Applied Energy xxx (2017) xxxxxx

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Impacts of microgrids with renewables on secondary distribution

networks q
Mustafa Alparslan Zehir a,, Alp Batman a, Mehmet Ali Sonmez a, Aytug Font a, Dimitrios Tsiamitros b,
Dimitris Stimoniaris b, Theofano Kollatou b, Mustafa Bagriyanik a, Aydogan Ozdemir a, Evangelos Dialynas c
a
Department of Electrical Engineering, Istanbul Technical University, Ayazaga Campus 34469, Istanbul, Turkey
b
Department of Electrical Engineering, Western Macedonia University of Applied Sciences, Koila Kozanis 50100, Kozani, Greece
c
Department of Electrical Engineering, National Technical University of Athens, Zografou 15780, Athens, Greece

h i g h l i g h t s

Low voltage distribution networks are severely affected from high DG penetrations.

High-resolution and stochastic supply and demand profiles provide valuable insight.

Determination of DG capacity depending on the worst-case scenarios is limiting.

The voltage increases and line losses do not reach extreme levels.

a r t i c l e i n f o a b s t r a c t

Article history: High penetration of distributed generation (DG) in secondary (low-voltage) distribution networks (SDN)
Received 25 August 2016 is one of the primary challenges of network operators. Particularly in radial LV networks, renewable
Received in revised form 7 December 2016 intermittency and bidirectional power flow may cause critical problems such as under/over voltages, line
Accepted 27 December 2016
overloads and high energy losses. System operators usually consider the worst-case scenarios for the
Available online xxxx
assessment of the impacts of new integrations and usually they prefer passive methods to mitigate the
negative effects. However, analysis of daily operation with high resolution, stochastic DG supply and res-
Keywords:
idential demand profiles can reveal both the occurrence rates and the severities of the critical events,
Distributed generation
Microgrids
while paving the way for the development of active mitigation methods. This study is focused on detailed
PV systems daily operation analysis of SDNs with renewable-based microgrids. Seasonal DG output and residential
Renewable energy demand profiles throughout a day were simulated using a virtual test bed to investigate the critical cases.
Secondary distribution networks 2016 Elsevier Ltd. All rights reserved.

1. Introduction renewables [9] and limit the penetration of DGs in generation

The concept of distributed generation grows in popularity
throughout the world [1]. There are numerous technical, financial
and environmental advantages that foster its development [2,3].
Contrarily, high penetration of renewable-based DG into existing
power systems can breed new issues. The prominent challenges
are increasing volatility in daily net load profiles (which is named
as duck curve) [46], extreme surplus supply from DGs [7] and
bidirectional power flows [8]. These stimulate the use of high
cost/low efficient peaking plants, cause periodical avoidance of
q DG impact on LV networks. The studies are focused on the analysis
The short version of the paper was presented at REM2016 on April 1921,
Maldives. This paper is a substantial extension of the short version.
Corresponding author.
E-mail address: zehirm@itu.edu.tr (M.A. Zehir).
mix [10,11].
Low voltage (LV) radial distribution networks with high R/X
ratios are more severely affected by DG penetration than the distri-
bution and transmission networks with higher voltage levels.
While bidirectional power flows in a radial network may cause
overvoltage and line overloading problems, they can also increase
the system losses. In addition to the point of connection, neighbor-
ing busses are also being affected [8]. Since distribution networks
are not designed for bidirectional power flow, impact analysis is
a vital assessment prior to new DG integrations into existing distri-
bution networks.
There are plenty of studies in the literature that concentrate on
of technical challenges [12], evaluation of hosting capacity [13],
investigation of negative impact mitigation options [14,15] and

http://dx.doi.org/10.1016/j.apenergy.2016.12.138
0306-2619/ 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
2 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx
sizing of the devices that will be used for mitigation [16]. It is sta-
ted in [8,17,18] that the most severely affected bus in a LV distri-
bution system was the furthermost one to the substation, while
the negative impact on the closest bus was very minor. Doumbia
et al. focused on the voltage variation in a small scale network
for three critical cases; namely, minimum generation-maximum
demand, maximum generation minimum demand and maximum
generation-maximum demand [19]. Conventionally, distribution
network operators consider the most threatening operating condi-
tions to scale the capacity of new DG installations. However, deter-
mination of tolerable DG capacity depending solely on the worst-
case scenarios is very constraining for the utilities [20,21]. Wind
turbines generally operate at a level less than their rated capacities,
because of variation in wind speed. The highest output from wind
turbines is generally provided in a couple of hours after midnight,
while photovoltaic (PV) panels reach their peaks during midday
just for a short period of time [22]. Therefore, peak outputs of wind
generation and photovoltaic generation rarely coincides. Moreover,
recent developments in distributed storage facilities provide more
uniform generating profiles and more effective utilization of
renewable generation. Besides, common load factors for buildings
are approximately 40% [23] which reduces the occurrence possibil-
ity of extreme (minimum generation-maximum loading)
conditions.
Most of the precautions taken according to the worst-case sce-
narios are hard to implement and costly. Reduction of line impe-
dance by reinforcing the distribution feeder or moving
connection point closer to the substation may not be applicable
due to physical and economic conditions in small scale systems
[8]. Power factor control (PFC) may not be very effective as well,
particularly in networks with low X/R ratio (compared to active
power control) [22]. Advanced control of tap changer at SDN sub-
station is another prospective solution, since it is an existing con-
trol option in many places [24]. However, it may not be effective
for local under/over voltage issues at feeders with multiple DGs
[25]. Therefore, more proactive and effective solutions are
required. Additionally, microgrids including renewable based DGs
and improved measurement and control capabilities can be bene-
fited in proactive studies. According to a comprehensive survey
study on DG and microgrids, the primary areas that the microgrids
will be deployed in the following years are public buildings, hospi-
tals, military facilities and rural areas with LV and MV distribution
networks [26]. According to the results of the same study, the key
enablers of microgrids are distribution management and energy
management systems which can play significant role in mitigation
of negative impacts of DG penetration. Thus, there is an increasing
need for detailed analysis in which daily supply/demand profiles
are considered [27]. Advanced analysis of LV networks with vari-
able daily generation and load profiles can provide valuable insight
into long term operation of the system and enable development of
active mitigation methods. In addition, stochastic and quasi-steady
state simulations can reveal magnitudes of critical events in daily
operation, contributing to studies on evaluation and sizing of solu-
tion options. The works proposed in [2830] include consideration
of daily PV and load profiles in LV distribution network operation.
However, interruptions in PV panel output due to cloudiness are
not reflected to the PV models. Cloudiness is taken into account
in some studies as in [31], without covering daily profiles. Further-
more, studies with daily profiles mostly use hourly data, neglecting
the minutely changes in PV supply and residential consumption,
which are of importance for the development of corrective power
management methods [3234]. Moreover, most of the studies
use aggregated load profiles, while individual consumption behav-
ior of each consumer can be used to simulate more realistic cases
[3536]. A study of the International Energy Agency (IEA) states
that, the level of PV penetration on a specific feeder depends on
Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
the minimum load on that feeder [37]. Relevant literature lacks
of the studies that compare the worst case analysis with high res-
olution and stochastic daily profile analysis in LV networks. Inves-
tigation of the occurrence rates and magnitudes of extreme
conditions in daily operation can foster the development and eval-
uation of active mitigation methods and sizing of solutions. What
is more, seasonal differences in PV profiles are rarely included in
the related studies, restricting possible variabilities of analysis
results. Additionally, microgrids require particular attention since
local generation and consumption are at the same point of connec-
tion to distribution feeder, with more possibility of hosting higher
penetration of renewables.
This paper is therefore devoted to detailed daily operation anal-
ysis of SDNs including renewable-based microgrids. Note that the
analysis is also valid for a SDN where renewable DG and loads
are located at the same feeder with short distances. Profiles of PV
panels in the microgrid and typical residential loads in a SDN were
simulated using a virtual test bed to monitor the changes in bus
voltage magnitudes and network losses. Steady-state voltage vari-
ation in radial SDN including microgrids or DG penetration is
explained in section 2. Section 3 describes the virtual test bed used
in the study and section 4 includes the detailed daily analysis of a
sample system. The last section is devoted to the discussion of the
results and real applications.
2. Steady-state voltage variation and line losses in distribution
networks with renewable-based microgrids
Conventionally, secondary distribution networks have been
designed for one-way power flow from substation to consumers
using a radial topology (Fig. 1).
The cross-sectional areas of the LV distribution lines are smaller
and their R/X ratios are higher when compared with MV and HV
lines [8]. Therefore, the relative voltage drops are high in these sys-
tems. The substation voltage phasor of a distribution network, V_ s ,
can be expressed as in Eq. (1).
P  jQ
V_ s V_R R jX 1
V_ 
S
Here, P and Q denote the active and the reactive powers sup-
plied by the utility, respectively. V_R represents the voltage phasor
at the receiving end. R and X are the line parameters and the super-
script denotes the conjugate of a complex quantity. Due to the
fact that the angle difference between the substation and con-
sumer voltage phasors is very small, the real part of the voltage
drop dominates the expression (2).
RP XQ
DV V_ s  V_R 2
VS
The expression can be adopted to represent the voltage relation
between each consecutive bus pair of an N bus system (Fig. 2).
Active line losses for this operational condition can be calcu-
lated using Eq. (3).
Fig. 1. Unidirectional power flow in a traditional radial distribution system with
two busses.
 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx 3

Fig. 2. Unidirectional power flow in a radial distribution system with multiple busses.

X
N
PLoss Rij  I2ij 3
i1

Iij is the magnitude of the current that flows through the line
between two neighboring busses (i and (j) in a system with mul-
tiple busses.
If a distributed generator or a microgrid with surplus generation
is established at one of the SDN busses, the microgrid connection
bus serves as a sending bus for some neighboring busses (Fig. 3).
Note that the power flow directions determine the receiving and
sending ends for the branches in the network.
The voltage relation can be written as in (4).

PN  jQ N
V_R V_S R jX 4
V_ R

PN PDG  PL 5

Q N Q DG  Q L 6
Fig. 4. The difference between the real data and fitted curve for the PV model.
In Eqs. (5) and (6) the subscripts DG and L denote distributed
generation and load, respectively. It can be stated that the load X
N

profile of a renewable-based microgrid will significantly change PLossnew Rij  I2ijnew 8

the net power exchange (PN and Q N ) in the network. Voltage drop
for reverse power flow conditions can be calculated using (7).
RP N XQ N
DV
VR
In this case, V_R represents the voltage phasor at the bus of the
microgrid. PN and Q N are the net amount of active power and reac-
tive power supplied by the renewable-based microgrid or equiva-
lently by the DG. As can be inferred from Eq. (7), the DG or
microgrid bus voltage should be greater than the substation volt-
age to supply the net amount of power over the line. The amount
of voltage difference depends on the net supplied power (P N and
Q N ), line parameters (R and X) and substation voltage (V S ).
In conventional radial distribution system, line cross-sectional
areas are generally in descending order from the substation to
the consumers assuming that the line currents will decrease along
the sending-receiving route. Such a configuration decreases the
installation costs. However, when a microgrid (or equivalently a
DG) is connected to the network, it will primarily feed the neigh-
boring loads through the gradually thinner distribution lines. This
case will increase the active power losses, as formulated in (8).
i1
P Lossnew and Iijnew are the new active power losses and branch cur-
rents following the connection of the microgrid. Depending upon
7
the network configuration and net power conditions there is a risk
of increasing the line losses.
3. Virtual microgrid test bed
The virtual test bed used in this study was previously developed
as a part of Smart Electric Energy Management and Buildings
Energy Efficiency Technologies (SEEMBEET) bilateral research pro-
ject conducted by the authors [38]. During its design and testing
stages, field data of a microgrid laboratory were used. The main
idea behind the virtual test bed development is to make research
beyond the laboratory scale. Experimental small scale systems
are limited by physical factors, such as single point of connection
to the grid, available device number, capacity and dependency on
local conditions. On the other hand, a virtual system that imitates
the operation of the real system can be connected to different parts
of a larger system, several scenarios can be implemented to

Fig. 3. Bidirectional power flow in a radial distribution system with DG.

Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
4 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx

simulate different operating conditions, and long term operation Table 1

can be inspected. This section briefly describes the virtual test R/X values for the line types of the sample LVDN.

bed and the purpose of its use in this study. Line type R/X ratio
Model development and simulations were performed using AMKA 3  35 + 50 8.35
PSCAD software [39]. At the first stage, local generation, storage AMKA 3  25 + 30 11.32
and loads were modeled individually, representing their quasi AMKA 3  16 + 25 17.69
steady-state operational behavior. A data set of PV outputs (gener- AXMK 4  50S 7.28
MCMK 3  10 + 10 20.8
ated using the existing PV array block of the software together with MCMK 3  6 + 6 34.22
the developed MPPT and inverter models) for several source input

Fig. 5. Overall structure of the virtual test bed.

AMKA 35 AXMK 50
95 m 571 m B
AMKA 35
60 m
AMKA 35 A MC 6
1 AMKA 35
89 m MC 10 AMKA 16 11 m
38 m
248 m 45 m AMKA 35
137 m
50 kVA
AMKA 35 33 m
Ulity
MC 10
20/0.4 kV
38 m

AMKA 35 AMKA 35 AMKA 25

2
85 m 52 m 58 m

Fig. 6. Sample LVDN with two feeders and 10 consumer nodes.

Power Demand (W)

12000
10000
8000
6000
4000
2000
0
00:42

05:36

16:06
00:00

01:24
02:06
02:48
03:30
04:12
04:54

06:18
07:00
07:42
08:24
09:06
09:48
10:30
11:12
11:54
12:36
13:18
14:00
14:42
15:24

16:48
17:30
18:12
18:54
19:36
20:18
21:00
21:42
22:24
23:06
23:48

Time (hour)
House 1 Demand House 2 Demand House 3 Demand House 4 Demand
House 5 Demand House 6 Demand House 7 Demand House 8 Demand
House 9 Demand House 10 Demand House 11 Demand

Fig. 7. Daily load profiles of 11 houses in spring season case.

Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx 5

(solar irradiation and temperature) values was used to model PV of the system are shown in Fig. 4. The maximum difference
generation. It covers the whole operational range of the PV system between the simulated and the actual output powers of PV gener-
and fill factor of the PV panel and is used to derive a first order Eq. ation is about 15 W (1%) at the operational limits of solar irradia-
(9). tion values; i.e. 0 and 1400 W/m2.
The loads in the microgrid laboratory were modeled by modify-
Ppv 8:542173377  0:1719887T 0:170425778Ir 9 ing the existing load block in the software. The modified model can
In Eq. (9), Ppv is the output power in W, while T is the temper- read the values from an external file and reflect the changes in sys-
ature of panels in C and Ir is solar irradiance in W/m2. The R- tem voltage and frequency to its profile. All the models were
squared fitting performance for PV generation is found as 0.9968. equipped with measurement and control devices to monitor
The test data and the fitted curve for the whole operational range supply/demand behavior.

(a) Spring
27000
Power Demand (W)

24000
21000
18000
15000
12000
9000
6000
3000
0
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
Time (hour)

(b) Summer
27000
Power Demand (W)

24000
21000
18000
15000
12000
9000
6000
3000
0
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
Time (hour)

(c) Autumn
27000
Power Demand (W)

24000
21000
18000
15000
12000
9000
6000
3000
0
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00

Time (hour)

(d) Winter
27000
Power Demand (W)

24000
21000
18000
15000
12000
9000
6000
3000
0
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00

Time (hour)

Fig. 8. Total daily load profile of 11 houses in the season of (a) spring, (b) summer, (c) autumn, (d) winter.

Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
6 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx

At the second stage of test bed development, all the blocks were 4. Case study
connected together through a main bus like in a real system and
formed the microgrid (Fig. 5). Their interconnected operation was The developed test bed was used to investigate the intercon-
already tested in a previous study [39]. More details about the nected operation of a microgrid with a LV distribution network.
development process and performance testing of the virtual test This section summarizes the details of the distribution network,
bed can be found in [39]. daily supply/demand profiles for different seasons, simulation
and the results of daily operation of the system.

(a) Spring
1200
Solar Irradiance (W/m2)

1000
800
600
400
200
0
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
Time (hour)

(b) Summer
1200
Solar Irradiance (W/m2)

1000
800
600
400
200
0
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
Time (hour)

(c) Autumn
1200
Solar Irradiance (W/m2)

1000
800
600
400
200
0
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00

Time (hour)

(d) Winter
1200
Solar Irradiance (W/m2)

1000
800
600
400
200
0
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00

Time (hour)

Fig. 9. Daily solar irradiance profiles used to represent a day of (a) spring, (b) summer, (c) autumn, (d) winter.

Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx 7

4.1. Sample SDN and daily supply/demand profiles
The microgrid was aimed to be connected to a small scale SDN
which represents a rural, an islanded or a small part of an urban
network. For this purpose, a 50 kVA, 400/230 V, 50 Hz typical
SDN, which was previously proposed and used in [4] was chosen
for the analysis (Fig. 6). Its two feeders with 10 load busses were
modeled in PSCAD program.
The network has long lines with small cross-sectional areas. The
R/X value for each line type of the sample SDN is illustrated in Table 1.
A realistic analysis for daily operational behavior requires the
adequate modeling of residential loads. It was previously stated
in several studies that the low resolution load models (above
10 min) were not sufficient to represent the steady state operation
of the system accurately and high resolution load models (1 min)
were recommended [4042]. In order to use different load profiles,

(a) Spring
250
240
Voltage(V)

230
220
210
200
190
00:00
00:45
01:30
02:15
03:00
03:45
04:30
05:15
06:00
06:45
07:30
08:15
09:00
09:45
10:30
11:15
12:00
12:45
13:30
14:15
15:00
15:45
16:30
17:15
18:00
18:45
19:30
20:15
21:00
21:45
22:30
23:15
Time (hour)

(b) Summer
250
240
Voltage (V)

230
220
210
200
190
00:00
00:45
01:30
02:15
03:00
03:45
04:30
05:15
06:00
06:45
07:30
08:15
09:00
09:45
10:30
11:15
12:00
12:45
13:30
14:15
15:00
15:45
16:30
17:15
18:00
18:45
19:30
20:15
21:00
21:45
22:30
23:15
Time (hour)

(c) Autumn
250
240
Voltage (V)

230
220
210
200
190
00:00
00:45
01:30
02:15
03:00
03:45
04:30
05:15
06:00
06:45
07:30
08:15
09:00
09:45
10:30
11:15
12:00
12:45
13:30
14:15
15:00
15:45
16:30
17:15
18:00
18:45
19:30
20:15
21:00
21:45
22:30
23:15

Time (hour)

(d) Winter
250
240
Voltage (V)

230
220
210
200
190
00:00
00:45
01:30
02:15
03:00
03:45
04:30
05:15
06:00
06:45
07:30
08:15
09:00
09:45
10:30
11:15
12:00
12:45
13:30
14:15
15:00
15:45
16:30
17:15
18:00
18:45
19:30
20:15
21:00
21:45
22:30
23:15

Time (hour)

Fig. 10. Voltage variation of the microgrid bus for the cases of (a) spring, (b) summer, (c) autumn, (d) winter.

Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
8 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx

a previously proposed, high-resolution residential load profile gen-
erator was preferred [40]. It is an open-source tool that was devel-
oped using Excel, based on the household usage statistics of UK,
meteorological data and appliance parameters. It provides the
specification of the number of people, the day of the year and
assignment of appliances in a dwelling. Following a bottom-up
approach and using stochastic programming methods, it aggre-
gates the consumption behavior of each appliance and produces
residential load profile for a day with 1 min resolution. The details
of the tool together with its performance verification can be found
in [40]. The tool can also generate daily solar irradiance data, based
on meteorological statistics, including cloudiness [43]. Daily
demand profiles of 11 houses (10 for the distribution network
busses and 1 for the microgrid) and solar irradiance that represent
a day of each season are generated using the tool. Demand profiles
for each house in spring case is provided in Fig. 7. Total demand
data and solar irradiance profiles for PV system are given in
Figs. 8 and 9 respectively.

(a) Spring
1,5
1
0,5
Power (kW)

0
00:00
00:45
01:30
02:15
03:00
03:45
04:30
05:15
06:00
06:45
07:30
08:15
09:00
09:45
10:30
11:15
12:00
12:45
13:30
14:15
15:00
15:45
16:30
17:15
18:00
18:45
19:30
20:15
21:00
21:45
22:30
23:15
-0,5
-1
-1,5
-2
-2,5
Time (hour)

(b) Summer
1,5
1
0,5
0
Power (kW)

00:00
00:45
01:30
02:15
03:00
03:45
04:30
05:15
06:00
06:45
07:30
08:15
09:00
09:45
10:30
11:15
12:00
12:45
13:30
14:15
15:00
15:45
16:30
17:15
18:00
18:45
19:30
20:15
21:00
21:45
22:30
23:15
-0,5
-1
-1,5
-2
-2,5
-3
Time (hour)

(c) Autumn
2
1,5
1
0,5
Power (kW)

0
00:00
00:45
01:30
02:15
03:00
03:45
04:30
05:15
06:00
06:45
07:30
08:15
09:00
09:45
10:30
11:15
12:00
12:45
13:30
14:15
15:00
15:45
16:30
17:15
18:00
18:45
19:30
20:15
21:00
21:45
22:30
23:15

-0,5
-1
-1,5
-2
-2,5
-3
Time (hour)

(d) Winter
1,5
1
0,5
Power (kW)

0
09:00
09:45
10:30
11:15
12:00
12:45
13:30
14:15
15:00
15:45
16:30
17:15
18:00
18:45
19:30
20:15
21:00
21:45
22:30
23:15
06:45
07:30
08:15
00:00
00:45
01:30
02:15
03:00
03:45
04:30
05:15
06:00

-0,5
-1
-1,5
-2
-2,5
Time (hour)

Fig. 11. Power exchange between the microgrid and the distribution network for the cases of (a) spring, (b) summer, (c) autumn, (d) winter.

Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx 9

4.2. Critical and daily operation analysis
At the first stage of the analysis, system operation was simu-
lated for 5 min (a time range at which transient phase following
a switching from grid-connected mode to islanded mode is sup-
posed to finish and the system switches to steady-state) for the
furthest and nearest connection of renewable-based microgrid.
The furthest connection point is the right end of 571 m long AXMK
50 line (Point B), while the nearest connection point is the right
end of 89 m long AMKA 35 line (Point A). This analysis is used to
infer the highest and the least impacts of the microgrid on the net-
work. In the case that the microgrid was connected closer to the
substation, R/X ratio between the microgrid bus and the substation
was much lower when compared with of the far point of connec-
tion. Such a case resulted in less severe negative impact on the
neighboring lines and the busses. The highest impact occured
when the microgrid is connected from the furthest bus. Two
extreme cases were simulated. The first was the maximum local

(a) Spring
5

0 03:38

08:43

13:49
00:00
00:43
01:27
02:10
02:54

04:21
05:05
05:49
06:32
07:16
08:00

09:27
10:11
10:54
11:38
12:22
13:05

14:33
15:16
16:00
16:43
17:27
18:11
18:54
19:38
20:22
21:05
21:49
22:33
23:16
-5
Power (kW)

-10

-15

-20

-25
Time (hour)

(b) Summer
5

0
03:38

08:43

13:49
00:00
00:43
01:27
02:10
02:54

04:21
05:05
05:49
06:32
07:16
08:00

09:27
10:11
10:54
11:38
12:22
13:05

14:33
15:16
16:00
16:43
17:27
18:11
18:54
19:38
20:22
21:05
21:49
22:33
23:16
-5
Power (kW)

-10

-15

-20

-25
Time (hour)

(c) Autumn
5

0
13:49

18:11
18:54
19:38
20:22
21:05
21:49
22:33
23:16
14:33
15:16
16:00
16:43
17:27
03:38

08:43
00:00
00:43
01:27
02:10
02:54

04:21
05:05
05:49
06:32
07:16
08:00

09:27
10:11
10:54
11:38
12:22
13:05

-5
Power (kW)

-10

-15

-20

-25
Time (hour)

(d) Winter
5

0
03:38

08:43

13:49
04:21
05:05
05:49
06:32
07:16
08:00

09:27
10:11
10:54
11:38
12:22
13:05

14:33
15:16
16:00
16:43
17:27
18:11
18:54
19:38
20:22
21:05
21:49
22:33
23:16
00:00
00:43
01:27
02:10
02:54

-5
Power (kW)

-10

-15

-20

-25
Time (hour)

Fig. 12. Power exchange between the distribution network and the utility for the cases of (a) spring, (b) summer, (c) autumn, (d) winter.

Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
10 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx

14%

Dierence with respect to the

operaon without MG (%)

13,0%
12%

12,1%
11,8%
10%

10,1%
9,5%
9,0%
08%

8,0%
06%
04%

4,5%
3,8%

3,5%
3,3%

3,0%
2,7%
2,7%

2,8%
02%
00%
Maximum Voltage Minimum Voltage Network Losses

Extreme Case Analysis Spring Daily Operaon Summer Daily Operaon

Autumn Daily Operaon Winter Daily Operaon

Fig. 13. Comparison of the results of the extreme case analysis with the daily operation analysis of different seasons.

100

74,3
90
80

56,7
70
Duraon (%)

49,6

49,4
44,6
60
40,1

36,2
50
40

21,6
30
9,3

20
6,5

5,2
3,3
0,7
1,1

0,7
0,6

10
0
200-210 210-220 220-230 230-240
Voltage Range (V)
Spring Summer Autumn Winter

Fig. 14. Microgrid bus voltage level duration for each simulated day.

generation-minimum demand in both the microgrid and SDN,
while the second was minimum local generation-maximum
demand in both microgrid and SDN. The resulting maximum volt-
age increase for these extreme operational cases was found to be
9%, exceeding 250 V, while the highest voltage decrease was 13%
and below 205 V. Simulations showed that, furthest point of micro-
grid connection to the SDN with maximum generation and mini-
mum demand can cause line losses up to 8%.
At the second stage of the analysis, minutely residential load
profiles and solar irradiation data that represent a day from each
season were used in daily operation simulations. The impacts of
far point of connection was investigated. The results showed that,
the microgrid bus voltage magnitude did not exceed 240 V, but it
decreased below 210 V at some critical times of operation (Fig. 10).
Power exchange between the microgrid and the distribution
network for daily operation cycles are illustrated in Fig. 11. Power
exchange between the distribution network and the utility is
shown in Fig. 12.
The increase in microgrid bus voltage did not exceed 4.5%, while
the decrease was up to 12.1%. The voltage rises were relatively less
when compared to the critical case analysis results, which was up
to 9%. This was due to short period of maximum DG output during
midday period which did not coincide with the minimum network
loading durations (late night and early evening). However, voltage
drop was still considerable. The reason was longer periods of min-
imum DG output duration which intersected with the maximum
network loading periods (evening hours). System losses were
found to be up to 3.5% more when compared to system operation
without microgrid. The results of extreme case analysis and daily
operation analysis are comparatively illustrated in Fig. 13.
The microgrid main bus voltage level durations according to 10
V ranges for each simulated day are given in Fig. 14. The results
indicate that, although the highest voltage drop is close to the
extreme case analysis, it is for short periods of time. On the other
hand, voltage rise is for longer periods; but far better from the
worst case results as magnitude.
5. Conclusion and discussions
The study has presented the impacts of a renewable-based
microgrid on a secondary distribution network. By the use of a vir-
tual test bed and a LV network model, high resolution solar irradi-
ance data considering cloudiness and residential consumption
profiles that represent a day of each season were simulated. The
severity of critical operating conditions in terms of voltage varia-
tion and line losses was investigated for daily operation. The
results of the daily analysis are compared with those of the
extreme-case analysis.
It can be inferred from the results that the voltage increases do
not reach extreme levels since the duration of maximum microgrid
supply does not coincide with the duration of minimum overall
consumption. Similarly, active power losses for daily analysis are
much smaller than those of the extreme cases. On the other hand,
similar to the critical case analysis, under voltage problems can
reach extreme levels. It is because of high coincidence of maximum
demand times with minimum generation periods. Duration

Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx
analysis of voltage values in daily operation revealed that the volt-
age drop can still reach extreme levels after the integration of PV
generation. The main improvement was shortened durations. On
the other hand, the voltage increases were found far lower than
the extreme case analysis; but with longer durations. Similarly,
the power losses were lower compared to the worst case analysis.
The analysis provides important insight into the operation of
renewable-based microgrids in weak LV networks. It can be stated
that the load factor of microgrid connection bus and neighboring
busses allows more renewable integration into the system without
causing overvoltage and significant power loss problems. Investi-
gation of the magnitudes and the durations of voltage drops and
rises can contribute to the development and sizing of mitigation
options. There are some preliminary field applications on real time
monitoring and mitigation of voltage changes in secondary distri-
bution networks and microgrids. A company in US provides real-
time voltage monitoring service at end-user connection points to
the utility operator [44]. Moreover, battery storage facilities at cus-
tomer sites can be remotely controlled by the company for
responding to events happening on the utility side. According to
the results of the experimental studies proposed in [25], batteries
can mitigate power and voltage fluctuations, influencing the net
power supplied to the network. The researchers in [24] investi-
gated tap changer operation performance with battery storage on
a hardware-in-the-loop (PHIL) test bed. Active power management
can be another mitigation option. In [45], lighting load was
dimmed to improve voltage profile of the main bus in an experi-
mental microgrid system benefiting from building automation
devices. The study proposed in [46] showed that a group of PV pan-
els can be divided into several dispatchable units through addition-
ally installed switches (rather than being switched as a whole) and
controlled to improve the main bus voltage profile of a microgrid.
The researchers in [47] tested their management approach for net-
worked microgrids and reduced the voltage and power fluctuations
in the active distribution system. The results found in this paper
can contribute to studies on sizing of devices that will be used in
real systems by revealing the critical cases that will more likely
to occur during daily operation.
The future studies will be on mitigation of negative impacts of
renewable-based microgrids on distribution networks using
microgrid energy management systems.
Acknowledgement
This research is funded as a part of Smart Electric Energy Man-
agement and Buildings Energy Efficiency Technologies-SEEMBEET
project under the framework of the Bilateral Research and Tech-
nology Cooperation Turkey-Greece 2013-2014 organized by
The Scientific and Technological Research Council of Turkey
(TUBITAK) and General Secretariat for Research and Technology
(GSRT).
References
[1] U.S. Energy Information Administration. International Energy Statistics-
Renewables, <https://www.eia.gov/cfapps/ipdbproject/IEDIndex3.cfm?tid=6&
pid=29&aid=12> [accessed 21.06.2016].
[2] Colmenar-Santos A, Reino-Rio C, Borge-Diez D, Collado-Fernndez E.
Distributed generation: a review of factors that can contribute most to
achieve a scenario of DG units embedded in the new distribution networks.
Renew Sustain Energy Rev 2016;59:113048.
[3] Stadler M, Cardoso G, Mashayekh S, Forget T, DeForest N, Agarwal A, et al.
Value streams in microgrids: a literature review. Appl Energy 2016;162:9809.
[4] California ISO. Flexible Resources Help Renewables - Fast Facts, <http://
www.caiso.com/Documents/FlexibleResourcesHelpRenewables_FastFacts.
pdf> [accessed 13.07.16].
[5] Bomann T, Staffell I. The shape of future electricity demand: exploring load
curves in 2050s Germany and Britain. Energy 2015;90:131733.
Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
11
[6] Janko SA, Arnold MR, Johnson NG. Implications of high-penetration renewables
for ratepayers and utilities in the residential solar photovoltaic (PV) market.
Appl Energy 2016;180:3751.
[7] Moura PS, Almeida AT. The role of demand-side management in the grid
integration of wind power. Appl Energy 2010;87:25818.
[8] Mahmud MA, Hossain MJ, Pota HR. Voltage variation of distribution
networks with distributed generation: worst case scenario. IEEE Syst J
2014;8:1096103.
[9] McKenna E, Grnewald P, Thomson M. Going with the wind: temporal
characteristics of potential wind curtailment in Ireland in 2020 and
opportunities for demand response. IET Renew Power Gener 2015;9:6677.
[10] Ackermann T, Carlini EM, Ernst B, Groome F, Orths A, OSullivan J, et al.
Integrating variable renewables in Europe. IEEE Power Energy Mag
2015;13:6777.
[11] Stadler I. Power grid balancing of energy systems with high renewable energy
penetration by demand response. Utilities Policy 2008;16:908.
[12] Karimi M, Mokhlis H, Naidu K, Uddin S, Bakar AHA. Photovoltaic penetration
issues and impacts in distribution network a review. Renew Sustain Energy
Rev 2016;53:594605.
[13] Gaunt CT, Namanya E, Herman R. Voltage modelling of LV feeders with
dispersed generation: limits of penetration of randomly connected
photovoltaic generation. Electr Power Syst Res 2017;143:16.
[14] Shivashankar S, Mekhilef S, Mokhlis H, Karimi M. Mitigating methods of power
fluctuation of photovoltaic (PV) sources- a review. Renew Sustain Energy Rev
2016;59:117084.
[15] Haque MM, Wolfs P. A review of high PV penetrations in LV distribution
networks: present status, impacts and mitigation measures. Renew Sustain
Energy Rev 2016;62:1195208.
[16] Bucciarelli M, Giannitrapani A, Paoletti S, Vicino A, Zarilli D. Energy storage
sizing for voltage control in LV networks under certainty of PV generation. In:
IEEE 2nd international forum on research and technologies for society and
industry. p. 16.
[17] Hoke A, Butler R, Hambrick J, Kroposki B. Steady-state analysis of maximum
photovoltaic penetration levels on typical distribution feeders. IEEE Transact
Sustain Energy 2013;4(2):3507.
[18] Kabir MN, Mishra Y, Ledwich G, Dong ZY, Wong KP. Coordinated control of
grid-connected photovoltaic reactive power and battery energy storage
systems to improve the voltage profile of a residential distribution feeder.
IEEE Trans Industr Inf 2014;10(2):96777.
[19] Doumbia ML, Agbossou K. Voltage variation analysis in interconnected
electrical network-Distributed generation. In: IEEE Canada electrical power
conference. p. 52530.
[20] Scott NC, Atkinson DJ, Morrell JE. Use of load control to regulate voltage on
distribution networks with embedded generation. IEEE Trans Power Syst
2002;17:5105.
[21] Hatziargyriou ND, Karakatsanis TS, Papadopoulos M. Probabilistic load flow in
distribution systems containing dispersed wind power generation. IEEE Trans
Power Syst 1993;8:15965.
[22] Wong J, Lim YS, Tang JH, Morris E. Grid-connected photovoltaic system in
Malaysia: a review on voltage issues. Renew Sustain Energy Rev
2014;29:53545.
[23] Peacock AD, Newborough AD. Controlling micro-CHP systems to modulate
electrical load profiles. Energy 2007;32:1093103.
[24] Liu X, Aichhorn A, Liu L, Li H. Coordinated control of a distributed energy
storage system with tap changer transformers for voltage rise mitigation
under high photovoltaic penetration. IEEE Transact Smart Grid 2012;3
(2):897906.
[25] Hara R, Kita H, Tanabe T, Sugihara H, Kuwayama A, Miwa S. Testing the
technologies. IEEE Power Energy Mag 2009;7(3):7785.
[26] Zpryme and IEEE. Power Systems of the Future: The Case for Energy Storage,
Distributed Generation, and Microgrid, <https://building-microgrid.lbl.gov/
sites/all/files/the_case_for_energy_storage_distributed.pdf> [accessed
09.05.16].
[27] Gozel T, Navarro A, Ochoa LN. Deliverable 3.5 Creation of aggregated profiles
with and without new loads and DER based on monitored data, <https://www.
escholar.manchester.ac.uk/api/datastream?publicationPid=uk-ac-man-scw:
244636&datastreamId=full-text.pdf> [accessed 18.06.16].
[28] Elam MJE, Muttaqi KM, Sutanto D. A comprehensive assessment tool for solar
PV impacts on low voltage three phase distribution networks. In: 2nd
International conference on the developments in renewable energy
technology. p. 15.
[29] Baran ME, Hooshyar H, Shen Z, Huang A. Accomodating high PV penetration on
distribution feeders. IEEE Transact Smart Grid 2012;3(2):103946.
[30] Solanki SK, Ramachandran V, Solanki J. Steady state analysis of high
penetration PV on distribution feeder. In: IEEE PES transmission and
distribution conference and exposition. p. 16.
[31] Schoene J, Zheglov V, Houseman D, Smith JC, Ellis A. Photovoltaics in
distribution systems- integration issues and simulation challenges. In: IEEE
power and energy society general meeting. p. 15.
[32] Kabir MN, Mishra Y, Ledwich G, Xu Z, Bansal RC. Improving voltage profile of
residential distribution systems using rooftop PVs and battery energy storage
systems. Appl Energy 2014;134:290300.
[33] Mandelli S, Brivio C, Colombo E, Merlo M. Effect of load profile uncertainty on
the optimum sizing of off-grid PV systems for rural electrification. Sustain
Energy Technol Assess 2016;18:347.
12 M.A. Zehir et al. / Applied Energy xxx (2017) xxxxxx
[34] Hung DQ, Mithulananthan N, Bansal RC. Integration of PV and BES units in
commercial distribution systems considering energy loss and voltage stability.
Appl Energy 2014;113:116270.
[35] Yan R, Saha TK. Investigation of voltage stability for residential customers due
to high photovoltaic penetrations. IEEE Trans Power Syst 2012;27(2):65162.
[36] Stenzel P, Linssen J, Fleer J. Impact of different load profiles on cost optimal
system designs for battery supported PV systems. Energy Procedia
2015;75:18628.
[37] Povlsen A. International Energy Agency report, IEA PVPS T5-10, <http://www.
hme.ca/gridconnect/IEA_PVPS_Task_5-10_Impacts_of_PV_Power_Penetration.
pdf> [accessed 01.12.16].
[38] Istanbul Technical University (ITU), Western Macedonia University of Applied
Sciences (TEIWM). Smart Electric Energy Management and Buildings Energy
Efficiency Technologies (SEEMBEET), <http://www.edalab.itu.edu.tr/
seembeet> [accessed 27.06.16].
[39] Zehir MA, Batman A, Sonmez MA, Font A, Tsiamitros D, Stimoniaris D, et al.
Development of a field data-based virtual test bed for microgrid integration of
building automation technologies. In: 9th International conference on
electrical and electronics engineering. p. 515.
[40] Richardson I, Thomson M, Infield D, Clifford C. Domestic electricity use: a high-
resolution energy demand model. Energy Build 2010;42:187887.
Please cite this article in press as: Zehir MA et al. Impacts of microgrids with renewables on secondary distribution networks. Appl Energy (2017), http://
dx.doi.org/10.1016/j.apenergy.2016.12.138
[41] Marszal-Pomianowska A, Heiselberg P, Larsen OK. Household electricity
demand profiles- a high-resolution load model to facilitate modelling of
energy flexible buildings. Energy 2016;103:487501.
[42] Wright A, Firth S. The nature of domestic electricity-loads and effects of time
averaging on statistics and on-site generation calculations. Appl Energy
2007;84:389403.
[43] Richardson I, Thomson M. Integrated simulation of photovoltaic micro-
generation and domestic electricity demand: a one-minute resolution open-
source model. J Power Energy 2011;227:7381.
[44] Stem. Grid services with Stem, <http://www.stem.com/grid/> [accessed
30.11.16].
[45] Stimoniaris D, Kollatou T, Tsiamitros D, Zehir MA, Batman A, Bagriyanik M,
et al. Demand-side management by integrating bus communication
technologies into smart grids. Electric Power Syst Res 2016;136:25161.
[46] Stimoniaris D, Tsiamitros D, Dialynas E. Improved energy storage management
and PV-active control infrastructure and strategies for microgrids. IEEE
Transact Power Syst 2016;31(1):81320.
[47] Lv T, Ai Q. Interactive energy management of networked microgrids-based
active distribution system considering large-scale integration of renewable
energy resources. Appl Energy 2016;163:40822.