Bench testing
 
Web windpower.org.za
Logo
Back to windpower

Once the generator had been constructed it obviously needed to be tested. Testing by waiting for the wind to blow is frustrating, time consuming, dangerous, etc... so it was necessary to construct a bench testing set up. This allows consistent speeds, and repeatable experimental measurements.

So I cut three disks out of 5mm plywood, 2 with 20cm diameter and one of 15cm. I glued them together, and drilled holes for the drive bolts. I then went to an auto spares shop (Tonnesons) and asked for a fan belt of about 2m in length. The guy there looked at me as if I was from Mars, had three heads, and was eating live mice. "What car is it for?", he kept repeating. Eventually he grudgingly fetched a table of fan belt lengths, and I chose one that was close to what I wanted.

Next I bolted a piece of wood to the mounting frame (which I cut from an old steel frame table), positioned a 560W motor (see left) so that the belt was sort of loosish/taut, drilled holes in the wood and bolted it all down (see photo below). To my surprise, it worked well first time. The fan belt is easy to slip on, and (usually) stays on when I switch on the motor.

Subsequently, I rigged up a smaller 90W variable speed motor (see right) and used the two 7.2AmpHour batteries as the test load (at 24V). A small magnet on the back of the rear rotor triggered an electronic bicycle speedometer (calibrated to read Rpm). This motor is much less powerful than the generator, but the variable speed control allowed more controlled measurements to be taken. For the detailed results from the 560W motor and bulbs, see below. For stator 4 and 50x20x8mm N35 magnets and other results, see here)

Open circuit voltage test

A good place to start is to record the voltage generated with no load, also called the open circuit voltage. This gives an accurate measure of the 'cut in' RPM (revolutions per minute). This is the RPM at which the generator will start to drive current into the batteries. For a 24V system, this is read by tracing a horizontal line from the vertical axis at 24V until it meets the graph. Then drop down to the horizontal axis to read the RPM. There are different values, depending on the number of turns per coil, and the size and strength of the magnets (grade N35 is weaker than grade N42). It is also affected by how close the magnets are to each other - the less the clearance, the greater the voltage.



Test with stator 5 and 46X30X10 grade N35 magnets

This test showed that 24V is reached at about 120 rpm. Extrapolating the watts line, 500 rpm should give about 865 watts.

500 rpm for a 1m radius blade gives a tip speed of 3.14*2*500/60 = 52m/s. Ignoring drag, and with a tip speed ratio of 7, this is a 7.5m/s wind. With a TSR of 5, it is a 10.5m/s wind.

Test with stator 4 and 50x20x8mm grade N35 magnets (back to top)

The unloaded output (OC = open circuit) looks perfectly linear when plotted against Rpm. The loaded power output using th 90W motor tapers off at around 60W. Since the motor is 90W, and some power must be lost in the motor bearings and especially the fan belt drive and pulleys, I was pretty pleased with this result. After all the power loss from the drive mechanism, about 66% of the power going into the motor was coming out of the generator.
.

Results from 560W motor using bulbs as a load

The motor is 560W, and so is probably more powerful than the turbine, hence these readings may not be what happens live. Then again, the turbine may spin a lot faster than the test bench, so may deliver more power.

Here are the results of the tests, using 12V downlighting globes. (or click here to see summary graph at bottom of page)

Load used:
2 x 50W
3 x 50W
Rotor spacing 5+2+2+10+10 = 29mm4.15A23.2V = 96.28W3.62A26.0V= 94.12W
Removed one spacing washer = 27mm4.5A26.5V= 119.25W3.85A29.5V= 113.58W
2.5+10+10+5 = 27.5mm4.45A

25.9V

= 115.3W3.86A29.4V= 113.5W
Note that the magnets are 8mm high, so that the gap between the magnets would be 13,11, and 11.5mm.

From the above figures can be seen how reducing the distance between the rotors dramatically increases the power output. But of course this reduces the tolerances, and can result in the magnets scraping against the stator, especially when high loads cause stress and distortion of the mechanism. Also the rotor plates have quite a wobble for some reason, which means I need the 2 x 1.5mm clearance (27 - 8x2 - 8).

To measure the rotational speed, I bought a cheap (R100) electronic bicycle attachment, which has a magnet, a lead, and a LCD display. You enter the size of your bicycle wheel, and it then tells you speed. Via what was for some reason a quite intricate calculation (something to do with beer perhaps), I worked out how big to tell it my 'bicycle wheel' was, so that it gave a reading (in 'km/h') which was actually RPM. This gave 384 RPM.

Alternatively, the electric motor runs at 1420 RPM, and the belt and drive wheel made approximately a 5cm diameter, with a 20cm wheel on the front of the generator (in place of the hub). This gives 355 RPM, which is close. 384 RPM is probably far more accurate (so perhaps the drive wheel is actually effectively 5.4cm). Due to the V shape of the drive belt, it difficult to know the exact effective diameter.

2 August 2007

Here are the results for a bench test of stator number 4 with rotor spacing 28mm:

Star wiring: Load usedAmpsVoltsWattsRpmV/RpmW/Rpm
None
 
58.6
 
384
0.1526 
2 x 50W bulbs
4.76
27.6
131.4
291
0.09480.4515
3 x 50W bulbs
4.12
34.1
140.5
328
0.10400.4183
4 x 50W bulbs
3.82
38.2
145.9
345
0.11070.4230
2 x 2 50W bulbs (parallel)
6.10
12.4
75.6
231
0.05370.03274
2 x 2 50W bulbs (parallel)
6.85
15.6
106.9
264
0.05910.4048
Flat 12V car battery
6.10
15.2
92.7
241
0.06310.3847
Flat 12V car battery
6.75
15.6
105.3
260
0.06000.4050


Note how the rpm changes at different loads, probably mainly due to belt slippage. From this I conclude that 80 winds per coil is too much for a 24V system, in that the amount of force needed to turn the generator is too high. Luckily stator 5 (already made) has thicker wire and 64 winds per coil, so I will bech test that when it is assembled.

It should allow higher rotational speeds, and so more power output. I haven't devised a way of measuring the rpm with the turbine flying (maybe meauring the frequency of the DC ripple would be a way?), but since it can easily produce 40 Volts open circuit, it must be able to spin faster than about 350rpm. I would take a guess it could reach 500rpm.

Theoretically speaking, with a TSR (tip speed ratio) of about 5, and blade radius of 1m, a 10m/s wind would cause a tip speed of 50m/s, or 16rps which is 954rpm. That would be frightenly fast!

Star and Delta
I tried to test the generator in Delta wiring configuration, but something seemed to be wrong. There was physical resistance to turning by hand, even with open circuit. I have been told this is common wih delta wiring when the coils are not precisely matched. An alternative is to use a full bridge rectifier for each coil (which I refer to as "Delta"), so I did some testing on the bench (at 384 Rpm, open circuit)

RectifierPhase 1Phase 2Phase 3
SB356 (a)22.0 V22.0 V22.3 V
SB356 (b)22.1 V22.0 V22.3 V
KBPC3510 (a)20.9 V20.9 V21.1 V
KBPC3510 (b)20.9 V21.0 V21.2 V
None23.1 VAC23.3 VAC23.5 VAC

So it seems clear that an extra winding or two has crept onto phase 3 somewhere.
Also, the different types of rectifiers have a 1.1 volt difference! Looking up the datasheets I see the "Max forward voltage drop" is 1.1V for the SB356 and 1.2V for the BPC3510.

I only had two of each, so I used two SB356 and one of the other, that is, one full bridge rectifier per phase.
"Delta" wiring: Load usedAmpsVoltsWattsRpmV/RpmW/Rpm
None
 
40.3
 
384
0.1049 
2 x 50W bulbs
4.51
25.8
116.4
342
0.07540.3395
3 x 50W bulbs
3.80
28.1
106.8
354
0.07940.3016
4 x 50W bulbs
3.32
29.3
97.3
354
0.08280.2748
2 x 2 50W bulbs (parallel)
6.92
15.9
110.0
281
0.05660.3916
2 x 2 50W bulbs (parallel)
7.45
18.4
137.1
316
0.05820.4338
Flat 12V car battery
7.38
15.0
110.7
285
0.05260.3884

So looking at the Watts, at this region of Rpm the "Delta" wiring is defintely not a good idea.

Here is a graph summarising the results:


Back to windpower