How-to
Motor starting voltage dip calculation on an 11 kV bus
Short answer
A starting medium-voltage motor dips the bus in proportion to its locked-rotor kVA against the source's short-circuit strength. Screen it as dip ≈ 100 × S_start/S_sc at the point of common coupling, then refine with ΔV = I_start(R cosφ + X sinφ) per phase. In the worked 11 kV, 1.5 MVA chiller example a direct-on-line start gives 3.6% at the PCC and 4.0% at the plant bus, failing a 3% limit; a 65% tap autotransformer or a soft starter limited to 3× rated current holds both points under 2%.
On this page
- What is the formula for the voltage dip a starting motor causes?
- What source impedance does the utility fault level give me?
- Which current and power factor go into the dip for a starting motor?
- How does the 11 kV chiller example work step by step?
- Where does the dip limit apply: at the point of common coupling or at my plant bus?
- What dip limit should the start be designed to?
- Which starting method keeps the dip inside the limit?
- How do I verify the dip once the starter is installed?
- Questions asked
A starting motor's voltage dip is one comparison: the motor's locked-rotor kVA against the short-circuit strength of the source behind it. Screen any medium-voltage start in thirty seconds as dip ≈ 100 × S_start/S_sc, then refine it per phase with ΔV = I_start(R cosφ + X sinφ). Worked for a 1.5 MVA, 11 kV chiller on a 250 MVA source: direct-on-line dips the point of common coupling 3.6% and the plant bus 4.0%, which fails the 3% limit this example assumes, while a closed-transition autotransformer on its 65% tap or a soft starter limited to 3× rated current holds both points under 2%. Every input below is an illustrative example — substitute your utility fault letter, cable schedule and motor datasheet before choosing hardware.
What is the formula for the voltage dip a starting motor causes? #
Screen the start as dip ≈ 100 × S_start/S_sc, where S_start = √3 × V_LL × I_locked-rotor; on the 250 MVA source of the worked example the 9.0 MVA inrush reads 3.6%, and the per-phase formula ΔV = I_start(R cosφ + X sinφ) confirms 227 V, also 3.6%.
The per-phase formula is the incremental voltage-drop estimate that IEEE Std 141, the recommended practice for industrial plant distribution, walks through for motor-starting studies: the locked-rotor current flows through the total series resistance R and reactance X between source and bus, and cosφ and sinφ project that lagging current onto the resistive and reactive parts of the impedance. At a 0.25 starting power factor the reactive term dominates — in the worked path X sinφ contributes 0.506 Ω-equivalent against 0.027 from R cosφ — which is why a calculation that ignores reactance misreads the dip badly.
Treat the result as a screening estimate: it assumes a balanced three-phase system, a dip small enough that the pre-start phase voltage stands in for the bus voltage, and every impedance referred to one voltage base. It is the same arithmetic that sizes a feeder against voltage drop, written with starting current instead of running load; the feeder version is worked through in fixing voltage drop on a long 277 V lighting feeder.
What source impedance does the utility fault level give me? #
Convert the three-phase fault level to ohms with Z_source = V_LL²/S_sc per phase: 250 MVA at 11 kV is 0.484 Ω, and at the X/R of 8 the utility reports, that splits into 0.060 Ω of resistance and 0.480 Ω of reactance.
The conversion falls out of the fault level's own definition: three-phase short-circuit power S_sc = √3 × V_LL × I_sc, and the source impedance is phase voltage over fault current, which collapses to V_LL²/S_sc, with volts over volt-amperes giving ohms. Split the magnitude with the reported X/R: R = Z/√(1 + (X/R)²) and X = Z × (X/R)/√(1 + (X/R)²); for 0.484 Ω at X/R 8, √65 = 8.06, so R = 0.060 Ω and X = 0.480 Ω. Get the fault level and X/R in writing from the utility's connection data.
If the letter gives only megavolt-amperes, screening can proceed anyway: between X/R 5 and 15 the worked example's dip at the point of common coupling moves less than 0.1 percentage point, because the reactance dominates either way. Add every series element downstream on the same base — each cable run, each transformer's nameplate impedance referred through the square of its voltage ratio — and the same screen works at any bus against that bus's own fault level: the worked plant bus has about 227 MVA behind it, and 100 × 9.0 ÷ 227 reads 4.0% again.
Which current and power factor go into the dip for a starting motor? #
Use the locked-rotor current from the motor datasheet — 6.0 × 78.7 A = 472 A in the example — with the starting power factor it quotes, 0.25 lagging here; running full-load values would understate the dip by a factor of six.
The rated current comes from the nameplate: 1,500,000 VA ÷ (√3 × 11,000 V) = 78.7 A. Datasheets quote locked-rotor current either as a multiple of rated current, 6.0× here, or as a NEMA MG-1 code letter giving a locked-rotor kVA per horsepower band; the code-letter conversion is in the FAQ below. The starting power factor comes from the same datasheet or factory test report, and 0.25 is a defensible screen value for a medium-voltage induction machine when nothing else is documented.
The power factor is also forgiving: between 0.20 and 0.30 the worked example's bus dip stays at 4.0%, because the reactive term carries the drop. What is not forgiving is substitution — the running 78.7 A at 0.85 power factor is a different problem. Use the locked-rotor values quoted at rated voltage, and let the starter choice scale them, as the sections below do.
How does the 11 kV chiller example work step by step? #
With the inputs below, the direct-on-line start draws 472 A and drops 252 V, which is 4.0% at the plant bus and 3.6% at the point of common coupling — the start fails the 3% limit as drawn.
Work it in this order:
- Collect inputs and their sources: an 11 kV, 50 Hz plant bus; a utility three-phase fault level of 250 MVA with X/R 8 at the point of common coupling; 500 m of 240 mm² feeder cable at 0.095 Ω/km resistance and 0.085 Ω/km reactance; a 1.5 MVA motor with locked-rotor current 6.0× rated and starting power factor 0.25.
- Rated line current: 1,500,000 VA ÷ (√3 × 11,000 V) = 78.7 A; starting current: 6.0 × 78.7 = 472 A.
- Source impedance: 11,000² ÷ 250,000,000 = 0.484 Ω per phase, splitting at X/R 8 into 0.060 Ω and 0.480 Ω.
- Cable over 500 m: 0.095 × 0.5 = 0.048 Ω and 0.085 × 0.5 = 0.043 Ω, so the totals to the plant bus are 0.108 Ω and 0.523 Ω.
- Pre-start phase voltage: 11,000 ÷ √3 = 6,351 V; with cosφ = 0.25, sinφ = 0.968.
- Dip at the plant bus: 472 × (0.108 × 0.25 + 0.523 × 0.968) = 472 × 0.533 = 252 V, and 100 × 252 ÷ 6,351 = 4.0%.
- Dip at the point of common coupling, source impedance only: 472 × (0.060 × 0.25 + 0.480 × 0.968) = 472 × 0.480 = 227 V, and 227 ÷ 6,351 = 3.6%.
- Cross-check with the screen: S_start = √3 × 11,000 × 472 = 9.0 MVA, which reads 3.6% against the source's 250 MVA and 4.0% against the bus's 227 MVA — both matching the phasor result.
The inputs are illustrative; the arithmetic is not. The result is an estimate to be checked against the equipment documentation and the connection agreement, and a start that carries contractual weight — four chillers against a negotiated dip limit — deserves a motor-starting study in power-system software that models every bus, the standing load and the full acceleration, as IEEE Std 141 recommends for industrial systems. The hand calculation's job is to set the expectation and kill a wrong starter choice before it is ordered.
Where does the dip limit apply: at the point of common coupling or at my plant bus? #
Both points get their own number from the same formula: the utility limit applies at the point of common coupling, which sees only the source impedance (3.6% in the example), while the plant bus adds the cable and any transformers (4.0%), and the motor terminals dip deeper still.
The point of common coupling is where the dip becomes the utility's business, because other customers see it. The plant bus is where it becomes your equipment's business: every contactor coil, control power supply and running motor rides through that 4.0%, and the motor feeder adds its own fraction of a percent to the terminal value.
Standing load moves the starting point, not the formula: with three chillers already running at 0.85 power factor, their combined 236 A pulls the worked bus down about 1.4% before the start, and the fourth start's incremental dip lands on that depressed voltage. Check the three-running case in the study as well as the cold-bus case; for this plant it is the realistic one.
What dip limit should the start be designed to? #
Design to the figure written into your utility connection agreement at the point of common coupling — 3% in this example — and to the ride-through of your own contactors, control power and running motors at the plant bus.
No code writes a universal starting-dip percentage. NFPA 70, the National Electrical Code, 2023 edition, governs the motor circuit's conductor sizing and protection and leaves dip to engineering practice; the binding numbers are the utility's agreement at the point of common coupling and your process requirements. Confirm both with the authority having jurisdiction and the equipment documentation before ordering. At the bus the requirement is ride-through: a dip that lasts the whole acceleration holds coil voltages and control supplies below nominal for seconds, so check the starter's minimum holding voltage in its documentation and arrange control power so a start cannot drop its own contactor out. Chillers add restart timers and minimum off-times that the starter's control scheme has to respect.
Which starting method keeps the dip inside the limit? #
In the worked example direct-on-line fails at 3.6% at the PCC, while a closed-transition autotransformer on the 65% tap draws about tap² × 472 = 199 A and lands at 1.5%, and a soft starter limited to 3× rated current draws 236 A and lands at 1.8% — both pass.
| Starting method | Supply current at start | Dip at the PCC | Dip at the plant bus | Verdict against 3% |
|---|---|---|---|---|
| Direct-on-line | 472 A | 3.6% | 4.0% | Fails |
| Closed-transition autotransformer, 65% tap | 0.65² × 472 ≈ 199 A | 1.5% | 1.7% | Passes |
| Same, 80% tap if torque demands it | 0.80² × 472 ≈ 302 A | 2.3% | 2.5% | Passes |
| Soft starter, current limit 3× rated | 236 A | 1.8% | 2.0% | Passes |
The autotransformer's numbers follow from its tap ratio a: the motor sees a times full voltage, its current falls with a, and line current and torque both fall with a², which is why the 65% tap draws about 199 A from the line while delivering about 0.42 of locked-rotor torque — the best torque per line amp among the tapped methods. Closed transition, the Korndorfer connection, keeps the motor energized through the tap change and avoids the open-transition spike. The torque check is decisive: 0.42 of locked-rotor torque must exceed the compressor's breakaway demand and accelerate it, which an unloaded centrifugal chiller manages; a loaded start pushes you to the 80% tap, and 2.3% at the PCC still passes.
A soft starter reaches its current limit by holding back terminal voltage, so at 3× rated current the terminal voltage sits near half and the torque near a quarter of locked-rotor torque — workable unloaded, wrong for a loaded start. The current limit lives among the starter's motor-start adjustments; set it from the study, not the default. Medium-voltage soft starters are built in 3.3, 6.6 and 11 kV classes: confirm the exact voltage class, duty and ratings in the manufacturer's documentation before ordering, and remember that stepping down from 11 kV to feed a lower-voltage unit inserts a transformer whose impedance joins the path — recalculate the dip with it included. Wye-delta starting rarely survives at medium voltage because it needs six terminal leads and a winding designed to run delta at line voltage, which single-voltage star-wound MV windings usually are not; check the connection diagram.
Decision: against a 3% limit at the point of common coupling, the worked plant passes with either the closed-transition autotransformer on the 65% tap or an 11 kV soft starter limited to 3× rated current — take the soft starter when acceleration control matters, the autotransformer when the compressor needs the higher torque per line amp.
How do I verify the dip once the starter is installed? #
Record the first start with a power-quality meter at the bus and at the point of common coupling, measuring dips to IEC 61000-4-30 so the magnitude and duration are comparable with the utility's own records, and compare the worst measured residual voltage with the study and the 3% agreement.
IEC 61000-4-30 defines how dips are measured — magnitude, duration and instrument class — and its Class A instruments are the ones to use when a number will be compared against a contractual limit. Work around an 11 kV switchboard is energized medium-voltage work: stage the metering under NFPA 70E, 2024 edition, which covers safe work practices for energized electrical work, with qualified switching personnel, and confirm every step with the site's authority having jurisdiction and the switchgear documentation.
Then make the record permanent: a power-monitoring station at the switchboard line-up, built around an industrial panel PC with the computer built in, logs every start's dip with a time stamp against the chiller that caused it, so the next dip complaint is answered from data instead of the hand formula. Re-run the study when the utility strengthens or weakens its network or a fifth chiller joins the bus — the fault level in the denominator is the utility's number, and it changes.
Questions asked
How do I get locked-rotor current from a NEMA code letter?
A NEMA MG-1 code letter gives a band of locked-rotor kVA per horsepower. Take the top of the band, multiply by the motor's horsepower rating, and divide by root three times the line voltage in kV. Code G spans 5.6 to 6.29 kVA per horsepower, so a 2,000 hp machine works out to 6.29 × 2,000 = 12,580 kVA, about 660 A at 11 kV. Screening with the top of the band is the conservative choice.
Does the dip stay constant while the motor accelerates?
No. The locked-rotor value persists only until the rotor begins to turn; drawn current then falls and power factor rises, so the first few cycles at standstill are the worst case the formula screens. On a high-inertia load the current stays elevated for many seconds, which matters for cable heating, starter duty and contactor hold-in as much as for the dip itself. A motor-starting study plots current and torque against speed for the whole acceleration and reports the dip envelope.
What happens to the three running chillers when the fourth starts?
They ride through the dip and reaccelerate afterward, each drawing recovery current as it pulls back to speed. A realistic study case is therefore three 1.5 MVA machines recovering plus the fourth starting, not a lone motor on an empty bus. That combination usually sets the true worst case for the plant bus, while the utility limit at the point of common coupling is judged on the start's own incremental dip. Check both scenarios against their respective limits.
When can I trust the quick screen of starting kVA over fault level?
As a screen, and a slightly conservative one: because a starting induction motor is strongly lagging and the source mostly reactive, the magnitude ratio sits close to or above the true dip — in the worked example the screen and the phasor formula agree at 3.6%. It turns optimistic only where it does not apply: capacitive compensation near the motor, a mostly resistive source, or a bus whose fault level you guessed. It never covers the impedance between the point of common coupling and your bus.
Should I add the standing load current to the starting current before computing?
No — add the effects, not the currents. The formula already returns the incremental dip the start itself causes; the standing load's job is to set the pre-start voltage that dip lands on. When that pre-start voltage is depressed by running load, judge the dip percentage against the pre-start phase voltage rather than nominal, because a dip measured from a lower base is slightly deeper. The study should carry both effects together, since recovery currents and the start overlap.