Georgios Demetriades
AbstractThis paper proposes a distributed multi-agent cooperative control system for dynamic energy balancing between
storage devices in droop controlled DC microgrids. With the
traditional droop control strategy, line resistances between DC
microgrid energy storage devices and loads will cause unequal
power sharing. The proposed control system modifies the output
power of the droop controlled storage devices so that they reach
a balanced energy level. Once a balanced energy level has been
reached, the cooperative control system maintains equal power
sharing between the storage devices. This ensures that the storage
devices will not prematurely run out of energy, so their full output
power capacity is available to regulate the microgrid voltage.
Simulations have been completed showing that the cooperative
control system is able to maintain voltage regulation in situations
for which the traditional droop control fails.
Index TermsBatteries, DC microgrids, distributed cooperative control, droop control, multi-agent systems.
I. I NTRODUCTION
Distributed renewable generation sources can provide increased network reliability, pollution reduction and facilitate
electrification in remote areas. However, the intermittent nature
of renewable generation sources can reduce network power
quality, stability and availability. Microgrids have gained popularity as a means for increasing the penetration of intermittent
distributed renewable generation sources while maintaining
network power quality [1]. A microgrid is a group of loads,
generation sources and energy storage (ES) devices which can
operate while connected to the main grid, or autonomously in
an islanded mode when disconnected [2].
DC microgrids present a range of benefits over AC microgrids for low voltage distribution systems. Many renewable
generation sources including photovoltaic cells and fuel cells,
as well as ES technologies such as batteries and flywheels, produce DC power and require inverters in an AC microgrid [3].
Also, DC loads such as computers, LED lighting and variable
speed drives are becoming more prevalent. With fewer power
converters network efficiency and reliability are improved [4].
DC networks do not suffer from the skin effect [5] and since
the voltage level of a DC system is stable, a higher RMS voltage level can be used than in an AC system for a given level of
Australasian Universities Power Engineering Conference, AUPEC 2014, Curtin University, Perth, Australia, 28 September 1 October 2014
V A
Battery
+
-
PI Voltage
Controller
PWM
Cooperative
Controller
Local State
Neighbour
States
(1)
0
0
0
Be
0
A
0
B
p
p
A=
0
B1
0
B2
0 Bdc Cd1 Cv1 Bdc Cd3 Adc + Bdc Cd2
T
B = 0 0 B3 Bdc Cd1 Cv2 .
B. Droop Control
The BESS droop control is based on a filtered measurement
of the DC-DC converter output power P ,
c
p, where p = v1 i1 .
(2)
P =
s + c
The V P droop control equation is given by
v = VM G m(P P ).
(3)
Australasian Universities Power Engineering Conference, AUPEC 2014, Curtin University, Perth, Australia, 28 September 1 October 2014
iL
P = Ap P + Bp v1
(5)
v2
= Cv1 P + Cv2 P
Ap = c , Bp = 0
c VH
R1
(6)
0 , Cv1 = m, Cv2 = m.
C. Voltage Controller
A PI voltage controller is used to set the duty cycle of
the PWM DC-DC converter to achieve the voltage reference
set by the droop control. The state equations for the voltage
controller are given by
= v v1
(7)
d = Do + kpv (v v1 ) + kiv .
(8)
iL
= B1 P + B2 v1 + B3 P
(9)
v2
d = Cd1 v
iL
+ Cd2 v1 + Cd3
v2
B1 = m, B2 = 0
Cd1 = kpv , Cd2 = 0 kpv
(10)
E. DC-DC Converter
The DC-DC converter model parameters include the inductance L, inductor parasitic resistance RL , MOSFET onresistance Rdson , low voltage side capacitance CL , high
voltage side capacitance CH , bus connection resistance R1
and battery internal resistance R2 . A small-signal state-space
model for the DC-DC converter is provided in [19],
iL
iL
v1 = Adc v1 + Bdc d .
(13)
v2
v2
III. C OOPERATIVE C ONTROL
In a distributed multi-agent cooperative control system autonomous agents pursue a common goal through communication with their neighbours in a sparse communication network.
In this paper the goal of the cooperative control system is for
the BESS agents to reach a balanced energy state and once it
has been achieved to maintain it through equal power sharing.
The communication network can be described by a directed
graph G = {V, E}, where V = {1, ..., N } is the set of N
nodes, each associated with a BESS, and E is the set of edges
of the graph. An edge (i, j) E if there is a link allowing
information to flow from node i to node j. Let the neighbours
of node i be Ni , where j Ni if (j, i) E. Assume the graph
is strongly connected, so there is a set of edges providing a
path between every pair of nodes. Let the graph adjacency
matrix be
(14)
A = [aij ] RN N ,
1, (j, i) E
aij =
0, otherwise.
The graph degree matrix is given by
D = diag(di ), where di =
aij .
(15)
j=1
0 , B3 = m,
0 , Cd3 = kiv .
D. Battery
The first order lithium ion battery energy model from [20]
is used. To prevent significant deterioration of the batterys
life, the energy level should be kept between 20% and 80% of
the full capacity. Within these bounds the battery is modelled
as an ideal 48V source. The battery energy state equation is
given by
2
vbatt v2 vbatt
.
(11)
E =
R2
The battery small-signal state-space model is given by
iL
batt
E = Be v1 , Be = 0 0 vR
.
2
v2
(12)
Australasian Universities Power Engineering Conference, AUPEC 2014, Curtin University, Perth, Australia, 28 September 1 October 2014
Islanded
Main Grid
Bus 1
Bus 2
Bus 3
Bus 4
380VDC/
48VDC
Transmission Line
Rline
Lline
0.5
30 H
Droop Controller
Pmax
m
VM G
c
20 kW
1.0103
380 V
10 rad/s
Transmission Line
Communication Link
vbatt
Etotal
Emax
Emin
Voltage Controller
kpv
kiv
-1.0106
-1.0105
DC-DC Converter
L
RL
Rdson
R1
R2
CH
CL
6.25
9.00
8.75
50.0
1.00
72.0
7.20
H
m
m
m
m
mF
mF
Initially the loads are evenly distributed in the DC microgrid, and are each fed by their local BESS so the traditional
droop control strategy provides equal power sharing. The
BESS energy levels fall at an equal rate as shown in Fig. 3(a).
The bus 4 BESS reaches the 20% minimum acceptable energy
level of 5kWh after 80 minutes. To prevent deterioration of
the lithium ion battery lifetime a PI controller is used to
reduce the BESS output power to zero. As shown in Fig. 3(b)
in accordance with the droop control strategy the microgrid
bus voltages fall in response to the change in supply power
until power is once again balanced in the microgrid. However,
since the loads are no longer evenly distributed with respect
to the remaining BESS in the microgrid, the load is not
shared equally between them, as shown in Fig. 3(c). After 200
minutes the bus 2 and 3 BESS have also run out energy, and
the bus 3 and 4 voltages fall below the minimum acceptable
limit of 360V.
B. Case B: Cooperative Control
Linear quadratic regulator design was used to design the
cooperative control system state variable feedback control
matrix K. With the cooperative control system the energy
levels of the BESS converge so that none of the BESS reach
the minimum energy level over the four hours of islanded
operation, as shown in Fig. 4(a). Fig. 4(b) shows that the
microgrid bus voltages stay within the limits of 360V to 400V.
As the energy levels of the BESS converge the microgrid load
is equally shared between the BESS, as shown in Fig. 4(c).
The dynamic energy balancing cooperative control system
successfully regulates the microgrid voltage and provides
equal power sharing between the BESS in steady state.
V. C ONCLUSION
In this paper a distributed multi-agent cooperative control
system has been proposed which provides dynamic energy
balancing between storage devices in droop controlled DC
microgrids. By balancing the energy levels of the storage
devices the cooperative control system ensures that none will
prematurely run out of energy, so their full output power
capacity is available to regulate the microgrid voltage. Simulations have been completed demonstrating the improvement
the cooperative control system can provide in terms of voltage
regulation over traditional droop control. Once the storage
devices reach a balanced energy level the proposed control
system maintains equal power sharing between them, which
is not provided by traditional droop control. The cooperative
control system only requires a sparse communication network
providing neighbour to neighbour communication, offering
advantages in terms of robustness, extensibility and flexibility
Australasian Universities Power Engineering Conference, AUPEC 2014, Curtin University, Perth, Australia, 28 September 1 October 2014
20
Bus
Bus
Bus
Bus
15
1
2
3
4
20
10
60
120
Time (min)
180
15
60
120
Time (min)
180
240
400
400
390
390
380
380
Voltage (V)
Voltage (V)
1
2
3
4
10
240
Bus
Bus
Bus
Bus
370
360
350
340
370
360
350
60
120
Time (min)
180
340
240
60
120
Time (min)
180
240
12
10
Output Power (kW)
10
8
6
4
2
0
-5
0
60
120
Time (min)
180
240
over centralised control strategies. The cooperative control system design procedure guarantees system stability and allows
the speed of energy convergence to be traded-off against the
power used for energy balancing.
R EFERENCES
[1] R. H. Lasseter, Smart Distribution: Coupled Microgrids, Proceedings
of the IEEE, vol. 99, no. 6, pp. 10741082, Jun. 2011.
[2] R. Lasseter, MicroGrids, in 2002 IEEE Power Engineering Society
Winter Meeting. Conference Proceedings (Cat. No.02CH37309), vol. 1.
IEEE, 2002, pp. 305308.
60
120
Time (min)
180
240
Australasian Universities Power Engineering Conference, AUPEC 2014, Curtin University, Perth, Australia, 28 September 1 October 2014