In quick "layman's terms", assuming from your question that you are not familiar with harmonics:
Harmonics are caused by the non-linear way power is extracted by a rectifier from an AC system. It's not really pulling power out continuously from the sine wave, but rather in "gulps" near the peaks. This causes a distortion of the incoming line power sine wave. The shape of the distortion can be analyzed and broken down into a series of "harmonics" which can be thought of as superimposed frequencies existing on top of the fundamental frequency of 60 (or 50) Hz, referred to as multiples of the fundamental. So a harmonic frequency of 120Hz would be a "2nd" harmonic, because it is at 2X the fundamental, a 3rd would be 180Hz, a 5th, 300hz etc. etc. In a 3 phase system, harmonic frequencies on individual phases that are "opposite" each other in each phase will either cancel each other out or add together. Even "order" harmonics, i.e. 2nd, 4th, 6th etc. will cancel each other out. Odd orders that are multiples of 3 will add together in the neutral of a transformer and that is bad (I'll forgo further details on this for this discussion). "Non-triplen odd order harmonics", i.e. 5th, 7th, 11th etc. are also additive but represent a less significant problem for power distribution equipment.
By using multiple rectifiers and feeding them with phase shifting transformers, you "rotate" these odd-order triplen harmonics with relation to each other by the induced phase shift, i.e. 30deg, 60deg, etc. etc. So the more you can shift, the more they become either even numbers with relation to each other, or non-triplen and therefore not as destructive. taht video shows the net effect as they increase the number of phase shifts; the more the merrier!
"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376
"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376
Consider basic power supply circuits.
A half wave rectifier with a capacitor filter will produce a sort of sawtooth waveform. The voltage will rise with the sine wave and drop as the capacitor discharges until it intersects the next rising sine wave cycle. The more load, the faster the voltage will drop.
A full wave rectifier with the same capacitor will have a similar wave form, but the voltage will not drop as far before it intersects with the next rising waveform.
With a three phase rectifier the sine waves overlap. This means that the voltage can never fall to zero. The more sine waves that there are overlapping in a cycle, the smoother the rectified output.
When full wave circuits are used on three phase, there are 6 pulses per cycle. If a wye transformer is combined with a delta transformer, 12 pulses per cycle may be obtained.
Phase shifting schemes may be used to develop even more pulses per cycle.
With a 6 pulse scheme, each phase will conduct for a maximum of 360/6 = 60 degrees.
With a 12 pulse scheme, the maximum conduction is 30 degrees per cycle.
With an 18 pulse scheme, the maximum conduction is 20 degrees per cycle.
At this point we are getting pretty smooth DC and very little if any filtering is needed.
Phase shifting to cancel harmonics is another matter. Loads such as switching power supplies create a lot of harmonics. If there are a lot of such loads anticipated in a facility, partial Zig-Zag transformers may be used to cancel the harmonics.
One transformer may have a 10% section of "B" phase connected in series with "A" phase, and so on for the other phases.
Another transformer may have a 10% section of "A" phase connected in series with "B" phase to induce a shift in the opposite direction.
Other transformers may have 20% sections cross connected.
By using sets of these transformers throughout the building, much of the harmonic content may be reduced or canceled.
Hospitals are particularly sensitive. They must have backup power, usually diesel generators. A harmonic content that may be acceptable on utility power may cause serious issues with a standby generator. The impedance of the generator may be 5 or 6 times as much as the supply transformer. That means that current harmonics will develop 5 or 6 times the amplitude of voltage harmonics when on the gen-set. This may cause AVR issues, synchronizing issues, and heating issues.
Bill
--------------------
"Why not the best?"
Jimmy Carter
If you have a 6-pulse rectifier, the two strongest harmonics it will produce will be the 5th and the 7th. If you take two 6-pulse rectifiers and connect both to the larger system each through its own transformer, and the transformers are the same, you will get the 5th and 7th harmonics onto the system through the transformers.
If, on the other hand, you connect one through a transformer with no phase shift and the other through a transformer with a 30 degree phase shift you will see your 5th and 7th harmonics essentially disappear. The trick is that the 30 degree phase shift is only at 60Hz. At the 5th harmonic the phase shift is -150 degrees and for the 7th harmonic the phase shift is 210 degrees. Taking into account the affects of the 30 degree shift of the fundamental, you find the 5th and 7th from the 30 degree transformer to be each 180 degrees from those of the non-shifting transformer. Cancellation.
Feed both those 6-pulse rectifiers into the same DC system and you have a 12-pulse system and your highest harmonics are the 11th and 13th, but they are inherently smaller than the 5th and 7th would be.
There is one large regulating transformer that feeds 9 smaller transformers all of which are connected in parallel to each other. Each transformer has a specific degree shift, ranging from -12deg to +10. And about 1/2 have the primary connected in delta, the other half wye. There is a chart that describes the "net shift" too.
Are we doing all this to smooth out the sine wave on the primary side? And what if the primary is delta connected?
This is a typical design for aluminum reduction rectifier stations and other high current rectifiers. The secondaries of the rectifier transformers are usually double wye with the neutrals connected to the negative DC bus. The wye windings are connected 180 degrees out of phase. That gives a six pulse output from each transformer to the rectifiers and the positive bus. The phase to neutral output waveform for the six phases has a peak every 60 degrees. (60=360/6)
With two transformers, we want to shift one of them 30 degrees to fill in the valleys between the 60 degree peaks. That is easily done by having one transformer with a wye primary and one with a delta. Output has 12 peaks, one every 30 degrees. (30=360/12)
If we use four transformers, two with delta primaries and two with wye, we would want to shift the second pair to fill in the valleys between the twelve peaks. We need a +15 degree phase shift and a -15 degree phase shift to do that. Using a +15 phase shifting delta-extended delta transformer in front of one delta primary and a -15 in front of one wye primary will do it. Now we have a 24 pulse system with a pulse every 15 degrees. (15=360/24).
Most designs go with six, six-phase output transformers with a pulse every 10 degrees. Older designs used eight transformers with a 72 pulse system. (+/-7.5 degrees) One reason was to divide the load to keep within the capability of the available rectifiers. Besides reducing harmonics, six or 8 units also give a better (n-1) rating to operate near full load on the system with a unit down for maintenance. More units = higher switchgear and installation costs but smaller ratings, better n-1 rating and lower harmonics.
The phase shifting transformer winding can be a separate transformer or built into the rectifier transformer primary using zig-zag or extended delta configurations. The added impedance of the phase shifting transformers affects the load sharing on the rectifier output. Some designs add reactors on the non-phase shifted units to improve laod balance.
The delta primary winding of the one regulating transformer feeding the rectifier transformers will have a current waveform close to a sine wave. Filters are added to reduce harmonics.
In some plants with multiple potlines, the main regulating transformers are also shifted between lines to reduce harmonics on the utility.
(An 8 potline plant I worked at was built in WWII. Each line had 8 rectifier transformers feeding 16 mercury arc rectifiers, later replaced with diodes. One line always had terrible load current sharing when a unit was out for maintenance. One of my young engineers put a phase angle meter on the outputs and found a phase shifter had been installed backwards for 40 years.)
