A 20 tonne double girder is five drives and 24.1 kW on one busbar, not one crane's worth of load. Even with a 5 percent reactor on every drive the busbar lands near 18 percent total demand distortion against an 8 percent limit. Here is what actually adds up, and which fix belongs to which crane.
A 20 tonne double girder is not one crane's worth of electrical load. A typical catalogue entry lists a 16 kW main hoist, a 2.6 kW auxiliary hoist, two 0.55 kW trolley motors and four 1.1 kW long-travel motors. That is 24.1 kW spread across five drives hanging on one busbar. When they run together the harmonic currents add before they ever reach your transformer, and no single-drive calculation predicts what your meter will read.
I have been on three jobs this year where a new double girder tripped a plant that had run single girder cranes for a decade. All three were argued about as if the crane were defective. Two of them were busbar problems, one was a panel cooling problem, and none of them were the crane.
The reason double girders behave differently comes down to two things the single girder article on this site did not have to deal with. There are more drives, and above 16 tonnes there are two hoists. Two hoists means two rectifiers, two sets of brake control and two chances for a contactor to drop out at the wrong moment. The arithmetic that follows is about what those drives do to each other.
Everything below is either a published figure with the source named, a standard clause, or a number we computed. Computed numbers are labelled as planning figures. Nobody has published a survey of harmonic distortion on a 20 tonne double girder in a Vietnamese steel mill, and I am not going to pretend otherwise.
More than buyers count. A single girder at 10 tonnes has one hoist, one trolley set and two long-travel motors. A double girder at the same tonnage has four long-travel motors because each end carriage carries two, and from 16 tonnes upward it also has a second hoist for the light lifts. The published motor breakdown below is from a standard double girder catalogue covering 5 to 40 tonnes, with the totals as printed.
| Capacity | Hoisting motor | Trolley motors | Long travel motors | Crane power | Drives on the bar | Current at 400 V |
|---|---|---|---|---|---|---|
| 5 t | 6.1 / 1 kW | 2 x 0.31 kW | 2 x 0.55 kW | 8.82 kW | 3 | 16.6 A |
| 10 t | 9.5 / 1.5 kW | 2 x 0.44 kW | 2 x 1.1 kW | 14.08 kW | 4 | 26.6 A |
| 20 t | 16 / 2.6 kW | 2 x 0.55 kW | 4 x 1.1 kW | 24.1 kW | 5 to 7 | 45.5 A |
| 32 t | 20 / 5.0 kW | 2 x 1.1 kW | 4 x 1.5 kW | 33.2 kW | 5 to 7 | 62.6 A |
| 40 t | 38 kW | 2 x 1.5 kW | 4 x 2.2 kW | 47.0 kW | 5 to 7 | 88.7 A |
| 50 t (European light) | 4 / 0.67 m/min, 2 speed | 20 / 5 m/min | 40 / 10 m/min, pole change | 44.4 kW | 4 | 83.8 A |
Currents are our calculation at 400 V, power factor 0.85 and motor efficiency 0.90, using the catalogue kilowatts. The drive count is the number of separately switched motors, which is what matters for harmonics: each motor with its own inverter injects its own current distortion at its own point on the busbar. Our own double girder crane listings publish the same breakdown per capacity, and it is the first thing worth comparing between two quotations.
The 50 tonne line is worth pausing on. It is a real published machine, 13.5 m span, 6 m lift, 8,950 kg bridge and 12,810 kg total weight, 134.5 kN maximum static wheel load, classification U4/Q2 to EN 13001-1, and 44.4 kW of electrical equipment. It has no inverter anywhere. Hoisting is 4/0.67 m/min on a two-speed pole-change motor and long travel is 40/10 m/min the same way. Keep that in mind, because it changes what you should worry about.
Two hoists on one bridge is not a redundancy feature. It is a productivity feature: the auxiliary hook moves at 2 to 5 times the speed of the main hook and handles everything under about a quarter of rated capacity. On a machine that spends most of its day lifting 2 tonnes, the main hoist is dead weight.
The published configuration list of a heavy double girder range makes the pattern obvious. From 16 tonnes upward every capacity is written as a pair.
| Listed capacity | Main / auxiliary | Auxiliary share of main |
|---|---|---|
| 5 t | single hook | none |
| 10 t | single hook | none |
| 16 t | 16 / 3.2 t | 20 percent |
| 20 t | 20 / 5 t | 25 percent |
| 32 t | 32 / 5 t | 16 percent |
| 50 t | 50 / 10 t | 20 percent |
| 75 t | 75 / 20 t | 27 percent |
| 100 t | 100 / 20 t | 20 percent |
| 125 t and above | 125 / 32 t and upward | 25 to 30 percent |
This is where the electrical story departs from the single girder. The auxiliary hoist is a second rectifier, a second brake control circuit and a second motor cable, and it is deliberately sized small. Small drives are the ones with the worst current distortion relative to their own load, because the DC bus capacitance and the reactor, if there is one, are chosen against a small frame size rather than against the harmonics. The same logic applies to the hoist and accessory side of the order: the small unit is not the simple one.
Two 6-pulse rectifiers on one busbar do not average out. A published laboratory demonstration on two drives from a common AC bus measured 10.9 A of fifth harmonic from the first drive and roughly twice that with the second drive running. One plus one equals two, as the Eaton application note puts it. The waveforms are similar, the phase angles are similar, and similar angles add.
Between the two extremes, and the gap is the whole discussion. The convenient assumption is arithmetic addition: five drives, five times the harmonic current. The optimistic assumption is root sum square: five drives, 2.24 times the harmonic current. Reality sits between them and depends on how similar the drives are and whether their rectifiers are phase shifted.
The industry quantifies this with a summation exponent, often written alpha. An exponent of 1 is arithmetic addition with no cancellation. An exponent of 2 is the root sum of squares, which is what you get when phase angles are randomly distributed. A 2013 IEEE paper by Cuk, Cobben, Ribeiro and Kling measured parallel induction motor drives at two sites for a week with one-second resolution, covering both 6-pulse and 12-pulse front ends, and derived coefficients for simplified summation precisely because neither extreme is right.
What matters practically is total demand distortion at the busbar, not total harmonic distortion of one drive. Those are different numbers and buyers mix them up constantly. Here is what the two addition rules produce for a group of identical drives.
| Drives on the busbar | TDD if arithmetic, no reactor (THDi 100 percent) | TDD if root sum square, no reactor | TDD if arithmetic, 5 percent reactor (THDi 40 percent) | TDD if root sum square, 5 percent reactor |
|---|---|---|---|---|
| 1 | 100 percent | 100 percent | 40 percent | 40 percent |
| 2 | 100 percent | 71 percent | 40 percent | 28 percent |
| 3 | 100 percent | 58 percent | 40 percent | 23 percent |
| 5 | 100 percent | 45 percent | 40 percent | 18 percent |
| 7 | 100 percent | 38 percent | 40 percent | 15 percent |
| IEEE 519-2022 limit, Isc/IL 20 to 50 | 8 percent | 8 percent | 8 percent | 8 percent |
Our own calculation, using the per-drive THDi ranges published by ABB for HVAC drives and the standard addition rules. The table says something uncomfortable. Even if every drive has a 5 percent reactor and the phase angles happen to distribute perfectly, a five-drive double girder lands at 18 percent TDD against an 8 percent limit. Adding reactors helps the individual drive enormously and still leaves the busbar out of compliance.
That is not a reason to skip the reactor. It is a reason to stop believing the reactor is the whole answer. On a single drive the same table shows 40 percent falling to 18 percent, which is why the reactor is worth its two hundred dollars. On a five-drive machine the same reactor buys you from 100 to 18, which is a different kind of number but still a long way from 8.
Phase shifting is the cheapest real harmonic fix for a group of drives, and it is almost never specified on a crane. The idea is simple: two 6-pulse rectifiers fed from windings that are 30 electrical degrees apart produce fifth and seventh harmonics that are 180 degrees apart, so they cancel at the shared busbar instead of adding. One 6-pulse plus one phase-shifted 6-pulse behaves like a 12-pulse rectifier at the point where they meet.
Published current distortion figures for the available front ends, from ABB and from Siemens harmonic data for drives at full load, explain why this matters.
| Front end | Typical current THDi at full load | What it does to a five-drive busbar | Realistic crane use |
|---|---|---|---|
| 6-pulse, no reactor | 90 to 120 percent | worst case dominates | very common in low-cost control panels |
| 6-pulse, 2 to 3 percent reactor | 43 to 44 percent | helps, still above 8 | common retrofit |
| 6-pulse, 5 percent reactor | 34 to 45 percent | 15 to 18 percent if angles spread | standard in HVAC, uncommon on cranes |
| 12-pulse | 8.8 percent | near limit, needs a dual-winding transformer | rare on cranes, common in steel plants |
| 18-pulse | 3.9 percent | below limit | occasionally, for large hoists |
| Active front end, two quadrant | 1.3 to 5 percent | below limit, unity power factor | increasingly specified for high duty hoists |
| Active front end, four quadrant | 1.2 percent, plus regeneration | below limit and recovers braking energy | premium, justified above about 50 tonnes |
One published demonstration is worth more than the table. Two identical 6-pulse drives on a shared bus, one phase-shifting transformer, fifth harmonic from 10.9 A per drive to roughly 21.8 A combined without shifting and to near zero with it. The transformer did not filter anything. It turned two in-phase waveforms into two opposite-phase waveforms.
On a double girder this is genuinely practical, because the two hoists are separate drives that do not have to be identical. A 20 tonne machine with a 16 kW main hoist and a 2.6 kW auxiliary can put the auxiliary on a phase-shifted feed, or simply on a separate small transformer with a different vector group. That cancels the fifth and seventh between the two largest sources for the price of a winding rather than an active front end.
What it does not do is cancel the four long-travel motors, which are identical, fed from one inverter pair in master/slave, and therefore perfectly in phase with each other. Identical drives fed from one inverter are not four harmonic sources. They are one source with four loads, which is why long travel is usually the least of the problems and the two hoists are usually the most.
No, and this is the point where most buyers get the diagnosis backwards. The 50 tonne machine described earlier has 44.4 kW of electrical equipment and no inverter anywhere. Hoisting is two-speed pole change at 4 and 0.67 m/min. Long travel is two-speed pole change at 40 and 10 m/min. Trolley is 20 and 5 m/min. Every motion is contactor switched.
That machine does not produce significant harmonic current, because a pole-change induction motor is a linear load. It produces something else: a starting current of six to eight times full load current sustained until the motor reaches pull-up speed, with a first-cycle peak that can reach ten to twelve times according to ABB drive documentation. On a 44 kW machine that is a 500 to 800 A event at every start, and it happens a few hundred times an hour at M6 duty.
The two designs therefore need opposite mitigations.
| Symptom | Pole-change crane | VFD crane | The wrong answer |
|---|---|---|---|
| Lights flicker on each lift | yes, starting current | less, unless the DC bus is at limit | buying a bigger transformer |
| Generator trips on start | very likely | possible if undersized | adding a harmonic filter |
| Neutral or earth current warm | no | yes, triplen harmonics | replacing the cable |
| Capacitor bank blows fuses | unlikely | classic harmonic resonance | fitting a bigger capacitor |
| Contactor contact pitting | high, from start current | lower | changing contactor brand |
| Correct mitigation | soft starter or VFD retrofit | reactor, then AFE or 12-pulse | both exist, they are not interchangeable |
Our planning figures, based on the starting current ratios published by ABB and the motion speeds from the published 50 tonne data sheet. The takeaway is that the first question on any double girder power problem is not what the distortion is. It is whether there is an inverter at all. I have seen a harmonic filter quoted for a contactor-switched crane, which is a polite way of selling someone a transformer they do not need.
Because the busbar is a resistor with an inductance, and the crane collects current at whichever point along it the bridge happens to be. A crane parked at the feed end sees almost no drop. The same crane at the far end sees the full conductor length, and on a long bay the numbers stop being academic.
Published impedance figures for shrouded conductor systems used on crane runways, at 35 degrees C, per metre.
| Conductor | Rating | Impedance, ohm per metre | DC resistance, ohm per metre |
|---|---|---|---|
| Aluminium, stainless steel contact face | 500 A | 0.000158 | 0.000098 |
| Aluminium | 800 A | 0.000145 | 0.000075 |
| Aluminium | 1000 A | 0.000138 | 0.000052 |
| Copper | 500 A | 0.000162 | 0.000105 |
| Copper | 800 A | 0.000138 | 0.000058 |
| Copper | 1000 A | 0.000128 | 0.000035 |
The manufacturer's own worked examples on the same sheet are the useful part, because they show what happens when a bay is shared. Two cranes drawing 900 A total on a 1000 A aluminium busbar, 245 m bay length, end fed: 52.6 V of drop. The same bay in copper: 48.8 V. One crane drawing 1000 A on a 250 m 1000 A aluminium bar: 59.7 V. On a 400 V system that is 13.2 percent and 14.9 percent.
That is far outside any sensible limit, and it is not caused by the crane. It is caused by feeding a long bay from one end at close to the bar's rating. What the example really demonstrates is that a shared runway is an electrical design problem before it is a crane problem, and that the feed position matters more than the conductor material.
IEC 60364-5-52 Annex G is informative rather than mandatory, but it is the number everybody designs against: up to 3 percent for lighting circuits and up to 5 percent for other final circuits, measured from the point of supply. The same annex says explicitly that motor starting dip has to be checked separately, because meeting the steady-state limit does not guarantee a tolerable dip during a start.
So the question for a double girder is how much of that 5 percent the busbar eats. Our calculation below uses the published impedances above, end feed, 400 V, and the crane currents from the first table.
| Crane | Current | Runway length | Busbar | Voltage drop | Share of 400 V |
|---|---|---|---|---|---|
| 20 t | 45.5 A | 150 m | 1000 A Al | 1.63 V | 0.41 percent |
| 20 t | 45.5 A | 300 m | 1000 A Al | 3.26 V | 0.82 percent |
| 50 t | 83.8 A | 200 m | 1000 A Al | 4.01 V | 1.00 percent |
| 50 t | 83.8 A | 400 m | 1000 A Al | 8.01 V | 2.00 percent |
| 100 t | 170 A | 300 m | 1000 A Al | 12.2 V | 3.05 percent |
| 100 t | 170 A | 400 m | 500 A Al | 18.6 V | 4.65 percent |
| 2 x 50 t together | 167.6 A | 400 m | 1000 A Al | 16.0 V | 4.01 percent |
| Published example, 2 cranes | 900 A | 245 m | 1000 A Al | 52.6 V | 13.2 percent |
Our calculation from the published per-metre impedance, using the standard three-phase relation with conductor length equal to half the bay for a centre feed and the full bay for an end feed. All rows are end fed so the comparison is like for like.
The results are reassuring for the crane and uncomfortable for the runway. A single double girder at 45 to 84 A eats 0.4 to 2 percent of the budget, which is fine. Two of them sharing a 400 m bar eats 4 percent, which leaves almost nothing for the plant's own feeders. Three cranes at 900 A eats 13.2 percent, which is the published example and which is why a shared bay fails while each individual crane looks correct on paper.
One more thing the drop does that a meter will not show. It happens in the moving connection, at the collector shoes, not in a terminal box. Intermittent collector contact on a long bar is a classic source of what the industry calls weak motion at the far end: the contactor holds, the motor turns, and the load still will not lift. If your symptom appears only when the bridge is at one end of the bay, measure the voltage at the crane, not at the panel.
They cook it. Inverter losses are published at 40 to 50 W per kVA of rating, or 3 to 5 percent of rated power. A 45 kW hoist drive at power factor 0.85 looks like about 53 kVA, so its full-load loss lands between 2.1 and 2.7 kW of heat inside a steel box.
A published worked example from a crane panel designer makes this concrete. A QDY type metallurgical main hoist panel accumulates about 1.68 kW of loss: 1.5 kW in the inverter, 0.07 kW in the busbar, 0.04 kW in contactors, 0.02 kW in the power supply and 0.05 kW of small items. The enclosure is 2,000 by 800 by 600 mm, installed free standing, which gives an effective heat transfer area of 5.712 square metres by the IEC TR 60890 relation. Painted steel transfers at about 5.5 W per square metre per kelvin. At 15 K rise the box sheds 471 W.
So 1.68 kW of loss in a box that naturally sheds 471 W settles at a rise of 1,680 divided by the product of 5.5 and 5.712, which is 53.5 K. In a 42 C workshop the inside of that panel runs at about 95 C.
Now add the second hoist. A double girder above 16 tonnes puts two inverter groups in the same enclosure, and the auxiliary is not negligible because the panel is shared.
| Crane | Installed drive power | Heat at 3.5 percent | Natural shedding at 15 K | Settled rise, no fan | Inside at 40 C ambient |
|---|---|---|---|---|---|
| 10 t single girder | 14.1 kW | 0.49 kW | 471 W | 16 K | 56 C |
| 20 t double girder | 24.1 kW | 0.84 kW | 471 W | 27 K | 67 C |
| 32 t double girder | 33.2 kW | 1.16 kW | 471 W | 37 K | 77 C |
| 50 t double girder | 44.4 kW | 1.55 kW | 471 W | 49 K | 89 C |
| 50 t with one fan kit | 44.4 kW | 1.55 kW | 1,200 W with 250 m3/h of filtered air | about 15 K | about 55 C |
Our calculation, using the published loss rate, the published enclosure relation and the published derating curve. The last row uses a typical 250 cubic metre per hour filtered fan set and is indicative only; the fan supplier's curve governs.
Why it matters for distortion specifically. Published inverter derating is 1.5 percent of rated output current per degree above 40 C ambient. At 55 C inside the enclosure the drive has 77.5 percent of its rating left, so a 45 kW inverter behaves like a 35 kW inverter. A drive running at 78 percent of nameplate into a motor selected for the full load either trips on overcurrent or, worse, runs at a distorted operating point where the current waveform has more harmonic content than the datasheet figure it was specified against. Panel cooling is not a comfort item on a double girder. It is part of the harmonic compliance argument.
Two related habits from the same source are worth copying. Put the brake resistor outside the enclosure, because its surface reaches above 300 C during a regeneration cycle. And never mount a panel against a wall without redoing the heat calculation, because a wall mounted single enclosure loses about 11 percent of its effective area compared with a free-standing one.
Sometimes, and the margin is narrower than most buyers assume. A contactor's coil has a pick-up range and a release range, and the release range is wide. One published miniature contactor datasheet gives pick-up at 85 to 110 percent of rated coil voltage and release anywhere from 20 to 75 percent. If your particular unit releases at the top of that band, a 25 percent dip drops it out.
On a double girder that matters more than on a single girder because there are more contactors and more motions to lose. Losing one of four long-travel contactors leaves three motors driving one end and one motor coasting, which the master/slave control then tries to correct by pushing the other three harder. That is how a rail climbs.
| Source | Figure | What it means on site |
|---|---|---|
| Published contactor datasheet | pick-up 85 to 110 percent, release 20 to 75 percent of coil rating | a 25 percent dip can release a worst-case coil |
| IEC 60034-1, zone A | plus or minus 5 percent voltage, plus or minus 2 percent frequency | motor is safe, contactor may not be |
| IEC 60034-1, zone B | plus or minus 10 percent voltage, plus 3 to minus 5 percent frequency | rated torque available but hotter, short time only |
| NEMA MG 1 | plus or minus 10 percent voltage, plus or minus 5 percent frequency, combined not over 10 | the North American equivalent, slightly looser |
| EPRI ride-through study | most common sags last 5 to 15 cycles, 83 to 250 ms | long enough for a coil to drop and re-close |
| Generator sizing practice | motors tolerate 30 to 35 percent, specifiers limit to 20 percent | the 20 percent is about contactors, not motors |
The practical fix costs almost nothing and almost nobody does it: feed the control circuit from a small constant-voltage transformer or from a DC supply with a capacitor, so coil voltage does not track the power busbar. The other fix is to command motions sequentially rather than simultaneously, which the PLC can do for free.
It depends entirely on whether the crane sets your monthly peak. Where it does, a published industrial tariff shows the damage. One rural electric cooperative rate sheet charges USD 11.56 per kW of purchased power demand plus USD 11.81 per kW of distribution demand, so about USD 23.37 per kW per month. The same sheet adjusts billing demand upward when the power factor at maximum demand is below 97 percent, multiplying recorded demand by 97 divided by the actual power factor. It also charges USD 0.01009 per excess kVARh when the monthly average falls below 95 percent.
A 6-pulse drive with no reactor has a true power factor near 0.69 even though the drive's displacement factor is 0.98, because the distortion term drags the total down. Here is what that does to a 24.1 kW crane.
| Power factor | Apparent demand | Billed demand | Penalty per month | Penalty per year | Likely cause |
|---|---|---|---|---|---|
| 0.98 | 24.6 kVA | 24.1 kW | none | none | active front end |
| 0.93 | 25.9 kVA | 25.1 kW | USD 23 | USD 281 | 5 percent reactor |
| 0.85 | 28.4 kVA | 27.5 kW | USD 79 | USD 953 | 2 to 3 percent reactor |
| 0.69 | 34.9 kVA | 33.9 kW | USD 229 | USD 2,748 | 6-pulse, no reactor |
| 0.69, two cranes peaking together | 69.8 kVA | 67.8 kW | USD 458 | USD 5,496 | same, doubled |
Our calculation applied to the published tariff and the crane's 24.1 kW, at USD 23.37 per kW per month. Two honest caveats. This only bites where the crane is coincident with the billing peak, which is typical in a foundry or a precast yard and atypical in a warehouse where the crane runs occasionally. And it assumes the utility applies the letter of that particular tariff, which varies by region.
Where it does bite, a reactor costing a few hundred dollars returns four figures a year. That is the entire commercial case for specifying reactors on a double girder, and it has nothing to do with being a good citizen.
Heavy overhead cranes run up to four travel motors synchronously, with synchronisation provided by one crane inverter in master/slave mode. Published drive data for crane duty puts travel motors at 1.8 times rated torque for starting, S4 duty at 40 percent cyclic duration factor, up to 360 starts per hour, and NEMA MG 1 Part 31 insulation for inverter compatibility.
Two electrical failure modes belong to this arrangement and to nothing else on the crane. The first is torque skew: if one slave lags, the master increases torque on all four, the leading end carriage slips or climbs the rail, and the crane skews in the runway. The second is circulating current between motors connected to the same DC bus when their cable lengths differ significantly, which shows up as unequal heating in the motors that a thermal camera finds long before a fault code does.
Neither is a harmonic problem, and both are frequently blamed on harmonics because the symptom is a trip. If your trips happen during long travel and never during hoisting, you are looking at a drive coordination problem, not a power quality problem.
Almost entirely. Published design limits for hoists, cross referenced across ISO, FEM, CMAA and AS classifications, put a hard ceiling on running hours and starts per hour for each class.
| Class | ISO / AS | FEM | CMAA | Running hours per day | Starts per hour |
|---|---|---|---|---|---|
| 4 | M4 | 1Am | B | 2 | 180 |
| 5 | M5 | 2m | C | 4 | 240 |
| 6 | M6 | 3m | D | 8 | 300 |
| 7 | M7 | 4m | E | 16 | 360 |
| 8 | M8 | 5m | F | 16 plus | 360 |
A machine at M5 accumulates roughly 8 million starts over its life at four hours a day. A machine at M7 accumulates about four times that, because the class ceiling is four times higher. The harmonics per start do not change. The number of starts in which harmonics and starting current coincide with everything else on the plant does.
This is where a 5 to 100 tonne range splits, and it splits in an awkward place. Harmonic emission limits for equipment connected to public low-voltage systems at 16 to 75 A per phase are set by IEC 61000-3-12. A 20 tonne double girder at 45.5 A sits inside that window. A 50 tonne at 83.8 A sits outside it, and the same product family is then judged against the utility's own distortion limit rather than the equipment standard.
| Document | Scope | Where a double girder lands |
|---|---|---|
| IEC 61000-3-2 | input current up to 16 A per phase | only the smallest machines, 5 t at 16.6 A is already outside |
| IEC 61000-3-12 | 16 to 75 A per phase, up to 690 V, 50 or 60 Hz, public low voltage | 5 to 32 t fall inside; 40 t and 50 t fall outside |
| IEEE 519-2022 | current and voltage distortion at the point of common coupling, limits scaled by short-circuit ratio | the applicable limit in most export contracts, TDD 5 to 20 percent depending on ratio |
| IEC 60204-32:2023 | electrical equipment of hoisting machines from the crane supply switch onward, including conductor bars and festoon outside the machine, up to 1000 V AC | covers the runway supply, which is the part most often left out of the crane contract |
| IEC 60364-5-52 Annex G | recommended voltage drop, 3 percent lighting, 5 percent other final circuits, starting dip checked separately | the budget the runway eats into |
| IEC 61800-5-1 | electrical safety of power drive systems, including earth leakage protection | drives can produce smooth DC residual current, so a Type B or B+ device is required |
The IEEE 519-2022 voltage table is worth carrying around, because it is the one buyers quote without the qualifier. At the point of common coupling below 1 kV the limit is 5.0 percent for any single harmonic and 8.0 percent total. From 1 to 69 kV it tightens to 3.0 and 5.0. The current table that everybody forgets scales with the short-circuit ratio: 4.0 percent for the fifth harmonic and 5.0 percent TDD when the ratio is below 20, rising to 15.0 and 20.0 when it is above 1,000. A crane in a plant with a stiff supply has more headroom than the same crane at the end of a long rural feeder, and the equipment is identical.
On earth leakage, the crane-specific point is that a Type AC or Type A device can be blinded by the smooth DC component a drive produces under a fault. Published device literature for crane installations recommends Type B or Type B+ with immunisation against nuisance tripping on long cables, and notes that an integrated EMC filter is only declared effective within about 5 to 10 m of shielded motor cable. On a double girder with a panel at one end and a hoist at the other, that distance is routinely exceeded, which is exactly the situation the immunised devices exist for.
Ten lines, all of them answerable on one page by any competent supplier. I would rather have these than a compliance certificate with no numbers behind it.
1. Number of separately switched motors and, for each, whether the front end is 6-pulse, 12-pulse or active. 2. Reactor or filter fitted, percentage impedance, AC side or DC link. 3. Busbar material, current rating, number and position of feed points, and the calculated drop at the crane's current over the actual bay length. 4. How many cranes share the bar and what simultaneity the design assumes. 5. Panel heat load in kW, cooling method, and the ambient temperature at the panel location in summer. 6. Coil voltage and hold-in tolerance, and whether the control circuit has its own transformer. 7. Motor insulation to NEMA MG 1 Part 31 if inverters are fitted. 8. Earth leakage device type and setting, and whether it is immune to multifrequency residual current. 9. Duty class to ISO 4301, FEM or CMAA, with starts per hour. 10. Which harmonic limit the supplier is designing to, and at which point of measurement.
If you get a quotation that says only that the crane is CE marked, none of those ten have been answered. That is not a criticism of CE marking. It is an observation that CE marking says nothing about the tenth crane in your bay.
For machines already running, the order of work below is what I would actually do, sorted by return rather than by elegance.
| Action | Cost | What it fixes | Payback |
|---|---|---|---|
| Move the panel out of the hot zone and add a fan kit | USD 400 to 1,500 | 20 to 25 K of internal rise, restores full drive rating | under 1 year where trips are frequent |
| Feed the control circuit from a constant-voltage source | USD 200 to 600 | contactor dropout on dips | immediate |
| 5 percent reactor on every drive with none | USD 200 to 600 per drive | THDi from about 100 to about 40 percent, power factor from 0.69 to 0.93 | 6 to 18 months on demand charges |
| Second feed point at the mid-span of the runway | USD 1,500 to 5,000 | halves the drop on a long bay | the only real cure for weak motion at the far end |
| Phase-shifted feed for the auxiliary hoist | USD 2,000 to 6,000 | cancels 5th and 7th between the two largest drives | 2 to 6 years depending on the limit |
| Active front end on the main hoist | USD 3,000 to 9,000 premium | THDi under 5 percent, unity power factor, regeneration | 3 to 8 years, faster on high duty |
| Harmonic filter on the plant busbar instead of the crane | USD 4,000 to 12,000 | the shared problem, if several cranes contribute | depends entirely on what else is on the bus |
| Fitting a harmonic filter to a contactor-switched crane | USD 3,000 to 8,000 | nothing, because there are no rectifiers | never. Fit a soft starter instead |
Costs are our planning figures for 2026, European and Middle East supply. The last row is in the table on purpose. It is the mistake I see most often, and it is the one that gets signed off because the word filter sounds like engineering.
Yes, but not for the reason people assume. The two hoist drives add their harmonic currents at the busbar before the current reaches your transformer, and similar drives with similar phase angles add close to arithmetically. A published laboratory demonstration measured the fifth harmonic from one drive at 10.9 A and roughly double that with a second drive running. The auxiliary is the more awkward of the two because it is small and runs at a poor operating point. Our double girder crane range lists both hoists separately for exactly this reason.
No, and this is the part that surprises buyers. A reactor takes an individual drive from about 100 percent current THDi to about 40 percent, which is a large improvement and worth the money. On a five-drive busbar the resulting total demand distortion still lands around 15 to 18 percent even if the phase angles distribute perfectly, against an IEEE 519 limit of 8 percent. The reactor is necessary and not sufficient. Phase shifting or an active front end is what closes the rest of the gap.
Because the busbar is the circuit, and the crane collects current at whichever point it has reached. A published conductor sheet gives 0.000138 ohm per metre for a 1000 A aluminium bar, which puts a 245 m bay shared by cranes drawing 900 A at 52.6 V of drop, or 13.2 percent of 400 V. A single double girder at 45 to 84 A eats only 0.4 to 2 percent, so the symptom you are seeing is almost always a shared bay. Add a mid-span feed point before you replace anything.
Yes, and many do. A published 50 tonne, 13.5 m double girder lists 44.4 kW of electrical equipment with two-speed pole-change motors on hoist, trolley and long travel, and no drive anywhere. That machine produces almost no harmonic current, because a pole-change induction motor is a linear load. It produces a starting current six to eight times full load instead, and the correct mitigation is a soft starter, not a filter. Ask which design you are buying before you accept a harmonic solution.
Because two inverter groups share one enclosure and inverter losses run at 3 to 5 percent of rated power. A published worked example puts a single 45 kW hoist panel at 1.68 kW of internal loss in a box that sheds 471 W naturally at 15 K rise, which settles at a 53.5 K rise and about 95 C inside at a 42 C workshop. Inverters derate 1.5 percent of output current per degree above 40 C, so a 45 kW drive behaves like a 35 kW drive. Fit the fan kit and put the brake resistor outside the cabinet.
Panel cooling, then control-circuit voltage, then reactors, in that order. The fourth item that pays is a second feed point on the runway, but only if the symptom is weak motion at the far end rather than distortion. If your plant applies a power factor adjustment, a few hundred dollars of reactor on a 24.1 kW crane saves around USD 2,700 a year at a true power factor of 0.69, based on a published tariff charging about USD 23.37 per kW of demand.
Related articles:
Crane Power Quality: What a Single Girder Crane Does to a Weak Grid
Double Girder Crane Energy Consumption: What a 5 to 100 Tonne Machine Really Costs to Run
Double Girder Crane Installation: Assembling the Bridge on Site
Crane Embodied Carbon: What a 20 Tonne Double Girder Actually Carries
Hoist Energy Consumption: Where the Kilowatt Hours Actually Go