PROTECTION OF DOMESTIC SOLAR PHOTOVOLTAIC

BASED MICROGRID
M.P Nthontho*, S.P Chowdhury*, S. Winberg*, S. Chowdhury*

*University of Cape Town,South Africa

monontsi.nthontho@uct.ac.za | sp.chowdhury@uct.ac.za | simon.winberg@uct.ac.za | sunetra.chowdhury@uct.ac.za

Keywords: Differential protection, solar PV microgrid. mode ± going from grid connected to islanded mode. Most
common disorders are transients, voltage sags and swells,
Abstract over-voltage and under-voltage as well as under-current and
over-current faults [5]. It was in response to these problems
A large-scale implementation of distributed generation (DG) that power electronic systems were employed to mitigate
for households involves each house generating its energy faults and to coordinate renewable sources and loads. This
from photovoltaic (PV) cells. This approach is particularly formed an almost self-healing network called a microgrid.
suitable in rural electrification projects where demand is Proposing an optimal technique that implements adaptive
relatively low and yet grid connection is costly. A meshed control of the power electronic systems used in the interface
microgrid is an attractive solution for energy generation and protection of microgrids is the aim of this paper.
sharing in domestic electrification projects. Solar PV systems
located on different houses can be interconnected together Adaptive control is achieved by protection schemes
into a meshed microgrid. This interconnection creates an appropriate to microgrid network constraints and
integrated system that can be treated as a single DG. While requirements supported by communication networks. This
they remain autonomous, the systems can operate as one technique leads to the self-healing feature in microgrids. It is
microgrid. This approach provides a more reliable and robust therefore the focus of this paper to review protection schemes
grid as the systems can supplement each other. However, of distribution networks integrated with microgrids.
protection and control in a meshed power network setup is a Due to the bi-directional power flow characteristic of
challenge. This is because a meshed microgrid has more microgrids, conventional protection schemes used in radial
interconnections and interfaces compared to radial architecture grids are unlikely to be applicable for meshed
architecture power grid. This is further complicated by effects microgrid [6]. Furthermore, conventional protection schemes
of DG on power quality, such as transients: voltage sags and used in radial distribution networks use load distribution,
swells, under and over-current faults amongst other direction and magnitude of fault current, and characteristic
difficulties. The challenges necessitate robust protection architecture of the network. As aforementioned, there is no
schemes supported by effective control and fault location predetermined direction of flow of energy in meshed grid-
identification facilities. This paper discusses implementation connected microgrids. A simulation performed in [6] shows
of islanded or grid-connected microgrids formed by solar that due to the jittering magnitude of fault current, traditional
systems installed in homes. The paper discusses microgrid current protection scheme becomes inapplicable in
protection, adaptive control and fault location identification. microgrids. Fault current swings frequently and unpredictably
In grid-connected mode, power flow in microgrids is bi- as the system changes from grid connected to islanded mode
directional. This characteristic rules out many of the and vice versa. Conventional inverter systems used with DGs
traditional protection schemes based on current direction. have very low fault current. In islanded mode, the fault
Thus differential current protection scheme is the focus of this current is two times less than the nominal operative value set
paper. In conclusion, this scheme overcomes many protection in conventional relays [4] [7]. This current requires very
problems including the low fault current nature of sensitive relays and may not cause a trip with conventional
conventional inverters. relays [4] [8]. Nevertheless, installing sensitive relays is not a
viable answer as these can lead to instability of the system by
1 Introduction responding to spurious currents. In addition, in grid connected
operation, even load distribution is not predetermined since
Distributed generation systems (DGs) forming a microgrid the microgrid can either be feeding only the local load or
and their integration into the utility grid is a popular topic in feeding the local load and the utility grid. It is therefore
the renewable energy research sphere. The main advantages mandatory to have protection and control systems that adapt
of microgrids are environmentally friendly distributed power to the prevailing situation in the network. Differential current
generation technologies and peak demand shaving [1] [2] [3]. protection mechanism is seen as a plausible solution for
Nevertheless, DGs have been found to pose unwanted current control as will be discussed later in this paper.
disturbances on utility power distribution networks [1] [3] [4].
The disturbances can occur during a change of operation
2 Adaptive Control depending on the installed capacity. A solar system can be
installed in many configurations that give different power
There are two types of control: hard programmed control and output suitable for various user needs. A typical single solar
adaptive control ( often referred as knowledge based control). panel produces a peak output voltage of 17 volts. Two panels
Hard programmed control (also called hard wired control) connected in series can to produce an output of 24V. To get
makes use of static relays. Static relays are programmed with more power at the load (usually the battery bank), the panels
hard-wired settings such as values for over-current protection. can be connected in parallel [10]. In larger systems, the
Shortcomings of fixing these values have already been disadvantage of a parallel configuration is that thicker wires
indicated above. On the contrary, knowledge based control will be required because of the larger current. Figure 1 below
makes use of intelligent sensors and advanced communication shows how the system is connected up for domestic
networks built in intelligent electronic devices (IEDs). The installation. The system consists of seven main components:
IEDs have sensors that determine real-time conditions of the solar panels, charge controller, AC breaker panel, a meter, a
network and communicate the information to the control circuit breaker, battery bank and a power inverter. There is an
centre which invokes smart algorithms that determine the option of integrating an AC generator or connecting to the
protected unit to isolate. In this way, control measures adapt grid 7KH FKDUJH FRQWUROOHU¶V PDLQ IXQFWLRQ LV WR UHJXODWH
to the situation at hand. In addition, this suggests that a battery charging to ensure that the batteries do not get over-
decentralized or agent based control system is needed. The charged. The inverter converts DC voltage from the batteries
agent based system is part of the supervisory control and data to AC voltage. E.g. it can convert 12V from the batteries to
acquisition (SCADA) system. 120V AC. The inverter can also act as a point of common
coupling (PCC) linking the system to the utility grid via the
Communication networks on which the SCADA runs is
AC breaker panel as shown in the figure. In grid-connected
important as emphasised in Section 1. Wireless mobile
mode the inverter acts as a charge controller and charges the
broadband communication networks may be used for LAN of
battery bank with power from the utility grid. This is where
IEDs and internet as the backhaul for linking the IEDs and the
the bi-directional nature of power flow in microgrids is
central control agent. Security, reliability and speed of the
encountered.
communication network are of critical importance in ensuring
integrity of control commands data transmitted. The
commands must also reach their destination in a timely
manner. In the past, wireless networks were not popular in
critical applications because of reliability issues caused by
jitter in wireless connectivity. Nevertheless, as new research
shows, connection oriented sessions, and use of cognitive
radios [9], are solutions that can improve reliability. A vital
task that the communication network will perform is
coordination of the buses. This coordination is crucial in
identifying faulted sections of the network [6].
Identifying a faulted section is an important function an
adaptive control system must perform. The issue here is that
when a fault occurs on a certain line connecting two buses,
other relays around the fault may also detect the fault and
want to trip. A challenge is therefore to identify the exact line
that has faulted, attempt to apply automatic re-closers on it; or
to isolate only that line in a timely manner to avoid chain trips
of other breakers to avoid total blackout in a protected area. Figure 1: Overview of a domestic solar system [10], et al
This problem is even more difficult in a meshed microgrid.
Here onwards, this system will be alternatively referred to as
This remainder of this paper is organised as follows. First an a single domestic solar system. In this paper, we assume that
overview of a domestic photovoltaic (PV) system is the system is piloted in a new modern housing area that has
presented. A discussion of the proposed protection scheme is broadband connectivity (e.g. wireless general packet radio
then given. A differential current technique proposed for service (GPRS) or 3G). Households are also assumed to be of
identification of a faulted section of the network is also similar income group hence their energy requirements
discussed. Conclusions are then drawn from the review and relatively equal. In this case, each household has one solar
the findings from the simulations. system. In an ideal setup, each household would have its own
PV array supplying its own load of batteries. Therefore each
3 System overview household would become an autonomous microgrid on its
own. However, to save on wiring and protection equipment,
Installation of photovoltaic panels can provide an effective and to achieve a more reliable system through redundancy,
means to complement energy needs of households. Excess two to three households may be regarded as a single load.
energy that may result can then be sold back to the utility grid Inherent redundancy in this set-up will ensure that when one
DG unit faults, the others will share its load [3]. Figure 2 further in this paper as focus is on differential current
below shows an abstracted view of the envisaged meshed protection.
microgrid.
The most important aspect to be discussed in this section is
The system shown in Figure 2 consists of abstracted solar how to detect a contingency when it occurs. Mitigation of a
systems on households, IEDs, bus bars and sections of the contingency, which follows its detection, is a function of
main utility grid. The solar panels are configured to act as a adaptive control facilitated by communication networks.
single source charging one battery bank and feeding or being
fed from a single bus bar. This retains autonomy of the A meshed microgrid protection is achieved by protecting each
system and simplifies the network, unlike situations where grid element individually. Grid elements are classified into
each household is regarded as a separate source. Section 4 protected zones or protected units. These units are: a line, a
below looks at a protection scheme for this microgrid. source, a bus and a load [6]. The zone boundaries are
protection units such as relays and circuit breakers. In the
introduction, we established that protection units make use of
sensors that detect abnormalities in the characteristics of the
power flowing. Also established was that these characteristics
are different in meshed microgrids to those known from radial
power grids. Nevertheless, in this paper, we propose the use
of direction and magnitude of fault current detected at point
of common contact relay (PCC) as a means of telling whether
the detected fault occurred within the microgrid or on the
main grid. If a fault occurred on the main grid, fault current
will be negative with reference to the microgrid and vice
versa [1]. Knowing whether the fault is within the microgrid
RU RQ PDLQ JULG¶V GLVWULEXWLRQ OLQHV LV YHU\ FULWLFDO ZKHQ
operating in grid-connected mode. It will assist in
determining the islanding necessity. Hence prevent undesired
service interruptions especially when the fault does not affect
the microgrid directly.

Figure 2: Structure of the proposed meshed microgrid 5 Protection Scheme Design

Differential current protection scheme is effective in
situations where many traditional protection methods fail.
4 Protection Scheme This scheme seems to be robust against the dynamics of a
microgrid such as low fault current condition brought by
A power grid is vulnerable to contingencies caused by power inverters. The over-current protection scheme suggests
abnormal variations in the characteristics of supplied power use of directional over-current relays installed at each end of a
as introduced in Section 1. A good practise is to detect, protected unit or phase line. For network protection, this
mitigate and recover from such incidents to continue service method calculates differential current between two ends of a
provision even when contingencies occur. Various protection line [12]. Magnitude and direction of fault current
schemes that deal with these contingencies have been information suffices to provide protection against low and
proposed in literature. Some of this new protection schemes high impedance faults [6]. Zero status of the differential
are: abc-dq0 transformation, THD protection, voltage current means that the current exiting the line is equal to the
restrained over-current relays with inverse time delay, and FXUUHQWHQWHULQJWKHOLQHE\.LUFKKRII¶V/DZ+HQFHWKLVLVD
addition of supplementary current sources [11]. The abc-dq safe condition for the network. If the value of differential
monitors voltages on the dq axis which change during a fault. current is not zero, this condition suggests a problem in the
With THD protection, three phase voltages on from the network. Nevertheless, this scheme will not be effective on its
inverter are Fourier transformed for analysis to determine own due to the fact that DGs frequently connect and
harmonics. There are more harmonics in the voltage signals disconnect from the network bringing uncertainty in the relay
when there is a fault. Voltage retrained method uses relays settings [4]. Many researchers have therefore come to a
that recalibrate automatically when a fault changes the normal conclusion that knowledge based differential current
voltage levels. This allows low current faults to be detected. protection (instead of programmed protection schemes) is the
The last method which is addition of supplementary current solution that meets many of the dynamics of the microgrid.
sources is the arguably the simplest. It was mentioned earlier To adapt to the dynamic characteristics of grid connected
that low fault current is one of the unique problematic microgrid, a new advanced differential protection referred to
characteristics of microgrids. Hence additional current as wide-area differential current protection system is
sources boost the fault current so that it can be detected by proposed [8] [12]. This system was described in Section 2.
traditional relays. Nonetheless, this will not be discussed The system consists of a network of IEDs and digital
differential relays. IEDs implement grid control algorithms on
programmable logic circuits and send control actions to phases were monitored by PV-B1 and B1-Grid scopes as
digital relays. As a result, communication networks are very illustrated in the figures below.
important in a successful implementation of this adaptive
protection scheme. Section 6 describes simulations of a grid
connected PV system performed to demonstrate protection of
a single point of common contact (PCC).

6 Simulation of a PCC Protection

A grid connected solar PV system built on Matlab Simulink
was used for the demonstration. In this simulation, focus was
placed on protecting a single PCC. The PCC is represented
by bus bar B1 illustrated in Figure 3 below. In the proposed
meshed micro-grid, there will be many of these PCCs. Similar Figure 3: Monitoring current sensors around a PCC ± no
principle of differential protection combined with wide area fault condition
monitoring, protection and control can be applied to
coordinate communication between the PCCs for protection
of the entire grid.
Figure 3 below is a picture of the grid connected solar PV
system used. The system consists of the irradiance model, PV
model, inverter model with maximum power point tracking
(MPPT) and a bus/sub-station model acting as the PCC. The
grid basically has two sections: the microgrid section and the
main grid section. The simulation therefore monitors current
in these two sections.

Figure 4: Monitoring current sensors around a PCC ± fault

condition
The second scenario represented a situation where there is a
fault in the PV system side of the grid. This situation was
modelled by a three-phase fault was injected in the PV side so
that it is detected by the IDEs before the bus on the micro-
grid. Figure 5 shows a snapshot of the second scenario.
Figure 5: Matlab Simulink model of the grid-connected
solar PV system
7 Results and Discussions
The simulation aims to demonstrate the unpredictability of The purpose of the simulations described above was to
direction and magnitude of current across the PCC. Hence illustrate how the state of a grid connected microgrid can be
this shows that there is dire need for the proposed adaptive monitored at the PCC. Monitoring the current flowing across
protection schemes that adopt to the continuously changing the PCC (B1 in this case) is the basis of the proposed
situation of the current across the PCC. differential current based WAMPAC. The results described in
this section were captured from the two scenarios introduced
The simulation modelled two scenarios: 1) no-fault and 2) a above. The first set of results show the state of current
three-phase fault scenario. The no-fault scenario modelled the between the PV microgrid and B1 and between B1 and the
normal operation of the network. In Figure 4 below, the two main grid when there is no fault in the network. The second
three-phase V-I measurement units around the bus bar set illustrates the state of phase currents when there is a three-
represent functions of IEDs. The IEDs were simply used to phase fault in the network. Figures 6a and 6b depict the plots
measure and monitor current flowing across B1. Thus of the phase current in the two sections of the network. Figure
measurements of the state of current between the solar PV 6a shows phase currents in the microgrid sections while
microgrid and the PCC (bus bar B1) and between the PCC Figure 6b shows the phase currents in the main grid section of
and the main grid were taken. The currents on the three the network.
 Figure 7a: Phase currents measured between the microgrid Figure 7b: Phase currents measured between the
and the bus bar microgrid and B1

Figure 6b: Phase currents measured between the bus bar Figure 6a: Phase currents measured between the
and the main grid microgrid and B1
Figures 7a and 7b present the second scenario where there is a after 30 seconds of the simulation time. This irregularity in
fault in the network. In the same way as in the first scenario the phase currents detected across the PCC suggests an
described above, phase currents were monitored from both abnormality in the power grid.
sides of the PCC and plotted as shown below.
Speed of the communication networks determines
The graph of Figure 7a above shows that there is a huge surge responsiveness of the protection system. Central microgrid
of current on each of the phases. The current surge dies down control computer system sends commands to relays,
to zero before 0.1 min of the simulation. After the 0.1 min attempting to apply automatic reclosers as the first line of
mark, the current increases slowly until it stabilizes for the protection. If this fault correction fails, protection algorithms
further 24 seconds. At 0.7 min mark, it dies until reaching will then locate and identify the fault using information from
almost zero at 0.8 min mark. the current sensors plus time-stamped GPS coordinates. The
faulty line/phase will then be isolated.
The similarity between Figures 6a and 6b, shows that the
phase currents across B1 are equal in magnitude and have
similar direction across the PCC when there is no fault in the 8 Conclusions and Future Work
network. The result LV LQ KDUPRQ\ ZLWK .LUFKKRII¶V FXUUHQW This paper introduced the concept of distributed generation as
law. This shows a pleasant condition in the network. a means to mitigate the concerning environmental changes.
However, Figures 7a and 7b give a different picture. There is This was motivated as a way of reducing dependence on
no similarity between the figures. This shows that the IED on fossil fuel generated power by supplementing it and shaving
the microgrid side and the IED on main grid section detected peak demand. Furthermore, this is expected to have an impact
different current readings. Hence the difference between the of delaying need for costly investment into transmission lines
magnitudes of the phase currents across the PCC will not be expansions. Domestic solar PV DGs were the focus of the
]HUR7KXVWKLVFRQGLWLRQLVQRWLQKDUPRQ\ZLWK.LUFKKRII¶V paper. This was looked at from the perspective of a small,
law. The sensor on the main grid side only detects current independent and intelligent system called microgrid.
However, the introduction recognised and acknowledged the
negative effects of integrating DGs with the main grid. The [2] Marjan Popov, Domenico Villacci, Vladimir Terzija,
dynamic changes power production capacity of solar PV Alfredo Vaccaro, "An Integrated Framework for Smart
which dictates the direction and magnitude current flowing in Microgrids Modeling, Monitoring, Control,
the grid was seen as the driving force behind the effects. This Communication and Verification," Proceedings of the
effects result in an unstable grid with properties that challenge IEEE, vol. 99, no. 1, pp. 119-132, January 2011.
the existing and well known grid protection techniques. [3] Zhongqiu Wang, Gengyin Li, Ming Zhou, K. L. Lo, Xi
Li, "The Generating Mechanism and Optimal Control of
New microgrid protection schemes that were highlighted in
Power Quality in Micro-grids," in International
this paper were proposed as a result these requirements.
Conference on Power System Technology, 2010, pp. 1-6.
Protection schemes such as: THD protection, addition of
current sources, abc-dq0 transformation and the voltage [4] He Zheng-you, Jiang Wei, Chen Jian, "The overview of
constraint methods were highlighted. Moreover, adaptive protection schemes for distribution systems containing
monitoring and control was introduced as the most viable micro-grid," in Power and Energy Engineering
technique that caters for the unique demands of microgrids Conference (APPEEC), 2011 Asia-Pacific, 2011, pp. 1-
such as solar PV microgrids. The adaptive monitoring and 4.
control technique discussed is called differential current [5] Julio Romero aguero Farid Katiraei, "Solar PV
protection based wide area monitoring and protection scheme. Integration Challenges," IEEE Power & Energy
Magazine, vol. 11, pp. 62 -71, April 2011.
Regardless of problems identified, the paper further refined
its scope to focus on differential current protection scheme. A [6] Du Yi, Paquette A., Buck E., Harley R., Divan D., Prasai
simulation was performed on Matlab Simulink to demonstrate A., "Protection of meshed microgrids with
how this scheme works. The simulation showed the essence communication overlay," in Energy Conversion
of differential protection scheme which entailed current Congress and Exposition (ECCE), 2010, pp. 64-71.
sensors placed across the protected unit and using the state of [7] S. Chowdhury, S.P. Chowdhury, P. Crossely, Microgrids
the current detected by the sensor to determine whether there and Actrive Distribution Networks. London, United
is a fault or not. Also discussed was that the information from Kingodom: Institude of Engineering and Technology,
the sensor will then be used as input to protection and control 2009.
algorithms executed by centralized control and protection [8] Song Shaoqun, Wang Dewen Zhu Yongli, "Multiagents-
computer systems. based wide area protection with best-effort adaptive
The simulation results presented in Section 7, the following strategy," Electrical and Power Energy Systems, vol. 31,
conclusions can be drawn. Differential protection scheme is pp. 94-99, October 2009.
the simplest and effective protection scheme for microgrid [9] Jean-Francois Frigon Arash Azarfar, "Improving the
protection. It is reliant on the stability of the supporting Reliability of Wireless Networks Using Cognitive
communication networks. It makes use of IEDs and sensors. Radios," IEEE Communications Surveys & Tutorials,
Because of these features, protection and control decisions are vol. PP, no. 99, pp. 1-17, March 2011.
inherently performed from a central computer system. [10] Free Sun Power. (2011, July) Free Sun Power. [Online].
http://www.freesunpower.com/index.php
Further work leading on from these initial experiments can
focus on extending the model of the grid-connected PV [11] Dysko A.,Burt G.M, Xinyao L., "Application of
system. The extension can add complexity and help further Communication Based distribution Protection," in 45th
understanding of the dynamics of current in a meshed PV International Universities Power Engineering
microgrid. Furthermore, as highlighted before, Conference, Cardiff, 2010, pp. 1-6.
communication networks are vital for the protection scheme [12] K.K. Li, W.L. Chan, Xiangjun Zeng, Dongyuan Shi,
discussed in this paper. Therefore, more work needs to be Xianzhong Duan, Su Sheng, "Adaptive Agent-Based
done to find an optimal networking solution to coordinate the Wide-Area Current Differential Protection System,"
IEDs. IEEE Transactions on Industry Applications, vol. 46, no.
5, pp. 2111-2117, September/October 2010.
References
[1] S. Raiti, S. Conti, "Integrated Protection Scheme to
Coordinate MV Distribution Network Devices, DG
Interface Protections and Micro-Grids Operation," in
Clean Electrical Power, 2009 International Conference
on, 2009, pp. 640-646.