>






Proposal for New Work Item

Joint Investigation by the Transformer Committee
and the Switchgear Committee to determine what design factors in the
development of a transformer design lead to natural frequencies greater
than (to be determined)
kHz when viewed from the high side with the low side shorted. 
How is this related to voltage rating and to the MVA size and type of
construction used to build the transformer?  We are finding more
and more applications of this type. A fault on the low side or opposite
side of a transformer may well be within the interrupting rating of
the breaker but the breaker may not properly interrupt this fault because
the transient recovery voltage frequency is too great.  We know
that the form of the recovery voltage will be a 1 cosine waveshape
with peak amplitude around twice the peak line to neutral voltage. Depending
on the rated voltage and the frequency of oscillation the recovery voltage
may be beyond the capability of most standard breakers of a given class
and at the same time the fault current level is well within the interrupting
rating of the breaker.

Discussion

There is very little hard data on this paradox other
than the Harner /Rodriquez paper IEEE Transactions PAS 1972 Sept/Oct
pages 1887-1869.  This paper looked at 117 transformer installations
of different voltage classes and MVA ratings.  Less than 50 of
these units were for voltages 46 kV and above. By the method of current
injection the authors were able to predict the recovery voltage of these
installations.  Based on this data they made assumptions on how
the recovery voltage may vary as a function of voltage rating and MVA
size.  As an industry we are trying to extend this data into a
rating structure which may or may not be valid. What we do know is that
there have been failures of breakers and damage to other equipment as
a result of these phenomena. 

At the present time neither ANSI/IEEE or IEC Standards
properly address this issue. The ANSI Standard C37.06.1 1997 provided
some trial use recommendations for tests and ratings of a class of definite
purpose breakers (this standard has now been adopted without significant
modifications or test experience).  This rating structure was created
from engineering judgment of the situation.  It may or may not
be adequate for the task.  What is sure is that the rates of rise
of recovery voltage given still do not cover some applications and the
times to peak are so fast that synthetic circuits must be used in most
development testing.

During the factory QA test program most transformers
are injected with a fast front high voltage impulse type (typically
a 1.2x50 microsecond) wave which is used to establish the internal condition
of the transformer before testing. This process is repeated after testing
to assure that the testing has not damaged the transformer.  This
test is also used as a record to determine if damage has occurred due
to a through fault later in the life of the transformer. Among the tests
done are a measurement of impedance between the various windings and
a measurement of the capacitance from the various terminals with other
terminals open or grounded.  However, this data does not provide
the natural ringing frequency to be expected during a short circuit. 
This means that the data we desire typically is not being measured or
recorded.  Although most of the required setup to make such a measurement
is being used.

Beyond this to do a complete investigation we need
help from the Transformer Committee in cataloging the data and trying
to reconcile the driving factors that lead to the very fast recovery
voltages.  Are economical alternate designs available that would
avoid this problem? Is this just a matter of informing the transformer
designer? Is this the result of typical design of certain types of transformers?
Dry type transformers as an example. Will this be true for all voltage
ratings? What problems do the transformer designers see in adding capacitance
at the terminals of the transformer?  We should be able to develop
a L*C value as a gauge equal to or less than some number (where
L is in henries and C is in farads) which defines a transformer design
that does not cause a problem.

If we can develop answers to these questions and
quantify the conditions that lead to the need for special breakers in
certain applications we would be able to inform the application engineers
when to be on alert. This may save the industry money by developing
and testing only those ratings that are required. The situation we are
in at the moment is that the breaker manufacturers will be testing breakers
for ratings that may not be the most needed values.  We lack experience
to show that the values given in C37.06.1 2000 provide adequate coverage
by voltage class for most applications.

The data that we need can be developed using any
one of several methods currently being used in the process of testing
transformers. The inductance required is the transformer leakage impedance,
which is always recorded at the time of testing now. The measurement
that we need that is not currently being done is to determine the transformer
effective capacitance. This is a distributed capacitance across the
winding, between windings and from the windings to ground. It is this
capacitance when combined with the transformer leakage inductance that
determines the frequency of the circuit breaker recovery voltage for
a fault on the secondary side of the transformer.

Commercially available measurement systems such as
FRAMIT based on frequency response analysis can be used. This system
is primarily used to detect small physical distortions in the transformer
winding. It can be used to determine the recovery voltage by lowering
the search frequency to determine the primary frequencies of oscillation.
Doble Engineering has recently developed a Sweep Frequency Analyzer
M5100 that could also be used for this purpose. In addition bench top
circuits as shown below can also be used for this purpose.

CIRCUITS USED TO DETERMINE NATURAL

FREQUENCY OF TRANSFORMER

HV

LV

SCOPE

3X 0-400 V<span class="Heading-00201--Char">Current
Injection Circuit

Signal Generator Circuit

HV

LV

Resistor value large
relative to leakage impedance

SCOPE

0÷ 100 kHz

We would like to request that the transformer manufacturers
routinely make such measurements and publish this information as a part
of the test report for each transformer that they build. It seems that
most manufacturers already have the equipment to do this test but at
this time are not making this specific measurement. If we collect such
data it will be easy for the user to identify those special cases where
additional capacitance is needed and this can be added at the time of
installation. A collection of this data will give standards writing
organizations confidence that the standards properly reflect current
practice.

In all of the above discussion we have not mentioned
the voltage distribution across, between and to ground of windings that
may be associated with switching the transformer. It is our understanding
that the Transformer Committee already has a working group addressing
problems of this type.

Mel Smith

May 8, 2001