AN IMPROVED ADAPTIVE NONLINEAR DROOP

CONTROL METHOD IN DC MICROGRID

*
Yaoyao Zhang, Maodong Tang, Xiaohui Qu , Wu Chen
Southeast University, Nanjing, China
*[email protected]

Keywords: DC MICROGRID, LOAD SHARING, VOLTAGE REGULATION, IMPROVED

ADAPTIVE NONLINEAR DROOP CONTROL

Abstract impedances to control the voltage within the allowable

fluctuation range of the bus voltage [10]-[14]. However, the
It is well know that the load sharing among several paralleled traditional droop control exists a design tradeoff between
distributed sources can be achieved by the conventional load sharing and voltage regulation, which may result in
droop control in the DC microgrid. However, there is a overloading for paralleled converters. In [15], a translational
design tradeoff in the conventional droop control method droop curve is added into the traditional droop control to
with respect to load sharing and voltage regulation. To stabilize the DC bus voltage and to improve the voltage
improve both two performances, this paper proposes an regulation. However, this method is more complicated to
improved adaptive nonlinear droop control by the means of control and the output voltage cannot be automatically
dividing the load condition into two categories: the heavy tracked when the load changes. In [16], a nonlinear droop
load and the light load. At the light loading range, a good control method is proposed to balance the tradeoff between
voltage regulation is prioritized due to no overload voltage regulation and load sharing. However, the voltage
phenomenon. At the heavy loads, the load sharing is regulation is still not good at heavy loads.
prioritized since a small load sharing error may cause the risk Solar Array Wind Turbine Plug-in Vehicle
of overload. Moreover, the good load sharing performance Grid

also improves the voltage regulation at the heavy loads,

which avoids the additional compensation circuits. This paper
also gives the design method to define the boundary between Grid-Interfacing
light load and high load. Then a curve fitting method is used Converter

to permit an automatic track during different load ranges.

Finally, two paralleled DC converters are built and the DC Bus ĊĊ
experimental results verify the theoretical analysis well.

1. Introduction
DC Loads DC Loads
Recently, the renewable sources are becoming more and
more popular. Meanwhile, DC loads are more common, such Energy Storage
S

as LED lighting systems, DC data centres, electric vehicle

Fig. 1 A typical DC microgrid.
charging stations, and so on. With such sources and loads, the
DC bus is more simple and efficient [1]. Fig 1 shows a To improve the performances of both voltage regulation and
typical configuration of a DC microgrid including the solar load sharing, this paper proposes an improved adaptive
array, wind turbine, and energy storage [2]-[3]. nonlinear droop control method in DC microgrid. This
When different kinds of sources are paralleled to the DC grid method divides the load into two categories: light loads and
via the interface converters, the voltage regulation and load heavy loads. At the light loading range, a good voltage
current sharing among sources may be not good due to regulation is prioritized due to no overload phenomenon. At
different output impedances of converters and the line the heavy loads, the load sharing is prioritized since a small
impedances to grid. To permit a good voltage regulation and load sharing error may cause the risk of overload. Moreover,
load current sharing among different converters, there are the good load sharing performance also improve the voltage
some existing methods, including the centralized and regulation at the heavy loads, which avoids the additional
distributive methods The centralized method usually uses a compensation circuits. This paper also gives the design
central controller to optimize the control of the DC microgrid, method to define the light load and high load. Then a curve
based on the information acquisition, prediction, and analysis fitting method is used to permit an automatic track during
[4]-[5]. But when the central controller fails or stops working, different load ranges. Finally, two paralleled DC converters
the whole system cannot work normally. The distributive are built and the experimental results verify the theoretical
method can be the V-I droop control using the local analysis well.
information to avoid the use of communication lines [6]-[9]. This paper is organized as follows: Section 2 analyses the
The traditional droop control scheme constructs virtual conventional droop control, sectional droop control and the

Authorized licensed use limited to: University of Exeter. Downloaded on July 15,2020 at 15:30:54 UTC from IEEE Xplore. Restrictions apply.
proposed improved adaptive nonlinear control. The design Therefore, it is hard to satisfy the requirements of high
and implementation of this proposed scheme is described in voltage regulation and low current sharing error.
Section 3. Section 4 shows the experiment results to verify 14

Normalized I or V (%)
voltage regualtion
the theoretical analysis. Section 5 concludes this paper. 12 current sharing error
10
2. Droop control analysis
8
In a DC microgrid, multiple sources are connected in parallel 6
to share the power, as shown in Fig. 1. Several converters in
4
parallel may result in uneven distribution of output current
and power. To simplify the analysis, this paper establishes a 2
simple dc microgrid system consisting of two parallel-
2 30 1 4 5
connected converters as shown in Fig. 2, where V1 and V2 are Droop Gain (Ω)
equivalent voltage of the converter 1 and converter 2. RL1 and Fig. 3 Current sharing error and voltage regulation versus
RL2 are the cable resistances between converters output droop gain.
terminal and load terminal.
2.2 Sectional droop control method
Io1 RL1 RL2 Io2
From the above analysis, it can be seen that there is a design
Io tradeoff between current sharing and voltage regulation in the
Rd1 Rd2 conventional droop control. If a predesigned voltage
Vdc1 Ro VBus Vdc2 regulation is given, the current sharing error can be improved
V1 V2 by optimizing the droop curves. To solve the issue, the
loading range can be divided into heavy loads and light loads.
Without the overload issue, the voltage regulation should be
Fig. 2 Simplified DC-MG with two converters in parallel. prioritized in light load regions. In heavy load regions,
2.1 Conventional droop control method current sharing accuracy is prioritized because a small current
sharing error may result in the overload in converters. In
The conventional droop control can be mathematically
addition, the load boundary between the light load region and
expressed as
heavy load region is initially set at about 0.5 p.u.
Vdci  VN There are many types of optimization curves from no load
Vdci VN  Rdi I oiˈ where Rdi (1)
I oi max operation to full load operation. A droop curve between two
Here, Vdci is the reference voltage of the ith converter, VN is points can be expressed by the following equation [17]:
the nominal voltage, Rdi is the droop gain or virtual Vdci VN  ki u ( I oi ) n ˈ
impedance, Ioi is output current of the ith converter, and Iimax (Vdci  VN ) (8)
is the maximal output current. where ki
According to KVL, we have ( I oi max ) n
Vdc1  I o1 RL1  I o Ro 0 (2) Where, n is the time of different droop curve. From (8), the
equivalent droop resistance is given by
Vdc 2  Io 2 RL 2  Io Ro 0 (3)
Obviously, Io is the load current, i.e., Io = Io1 + Io2. According dVdci
Rdi n u ki u ( I oi )n 1 (9)
to KCL, output currents of converters are given by dI oi
Ro (Vdc1  Vdc 2 )  Vdc1 RL 2 From (8) and (9), various droop curves and equivalent droop
I o1 (4)
RL1 RL 2  RL1 Ro  RL 2 Ro resistances are plotted for different value of n, as shown in
Fig. 4 and Fig. 5. The nominal voltage (VN) is 48 V, voltage
Ro (Vdc 2  Vdc1 )  Vdc 2 RL1
Io2 (5) deviation is set as 5% (46.8V ≤ VN ≤ 49.2V) and the
RL1 RL 2  RL1 Ro  RL 2 Ro maximum output current (Ioimax) is 6 A in Fig. 4. From Fig. 5,
The cable resistances RL1 and RL2 may not be equal. The it can be seen that the equivalent droop resistances of the
difference of the cable resistance or the mismatch in terminal droop curve with n = 2, 3, 4, 5 is close to 0 in the light load
voltages results in a current-sharing error (ΔI) and a voltage conditions, which will result in a very poor current sharing
deviation (ΔV), which are defined as follows: accuracy. In the heavy load conditions, the droop curve with
I o1  I o 2 n = 3 intersects the droop curve with n = 1 at 0.57 p.u., which
'I u100% (6) is close to 0.5 p.u. compared with other curves. Meanwhile,
Io the droop curve with n = 3 offers more droop gain than the
VN  Vdci droop curve with n = 2.
'V u100% (7) As shown in Fig. 5, it can be seen that the droop values of the
VN
non-linear curves (n = 2, 3, 4, 5) are close to zero under light-
From (6) and (7), assuming that Rd1 equals Rd2 and Vdc1
loading conditions, which deteriorates the current sharing
equals Vdc2, the voltage regulation, current sharing error and
accuracy. In summary, if the value of n is small, good current
droop gain can be drawn as shown in Fig. 3 [16]. From Fig. 3,
sharing is achieved for light-load conditions, but the current
it can be seen the voltage regulation deteriorates as the droop
sharing is poor for heavy-load conditions.
gain increases. However, the current sharing error decreases.

Authorized licensed use limited to: University of Exeter. Downloaded on July 15,2020 at 15:30:54 UTC from IEEE Xplore. Restrictions apply.
 method, a better voltage regulation is provided by the
48 sectional droop control method in the light load conditions.
However, the sectional droop control method performs poor
current sharing near the intersection point, where it may
Voltage(V)

47.5 occur out of control. When converters reach 0.57 p.u., one
n=1 converter may operate with the linear droop curve (n = 1), the
n=2 other converter works with the nonlinear droop control curve
47 n=3
n=4 (n = 3).
n=5 48.5
Droop curve (n=1)
Droop curve (n=3)
0.25 0
0.5 0.75 1
Current(p.u.) Sectional droop curve
48
Fig. 4 Output voltage versus output current.

Voltage (V)
1
47.5
n=1
Droop Gain(Ω )

n=2
n=3
n=4 47
0.5 n=5

46.5
0 0.25 0.5 Is 0.75 6
Current (p.u.)
0 0.5 1 Fig. 6 Output voltage versus output current for sectional
Current (p.u.) droop control method.
2.3 Improved adaptive droop control method
Fig. 5 Droop gain curves.
In the previous section, it is seen that the sectional droop
To overcome the drawback of the nonlinear method control method has a better performance at the light-load
mentioned above, a combination of linear droop curve with n conditions and heavy-load conditions. However, the
= 1 and the droop curve with n = 3 is proposed. Droop performance of the sectional droop control method is not
characteristics for the sectional droop control are shown in good in the vicinity of 0.57 p.u. To overcome this problem, a
Fig. 6. polynomial droop control method is used by fitting the linear
From (8) and (9), the sectional droop control method are droop control curve (n = 1) at light load regions and the
given as follows: nonlinear droop control curve (n = 3) at the heavy load
­V  p1 I oi I oi  [0, I s ] regions in the least squares method. However, the polynomial
Vdci ® N
¯VN  p2 I oi I oi  [ I s , I N ]
3 droop control method has a large voltage deviation which
exceeds the normal fluctuation range of the bus voltage
Vdci  VN deviation heavy load conditions. So the polynomial droop
Rdi _ L p1 (10)
I oi max control needs compensation to ensure the voltage deviation
(Vdci  VN ) I oi2 within the allowance range, which makes the control
Rdi _ H 3p2 I oi2 =3 u complicated.
( I oi max )3
In this section, an improved adaptive droop control method is
Here, Rdi_L and Rdi_H are the droop constants (in Ohms) in the proposed. This method takes the line resistances and fits the
light load region and heavy load region, respectively. Is is the line resistances and sectional droop control method through
boundary current value. When Rdi_L and Rdi_H are equalˈIs least squares. The improved droop control equation can be
can be achieved. given as follows:
If there are two converters p and q, the ideal current sharing Vdci VN  n1 Ioi  n3 I oi3  n5 I oi5 (13)
ratio between the converters is given by:
In (11), the value of ni (i = 1, 3, and 5) are obtained by the
I p Pp
(11) least square curve fitting method. The value of ni (i = 1, 3,
I q Pq and 5) are obtained by (12) to ensure a proportional load
For the sectional droop control method, droop coefficients for sharing.
converter p and converter q are designed as follows: From (13), the equivalent droop gain of the improved
3 adaptive nonlinear droop control method is given by
§ Pq ·
Rdp _ L Pq p2 p
¨ ¸ ˈ (12) dVdci
Rdq _ L © PP ¹ Pp p2 q Rdi (14)
dI oi
Here, p2p and p2q are droop coefficients for converter p and To ensure a stable system, the virtual resistances of the
converter q when n is equal to 3. improved adaptive droop control method must be less than
From Fig. 6 and (12), it can be seen that the sectional droop line resistances.
control method offers a better current sharing for heavy load
conditions compared with the linear droop control method. In
comparison to the conventional nonlinear droop control

Authorized licensed use limited to: University of Exeter. Downloaded on July 15,2020 at 15:30:54 UTC from IEEE Xplore. Restrictions apply.
 dVdci Ioi
Rv d RLi (15)
dI oi
f(Ioi)
Where, Rv is the equivalent virtual resistance.
Voi
Fig. 7 shows the improved adaptive nonlinear droop control
curve. From (13) and (14), the relationship of droop gain and VN Vref Voltage D
current is shown in Fig. 8. From Fig.8, it can be seen that the Loop
droop gain is constant in the light load conditions, which
promises better voltage regulation. In the heavy load Fig. 9 General droop control structure
conditions, the droop gain become large to promise well The proposed method utilizes a single voltage loop to control
current sharing accuracy and the converter output voltage the voltage. Because no real resistors are used, no extra
reference is inflated to ensure better voltage regulation. Since power losses are introduced.
the performance of the improved adaptive droop control
method is better than the linear and nonlinear droop control
methods.
49.2
Converter 1
48.8
Input
Vdci (V)

48.4
Drive Control
source board
Converter 2
48
0 0.25 0.5 0.75 1 DSP
Current (p.u.)
Fig. 7 Droop characteristics for improved adaptive nonlinear
droop control method.
0.4
0.3
Fig. 10 Photograph of the experimental setup.
Droop gain(Ω )

0.2
0.1 4. Experiment Verification
0
-0.1 Table 1 Converter Parameters
-0.2 Parameter Value
-0.3
0 0.25 0.5 0.75 1 Swithing frequency fs/kHz 50 kHz
Current (p.u.)
Fig. 8 Droop gain curve. Line resistance RL1ǃRL2 0.3ǃ0.35 Ω
Nominal current I 6A
3. Control scheme
Output capacitor C 800 uF
The control diagram of the proposed improved adaptive Input voltage V1ǃV2 36 V
nonlinear droop control method is shown in Fig. 9. From Fig.
9, it can be observed that the improved adaptive nonlinear Nominal voltage VN 48 V
droop control method utilizes a single voltage control loop to 0.22866804ǃ
realize the converter control and the output voltage reference
Droop coefficients n1ǃn3ǃn5 0.00188899ǃ
of the converter varies with the output current. The difference
value of the converter output voltage reference Vref and real -0.00012553
value Voi passes across the voltage loop and duty cycle is To verify the performance of the proposed droop control
obtained. Here, f (Io) represents the droop equation, while Voi method, two voltage-controlled converters with a total power
and Ioi are the converter output voltage and current, of 600 W are connected to form a simple DC microgrid as
respectively. shown in Fig. 10. The voltage-controlled converters are boost
For the ith converter, f (Io) can be expressed as follows: converters. These converters operate in voltage control mode
f ( I oi ) n1 I oi  n3 I oi3  n5 I oi5 (16) and regulate their output voltages by incorporating the
proposed droop control strategy. The converter parameters
are listed in table 1. The control of the individual converter is

Authorized licensed use limited to: University of Exeter. Downloaded on July 15,2020 at 15:30:54 UTC from IEEE Xplore. Restrictions apply.
carried out using TMS320F28335 DSP microcontroller from the output current waveform of converter 1 and converter 2,
Texas Instruments. respectively. It can be observed that the proposed method
remain stable when load change.
3.1 Light load conditions
Fig. 11 shows the experimental dc bus voltage VBus and 5. Conclusion
converters output currents Io1 and Io2 for a total load of 200 W.
In this paper, an improved adaptive nonlinear droop control
Here, the voltage value is 47.9 V and voltage deviation is
method is proposed to improve the load sharing and voltage
0.2%, which is within limit (±5%) for the improved adaptive
regulation in the DC microgrid, which can achieve a better
nonlinear droop control method.
tradeoff between current sharing and voltage regulation. The
tight voltage regulation is promised under light load
conditions. In the heavy load conditions, the better current
sharing accuracy is promised to prevent overloading and to
VBus improve the converter output voltage regulation without
Io1 adding the extra compensation loop. A fitted droop gain is
Io2 given to make sure the automatic track in the different load
ranges. The experimental results agree well with the
theoretical analysis.
Fig. 11 Converter output currents and dc bus voltage for a
load of 200 W. 6. Acknowledgements
3.2 Heavy load conditions This work was supported by the National Key Research and
Fig. 12 shows the experimental dc bus voltage VBus and Development Program of China (2018YFB0904100) and
converters output currents Io1 and Io2 for the proposed droop Science and Technology Project of State Grid
control methods for a total load of 500 W. From Fig. 12, it (SGHB0000KXJS1800685).
can be seen that the current sharing error is within limit for
the improved adaptive droop control method, which is 3.2%. 7. References
In addition, the bus voltage value is 47.6V and voltage
[1] J. M. Guerrero, P. C. Loh, T. Lee and et al, “Advanced
deviation is 0.83%, which is within limit for the improved
control architectures for intelligent microgrids—Part II:
adaptive nonlinear droop control method.
power quality, energy storage, and AC/DC microgrids,”
IEEE Transactions on Industrial Electronics, vol. 60, no. 4,
pp. 1263-1270, Apr. 2013.
[2] X. Lu, J. M. Guerrero, K. Sun and et al, “An improved
VBus droop control method for DC microgrids based on low
Io1 bandwidth communication with DC bus voltage restoration
Io2 and enhanced current sharing accuracy,” IEEE Transactions
on Power Electronics, vol. 29, no. 4, pp. 1800-1812, Apr.
2014.
Fig. 12 Converter output currents and dc bus voltage for a [3] P. C. Loh, D. Li, Y. K. Chai and F. Blaabjerg, “Hybrid
load of 500 W. AC–DC microgrids with energy storages and progressive
energy flow tuning,” IEEE Transactions on Power
In heavy load conditions, under the premise of ensuring a
Electronics, vol. 28, no. 4, pp. 1533-1543, Apr. 2013.
good current sharing accuracy, no compensation link is
[4] Roytelman and V. Ganesan, “Coordinated local and
introduced to make the voltage deviation within the specified
centralized control in distribution management systems,”
range.
IEEE Transactions on Power Delivery, vol. 15, no. 2, pp.
3.3 Dynamic loading conditions 718-724, Apr. 2000.
[5] P. N. Vovos, A. E. Kiprakis, A. R. Wallace and et al,
VBus “Centralized and distributed voltage control: impact on
distributed generation penetration,” IEEE Transactions on
Power Systems, vol. 22, no. 1, pp. 476-483, Feb. 2007.
Io1 Io1 [6] J. Schonberger, R. Duke and S. D. Round, “DC-bus
signaling: a distributed control strategy for a hybrid
Io2 Io2 renewable nanogrid,” IEEE Transactions on Industrial
Electronics, vol. 53, no. 5, pp. 1453-1460, Oct. 2006.
Fig. 12 Converter output currents and dc bus voltage during [7] S. Anand, B. G. Fernandes and J. Guerrero, ͆Distributed
change in load from 300 W to 500 W. control to ensure proportional load sharing and improve
Fig. 13 shows the converter currents and dc bus voltage when voltage regulation in low-voltage DC microgrids,” IEEE
load change from 300 W to 500 W. From Fig. 13, it can be Transactions on Power Electronics, vol. 28, no. 4, pp. 1900-
seen that the voltage remain stable and change from 47.9 V to 1913, Apr. 2013.
47.7 V when load changes. The green line and purple line are [8] V. Nasirian, A. Davoudi, F. L. Lewis and et al,

Authorized licensed use limited to: University of Exeter. Downloaded on July 15,2020 at 15:30:54 UTC from IEEE Xplore. Restrictions apply.
“Distributed adaptive droop control for DC distribution
systems,” IEEE Transactions on Energy Conversion, vol. 29,
no. 4, pp. 944-956, Dec. 2014.
[9] J. He, Y. Li, B. Liang and et al, “Inverse power factor
droop control for decentralized power sharing in series-
connected-microconverters-based islanding microgrids,”
IEEE Transactions on Industrial Electronics, vol. 64, no. 9,
pp. 7444-7454, Sept. 2017.
[10] B. T. Irving and M. M. Jovanovic, “Analysis, design,
and performance evaluation of droop current-sharing
method,” APEC 2000. Fifteenth Annual IEEE Applied Power
Electronics Conference and Exposition (Cat.
No.00CH37058), vol.1, pp. 235-241, 2000.
[11] J. Beerten and R. Belmans, “Analysis of power sharing
and voltage deviations in droop-controlled DC grids,” IEEE
Transactions on Power Systems, vol. 28, no. 4, pp. 4588-
4597, Nov. 2013.
[12] N. L. Diaz, T. Dragičević, J. C. Vasquez and et al,
“Intelligent distributed generation and storage units for DC
microgrids—a new concept on cooperative control without
communications beyond droop control,” IEEE Transactions
on Smart Grid, vol. 5, no. 5, pp. 2476-2485, Sept. 2014.
[13] N. M. Dehkordi, N. Sadati and M. Hamzeh, “Robust
tuning of transient droop gains based on Kharitonov’s
stability theorem in droop-controlled microgrids,” IET
Generation, Transmission & Distribution, vol. 12, no. 14, pp.
3495-3501, Aug. 2018.
[14] D. Chen, L. Xu and L. Yao, “DC voltage variation based
autonomous control of DC microgrids,” IEEE Transactions
on Power Delivery, vol. 28, no. 2, pp. 637-648, Apr. 2013.
[15] X. Yu, X. She, X. Ni and et al, “System integration and
hierarchical power management strategy for a solid-state
transformer interfaced microgrid system,” IEEE Transactions
on Power Electronics, vol. 29, no. 8, pp. 4414-4425, Aug.
2014.
[16] A. Khorsandi, M. Ashourloo, H. Mokhtari and et al,
“Automatic droop control for a low voltage DC microgrid,”
IET Generation, Transmission & Distribution, vol. 10, no. 1,
pp. 41-47, Jan. 2016.
[17] A. Khorsandi, M. Ashourloo, and H. Mokhtari, “A
decentralized control method for a low-voltage DC
microgrid,” IEEE Trans. Energy Convers., vol. 29, no. 4, pp.
793–801, Dec. 2014.

Authorized licensed use limited to: University of Exeter. Downloaded on July 15,2020 at 15:30:54 UTC from IEEE Xplore. Restrictions apply.