Parallel Operation of Small Interior Permanent Magnet Alternators


Small (kW range) portable AC generating units are used to drive isolated AC loads and normally consist of a diesel or petrol prime mover and a wound-field synchronous generator.

The ability to operate these units in parallel is convenient as it allows operational and efficiency advantages. Replacing the wound-field generator with a permanent magnet generator improves the system efficiency and reliability. This paper experimentally investigates the parallel operation of the proposed permanent magnet generator and demonstrates good parallel operating performance.


I. Introduction

Portable AC generating units designed to drive small (kW range) isolated AC loads and normally consist of a diesel or petrol prime mover and a wound-field synchronous alternator. A control unit is used to regulate both the AC output frequency by varying the speed of the prime mover, and the output voltage by varying the field current of the synchronous generator.

It is often preferable to have two or more smaller generating sets (gen-sets) operating in parallel rather than a single larger unit as smaller units are easier to transport and it allows one of the units to be stopped at light loads which can significantly improve the fuel efficiency. Parallel operation also enables easier maintenance, future expansion and improved reliability based on the ability to have a level of spinning reserve or redundancy.

When alternators are paralleled, they operate at the same voltage and frequency and it is important that they share the real (kW) and reactive (kVAr) power load in proportion to their rating. The real power sharing is related to the prime mover speed (throttle) control and the reactive power sharing is related to the field current excitation control [1].

A common passive means to parallel alternators is using droop control, where the gen-sets are set to have the same no-load speed and voltage, and the speed and voltage of both units is designed to fall (droop) by the same amount (typically 3 to 5%) at full-load. Alternatively a more complex active parallel control system can be used which monitors the real and reactive output of each generator and controls them to be in proportion to their rating without requiring the need for a change in the output voltage or frequency with load [1].

Recently a novel interior permanent magnet (PM) alternator has been developed as a replacement for the wound-field alternator for small (kW range) gen-sets [2]. Despite its lack of field control, by careful design of the alternator it has an acceptable voltage regulation. Compared to the conventional wound-field alternator it offers higher efficiency and smaller size and weight.

This paper examines the parallel operation performance of the interior PM alternator.

Fig 1 and 2

The cross-section of the two-pole, 3,000 rpm, 8 kVA interior PM alternator is shown in Fig. 1 and a photo of the complete alternator unit is shown in Fig. 2.

Tests were performed using two small single-cylinder engines, one diesel and one petrol driven, both with identical two-pole three-phase interior permanent magnet (PM) alternators attached. The diesel engine was rated at 12 hp (9 kW) while the petrol engine was rated at 13 hp (9.7 kW). The engines were made by different manufacturers. The different engine types were chosen to test the ability of paralleling systems with different load characteristics.

A paralleling switch was set up with a three-phase contactor and a simple electronic circuit that detected the point when the phase waveforms were in synchronism. The neutrals of the three-phase circuits were connected together, leaving the three phases open to be connected when the contactor closed for paralleling.

This paper describes the series of tests done to verify the parallel operating performance of the alternators. Section II describes the steady-state load sharing testing, section III discusses the step load transient waveforms and section IV investigates the capability of the paralleled alternators to start induction machines.


II. Steady-State Parallel Testing

This section of the testing investigated the ability of the paralleled alternators to share load under steady-state conditions with a three-phase resistive load bank.

The first step was to conduct load tests on each of the two generating sets (gen-sets) separately and generate a load voltage droop curve for each. These are shown in Fig. 3 (a). Although the alternators are the same, the engine throttle responses are quite different. This results in different voltage versus load curves for the two engines. This would generally make it difficult to parallel the two generating sets (gen-sets) with simple controls.

The two gen-sets were then paralleled using the simple switching unit described earlier. The paralleling method used in this test was droop paralleling. This method requires the voltage of the alternator to drop as the load current increases. This allows the alternators to automatically balance the load between the two paralleled alternators. Conventional wound-rotor alternators need to add a droop current transformer sensor with feedback to the automatic voltage regulator (AVR) to do droop paralleling

The first tests were taken without any attempt to adjust the engine throttle to balance the voltages at no load.

Fig. 3 (a) also shows the load voltage curve when the two alternators were running in parallel. The paralleled curve follows the petrol engine voltage load curve until the load gets to 3,500 W, and then the diesel engine begins to share the load. From 4,500 W to the maximum load of 12,500 W the diesel engine takes an increasing share of the load (based on its output current) until it is in proportion to the engine ratings at maximum load. The line THD stayed in the range 1.4-2% while the phase THD increased from 2 to about 6%.

Fig. 3 (b) shows the output currents of the two paralleled alternators as well as the circulating current between them. A difference between the no-load output voltage operating points of the two alternators produces a circulating current of approximately 6 A between them at no-load. This produces a circulating reactive power between the gen-sets. The no-load winding copper loss was calculated at around 150 W total, 75 W for each alternator. This is well within the limits of the alternator windings. At higher loads the circulating current falls to about 2 A.

Next, the petrol engine’s throttle was adjusted to get a lower circulating current at no load. The results are given in Fig. 4 and show that the circulating current is much lower, less than about 2A over the entire load range. The alternators now appear to share the output current much more in line with the engine ratings. The nett output power versus droop curve is comparable to the case before the throttle adjustment.

Fig 3

Fig. 3.   Before throttle adjustment. (a) voltage droop versus output power for two gen-sets independently and in parallel. (b) current from each alternator and circulating current.


Fig 4

Fig. 4. After throttle adjustment. (a) voltage droop versus output power for two gen-sets independently and in parallel.       (b) current from each alternator and circulating current.


Figure 5 shows examples of the line voltage (larger amplitude, blue) and phase voltage (smaller amplitude, red) waveforms under parallel operating conditions at no-load (a) and with a 3ph, 12 kW balanced resistive load (b). They show reasonable quality sinewaves with acceptable total harmonic distortion.

Fig 5

Fig. 5. Parallel waveforms. (a) no-load voltage waveforms (scales 10ms/div). (b) voltage waveform under 12 kW load (scales 160V/div).


III. Load Change Transients

An investigation was undertaken to look at the voltage waveform transients with a load change of the paralleled gen-sets.

Figure 6(a) shows the waveform when applying a resistive load of 12 kW. There is only a small voltage glitch, as the waveform “shifts” to the right under the new load. This is caused by the flux drag effect of the alternators where the load causes a rotation of the stator magnetic field with respect to the rotor from its open-circuit position. The effect is instantaneous. Note that the engine flywheel momentum prevents frequency change or extra voltage drop due to speed change within the time-scale of the graphs.

Figure 6(b) shows the waveform when a resistive load of 13 kW is suddenly removed. You can just see the point where the load is removed. It is most noticeable in the phase voltage waveform where the waveform shows a sharper peak as the waveform moves back to the no-load position. It happens in a fraction of a cycle. There is no voltage overshoot, and very little waveform distortion.


Fig 6

Fig. 6. Transient line and phase voltage waveforms under step load changing. (a) 0 to 12 kW (scale 10ms/div and 160V/div) (b) 13 to 0 k W (scales 7.8ms/div and 130V/div)


IV. Motor Starting Transients

An investigation was carried out to look at the improvement in motor starting capability produced when the two gen-sets were connected in parallel.

When starting integral kW induction machines direct-on-line, they draw a high starting current which is typically several times their rated current for a fraction of second. It is important that the alternator output voltage does not drop by more than 30% during starting as this can cause other equipment connected to the alternator to malfunction. For instance fluorescent lighting may stop operating (blank out) and contactors may drop out.

The starting of a standard 5.5 kW, three-phase, 3000 rpm induction motor was considered. From the measured characteristics from Fig. 4, one alternator alone would be expected to have difficulty running the motor under full-load and so would likely have difficulty starting it where the currents are much higher.

Figure 7 (a) shows the waveforms for the induction motor start when the petrol engine was used alone. When the motor is connected, the voltage drops to around 190 V (rms), an approximately 50% reduction which is unacceptable. Thus the petrol gen-set is not appropriate to start a 5.5 kW three-phase motor by itself.

Figure 7 (b) shows the waveform when the two gen-sets are connected in parallel. Now the voltage drops to around 290 V which is within the 30% voltage drop requirement. The noticeable voltage increase on the right-hand side of the plot shows that the motor is accelerating more rapidly than with one alternator alone. The increase in speed causes the current to drop and thus the voltage to rise. This demonstrates that two gen-sets operate effectively in parallel mode during transient conditions such as motor starting.


Fig 7

Fig. 7. 5.5 kW motor starting transients. (a) petrol alternator alone (scale 31ms/div and 138V/div) , (b) both alternators in parallel (scale 38ms/div and 135V/div)


V. Conclusions

The two generating sets (gen-sets) with interior permanent magnet (PM) alternators were able to be operated in parallel without any problems. They shared the load roughly in proportion with their rated engine power and operated together like a single larger gen-set as far as output power and motor start capability. The different engine types made no difference in the paralleling operation. The testing showed that paralleling these types of interior PM alternators was simple and effective, even when the driving engines were different. Little waveform distortion was noted when alternators were operating in parallel or alone. They ran smoothly together under all loads.

Although this digest only shows three-phase paralleling of two 8 kVA alternators, single-phase alternators up to 20 kW each, and four-pole three-phase 24 kW alternators of the same type, have been successfully paralleled with no issues.

The full paper will contain further experimental results including examination of the current sharing between the alternators under transient conditions.



[1] G. Olsen, “Paralleling Dissimilar Generators : Parts 1 to 3,” Cummins Power Generation White Paper, Available from:,%20Case%20Histories,%20Technical%20Papers

[2] “Interior PM Generator for Portable AC Generator Sets”, ECCE 2014.

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *