Active and Reactive Power Sharing in Inverter Based

Droop-Controlled Microgrids
Ogbonnaya Bassey, Student Member, IEEE, Karen L. Butler-Purry, Fellow, IEEE, and Bo Chen, Member, IEEE

Abstract—Accurate power sharing is essential for the successful
operation of an islanded microgrid with droop-controlled
inverters particularly at times when the total loads served in the
system change significantly like during microgrid black start. In
this paper, two paradigms for active power load sharing were
derived: 1) load sharing by the proportion of active power limit
set by the DC primer mover of DGs and 2) load sharing by the
proportion of optimized reference active power. Further, a
method was derived for approximate sharing of reactive power
by choosing the coupling inductance of the droop-controlled
inverters, which though not as precise as active power sharing,
could be coordinated by a microgrid central controller to ensure
good voltage regulation. The load sharing method was validated
by a simulated black start test case with droop-controlled
inverters and aggregated loads restored in sequence.
Index Terms—Droop control, frequency control, load sharing,
microgrid, reactive power, voltage control
I. INTRODUCTION
A microgrid is a localized interconnection of distributed
energy resources and loads intelligently controlled by a central
controller. Microgrids can work in island or grid-connected
mode. When operating in island mode, the microgrid must
share loads and maintain quality frequency and voltage profile.
The work in this paper focuses on active and reactive power
load sharing, and how to control the frequency and voltage of
an islanded droop-controlled microgrid.
The authors in [1] presented two operation modes in which
islanded microgrids could operate. The first mode is the single
master operation (SMO) mode. This mode has one master
distributed generator (DG) tasked with real-time load balancing
and voltage/frequency control thereby offering a simpler
control. The second mode is the multi-master operation (MMO)
mode. This mode can have more than one master DG tasked
decoupling control used in the bulk grid. Decoupling power
transformation and virtual impedance methods have been
proposed as better alternatives to the use of coupling inductors
[3-5]. However, understanding load sharing in conventional
droop control with physical inductive coupling could help in
developing a better load sharing paradigm for systems with
virtual impedance and power transformation methods.
In our previous work, we have studied black start restoration
by the sequential formation of multiple microgrids operating in
SMO mode [7-9]. In trying to adapt the black start restoration
method for multi-master microgrids, it is necessary to study
active and reactive power sharing in MMO mode. This paper
presents active and reactive power sharing paradigms in
islanded microgrids with conventional droop-controlled
inverters by giving insights on how microgrid power dispatch
conditions decide the choice of droop settings. The outline of
the rest of the paper is as follows. In section II, we present
power transfer equations for an inverter with inductive
coupling. Section III, a review and revision of active load
sharing is proved. In addition, an approximate method for
reactive power sharing through the choice of coupling
inductance for the system is derived in section IV. Section V
and section VI consist of case studies and conclusion
respectively.
II. POWER TRANSFER THROUGH INDUCTIVE COUPLING
Fig. 1 shows the connection of an inverter to a microgrid via
inductive coupling. 𝑉𝑐 = |𝑉𝑐 |∠𝛿 is the voltage at the converter
terminal and 𝑉𝑚 = |𝑉𝑚 |∠0 is the voltage at the point of
common connection (PCC) with the microgrid. 𝐼𝑐 is the current
output from the converter. All quantities are in per unit.
Applying ohms law across the inductor gives the relation
below.
|𝑉𝑐 |∠𝛿−|𝑉𝑚 |∠0 |𝑉𝑐 | |𝑉𝑚 | 𝜋

with providing real-time load sharing and coordinated 𝐼𝑐 = = 𝑒 𝑗(𝛿−𝜋/2) − 𝑒 𝑗(− 2 ) (1)
𝑗𝑋 𝑋 𝑋
voltage/frequency control. This shared responsibility is
typically accomplished by incorporating droop control in the
interfacing inverters [2-5]. The conventional droop control
applied to microgrids tries to mimic the operation of generators
in the bulk grid. In the bulk power system, the transmission
lines have small resistance compared to the inductive reactance
[6], and thus enables decoupling of the active and reactive
power control. Since the distribution lines in microgrids are
significantly resistive, an inductive coupling at the droop-
controlled inverter output can help in realizing similar Fig. 1. Inductive Coupling for Control of Power Transfer.

O. Bassey and K. L. Butler-Purry are with the Department of Electrical

and Computer Engineering, Texas A&M University. B. Chen is with the
Energy Systems Division, Argonne National Laboratory.
Accepted to the IEEE PES GM 2019. © 2019 IEEE. Personal use of this
material is permitted. Permission from IEEE must be obtained for all other
users, including reprinting/ republishing this material for advertising or
promotional purposes, creating new collective works for resale or
redistribution to servers or lists, or reuse of any copyrighted components of
this work in other works.
 𝜋
|𝑉𝑐 |2 |𝑉𝑐 ||𝑉𝑚 | In the following sections III A and III B, we present two
𝑆 = 𝑉𝑐 𝐼𝑐 ∗ = 𝑒 𝑗𝜋/2 − 𝑒 𝑗(𝛿+ 2 ) (2)
𝑋 𝑋 active power sharing paradigms. The paradigm in section A was
|𝑉𝑐 ||𝑉𝑚| |𝑉𝑐 ||𝑉𝑚 | derived in [10, 11], however, we present an alternative
𝑃 = 𝑅𝑒(𝑆) = 𝑠𝑖𝑛𝛿, 𝑃𝑚𝑎𝑥 = (3) perspective and proof of the theorem. In section B, we extend
𝑋 𝑋
|𝑉𝑐 | the theorem to load sharing in a microgrid with optimized
𝑄 = 𝐼𝑚(𝑆) = (|𝑉𝑐 | − |𝑉𝑚 |𝑐𝑜𝑠𝛿) (4) reference power.
𝑋

Since 𝛿 is typically small, 𝑠𝑖𝑛𝛿 ≈ 𝛿, 𝑐𝑜𝑠𝛿 ≈ 1, then (3) and A. Load Sharing Based on Active Power Limit
(4) can be approximated as:
Assuming the active power constraint is placed by the DC
|𝑉𝑐 ||𝑉𝑚 | |𝑉𝑐 |
𝑃≈ 𝛿, 𝑄 ≈ (|𝑉𝑐 | − |𝑉𝑚 |) (5) prime mover of each inverter, it can be formulated as,
𝑋 𝑋

𝑃𝑚𝑎𝑥 is the maximum power when 𝛿 = 90° and is called 0 ≤ 𝑃𝑖 ≤ 𝑃𝑖 , 𝑖 ∈ {1,2, … , 𝑛} (10)
the static transmission capacity [6]. Typically, 𝛿 is desired to be The droop coefficients are said to be selected proportionally
kept small far below 90° such that the power transferred is far based on the active power constraint imposed by the DC prime
less than the transmission capacity. The coupling inductance is mover if for each 𝑖𝜖{1,2, … , 𝑛} [2],
inversely proportional to the transmission capacity and has to
be carefully chosen such that it is high enough to turn the 𝑛𝑓,𝑖 𝑃𝑖 = 𝑘𝑝 , 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 ∀ 𝑖 ∈ {1,2, … , 𝑛} (11)
network to be mostly inductive and low enough not to cause
excessive voltage drop and to keep the 𝛿 small. Where 𝑃𝑖 is the maximum allowable active power from the
𝑖 𝑡ℎ inverter’s DC prime mover and 𝑘𝑝 is a constant. The
III. ACTIVE POWER SHARING reference frequency, 𝑓 ∗ , is assumed to be the same for all the
The frequency droop method of sharing active power is a droop-controlled inverters in the system. The MGCC could be
precise power sharing scheme. This is because, though the operated in such a way to realize 𝑓 ∗ as the system’s frequency
frequency may differ at various buses in the system during and as such, 𝑃𝑖∗ and ∑𝑛𝑖=1 𝑃𝑖∗ , the reference power and its
transients, the frequency would usually settle to a common summation should be calculated to match closely with the
value afterward and could be used for local control by each of active power output of each inverter 𝑃𝑖 and the estimated total
𝑒𝑠𝑡
the droop controlled DGs without the need for communication. active power of load, 𝑃𝐿,𝑡𝑜𝑡𝑎𝑙 , respectively,
The frequency droop selection used in this paper is an extension ∑𝑛𝑖=1 𝑃𝑖∗ = 𝑃𝐿,𝑡𝑜𝑡𝑎𝑙
𝑒𝑠𝑡
, 𝑖 ∈ {1,2, … , 𝑛} (12)
of the derivations in [10, 11]. For 𝑛 droop inverter
interconnection, let the per unit active power output of each However, due to some uncertainties in load estimation,
inverter be 𝑃𝑖 , 𝑖 ∈ {1,2, … , 𝑛}, then clearly 𝑃𝑖 should satisfy this time-varying load, and losses in the system, there will usually
inequality: be a mismatch and hence a deviation from the reference
frequency. It is desired that any mismatch between ∑𝑛𝑖=1 𝑃𝑖∗ and
𝑃𝑖 ≤ 𝑃𝑚𝑎𝑥,𝑖 , ∀ 𝑖 ∈ {1,2, … , 𝑛} (6)
total active power of loads be shared proportionately according
Where 𝑃𝑚𝑎𝑥,𝑖 is the static transmission capacity defined for to their DC prime mover active power rating.
each inverter based on (3) rewritten with subscript 𝑖 as: 1) Theorem 1 (Proportional Power Sharing Based on
|𝑉𝑐 |𝑖 |𝑉𝑚 |𝑖 Active Power Limit):
𝑃𝑚𝑎𝑥,𝑖 = , 𝑖 ∈ {1,2, … , 𝑛} (7)
𝑋𝑖 Consider statements (a) and (b) below.
|𝑉𝑐 |𝑖 is the per unit RMS voltage at the 𝑖 𝑡ℎ inverter terminal, a) Proportional reference active power
𝑋𝑖 is the per unit coupling inductive reactance and |𝑉𝑚 |𝑖 is the
per unit RMS voltage at the point of connection to the 𝑃𝑖∗ = 𝑘𝑃𝑖 , ∀ 𝑖 ∈ {1,2, … , 𝑛} (13)
microgrid. Equation (6) ensures that the power at each inverter 𝑒𝑠𝑡
⁄∑𝑛𝑖=1 𝑃𝑖
𝑘= 𝑃𝐿,𝑡𝑜𝑡𝑎𝑙 (14)
is realistic.
Each droop-controlled inverter 𝑖 controls its instantaneous Where 𝑘 (0 ≤ 𝑘 ≤ 1) is a constant and same for each droop
frequency, 𝑓𝑖 , at its output by the following relations. DG. k is clearly reasonable in this range to ensure that reference
power is chosen within the active power constraints.
𝑓𝑖 = 𝑓 ∗ − 𝑛𝑓,𝑖 (𝑃𝑖 − 𝑃𝑖∗ ), 𝑖 ∈ {1,2, … , 𝑛} (8)
b) Proportional droop coefficients of (11) are satisfied
Where 𝑓 ∗ is the reference frequency in Hz, 𝑃𝑖∗ is the [2]
reference active power in per unit, 𝑃𝑖 is the active power output If (a) and (b) are true, then the loads are shared
in per unit, 𝑛𝑓,𝑖 is the frequency droop co-efficient of the 𝑖 𝑡ℎ proportionately in accordance with the rating of the DC prime
inverter in Hz per unit power. Define the frequency deviation mover, that is,
by,
𝑃𝑖 ⁄𝑃𝑖 = 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡, ∀ 𝑖 ∈ {1,2, … , 𝑛} (15)
𝛥𝑓𝑖 = 𝑓𝑖 − 𝑓 ∗ , 𝑖 ∈ {1,2, … , 𝑛} (9)
Proof:
For an exponentially stable solution, the frequency
deviation will settle to the same constant value for every node The proof follows from rearranging (8) to,
in the system, thus, 𝛥𝑓𝑖 = 𝛥𝑓.
 𝛥𝑓−𝑛𝑓,𝑖 𝑃𝑖∗ inverter. 𝑛𝑣,𝑖 is the voltage droop co-efficient in per unit
= 𝑃𝑖 (16)
−𝑛𝑓,𝑖 reactive power.
𝛥𝑓−𝑛𝑓,𝑖 𝑃𝑖∗ Reactive power sharing is not as precise as active power
𝑃𝑖 ⁄𝑃𝑖 = (17)
−𝑛𝑓,𝑖 𝑃𝑖 sharing. This is because, unlike that of frequency which
Substituting (11) and (13) into (17) gives, eventually settles to a common value for an exponentially stable
system, the system does not have uniform voltages at the nodes.
𝛥𝑓−𝑘𝑘𝑝 Several papers have proposed ways for accurate reactive power
𝑃𝑖 ⁄𝑃𝑖 = , 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡, ∀ 𝑖 ∈ {1,2, … , 𝑛} (18)
−𝑘𝑝 sharing like in [12] where the authors proposed novel droop
method which adaptively controls the reference voltage of the
B. Load Sharing Based on Optimized Reference Active inverters and in [13] which argued for increased droop gain. In
Power the following analysis, we present a simple way to coordinate
When the MGCC adds some optimization layer (e.g voltage regulation and accurate reactive power sharing by
economic, emission optimization, etc.) to calculate optimal routine coordination of the conventional voltage droop
reference power for each droop DG, then the reference power setpoints and sizing of inductive coupling by the MGCC.
expression in (13) may not have the same k for each droop
inverter. In such a situation, it may be desired that the power B. Voltage Droop Reference Calculation
mismatch be shared proportionately to the optimal reference As stated in section III, active power is constrained by the
active power calculated by the MGCC. Theorem 2 is motivated capacity of the DC prime mover. Reactive power, on the hand,
by this situation. could be seen as being constrained by the VI rating of each
1) Theorem 2 (Proportional Power Sharing Based on inverter, 𝑆𝑖 , and the DC prime mover rating, 𝑃𝑖 , by the
Optimized Reference Active Power): following relation,
a) Optimized reference active power definition: 0 ≤ 𝑄𝑖 ≤ 𝑄𝑖 (24)
𝑃𝑖∗ = 𝑘𝑖 𝑃𝑖 , ∀ 𝑖 ∈ {1,2, … , 𝑛} (19)
2 2
Where 𝑘𝑖 (0 ≤ 𝑘𝑖 ≤ 1) is a constant and not necessarily 𝑄𝑖 = √𝑆𝑖 − 𝑃𝑖 , 𝑖 ∈ {1,2, … , 𝑛} (25)
the same values for each droop inverter. Equations (24) and (25) ensure that the VI rating of each
b) Proportional droop coefficients based on optimized inverter is not exceeded. There could be several paradigms for
reference active power: determining the reference reactive power, 𝑄𝑖∗ , for each inverter.
𝑛𝑓,𝑖 𝑃𝑖∗ = 𝑘𝑣 , 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 ∀ 𝑖 ∈ {1,2, … , 𝑛} (20) It could be chosen proportionately to 𝑄𝑖 or based on some
optimal network flow considerations. Here we consider the first
If (a) and (b) are true, then the loads are shared case for inductive type load (𝑄𝑖∗ , 𝑄𝑖 > 0), of which the second
proportionally in accordance with the optimized reference case is just a little modification of the first case. Thus,
active power, that is,
𝑄𝑖∗ = 𝑞𝑄𝑖 (26)
𝑃𝑖 ⁄𝑃𝑖∗ = 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡, ∀ 𝑖 ∈ {1,2, … , 𝑛} (21)
𝑒𝑠𝑡
𝑞 = 𝑄𝐿,𝑡𝑜𝑡𝑎𝑙 ⁄∑𝑛𝑖=1 𝑄𝑖 , 𝑖 ∈ {1,2, … , 𝑛} (27)
Proof:
𝑒𝑠𝑡
Where 𝑞 is a constant and 𝑄𝐿,𝑡𝑜𝑡𝑎𝑙
is the total estimated
Similarly, the proof follows from rearranging (8) to get (16)
reactive power to be shared. Assuming small power angle for
and then dividing through by 𝑃𝑖∗ ,
the active power output from each inverter, (5) can be rewritten
𝛥𝑓−𝑛𝑓,𝑖 𝑃𝑖∗ 𝛥𝑓−𝑘𝑣 for the reference reactive power output of each inverter as,
𝑃𝑖 ⁄𝑃𝑖∗ = = , 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 (22)
−𝑛𝑓,𝑖 𝑃𝑖∗ −𝑘𝑣
|𝑉𝑐 |∗𝑖
𝑄𝑖∗ ≈ (|𝑉𝑐 |∗𝑖 − |𝑉𝑚 |𝑖 ), 𝑖 ∈ {1,2, … , 𝑛} (28)
Assuming little power mismatch, the system operating point 𝑋𝑖
may deviate from the optimal dispatch point for a general non- Where 𝑋𝑖 is the inductive reactance in per unit of the
linear relationship between the active power and objective coupling inductor, |𝑉𝑚 |𝑖 is the per unit RMS voltage at the point
function, however, it would give a reasonably close to optimal of connection to the microgrid network for each inverter. One
dispatch for small active and reference power mismatch. of the goal is to ensure that we have close to 1 per unit at the
point of connection to the network. Assuming |𝑉𝑚 |𝑖 = 1𝑝𝑢,
IV. REACTIVE POWER SHARING then (28) becomes,
A. Voltage Droop Basics |𝑉𝑐 |∗𝑖
2
|𝑉𝑐 |∗𝑖
Reactive power sharing is typically accomplished through − − 𝑄𝑖∗ ≈ 0 (29)
𝑋𝑖 𝑋𝑖
the voltage droop method represented mathematically as [2],
Solving for |𝑉𝑐 |∗𝑖 in (29) and discarding the negative non-
|𝑉𝑐 |𝑖 = |𝑉𝑐 |∗𝑖 − 𝑛𝑣,𝑖 (𝑄𝑖 − 𝑄𝑖∗ ), 𝑖 ∈ {1,2, … , 𝑛} (23) physical result,
1 1
Where |𝑉𝑐 |𝑖 is the output RMS voltage, |𝑉𝑐 |∗𝑖
is the reference |𝑉𝑐 |∗𝑖 = + √1 + 4𝑄𝑖∗ 𝑋𝑖 (30)
2 2
output RMS voltage, 𝑄𝑖 is the output reactive power, 𝑄𝑖∗ is the
reference output reactive power, all in per unit of the 𝑖 𝑡ℎ
 By choice of 𝑋𝑖 , 0 < 4𝑄𝑖∗ 𝑋𝑖 ≪ 1, and therefore, we can
apply binomial approximation to (30) to get,
1 1 1 There are three DGs in the test system: DG1, DG2, and
|𝑉𝑐 |∗𝑖 ≈ + (1 + ∗ 4𝑄𝑖∗ 𝑋𝑖 ) = 1 + 𝑄𝑖∗ 𝑋𝑖
2 2 2 DG3. Each is modeled as a droop-controlled inverter interfaced
Equations (26), (27), and (31) can be used by the MGCC to
calculate 𝑄𝑖∗ and |𝑉𝑐 |∗𝑖 for each of the inverter’s voltage droop
controller. If there is a good estimation of the total reactive
power in the system, the inverter’s output voltage, |𝑉𝑐 |𝑖 , and
reactive power, 𝑄𝑖 , will closely match the calculated reference
values |𝑉𝑐 |∗𝑖 and 𝑄𝑖∗ respectively, and |𝑉𝑚 |𝑖 ≈ 1.
C. Approximate Reactive Power Sharing by Choice of
Inductive Coupling
Assuming that there is a small deviation in total reactive
power served from the estimated total reactive power, this
deviation is usually manifested as small deviations in bus
voltages. Define the new |𝑉𝑚 |𝑖 = 1 + ∆|𝑉𝑚 |𝑖 , then the reactive
power output can be written as,
|𝑉𝑐 |𝑖
𝑄𝑖 = (|𝑉𝑐 |𝑖 − 1 − ∆|𝑉𝑚 |𝑖 )
𝑋𝑖
simplicity, the loads are modeled as constant impedance loads
and have values as shown in Table 1 at nominal voltage.
(31)
DG. The droop control inverter model was developed in
PSCAD and made use of the model for the grid supporting
inverter as found in [3]. Table 2 shows the major parameters of
each DG. Each of the inverter interfaces for the DG outputs a
base line to line voltage of 0.48KV and then the matching
transformer transforms the voltage to the distribution network
level of 4.16KV L-L. The base voltage on the DG side is
0.48KV L-L and base three phase power is 1 MVA. The
coupling inductive reactance was chosen to be in inverse
relation to the 𝑄𝑖 as earlier analyzed in section IV.
B. Case Study Results
In the case study simulated in PSCAD, there was five
sequence of events with each lasting 3 seconds. The MGCC
(32)

For a relatively small voltage droop coefficient, then |𝑉𝑐 |𝑖 ≈

|𝑉𝑐 |∗𝑖 , then,
∆|𝑉𝑚 |𝑖 |𝑉𝑐 |∗𝑖 ∆|𝑉𝑚 |𝑖 ∆|𝑉𝑚 |𝑖
𝑄𝑖 − 𝑄𝑖∗ ≈ − ≈− (1 + 𝑄𝑖∗ 𝑋𝑖 ) ≈ −
𝑋𝑖 𝑋𝑖 𝑋𝑖

(33)
We assume that the coupling inductor was chosen to
dominate over the line impedance and thus voltage drops Fig. 2. Modified 4 bus IEEE test feeder.
significantly only across the coupling inductor, therefore, the was assumed to adjust the droop setpoint every 6 seconds
drop across the network can be assumed relatively small. This starting from the beginning of the simulation. The results are
leads to the approximation, shown in Fig. 3 and Fig. 4.
∆|𝑉𝑚 |𝑖 = ∆|𝑉𝑚 |𝑗 = ∆|𝑉𝑚 |, ∀ 𝑖, 𝑗 ∈ {1,2, … , 𝑛} (34) Sequence 1: The 3 DGs were turned on together with load
∆|𝑉𝑚 | 1, while the other two loads were off. Loads were shared
∆𝑄𝑖 = 𝑄𝑖 − 𝑄𝑖∗ ≈− (35)
𝑋𝑖 according to the DC prime mover ratings (𝑃𝑖 , see Theorem 1)
1 1 in the ratio of 2:1:1 for DG 1, 2, and 3 respectively. Reactive
|∆𝑄𝑖 |: |∆𝑄𝑗 | ≈ : , ∀ 𝑖, 𝑗 ∈ {1,2, … , 𝑛} (36) power was shared in the same ratio. Reference points for the
𝑋𝑖 𝑋𝑗
droop controllers were calculated as earlier discussed and sent
1 1
: = 𝑄𝑖 : 𝑄𝑗 , ∀ 𝑖, 𝑗 ∈ {1,2, … , 𝑛} (37) to the DGs by the MGCC.
𝑋𝑖 𝑋𝑗
Sequence 2: At 3 seconds simulation time, load 2 was
The implication of (36) is that deviations from the reference
turned on. There is no adjustment by the MGCC in the interval
reactive power can be shared by inverse proportion of the
spanning sequence 2. The additional power was shared
coupling inductive reactance. Other reactive power sharing
according to Theorem 1 and section IV C.
paradigm may require varying 𝑋𝑖 according to some optimal
reactive power reference. In such a case, an inverter control
TABLE I
with virtual impedance would offer more flexibility than the AGGREGATE LOAD VALUES
physical coupling inductance used here. L1 L2 L3
P (MW) 0.45 0.15 0.30
V. CASE STUDIES Q(MVAR) 0.3 0.075 0.120

A. Description of Test System TABLE II

AGGREGATE LOAD VALUES
The test system used to validate the load sharing capability DG1 DG2 DG3
of the methods presented in sections III and IV is based on the 𝑷𝐢 (MW) 0.6 0.3 0.3
IEEE 4 bus test feeder [14] and the one line diagram is shown 0.8 0.4 0.4
𝐒𝐢 (MVA)
in Fig. 2. 0.5292 0.2646 0.2646
𝟐 𝟐
𝐐𝐢=√𝐒𝐢 − 𝐏𝐢 (MVAR)
The modified system is made up of three aggregate loads,
L1, L2, and L3, and each is assumed to be switchable. For 𝐗 𝐢 (PU) 0.217 0.434 0.434
 (a)
Fig. 3. Plot of (a) Active Power (b) Reactive Power (c) Droop Frequency for DG 1, 2 & 3
(a)
Fig. 4. Plot of Inverter Output Voltage and Voltage at Primary Side of Matching Transformer (PCC) for (a) DG 1 (b) DG 2 (c) DG 3
Sequence 3: At 6 seconds, the MGCC sent optimal setpoints
to the DGs. It was assumed that the MGCC does some internal
calculation to determine that equal share of load by the DGs
(1:1:1) optimizes the system according to Theorem 2. The ratio
of reactive power sharing is assumed to remain the same. Notice
the frequency and voltage restoration action after setpoint
correction at 6 seconds in Fig. 3 (c) and Fig. 4 respectively.
Sequence 4: At 9 seconds, load 3 was picked up with the
sharing ratio of sequence 3.
Sequence 5: At 12 seconds, the MGCC recalculated the
setpoint to adjust the system frequency and voltage.
Analyzing the plots in Figs. 3 (a) and (b), we notice that the
active power was shared accurately, however, the reactive
power sharing, set at a constant ratio of 2:1:1 did not follow this
ratio accurately but still shared close to this ratio.
VI. CONCLUSION
In this paper, we have shown how to realize precise sharing
of active power of parallel DGs in islanded microgrid using two
paradigms: load sharing by the proportion of active power limit
and load sharing by the proportion of optimized reference
power. We also presented a simple way to realize an
approximate sharing of reactive power through the choice of
coupling inductance, which though not as precise as that of
active power, tracked sharing ratio fairly.
REFERENCES
[1] J. A. P. Lopes, C. L. Moreira, and A. G. Madureira, "Defining control
strategies for MicroGrids islanded operation," IEEE Transactions on
Power Systems, vol. 21, no. 2, pp. 916-924, 2006.
(b) (c)
(b) (c)
[2] M. C. Chandorkar, D. M. Divan, and R. Adapa, "Control of parallel
connected inverters in standalone AC supply systems," IEEE
Transactions on Industry Applications, vol. 29, no. 1, pp. 136-143, 1993.
[3] J. Rocabert, A. Luna, F. Blaabjerg, and P. Rodríguez, "Control of Power
Converters in AC Microgrids," IEEE Transactions on Power
Electronics, vol. 27, no. 11, pp. 4734-4749, 2012.
[4] K. D. Brabandere, B. Bolsens, J. V. d. Keybus, A. Woyte, J. Driesen, and
R. Belmans, "A Voltage and Frequency Droop Control Method for
Parallel Inverters," IEEE Transactions on Power Electronics, vol. 22, no.
4, pp. 1107-1115, 2007.
[5] W. Guo and L. Mu, "Control principles of micro-source inverters used
in microgrid," Protection and Control of Modern Power Systems, vol. 1,
no. 1, p. 5, 2016.
[6] H. Saadat, Power System Analysis McGraw-Hill Series in Electrical
Computer Engineering. 1999.
[7] B. Chen, C. Chen, J. Wang, and K. L. Butler-Purry, "Sequential Service
Restoration for Unbalanced Distribution Systems and Microgrids," IEEE
Transactions on Power Systems, vol. PP, no. 99, pp. 1-1, 2017.
[8] O. Bassey, B. Chen, and K. L. Butler-Purry, "Sequential service
restoration in distribution systems and microgrids integrating frequency
response and varying switching interval," in 2018 IEEE Texas Power
and Energy Conference (TPEC), 2018, pp. 1-6.
[9] B. Chen, C. Chen, J. Wang, and K. L. Butler-Purry, "Multi-Time Step
Service Restoration for Advanced Distribution Systems and
Microgrids," IEEE Transactions on Smart Grid, vol. PP, no. 99, pp. 1-1,
2017.
[10] J. W. Simpson-Porco, F. Dörfler, and F. Bullo, "Droop-controlled
inverters are Kuramoto oscillators," in IFAC Workshop on Distributed
Estimation and Control in Networked Systems, 2012, pp. 264-269.
[11] J. W. Simpson-Porco, F. Dörfler, and F. Bullo, "Synchronization and
power sharing for droop-controlled inverters in islanded microgrids,"
Automatica, vol. 49, no. 9, pp. 2603-2611, 2013/09/01/ 2013.
[12] K. Jung-Won, C. Hang-Seok, and C. Bo Hyung, "A novel droop method
for converter parallel operation," IEEE Transactions on Power
Electronics, vol. 17, no. 1, pp. 25-32, 2002.
[13] A. Haddadi, A. Shojaei, and B. Boulet, "Enabling high droop gain for
improvement of reactive power sharing accuracy in an electronically-
interfaced autonomous microgrid," in 2011 IEEE Energy Conversion
Congress and Exposition, 2011, pp. 673-679.
[14] PES Test Feeder [Online]. Available: http://sites.ieee.org/pes-
testfeeders/resources/