The generation of electrical power is mostly done by a synchronous generator. This machine is robust, simple to control and almost maintenance free in a brushless (rotating rectifier) version. It generates electrical power with a frequency proportional to the speed of the rotor, so the electrical frequency and mechanical speed are synchronous, which explains the machine's name.
1.1 Working principles
The machine generates electrical power in the stator coils by a rotating magnetical field which is generally produced by a DC magnet. The DC power for this magnet can be supplied externally, but it is convenient to integrate a small generator, the exciter, on the same shaft within the same housing. In any case to feed this magnet, the DC current has to be supplied to the rotor. A simple way is through two slip-rings, but this involves wear and regular maintenance. A more sophisticated, but maintenance free solution is to integrate a small AC generator (exciter) on the rotor, rectify the current and supply it to the rotor coil, all on the same shaft, avoiding any slipping contacts (therefore rotating rectifier). The excitation current for the exciter is provided by a DC current to the exciter stator which is generated by the output power of the main generator.
This method allows to design a clever system, integrating even a simple but efficient voltage control. Since the voltage of the main generator drops with an increasing load, its excitation has to be increased to compensate. This requires more current from the exciter, which is achieved by increasing its excitation. So the DC supply for the exciter stator can be used to control the main generator's voltage. Such a control is done electronically by an automatic voltage regulator (AVR). A certain output voltage is preset and the control adjusts the excitation to keep the output voltage under all loads close to the preset value. Today's brushless, synchronous generators include an AVR.
The machine is able to self-start by its residual magnetism: as soon as the machine rotates, a small output voltage is generated starting to feed a small current through the AVR to the exciter, which in turn increases the output voltage. This control loop develops the nominal output voltage within a few machine sums. It is possible that the residual magnetism was lost. In this case it has to be rebuilt. The same method as for asynchronous machines can be used and is described in the next chapter.
The frequency, however, has to be regulated, as an increased load will decrease the rotor speed. Such a speed control could be most simply done by hand or automatically by an automatic governor.
At least two important design/selection considerations for generators related to the control of MHP sets shall be pointed out here:
- the generator must be equipped with special bindings able to withstand the runaway speed of the turbine (in case of crossflow-turbines approx. 1.8 times the rated speed).
- if possible use standard generators. BYS Nepal for instance uses wherever possible 4- pole, brushless-type generators, equipped with an electronic automatic voltage regulator (AVR). These provide fairly constant voltage over a wide speed range which is important for manual or simple governor-based flow controls. Moreover they require virtually no maintenance.
Fig 1 Principle diagram of a synchronous generator.
As described in the previous chapter, the conventional generator for MHP (or PHP) is the synchronous generator. Specially for isolated grids it is often thought to be the simplest and adeguate solution. There are, however, some situations where an asynchronous generator offers the same or even better possibilities, despite its well known disadvantages described below. Two such cases are:
- the system is (even if not immediately, but in some near future) to be connected to a larger grid (e.g. regional, national supply...)
- investment capital is scarce and the planned system must be simple.
This chapter shall introduce the reader to the basics of asynchronous generation and help to make decisions.
2.1 The Asynchronous Generator: an Induction Machine
The induction machine is very simple in construction, but tricky to understand and control. To get a feeling for this machine a short comparison with its competitor, the synchronous generator is given here.
- easily available in all power categories.
- cheap (compared with a synchronous generator at the same power rating). simple and robust construction.
- does not need DC excitation. does not need synchronization when paralleled to a grid. needs little maintenance (no slipping electrical contacts).
- self-protecting. In case of a short circuit or overload, it de-excites automatically and remains de-excited.
- simple operation. wide spread. Almost every electro workshop can handle it.
- needs external magnetization energy -> additional equipment is needed for isolated operation.
- less efficient, especially below nominal power.
- less rotating mass -> less inertia, riskier toover speed.
- power characteristics are highly sensitive on the load -> complicated electronic control is needed in case the load is not constant.
- loss of residual magnetism in a short circuit or overload -> loss of self-start capacity.
Talking about an asynchronous generator in this chapter, we will actually always refer to an asynchronous motor (same as induction motor) run as a asynchronous generator (IMAG = Induction Motor used as Asynchronous Generator). Almost any induction motor can be used as a generator, as seen by studying its symmetric torque/slip diagram. Note: the slip is the difference of the electrical frequency and the mechanical frequency (speed) normed with (divided by) the electrical frequency.
Fig 2 Torquel slip (speed) diagram of an asynchronous machine.
slip<0 over-synchronous speed -> rotor is forced to turn faster than the magnetic field, mechanical power is transformed into electrical power. GENERATOR range. slip = 0 synchronous speed -> rotor is running idle, no power transfer, no torque.
0 < slip < 1 under-synchronous speed -> rotor is driven
by the faster rotating magnetic field, electrical power is transformed into
mechanical power. MOTOR range.
slip = 1 standstill -> rotor is blocked, no power transfer.
slip > 1 reverse speed -> rotor is forced to turn against the magnetic field. BREAK range.
The slip is a relative value and characteristic for asynchronous machines. The generated power (mechanical or electrical) depends on the difference between the electrical frequency (rotation of the internal magnetic field) and the mechanical frequency (rotation of the rotor). Hence the electro mechanical coupling of such a machine resembles a mechanical clutch, where the torque is transmitted over to slipping surfaces. A difference is that an asynchronous machine does not couple without slip. The power transfer is zero if both the electrical and mechanical frequency are equal (at s = 0 or synchronous speed). The clutch on the other hand couples best at synchronous speed! This behavior makes the asynchronous machine difficult to imagine.
To work as a generator the following main conditions must be fulfilled:
- the prime mover (turbine, motor ...) must speed up higher than the synchronous speed in order to produce slip and torque.
- the IMAG is not able to produce the energy for magnetizing its coils. It must be supplied the required reactive power from an external source (like a grid or a capacitor bank).
The torque/slip relation shows a pronounced maximum torque. As a motor this means loading the machine with a torque higher than this maximum will stop it; as a generator this means the break down of the power generation.
2.2 Working Principles
Induction machines are often called 'rotating transformers'. The shorted (secondary) winding on the rotor, the 'squirrel cache', is excited by the (primary) stator winding, which in turn is excited by the rotor. That is a transformer and in fact, with a blocked rotor, the asynchronous machine is a normal transformer with a shorted secondary winding.
Fig 3 Principle diagram of a asynchronous generator.
We see (Fig 3) that the trick to use the shorted rotor 'coil' to produce the excitation current results in an amazingly simple construction (compare with Fig 1). Besides the stator and the rotor coil nothing else is needed. The rotor coil degenerates to some aluminum or copper bars embedded in the rotor iron, shorted at the ends with Cu or Al rings (therefore squirrel cache). This simplicity of construction is unique but also disables a direct influence on the generated voltage (via excitation current) and moreover there is no simple relation between speed and frequency.
Fig 4 A simplified physical representation of a transformer. is the magnetic flux separated in a coupling component (index m) and two stray components (index a). This flux is modelled in a lumped circuit as inductance.Resistive losses are indicated as resistors (index L for coils, index Fe for iron)
To describe the behavior of the IMAG a simple electrical circuit is used. The machine is modelled by a transformer T-circuit (because it resembles a T see Fig 4), including coil losses (RL) and stray losses (L) as well as iron losses (RFe) and the machine's inductance (Lm)
Working as a motor, the mechanical load is indicated as a resistor R which is a function of the slip s. Working as a generator s and hence R becomes negative, which means it becomes a power source (a negative resistor produces power = generator).
But a real, negative load cannot produce reactive power, therefore in generating mode the IMAG needs an external source for reactive power.
In case of a parallel operation to a larger grid, the reactive power can be drawn from it and there is no need for additional care (besides normal power factor correction). For an isolated operation, however, we need to look at the machine a bit closer:
In the part 'power factor correction' we see that the rcactive power of an inductance can be supplied (compensated) with a corresponding capacitor C. For a complete compensation, we would need an exact value of C in order to have identical Ullines (both are stright lines) in the current/voltage diagram (see part 'power factor correction' Fig 6). Fortunately, in
Fig 5 Ul curve for the saturating excitation current and U1 lines for two different capacitors at constant frequency. The resulting two stable operating points (OPT and OP2) are reached at very different output voltages U1 and U2. The circuit besides shows the pow of power. The required magnetization power (reactive Q) is supplied by the excitation capacitor.
Fig 6 Relations of UI curves for varying excitation currents at constant lo,,, and the corresponding UI curves for the output currents at constant C. For a constant capacitor value the voltage varies with the output current along the C-lines (a-a', b-b'). the point c indicates that for C3 the maximum output current is reached. All curves at constant frequency.
induction machines the reactive power need depends on the load and hence on the excitation current. This is due to the magnetic saturation effect of the iron. If we draw both the UI line of the excitation (bent line) and a capacitor (stright line), we see that a stable operation point, namely the point where the two lines cross, is possible (see Fig 5).
Unfortunately, as soon as a (variable) load is added and an additional current flows, two things happen: the generator voltage drops (see Fig 6).
- the frequency drops too (as it is proportional to the difference of rotor speed and slip, and the slip increases with the load (increased torque). Is the reactive power supplied by a fixed capacitance, the output voltage U varies widely with the output current I and there is a maximum current Imax. Each chosen C will produce another curve, with a higher I for greater C.
Under load the IMAG changes voltage and frequency and shuts off at a certain current Imax ( = a certain load). For almost all situations a control of U and I is needed. Only where the load is not varying (e.g. a single, permanent load...) or the generator is parallel to a large, rigid grid, we will not need any.
2.3 Voltage & Frequency Control for IMAGs
A control would measure at least the following values: voltage,
current, phase, frequency and speed to determine the values for C, speed, dump
load... to prevent the voltage U and the frequency f from exceeding nominal
values (see Fig 7).
We actually need two controls: one for the frequency and one for the voltage, but both are interlinked which needs some care. Changing for instance the speed to adjust the frequency will also change the
output voltage and vice versa. To avoid instabilities the control system has to be designed carefully. Here we will only mention some strategies to control the output voltage.
Fig 7 Control scheme for an IMAG in a MHP.
The following table (see Table 2) shall indicate some of the possibilities for a voltage control, which are described in some details afterwards. This will deepen the understanding of an IMAG and might allow to select a adequate control system:
Fig 8 Reduce the output voltage swing by adding a saturating inductance.
Table I Possibilities to control the output voltage for a supply system with IMAG. The sequence corresponds to the paragraphs below.
a) Passive regulation with saturating inductance
The inconvenience with a constant capacitor is its excess reactive current at no load, when it is correctly dimensioned for nominal load. This excessive current overexcites the generator and results in high overvoltages.
By adding an inductance, which saturates at the nominal output voltage, some of the excess reactive current will be absorbed and the voltage stabilized. The capacitor has to be increased to compensate the decreased excitation current at the rated output. Besides the costly coil and moderate efficiency, this compensation ('control') is very simple not needing any further active control elements.
b) Voltage control by regulating solely the speed Omega r
Is the frequency of no interest (for instance in DC applications like battery loader, lighting with bulbs...), the voltage can be adjusted by solely changing the rotor speed. The capacity is prefixed for no load.
Fig 9 Controlling solely the voltage by regulating the speed. The frequency doesn't matter.
The control senses the voltage error AU and corrects the speed. In a simple setup this can be done by hand.
c) Voltage control by regulating C
In most cases we need also a good enough frequency stability (for instance 50Hz+-1%). Besides the rotor speed we need to control another parameter, conveniently the exciting current by varying C. If we tolerate voltage 'jumps', the control may be discrete, this means C could be switched to lower and higher values at certain currents (see Table 2), ensuring that the voltage remains within a defined range.
Fig 10 Switching between four combinations of excitation capacitor values to reduce the voltage swing.
The values for C are selected to keep U-range in acceptable limits.
A possibility to switch between four different C's is shown in Fig 10.
CO, Cx and Cy have to be determined experimentally. The switching to new C's will change the frequency and so the rotor speed has to be readjusted (which changes the voltage again, which is adjusted by a change of C, which changes again the frequency, which... you see the risk for instability?).
This system is relatively simple, even though using some control electronics. Such kinds of controls are widely used in voltage stabilizers ('fridge guards'...) and readily available.
d) Reducing the voltage 'jumps' in c) by combining a) and c)
We have seen in a) that the voltage variation for different loads can be reduced with a saturated inductance. Combining this solution with e) results in a smoother voltage regulation (see Fig 11).
e) Voltage and frequency control by load regulation
If the primary energy is abundant, the system could always be run at nominal power, avoiding any adjustment of C or Omega (similar to b). If the load varies, we could add dump loads to make up for the difference.
Fig 11 Reducing voltage variation by combining switchable C's with a saturating inductance.
Table 2 Capacitor bank switching: status of switches and the corresponding total values for C.
The control triggers at the prefixed output currents Iswitch. Two switches x and y; position 0 = off 1 = on.
Fig 12 Stabilizing the voltage and frequency by keeping the load constant. By adding dump loads the power output swing is reduced. The number of resistors determines the swing amplitude.
Table 3 Dump load switching: status of switches and the corresponding total values for R.
Two switches x and y; position 0 = Off 1 = on;
Table 4 Jet switching: status of switches and the corresponding input power and total values for C.
The control system is very much the same as in c) except we switch resistors instead of capacitors.
This dump resistors can of course be for any kind of non-time critical use (for instance heat or freeze water in a storage tank, storage cooker...) provided the average dump power is sufficient for this purpose.
This control system is cheap and convenient in most cases. It eliminates the need for a speed control system. The dump loads can also be connected anywhere on the grid, they don't add reactive power to the distribution lines.
f) Voltage and frequency control by load and input power regulation
If prime energy is scarce, the control system e) is not feasible. In some situations, however, it is easily possible to regulate the input power discretely. Suppose we use a pelton turbine with fourjets. We could vary the power by simply switching the jets on or off in the range from l/4Pmax to Pmax (see Fig 12). Each of this input power level is chosen to produce a certain amount of electrical power. Each will need its particular, prefixed C, which could be switched together with the valves in the same way as in c).
The varying power demand within one input power level could be compensated by switched load resistors as described in e).
We would use two nested control circuits: the first selects the number of jets (and the appropriate C), the second control (not displayed here) makes the fine adjustment by varying dump loads. A particular disadvantage of this solution is the poor efficiency of IMAGs at low power output.
2.4 Particulars & Operation of IMAG
IMAGs are competitive as long as generation does not exceed some 10-100kW and the control circuitry remains cheap, simple and reliable (or omittable). The machine itself is about 50% cheaper than synchronous generators in this power range.
IMAGs can be wired for single or three phase.
Starting after loss of remanence
It is possible to lose the permanent magnetic field (remanence of the iron), which disables the machine from self-starting. This happens for new machines or after overloading the running machine (break down of current). In this case it is sufficient to supply some excitation current by a battery to the connectors of the running generator. A short pulse is required. The quickly growing generator voltage will burn the battery if it remains connected!
Fig 13 Reduce power loss by varying Pmech, and combine it with a load controller.
Less risky is the build up of a residual magnetic field in the machine by connecting a battery for a short moment to the machine at stand still. The DC current will produce a magnetic field, which will partly remain after switching the battery off.
IMAGs will not start under load!
Connecting an IMAG to a rigid grid is very simple and represents the most economic way of using it.
-neither voltage nor frequency control is necessary, both are maintained by the grid.
-no synchronization is needed. Speed up the generator to approximately synchronous speed, switch it to the grid, increase the speed to full power, that's all.
-no reactive power source is needed for the magnetization. the grid supplies it. To improve the power factor, however, a compensation capacitor should be connected (the same as for any other motor. Its value is optimized for the nominal conditions). Careful: the capacitor must not be connected before switching the generator to the grid. The generator would excite end tee damaged,ifswitched on unsynchronized !