent dip with no other load and 19 per cent with the 300 kVA still connected, which a stiff alternator will take. Whether that is worth the money depends on how often the motor starts and how much of the other load is genuinely interruptible.
The sets used in this example are the 500 kVA diesel generator and the 800 kVA diesel generator. Send the motor schedule with the enquiry and the same calculation is done with the real figures rather than with the planning basis.
Mistakes that cost money
- Sizing the set on the motor’s kW and not on its starting kVA. A 90 kW motor is 122 kVA running and 734 kVA at the instant of starting. A set bought on the first number will not start the motor, and the failure appears on the first run rather than in the office.
- Adding the largest motor to the total without naming which motor it is. The starting kVA depends on the method fitted to that motor, and the method is a decision. Two tenders for the same plant can need two different sets and both be correct.
- Never asking for the alternator’s transient data. Two 500 kVA machines from two manufacturers can differ by half again in subtransient reactance, and it is that figure, not the kVA, that decides whether the motor starts. It is free to ask for and expensive to discover afterwards.
- Choosing star-delta because it is cheaper than a drive. On a conveyor or a crusher it removes two thirds of the torque at zero speed, which is the only speed where the load needs torque at all. The starter that was meant to save the set ends up unable to start the plant.
- Testing on an empty board. A start that works with nothing else connected tells you almost nothing about the same start on a loaded afternoon. The test has to be run with the load that will really be on, at the temperature the site really reaches.
- Looking for the fault in the set. A generator that trips, stalls or drops its voltage on every start of one particular motor is usually being asked to do something the starting method should have been doing. The checks that separate the two cases are set out in diesel generator fault finding.
Questions we are asked
How much starting current does a motor draw?
A standard three phase induction motor draws six to eight times its full load current while it accelerates, and a high starting torque design draws eight to ten times. The figure is a property of the motor and its rotor design rather than of the machine it drives, and the motor datasheet is the authority. A slip ring motor is the exception, at one and a half to two and a half times, because resistance is switched into the rotor circuit during the start.
How do I calculate motor starting kVA?
Convert the motor’s shaft kW to running kVA first, by dividing by the efficiency and the power factor, then multiply by the starting current multiple. At 92 per cent efficiency and 0.8 power factor the conversion is kW x 1.36, so a 90 kW motor is 122 kVA running and six times that is 734 kVA. The multiple is almost the whole of the calculation, which is why the starting method matters more than the motor rating.
Can a 500 kVA generator start a 90 kW motor?
It depends entirely on the starting method, and the worked example on this page runs the four cases. Direct on line it cannot, because the start is 734 kVA against a 500 kVA set and the estimated dip is 22 per cent. With a soft starter or star-delta it can, with the dip falling to 11 and 7 per cent respectively, and with a drive the start is comfortable. The set is rarely the wrong size; more often the starter is missing.
Does star-delta really reduce the starting current to one third?
Yes, and it reduces the torque to one third as well. In star each winding sees 58 per cent of the line voltage, and the current falls with the voltage twice over, which is the factor of three. Torque falls with the square of the winding voltage, which is the same factor of three. The two reductions are one fact, and a method chosen for the current has to be accepted for the torque.
What voltage dip is acceptable on a motor start?
Ten to fifteen per cent at the alternator terminals is normal for a well matched set and is what a motor start has always looked like on a generator. Below that, the answer belongs to the load rather than to the set: a plant of motors and contactors will ride through a dip that a hospital theatre or a data hall will not. The thresholds for those two are set out in how to size a generator for a hospital and in data centre backup power, and they are stricter than the set’s own limit.
Does the generator or the alternator decide whether the motor starts?
Both, but for different things. The engine and its governor decide whether the set holds frequency and whether the load step can be accepted at all, and the alternator’s subtransient reactance decides how deep the voltage dip is at the instant of the start. A set can be perfectly adequate on kW and still show a poor start, because the alternator that was fitted is softer than the one that was assumed. That is why the transient data belongs in the specification rather than in the post order paperwork.
What happens to the other loads when a large motor starts?
They see the same dip the motor does. Contactors drop out below about 70 to 80 per cent of their coil voltage, drives trip on undervoltage, a UPS transfers to battery and discharge lighting may extinguish and need to restrike. Anything already running that is a motor can stall and add its own locked rotor current to the start. The table on this page sets out the thresholds, and the worst combination is the one to design for.
What do you need from me to check a motor start?
The motor schedule rather than the kW total: each motor’s rating, its starting method or the method you are willing to accept, and the largest one named separately. Add the running load that will be connected when that motor starts, the set rating you have in mind, and the supply voltage and phase arrangement. With those we will come back with the starting kVA, the estimated dip and the method or the set size that makes the start work.
Send the motor schedule, not only the kW
Tell us the largest motor’s rating and how it is started, the running load that will be on the board when it starts, and the set rating you have in mind. We will come back with the starting kVA, the estimated voltage dip and the starting method or set size that carries the start.
Written and maintained by the KVA Grid sales engineering desk. The starting multiples on this page are planning values for three phase induction motors: six to eight times full load current for a standard direct on line start. Running kVA is calculated from the shaft kW at 0.8 power factor and 92 per cent efficiency, and the current per phase as kVA x 1000 divided by 1.732 times the line voltage. The voltage dip figures are the planning relation dip equals subtransient reactance in per unit multiplied by starting kVA divided by set kVA, shown at 0.12, 0.15 and 0.18 per unit, and they are an approximation that ignores saturation, exciter response and the fall in starting current as the voltage drops. Real values come from the motor datasheet, the starter settings and the alternator manufacturer’s transient data. This page explains how a motor start is assessed on a generator supply and is not a quotation or advice on any specific project. Reviewed 8 October 2026.
