How does changing a transformer's impedance during ordering change its construction?

If
impedance was not a concern (totally disregarded) during design and manufacturing, what would it
typically come out to be?
40/50/60MVA units are what I have in mind but really this applies to any power transformer.

Fewer turns will drop both the resistance and the inductive reactance. This will probably require a
greater cross section of core.
That will take away some of the saving in resistance.

The impedance is basically the leakage reactance - flux that does not link both windings.
Transformer designers should weigh in, but it can be modified to some degree by winding
techniques, wire type, and spacing. I think the BIL is a big factor in the impedance. The
higher the BIL, the greater the leakage reactance. ANSI C57 has standard impedances based
on the voltage level of the primary (high side).

DPC is right on.

Core shape (# of legs or shell form) and core steel material also impact leakage reactance. The order
of windings also impacts leakage. For example transformer A would likely have lower leakage
impedance than B:
A)core, LV LTC winding, LV winding, HV winding, HV DETC winding
B)core, LV winding, LV LTC winding, HV DETC winding, HV winding
For 3 winding transformers, the impedance to the tertiary is highly customizable based on the order
of the winding.
When I asked a transformer manufacture which factors in my transformer specification could be
relaxed to save cost, they seem to regard adjusting the impedance up or down by a couple of percent
as a negligable-cost design contraint.

Can I ask about a CWI ballast or neon transformer being put into perspective in terms of leakage
reactance? To my understanding these can be operated short circuited, sometimes indefinitely.

DETC winding? Sorry lol

The transformer impedance (for large ratings this is almost same as % reactance, % resistance being
very small) of a transformer depends on the number of turns in winding(vary as square of number of
turns)and physical size of HV & LV windings(width and height) and the insulation radial gap between
windings (depends on HV voltage ie BIL) During the first decades of transformer production in 19th
century, engineers were trying to achieve as low a impedance as possible to minimize the voltage
regulation. Later engineers understood that impedance has a beneficial effect- it gives protection
during line short circuits( by reducing the short circuit current flowing through windings -it
is(100/%impedance ) times the rated current of transformer).
For a particular MVA and HV voltage there is an optimum % impedance where the cost of
transformer is minimum. When we increase or decrease from this value cost increases. But there is a
minimum value beyond which it is extremely difficult to achieve lower values. When impedance goes
up, transformer becomes a copper machine ie turns are increased, copper weight goes up, core
weight comes down, copper loss goes up, core loss comes down-over all transformer weight comes
down, height goes up. With lower impedance, transformer becomes core machine ie turns are less,
core weight more, copper less, core loss more, copper loss less -, overall height reduced but becomes
heavier. For easy memory -high impedance a slender, tall lady,low impedance fat, heavy and short
lady.
Usually impedance varies between 2-5 % for distribution transformers and 6-25 % for Power
Transformers. There are exceptions- for power transformers used in Power Lab (for short circuit
testing)impedance will be low 1-2 % at 500-800 MVA , 30-60 % for small transformers used with
power electronics ( traction transformers used inside rail coaches)
When transformer winding is provided with taps, physical radial position of regulating winding with
respect to core and other windings is done such a way to get constant % impedance at all taps,
increasing % impedance with increased tap numbers or decreasing % impedance with increasing tap
numbers. This can become critical when parallel operation with existing transformers is required.
Buyer should inform manufacturer the pattern of impedance variation of existing transformer with
tap changing.
Below is an illustration showing some of the impacts changing the Impedance in power transformers

prc & Cucky: Exactly what I was looking for!

Lastly, how does resistance and reactance change as impedance changes? Is voltage regulation
linear? I'd assume more copper means higher R and the core being smaller leads to less X but I'm
sure there is way more too it.

Impedance (pu)= square root of (resistance squared +reactance squared ) pu resistance = copper
loss(kW)/Transformer rating (kVA);reactance is calculated based on parameters mentioned earlier.
Voltage regulation = IR cos phi + IX sin phi (pu) Cos phi=power factor of load.

Very interesting thread. Thanks guys.

prc - IIRR, higher impedance means longer limb and hence higher transformer. Is this correct?
Thanks everyone! This thread has been beyond words a HUGE help
@Cuky or anyone for that matter: is flux leakage that much of an issue? How much does it add to the
price to mitigate overheating? And what is done exactly to mitigate it?

Impedance is the result not the cause. When either the resistance or the reactance change then the
impedance changes.

I have seen more than one refinery with series reactances to limit fault current on low impedance
transformers. Smooth move on that one.
I have also seen a plant that had breakers to short out series reactances when the plant wasn't being
fed from two sources. That one is a little bit more forgivable but load break switches could have been
used instead of full sized breakers.

Hi MBrook,
Q1) is flux leakage that much of an issue?
a-In most cases this leakage flux is not an issue.
b-This flux links the primary and secondary winding and is responsible for the self-reactance X1 & X2.
This will create a voltage drop in both windings that limit the short circuit effect.
c-The IEEE Std C57.12.10 and the IEC Std 60076-5 provide guidance to determine standard
impedance. (See enclosed Impedances Graphs)
Outside of the standard impedance range, the cost of the transformer most likely will increase.
Q.2) How much does it add to the price to mitigate overheating?
a) The flux that is not linked to the winding will do with the core, tank or any other magnetic steel
part inducing emf and circulating current (Eddy current) that create losses in form of heat.
b) Stray Loss inside and outside the winding is another source of heat losses
c) This heat can be localized creating undesirable hotspots
Q.3) What is done exactly to mitigate it?:
a) Transformer manufacturer optimized design sophisticated finite element analysis computer
software to simulate losses and guaranty data.
b) Transformer manufacturer manages the price risk of the volatility of copper/aluminum and steel at
the time of procurement. Any uncertainty will add contingency cost.
c) Use high permeability-grade, domain-refined “H” grade, cold-rolled grain-oriented silicon steel
(“M” grade steel is used in some applications). Annealing all core steel after slitting provides optimal
loss performance.
d) During design and manufacturing some of the strategy used is:
Reduce magwire conductor size with suitable transpositions, Use of CTC (Continuously Transposed
Cable), for Tie plates/Ebars use segment steel or use stainless., Tank & Clamp - magnetic core pack
shielding, use high current busbar/bushings – shield magnetic structures w/ aluminum or copper
shielding or use stainless steel.
One place that I worked at had issues with distribution transformer shell heating on wye-wye
transformers with loss of phase. You get zero sequence flux that passes through the shell, the outside
of the transformer, and it will heat it up until it fails. One of the newbie engineers from an ivory
league school with an average engineering program thought it worthwhile to model the transformer
to try to determine how much time the operator had to isolate the transformer before it was
damaged. I admired his optimism to solve the problem but groaned over his underestimation of the
difficulty of the problem and how much time it would take for someone to do it right. The only way
that it made sense to me is that he had a supervisor that knew he was bored and just wanted him
out of his hair and let him go out on a snipe hunt.

HH
Are you sure that was not an overload issue?
Was this a three phase transformer with a three legged core? Phantom delta?
I ran a community of about 5000 customers on two phases with wye:wye connections on single
phase transformers for a long time with no issues.
We lost one phase of an undersea cable and we were on a paupers budget.
As I remember it was over a year until a surplus undersea cable within our budget was located.
Actually we did eventually service a generator rotor (preventative, not a failure) due to overheating
from the unbalanced loading.

Yes, it was a three leg core. The shell (not the core) creates a one turn delta tertiary. What was
happening, was that a phase would be lost on the distribution system because the field techs would
just use whatever on hand to repair the distribution lines. Some of the bolts and whatnot were not
the same metal as the conductor and galvanic corrosion would rust up and form bad connections
that would be worked loose by constant heating and cooling until the line fell. This was a systemic
problem that was all over the system and could not be fixed easily. When this happened, all the
transformers on a string would be damaged. Millions of dollars in distribution transformers. No one
fixed the problem even though it was well known because no one would have seen the problem
more than once in their entire career. The utility only lost a whole string of transformers every 20-25
years so it was not a problem that was begging to be fixed even if it was costly. The reason that a
solution was being pushed for was that it was getting to be so bad that if some of the longer strings
failed, they wouldn't be able to ready replace all those transformers. Some of the longer strings I
think had like 20+ transformers. The solution that was come to was to put communication devices
out on the strings so that lose of phase could be monitored. I proposed using IR cameras to try to
find all the connections in the system that were weak to try to be proactive on fixing those lines
before they fell and they went with doing that too. I was really hoping that they could just pull in all
the customer smart meter data to determine if a a phase had been loss but they didn't have smart
meters everywhere in their system yet. I too thought they might have been able to just listen to
customer complaint calls as well because there had to be a bunch of customers on these long strings
that would noticed when their 3 phase equipment (likely air conditioners) just quit working. The
window the manufacturer gave us for how long it took our transformers to fail was around 2-3 hours.
One way manufacturers deal with leakage flux is by attaching layers of core steel type material to the
tank wall. During a heat run test, the extent of the sheilding is quite evident as seen in the attached
thermal image. The sheilding begins about 0.6 meters in from the sides and bottom of the
transformer. At the bottom left corner without shielding, the tank wall was about 15 C hotter that
areas without stray flux.

HH Thanks for the info.

I was faced with wye:delta connections. In that case the delta would back feed the missing phase.
Combine that with fuse protected distribution circuits, a work outage every Sunday and single phase
switching (one fuse at a time to restore power) and the problems were endless. I suspect that you
may have had back-feed overload issues as well as leakage flux. We may never know exactly.
The condition is described in the three leg section of this link. This isn't a nackfeed issue. Three
legged cores always pass flux through the tank walls or oil. It just usually is not enough to matter.
https://generalpac.com/transformers/open-phase-con...
To me, what they should have done was specify a 4 leg core so that there would be a path for the
zero sequence flux and it wouldn't path through the tank. Why they didn't start making 4 leg cores
the company standard after the first incident is beyond me. Utilities get paid here on capital
investment. It would not have been hard to get the cost of an extra leg included in the capital
investment.

Edison123 said:
higher impedance means longer limb and hence higher transformer. Is this correct?
In my opinion it is not.
The reactance of two windings [primary and secondary] transformer it is[approximate]:
X=2πfμoN^2/Lc*Lmt((d1+d2)/3+s) where d1 and d2 are the thickness of the primary respective
secondary bobbin , s it is the space distance, N it is primary windings number of turns, Lmt average
length of a turn and Lc bobbin height [or length].

7anoter4 - That statement was from me. Edison only wanted to reconfirm. You are right and I am
sorry I misled. It is the reverse ie higher impedance shorter limb and lower impedance higher limb.
But frankly this will happen only for substantial change in impedance as impedance variation is
always achieved by changing the number of turns. For large transformers limb height is fixed from
transport limit considerations. Thanks for correcting me.

HH- I think this problem of exorbitant tank heating will occur only with Ynyn connected distribution
transformers and not with Dyn (universally used in India and we never heard of such a problem )or
Ynyn units with a stabilizing winding. There is an excellent tutorial by Westinghouse engineers of
1978- Distribution Transformer Application considerations presented at Power System District
engineers conference. I could not locate the link and don't know whether it can be posted here. Parts
of it were later incorporated in C57.105 standard.
Them my query is- in India all sub transmission units ( 220~66 / 11 kV ) units are Ynyn without
stabilizing tertiary but with 3 phase 3 limbed core. Here also we never heard about tank heating but
sure there must be unbalanced loading ,but may be no operation on two phases. Or with solid
grounding, with low earthing factors (Xo/X1) this will not happen?
bacon4life - How this thermography was taken? From outside tank? The red patches are of shield?
Then it looks bit strange for me. Usually the shields will be cooler than near by tank.Normally
transformer designers follow two methods to mitigate the impact of leakage flux (extra load losses
from eddy currents and hot spot temperatures in tank)- magnetic shielding where packets of silicon
steel laminations fixed on tank(10-40 mm thick) covering the winding height + 1000 mm.The flux
coming from top of winding enters the shield moves down and re-enter at bottom of winding. Since
silicon steel provide a high permeability path with very low losses, leakage flux will not enter in to
steel to create hot spots. Another way of creating magnetic path is providing shields (made of silicon
steel laminations) at winding top to direct winding flux back in to core.
Second method is electro-magnetic shielding where conducting material(copper 5-6 mm thick) or
aluminum ) is used to cover tank on areas where flux impinges. Flux will create an eddy current in
copper sheet that will repel flux back in to the gap between tank and winding thereby shielding the
tank surface. This will be effective where high MMF is developed on thank eg: high current leads, LV
pockets of GSU etc.

@PRC: Why was wye-wye chosen over delta wye? Soley because of the insulation reduction? Or
other reasons.
Desperate question if anyone can answer it: How does impedance play out in a 3 winding delta-wye-
wye 230/13.8/13.8kv? I'm seeing PSE&G doing this to limit fault current while using larger trafos and
am considering the same.

Mbrooke: Frankly I don't know. From Davidbeach I learnt that in US they have many different
connections for distribution transformers. Probably most of the technology developed there and this
may be an evolutionary issue. In India we have only Dyn11 for DTs.
You must get HV-LV1,HV-LV2 and LV1-Lv2 impedances ( from rating plate or test report) so that one
can arrive at H,L1 & L2 impedances. In case you have only one or two, I can make a guesstimate
provided you give me as much details as possible- application , rating plate etc.
Impedance is first dependent on inductive reactance and resistance. Leakage flux is an additional
factor.
Inductive reactance depends in part on the square of the number of turns, however the Volts per
Hertz ratio must be observed.
Reducing the number of turns requires an adjustment to the core dimensions to adjust the
acceptable V/Hz ratio.
The adjustment of the core along with changing the number of turns may in turn affect the length
per turn of the windings.
Changing the number of turns and possibly the length per turn will affect the resistance.
Now add to this the additional factors such as leakage flux and changed winding diameters due to
changed core dimensions and to allow for cooling passages.
Remember that the inductive reactance is governed by the square of the number of turns. As
impedance is predominated by inductive reactance, the most effective way to change the impedance
is often to change the number of turns. This will generally have an effect on the core dimensions and
the winding resistance.
My old transformer book states that there is an economic "Sweet spot" for transformer impedance
that is close to standard ratings.
Significant changes in impedance, either up or down generally incur added cost. (And often added
weight)

Manufacturers tend to favour the economic sweet spot first and impedance second. Standard ratings
are probably close to the economic sweet spot.
prc- I added an approximate outline of the silicon steel tank wall sheilding. The bottom and left of the
tank is warmer as indicated by the arrows where no sheilding exists. The radiators take up the top
half of the photo so the upper half of the tank wall is not really visible. We specified a high loss
factor, so I think the extent of the wall sheilding optimized to reduce losses more than to control for
temperature. This is a wye-delta-wye autotransformer.
@PRC: I do not have any actual 3 winding units on hand. Its a concept I just digging into as a possible
means to serve more load while reusing existing equipment. 115kv/13.8/13.8kv at 100MVA is what
I'm thinking of.
30 years ago when i had a job winding transformers i saw a pattern on the % of impedance of
transformers low to high side physical locations.
If the % impedance was > 3-5 the low side winding was next to core and there was a fixed distance
air/insulation barrier to the high side of transformer.
If the % impedance was <3 (down to 1%) part of the low side was next to core, with a solid physical
insulation barrier then the high side was wound on top of the 1st low side winding. Then a physical
barrier of insulation was put on high side, then a 2nd low side winding was wound on top of
insulation barrier.

bacon4life - Now I understand. What you have taken is a thermography from out side of transformer.
I am afraid we cannot say what you have seen is any hot spots from leakage flux from winding.
Normally the shielding will cover the winding top to bottom and it should not create any hot spots
outside. It can be from say, high current leads too.
Mbrooke- There are several types of three winding transformers. The unit that you mentioned may
be of two LVs of equal rating and equal impedance to HV. This is normally done when LV breaker level
exceeds. In this arrangements the LVs will be put one above the other (axially split LVs) and the
impedance between the LVs will be double of the LV to HV values. The LVs are loosely coupled and
hence loading of one LV will not affect the terminal voltage of other LV -a requirement for such
transformers used at Power stations as start up transformer.
When LVs are of different MVA ratings(but total same as HV) they will be arranged radially and the
individual circuit impedance (not LV to HV) shall be inversely proportional to their ratings to get
loading proportional to their nominal ratings.
Controlsdude: This is done to reduce the weight even with reduced impedances. In effect, instead of
X kVA , you are putting two X/2 kVA windings in parallel . This arrangement is followed even in very
large GSUs to limit transportation weight and height.

@PRC- how do they compare? Say I have a 115/13.8kv 50MVA 10% Z transformer, how would the
fault current/impedance appear on one secondary of a 115/13.8/13.8 100MVA unit?
Mbrooke - Impedances for the 100/50-50 MVA 115/13.8-13.8 kV axially split (LV2 placed axially
above LV1 with HV axially split one above the other)) transformer will be Impedance HV-LV1
=Impedance HV-LV2 =10 %on 50 MVA base
LV1-LV2 =20 % on 50 MVA base; Impedance HV/LV1+LV2 (ie both LVs simultaneously loaded) =11.5 %
on 100 MVA base.
Why LV1-LV2 impedance 20%? When you apply a voltage (as %of rated LV voltage) to LV1,with LV2
shorted (that is what is meant by LV1-LV2 impedance) , the power transfer will be from LV1 to HV
(lower half) and then from HV upper half ( it is in parallel to lower half) to LV2.So this is equal to two
transformers in series and hence impedance doubles.

Makes sense. Much thank you- pink star approved! :)

Before you part- are 3 windings typical in India? Or rare? Are these 3 winding units or 2 winding
units? They look like 2 winding- but the diagram almost hints at 3:

The diagram shows YNyn ie two winding transformers. In India up to 100 MVA YNyn transformers are
with out stabilizing winding. Above this rating, stabilizing delta tertiary is normal. In a way, you can
term them as three winding units. Multi secondary transformers are used - In thermal power stations
as start up transformers 6.6-6.6 kV or 11-11 kV. These are 50-100 MVA units. Another application is
large steel mills where there are many large motors at 6.6 or 11 kV. So multi secondary transformers
of 50-200 MVA with different voltages for secondary windings are common. But in transmission
substations multi secondary units are not preferred though was common early days. Many units used
to fail due to inappropriate application issues -like wrong selection of impedance pairs resulting in
heavy fault currents in one secondary, unbalanced load sharing etc.
Recently thousands of three winding transformers (2-12 MVA) are used here in solar farms where
two primaries accept power from two inverters stepping up to one HV 11 kV winding. Such large
GSUs are used in hydro-power projects to reduce number of transformers esp in South America. In
India we avoid such application.
A good tutorial on Multi-winding transformers is Chapter 5 of Transformer Engineering by L F Blume
(1951). This is based on a famous AIEE paper of 1924 -Theory of three winding transformers by
A.Boyajin,a GE,Pittsfield veteran.

I missed another recent application of multi winding GSU. In combined cycle thermal power stations,
normally there will be two GSUs one taking power from gas turbine generator and a smaller one
from steam generator. To reduce footprint at yard, it is possible to combine these in to one with two
primaries of different voltages and ratings with one secondary feeding power to grid. I was involved
in making such a 570 MVA GSU some time back. There also selection of impedance pairs is very
important to avoid surprises from reactive and active power sharing.
PRC wrote:

Quote:

In combined cycle thermal power stations, normally there will be two GSUs one taking
power from gas turbine generator and a smaller one from steam generator. To reduce
footprint at yard, it is possible to combine these in to one with two primaries of different
voltages and ratings with one secondary feeding power to grid.

Hmmm . . . technically very feasible, but advisable? In stations where the number of gas
turbines is >1, it would be better IMHO to have separate main output transformers [MOTs]
for each gas turbine and the steam turbine; that way in the event of the failure of a GTMOT
the balance of the plant can still operate, and if the STMOT fails the balance of the plant can
still be run in simple cycle if required [provided the necessary by-passing etc. is readily
available]. I would think that reducing the yard footprint would have to of very high and
overriding priority before such an approach would be warranted.

For a 1 + 1 combined cycle plant the shared GSU makes sense and is quite common. In a 6 +
1 plant I’m aware of pairs of CTs share GSUs and the ST has its own GSU.

6+1? Wow! Two GTs to 1 GSU makes sense in that configuration. I'm curious, though: does each GT
still have its own HRSG?

Yep, one HRSG per CT. My understanding is that when new it was the largest combined
cycle plant in the world but it's difficult to check something like that.