In anything other than an induction set, it is more like something between a continuation of the total load current at the time and more like twice whatever that was, perhaps rising after a few seconds as the controller realises the volts are dropping, and opens the throttle. The voltage collapses catastrophically, so  say if there was 100A of load current at 250v  before fault came on, then you may get no volts at all and about 200A, and hopefully that is enough to prompt clear the fuse, and then the volts can rise back up again, and it settles back. Users see the lights flicker, and maybe the genset coughs a bit.


However that  it is not enough to promptly clear the fuse then either the controller puts more fuel in, the genset slows and stalls, or a shear pin breaks depending on the design of the alternator and prime mover.

In an induction set the fault current rises until the volts drop to near zero, then it stops generating altogether, and the engine races away.


M.

Alright. According to Kohler its 300% of the FLC for 10 seconds at the top of page two:

http://resources.kohler.com/power/kohler/industrial/pdf/g5381.pdf

Also came across this:

The problem is that "Zs" is not single valued and depends on the magnetization state of the machine at the time the fault comes on.

In an induction machine the shaft rpm and frequency are related but not in step, there is a 'slip' as there is in an induction motor, and that frequency difference between revs*poles and the output waveform is the frequency of the AC that is circulating in the spinning armature. We like induction machines for pulse loads (things that have a 'firing pulse' nature) because the output frequency and the armature rate can drop almost at once, but power is still delivered, and  here you may get more current than I suggest.. However they do not like reactive loads, especially variable reactive loads, and are normal used only when the load is well known - floodlight trailers a commercial example that often use induction sets. (even so an impedance limit of 10 times FLC would be a heavy metal machine that is under-run for its FLC)


In the more common forcibly excited machine, (even one with no brushes, but an excitation winding and the rectifiers on armature), the shaft revs set the output mains frequency , and the voltage regulation is performed by winding the current in the rotating magnet up and down, which as a side effect makes the torque for the prime mover vary to scale with the amps. Fast faults happen on a time scale faster than the magnetic field can be changed, so the Zs of the moment is rather dependant on the current load, and the way you try to measure Zs..


All this is solved by having one or more earth fault relays that can be set to trip at several  amps, or even tens on a big set, but only after  300msec or 1000msec delay. For small faults fuses will work like normal, as will normal and S type RCDs. If a serious fault occurs and for whatever reason cannot be cleared, then the EFR will trip and serve like the main fuse for the branch it supplies.


Mike

Right, but when the fault happens a few cycles latter the excitation would increase the voltage, and the engine will try to keep up torque, increasing the output current. If the machine can output a value mostly in the 7x range for 5 seconds the MCB can open without the use of RCDs and the like.

This is getting very confused. The engine will only produce its rated power assuming the frequency (RPM) is constant. All other associated power can only come from the rotating inertia of the engine and generator rotor. A complete short on most sizes of generators will stop the machine very quickly, perhaps 10 or 20 revolutions. The moment the engine speed falls so does its output power, power = torque x RPM, and the torque is limited at all speeds and gets less as the speed falls in most machines at limiting power output. Your document is probably not very accurate for most reasonable-sized generators say <500kVA. Also, the AVR is designed to reduce the excitation if the frequency falls, and usually to make it zero at 40Hz for a 50Hz machine to protect the mechanical parts from failure. The forces involved may be very large, sufficient to break generator shafts, or piston rods or the crankshaft with a low impedance short circuit. Such failures do occur in generator sets sometimes because those first few revolutions are critical, the short circuit current is not in any way "free", it is a real very large amount of power. Just because the short collapses the voltage does not reduce the real power in any way, the alternator is still making 230V or whatever, it is almost all lost in the windings and the external circuit as resistive loss, although perhaps 10% may be lost in the reactance.


I'm afraid Mike has made a slight slip, a synchronous alternator does not suffer "slip" as such, the output exactly follows the machine rotation at all speeds, in the same way that a separately excited motor does. Slip in an induction motor is simply to make the rotor magnetic field, in an alternator, this is externally applied DC and only controlled by the AVR. Small machines do sometimes operate as asynchronous machines but these have very poor overload and voltage control characteristics.


One further point is that most machines that are brushless depend on simulated transformer action to get the AVR excitation power to the rotor, where it is rectified and gives the rotor field. This is a very important part of our size of the machines under discussion. There are a number of various ways to do this, which are usually covered in Patents, it is easiest to just look on this as I suggest, although there may be a secondary alternator action or similar involved. It makes little difference to the operation.

I'm afraid Mike has made a slight slip,

At the risk of appearing a bit of a pedant I disagree, and I stand my ground.


An Induction genset, like an induction motor is asynchronous and does slip - it is just negativein that you spin it faster than synchronous and get energy out. It  has a monobloc shorted squirrel cage rotor, just like an induction motor, and no brushes. The revs have to change to keep the volts in tolerance, and the phase angle of the load is very critical to get interaction of slip rate  with V and F right. There is no AVR winding, Such machines can have a relatively good response to pulse resistive  loads == low "Zs" if you like, up to the point of total field collapse but are a sod to stabilise for changing reactance loads. All you can do is look at the output voltage and retrospectively wind the throttle to change  revs and flick capacitors in and out of circuit. I have sat astride too many of these with instruments to be unsure how they work?. (the smell of diesel in the morning and all that) The effect of one phase seeing a different capacitance to the others is interesting but I'll leave that for now. Actually in the very short pulse domain (tens of micro seconds now )the high surge current is propped up by the capacitors as well. Speaking only as a happy customer I can say that Fisher Panda make some very good asynchronous machines using this principle. Just do not expect it to be spot on-freq. ever. They can however be made short proof, due to the field collapse and that is sometimes a nice feature to have in the field.


However most generators that are made to be used with any arbitrary load or to run in parallel are certainly not that sort, as the V-F interaction with load phase angle is too hard to manage. Instead they are indeed the synchronous sort that Dave is thinking of, and  have an excited rotor that is in effect a slowly changing DC magnet spinning at exactly pole frequency, driven either by brushes and slips or by rotating transformer and on-armature diodes. Then the rotor speed and freq are in perfect lock and the rotor current varies to maintain volts, at least over a limited range. However, as noted the effective 'Zs' is less stiff. I'm aware of controllers having low frequency lock off, but I defer to Dave's greater knowledge as to what level you would set it to.


My original point was to be very careful with rules of thumb about short circuit current comparing extracts from different text books unless you know a lot about the particular genset under discussion. It may be more or less than you think, usually less.


M

From the large print extract:

Generator can sustain 2.5 to 3 times rated current with a 3-phase fault applied for approximately 10 seconds without damage to alternator.


The alternator windings may well be able to, but I'm struggling to accept that the associated prime mover can produce 2.5 to 3 times it's rated output for a second, yet alone 10 seconds. What size is the flywheel?!!!!!


Regards


BOD

Ah Mike, but I did specifically talk about synchronous machines! As it happens I do have a small generator of the asynchronous kind, but to be honest (and as I said) it is not very good at either constant frequency or constant voltage. All of the generators we are talking about are of the synchronous kind, and although other asynchronous types are available (and sometimes the military use them for other reasons) machines in the 25kVA and up are normally fully synchronous types.


I agree with you BOD, there is no way that a 100HP diesel can produce 300HP even for a second or two, I can get that much given a bigger turbo and a lot more fuel injection (and black smoke) and.... but it is not there in the standard model. The overload characteristics are only the inertia and the rated power, and if the speed reduces for whatever reason the power may be as rated for a small change, but much and it goes very quickly. Engines for generators are very different in "tuning" to that in a car, the speed is expected to be constant at a particular value, and the fuel consumption is minimised at that speed to the loss anywhere else, particularly of torque at higher or lower speed. In the generator world, fuel consumption is everything, it is entirely driven by operating cost. A typical generator for film or television lighting is a peculiar beast, it may well have an external television sync input, and be crystal controlled for frequency to 3 decimal places for frequency (which might even be 48 or 50Hz, or 29.970*2 in the US). It is easy to think all generators are the same, they are not, and ones that can sync up and work in parallel are much more complex than one might expect.

My concern is that ProMbrooke seemed to be mixing quoted fault levels from various sources and was not aware of the huge effect that the generator type and design has on the transient output characteristics. It may be that larger async machines are more popular in Germany.


Flywheel moment of inertia is normally calculated large enough to store enough energy to smooth over between the firing of successive cylinders with acceptably low torque ripple (ripple p-p perhaps 1 to 3% of the average torque, probably a bit less on a Rolls-Royce job, and perhaps more vibration allowed on a lawnmower where a large diameter wheel would be hard to fit in, and anyway you may prefer it to stop dead if it hits something hard. ).

So you could say it stores the energy of maybe 30 to 100 revs at nominal motor output power. So  if the PM dies, it will run down to 1/e^1/2 of original speed in that many rotations.

edit sorry divide by no of cylinders that fire per rev.


In practice  in the case discussed the shaft speed will be well on the way down in a fraction of a second.


M.