Modeling and Control of a Dual Active Bridge for

Energy Storage in DC Microgrid Applications

Frederico L. F. Marcelino, Hans H. Sathler Thiago R. de Oliveira, Pedro F. Donoso-Garcia
Graduate Program in Electrical Engineering Department of Electronic Engineering
Federal University of Minas Gerais-Av. Antonio Carlos Federal University of Minas Gerais-Av. Antonio Carlos
6627, 31270-901, Belo Horizonte, MG, Brazil 6627, 31270-901, Belo Horizonte, Brazil
[email protected], [email protected] [email protected], [email protected]

Abstract—In the present paper the modeling and control of a The energy storage converter must exhibit mainly bidi-
Dual Active Bridge (DAB) converter with input and output filters rectional power flow capability and a high conversion rate
for energy storage systems (ESS) applied to DC microgrids has depending on the voltage levels employed. Many topologies
been described. The small-signal equivalent models are derived
from the averaged circuit of the converter. Since the DAB have been proposed to this sort of application with special
converter with filters exhibit distinct transfer functions for charge attention to the DAB converters. A DAB converter consists
and discharge modes, a double loop control and a method for of two full bridges connected to a high frequency transformer
smooth transition have been proposed. Simulations results are and an auxiliary inductor. Bidirectional power flow can be
presented in order to evaluate the proposed control technique. achieved controlling both bridges to apply phase-shifted square
Keywords—Dual Active Bridge, dc-dc converter, dc microgrid,
energy storage.
or quasi-square voltage waveforms across the terminals of
the transformer, which depend on the modulation techniques.
The most simple and traditional modulation scheme is the
I. I NTRODUCTION
Phase Shift Modulation (PSM) used in [1]. More elaborated
Microgrids are described as small distribution power sys- schemes have been developed to improve figures of merit such
tems which employ distributed generation and power man- as zero voltage switching (ZVS) and conduction/switching
agement to supply a set of local loads and that can operate losses across the entire operation range but at the expense
connected to or isolated from the utility grid. Lately, they of complexity [2].
have become an important topic in power electronics and Besides the minimum requirements for ESS applications,
electrical power systems fields, since they can promote greater the DAB converters have advantages over other dc-dc topolo-
penetration of renewable resources (RES) to the power grid, gies such as high power density, high power efficiency and low
reducing the dependency on fossil fuels, and enhance local number of devices [3], [4], which makes this converter a very
consumer power quality, since local network dynamics can attractive candidate to interface an ESS and a microgrid dc
be decoupled from the utility grid. Moreover, the microgrid bus. On the other hand, a relevant drawback is the high ripple
paradigm allows the development of local dc distribution, content of the converter input/output currents. To suppress the
usually associated with residential and commercial buildings, injected/output current ripples into the microgrid dc bus and
which can substantially increase the energy efficiency of the ESS, low pass filters are connected to both sides of a DAB
buildings, since most loads in these environments are intrin- converter. The circuit of the proposed converter with filters is
sically dc or may operate easily from a dc bus. Therefore, dc shown in Fig. 1.
microgrids can contribute towards net zero energy buildings. The control of the converter must take into account the
Renewable resources have a disadvantage of being strongly bidirectional power flow according to the microgrid operating
dependent on climate and environmental conditions and there- point. Since the DAB converter exhibit different transfer
fore the power generation may vary considerably. Isolated functions for charging and discharging modes, one current
microgrids based on RES may experience instability in case controller with adaptive parameters or two current controllers
of a mismatch between supply and demand. In order to with fixed parameters must be used. In this paper two current
mitigate this problem and to smooth out energy balance, controllers are employed. The microgrid dc bus and the ESS
ESS in microgrids have been considered independently of voltages are also controlled while discharging and charging
utility grid connection. Due to the natural dc behavior of the respectively. Seamless mode transition are achieved with a
accumulators, batteries and supercapacitors, a dc-dc converter finite state machine (FSM) based on [5].
must be used to interface the ESS and the microgrid dc bus.
This paper is organized as follows: Section II discusses
——————————————————————————————- the modeling of the DAB converter. Section III describes the
This work has been supported by the Brazilian agency CAPES. control of the DAB converter. Section IV presents the FSM
978-1-5090-5339-1/17/$31.00 ©2017 IEEE
 High Voltage Low Voltage
ii Lf1 Bridge Bridge Lf2 iess
i1 i2
+ +
+ S1 S3 S5 S7 +
Ri,ess
Llk
C1 Rf1 Cf1 I Cf2 Rf2 C2 VESS
Rd,BGIC
Zm vdc v1 Vac1 L Vac2 v2
Cb1 Cb2
Vo,BGIC Cess Rsd,ess
S2 S4
N:1 S6 S8

- - -
-

BGIC Microgrid High Voltage High High Voltage Energy Storage

Load Side Filter Frequency AC Side Filter System
Link

Fig. 1: DC microgrid and DAB converter with proposed filters.

used for smooth transition. Section V presents case studies. in [14] that the power transferred with PSM can be expressed
Finally, Section VI summarizes the contribution of this work. as
N V1 V2 d(1 − |d|)
P = 0.5 ≤ d ≤ 0.5 (1)
2Llk f
II. S MALL SIGNAL MODEL OF THE D UAL ACTIVE B RIDGE where N is the transformer voltage ratio, V1 and V2 are the
terminal voltages of the DAB converter, Llk represents the
The modeling of the DAB converter has been presented in auxiliary inductor plus the transformer leakage inductance, f is
the literature using different techniques, which can be divided the switching frequency and d is the phase shift. The literature
into simplified averaged model, continuous-time model and also uses the phase shift ϕ, which is given in radians. The
discrete-time model. In [3],[6],[7] the simplified averaged relation between d and ϕ is d = ϕ/π. From (1) the average
model is obtained from the input and output average currents, terminal currents are expressed as
related to each other by the transformer ratio. The small N V2 d(1 − |d|)
I1 = (2)
signal model is extracted through conventional perturbation 2Llk f
and linearization techniques [8]. In [9] an averaged model
N V1 d(1 − |d|)
that considers conduction and transformer power losses is pre- I2 = (3)
2Llk f
sented. In [10] the modeling and control of the DAB converter
has been achieved with gyrator theory, which resembles the Perturbing the average currents around an operation point,
method employed in [3],[6],[7]. A continuous-time averaged we need to consider the variables as averaged values plus a
model of the DAB converter which uses the dc and first small-signal such as: v1 = V1 +v̂1 , v2 = V2 +v̂2 and d = D+d, ˆ
order terms of transformer current and capacitor voltage as where the upper case letters are the dc operating points and the
state variables is presented in [4]. In [17] employed state- hatted letters are the small-signals variables. As only small-
space averaged over onde half cycle to derive the small-signal signal behavior is under evaluation, the dc terms are ignored.
transfer functions. In [11] an accurate small-signal model Assuming that the first order terms will be significantly more
discrete time for automotive DAB converter is detailed. The predominant than second order terms, the model nonlinearities
model yields highly accurate results at the expense of more can be neglected [8]. Therefore, the linearized small-signal
mathematical effort. Discrete-time modeling is also discussed model can be described as (4) and (5). The small-signal model
in [12],[13]. equivalent circuit is shown in Fig. 2.
This paper uses the averaged model presented in [3],[7]. In
this approach, the auxiliary inductor, high frequency trans- î1 = Gi1d dˆ + Gi1v2 v̂2 (4)
former and switches are substituted by dependent current î2 = Gi2d dˆ + Gi2v1 v̂1 (5)
sources and the converter is represented by its terminal average
currents I1 and I2 , which are identified in Fig. 1. In fact, where
neglecting those devices will lead to a model lower accurate î1 N V2 (1 − 2|D|)
Gi1d = = (6)
than the others. However, a simpler model facilitates the design dˆ v̂2 =0
2Llk f
of the controllers and may serve as the starting point for the
design of the converter after defining the microgrid and ESS î1 N D(1 − |D|)
Gi1v2 = = (7)
at system-level. v̂2 ˆ
2Llk f
d=0
The terminal average currents can be calculated from the
power flow expression for the DAB converter. Considering î2 N V1 (1 − 2|D|)
Gi2d = = (8)
fixed terminal voltages V1 and V2 , it has been demonstrated dˆ v̂1 =0
2Llk f
 i1 i2 35

Magnitude (dB)
+ +
30

v1 v2 Complete Circ.
Averaged Circ.
25
Small-Signal Mod.
Gi1dd Gi1V2v2 Gi2d d Gi2V1v1 Analytical Mod
- 20
- 102 103 104
Frequency (Hz)
Fig. 2: Dual Active Bridge small-signal model.
0

Table I: Low-pass filters parameters

Phase (º)
-50

Values -100
Parameters
High side filter Low side filter
-150
Lf 5µH 0.3µH 102 103 104
Cf 10µF 220µF Frequency (Hz)

Cb 33µF 660µF
Fig. 3: Frequency response of the control to ESS current
Rf 0.5Ω 0.026Ω
Gciessd transfer function for charging mode. Curve with “-
” represent the complete model, “∆” represent the averaged
î2 N D(1 − |D|) circuit,“+” represent the small-signal model and “o” represent
Gi2v1 = = (9) the analytical model.
v̂1 ˆ
2Llk f
d=0

Since the DAB is a bidirectional converter, the direction

of the power flow must be considered when using the small- circuit presents similar magnitude response until about 4kHz,
signal model. According to Fig. 1 and (1)-(2), if the phase but deviates from the other after this frequency. The small-
shift d is positive the power flow is from V1 to V2 and the signal model phase response is identical to the averaged circuit
small-signal model is identical to the circuit shown in Fig. 2. phase response. The complete circuit exhibit significative
On the other hand, if the power flow is from V2 to V1 , the phase deviation from the others due to the presence of the
directions of all dependent currents of the small-signal model modulator and switches. In order to adequately account the
shown in Fig. 2 are reversed, which are particularly important phase deviation the analytical expressions must include a time
for transfer functions extraction. delay transfer function e−sτ .
The DAB converter input and output currents i1 and i2 As shown in (6)-(9), the transfer functions derived from the
exhibit high current ripples, which are undesirable from the small-signal model are essentially static gains, which values
point of view of the microgrid and the ESS. Thus, input/output are highly dependent of the average phase shift D. Fig. 7
LC low-pass filters with a RC parallel damping branches shows this dependence for 0 ≤ D ≤ 0.5. The parameters used
have been used to mitigate current and voltage ripples [8]. are shown in Table III. As can be seen, the gains can vary
Calculations have been done to keep the maximum harmonic significantly according to the power flow of the converter.
currents of i1 and i2 at twice the switching frequency equal to
approximately one fourth of their nominal values. The design III. C ONTROL OF D UAL ACTIVE B RIDGE C ONVERTER
procedure employed is based on [14], but with a RC parallel
damping branch. The filter topology is shown in Fig. 1 and This paper considers a DAB converter modulated with PSM
the values are listed in Table I. to interface an ESS and a droop based microgrid dc bus. When
With the small-signal model and the filter topology we the Bidirectional Grid Interface Converter (BGIC) and/or RES
can finally derive the control-to-output transfer functions. The connected to the microgrid supply enough energy to charge
control to ESS current transfer functions for charging and the ESS, both current and voltage of the ESS are controlled
discharging modes Gciessd and Gdiessd respectively are shown during the process. On the other hand, if a lack of energy exists
in Fig. 3 and 5 respectively, whereas the control to ESS the ESS must discharge to supply the remaining power to the
voltage transfer function for charging Gcvod and the control load connected to the microgrid dc bus. When discharging, the
to microgrid dc bus Gdvid for discharging mode are shown in output current of the ESS and the dc bus are the controlled
Fig. 4 and 6 respectively for D = 0.3. The parameters used variables. Therefore, an inner current loop and an outer voltage
are shown in Table III. For comparison purposes the analytical loop for charge and discharge modes are employed, consisting
transfer functions are compared with the transfer functions of a double loop control as shown in Fig. 8. In Fig. 8 Gci and
extracted from the small-signal model, averaged circuit model Gcv are the current and voltage controllers for charging mode,
and complete circuit, which were obtained through ac sweep respectively; Gdi and Gdv are current and voltage controllers for
with PSIM. The Bode plots show that the averaged circuit, discharging mode, respectively; Gc1 is the ESS current to ESS
small-signal model and the analytical expressions have approx- voltage transfer function for charging mode; Gd1 is the ESS
imately the same results for magnitude response. The complete current to microgrid dc bus transfer function for discharging
 5 60

Magnitude (dB)

Magnitude (dB)
Complete Circ.
40 Averaged Circ.
Small-Signal Mod.
0 Complete Circ. Analytical Mod.
Averaged Circ. 20
Small-Signal Mod.
Analytical Mod.
-5 0
102 103 104 102 103 104
Frequency (Hz) Frequency (Hz)

0 -60
Phase (º)

Phase (º)
-50 -80

-100 -100

-150 -120
102 103 104 102 103 104
Frequency (Hz) Frequency (Hz)

Fig. 4: Frequency response of the control to ESS voltage Fig. 6: Frequency response of the control to microgrid dc bus
Gcvod transfer function for charging mode. Curve with “-” voltage Gdvid transfer function for discharging mode. Curve
represent the complete model, “∆” represent the averaged with “-” represent the complete model, “∆” represent the
circuit, “+” represent the small-signal model and “o” represent averaged circuit, “+” represent the small-signal model and “o”
the analytical model. represent the analytical model.

45
150
Magnitude (dB)

Complete Circ.
Averaged Circ.
40 100
Small-Signal Mod.
Gain
Analytical Mod.
35 50

0
30 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
102 103 104
Phase shift
Frequency (Hz)
(a)
50
0.1
Phase (º)

0
Gain

0.05
-50

0
-100 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
102 103 104
Phase shift
Frequency (Hz)
(b)

Fig. 5: Frequency response of the control to ESS current Fig. 7: Small-signal transfer functions variation with phase
Gdiessd transfer function for discharging mode. Curve with “- shift for (a) Gi1d in blue and Gi2d in red and (b) Gi1v2 in
” represent the complete model, “∆” represent the averaged blue and Gi2v1 in red.
circuit, “+” represent the small-signal model and “o” represent
the analytical model.
As the static gains of the small-signal model transfer functions
vary with D, the design procedure of the current controller
mode; Gmod is the modulator gain, Rd is the droop resistance
must take this variation into account. For both operation modes
of the DAB converter.
the current controllers must guarantee adequate crossover fre-
The transfer functions Gciessd and Gdiessd can be calculated
quency and minimum phase margin across the entire operation
as (10) and (11) respectively, where ZLf 2 , ZCf 2 , ZCb2 are
range. In this paper PI controllers are proposed for current
the low side voltage filter branch impedances, ZC2 is the
control and were tuned using the pole placement method with
capacitor C2 impedance and Zess is the ESS equivalent RC
parameters shown in Table II.
circuit. Since Gciessd and Gdiessd are different, separate current
control have been proposed. The compensated current open The compensated voltage open loop transfer functions for
loop transfer functions for charging and discharging modes charging and discharging modes Tvc and Tvd respectively, are
Tic and Tid respectively, are described as described as
Tic = Gc1 Iess,cl
c
Gcv (14)
Tic = Gciessd Gmod Gci (12)
Tid = Gdiess Gmod Gdi (13) Tid = Gd1 Iess,cl
d
Gdv (15)
 VESS* + iESS* + d c iESS c vESS
Gvc Gic Gmod Giessd G1
- -

(a)

Vdc,bus* + + d iESS* + d c iESS vdc,bus

Gv Gid Gmod Giessd G1d
- - -
-ii
Rd
(b)
Fig. 8: Dual Acive Bridge converter control block diagram.

îess ZCf 2 //ZCb2 ZC2

Gciessd = = Gi2d · · (10)
dˆ charge
ZCf 2 //ZCb2 + ZLf 2 + ZC2 //Zess ZC2 + Zess

îess ZCf 2 //ZCb2

Gdiessd = = Gi2d · (11)
dˆ discharge
ZCf 2 //ZCb2 + ZLf 2

c d
where Iess,cl and Iess,cl are the current closed loop transfer adaptively recalculated at every sampling time. The results
functions for discharging and charging mode respectively. The showed that the adaptive PI achieved consistent performance
transfer functions Gc1 and Gd1 are calculated as (16) and across all the operating conditions, but at the expense of more
(17) respectively, where Zm is the microgrid load impedance, effort in controller design.
ZLf 1 , ZCf 1 , ZCb1 are the high side voltage filter branch Assuming that the low-pass filters suppress sufficiently the
impedances, ZC1 is the capacitor C1 impedance. harmonic content of the currents no filter has been considered
to measure the controlled variables. In addition, for simplifica-
Gc1 = Zess (16)
tion purposes, sensors have been defined with unity gain and
As can be seen in (17) the voltage open loop transfer therefore provide unity feedback loop. Otherwise, the sensor
function in discharge mode may have its static gain varied gains would be inserted in the feedback loop and the forward
depending on the power flow. Likewise the method described path transfer functions would vary.
for the current loop, the voltage controllers must be tuned to
ensure converter stable operation for all phase shift allowed IV. T RANSITION CONTROL
values. The parameters of the proposed PI voltage controllers
were tuned using the same procedure used for the current Bidirectional control with two control loops is achieved
controllers and are shown in Table II. only if an adequate transition is accomplished [5]. Unfortu-
The operation range of the DAB converter is limited to nately most papers about DAB converter lack the power flow
0 ≤ |d| ≤ 0.3, since increasing the phase shift the conduction transition. In [16] a smooth transition between charging and
losses are increased due to the reactive power flow in the discharging operation modes was achieved with the use of one
transformer [3]. For this range and the controller gains tuned current controller and two competitive voltage loops. In this
the minimum phase margin and crossover frequency range are paper a double loop control is employed, thus the technique
shown in Table II for the inner current loops and outer voltage used in [16] is not applicable. Therefore to switch between
loops. As can be seen, the static gain variations of the small- charge and discharge modes a FSM based on [5] has been
signal model modify the current loops crossover frequencies. developed. The algorithm detects the microgrid operation point
On the other hand, the voltage crossover frequencies were not using the dc bus according to the microgrid droop design and
modified significantly. defines which control loop must be used. When a transition is
The major advantage of the fixed parameters is the ease detected, the algorithm enters in a special state. Neither control
to tune the gains of the PI controllers. On the other hand, it loops are used, but the last phase shift passed as reference to
is not possible to keep the same phase margin and crossover the modulator is sampled and directly increased or decreased
frequency for the entire operation range, since the small-signal to drive the ESS current gradually toward zero. The current
transfer functions modify the static gain of the forward path rate can be easily adjusted through a designer defined phase
loop according to the phase shift value. This means that the shift step value. When the current is near zero and dc bus
performance of the converter will vary across the operation voltage conditions are satisfied, charging or discharging mode
range. To improve performance, in [15] the proportional gain is set and the loop switching is effectively accomplished. To
of the PI controllers used to control the DAB converter is avoid instabilities, voltage and current hysteresis have been
 ZC1 ZCf 1 //ZCb1 Gi1d ZCf 2 //ZCb2 + ZLf 2
Gd1 = Zm · · · (17)
ZC1 + Zm ZCf 1 //ZCb1 + ZLf 1 + ZC1 //Zm Gi2d ZCf 2 //ZCb2

Table II: Controller parameters

15

Mode Loop Controller Phase Crossover 10

gains margin frequency
P M > 57° 1.1 − 2.3kHz 5
Current Kp = 0.0038
Charge Ki = 180.95
0
Voltage Ki = 300 P M > 76.9° 1.95Hz

iess (A)
Current Ki = 284.59 P M > 57° 1.95 − 7.7kHz -5
Discharge
Voltage Kp = 0.43 P M > 76.9° 195.7Hz -10
Ki = 338.02
-15
Table III: Parameters used in simulation
-20
Parameters
Symbol -25
Description Value 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
V1 High side nominal voltage 380V Time (s)

V2 Low side nominal voltage 48V Fig. 9: Energy Storage System current iess in (a) charging
N Transformer ratio 7.94 to discharging transition and (b) discharging to charging
Auxiliary inductance plus transition.
Llk 715.42 µH
transformer leakge inductance
f Switching frequency 20kHz
Pnom,DAD DAB nominal power 1kW ESS current, i ess (A) 20
Rd DAB droop resistance 3.6Ω
10
Pnom,BGIC BGIC nominal power 3kW
VBGIC BGIC open circuit voltage 380V 0

Rd,BGIC BGIC droop resistance 1.23Ω -10

0.048 0.0485 0.049 0.0495 0.05 0.0505 0.051 0.0515 0.052 0.0525
R = 144.4Ω Time (s)
Zm Microgrid load
C = 1000µF (a)
ESS current, i ess (A)

Ri,ess ESS internal resistance 40mΩ 10

Rsd,ess ESS self-discharge resistance 490.34Ω 0

Cess ESS stored charge capacitance 8.75F
-10
Gmod Phase Shift Modulator gain 1
-20
0.1515 0.152 0.1525 0.153 0.1535 0.154 0.1545 0.155 0.1555 0.156
Time (s)
used for mode detections. Simulations results are carried out (b)

to verify the proposed smooth transition control.

Fig. 10: Detailed Energy Storage System current iess during
V. S IMULATION RESULTS transition mode.
The circuit used for simulation is shown in Fig. 1. The ESS
was modeled as simplified RC circuit where Ri,ess is the ESS
internal resistance, Rsd,ess is the resistance for self-discharge is set and ESS current grows negatively down to the nominal
current and Cess is to represent stored charge with 105Ah discharge value (-22A) to supply the dc bus voltage according
of capacity. The BGIC used to connect the microgrid dc bus to the droop characteristic. At t=150ms, the switch is closed.
to the utility grid is modeled as a voltage source VBGIC in The algorithm detects a discharge-charge transition and drives
series with its droop resistance Rd,BGIC . The parameters used the phase shift to increase ESS current toward zero, as shown
in simulation are shown in Table III. Fig. 10b and 11b. When the current approaches -1A, charge
Fig. 9 shows the ESS current for charging to discharging mode is set and the ESS current rises until the nominal charge
and discharging to charging transition. Until 50ms, the DAB current (11A).
converter is operating in charging mode with nominal charge The dc bus voltage is shown in Fig. 12. During charge-
current (11A). At t=50ms, the BGIC is disconnected from discharge transition the voltage drops due to the microgrid
the microgrid. The control algorithm detects a charging- load and the ESS discharges to control de microgrid dc bus
discharging transition and the phase shift is linearly decreased voltage. At t=150ms the BGIC is reconnected and resumes the
to lead ESS current to zero as depicted in Fig. 10a and 11a. microgrid dc bus voltage control. Thus, the ESS can operate
When the current is sufficiently small (1A), discharge mode again in charging mode. In order to mitigate the dc bus voltage
 0.2 0.2
Phase Shift

Phase Shift
0.1 0

0 -0.2

-0.1 -0.4
0.048 0.0485 0.049 0.0495 0.05 0.0505 0.051 0.0515 0.052 0.0525 0.048 0.05 0.052 0.054 0.056 0.058 0.06 0.062
Time (s) Time (s)
(a) (a)
0.1 0.2
Phase Shift

Phase Shift
0
0
-0.1

-0.2 -0.2
0.1515 0.152 0.1525 0.153 0.1535 0.154 0.1545 0.155 0.1555 0.156 0.152 0.154 0.156 0.158 0.16 0.162 0.164
Time (s) Time (s)
(b) (b)

Fig. 11: Phase shift reference in (a) charging to discharging Fig. 13: Phase shift reference in (a) charging-to-discharging
transition and (b) discharging to charging transition with transition and (b) discharging to charging transition with
dd/dt=127.7/s. dd/dt=12.77/s.

380
derived from the averaged DAB converter was employed to
state the transfer functions necessary for controlling the con-
375 verter. Bode plots indicated that the control-to-output transfer
functions analytically derived are in accordance with the ones
DC Bus voltage, v dc (V)

370
extracted from simulation models. PI controllers with fixed
parameters are proposed for the double loop control employed.
A FSM is proposed to control the DAB converter operation
365
modes. When a transition is detected the FSM drives the phase
shift applied to the modulator adequately to achieve smooth
360 transition. The rate of the ESS current during transitions can
be easily adjusted by software.
355
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time (s)
ACKNOWLEDGMENT

Fig. 12: DC Bus voltage for charging to discharging and This work has been supported by the Brazilian agency
discharging to charging transition. CAPES.

R EFERENCES
drop during charging to discharging transitions, the capacitor
C1 can be increased. [1] R. W. A. A. De Doncker, D. M. Divan and M. H. Kheraluwala, ”A three-
phase soft-switched high-power-density DC/DC converter for high-power
Fig. 11 and 13 show ESS current for the same microgrid applications,” in IEEE Transactions on Industry Applications, vol. 27, no.
operation conditions but with derivatives of phase shift (dd/dt) 1, pp. 63-73, Jan/Feb 1991.
of 127.7/s and 12.77/s respectively. At t=50ms the FSM [2] Y. A. Harrye, K. H. Ahmed, G. P. Adam and A. A. Aboushady,
”Comprehensive steady state analysis of bidirectional dual active bridge
detected a charge-discharge transition. The last phase shift DC/DC converter using triple phase shift control,” 2014 IEEE 23rd
passed by the charging loop is sampled and is used as reference International Symposium on Industrial Electronics (ISIE), Istanbul, 2014,
for the DAB converter. The phase shift is reduced step by step pp. 437-442.
[3] A. Rodriguez Alonso, J. Sebastian, D. G. Lamar, M. M. Hernando and
until the ESS approximates zero. With dd/dt=12.77/s 8.6ms A. Vazquez, ”An overall study of a Dual Active Bridge for bidirectional
were necessary to lead to ESS current to zero, whereas with DC/DC conversion,” 2010 IEEE Energy Conversion Congress and Expo-
dd/dt=127.7/s approximately 1.0ms passed. Therefore, higher sition, Atlanta, GA, 2010, pp. 1129-1135.
[4] H. Qin and J. W. Kimball, ”Generalized Average Modeling of Dual Active
phase shift derivatives lead the ESS current to zero faster. The Bridge DCDC Converter,” in IEEE Transactions on Power Electronics,
phase shift derivative is easily adjusted by software and can be vol. 27, no. 4, pp. 2078-2084, April 2012.
used to control the rate ESS current during transition mode. [5] Saichand, K; Vinod, John. Virtual resistance based control for Ultracapac-
itor based Bidirectional dc-dc Backup System. Department of Electrical
Engineering Indian Institute of Science, Bengaluru, 2015.
VI. C ONCLUSION [6] H. K. Krishnamurthy and R. Ayyanar, ”Building Block Converter Module
for Universal (AC-DC, DC-AC, DC-DC) Fully Modular Power Conver-
In this paper the modeling and control of a DAB converter sion Architecture,” 2007 IEEE Power Electronics Specialists Conference,
with input/output filters is presented. The small-signal model Orlando, FL, 2007, pp. 483-489.
[7] Guacaneme, J., Garcer, G., Figueres, E., Patrao, I., and Gonzlez-Medina,
R. (2015) Dynamic modeling of a dual active bridge DC to DC converter
with average current control and load-current feed-forward. Int. J. Circ.
Theor. Appl., 43: 13111332.
[8] Erickson, R. W.; Maksimovic, D. Fundamentals of Power Electronics.
2ed. Springer, 2001.
[9] K. Zhang, Z. Shan and J. Jatskevich, ”Large- and Small-Signal Average-
Value Modeling of Dual-Active-Bridge DCDC Converter Considering
Power Losses,” in IEEE Transactions on Power Electronics, vol. 32, no.
3, pp. 1964-1974, March 2017
[10] Martins, W. M. D. S. e D. C. Modelagem e controle dos conversores
dab e tab utilizando a teoria do gyrator. In: XIX Congresso Brasileiro de
Automatica. [S.l.: s.n.], 2012.
[11] F. Krismer and J. W. Kolar, ”Accurate small-signal model for an auto-
motive bidirectional Dual Active Bridge converter,” 2008 11th Workshop
on Control and Modeling for Power Electronics, Zurich, 2008, pp. 1-10.
[12] Ling Shi, Wanjun Lei, Jun Huang, Zhuoqiang Li, Yao Cui and Yue
Wang, ”Full discrete-time modeling and stability analysis of the digital
controlled dual active bridge converter,” 2016 IEEE 8th International
Power Electronics and Motion Control Conference (IPEMC-ECCE Asia),
Hefei, 2016, pp. 3813-3817.
[13] D. Costinett, R. Zane and D. Maksimovi, ”Discrete-time small-signal
modeling of a 1 MHz efficiency-optimized dual active bridge converter
with varying load,” 2012 IEEE 13th Workshop on Control and Modeling
for Power Electronics (COMPEL), Kyoto, 2012, pp. 1-7.
[14] Krismer, F. Modeling and optimization of bidirectional dual active
bridge DC-DC converter topologies. Thesis, Swiss Federal Institute of
Technology, 2010.
[15] Segaran, Dinesh Sekhar. Dynamic Modelling and Control of Dual Active
Bridge Bi-Directional DC-DC Converters for Smart Grid Applications.
Thesis, Royal Melbourne Institute of Technology University, 2006.
[16] W. W. A. G. Silva, P. F. Donoso-Garcia, S. I. Seleme, T. R. Oliveira, C.
H. G. Santos and A. S. Bolzon, ”Study of the application of bidirectional
dual active bridge converters in dc nanogrid energy storage systems,” 2013
Brazilian Power Electronics Conference, Gramado, 2013, pp. 609-614.
[17] G. D. Demetriades and H. P. Nee, ”Dynamic modeling of the Dual-
Active Bridge topology for high-power applications,” 2008 IEEE Power
Electronics Specialists Conference, Rhodes, 2008, pp. 457-464.