CIRED Workshop - Ljubljana, 7-8 June 2018

Paper 0288

EVALUATION OF THE IMPACT OF HIGH PRESENCE OF SMALL DERS

CONNECTED TO THE URBAN LV NETWORK

Juliana KATIC Vladimir KATIC

University of Novi Sad - Serbia University of Novi Sad - Serbia

ABSTRACT
METHODOLOGY
The subject addressed in this paper is the evaluation of the
impact of small Distributed Energy Resources (DER) LV network modelling
connected to the low voltage network in an urban area of Real urban LV network, supplied by one 10/0.4 kV
the distribution network. The aim of this paper is to show (MV/LV) substation with one transformer, was used for
the impact of the DERs by analysing key performance the simulation. Thanks to the collaboration of the local
indices of the network, with and without DER connected, distribution company, the authors were able to get:
and by analysing different DER output. To address the - The schematic and geographic scheme of the LV
problem, the authors have used advanced distribution network (AutoCAD files);
management system (ADMS) and performed analysis in - Technical data about the LV underground cables
study mode on real network model. Results obtained by - Exact number of customers supplied by each feeder
software simulations were verified with real measurements and by supply point
data from the field. - Measurement data of the max. current on the 10 kV
feeder head, in 35/10 kV supply substation,
INTRODUCTION - Measurement data of the active energy on all MV/LV
transformers of the 10 kV feeder,
Energy supply is one of the major issues of modern - Typical load curves of the consumer group.
society. Shifting from fossil fuels to renewable ones is on- The authors have also retrieved the min. of necessary
going and solar energy is expected to have leading role in
the future, anticipations are that by 2050 it will contribute technical data for MV network from which test network is
up to 35.8% to the global generation [1,2]. supplied (MV feeder and supply substation), as necessary
In USA and India there are already PV plants with over to run the load flow (LF) calculation in ADMS. The logical
100 MW of installed power and solar projects exceeding 1 scheme of the real LV network used for testing (test
GW are expected to be deployed in near future [1]. network) is presented on Fig. 1.
However, most of the deployed PVs are in small rooftop
arrays of less than 20 kW and their number is growing fast.
Most of the distribution network operators (DNO) monitor
and control the DN in real-time using some type of
SCADA and, in some cases, distribution management
system (DMS). In most cases these solutions cover models
of the medium voltage (MV) networks, while low voltage
(LV) networks are rarely modelled. This also means that
the impact of the DER connected to LV network are not
considered in full amount.
Furthermore, in most cases, utilities are still neglecting the
DER impact on network capacity in long term planning.
Great Britain is the only country in the world today that
has defined the rules for considering DER impact on
network capacity when planning its reinforcement [3].
This means that most (if not all) utilities still consider Fig. 1 LV test network scheme
traditional energy resources as the only source of supply
when planning costly network investments. The test network consists of:
This paper will focus on the evaluation of the impact of - MV/LV substation:
DER connected to the LV network in an urban area of the - 1 x 10kV normally closed disconnector
DN. It will show that high presence of modern PV systems - 1 x 10kV fuse
connected to LV network, may have significant impact on - 1 x 10/0.4 kV (630 MVA) transformer
network operation, and cannot be neglected in analysis. - 1 x 1kV normally closed disconnector
To do this, authors have modelled a test LV network, based
on a real LV network data, using one commercial solution - 8 LV feeders underground cables (PPOO 4x95 Cu):
of advanced distribution management system (ADMS) to - LV feeder 1: 1 building, 30 customers,
analyse different test cases. - LV feeder 2: 1 building, 2 entrances x 16 customers
- LV feeder 3: 1 building, 2 entrances x 16 customers
- LV feeder 4: 1 building, 3 entrances x 16 customers

Paper No 0288 Page 1 / 4

 CIRED Workshop - Ljubljana, 7-8 June 2018
Paper 0288

- LV feeder 5: supplies post office (PO), load for winter day, 0 PV production
- LV feeder 6: 1 building, 88 customers and a school, b. Time: 10:00 am, 0°C, insolation 108W/m2 ,
- LV feeder 7: 1 building, 88 customers, average power load for winter, average to max. PV
- LV feeder 8 is normally open and serves as backup production for winter, same cases as in ‘1. b, i-iv’
supply to the neighbouring MV/LV substation. c. Time: 02:00 pm, 2°C, insolation 42 W/m2, average
- LV feeders 2 and 3 as well as 6 and 7 are connected power load for winter, small to average PV for
via normally open 1kV point, winter, same cases as in ‘b, i-iv’
- To each supply point a set of normalized daily load d. Time: 08:00 pm, 0°C, No insolation, power system
diagrams (load curves) is assigned. Different load max. load, 0 PV production
curves are available for different seasons, characteristic
days and temperature ranges. RESULTS AND DISCUSSION
PV modelling Case study results
In the same area is the University of Novi Sad, which is Impact to key performance indices
supplied from the neighbouring MV/LV substation. The When PV connects to the LV grid, it supplies some amount
University of Novi Sad has a roof-top PV plant consisting of power, and the power injected (Pi) from the MV/LV
of 40 modules with total installed power of 9600 Wp [4]. transformer is diminishing. Unless the software can
The PV modules are connected to the public grid through account for the growing number of PVs connecting to the
the grid-tied inverter and the data regarding its generation LV grid – the utilities deal with discrepancy of injected and
are monitored in real time. The authors have used these consumed power data.
data to test connection of the PV in every consumer point Results of the generated power (Pg) by PVs (in kW), for
of the test network. analyzed cases are presented in Tab. 1 and Tab.2
Tab. 1 PV generation, summer day
Software for the simulation
The test network and PVs were modelled using one Pc T h Ins Pg 0 Pg 5 Pg 9 Pg A
commercial ADMS solution. Authors have used the study
198.6 26 3 0 0 0 0 0
mode to run LF calculation (and other advanced DMS
applications) for different test cases. 308.3 30 9 300 0 12.2 21.9 29.2
As no telemetry is available in the LV network, ADMS is 356.5 40 15 1020 0 35.2 63.4 84.6
using virtual load curves in order to determine the
397.2 30 21 0 0 0 0 0
consumption on supply points at any moment.
Furthermore, ADMS integrated with weather system can Tab. 2 PV generation, winter day
retrieve weather signals such as insolation and temperature
in (near) real time and estimate the production of PV in Pc T h Ins. Pg 0 Pg 5 Pg 9 Pg A
accordance with the typical PV output curve. 223.1 -4 2 0 0 0 0 0
Case study 421.4 0 10 108 0 4.7 8.4 11.2
The impact of PV on LV network operation was tested 394.5 2 14 42 0 1.8 3.3 4.4
through simulation of 20 use cases using ADMS software: 518.7 0 20 0 0 0 0 0
1. Scenario 1 - Workday, summer, 18.7.2017 - hot
summer, long days, min. or average power load and The meaning of variables are as follows:
max. PV production: - Pc – Power consumed by the LV test network (kW),
a. Time: 03:00 am, 26°C, no insolation: system min. calculated at the LV busbar (source),
load, 0 PV production - T – temperature (ºC)
b. Time: 09:00 am, 25°C, 300W/m2 insolation, - h – hour
average power load, small PV production – 4 - Insolation (W/m2)
cases were observed: - Pg – x – Pi (kW) by PVs, x PVs, A – all 12 PVs
i. no PV connected, The highest generation is, of course, detected for the max.
ii. PV connected in 5/12 conn. points insolation value – in the hot summer day.
iii. PV connected in 9/12 conn. points, Next, the impact on the total Pi (in kW), will be
iv. PV connected in 12/12 conn. points commented - Tab. 3 and Tab. 4.
c. Time: 03:00 pm, 35°C, 1020 W/m2, average Tab. 3 – Pi - summer day
power load, max. PV production, same cases as
in ‘b, i-iv’ Pc T h Ins. Pi 0 Pi 5 Pi 9 Pi A
d. 09:00 pm, 25°C, no insolation, max. power load 198.6 26 3 0 199.5 199.5 199.5 199.5
for summer day, 0 PV production 308.3 30 9 300 310.6 298.4 288.8 281.5
2. Scenario 2 - Workday, winter, 15.1.2018 - cold winter, 356.5 40 15 1020 359.5 324.5 296.5 275.5
short days, max. power load and min. PV production:
a. Time: 02:00 am, -4°C, no insolation min. power 397.2 30 21 0 401.1 401.1 401.1 401.1

Paper No 0288 Page 2 / 4

 CIRED Workshop - Ljubljana, 7-8 June 2018
Paper 0288

Tab. 4 - Pi - winter day Tab. 6 - Pl, summer day

Pc T h Ins. Pi 0 Pi 5 Pi 9 Pi A Pc T h Ins. Pl 0 Pl 5 Pl 9 Pl A
223.1 -4 2 0 224.2 224.2 224.2 224.2 198.6 26 3 0 0.9 0.9 0.9 0.9
421.4 0 10 108 425.7 421.1 417.4 414.6 308.3 30 9 300 2.2 2.1 2.1 2
394.5 2 14 42 398.3 396.5 395 394 356.5 40 15 1020 3 2.6 2.5 2.2
518.7 0 20 0 525.5 525.5 525.5 525.5 397.2 30 21 0 3.8 3.8 3.8 3.8
When the insolation is small and the consumption is high Tab. 7 - Pl, winter day
(winter), total Pi diminished by barely 1%. However,
when the insolation is high and the consumption is low Pc T h Ins. Pl 0 Pl 5 Pl 9 Pl A
(summer), the Pi can be diminished over 23%! This delta 223.1 -4 2 0 1.1 1.1 1.1 1.1
is nicely illustrated in Fig. 2 and Fig. 3.
421.4 0 10 108 4.3 4.3 4.2 4.2
394.5 2 14 42 3.8 3.7 3.7 3.7
518.7 0 20 0 6.8 6.8 6.8 6.8
When the insolation is small and the consumption is big,
Pl diminish hardly over 2.5%. However, when the
insolation is high and the consumption is low, the
reduction of Pl is over 26%! This impact of PVs is
completely neglected if utilities are unable to account for
their presence. The delta is illustrated in Fig. 4.
Fig. 2 Pi, summer day

Fig. 4 Pl, summer day

Fig. 3 Pi, winter day

Finally, the change of the voltage level (% of rated voltage)
will be observed on the LV network entry point (secondary
Furthermore, the impact on the power factor (Pf), in of the MV/LV transformer) -
relative units is presented in Tab. 5.
Tab. 5 Impact to power factor – summer day Tab. 8. As the Pi diminishes, the current passing through
the grid is smaller, thus creating smaller voltage drops, and
Pc T h Ins. Pf 0 Pf 5 Pf 9 Pf A the voltage on LV entry slightly increase [5,6].
198.6 26 3 0 0.94 0.94 0.94 0.94 Tab. 8 - Voltage on the entry point - summer day
308.3 30 9 300 0.94 0.94 0.93 0.93
Pc h Ins. V V V V
356.5 40 15 1020 0.94 0.93 0.92 0.91 0PV 5PV 9PV 12PV
397.2 30 21 0 0.94 0.94 0.94 0.94 198.6 3 0 102.53 102.53 102.53 102.53
Modern PV units produce active power, with small 308.3 9 300 101.28 101.31 101.34 101.36
capacities in reactive power. With the high production of 356.5 15 1020 100.7 100.81 100.89 100.95
PVs, the injected power in LV network diminishes, and so 397.2 21 0 100.2 100.2 100.2 100.2
does the Pf on the entry point. The effects are subtle but
persistent in all cases. The changes to the voltage balance (and thus voltage
As the Pi diminishes, the current through LV feeders is drops) are more subtle (moving in range 0.02% - 0.25%),
smaller, thus creating smaller active power losses (Pl) but persistent in all cases.
[5,6]. Tab. 6 and Tab. 7 show the change in the total As PV units do not produce reactive power and consume
amount of Pl (in kW). very little of it from the network, the case study results
have shown little or no impact to the reactive power flow
in the test network, on the level of decimal values, and
therefore they will not be presented in details.

Paper No 0288 Page 3 / 4

 CIRED Workshop - Ljubljana, 7-8 June 2018
Paper 0288

Other effects results are presented in the Tab. 10. The error was in no
Reliability indices – in no use case the Pg by PVs could case greater than 10%.
supply the total demand of the LV feeders – in urban Tab. 10 PV output validation
network many customers are connected to the same supply
point creating a big demand, impossible to satisfy with Ins. Sim. Pg Meas. Pg Error
rooftop PV. Therefore, analysis have shown that PVs have 250 2.4 2.2 8.33%
no potential as a backup source of supply in case of an
400 3.5 3.5 0.00%
incident on LV laterals. However, they have proven to
have great impact on the installed capacity – which should 440 3.8 3.8 0.00%
be considered when planning network reinforcement. 580 4.7 5 6.38%
Range of change to the power flow – in no use case the Pg
700 5.9 6.1 3.39%
by a single PV could supply the demand of one supply
point. For small rooftop PV units connected in the urban 800 6.9 6.9 0.00%
area of the grid energy flow remains radial from the grid Authors concluded that the software LF and PV production
(source) to LV consumers. This can be a green light to results are eligible to draw conclusions in this paper.
install small PV units on the urban building rooftops – as
there are no disturbances to the energy flow. CONSLUSION
Fault Calculation and impact to protection settings –
Using Fault Calculation in ADMS - short circuit analysis Using software simulations, 20 use cases with different
was tested. As PV units are not synchronous machines - load and generation were analyzed to show thT impact of
the results are showing that connection of PV units has high presence of small PVs in the LV network is not
little impact on the results, regardless of the fault place, insignificant. The results have shown that Pi from the MV
type and number of PVs connected. Characteristic results network can be diminished over 23%, Pl in the LV
are summarized in the Tab. 9 network can be reduced over 26% and voltage profiles can
Tab. 9 - Fault Calculation be improved in the range of 1%. Simulation results were
validated with field measurements, showing that modern
I V - F. V - E. software can accurately analyze LV network operation.
3Ph 0 PV 4,094.35 0.00 6.27 This is especially important to utilities with high presence
of PVs in LV network, which is a growing trend in Europe.
3Ph 12PV 4,100.14 0.00 6.27 Unless they account for the effect of PVs they may:
1 Ph 0 PV 2347.14 0.00 6.13 - Produce wrong reports on network operation analysis:
1 Ph 12 PV 2,349.10 0.00 6.13 discrepancy in energy injected and energy consumed;
energy losses division and causes; voltage profiles
3 phase (3 Ph) and 1 phase to ground fault (1 Ph) are - Produce over estimations of the costly future
presented, for no PV (0 PV) and all PVs connected (12 investments in network capacity reinforcement
PV). The place of the fault was on the LV Feeder 5, during - Lack to produce sufficient data/motivation to seek
summer day. The columns show values of the current and investments in the renewable technology
voltage in the place of the fault for the affected phase – I
(A); V– F (kV) and voltage on the entry point– V– E. (kV).
REFERENCES
This also can be seen as the green light, as rooftop PVs do
not have impact on incidents on LV feeders or to the [1] REN21, "Renewables 2017 Global Status Report"
protection settings. Paris, 2017.
[2] D. Engel, "Renewables, Power and Energy Use:
Validation of case study results Forecast to 2050", DNV-GL Energy, Arnhem, 2017.
In order to validate the results, the input data for the [3] ENA: "Engineering recommendation P2/6 Security of
software simulation was compared to the field Supply", Energy Networks Association Engineering
measurements. Directorate, 2006
Firstly, to verify LF calculations of the ADMS software, [4] V.A. Katić, et all., 2015 „Realization and operation of
authors have used real measurements by Utility: max. load roof-top photovoltaic power plant at the Faculty of
(current) on the 10 kV feeder head and active energy on Technical Sciences in Novi Sad”, Technics – Electrical
each MV/LV transformer. The max. power on the MV/LV Engineering, Novi Sad, Serbia pp.90-97.3
transformer supplying test network, according to real [5] N. Zolkifri, et al., 2017, “Impacts of Residential Solar
measurements, was approx. 500kW in the winter. The Photovoltaic Systems on Voltage Unbalance and
max. power calculated by the software never exceeded the Network Losses”, IEEE Conference (TENCON),
real max. power value - in case of highest demand the Malaysia
value calculated was 518kW, matching the expectations. [6] F. Tursun, et al., 2013, “Impacts of PV Installation on
Furthermore, the university provided the PV measured the Low Voltage Residental Distribution Networks”,
data and authors compared the estimated software 4th International Conference on Power Engineering,
production with the measurements. Characteristic test Energy and Electrical Drives Istanbul, Turkey

Paper No 0288 Page 4 / 4