Efficiency and Cost Optimization of Dual Active

Bridge Converter for 350kW DC Fast Chargers

Sadik Cinik Fangzhou Zhao Giuseppe De Falco Xiongfei Wang

Department of Energy Department of Energy Infineon Technologies Division of Electric Power
Aalborg University Aalborg University Austria AG and Energy Systems
Aalborg, Denmark Aalborg, Denmark Villach, Austria KTH Royal Institute of
[email protected] [email protected] Giuseppe.DeFalco@infine Technology
on.com Stockholm, Sweden
[email protected]

Abstract— This study focuses on optimizing the design modulation techniques [4] or hardware modifications [5,6].
parameters of a Dual Active Bridge (DAB) converter for use in However, these studies overlook two critical aspects. First, the
350 kW DC fast chargers, emphasizing the balance between balance between efficiency and cost is often neglected, which
efficiency and cost. Addressing the observed gaps in existing is crucial for ensuring the economic viability of charging
high-power application research, it introduces an optimization systems. Second, while converters are designed and optimized
framework to evaluate critical design parameters—number of for lower power levels, the trade-offs for high-power scenarios
converter modules, switching frequency, and transformer turns like 350 kW DC fast chargers remain under-investigated. For
ratio—within a broad operational voltage range. The analysis
example, the optimal configuration and number of converter
identifies an optimal configuration that achieves over 95%
modules in building a 350 kW system require further research.
efficiency at rated power across a wide output voltage range,
comprising seven 50 kW DAB converters with a switching To address the identified need, this paper aims to extend
frequency of 30 kHz, and a transformer turns ratio of 0.9. the isolated DC/DC converter optimization analysis
specifically for 350 kW DC fast chargers. It employs a
Keywords— DC Fast Charging, Dual Active Bridge Converter, detailed framework to evaluate how various design
Wide Output Voltage, Efficiency and Cost Optimization parameters—such as converter module count, switching
I. INTRODUCTION frequency, and transformer turns ratio—affect efficiency and
cost across different output voltages. For this optimization, the
Over the last decade, there has been a significant increase Dual Active Bridge (DAB) converter is chosen due to its
in electric vehicle (EV) adoption, driven by substantial widespread utilization in the industry, simplicity, inherent
incentives and investments, despite challenges such as limited buck-boost operation capability, and bidirectionality. Section
range capacities and long charging times [1]. DC fast charging II details the optimization methods and converter modelling,
technology, particularly through the deployment of 350 kW while Section III discusses how these parameters impact
ultra-high-power units, is viewed as a key solution, aiming to efficiency-cost trade-offs in varied voltage operations.
reduce charging times to below 10 minutes.
II. METHODOLOGY
To achieve this goal, the industry is shifting from 400V to
800V systems in passenger vehicles to enable high-power This section presents a systematic optimization procedure,
charging without increasing current drawn by batteries [2]. depicted in Fig. 1, aiming to identify the optimal design
This evolution requires DC fast chargers that support a wide parameters for the DAB converter. These parameters include
output voltage range, from 150V to 1000 V. However, the the switching frequency, the transformer turns ratio, and the
demands for high power and broad output voltage range, number of converter modules necessary to achieve a total
coupled with pre-existing necessities such as galvanic power output of 350 kW. The primary goal is to maximize
isolation, introduce complex design challenges. These efficiency and cost for the converter design.
challenges, directly tied to converter technology critical for A. Overall System Requirements
charging station functionality, include efficiency degradation
over wide voltage range operations and increased costs. The optimization flow begins with the overall system
requirements, which cover CHARIN's DC fast charger
Today, DC fast chargers utilize two power conversion specifications [7]. It specifies a maximum power (P ) of 350
stages: an AC/DC unit for AC to DC conversion and a DC/DC kW and a maximum current (I ) of 500 A, with an output
unit to regulate this DC for EV battery voltage needs. These voltage range (V ) of 150 V to 1000 V.
chargers integrate isolation features, through a large line-
frequency transformer prior to the AC/DC stage or a smaller, B. Optimization Design Space
more efficient one within the DC/DC converter stage. The The study considers three critical parameters that
latter is favored for its reduced size and weight, providing significantly impact the cost and efficiency of converters for
necessary isolation [3]. Additionally, the DC/DC converter DC fast charging applications. These parameters are varied
must sustain high efficiency across a broad output voltage within predefined ranges to determine the optimal
range, a critical aspect for optimizing the charging process. configuration that balances efficiency with cost-effectiveness:
Existing literature, therefore, predominantly focuses on Number of Converter Modules (N): Explored within a
enhancing the efficiency of isolated DC/DC converters across range from 5 to 35, this parameter determines how many
a wide output voltage ranges, with numerous studies individual converters are required to be stacked together to
demonstrating efficiency improvements through advanced reach the target power output of 350 kW.
 Fig. 2. (a) DAB Circuit Configuration. (b) Equivalent Circuit Model.
(c) Time Domain Representation of Instantaneous Inductor Current.

offering direct power flow control. While various modulation

strategies could refine performance through these phase shifts,
this research uses the Single-Phase Shift (SPS) approach for
its ease and straightforward implementation.
D. Electrical Model
Fig. 1. Optimization Flowchart for a Converter in 350 kW DC Fast Charger
The initial phase in developing the electrical model is to
Switching Frequency (𝐟𝐬𝐰 ): Analyzed between 5 kHz compute the necessary leakage inductance, a key factor
and 100 kHz, f is crucial for the efficiency and cost of the impacting subsequent current waveform calculations essential
converter components, directly affecting its performance, for optimization.
size, and thermal management. Leakage Inductance Determination: The leakage
Transformer Turns Ratio (n): Examined from 0.2 to 1.7, inductance L is a key parameter for the DAB converter,
n is crucial for voltage transformation and isolation, impacting directly impacting power delivery capability. To ensure the
the transformer's cost, efficiency, and design complexity. required rated power can be reached for each output voltage
operation, the leakage inductance is computed as:
C. Converter Specification
V V D(1 − D)
This section details the specific parameters and L = (2)
operational principles that define the converter's design within 2nf P ,
the context of this optimization study. Voltage and Current Waveforms Determination: The
Power and Current Ratings: The power (P , ) and primary and secondary voltage waveforms are defined as
current (I , ) ratings for the converter are determined by square waves, with their amplitudes adjusted based on the
dividing the total target of 350 kW by N. input and output voltages. Given the 150 - 1000 V range
needed for fast charging, the converter operates in both boost
Input and Output Voltage Ratings: The study assumes and buck modes. This functionality results in different current
that input voltage (V ) can be regulated between 600 V and waveform patterns in the converter. Utilizing these current
850 V to enhance the converter efficiency under wide output waveforms, formed by using (1), (2), enable precise current
calculations for any active switch or diode. Such detailed
voltage range operation. The output voltage (V ) maintained analysis is critical for the subsequent converter component
as specified in the system requirements (150 V - 1000 V). models, directly influencing the overall optimization flow.
Temperature Constraints: The optimization stipulates E. Global Optimization Strategy
that the maximum temperatures for both the semiconductor In the optimization strategy, costs and average operational
case (T , ) and the transformer (T , ) should not exceed losses for key converter components—semiconductors,
100 ℃, considering an ambient temperature (T ) of 25 ℃. cooling systems, transformers, and capacitors—are evaluated
Topology and Modulation: The DAB converter is under various design scenarios determined by configurations
chosen for its notable features [4], illustrated in Fig. 2.a, of N, n, and f . The strategy comprises two core analyses for
showcasing its dual-active bridge architecture linked by a each variable set:
high-frequency transformer. The bridges are modelled as AC 1. Cost Calculations: The costs for each component are
voltage sources in rectangular waveforms, as depicted in Fig. determined for every design variable combination to
2.b. The inductor L , essential for the power transfer, is the quantify the financial impact of each configuration.
transformer's leakage inductance. Its instantaneous current
shown in Fig. 2.c is described in (1). 2. Operational Loss Calculations: Component losses are
calculated across different voltage (150 V to 1000 V) and
1 v , (t) power levels (10% and 100% of rated power). By
i (t ) = i (t )+ v , (t) − dt (1) analysing losses at each voltage and power level
L n
intersection, a comprehensive profile of the converter's
where i (t ) represents the inductor current at any given operational behaviour is established. These losses are
time, L is the leakage inductance, v , (t) and v , (t) are averaged to establish a benchmark loss metric for each
the transformer’s primary and secondary voltages, design variable set.
respectively, and n is the turns ratio. The inductor current i ,
These collective cost and loss evaluations for each design
within the DAB converter can be modulated by altering the
variable set (N, n, f ) are then integrated into an optimization
phase shift ratio (D) between the dual full bridges, and the
synthesis block. This block consolidates the data, supporting
individual phase shifts (D1, D2) within each bridge's legs,
strategic decisions to optimize the converter's cost and
operational efficiency, aiming for an overarching optimized (E ) and turn-off (E ) energy values are provided by
solution. manufacturers for typical conditions, which may not reflect
the actual operating environment. To bridge this gap, scaling
F. Converter Component Models
factors for gate resistance (K ), junction temperature K ,
This section delves into detailed analyses of the key
and switching voltage K are employed to modify these
converter components: semiconductors, cooling systems,
transformers, and capacitors. values, thus allowing for an accurate determination of
switching losses:
1) Semiconductor Model 1
Semiconductor devices are pivotal in power converter P = E (I ) ∗ K ∗K ∗K +
designs, influencing both efficiency and cost. This study T
utilizes 1200 V SiC MOSFETs from Infineon, noted for their 1
reliability and efficiency, detailed in Table I. These devices, E (I ) ∗ K ∗K ∗K (4)
T
selected across various current ratings, accommodate the
DAB converter's 10 kW to 70 kW output range, aligned with with
the N values in design variables. Component pricing, essential
for cost analysis, was obtained from Mouser Electronics, K ( / ) ∶E / (R )⁄E / R
reflecting the cost for 100 units as of March 1, 2024, K ∶ E T E T
( / ) / /
facilitating our cost optimization evaluation.
K ( / )
∶ V ( / ) V ( / )
TABLE I. INFINEON 1200 V SIC MOSFET LIST
Here, the subscript "DS" denotes datasheet values, while
Part Number 𝐑 𝐝𝐬(𝐨𝐧) 𝐈𝐝 @𝟏𝟎𝟎 ℃ 𝑪𝒐𝒔𝒕𝟏𝟎𝟎
"user" denotes the actual operating conditions. If the diode's
IMZA120R007M1H 7 mΩ 168 A $62.39 reverse recovery energy E is not included in E or E , its
reverse recovery loss P needs to be calculated using a
IMZA120R014M1H 14 mΩ 89 A $32.34
formula similar to (4). This loss should then be added to P .
AIMZH120R020M1T 20 mΩ 71 A $29.15
Subsequently, the total loss per switch, P , , ,
AIMZH120R030M1T 30 mΩ 49 A $20.62 calculated as the sum of P and P , for each switch listed
in Table I, is integrated into the cooling system model to
optimize thermal management.
Semiconductor losses are mainly categorized into
conduction and switching losses. The upcoming subsection 2) Cooling System Model
outlines the method to calculate these losses, leveraging the Effective cooling management is vital for semiconductor
semiconductor device specifications from manufacturer performance, keeping its junction temperature below the
datasheets and the converter's operational conditions. These datasheet-specified T , . While cooling systems are often
conditions cover switching frequency, ambient temperature, overlooked in efficiency optimizations due to minimal losses,
and previously determined current/voltage waveforms. their significant cost impact necessitates integrating a detailed
cooling system model into the optimization process. Inspired
a) Conduction Losses:
by [8] that optimizes the weight of forced convection cooling
Semiconductor characteristics i = f(v ) and anti- systems for airborne wind turbines, this study adopts a similar
parallel diode characteristics i = f(v ) at various approach to assess and reduce the cooling system's cost,
temperatures are derived from datasheets and organized into featuring a fan, an adapter, and a straight-fin aluminum heat
lookup tables. These tables, combined with respective current sink as shown in Fig. 3.b.
profiles, allow accurate conduction loss calculations for both
semiconductors and diodes by interpolating across diverse Distinct cooling system designs for the primary and
operating conditions. The total conduction loss is then secondary side switches of the DAB converter are required
determined using (3). due to the potential variance in thermal behaviour and power
loss between two bridges. The cooling system optimization
1 outlined in the overall optimization flow chart (Fig. 1) is
P = i (t) ∗ v i (t), T dt
T methodically structured with the following steps.

+ i (t) ∗ v i (t), T dt (3) a) Definition of Input Parameters

In addition to the input parameters like T , ,T ,
In this formulation, T represents the switching period. P , , already provided previously, the process begins
The i (t) and i (t) instantaneous currents through the with the definition of critical parameters that include:
semiconductor switch and diode, while the functions Heatsink Geometry: Key dimensions defining heatsink
v i (t), T and v i (t), T specify how the saturation design optimization, as illustrated in Fig. 3.a, are detailed with
voltage of the semiconductor and the forward voltage of the their respective ranges in Table II:
diode vary with their respective currents and junction
temperatures (𝑇 ). TABLE II. HEATSINK GEOMETRY RANGES

b) Switching Losses: Sink length (𝐋𝐬) [50-200] mm Fin height (𝐡𝐟 ) [10-𝑓 ] mm
Switching losses, which occur during the transient states Sink width (𝐰𝐒 ) Fan width (𝑓 ) Fin width (𝐭 𝐟 ) [1-20] mm
of turn-on and turn-off events, are inherently more complex Baseplate
than steady-state conduction losses. Standardized turn-on 3 mm Fin number (𝐧𝐟 ) [3-50]
thickness (𝐝𝐒 )
Fig. 3. (a) Detailed Heatsink Geometry Specifying Key Design
Optimization Dimensions. (b) Cooling System Featuring a Fan, and an
Adapter, and a Straight-fin Aluminum Heat Sink.
Fig. 4. (a) Detailed Transformer Geometry Specifying Key Design
Optimization Dimensions.
Heatsink Cost Data: A dataset enumerating various
aluminium heatsink options sourced from Mouser as of March
1, 2024, including their costs and weights. d) Optimal Configuration Determination
The process results in comparing the stored configurations
Fan List: A list of fans given in Table III, detailing their to identify the most cost-efficient choice that also satisfies
specifications to be iteratively evaluated in the optimization. thermal performance criteria.
TABLE III. FAN LIST CONSIDERED FOR THE OPTIMIZATION 3) Transformer Model
This phase, as outlined in the overall optimization flow
Part Number Size(HxWxL) Air Flow 𝑪𝒐𝒔𝒕𝟓𝟎
chart (Fig.1), systematically evaluates various design
0.36 parameters to identify configurations that deliver the highest
9GA0412P7G001 40x40x15 mm $13.33
m3/min
0.67
efficiency and the lowest cost, while adhering to all necessary
9GA0412P3J01 40x40x28 mm
m3/min
$14.97 operational constraints.
1.13
04028DA-12V-A6-KG 40x40x28 mm
m3/min
$20.03 a) Definition of Input Parameters
Beside the parameters previously defined such as T , ,
T , and the DAB converter's operational conditions
b) Calculation of Maximum Heatsink to Ambient
P , ,I , , V , V , n, f , L , this phase integrates
Thermal Resistance (𝑅 , )
, following additional details:
The design constraint, R , , , central to the cooling
system optimization, is calculated as: Core Data: Details the core's architecture, material, and
key characteristics like 𝐵 and Steinmetz parameters, vital
T, − P , , ∗R , −T for subsequent computations. The study opts for an EE core
R , , = (5) structure with N87 ferrite material to precisely estimate
P , ,
leakage inductance. Cost information is derived from the latest
Here, the 𝑅 , denotes the thermal resistance between Mouser listings as of March 1, 2024.
semiconductor case and heatsink, determined by thermal
interface material. The P , , is the maximum total loss Winding Data: Includes information on the winding
configuration, conductor material, and essential attributes
of the switches mounted on the same heatsink.
such as density, resistivity, and temperature coefficient,
c) Iterative Evaluation Steps necessary for ensuing calculations. Copper foil winding is
The optimization methodology iteratively evaluates each selected due to their superior copper fill factor and thermal
fan from the specified list against a spectrum of heatsink performance despite the slight increase in eddy current losses.
geometries within defined parameter ranges. For each unique Cost data is taken from the latest copper foil prices on [9] as
fan-heatsink pairing, the process includes: of March 1, 2024.

Volume Flow and Heatsink-to-Thermal Resistance Design space: Includes variables related to core geometry
(𝐑 𝐭𝐡,𝐬 𝐚 ) Calculation: The system calculates the volume and winding configurations, as detailed in Table IV and
flow and the R , for each unique fan-heatsink pairing showed in Fig. 4, essential for exploring different design
using analytical equations from [8]. possibilities and their impact on the transformer's
performance.
Configuration Validation: For the solutions where
R , <R , , , the volume, weight, and cost of the TABLE IV. TRANSFORMER DESIGN PARAMATERS
heatsink are then computed using the following relations: Core width Leakage width
[10-100] mm [10-50] mm
(𝐰𝐜𝐨𝐫𝐞 ) (𝐰𝐥𝐤)
V = d ∗ L ∗ w + (n + 1) ∗ h ∗ L ∗ t (6)
Core depth Foil width
[10-200] mm [0.1-2] mm
m =V ∗ρ (7) (𝐝𝐜𝐨𝐫𝐞) (𝐰𝐟𝐨𝐢𝐥 )

C =m ∗δ (8) Window height Primary Turn

[10-200] mm [3-50]
(𝐡𝐰𝐢𝐧𝐝 ) (Np)
Here, δ represents the cost per unit weight in $ /kg,
determined by examining the heatsink cost data, and ρ is the
density of aluminum (2.7 g/cm^3). Incorporating the fan cost b) Iterative Evaluation Steps
from Table III, sourced from Mouser as of March 1, 2024, we At this phase, the goal is to thoroughly assess each design
record the total cost for each configuration for later evaluation within the specified design space, using the core and winding
information to systematically evaluate transformer
configurations. This iterative process excludes any design that 4) Capacitor Model
does not meet the predefined standards, advancing only Due to the wide output voltage range operation, output
feasible options for in-depth examination and optimization. capacitors might be subjected to high-frequency ripple
currents with significant amplitudes. These currents can
Calculate and Evaluate 𝐁𝐦𝐚𝐱 and 𝐋𝐥𝐤,𝐭𝐱 : For each
significantly influence the capacitor's loss, volume, and cost,
potential transformer design, maximum magnetic flux density
impacting the overall system design. Moreover, to maintain
(B ) and leakage inductance (L , ) is calculated, ensuring system stability, it is essential to limit the voltage ripple across
that B stays below the core material’s saturation (B ) and the capacitor to a maximum allowable variation, denoted as
L , is within 10% of the specified leakage inductance, L . ∆V , . This threshold is set to maximum 5V for any voltage
Any design not complying with these parameters is excluded operation from 150 V to 1000 V. The necessary output
from further analysis. capacitance value, C , is determined as:
Calculate Core Loss: The core losses (P ) in this study 1
are calculated by using the improved generalized Steinmetz C = i (t)dt (13)
equation as described in [10] which allows to determine the 4∆V ,
core losses also for piece-wise linear waveforms. Here, i (t), shown in Fig. 2.a represents the instantaneous
Calculate Winding Loss: Winding losses in magnetic capacitor current, calculated as the difference between the
components, especially at higher switching frequencies, are rectified secondary current i , (t) and the ouput current i (t).
mainly dictated by skin and proximity effects that amplify Following the capacitance requirement identification, the
ohmic losses through non-uniform current distribution. The appropriate capacitors is selected. Film capacitors were
winding power losses per unit length (P ) in foil conductors chosen for their favourable characteristics in handling the
are calculated as described in [11].
described conditions. Cost estimation ( C ) for these
Calculate and Evaluate Temperature: Along with the capacitors is conducted using a curve fitting method based on
calculated core and winding losses, a detailed thermal model rated capacitance (C ) and voltage(V ), as outlined in reference
is essential to accurately compute the temperature distribution [13]. The method yields a cost function expressed as:
within the transformer. This research adopts the methodology
from [12], providing a comprehensive structure for evaluating C =a , ∗C +a , ∗V +a , (14)
temperatures at critical locations. It considers heat generation Furthermore, power losses (P ) within film capacitors,
from conduction, convection, and radiation, differentiating predominantly attributed to their equivalent series resistance
between core and winding loss impacts. By employing (R ) , are approximated by:
thermal resistance values, the model accurately forecasts
steady-state temperatures, verifying they comply with the P =R ∗I , (15)
maximum temperature limit T , .
To accurately assess these losses, the R estimation
Cost Calculation: Once it's confirmed that the design formula, tailored from [13], is:
parameters meet the predefined constraints, this phase
1 1
estimates the costs for the core and windings for each feasible R =a , ∗ +a , ∗ +a , (16)
set of parameters. C C ∗V
Core Cost ( C ): Calculated from the core volume The empirical coefficients a , , a , , a , in (14) and
(V ) using the formula: a , , a , , a , in (16) are extracted from the curve fitting
method by using the 1.2kV Kemet C4AQ-M capacitor series
V =2∗w ∗d ∗ (h +w +w ) (9) and their associated costs on Mouser as of March 1, 2024.
C =V ∗δ (10) G. Optimization Synthesis
Here, δ is the cost coefficient, expressed in $/m , which The optimization synthesis integrates all the data from the
is derived from a curve-fitting analysis of the costs associated component models, calculating total cost and loss for each
with N87 E cores as obtained from the core data. possible configuration of the system's design parameters. Each
configuration's data are scaled to the 350kW target output
Winding Cost (C ): Derived from the total volume of
power of the DC fast charger, which comprises N DAB
primary and secondary windings:
modules. Then, the normalization process is applied to all
V =l , ∗A , +l , ∗A , (11) data, eliminating any bias toward configurations with
inherently more or fewer modules, ensuring fair comparison
C =V ∗δ (12) in terms of cost and efficiency per unit of power provided.
where l , and l , represent the mean length for primary A Pareto front optimization then evaluates these
and secondary windings, and A , and A , are their normalized, scaled values to locate the most optimal designs,
respective cross-sectional areas. The cost coefficient δ , also which are visualized in Fig. 5. This figure illustrates the
expressed in $/m , is determined using curve fitting analysis inherent trade-offs between cost and loss, highlighting that
on cost of copper foils previously indicated in winding data. improvements in one often entail concessions in the other.
Therefore, the optimization does not converge on a singular
c) Store Solutions
optimal solution but rather identifies a set of potentially
In the final phase, all potential solutions that have optimal configurations along the Pareto front.
successfully passed the previous evaluations, along with their
associated cost and loss data, are aggregated and forwarded to
the optimization synthesis block.
 Fig.6. Performance and Cost Breakdown for 50kW DAB Converter Unit.
(a) Efficiency comparison at light load (10% of full power, depicted with
lower color intensity) and full load across various voltage levels (b)
Fig.5. Pareto Front Optimization with Color Map Indicating Converter Breakdown of the converter's cost components.
Module Counts Necessary to Reach 350 kW. drop at light loads, attributed to the SPS modulation strategy
used. Future work should investigate alternative modulation
III. OPTIMIZATION RESULTS techniques to overcome this issue, further optimizing DAB
The Pareto front optimization, illustrated in Fig. 5, converter performance for all load conditions.
provides a critical understanding of the trade-offs between REFERENCES
loss and cost for different DAB converter configurations in the
[1] “Global EV Outlook 2023,” International Energy Agency (IEA), Paris,
350 kW DC fast charger system. Configurations with a higher France, Tech. Rep.,2020. [Online]. Available: https://www.iea.org
power rating per module (resulting in a lower N) align more /reports/global-ev-outlook-2023. [Accessed: Mar. 14, 2024].
closely with the Pareto front, signifying a more optimal [2] C. Jung, "Power Up with 800-V Systems: The benefits of upgrading
balance between loss and cost. On the other hand, voltage power for battery-electric passenger vehicles," in IEEE
configurations with lower power ratings per module, like 10 Electrification Magazine, vol. 5, no. 1, pp. 53-58, March 2017.
kW (N=35), are situated further from the Pareto front, [3] S. Rivera et al., "Electric Vehicle Charging Infrastructure: From Grid
indicating suboptimal performance. to Battery," in IEEE Industrial Electronics Magazine, vol. 15, no. 2, pp.
37-51, June 2021.
To further refine the selection process and adhere to [4] D. Lyu, C. Straathof, T. B. Soeiro, Z. Qin and P. Bauer, "ZVS-
industry standards, we introduce an additional benchmark Optimized Constant and Variable Switching Frequency Modulation
criterion: achieving at least 95% efficiency across a broad Schemes for Dual Active Bridge Converters," in IEEE Open Journal of
Power Electronics, vol. 4, pp. 801-816, 2023.
output voltage range at rated power—a well-established target
in DC fast charger systems. Among the configurations [5] O. Zayed, A. Elezab, A. Abuelnaga and M. Narimani, "A Dual-Active
Bridge Converter With a Wide Output Voltage Range (200–1000 V)
examined, the design employing seven 50 kW DAB modules for Ultrafast DC-Connected EV Charging Stations," in IEEE
with the switching frequency of 30 kHz and the transformer Transactions on Transportation Electrification, vol. 9, no. 3, pp. 3731-
turns ratio of 0.9, stands out for meeting this efficiency 3741, Sept. 2023.
threshold while also optimizing cost. [6] D. Lyu, T. B. Soeiro and P. Bauer, "Design and Implementation of a
Reconfigurable Phase Shift Full-Bridge Converter for Wide Voltage
This selected design's performance, showcased in Fig. 6.a, Range EV Charging Application," in IEEE Transactions on
maintains an efficiency above 95% at full power across Transportation Electrification, vol. 9, no. 1, pp. 1200-1214, March
varying operational conditions. However, the average 2023.
efficiency across all conditions, including both 100% and 10% [7] Charging Interface Initiative e.V., "DC CCS Power Classes V7.2,"
Dec. 9, 2021. [Online].Available:https://www.charin.global/media/
of full power within a wide voltage range, falls to around 80%. pages/technology /knowledge-base/c6574dae0e-1639130326/charin_
The SPS modulation technique mainly causes this efficiency dc_ccs_power_classes.pdf. [Accessed: Mar. 14, 2024].
reduction due to the wide voltage and load variations. [8] C. Gammeter, F. Krismer and J. W. Kolar, "Weight optimization of a
cooling system composed of fan and extruded fin heat sink," 2013
The cost breakdown, detailed in Fig. 6.b, highlights the IEEE Energy Conversion Congress and Exposition, Denver, CO, USA,
main contributors to the converter's cost. The largest part 2013, pp. 2193-2200.
comes from the transformer, followed by costs for switches, [9] Basic Copper, "Copper Sheet & Rolls," Basic Copper. [Online].
heatsinks, and capacitors. The selected 30kHz switching Available: https://basiccopper.com/copper-sheet--rolls.html.
frequency is a strategic choice to maintain the system [Accessed: Mar. 14, 2024].
efficiency above the 95% threshold at rated power while [10] K. Venkatachalam, C. R. Sullivan, T. Abdallah and H. Tacca,
concurrently optimizing cost. This analysis acknowledges that "Accurate prediction of ferrite core loss with nonsinusoidal waveforms
while higher switching frequencies can potentially lower the using only Steinmetz parameters," 2002 IEEE Workshop on
Computers in Power Electronics, 2002. Proceedings., Mayaguez, PR,
transformer cost, the priority was to avoid compromising USA, 2002, pp. 36-41
efficiency. [11] J. Mühlethaler, “Modeling and multi-objective optimization of
inductive power components”, Ph.D. dissertation, ETH Zürich,2012.
IV. CONCLUSION [12] M. Bahmani, “Design and Optimization Considerations of Medium-
This study identifies an optimal design configuration for Frequency Power Transformers in High-Power DC-DC Applications,”
DAB converter in 350 kW DC fast chargers, achieving over PhD thesis, Chalmers University of Technology, Gothenburg, 2016.
95% efficiency at rated power across various voltages. It [13] D. Christen, “Analysis and Performance Evaluation of Converter
Systems for EV-Ultra-Fast Charging Stations with Integrated Grid
illustrates the substantial impact of optimizing DAB Storage” , Ph.D. dissertation, ETH Zürich,2017.
converters on the efficiency and cost-effectiveness of fast
charging stations. A key limitation identified is the efficiency