A2-302

21, rue dArtois, F-75008 PARIS

CIGRE 2008

EXPERIENCE OF SPECIFYING AND USING REACTORS

John Lapworth, Doble Powertest UK

Benjamin Marshall, National Grid UK

Christoph Ploetner, Siemens Germany

Nasser Tleis, National Grid UK

SUMMARY
Reactors are used in the UK transmission system for three principal reasons, controlling short circuit
fault current levels, controlling the current in parts of the network following a failure, and voltage
control. In the past few years reactors up to the largest sizes have been procured for all these uses
including an asset replacement programme for certain shunt reactors.
Series reactors are used to control fault levels on the system in regions with high concentrations of
generation without sectionalising the network which could reduce efficiency or security where
network capacity is limited. They are also used to control post-fault power flows again increasing
system security and efficiency. A particular example of the situation where series reactors can be used
effectively is in the Thames Estuary region of England where three 400kV 2000MVA throughput
power 40% impedance on rating reactors are to be installed to allow the connection of new generation
and interconnection infrastructure. An important consideration when using series reactors is that
increasing system reactance will increase X/R ratios and DC time constants; this can have implications
for circuit breaker duty even with reduced short circuit current levels.
The main use for shunt reactors in the UK transmission network is for controlling voltage levels
associated with cable circuits under light load conditions, preventing the need to switch circuits and
thereby increasing system security. Shunt reactors are also required for effective system restoration
strategies under black start conditions. Three ratings of reactor are used 275kV 100Mvar, 400kV
200Mvar and 13kV 60Mvar. The 13kV reactors are connected to the system via a transformer delta
tertiary winding.
The impedance of series reactors must be carefully specified to take account of the reduction in
impedance experienced when the magnetic shield saturates under short circuit conditions. The
overload requirements are also of particular importance because magnetic as well as thermal aspects
need to be considered in the design.
Large reactors can pose particular difficulties for factory testing. There is a new international standard
for reactors IEC 60076-6 (formerly IEC 60289) which takes this into account, but large reactors still
require close co-operation between supplier and purchaser to ensure appropriate tests are chosen.
Some installation conditions require particular design and test requirements, for example bypassed

[email protected]

operation requires a double ended lightning impulse test, and the dielectric capability under short
circuit conditions requires consideration if it cannot be directly tested.
Many shunt reactors are either in permanent service or switched daily, and so represent some of the
most thermally and electrically stressed plant in the transmission network. Despite this, operating
experience of most shunt and series reactors has been good and similar to or better than equivalent
transformers. A particular design of 275kV shunt reactor utilising a conductive shield arrangement
proved to have a lifetime somewhat shorter than expected for equivalent transformers, these units have
now been replaced.

KEYWORDS
Reactor Series Shunt Design Specification - Testing

INTRODUCTION
National Grid owns and operates the 400kV and 275kV transmission system in England and Wales
and operates the Scottish transmission network.
The network is highly interconnected with high concentrations of load (the Greater London network
for example, supports some 20% of the total England and Wales demand with 20% of the substation
infrastructure of the network but less than 5% the circuit route length of the overall network within a
40km radius) There is a longstanding need to manage the network loads, voltage and fault levels,
whilst maximising the use of existing network infrastructure which is frequently servicing areas of
high urban density and/or subject to specific environmental conditions which make it impractical or
prohibitively costly to expand. Series and shunt reactors are a vital part of the system design strategy
used to overcome these problems.
National Grid has 5 series reactors installed on the 400kV network, with throughput ratings from
1320MVA to 2640MVA and an impedance of around 20% on rating. A further 5 series reactors are
installed on the 275kV network with a throughput rating of 750MVA and an impedance of around
20% on rating.
In order to accommodate significant additional generation within the Thames Estuary region of South
East England a further 3x 400kV 2000MVA, 40% impedance on rating series reactors are in
manufacture. These will be the first three phase banks of single phase reactors installed on the system.
This will address the challenge of integrating approximately 13% of the required generation capacity
to satisfy demand within England and Wales within a radius of only 20km.
Some 8940Mvar of shunt reactors are installed on the National Grid System. This capacity is made up
of 76, 60Mvar units installed on 13kV SGT tertiary connections, 3 60Mvar 33kV units, 16 100Mvar
275kV units and 13 200Mvar 400kV units. Many of these reactors were installed in the 1960s and the
early 1970s and are approaching the end of their design lifetime. Although some service failures have
been experienced this is mostly limited to one particular design. A condition based replacement
program is being implemented to ensure sufficient reliable reactive compensation is available into the
future.
SYSTEM REQUIREMENTS
Series Reactors
The MVA throughput rating of series reactors is coordinated with the thermal rating of the adjacent
network equipment including switchgear. In continuing to design and operate a secure and economic
transmission system, the system designer is faced with many challenges including:

Potential overstressing of switchgear during fault conditions under conditions of high generation/
demand availability
Restrictions to the running arrangements available for managing depleted network conditions
leading to insecure/ uneconomic operating conditions
Insufficient capacity to securely sectionalise the network under pre or post fault network
conditions

To take a specific situation as an example, the Thames Estuary region of England already contains a
large amount of generation and a significant amount of new generation is contracted to connect in the
area. Figure 1 shows a simplified diagram of the network in this area with the new connections
contracted to connect by 2011. This development has created the requirement for installing three new
reactors one of which is a replacement for an existing reactor. These reactors are rated at 400kV,
2000MVA continuous throughput power and have a 40% impedance on rating. The reactors will have
post-fault ratings of up to 3000MVA. The reactor sites and connections are shown in Figure 1.
/

Rayleigh Main
Coryton
GrainCCGT
(2645MW)

Britned interconnector
(+1320MW/ -1390MW)

Tilbury
Grain

Barking
West Thurrock

Kingsnorth

West Ham
Hurst

Kingsnorth

Kemsley

Rowdown
Littlebrook
London Array wind farm
(1000MW)

Figure 1 Simplified diagram of the transmission network in the Thames Estuary region of
England showing the use of 2000MVA series reactors at Grain and Kingsnorth.
Note: colours denote system development stages between 2007 and 2011.

The reactor at Grain (shown in blue in figure 1), is a replacement for a pre-existing 2000MVA reactor
having 20% impedance on rating. The rating is dictated by post-fault power flow requirements. A
further reactor (shown in red in figure 1), again sized by the power flow requirement is proposed at
Kingsnorth 400kV substation some 8km from Grain. This reactor will normally be operated on
standby and is designed to be switched in following a network fault to remove overloads of network
elements. This mode of operation results in lower active power losses compared to permanently
inserted reactors. This reactor offers further security during depleted network conditions where it can
be switched into service to provide operational running arrangements that maximise generation
availability at that time. A third reactor (shown in green in figure 1) is to be installed at Kingsnorth
400kV substation and will operate with a similar design concept to the Grain unit, that is between
substation busbar sections and sized to control fault levels. Figure 1 also shows the two quadrature
boosters to be installed at Kemsley which are a more expensive but more flexible solution to
controlling power flows.

The increased generation in the area will result in much of the transmission infrastructure operating
within 10% of its fault level capacity and some within 5% of its capacity. Series reactors are generally
efficient devices that have very high X/R ratios. At 400kV, a typical X/R value is in the order of 100
which corresponds to a DC time constant of 318.3ms. Whilst the reactance reduces, as intended, the
AC fault current component, the reactors high DC time constant may increase the time constant of the
DC fault current component to values beyond 45ms to 60ms which are the normally tested values for
400kV circuit breakers. This may necessitate reducing the AC fault current rating of the switchgear
and substation infrastructure equipment, necessitating a further reduction in fault current to ensure that
the total fault current is within the reduced rating. The system designer needs to consider the balance
between low reactor resistance that results in lower active power losses for permanently inserted
reactors but a high DC time constant and possible overstressing of equipment, and a higher reactor
resistance that results in higher losses but a lower DC time constant. The high short-circuit reactance
of the Kingsnorth and Grain reactors is chosen to ensure sufficient reduction of the AC fault current
component so that the asymmetric fault current duty is well within the corresponding rating despite the
high X/R. The reactor ratings are also chosen to be robust to future possible system requirements.
Shunt Reactors
The main use for shunt reactors on the National Grid transmission system is for controlling voltage
levels associated with cable circuits under light load conditions. However, some reactors were
installed to support system restoration strategies, in particular the need to re-energise the network
under a black-start condition following a system shutdown. The reactor Mvar rating is chosen to limit
the magnitude of the steady state voltage step change during routine switching operations.
To maintain the voltage security of the transmission system the following three principles are adopted
to decide on the system development and asset replacement need for shunt reactors:

Achieve reactive power balance on a zonal basis to minimise active and reactive power losses,
voltage drops and depletion of reactive power reserves where needed most.
The requirement for and specification of reactors to be robust to a number of potential generation
availability patterns and mindful of future network topology changes under development.
The system is designed where possible to avoid the need for switching out of circuits for voltage
control purposes thereby maximising network integrity during circuit trips.

For black-start shunt reactors which are not frequently utilised in normal system operation,
consideration is being given to extending the life of these assets and associated connecting equipment.
This is against an evolving future network topology and uncertain generation background where the
present requirements for black-start energisation may not match those required for the future network.

REACTOR SPECIFICATION AND TEST

Reactance versus current

X [pu]

0,95

0,9

0,85
0

I [pu]

Figure 2 Relationship between reactance and

When a liquid-immersed series reactor is

reduce. The extent of this reduction in impedance with load depends on the reactor design, but can be
in the region of 10 to 20%. A curve showing the relation between load and impedance for a particular
series reactor is given in figure 2. To achieve the desired maximum fault current there must be a
minimum impedance specified at the short circuit current level. This parameter cannot be directly
tested on a large unit, because the full fault level would need to be applied. Instead it is possible to
either conduct an impedance measurement on a coil out of the tank and away from any ferromagnetic
material or perform a calculation based on as-built measurements of the coil geometry. Factory tests
have shown a very close correlation between these methods. The full-load reactance is normally
measured during test and for National Grid reactors this reactance must be less than the specified
maximum. Discussion about the achievable tolerance between maximum impedance at full load and
minimum impedance at short circuit was included in the design review of the 2000MVA reactors to
ensure that the specification could be met.
The overload rating of a series reactor is of particular significance because the flux density in the
shield is proportional to load. Overheating of frame and tank could occur quite rapidly if sustained
overloads that saturate the magnetic shield are applied. National Grid normally requires a 150% short
time overload capacity, but this can be varied to match the capacity of the transmission circuit if the
reactor is connected directly in series. The capability of the reactor to sustain the resultant flux levels
is incorporated into the design and checked at design review as this is difficult or impossible to test.
If the reactor is to be used for limiting current flows under some network conditions, but is operated
with its input and output terminals connected together for significant periods of time, as was the case
for the 2000MVA reactor installed at Elstree in London, then the reactor must be designed and tested
to withstand impulses applied to both input and output terminals simultaneously. This imposes a
significant additional dielectric stress at the middle of the winding where a voltage rise may be
expected. This double-ended impulse test is included as a special test in the new issue of IEC 600766 [1].
The winding of a series reactor, in common with other series windings, such as in autotransformers
and phase shifting transformers, will be subject to the system voltage from end to end during a closeup short circuit. The voltage and duration of this short time event will depend on the system
impedance and protection settings, but it is important to make sure that the dielectric withstand of the
winding as well as its mechanical strength is
considered. Often a direct turn-to-turn
dielectric test is not possible on large units
because of the low impedance, and calculation
and testing using lightning impulses and
switching impulses has been used as a
substitute. The low impedance means that some
compromises over the achievable waveshape
for impulse testing have been required, with the
emphasis being on achieving sufficient time
above 90% of peak voltage. Switching impulse
waveshapes are difficult to achieve with the
low magnetising inductance of many reactors.
The use of resistors to ground the un-tested
side of a series reactor during impulse test
should be discussed at an early stage as a high
resistance used with a low impedance winding
results in a good waveshape but only a small
proportion of the voltage across the reactor. In
some cases two impulse tests (with and without
Figure 3: Layout of one phase of the 400kV
the resistor) have been needed to properly test
2000MVA throughput series reactor for
both the insulation to earth and the turn-to-turn
Kingsnorth
insulation.

Transport considerations and site restrictions at the Thames Estuary sites meant that both single phase
and three phase options were considered. The eventual design of single phase reactor is shown in
Figure 3, incorporating both easier transport of the individual units and the ability to be closely spaced
on-site to reduce space requirements.
A loss evaluation formula was used in the specification to allow the most economic compromise
between loss and first cost in the design. Although the X/R ratio of the reactor is an important system
design factor, specifying a particular, or minimum R and thereby constraining the design loss would
need careful consideration and discussion with the manufacturer.
Shunt reactors are used at full rating when switched in, and this means that particular attention is
required in the specification and design review of the thermal aspects to ensure a full lifetime is
achieved. Some utilities including National Grid specify lower temperature rise guarantees for shunt
reactors than for other plant because of this. In order to gain the advantages of a lower operating
temperature and therefore longer life at relatively modest additional cost, the National Grid
specification provides for the use of fans to reduce temperatures during periods of high voltage or high
ambient temperature which in many parts of the network tend to occur together. To prevent tripping
on high temperature in the event of a loss of site auxiliary supplies the normal temperature rise must
be achieved without forced cooling.
For both series and shunt reactors of 200Mvar reactive power and above (for series reactors this is the
throughput rating multiplied by the impedance) testing becomes a significant issue. On a three phase
unit it may not be realistic to achieve the test levels for noise and temperature rise on all three phases
simultaneously. Acceptable compromises have been reached using a combination of three-phase tests
at reduced level, single phase tests and DC current tests to prove the capability of the largest units. It
remains a general requirement however, that a vital qualification for the supplier chosen to build the
reactor is the ability to test it appropriately.

SERVICE EXPERIENCE
National Grid has 250 unit years of operating experience of 400kV and 275kV series reactors over
nearly 40 years with no failures. This means that the reliability of these units is at least as good as that
of equivalent transformers. Indications from oil analysis and other tests are better than would be
expected from similar transformers. Many of the normal diagnostic tests on transformers do not work
well with the single winding of a reactor, as there is no winding to winding capacitance and winding to
earth capacitance depends on design
and is often low, but frequency
response analysis does work to check
mechanical condition.

Figure 4 Insulation degradation and winding damage

For shunt reactors at 400kV National

problem. A loose bushing shield was also found on one reactor in the SF6 busbar connection to the
unit. This was detected using routine partial discharge measurements on the GIS installation and
repaired. The overall failure rate of 400kV shunt reactors to date is very similar to that of equivalent
transformers on the UK system.

[2FAL], ppm

There are some 740 unit years of shunt reactor operating experience at 275kV, the oldest reactor in
service being 40years old. The only problems experienced, (apart from a bushing failure due to
vibration which did not require reactor replacement) have been due to a particular design of reactor
utilising a conductive magnetic shield (flux reflector) above the windings. This arrangement caused a
very high magnetic field in the winding a few
discs from the top, although the designer had
10
taken some steps to prevent a high stray loss in
the end turns by changing the conductor
8
arrangement, this was not continued far
enough down the winding. The higher than
6
expected stray loss lead, after 30 to 35 years in
service to thermal degradation of the paper
4
and turn-to-turn dielectric failure in this
region. (Figure 4 shows one phase of the
2
reactor of this type from Hurst substation
during scrapping). Out of an initial population
0
of 7 units, 3 failed in service and the
1/1/91
1/1/92
31/12/92
31/12/93
31/12/94
remaining units were replaced. This
experience means that the overall failure rate Figure 5: Increase in FFA (2FAL) levels in oil
is slightly worse than for similar transformers, experienced before the failure of the 275kV
but there were no failures before 30 years in 100Mvar reactor at Hurst.
service. It is hoped that modern designs using
sophisticated finite element modelling to
identify stray loss effects would not suffer from this problem. Figure 5 shows the increase in furfural
(FFA) concentration in oil that appears to be characteristic of the long term paper overheating fault
that caused the failures. The DP of the paper from this reactor was 141 in the overheated region.
At 13kV the operational experience of shunt reactors has been very good. Most of these tertiary
connected reactors are now approaching 40 years in service, giving an accumulated service experience
of some 2600 unit years. Of the three recorded failures two were due to loose connections probably
caused by vibration and the third by a foreign body in the winding. The overall failure rate is 0.1% per
unit year. A small number of these reactors are now due for asset replacement based on oil test results,
but the future replacement programme will be largely determined by the condition found from forensic
analysis of the first reactors to be replaced. The insulation is subject to relatively low levels of
dielectric stress and this may allow a different approach to asset replacement compared to the high
voltage units.
All the shunt reactor failures experienced to date have been low energy events compared to
transformer failures, because the turn-to-turn fault current is relatively low. Fault currents generally
within the reactor are limited except towards the line end.
CONCLUSIONS
Reactors form a vital part of the system design strategy required to maintain control of power flows,
voltage levels and fault levels on National Grids highly interconnected network. In particular they
provide solutions to integrating increasingly high concentrations of generation and demand into
already crowded parts of the network. Specifying reactors for these variable requirements leads
inevitably to higher impedance and higher rated designs of reactors being required.

The system demand for higher series reactor ratings pushes the boundaries of available capacity
particularly with respect to factory tests, but close co-operation between utility and manufacturer has
overcome potential problems without compromising the need for effective tests.
Service experience of both series and shunt reactors has been good overall, apart from one particular
design of 275kV shunt reactor which has now been replaced, although even this design provided 30
years of reliable service. Reliability is as good as or better than equivalent transformers. Vibration can
be a problem particularly for connections and components.
FFA analysis provided a good guide to the aged state of the insulation in certain reactors and can be
used to detect the kind of long term insulation overheating experienced on one particular design of
reactor.

BIBLIOGRAPHY
[1]

IEC 60076-6 Reactors Ed 1: 2008