Harmonic cancellation in power systems

[rconnett]

I am engaged with several clients who have large numbers (in some instances several hundred)of small VFD's (typically 20 - 40 hp) of similar inexpensive 6 pulse drives installed over a number of years. These are fed from either 12.5 kV or 34.5 kV overhead and underground electrical distribution systems. I'm also designing some new distibution systems for some new industrial customers who are installing multiple 20 - 40 hp drives in a new installation for pumping apluications. They are concerned about conformance to IEEE 519 at the point of common coupling.

According to my modelling, if I serve drives or groups of 6 pulse drives with a combination of wye-wye & delta wye transformers, I will get some cancellation of the 5/7 harmonics. Likewise if I purchase some custom transformers which will give me either a 15 or 20 degree shift, I can get further cancellation at the 11/13 & 17/19th harmonics.

Have any of you tried this or seen this done on a systemwide basis? I am not a fan of resonant (LC) filters, due to the possibility of the overall system resonance point shifting over time.

Thanks!

 

[jbartos]

Suggestions marked ///\\...a 15 or 20 degree shift, I can get further cancellation at the 11/13 & 17/19th harmonics.
///20 degree shift is associated with 18-pulse (3 6-pulse blocks) front end (360deg/18pulse=20degre shift) and 15 degree shift is associated with 24-pulse (4 6-pulse blocks) front end (360deg/24pulse=15degree shift).

The system approach is interpreted as a usage of one phase-shifter, e.g. 24-pulse for all downstream ac drives.
Normally, each ac motor drive has its own phase-shifting transformer (or sometimes called rectifier transformer or isolation transformer (by ABB)). If the phase shifter is used for more ac drives with different AC-DC front end pulse count, there will be interaction of harmonics on the phase shifter output. Therefore, it is not recommended. However, there is a way to avoid the phase shifter by mitigation of harmonics at Point of Common Coupling (PCC), globally, using harmonic filters. There are various types of harmonic filters, e.g. passive harmonic filters, active harmonic filters, and hybrid harmonic filters. Also, power line conditioners may be used. These may consist of AC-DC-AC conversions thus separating the upstream power supply with the down stream power supply. However, the power line conditioner may have a phase shifter or Active Front End (AFE) on its input.\\

 

[ozmosis]

Another point to be aware of if using phase-shift transformers designed to cancel out specific harmonic components is that if the load is not equal from the (6-pulse) drives you connect on each leg of the transformer (wye-wye & Delta-wye points) then cancellation will not take place and you will end up with severe overheating of the transformer.

 

[alehman]

When multiple small sources are combined there can be considerable cancellation of harmonics due to varied phase angles of the harmonic components. e.g. the 3rd harmonic from drive A may have a different angle than the 3rd harmonic from drive B.

What has been your experience with the clients you mentioned as to harmonics? Have there been problems or issues with 519 compliance?

 

[ozmosis]

If the loads are equal and balanced then using, what we would call a 'pseudo 12-pulse drive solution, the results are effectively the same as if you would be using a traditional 12-pulse drive converter (i.e. 2 x 6-pulse rectifiers per drive). The cost of a 'pseudo system' is far lower but the VSD control is so much limited as you need to monitor and check loading. The other issue is to be aware of THVD level as this system reduces your current harmonics of the 5th/7th & 17th/19th but the THVD level remains the same. I've only done this on 3-phase drives not on 1 phase input drives so wouldn't know the effects of 3rd harmonic. I suspect this would be slightly more complex for the tranx designer, not too sure on that point.
I've never had to work towards IEEE519, only the UK's G5/4 (at the time it was G5/3) and more recently IEC61000-3-6 for MV planning but our software tools do have the facility to look at IEEE519.2. As far as compliance, well, if the calculations show we cannot get to the specific levels we design it accordingly. Not being overly familiar with IEEE519 I can't comment but the UK G5/4 is a 'planning Guidline' and not a product standard: a fact often overlooked by a lot of consultants, and so if your calculations show you are going to be out, then you change your design or look at other methods of mitigation techniques.

 

[dpc]

The short answer is that the approach you are describing can greatly reduce harmonic distortion. This phase-shifting technique has been used for years by facilties with huge rectifier loads (aluminum smelters) with good results.

I also agree with alehman that there can be cancellation effect from multiple drives on a particular system due to varying phase angles of the harmonics produced.

As far as filters are concerned, if deemed necessary, I recommend looking into some of the newer broadband filters which are becoming more popular than the traditional resonant filters.

But you may want to do some measurements prior to buying any filters to make sure the distortion is as bad as might be supposed.

 

[busbar]

rconnett, it may be worth mentioning that an arrangement of “a combination of wye-wye & delta wye transformers” will not limit harmonic content within the isolation transformers, but only ‘above’ the interconnection upstream of the respective paralleled primaries. Depending on the desired end result, It may be appropriate to specify transformer(s) with ∆-YY or Y-∆∆ banks. Secondary windings would effectively be 6ø, like those in IEEE C57.12.70-2000 figures 17/18.

 

[hvcad]

I have measured harmonic levels at a large aluminimum smelter and can confirm that phase shifting transformers are very effective in limiting harmonic voltage distortion on the HV side on the rectifier transformer. In an aluminium smelter potline, the dc currents supplied by each rectifier are almost identical (due to the accurate control achieved using transductors), which ensures that almost ideal harmonic cancellation is achieved (72 pulse systems are not uncommon in this application).
In the system described, many different rectifier types/configurations may be used, with differwent harmonic phase angles so any cancelation may be far from ideal.
Before proceeding with the purchase of any harmonic filter equipment, it would be prudent to complete an extensive on site harmonic measurement study, to assess the characteristics of all existing rectifier loads and the present levels of distortion. Based on the results of such a survey, it would be relatively straight forward to develop a frequency domain harmonic model of the power system, with which different filter designs could be used to assess there effectiveness, under all system conditions.

Dr K S Smith
Mott-MacDonald, Power Systems Division
Glasgow, Scotland.

 

[jbartos]

Suggestion: The passive harmonic filters are not convenient for a power distribution that has varying system impedance due to load additions, changes and/or removals. The active harmonic filters or hybrid harmonic filters are preferred.
When there are many small VFDs, the worst case of harmonic vectorial additions should not be ruled out even though, statistically, it is very remote because of the probable harmonic currents cancellations. There will always be some risk that a harmonic "peak" will happen.

 

[electricuwe]

The pseudo 12-pulse arrangement will work well with a large number of drives in that power range if the following conditions are met:

-Three-phase input
-voltage source converters
-nearly equal power consumed by both of the groups
-transformers suitable to withstand harmonic currents

I wouldn't recommend to use a distributed approach for achieving operation with higher pulse-number for the following reasons:

-different phase shifts introduced by the length of the lines and transformer impedances make it difficult to guarantee that the higher order harmonics will really cancel each other at the PCC
- It is very uncommon to operate a distribution system with phase shifts which are not multiples of 30°. So it will be difficult to get approbiate transformers at reasonable cost.

Dedicated rectifiers systems work quite well with pulse numbers of 18, 24, 36, or even higher but to achieve proper cancellation of harmonics in such a system it requires lots of attention on the follwing issues:
-current balancing
-proper phase shift of the transformer
-impedances balanced properly
-influence of existing harmonics in the voltage

So, judged from my experience I wouldn't recommend to rely on harmonic cancellation in a distributed system if more than pseudo 12-pulse performance is required.

 

[rconnett]

As a point of interest, the application is in two oil field applications (one existing and one new).

The existing site utilizes widely distributed (low cost) 6 pulse VFD's (with a PWM output). There are approximately 6,000 drives installed over a 500 square mile area in this field. The drives are applied to 20hp - 30 hp submersible pumps each of which which initially operate at 100% load for a few months, then drop to 50-70% load for normal operation (for a number of years). The load is very predictable. These drives are fed from an overhead 3PH 4W 34.5 kV distribution system. The wells are spaced approximately 2,500 feet/800 meters apart (every 40 acres) The loads are operated by about 80 customers on the system, some of whom own their own 34.5 kV system. Traditionally, 480 volt loads are served by pole mounted Y-Y transformers to avoid possible ferroresonance, which does not (attenuate the triplen harmonics. The Utility did not have a harmonics requirement / did not enforce it when the units were intstalled, so they are having tremendous difficulties now.

A new similar field is expected to begin development in the next 18 months, so I am trying to advise my industrial clients in their system planning - an economical, maintainable operable and fault tolerant system. In this new field, the users will likely own a large portion of their own electrical distribution system. I was going to recommend that the Y-Y & delta-Y combinations be installed as a good first step (ferroresonance risk is low on a pole top), then install supplementary 480v-480v 15degree ph shift at selected locations. Transformer capacity is typically 75 - 150 kVA because of distance limitations to the loads. As I said, my models (on SKM Powertools) show a significant reduction. The only time cancellation would not occur is if a large number of drives are off, which means a low load-I don't see a problem here. Statistically this is very remote.

The customer wants to buy the cheapest drives he can, & I may not have control over this.

What are these broadband filters and active you refer to?

 

[jbartos]

Suggestion to the previous posting: The active filters and hybrid filters have a capability of injecting harmonic currents of opposite direction; therefore, harmonics become cancelled, at least to some extent.
Visit for example:

http://www.schneider-electric.ca/

http://www/en/techpaper/html/actharctrl.htm

http://www.squared.com/us/products/...F65D385256C7D006D6717/$file/activepfcpage.htm

etc. for more info on active harmonic filters