A Controller for Fair and Coordinated Active Power

Curtailment of Low Voltage PV Inverters

Aadil Latif 1,†, Prof. Wolfgang Gawlik 2 and Prof. Peter Palensky 3

1 Energy Department, Austrian Institute of Technology, Vienna, Austria; aadil.latif.fl@ait.ac.at

2 Institute of Energy Systems and Electrical Drives, Vienna University of Technology, Vienna, Austria;
wolfgang.gawlik@tuwien.ac.at
3 Department of Electrical Sustainable Energy, Delft University of Technology, Delft, Netherlands;
P.Palensky@tudelft.nl

Abstract: PV inverters provide voltage support using active power curtailment (Active power –
Voltage droop control). There is an inherent unfairness in the active power curtailment scheme.
When generation is high and consumption is low, voltages at the end of the feeder tend to be the
highest. This results in high curtailment of active power output of the inverters located at the end
of the feeder, and low or even no curtailment for the inverts located closer to the transformer. A
Secondary voltage controller has been implemented to mitigate this unfairness in active power
curtailment based voltage support schemes. The focus of this work is to quantify this unfairness
and develop methods that enable any two house owners serviced by the same feeder intending to
install rooftop PV systems as a potential source of revenue have equal opportunity in PV
generation and equal participation in voltage regulation.

Keywords: Coordinated voltage regulation; Active power curtailment; Fairness; Secondary

voltage control

1. Introduction

Advances in technology, ever increasing demand for green energy and favorable government
policies have resulted in rapid increase of photo voltaic (PV) systems in a number of energy
markets around the world [1]. A considerable percentage of the PV systems have been installed in
low voltage network (LVN). High PV penetration results in a number of engineering challenges like
harmonic distortion, voltage flicker, reverse power flow and over voltage [2, 3, 4] .These challenges
are a major limiting factor in increasing PV penetration in distribution networks.

Distribution systems have been designed to consume power generated at the high voltage side of
the electrical grid. Current distribution systems at the time of conception were designed to cater for
the voltage drop caused by consumer load and inherent impedance of the distribution lines.
Injection of power at the low voltage network results in over voltage which the networks were not
designed to cater for. Traditionally, network voltage has been regulated by the on-load tap changer
of the HV/MV transformers [5]. Distributed generation at medium and low voltage networks
reduces voltage regulation capability of such a scheme [6]. At times when there is high generation
from PVs and low load consumption, power is injected into the grid and line impedance causes the
voltage at the point of common coupling (PCC) to rise [7]. Distribution system operators (DSO)
therefore require additional control over voltage regulation.

In literature, a number of schemes have been suggested to improve voltage regulation in the
distribution network and thereby making it possible to increase the PV penetration. In [8] Turitsyn
et al proposed reactive power compensation (RPC) by PV owners. RPC reduces the grid’s power
factor thereby increasing losses. In several countries e.g. Canada, Indian, Bangladesh reactive power
compensation is not legal by rooftop PV owners and inverters are only allowed to operate at unity
power factor [9-12]. In these countries, active power curtailment is the only way of regulating
voltage at the point of common coupling.R/X ratio is typically > 1 in LVNs therefore, voltage is
more sensitive to active power injection rather than reactive power compensation [7]. On the other
hand, active power curtailment means loss of revenue for PV owners. In many papers it has been
suggested to combine the two schemes for better voltage regulation [13]. A number of articles
suggest using an OLTC on the MV/LV transformer, which traditionally does not regulate voltage
[14, 15]. Coordinated control for APC and RPC of PVs is another solution to the voltage regulation
problem proposed by a number of papers [16-20]. Implementation of these strategies would require
communication infrastructure.

The Voltage at the PCC is a function of the active and reactive power being injected at that point
and the grid impedance as seen by the inverter at PCC. As the electrical distance from a transformer
increases so does the voltage sensitivity. At times of high generation inverters connected at PCC
with higher voltage sensitivity may violate voltage limits. As a result, the inverters that are
violating voltage limits have their output curtailed. Essentially, PV owners located at the end of the
feeder are more susceptible to voltage violations and getting their power curtailed resulting in loss
of revenue. DSO commonly calculate a feeders’ housing capacity using the worst case over voltage
scenario which might further limit the size of PV that can be installed by the house owner connect
near the end of the feeder.

Different solutions have been put forward to mitigate this inherent unfairness in droop based APC
schemes for voltage control.

A) Policy based solutions – In Japan for example as of 2012, if power from a PV inverter is curtailed
for more than 30 days a year (8%), the owner has to be compensated by the DSO [21].

B) [19] suggested using batteries at times of low consumption and high DG generation. An added
advantage of using battery storage is that it minimizes fluctuations in DG output.

C) Researchers have also suggested using demand side management and increasing self-
consumption during the time power curtailment is an issue [23].

D) Tonkoski et al. [7] developed a coordinated algorithm to mitigate unfairness by experimenting

with different droop curves for DG owners and achieved excellent results. Perera et al. [24]
developed an iterative algorithm that improves fairness while keeping voltage with limits.

The scope of this work is to investigate whether it is possible to increase fairness of APC schemes
using coordinated control. It further investigates the cost of this increased fairness (increase in
curtailment). Section 2 details network model and simulation setup. Section 3 explains principles
active power curtailment. In section 4 the formulation of the KPIs used for comparison has been
presented. Section 5 details the proposed method. Simulation results are presented and discussed in
detail in section 6. Finally in section 7 conclusions have been drawn on the basis of this work.

2. Simulation Setup

2.1. Network Model

Unlike industrial or commercial feeders, active power demand in residential feeders typically peaks
at night which does not coincide with peak generation, which is usually mid-day. This is why,
residential feeders are more prone to over voltages. The test case chosen for this work consists of a
low voltage residential network connected to medium voltage network through a 75 KVA
transformer. The low voltage network consists of two feeders connecting 27 consumers. For the

2
purpose of the test case, 16 photo voltaic systems have been placed randomly within the network.
The PV systems have been randomly sized between 4 and 9 KVA. PV inverters provide voltage
support using active power curtailment. For the purpose of simulation one minute average load
and generation profiles have been used. Network data is listed in Tables 1 to 4. The topology of
residential test feeder used for this work is presented in Fig. 1.

Figure 1. Residential test feeder configuration

TABLE 1. LINE LENGTHS

LINE LENGTH LINE LENGTH LINE LENGTH

B1 – C1 69 m I1 – J1 46 m G4 – H3 53 m

C1 – D1 67 m J1 – K1 36 m H3 – I2 44 m

D1 – E1 67 m . I2 – J2 40 m

E1 – F1 85 m B2 – C2 88 m F2 – G5 67 m

F1 – G1 55 m C2 – D2 74 m G5 – H4 66 m

F1 – G2 46 m D2 – E2 90 m H4 – I3 42 m

F1 – G3 77 m E2 – F2 67 m I3 – J3 44 m

G2 – H1 76 m F2 – G4 105 m J3 – K2 38 m

H1 – I1 54 m G4 – H2 61 m J3 – K3 61 m

TABLE 2. TRANSFORMER DATA

3
 Voltage Rating Rated Power Type Tap changer Volts/Tap

20 kV / 0.4 kV 80 kVA Dyn11 -4 : +4 1.25 %

TABLE 3. GENERATOR OPERATIONAL LIMITS

INVERETER ID RATING - KVA INVERETER ID RATING - KVA

PV C1 4 PV D2 5
PV D1 5 PV H3 6
PV G1 9 PV H4 4
PV H1 7 PV J2-1 8
PV I1-1 6 PV J2-2 5
PV I1-2 5 PV J3 8
PV K1 7 PV K2 8
PV C2 5 PV K3 7

TABLE 4. LINE PARAMETERS

RESISTANCE REACTANCE CAPACITANCE

+SEQ 0,3211 OHM/KM 0,069 OHM/KM 0,5900 UF/KM

0SEQ 1.0844 OHM/KM 0.276 OHM/KM 0,3308 UF/KM

2.2. PowerFactory Python interfacing

The distribution network and the local voltage controller detailed in section 2.1 and 3 have been
implemented in PowerFactory a popular power network simulation tool. The secondary voltage
controller, detailed in section 5, has been implemeted in Python because of PowerFactory’s
limitations while dealing with matrices. During RMS simulation it is possible for controllers
modelled in DSL (DigSilent programming language) blocks to call functions from an external C++
library. A C++ library has been used to implement sockets to communicate with Python. In this
work socket based communication approach has been prefered as it can easily be extended to
incorporate a network simulator such as OMNET++ to study the impact of communication on
voltage control for future work. Fig. 2 is a graphical illustration of coupling scheme used.

Figure 2. PowerFactory and Python coupling via sockets using external DLL

3. Simulation Setup
Power flow through a line can be calculated using the following equations. Where, R + jX is the
impedance of the line, 𝐸1and 𝐸2are source and line end voltage respectively and
Φ is the phase angle between the two voltages.
𝐸1
𝑃 = [𝑅(𝐸1 − 𝐸2 cos Φ) + 𝑋 𝐸1 sin Φ], (1)
𝑅2+ 𝑋 2

𝐸1
𝑄 = [−𝑅 𝐸2 sin Φ + 𝑋 (𝐸1 − 𝐸2 cos Φ)] (2)
𝑅2 + 𝑋 2
In low voltage networks R/X ratio is typically high. High R/X ratio results in limited impact of
reactive power on voltage regulation making active power curtailment a more effective option for
voltage control. For high R/X ratios the reactance of the network can be neglected (X ≈ 0). The phase
angle Φ is generally small and sin Φ ≈ Φ and cos Φ ≈ 1 are reasonable approximations. Using these

4
approximations, the change in voltage at the PCC can be formulated as Eqn. (3) [24]. For PV
inverter units operating at unity power factor this expression can be reduced to Eqn. (4).
𝑃𝑅 + 𝑄𝑋
𝛥𝑉 ≈ (3)
𝑉
𝑃𝑅 (4)
𝛥𝑉 ≈
𝑉
A local droop based voltage controller has been implemented in PowerFactory. Below a critical
voltage 𝑢𝑐𝑟𝑖 inverter output follows maximum power point tracking (mppt). The active power is
curtailed linearly once the voltage at the bus exceeds the critical voltage. If voltage at PCC exceeds
the upper voltage limit 𝑢𝑡ℎ𝑟 the active power output of the inverter is reduced to the minimum
permissible value 𝑃𝑚𝑖𝑛 .
𝑃𝑐𝑎𝑙 = 𝑃𝑚𝑝𝑝𝑡 𝑢 < 𝑢𝑐𝑟𝑖
𝑐𝑎𝑙 𝑚𝑝𝑝𝑡 𝑡ℎ𝑟
{𝑃 = 𝑃 − 𝑚 (𝑢 − 𝑢) 𝑢 ≤ 𝑢 ≤ 𝑢𝑡ℎ𝑟
𝑐𝑟𝑖
(5)
𝑡ℎ𝑟
𝑐𝑎𝑙
𝑃 = 𝑃 𝑚𝑖𝑛 𝑢 ≤ 𝑢

3.1. Voltage sensitivity estimation

In this paper, additional logic has been implemented to estimate the voltage sensitivity of the PCC
to active power (du/dp). When the Calc signal is set high, a pulse generator generates pulses for a
predefined time interval. If 𝑃𝑐𝑎𝑙 is greater than a critical value 𝑃𝑐𝑟𝑖 , the inverter set point is
switched to zero temporarily. The change in voltage is then recorded and du/dp is estimated using
finite difference. The process is repeated to and the results are filtered to minimize estimation
errors. For this work, 𝑃𝑐𝑟𝑖 has been set to 70% of the rated power. 𝑃𝑐𝑟𝑖 is intentionally set high so
that a significant change in the voltage can be recorded and the impact of other inverters and the
load present in the system can be neglected. The main though process behind estimation of voltage
sensitivity is to provide a truly plug and play solution for voltage control.

Figure 3. Droop based APC controller with du/dp estimation logic

4. Formulation of key performance indices (KPIs)

To measure the performance of the coordinated control scheme and for the purpose of comparison
with currently existing schemes, two KPIs have been formulated namely total energy curtailed
(TCE) and curtailment unfairness index (CUI).

4.1. Total Curtailed Energy

5
Total energy curtailed is an important KPI for comparison as any proposed coordination control
scheme should not significantly increase loss of green energy as it results in loss of revenue for the
PV system owner and loss of green energy for DSO.
𝑚𝑝𝑝𝑡
𝑃𝑖𝑐𝑢𝑟 = 𝑃𝑖𝑖𝑛𝑣 − 𝑃𝑖 , (6)

𝐸𝑖𝑐𝑢𝑟 = ∫ 𝑃𝑖𝑐𝑢𝑟 𝑑𝑡 (7)

𝑐𝑢𝑟
TCE = ∑𝑁
1 𝐸𝑖 (8)

𝑓𝑜𝑟 𝑖 = 1, . . , 𝑁
𝑚𝑝𝑝𝑡
Where N is the number of active PV systems installed on a particular feeder. 𝑃𝑖 , 𝑃𝑖𝑖𝑛𝑣 , 𝐸𝑖𝑐𝑢𝑟 are
the maximum active power output, actual active power output and the curtailed energy of the i th
inverter. TCE is therefore total energy curtailed from all inverters present in the feeder performing
APC.

4.2. Fairness in active power curtailment

Fairness in active power curtailment can be examined from two perspectives. The first perspective
being, the loss of revenue for all PV system owners should be equal irrespective of the nominal
rating of the PV system. CUI1 index formulated in equation (9) has been used in this work to

∑(Eicur − ̅̅̅̅̅̅
Ecur )2
CUI1 = √ (9)
N−1

Where 𝐸̅̅̅̅̅̅
𝑐𝑢𝑟 is the average energy curtailed from all PV systems present in a feeder. This implies

that an owner of a PV system with a smaller rating would incur higher percentage reduction in
revenue. Another way to accommodate the unfairness is to ensure that percentage reduction in
revenue for every PV system owner is the same. A KPI has formulated to evaluate such a
perspective in equations (10) and (11).

ecur
i = Eicur /Pirated (10)

∑(ecur
i − ̅̅̅̅̅
ecur )2
CUI2 = √ (11)
N−1

Where 𝑃𝑖𝑟𝑎𝑡𝑒𝑑 is the rated power for the ith inverter and ̅̅̅̅̅̅
𝑒 𝑐𝑢𝑟 is average of the normalized curtailed
energies of all PV systems present in a feeder.

5. Proposed Algorithm
In this paper a coordinated control scheme has been proposed in which each feeder in a low voltage
network has a separate controller. This controller ensures that every PV system owner participates
equally in voltage regulation of a feeder. This work assumes a bidirectional ideal communication
channel between the coordinating controller and smart inverters. The controller has two modes of
operation. In normal operation, when maximum voltage at every point in feeder is less than 𝑢𝑙𝑏 , no
energy is curtailed. In distressed mode, the controller generates set points for every inverter. This
subsection details the working of the proposed controller.

5.1. Total Curtailed Energy

6
The impact of variation of active power injected by the inverters can be measured quantitatively
using the sensitivity matrix [25]. In this work, inverters have been modeled to operate at unity
power factor hence the impact of reactive power on voltage can be neglected.
𝜕𝑃1 𝜕𝑃1
⋯
𝛥𝑃1 𝜕𝑉1 𝜕𝑉𝑁 𝛥𝑉1
[ ⋮ ]= ⋮ ⋱ ⋮ [ ⋮ ] (12)
𝛥𝑃𝑁 𝜕𝑃𝑁 𝜕𝑃𝑁 𝛥𝑉𝑁
⋯
[𝜕𝑉1 𝜕𝑉𝑁 ]
Voltage sensitivity of a node to active power injection at any node in a radial distribution network
can be calculated using [25,26].
𝜕𝑃𝑖 𝑅𝑖𝑗
= − 𝑛𝑜𝑚 (13)
𝜕𝑉𝑗 𝑉

With 𝑖, 𝑗 = 1,2, … , 𝑁

The expressions suggest that longer the feeder length more susceptible it is to over voltages. The
resistance matrix can be expanded and rewritten as a function of line length and resistance per
kilometer.
𝑅𝑖𝑗 = 𝑟𝑖𝑗 . 𝐿𝑖𝑗 (14)
For i=j, 𝐿𝑖𝑗 is the minimum electrical distance from the transformer to the i th node. In case i≠j, 𝐿𝑖𝑗 is the
maximum overlap of the paths formed from the transformer to the ith and jth node. 𝑟𝑖𝑗 is the average
resistance per kilometer of the braches belonging to 𝐿𝑖𝑗 .
In rural low voltage networks, pole to pole distance are usually longer when compared to urban or
village networks [27]. For i>j or j>I, if overlapping path length is much greater than non-overlapping
path lengths the off diagonal elements of the sensitivity matrix can be approximated as the diagonal
elements of the matrix.

𝜕𝑃1 𝜕𝑃1 𝜕𝑃1

⋯
𝜕𝑉1 𝜕𝑉1 𝜕𝑉1
𝜕𝑃1 𝜕𝑃2 𝜕𝑃2
𝑆𝑖𝑗 = 𝜕𝑉1 ⋯
𝜕𝑉2 𝜕𝑉2 (15)
⋮ ⋮ ⋱ ⋮
𝜕𝑃1 𝜕𝑃2 𝜕𝑃𝑁
⋯
[𝜕𝑉1 𝜕𝑉2 𝜕𝑉𝑁 ]
As each inverter is capable of calculating the voltage sensitivity of the connected node to active
power (diagonal elements of the matrix), rest of the matrix can be populated.

5.2. Active power reduction calculation

At initial run, the coordinating controller requests for and receives rated powers, estimated
sensitivities and node ID from every inverter connected to the feeder. Rated output power of the PV
systems are stored in the matrix 𝑅𝑃𝑘×𝑙 , where k is the number of unique nodes in a feeder with a PV
connection and l is the maximum number of PVs connected to a single node. The sensitivities
arranged in an ascending order form the diagonal part of the sensitivity matrix 𝑆 𝑘×𝑘 ,. Non-diagonal
elements of the sensitivity matrix are populated using Eqn. (15).
Voltage at the secondary side of the MV/LV transformer is calculated using the current tap position
𝑇𝑃, percentage voltage change per tap 𝑢𝑡𝑎𝑝 and nominal voltage 𝑢𝑛𝑜𝑚 .

𝑢𝑡𝑟𝑙𝑣 = 𝑢𝑛𝑜𝑚 − 𝑢𝑡𝑎𝑝 . 𝑇𝑃 (16)

Maximum voltage rise at all nodes connected to PV systems is calculated using the sensitivity matrix
𝑆𝑖𝑗 and the rated powers 𝑃𝑖𝑟𝑎𝑡𝑒𝑑 .

7
 𝑘 𝑙
𝛥𝑢𝑖𝑚𝑎𝑥 = ∑ ∑ 𝑅𝑃𝑗𝑚 . 𝑆𝑖𝑗 (17)
𝑗=1 𝑚=1

With 𝑖 = 1,2, … , 𝑘

Voltages measured at PCC of each inverter are sent to the coordinating controller and continuously
monitored. Critical voltage for each inverter is set using expressions (18) and (19), where 𝑢𝑚𝑎𝑥 is the
maximum of the measured voltages.
𝑢𝑚𝑎𝑥 = max(𝑢1,2,…,𝑘 ), (18)

𝑢𝑖𝑐𝑟𝑖 = 𝑢𝑖 |𝑢 𝑚𝑎𝑥=𝑢𝑙𝑏 (19)

If 𝐼 is the index of the node with the maximum measured voltage, reduction in voltage required (𝛥𝑢 )
to ensure no over voltage occurs is calculated using (20).

𝛥𝑢 = 𝑢𝑡𝑟𝑙𝑣 + 𝛥𝑢𝐼𝑚𝑎𝑥 − 𝑢𝑢𝑏 (20)

Reduction required in active power injection for each inverter is calculated using either (21) or (22,23)
depending on which fairness perspective is currently under investigation.

A. Equal loss of revenue for each PV owner

𝛥𝑢
𝛥𝑃𝑚𝑎𝑥 = (21)
∑𝑘𝑖=1 𝑚𝑖 . 𝑆𝑖𝐼
Where 𝑚𝑖 is number of non-zero elements in the ith row of 𝑅𝑃 matrix (number of PV connected to a
node). It should be noted that 𝛥𝑃𝑚𝑎𝑥 is a single value hence the same every inverter irrespective of
its rated output.

B. Equal percentage reduction in revenue

𝛥𝑢
𝑧= (22)
∑𝑘𝑗=1 ∑𝑙𝑚=1 𝑅𝑃𝑖𝑚 . 𝑆𝑖𝐼

𝛥𝑃𝑚𝑎𝑥 = 𝑧 . 𝑅𝑃 (23)
Where 𝑧 is the percentage reduction in active power for each inverter. In this case, 𝛥𝑃𝑚𝑎𝑥 is k x l
matrix. In the final step, S function is used to reduce excessive curtailment.

𝛥𝑃𝑖𝑗𝑚𝑎𝑥
𝛥𝑃𝑖𝑗 = 𝐵−𝑢 𝑚𝑎𝑥 (24)
1+𝑒 𝐴

where

𝑢𝑢𝑏 − 𝑢𝑙𝑏
𝐵= (25)
2
Where 𝐴 is a constant that controls the rate at which 𝛥𝑃𝑖𝑗 increases from zero to 𝛥𝑃𝑖𝑗𝑚𝑎𝑥 . This ensures
that maximum possible curtailment occurs only when necessary. Figure 4 presents graphic overview
of the proposed scheme.

8
 Figure 4. Flowchart of the proposed control scheme

6. Results and discussion

In the first step, local voltage sensitivities (diagonal values of the sensitivity matrix 𝑆𝑖𝑗 ) to active
power are calculated by the PV inverter using the logic presented in Fig(3). Once Calc signal is set
high, inverter momentarily disconnects every 20 minutes 25 times. Change in output power and
voltage at the PCC is used to estimate the local sensitivity. The coordinating controller ensures no
two inverters calculate the sensitivity at the same time. Figure 5 shows inverter output and du/dp
estimation results of inverter at node K3.

Figure 5. (a) Inverter output and (b) du/dp estimation results

Figure 6 presents a comparison of the du/dP calculated by a PowerFactory PV connected nodes and
the values calculated by the inverters. For nodes with multiple PV systems connected, the value has
been averaged. The average error between the actual and the estimated value is 3.4 %.

Figure 6. A comparison of actual and estimated du/dp values

9
The base case results presented in figure 7 are for PV systems without local and coordinating
voltage controller. For a scenario with high PV generation and low consumption, the voltages for
measured from feeder 1 are within the permissible voltage band and hence do not require
additional control for regulation. The Feeder 2 exceeds the voltage maximum permissible 𝑢𝑢𝑏 and
requires additional voltage controls.

Figure 7. Base case voltage profiles for feeder 1 and 2

In this paper, two design approaches have been implemented for voltage regulation. Each approach
is discussed in detail in the subsequent subsections.

6.1. Local voltage regulation

In this approach, each inverter is capable of voltage regulation using APC. This approach requires
only local information hence no communication is required. For the experiment, it has been
assumed that maximum permissible reduction in active power output of the inverter (𝑃𝑚𝑖𝑛 ) is 50%
of the rated power. The critical (𝑢𝑐𝑟𝑖 ) and threshold voltages (𝑢𝑡ℎ𝑟 ) have been set at 1.03 and 1.05
p.u. respectively for every inverter on feeder 2. Voltage profiles for the inverters are presented in
figure 8 (a).

Figure 8. (a) Voltage profiles for feeder 2 (b) Curtailed power from each inverter

Figure 9 (b) presents power curtailed from each PV inverter. The figure shows the inherent
unfairness in local APC schemes. PV inverters connected close to the transformer get significantly
less energy curtailed in comparison to the ones located near the end of the feeder.

6.2. Coordinated voltage regulation

In this work two coordinating controller models have been implemented to minimize the two
perspectives of unfairness presented in section 4.2. Coordinating control schemes implemented in
this paper require information from every PV inverter installed on the feeder, which requires a
communication network. In this paper, an ideal communication channel has been assumed. For
simulation purposes, the upper 𝑢𝑢𝑏 and lower 𝑢𝑙𝑏 limits of the voltage regulation band have been
set at 1.035 and 1.05 respectively.

A. Equal loss of revenue for each PV owner

The first controller (CC1) generates set points for critical voltage critical 𝑢𝑐𝑟𝑖 in p.u. and the
reduction in active power 𝛥𝑃 in kW. 𝛥𝑃 is the same for all the inverters which ensures equal
curtailment for every PV system. Figure 9 presents voltage profiles at PCC and power curtailed
from each PV inverter.

10
 Figure 9. (a) Voltage profiles for feeder 2 (b) Curtailed power from each inverter

B. Equal percentage reduction in revenue

The second controller (CC2) like the controller implemented for equal revenue loss also generates
set points for critical voltage critical 𝑢𝑐𝑟𝑖 in p.u. and the reduction in active power 𝛥𝑃 in kW. In this
case however, 𝛥𝑃 is matrix and the set point for each inverter is a function of its rated power. Figure
10 shows feeder voltage profiles and power curtailed from each PV inverter. Comparing figures
8(b), 9(b) and 10(b), impact of different control schemes on curtailed power is visible.

Figure 10. (a) Voltage profiles for feeder 2 (b) Curtailed power from each inverter

Table 5 lists the energy curtailed from each PV owner for the three control schemes simulated for
the paper. Voltage regulation schemes using local APC are unfair to PV system owner connected
near the end of the feeder. The energy curtailed from PV C2 is only 1.5% of the total energy
curtailed from all inverters. PV K2 on the other hand gets 44.7 kWh curtailed which is over 17% of
the total curtailed energy. It therefore contributes over ten times more that PV C2 in the voltage
regulation process.
In the second scenario (CC1), energy curtailed from each PV ranges between 11.2 – 12.9%. In this
scenario, every PV owner contributes approximately equally in the voltage regulation process. One
drawback of this scheme however is that PV owners with smaller ratings incur a bigger percentage
decrease in revenue generated by the PV. At times of peak curtailment for example, (figure 9(b))
power curtailed from each PV system is 1.8 kW. PV K2 rated at 8 kW gets 22.5% of the output
curtailed, but PV H4 rated at 4kW loses 45% of the revenue generated by the PV at peak output.

TABLE 5. CURTAILED ENERGY OF EACH INVERETR FOR DIFFERENT CONTROL SCHEMES

Local CC1 CC2

cur cur cur cur
E e E e E cur ecur

(kWh) (kWh) (kWh)

PV C2 3.7 0.75 32.9 6.59 23.4 4.69

PV D2 8.1 1.63 32.9 6.59 23.4 4.69

PV H4 27.9 4.65 31.2 5.20 28.1 4.69

PV H3 19.2 4.81 35.1 8.87 18.7 4.69

PV J3 42.4 5.30 28.6 3.58 37.5 4.69

PV J2-1 26.5 5.30 32.9 6.59 23.4 4.69

PV J2-2 43.7 5.47 28.6 3.58 37.5 4.69

PV K2 44.7 5.58 28.6 3.58 37.5 4.69

PV K3 39.1 5.58 29.8 4.25 32.8 4.69

11
In the final scenario (CC2), energy curtailed from each PV is a function also the function of the
rating of the PV system. Energy curtailed from each PV ranges between 7.3 – 14.6% of total
curtailed energy. In this study case, All PV system owners do not participate equally in voltage
regulation. However, every PV owner on the feeder loses the same percentage of revenue.
To evaluate the performance of control schemes implemented in this paper, KPI formulated in
section 4 have been calculated (table 6) using data from table 5. Loss of green energy increases for
control schemes CC1 and CC2 by 9.9% and 2.7% respectively. Curtailment unfairness index 1
(CUI1) reduces by 84% for CC1 and by 53% for CC2. CC1 has negligible impact on the second
fairness index while CC2 reduces CUI2 to zero meaning every PV owner participates equally in
voltage regulation.

TABLE 6. CALCULATED KPI FOR DIFFERENT CONTROL SCHEMES

Local CC1 CC2

TCE - kWh 255,34 280,78 262,40

CUI1 - kWh 15,47 2,41 7,33

CUI2 - h 1,83 1,84 0,00

7. Conclusions

With increasing PV penetration in low voltage networks, voltage regulation is a challenge. Active
power curtailment is one possible solution for over voltages. Local APC schemes are inherently
unfair (figure 8(b), tables 5 and 6). In this paper a truly plug and play coordinated control scheme
has been implemented with two variations. Two perspectives of unfairness have been presented in
the paper and two indices were formulated to quantify the unfairness in APC schemes. The
effectiveness of the control schemes have been tested by implementing network in PowerFactory
and the controllers in Python. Both variants of the coordinating controllers were able to increase
fairness in APC significantly (table 6). As proof of concept, the proposed algorithms have been
implemented on a single feeder. Future work will be the implementation of the proposed algorithms on a
number of LV networks to better understand the effectiveness and the limitations of the proposed schemes.

References

[1] G. Masson, M. Latour, M. Rekinger, I.-T. Theologitis and M. Papoutsi, "Global market outlook for
photovoltaics 2013-2017," European Photovoltaic Industry Association, pp. 12-32, 2013.
[2] S. Ali, N. Pearsall and G. Putrus, "Impact of high penetration level of grid-connected photovoltaic
systems on the UK low voltage distribution network," in International Conference on Renewable
Energies and Power Quality, 2012.
[3] T. Stetz, Autonomous Voltage Control Strategies in Distribution Grids with Photovoltaic Systems:
Technical and Economic Assessment, vol. 1, kassel university press GmbH, 2014.
[4] J. Guisado, P. Carvalho, L. Ferreira, J. J. Santana and G. Marques, "Voltage control challenges and
potential solutions for large-scale integration of PV resources in LV networks," in Integration of
Renewables into the Distribution Grid, CIRED 2012 Workshop, 2012.
[5] T. Gonen, Electric power distribution engineering, CRC press, 2014.
[6] A. Hagehaugen, Voltage Control in Distribution Network with Local Generation, Institutt for
elkraftteknikk, 2014.
[7] R. Tonkoski, L. A. Lopes and T. H. El-Fouly, "Coordinated active power curtailment of grid
connected PV inverters for overvoltage prevention," Sustainable Energy, IEEE Transactions on, vol. 2,
no. 2, pp. 139-147, 2011.
[8] K. Turitsyn, S. Backhaus, M. Chertkov and others, "Options for control of reactive power by
distributed photovoltaic generators," Proceedings of the IEEE, vol. 99, no. 6, pp. 1063-1073, 2011.
[9] F. Olivier, P. Aristidou, D. Ernst and T. Van Cutsem, "Active Management of Low-Voltage

12
 Networks forMitigating Overvoltages Due to Photovoltaic Units".
[10] Tamimi, Behnam, Claudio Canizares, and Kankar Bhattacharya. "System stability impact of large-
scale and distributed solar photovoltaic generation: The case of Ontario, Canada." Sustainable
Energy, IEEE Transactions on 4, no. 3 (2013): 680-688.
[11] Mojumdar, Md Rejwanur Rashid, Arif Md Waliullah Bhuiyan, Hamza Kadir, and Md Nizamul
Haque Shakil. "Design & Analysis of an Optimized Grid-tied PV System: Perspective Bangladesh."
IACSIT international journal of engineering and technology 3, no. 4 (2011).
[12] "Guidelines for Grid-connected Small Scale (Rooftop) Solar PV Systems for Tamil Nadu", April
2014. Available: http://www.teda.in/pdf/Specification_Grid_Tie_SPV_plant.pdf
[13] M. Triggianese, J. Morren, S.W.H. de Haan and P. Marino, "Improved and extended DG capability
in voltage regulation by reactive and active power," in Power Engineering, Energy and Electrical
Drives, 2007. POWERENG 2007. International Conference on, IEEE, 2007, pp. 583--588.
[14] C. Gao and M. A. Redfern, "A review of voltage control techniques of networks with distributed
generations using On-Load Tap Changer transformers," in Universities Power Engineering Conference
(UPEC), 2010 45th International, 2010.
[15] Y. Li and P. Crossley, "Voltage control on unbalanced LV networks using tap changing
transformers," 2012.
[16] N. Safitri, F. Shahnia, M. Masoum and others, "Coordination of single-phase rooftop PVs to
regulate voltage profiles of unbalanced residential feeders," in Power Engineering Conference
(AUPEC), 2014 Australasian Universities, 2014.
[17] P. N. Vovos, A. E. Kiprakis, G. P. Harrison and others, "Centralized and distributed voltage control:
Impact on distributed generation penetration," Power Systems, IEEE Transactions on, vol. 22, no. 1,
pp. 476-483, 2007.
[18] T. Vandoorn, J. De Kooning, B. Meersman and L. Vandevelde, "Soft curtailment for voltage limiting
in low-voltage networks through reactive or active power droops," in Energy Conference and
Exhibition (ENERGYCON), 2012 IEEE International, 2012.
[19] V. M. Mansoor, P. H. Nguyen, W. Kling and others, "An integrated control for overvoltage
mitigation in the distribution network," in Innovative Smart Grid Technologies Conference Europe
(ISGT-Europe), 2014 IEEE PES, 2014.
[20] D. Reeves, G. Nourbakhsh, G. Mokhtari and A. Ghosh, "A distributed control based coordination
scheme of household PV systems for overvoltage prevention," in Power and Energy Society General
Meeting (PES), 2013 IEEE, 2013.
[21] D. Lew, L. Bird, M. Milligan, B. Speer, X. Wang, E. M. Carlini, A. Estanqueiro, D. Flynn, E. Gomez-
Lazaro, N. Menemenlis and others, "Wind and solar curtailment," in Int. Workshop on Large-Scale
Integration of Wind Power Into Power Systems, 2013.
[22] M. C. Bozchalui and R. Sharma, "Optimal operation of Energy Storage in distribution systems with
Renewable Energy Resources," in Power Systems Conference (PSC), 2014 Clemson University, 2014.
[23] M. Castillo-Cagigal, A. Gutierrez, F. Monasterio-Huelin, E. Caamano-Martin, D. Masa and J.
Jimenez-Leube, "A semi-distributed electric demand-side management system with PV generation
for self-consumption enhancement," Energy Conversion and Management, vol. 52, no. 7, pp. 2659-
2666, 2011.
[24] B. K. Perera, P. Ciufo, and S. Perera, "Power sharing among multiple solar photovoltaic (PV)
systems in a radial distribution feeder," in Power Engineering Conference (AUPEC), 2013 Australasian
Universities, IEEE, 2013, pp. 1--6.
[25] S. Conti, A. Greco and S. Raiti, "Voltage Sensitivity Analysis in MV Distribution Networks," in
Proceedings of the 6th WSEAS/IASME International Conference on Electric Power Systems, High Voltages,
Electric Machines, 2007.
[26] M. Brenna, E. De Berardinis, L. Delli Carpini, F. Foiadelli, P. Paulon, P. Petroni, G. Sapienza, G.
Scrosati and D. Zaninelli, "Automatic distributed voltage control algorithm in smart grids
applications," Smart Grid, IEEE Transactions on, vol. 4, no. 2, pp. 877-885, 2013.
[27] K. Pantziris, "Voltage support strategies in a rural low voltage network with high photovoltaic
penetration." PhD diss., TU Delft, Delft University of Technology, 2014.

13