23rd International Conference on Electricity Distribution Lyon, 15-18 June 2015

Paper 0254

CLOSED-RING OPERATION OF MEDIUM VOLTAGE DISTRIBUTION GRIDS –

THEORY MEETS PRACTICE

Robert J.W. DE GROOT Johan MORREN Johannes G. SLOOTWEG

TU/e - Netherlands Enexis B.V. - Netherlands Enexis B.V. - Netherlands
[email protected] [email protected] [email protected]

as ideally the current flowing through each feeder should

ABSTRACT be roughly equal (given that each half of the distribution
Closed-ring operation of distribution grids has several ring is equal in cable length, size and connected loads).
advantages over more common distribution grid
configurations incorporating normally-open points. As HV/MV-station or
MV-substation Industrial areas
power flows are able to naturally balance out between
feeders of a ring shaped distribution grid, it is expected
1 2 3 4
that peak loading will be reduced, and with that, grid 5
losses. As such, investment costs can be postponed or
avoided and operational costs can be reduced due to MV-distribution network
lower grid losses. This paper presents the results found in 6
a field test, performed within a live distribution grid, 10 1 9 8 7
which was operated in closed-ring configuration.

INTRODUCTION
Distribution system operators (DSO) face the challenge Residential
areas
of coping with aging grid assets on one hand, while on
the other hand consumer demands increase due to Figure 1: Typical lay-out of a MV distribution grid
increasing dependency on availability of electrical
energy. The distribution of electrical energy in a safe,
reliable and affordable manner is crucial for many aspects Distribution automation
of modern society. DA systems are all systems, consisting out of an
Distribution automation (DA) can be implemented into arbitrarily number of components, which contribute
the (smart) grid in order to improve reliability, decrease somehow to the automation and remote operation of the
the need for manual labour and reduce grid losses. This distribution network. The automation of switching is
paper focuses on grid loss reduction in the medium usually performed based on measurements done within
voltage (MV) distribution grid by operating the grid in a
that same grid, in order for the system to react on
closed-ring lay-out, supported by a DA system. The
results of a field test done in a live distribution network in changing circumstances. In this way DA can be deployed
The Netherlands are presented. The goal of this test was for a number of different purposes, such as, but not
to investigate the effect of closed-ring operation with limited to, load control, remote metering, power flow
regards to grid losses. A comparison is made between optimization, protection adaptability and self-healing
theory and practice, comparing earlier found results in capabilities.
simulation with the newly found results from the field In the study case presented later in this paper, the grid has
test in the grid. been equipped with DA for its self-healing capabilities.
Whenever a fault would occur within the grid, it is
CLOSED-RING GRID OPERATION possible to either remotely or automatically perform
Typically distribution grids are structured according to switching actions, in order to quickly and effectively
the lay-out depicted in Figure 1. Medium voltage restore power to unaffected parts of the grid after a fault.
distribution rings are usually fitted with a normally-open A major factor in selecting a grid for the field test
point (NOP), splitting the ring into two separate feeders presented in this paper was the presence of a DA system
from an electrical-technical point of view. The purpose of for quick recovery after a potential fault. Closed-ring
the NOP is to ensure selectivity for protection systems operation affects reliability in a negative manner. It will
and reduce the impact of faults in the grid by limiting the straight-away double the system average interruption
number of customers that are affected when protection frequency index (SAIFI), due to the tripping of a whole
trips. However, on the downside, the NOP might distribution ring instead of just one half. Furthermore, it
indirectly incur additional grid losses, by obstructing an will increase the system average interruption duration
optimal power flow. With a power flow that shows index (SAIDI), because detecting and locating a fault will
strongly dynamic behaviour throughout the day, the take longer, which will in turn have its effect on average
optimal location for the NOP would continuously change, restoration times.

CIRED 2015 1/5

 23rd International Conference on Electricity Distribution Lyon, 15-18 June 2015

Paper 0254

DA systems will mitigate the negative effects of closed- within figures.

ring operation, by offering quick detection and To mitigate the previously discussed downsides and risks
localization of faults. Furthermore, since switching can related to closed-ring operation, the time window that
be done remotely or even fully automatic, restoration was granted for actually running the grid in such a layout
times will greatly be reduced [1]. Within a risk-based for field-testing was limited.
asset management (RBAM) strategy [2], the reduction of
grid losses, customer minutes lost (CML) and labour that
comes with installing a DA system and running in closed-
ring layout, might actually outweigh the increase of
SAIFI.
Grid loss reduction
According to [3], a grid loss reduction of up to 10% can
be achieved, given the right circumstances. The Enexis1
grid has a total yearly grid loss of 1700 GWh. Taking a
conservative 5% reduction, and given that the MV
portion is responsible for 28% of the total losses in the
Enexis distribution grid, this would lead to a total grid
loss reduction of 1700 GWh * 28% * 5% = 23.8 GWh. Figure 3: RMU in the Enexis distribution grid, equipped for DA
Significant savings can potentially be made by
introducing closed-ring grids throughout the network. MEASUREMENTS & RESULTS
Other than that, power quality and voltage profiles will
both benefit from closed-ring operation [4]. Last but not For over a week’s duration, several measurements were
least, the grid will be more resilient when it comes to taken within the grid. The sample frequency of the
incorporating renewable energy sources such as measurements was 5 minutes, at 5 measurement points
photovoltaic systems or wind power. within the ring. Power, current and voltage were
measured at these points. The grid was operated in
Study case closed-ring lay-out starting Wednesday December 10,
In order to investigate the effect on grid-losses in 2014 @ 13:54 hrs, and it was switched back into regular
practice, a grid within the Enexis distribution grid was open-ring lay-out on Thursday December 18, 2014 @
selected. This grid has a ring shaped lay-out and contains 09:27 hrs. Inside the figures throughout the rest of the
roughly 50% (light) industrial/commercial consumers and paper, these two switching moments are indicated with
50% residential consumers, it is situated in the city of vertical, red dashed lines.
Breda in the south-west of the Netherlands. Figure 2
depicts a schematic lay-out of the grid under Measurements
investigation. The upper feeder (labelled ‘1’) has almost What can be clearly seen from the measurement results is
exclusively industrial consumers connected to it, while that a considerable power flow occurs at the original
feeder 2 has solely residential consumers connected to it. location of the NOP, indicating that quite a shift in
balance occurs the moment the NOP is closed. As can be
seen from Figure 4, the power flow through the closed
1 NOP is in the range of about 50 to 250 kW.
Power flow at KOP station
WOU
300

2 250
ACH 4
WER KOP
ACH 10
Figure 2: Grid under investigation in the network of the “Breda” 200

region. Measurement points encircled in red.

150
Power [kW]

Due to the mixed nature of consumers, with very

100
different load profiles, a relatively high reduction of grid
losses can be expected, according to [3]. 50
The selected grid is already equipped with DA (Fig. 3.)
and therefore quick restoration in the unlikely case of 0
faults can be ensured. Measurements within the grid are
performed at the DA-equipped stations in order to reveal -50
01/12 03/12 05/12 07/12 09/12 11/12 13/12 15/12 17/12 19/12
the effect on grid losses. These substations are encircled Time [days]

in red in Figure 2 and each of them is labelled, these Figure 4: Power measurement at location of NOP. Red vertical lines
labels will be used throughout the rest of the paper and indicate NOP switching moments.
1
One of the larger distribution system operators in The Netherlands

CIRED 2015 2/5

 23rd International Conference on Electricity Distribution Lyon, 15-18 June 2015

Paper 0254

Network characteristics
Closing the NOP has a similar effect on the power flow at In order to fully understand the phenomena seen in the
other locations within the grid. For example Figure 5 & 6 previous section, it is important to look at some essential
indicate a similar shift in power flow for the locations network characteristics. An important factor in the
labelled as WER and WOU. redistribution of power flows after closing the NOP are
600
Power flow at WER station
the cable resistances. In Table 1, an overview is given of
the cable resistance between several stations within the
500
distribution ring.
Cable section(s) Impedance [Ω] Length [m]
ACH4  1st station 0.38 2812
400
ACH4  WOU 0.60 3886
WOU  KOP 0.39 1779
Power [kW]

300 ACH10 – 1st station 0.06 279

ACH10  WER 0.84 4077
WER  KOP 0.59 2867
200
Table 1: Network characteristics

100 It is worth noting that the first cable section originating

from measurement point ACH4 is relatively long
0
01/12 03/12 05/12 07/12 09/12 11/12 13/12 15/12 17/12 19/12
compared to the average cable lengths throughout the rest
Time [days] of the distribution ring.
Figure 5: Power measurement at location of WER station. Red
vertical lines indicate NOP switching moments. Results
The power flows seen at the beginning of both feeders are
700
Power flow at WOU station
depicted in Figure 7 & 8 below.
Power flow at ACH10 station
600 1600

500 1400

400 1200
Power [kW]

300
1000
Power [kW]

200
800
100
600
0

400
-100
01/12 03/12 05/12 07/12 09/12 11/12 13/12 15/12 17/12 19/12
Time [days]
200
01/12 03/12 05/12 07/12 09/12 11/12 13/12 15/12 17/12 19/12
Figure 6: Power measurement at location of WOU station. Red Time [days]
vertical lines indicate NOP switching moments.
Figure 7: Power measurement at location of ACH10. Red vertical
lines indicate NOP switching moments.
The feeder containing the WOU station mainly has Power flow at ACH4 station
industrial consumers connected to it, something that can 2200

be clearly seen from the measurements shown in Figure 2000

6. During the weekend days (December 6-7), the power
1800
flow through the WOU station drops to a near-steady 100
kW base load. However, once the NOP is closed, a more 1600

dynamic, and at times even reverse power flow occurs,

Power [kW]

1400
due to downstream loads being mainly fed through the
other feeder. This effect is therefore obviously opposite at 1200

the WER station. As can be seen from the measurement 1000

results in Figure 5, with a closed NOP, power flows
800
through the WER station are considerably larger. This is
the result of loads in the feeder 1 being fed through the 600

feeder 2. As the WER station is positioned relatively far 400

01/12 03/12 05/12 07/12 09/12 11/12 13/12 15/12 17/12 19/12
down the feeder (compared to the WOU station), the Time [days]
effect is relatively large. Figure 8: Power measurement at location of ACH4. Red vertical
lines indicate NOP switching moments.

CIRED 2015 3/5

 23rd International Conference on Electricity Distribution Lyon, 15-18 June 2015

Paper 0254

CONCLUSIONS
What can be seen from the measurement results at the
beginning of both feeders, is that that during closed-ring In order to have an indication of the effect of closed-ring
operation, in ACH4 the average power decreases, while in operation on grid losses, a field test was performed within
case of ACH10, the average power increases. The average the Enexis distribution grid, located in the south-west of
power flows at both locations with and without a closed the Netherlands. Due to limits on the time window in
NOP are presented in Table 2. which the ring could be operated in closed-ring layout
(closed-ring operation inhibits certain reliability risks),
Open ring Closed ring Difference the resulting available data was limited as well.
ACH10 687.5 kW 897.1 kW +30.5% Nevertheless, from the data it can be concluded that
ACH4 1225.3 kW 1101.1 kW -10.1% closed-ring operation of distribution grids has a positive
Total 1912.8 kW 1998.2 kW +4.5% effect on grid losses within that grid. This can be
Table 2: Average power flows through ACH4 & ACH10, both open- concluded from the fact that part of the transported power
and closed-ring operation.
is redistributed to the less-heavy loaded feeder. As well
To determine the difference in grid losses, the as from the fact that even though electricity consumption
measurement results are used to calculate losses within (week-on-week) has risen significantly, the increase in
the grid. In order to assign different current values to the grid losses was smaller than that. However, it was not
(not measured) loads within the grid, the yearly possible to pin-point an exact quantification of loss
maximum current levels per load are taken, and used to reduction, as it is impossible to determine what effect
calculate to the peak load ratio (Eq. 1.). This peak-load attributes to which difference in the measured and
ratio is then used to distribute the measured total loads calculated results.
among each individual load accordingly. Yearly peak- Therefore, in future research, it is desirable to have a
loading values used for calculating the peak-load ratio larger time window to perform measurements, so that
were obtained from ongoing field-measurements that are irregularities and natural differences in electricity
part of standard asset monitoring practices. consumption can be taken into account for and can be
levelled out by averaging over larger time windows.
𝑆𝑝𝑒𝑎𝑘,𝑛
𝑃𝑒𝑎𝑘 𝑙𝑜𝑎𝑑 𝑟𝑎𝑡𝑖𝑜 = ∑𝑥 (1)
𝑛=1 𝑆𝑝𝑒𝑎𝑘,𝑛 FUTURE WORK
Once the current flowing through each of the loads is There’s a strong indication that closed-ring operation
determined, and with that the current through each cable leads to reduced grid-losses. However, the promise of
section, losses are calculated by simply applying Ohm’s grid loss reductions up to 10% [3], seen in simulation
law (Eq. 2.), and summating the losses found for each environments, is still nowhere to be found in the results
cable section accordingly. of this field-test. Not only might a single case study not
be representative of what can be expected throughout a
𝑃𝑙𝑜𝑠𝑠 = 𝐼 2 𝑅 (2) network. Also, the limited availability of measurement
data (due to practical reasons) and the obfuscation of this
The losses resulting from this calculation method are data due to variables that are outside the controlled
presented in Table 3 below. environment, complicates analysis.
Future work should comprise the collection of sufficient
Day NOP Losses Difference data, so that seasonal influences, random events, and
8/12 open 182 kWh - other uncontrollable phenomena can be factored out of
15/12 closed 188 kWh + 3.2% measurement results as much as possible. In a best-case
Table 3: Comparison of losses during open- and closed-ring situation, the time window in which the grid can be
operation.
operated in closed-ring layout is significantly larger.
What’s interesting to see is the fact that during closed- Ideally a year, since that would cover at least one cycle of
ring operation, the losses actually increase. On the seasonal influences, which can then be (partly) filtered
contrary, the average load increased by 4.5% during out of the end results.
closed-ring operation. As the losses increased less than
the increase in transported energy, losses have actually REFERENCES
decreased relative to the transported energy.
This strong increase in electricity consumption can be [1] T.C.A. Castelijns, R.J.W. de Groot, J. Morren, J.G.
(partly) attributed to the fact that the festive season (e.g. Slootweg, “Using Particle Swarm Optimization for
decorative lighting) is getting near and days are getting Placement of DA in Distribution Networks”,
shorter during the period in which the field-test was Universities’ Power Engineering Conference
performed. (UPEC2014), September 2014, Cluj-Napoca,
Romania.

CIRED 2015 4/5

 23rd International Conference on Electricity Distribution Lyon, 15-18 June 2015

Paper 0254

[2] M.J.C. Berende, J.G. Slootweg, J. Kuiper, “An asset

management approach to distribution automation”,
CIRED 2009, June 2009, Prague, Czech Republic.

[3] R.J.W. de Groot, J. Morren, J.G. Slootweg,

“Investigation of grid loss reduction under closed-
ring operation of MV distribution grids”, IEEE PES
GM 2014, July 2014, Washington D.C., USA.

[4] G.C. Schoonenberg, F. van Overbeeke, C.

Spoorenberg, “Future concepts for medium voltage
distribution networks: a new philosophy”, CIRED
1999, June 1999, Nice, France.

CIRED 2015 5/5