Beruflich Dokumente
Kultur Dokumente
Final Report
A 10KW Fuel Cell Inverter System
Submitted by
Faculty Advisor
TABLE OF CONTENTS
1. Introduction
2.
3.
4.
Management
2.1
Team Organization
2.2
Education Impact
2.3
2.4
Project Budget
3
4
5
5
Topology Evaluation
3.1
Two Topologies
3.2
3.3
Cost Evaluation
3.4
Efficiency Evaluation
Design Rationale
12
4.2
Inverter
4.3
14
14
21
28
5. Simulation
37
41
6.
Experimental Result
7.
Performance Evaluation
8.
Bill of materials
9.
Cost Analysis
10.
Conclusion
11.
Reference
43
46
47
47
49
ii
50
List of Figures
Fig. 2.1
Fig. 3.1
Fig. 4.1
14
Fig. 4.2
15
Fig. 4.3
15
Fig. 4.4
20
Fig. 4.5
21
Fig. 4.6
24
Fig. 4.7
25
Fig. 4.8
Display of RS-232
36
Fig. 4.13
39
Fig. 4.14
40
Fig. 4.15
41
Fig. 5.1
Simulated waveforms
42
Fig. 6.1
44
Fig. 6.2
45
Fig. 6.3
46
iii
List of Tables
Table 2.1
Project budget
Table 3.1
Table 3.2
Table 4.1
Table 4.2
Table 4.3
Table 4.4
Table 7.1
Table 8.1
Bill of Materials
48
Table 9.1
48
Table 9.2
Table 9.3
19
Appendices
Appendix
A.1 : Schematic for the sensing board
A.2 : Schematic for the sensing and protection
A.3 : DSP board
A.4 : Inverter gate driver
B
iv
51
Student Members
Minsoo Jang
Min Koo
Jaehyuck Jung
Sangmin Jung
Taehoon Kim
Jinwook Oh
Hyunjung Kim
Namki Lee
Jinhee Lee
Joonseo Lee
Jinsang Jo
Kangsuk Lee
Minkook Kim
Byungsoo Nho
Seungjoo Cheon
Summary
The objective of the 2003 Future Energy Challenge competition is to develop a low cost
10kW power processing unit for a fuel cell system. The SNUT team, which is composed of
senior undergraduate students, graduate students and faculty advisors, has been launched for the
competition.
This report discusses the power circuit topologies for the SNUT inverter system by evaluating
the topologies in a practical way. After researching several topologies a topology is chosen and
the component ratings are designed along with through analysis on the chosen topology. The
simulation is performed to verify the design and control of the proposed topology. A hardware
prototype capable of supplying 10kW load was built and tested in the laboratory of SNUT.
Experimental performances on some design items are compared to minimum target
requirements of the inverter system. The cost analysis is done based on the spreadsheets
evaluation forms provided in the 2003 FEC workshop. Some conclusions are made to meet the
minimum target requirement in the final competition.
1. Introduction
The environmental concern is now the driving force for alternative energy. Fuel cell power
generation systems are expected to see increasing practical use due to the several advantages
over conventional generation systems. These advantages include 1) low environmental pollution
2) highly efficient power generation 3) diversity of fuels(natural gas, LPG, methanol and
naphta) 4 ) reusability of exhaust heat 5) modularity and 6) faster installation [1]. Fuel cells are
generally characterized by the type of electrolyte that they use. Solid oxide fuel cells (SOFC)
have grown in recognition as a viable high temperature fuel cell technology. The most striking
quality of SOFCs is that the electrolyte is in solid state and is not a liquid electrolyte. The high
operating temperature up to 1000C allows internal reforming, promotes rapid kinetics with
non-precious materials and produces high quality byproduct heat for cogeneration or for use in a
bottoming cycle. A number of different fuels can be used from pure hydrogen to methane and
carbon monoxide. The major advantage of SOFC lies in its efficiencies raging from 55 to 60%
[2].
In general, the function of a power conditioning system in a fuel cell generation system is to
convert the DC output power from the fuel cell to regulated AC power. There may be two stages
of power converters. A DC-DC converter converts the low voltage DC output from the fuel cell
to a level at which an inverter can safely operate. The inverter is used to invert the DC output
from the DC-DC converter to a suitable AC voltage. The power conditioning unit that basically
consists of an inverter is required to have the following characteristics: 1) allowable for wide
output voltage regulation of fuel cell 2) controllability of output voltage 3) available for isolated
operation and line parallel operation 4) fast reactive power dispatch 5) low output harmonics 6)
high efficiency and 7) suitable for high power system [3]. Fuel cell production costs are
currently decreasing and have nearly achieved energy costs that are competitive with local
utility rates. The inverter cost must also decrease while at the same time increasing efficiency,
reliability, and power quality levels. The cost reduction of the power processing unit will enable
the fuel cell systems to penetrate rapidly into the utility market.
The objective of the Fuel Cell Inverter Challenge is to develop a 10kW low-cost power
processing systems that support the commercialization of a SOFC power generation system to
provide non-utility and ultra-clean residential electricity. The target cost of the stand-alone
10kW power processing unit will be less than $40/kW in high volume. Further, emphasis will
also be placed on high energy efficiency as this has direct impact of size and cost of the SOFC
system and overall system fuel efficiency. Another key objective of this competition is to
promote design education in the undergraduate curriculum at Seoul National University of
Technology in conjunction with faculties and industry experts in the power electronics and
energy conversion area.
2. Management
2.1 Team Organization
The Seoul National University of Technology (SNUT) has formed a multi-disciplinary team
consisting of nine undergraduate and six graduate(master) students and three faculty advisors.
The undergraduate students from Department of Control and Instrumentation Engineering(CIE),
Department of Electrical Engineering(EE), and Department of Mechanical Design(MD) have
been selected by public notice in our university. The students are divided into six sub-teams (1)
Inverter and DSP (2) Front-end DC-DC converter (3) Bi-directional DC-DC converter and
Battery control (4) Sensors and Protection (5) System integration and Interface (6) Heat sink
and Packaging.
approach will allow the team to address thermal management, packaging and case issues.
Dr. Sewan Choi, professor in the power electronics area, served as the lead Faculty Advisor
for the team. Other faculty advisors to this project include: Dr. Kiyong Kim with expertise in
control systems and Dr. Youngseog Kim, professor in the mechanical design area, with expertise
in heat sink design and packaging.
Faculty Advisors
Dr. Sewan Choi (CIE)
Fig. 2.1
to take all the courses necessary to complete the project. The 2003 Future Energy Challenge
provided a good topic for the undergraduate students to participate in the competition. All the
undergraduate students on the team will receive credits in Design Project courses(CIE322 and
CIE415).
The SNUT team has been holding weekly meetings with all of its members to discuss the
project. We have had a series of technical seminars mostly by ourselves to get the practical
background on design of the fuel cell power processing system. Each of graduate student led the
seminar on some specific topic. Some experts from industries had been sometimes invited to
give students a practical knowledge.
These kinds of efforts have been shown to be successful in attracting students. In fact, in the
spring semester of 2003 two student members, Jinhee Lee and Jinsang Jo, joined the graduate
program in power electronics of SNUT and continue participating this 2003 FEC competition.
In the spring semester of 2004 four student members will apply for the graduate program in
power electronics.
2.3 Project Timeline
The project time line for the SNUT team is sown in Appendix B.
construct the prototype in two steps ; the preliminary prototype by the mid of November 2002
and the final prototype by the end of February 2003. In November 2002, the team submitted the
written report and presented working prototype as a graduation requirement. The preliminary
prototype was evaluated by peer reviews from faculty advisors and industry experts. The team
incorporated feedback from reviewers and any design changes to final prototype.
2.4 Project Budget
The project budget for the SNUT Future Energy Challenge team is shown in Table 2.1.
The
budget only takes into account parts for the power processing system and travel for fundraising
and the competition excluding any labor and equipment. The undergraduate students take part in
the competition as a choice of topic for the Design Project course and the graduation
requirement. Also, the department of CIE, SNUT already has laboratory facilities equipped with
many power electronics instruments and equipments.
Table 2.1 Project budget for the SNUT Future Energy Challenge team
Classification
Amount
Power Device
(IGBT, MOSFET, DIODE)
$ 2,000
Battery
$ 700
DSP Evaluation
$ 500
PCB
$ 1,500
Sensors
$ 700
Inductors
$ 800
Capacitors
$ 500
Transformer
$ 2,000
$ 1,000
$ 2,000
Company Presentations
$ 1,000
Work shop
Final Competition
$ 5,000
$ 10,000
Copies
$ 300
Lab supplies
$ 500
Parts
Travel
Miscellaneous
Sub Total
$ 28,000
Support from our university and industries is essential to the teams successful project
performance not only for financial sponsorship, but also for industrial experience. The SNUT
team faculty advisors and students have been trying to secure the necessary funding for this
project from the school, industries including national laboratories. The SNUT team secured the
sponsorship and commitment from the school and some industries, and has been trying solicit
additional support for the proposed project.
3. Topology Evaluation
3.1 Two Topologies
The SNUT team decided to use the low voltage (48V) battery, which is supposed to be
provided at the competition test site, as a secondary energy storage to supply transient loads.
The SNUT team considered two types of power circuit topologies for the SOFC power
processing system : Scheme I shown in Fig. 3.1(a) and Scheme II shown in Fig. 3.1(b). In
Scheme I, as shown in Fig. 3.1(a), the DC voltage from the fuel cell, 29VDC nominal, is first
boosted to 48V via a non-isolated boost converter. The 48V battery bank in the fuel cell system
is connected to the 48V DC link so that power flow to and from the battery is controlled by the
current control of the boost converter. The 48VDC from the boost converter is then converted to
400VDC via an isolated high frequency DC-DC converter. The high frequency DC-DC converter
could be push-pull, half-bridge or full-bridge types. The full-bridge type is considered suitable
for 10KW of high frequency DC-DC conversion.
D1
L1
L2
S2
D2
S3
S6
Fuel Cell
22 ~ 41V
S1
S7
C2
T1
L4
D3
C4
120Vac
60HZ
Vdc
400v
240Vac
60HZ
C1
Battery
48V
S8
S4
D4
S5
D5
S9
C5
C3
120Vac
60HZ
L3
L5
(a) Scheme I
L3
S1
D2
D1
S2
L1
S5
S6
C1
T1
D3
C3
D4
120Vac
60HZ
Vdc
400V
C in
Fuel Cell
22 ~ 41 V
D5
S3
D6
L2
S7
S4
S8
C4
C2
D7
S11
S12
S13
S14
T2
L5
S9
S10
120Vac
60HZ
D8
Battery
48V
120Vac
60HZ
(b) Scheme II
Fig. 3.1 Proposed power circuit topologies
L4
The DC-DC conversion stage includes a transformer isolation for safety, protection and to
meet the stringent FCC Class-A standards. The 400V DC-DC converter output is then converted
to 120V/240V, 50/60 Hz, single-phase AC by means of a PWM inverter stage. An output L-C
filter stage is employed to reduce the ripple component and draw a low THD AC waveform.
Fig.3.1(b) shows an alternative power circuit configuration (Scheme II) for the SOFC inverter
system. The DC voltage from the fuel cell , 29VDC nominal, is first converted to 400VDC via an
isolated high frequency DC-DC converter. The high frequency DC-DC converter could be pushpull, half-bridge or full-bridge types. The full-bridge type with two diode bridges connected
in series at the secondary was chosen as a 5KW of front-end DC-DC conversion. The DCDC conversion stage also includes a transformer isolation for safety, protection and to meet the
stringent FCC Class-A standards as well. The 48V battery bank in the fuel cell system is
connected to the 400V DC link via a bi-directional DC-DC converter which is also operated at
high frequency. The low voltage(48VDC) battery side and the high voltage(400VDC) dc link side
of the bi-directional DC-DC converter could be current-source full-bridge type and voltagesource full-bridge type, respectively, or vice versa. A current-sourced push-pull type using
MOSFETs and a voltage-sourced full-bridge type using IGBTs were chosen at the battery
side and the dc link side, respectively. The 400V DC-DC converter output is then converted to
120V/240V, 50/60 Hz, single-phase AC by means of a PWM inverter stage. An output L-C filter
stage is employed to reduce the ripple component and draw a low THD AC waveform.
The common topology chosen for both of the schemes to provide the split phase output
was two half-bridge inverters. Both of the topologies have thoroughly been examined from
cost and efficiency standpoint, and one of them was adopted for this project.
3.2 Power Component Design
In this section power components of the two schemes are designed so that the designed
values are used to compare cost and efficiency of the two schemes. The output displacement
factor is assumed to be unity for design of the fuel cell power processing system rated at
10KW. Table 3.1 shows some system parameters for power circuit design of the two
schemes.
Table 3.2 and 3.3 list the designed ratings of the power components in the two schemes
using system parameters in Table 3.1. Based on the designed values, actual devices were
selected from some manufactures. Appropriate safety margins were considered for actual
device selection. The design procedure and device selection presented here is not unique,
but they could be used to relatively compare both of the schemes from cost and efficiency
standpoint.
3.3 Cost Evaluation
In this section the two schemes are compared each other from a cost standpoint. The
spreadsheets evaluation forms presented at the 2003 FEC workshop is used to perform
relative cost estimates. Cost factors have been obtained based on the designed vales, shown
in Table 3.2 and 3.3, without safety margin. The resulting cost estimates of the two schemes
is shown in Table 3.4. It can be noticed from Table 3.4 that power switches, capacitors,
transformers employed in both schemes did not give much difference in cost. However, the cost
of the inductor in the non-isolate boost converter section of Scheme I is significant due to its
high current capacity. Therefore, it can be concluded that Scheme II is superior to Scheme I in
cost.
Table 3.1 System parameters for power circuit design of the two schemes.
Scheme I
Scheme II
Non-isolated
Boost
Isolated
DC-DC
Inverter
Front-end
DC-DC
Inverter
Bi-directional
DC-DC
Switching
frequency
40kHz
20kHz
20kHz
25kHz
20kHz
20kHz
Input
voltage
22 ~ 41V
42 ~ 57.6V
400V
22 ~ 41V
400V
42 ~ 57.6V
Output
voltage
42 ~ 57.6V
400V
60Hz 120VAC
(Split phase)
400V
60Hz 120VAC
(Split phase)
400V
Section
Non
-isolated
Boost
Isolated
DC-DC
Inverter
Component
Designed value
MOSFET
(S1)
Vpeak (V)
57.6
Irms (A)
214.4
Diode
(D1)
Vpeak (V)
57.6
Irms (A)
266.2
Inductor
(L1)
Inductance
50 uH
Irms (A)
272.5
Capacitor
(C1)
Capacitance
3300uF
Vpeak (V)
77.7
MOSFET
(S2, S3, S4, S5)
Vpeak (V)
57.6
Irms (A)
336.66
Transformer
(T1)
Vrms (V)
47.3
Irms (A)
389.8
Diode
(D2 ~ D9)
Vpeak (V)
410
Irms (A)
13.75
Inductor
(L2, L3)
Inductance
100 uH
Ipeak (A)
27.3
Capacitor
(C2, C3)
Capacitance
5000 uF
Vpeak (V)
210
IGBT
(S6, S7, S8, S9)
Vpeak (V)
420
Irms (A)
50
Inductor
(L4, L5)
Inductance
93 uH
Irms (A)
42A
Capacitor
(C4, C5)
Capacitance
16uF
Digital Tech
Vpeak (V)
170V
10
IXYS
DSEI2x161-12P
(1200V, 2X128A, trr = 35ns)
MAGNETICS
43208 (EI)
Samwha
SZ2A338M35100
(100V, 3300uF, ESR 0.08)
IXYS
IXFN340N07
(70V, 340A, 4m)
MAGNETICS
49925 (U)
IXYS
DESI 30-10A
(1000V, 30A, trr = 50ns)
Chang-sung
CH270125E (Toroid)
Samwha
GF2G688M76160
(400V, 6800uF, ESR 0.04)
TOSHIBA
MG50Q2YS50
(600V, 50A, VCE(sat) 2.7V)
Chang-sung
CH572060E (Toroid)
Component
MOSFET
(S1, S2, S3, S4)
Transformer
(T1)
Front-end
DC-DC
Diode
(D1 D8)
Inductor
(L1, L2)
Capacitor
(C1, C2)
IGBT
(S5, S6, S7, S8)
Inverter
Inductor
(L3, L4)
Capacitor
(C3, C4)
MOSFET
(S9, S10)
Bi Directional
DC-DC
Transformer
(T2)
Inductor
(L5)
IGBT
(S11,S12, S13,S14)
Designed value
Vpeak (V)
41
IXYS
IXFN340N07
Irms (A)
177.4
Vrms (V)
47.3
Irms (A)
389.8
Vpeak (V)
410
Irms (A)
8.8
Inductance
100 uH
Irms (A)
12.6
Capacitance
5000 uF
Vpeak (V)
210
Vpeak (V)
420
Irms (A)
50
Inductance
93 uH
Irms (A)
42A
Capacitance
16uF
Digital Tech
Vpeak (V)
170V
Vpeak (V)
140
APT
APT20M20JFLL
Irms (A)
77.5
Vrms (V)
58
Irms (A)
73.3
Inductance
40 uH
Irms (A)
113
Vpeak (V)
420
Irms (A)
13.1
11
MAGNETICS
49925 (U)
IXYS
DESI 30-10A
(1000V, 30A, trr = 50ns)
Chang-sung
CH270125E (Toroid)
Samwha
GF2G688M76160
(400V,6800uF, ESR 0.04)
TOSHIBA
MG50Q2YS50
(600V, 50A, VCE(sat) 2.7V)
Chang-sung
CH572060E (Toroid)
MAGNETICS
49925 (U)
Chang-sung
CH572060E (Toroid)
TOSHIBA
MG50J2YS50
(600V, 50A, VCE(sat) 2.7V)
Table 3.4 Cost estimates of the two schemes according to FEC 2003 cost spreadsheet
Scheme I
Component
MOSFET
Scheme II
Desig.
Qty
Cost
Factor
Unit
Cost ($)
Extended
Cost ($)
Desig.
Qty
Cost
Factor
Unit
Cost ($)
Extended
Cost ($)
S1
12349
10,44
10.44
S1 ~ S4
7273.4
7.71
30.85
S2 ~ S5
22419
14.75
50.01
S9, S10
10850
9.64
19.27
D1
15333
3.45
3.45
D1 ~ D8
3608
2.38
9.55
D2 ~ D5
5637
2.57
10.30
S5 ~ S9
21000
8.43
33.7
S5 ~ S8
21000
8.45
33.7
S11~S14
5502
2.3
9.2
T1
18437
24.7
24.7
T2
4251.4
10.94
10.94
Diode
IGBT
T1
18437
26.55
26.55
Transformer
Inductor
Capacitor
L1
3.69
249.51
249.51
L1, L2
0.01
41.8
83.7
L2, L3
0.07
44.5
89
L4, L5
0.16
49.81
99.62
L4, L5
0.16
49.81
99.62
L5
0.5
68.7
68.7
C1
19.92
2.85
2.85
C1, C2
220
30.63
61.27
C2, C3
220
30.63
61.26
C4, C5
0.46
0.16
0.32
C4, C5
0.46
0.16
0.32
Total Cost
636.97
451.82
12
considered for inductor since copper loss dominates when an inductor is operated as a filter
at high frequency. Capacitor loss is calculated using ESR of the selected capacitor.
The efficiency of the both schemes can be obtained based on the power loss estimates in
Table 3.5. The estimated efficiencies are 85.4% for Scheme I and 94.6% for Scheme II,
respectively. Losses in the diode and the MOSFET in the non-isolate boost converter section of
Scheme I are significant. The losses in the two switches exceed 50% of the total loss of Scheme
I. Scheme II is superior to Scheme I in both cost and efficiency. In conclusion, the SNUT team
decided to choose Scheme II for the power circuit topology.
Table 3.5 Power loss estimates of the two schemes based on manufactures datasheet
Scheme I
Device
Switching loss
0.012
Conduction loss
63.3
Switching loss
29.8
Conduction loss
36.7
Switching loss
29.4
Core loss
15.9
Copper loss
24.87
Copper loss
0.074
Core loss
1.65
Copper loss
0.007
C1, C2
Capacitor loss
0.0036
C3, C4
Capacitor loss
69.2
Switching loss
0.875
Conduction loss
20.62
Switching loss
0.024
Conduction loss
122.1
Switching loss
40.69
Conduction loss
63.3
Switching loss
29.8
Conduction loss
36.7
Switching loss
29.4
Core loss
31.84
Copper loss
45.77
L1
Copper loss
7.77
L1, L2
L2, L3
Copper loss
0.148
L3, L4
Core loss
1.65
Copper loss
0.007
C1
Capacitor loss
29.15
C2, C3
Capacitor loss
0.15
C4, C5
Capacitor loss
69.2
S1
MOSFET
S2 ~ S5
S6 ~S9
T1
L4, L5
Total Loss
13.2
335.6
D2 ~ D5
Capacitor
Conduction loss
Conduction loss
Diode
Inductor
Extended
Loss (W)
Extended
Loss (W)
D1
Transformer
Loss
Loss
Desig.
IGBT
Scheme II
864.8
13
Desig.
D1 ~ D8
S1 ~ S4
S5 ~ S8
T1
284.12
4. Design rationale
According to the topology evaluation in Section 3, the SNUT team decided to choose
Scheme II, shown in Fig. 3.1(b) , for the power circuit topology. The block diagram for the
SNUT fuel cell inverter system is shown in Fig. 4.1. The inverter system consists of a front
end DC-DC converter, a DC-AC inverter and a bi-directional DC-DC converter. Both the
fuel cell current control and the dc link voltage control are performed for the front end DCDC converter to improve dynamic performance of the system during a transient state.
The bi-directional DC-DC converter is operated to charge or discharge the battery according
to the current reference and the mode of operation determined by the DSP. The two PWM
controllers are employed for charge and discharge modes of operation, respectively.
IFC
Fuel Cell
22~41V
IBatt
IBatt
VBatt
Idc
Io
Inverter
Vdc
+ Vo
-
DSP
320LF2407
UCC 3895
48V
400V
VFC
IFC
VFC
Idc
Vdc
Battery
Front-end
DC-DC
Load
120V/240V
60HZ
AC output
Io
Vo
Bi-directional
DC-DC
VBatt
UCC 3895
&
UCC 3825
Fig. 4.1 Block diagram of the SNUT fuel cell inverter system
4.1 Front end DC-DC converter
A Front end DC-DC converter is required to boost a unregulated fuel cell voltage of 29V
nominal to a regulated 400V, as shown in Fig. 4.2. The full-bridge type is a topology of
choice with which a phase-shift PWM technique can be implemented.
14
I L + VL +
S1
D1
S2
Fuel Cell
22 ~ 41 V
Vin
C in
Vpri2
D3
C1
D4
Vsec1
Vsec2
I dc
L1
Vd
T1
Vpri1
D2
D5
D6
D7
D8
Vdc
400V
L2
N p : Ns
S3
S4
C2
15
The reduce duty loss also reduces turns ratio of the transformer. This reduces the voltage
rating of diodes in the secondary side and current rating of MOSFETs in the primary side.
Fig. 4.3 shows the main waveforms of the front-end DC-DC converter.
From the inductor voltage VL an equation can be written as,
Ns 1
DT 1
T (1 D)
Vdc )
= Vdc
Np 2
2
2
2
(Vin
(1)
Therefore, the duty cycle of the proposed front-end DC-DC converter is obtained by,
D=
N p Vdc
(2)
4 N s Vin
According to eqn.(2), the duty cycle ranges 0.24 to 0.45 to regulate the dc link voltage of
400V when the fuel cell voltage varies between 22V and 41V.
4.1.1
The power components of the front end DC-DC converter are designed in this section
with the following system parameters.
I dc =
Pdc
= 12.5 A
Vdc
(3)
DT
, the inductor voltage becomes,
2
N
1
VL = Vin s Vdc
Np 2
16
(4)
(1 2 D ) (
L=
Ns
) Vin D
Np
(5)
I f s
Assuming a permissible ripple current on the inductor to be 50% of the maximum current
at the dc link or 6.25A peak to peak, the inductance can be calculated using eqn.(5) to be L
= 100 uH. The peak inductor current can also be calculated as,
1
I L, peak = I dc + I = 15.6 A
2
(6)
From Fig. 4.3, the rms value of inductor current can also be calculated as,
I L,rms = 12.6 A
Power switch design
The voltage rating of diodes and MOSFETs are calculated based on the moment at which
the fuel cell outputs a maximum of 41V at a minimum duty cycle of D = 0.24. The current
rating of diodes and MOSFETs are calculated based on the moment at which the fuel cell
draws a maximum current of 275A at a maximum duty cycle of D = 0.45.
a. Diodes
When the fuel cell voltage is 41V, the secondary winding voltage becomes 410V,
therefore, the peak voltage of a rectifier diode is 410V. A safety margin should be
considered due to the ringing phenomenon at the secondary winding of the high frequency
transformer. The voltage rating of the diode is determined to be 600V. A ultra-fast recovery
diode is chosen to lower the switching loss due to the high switching frequency
operation(25kHz). As the diode always conducts half of a switching cycle, the average
current rating of the diode can be obtained by,
I D ,av =
1
I dc = 6.25 A
2
(7)
The peak current of the diode is identical to the peak current of the inductor. Then,
(8)
17
A ultra-fast recovery diode with a rating of 600V, 20A was selected from a manufacture.
b. MOSFETs
Power MOSFETs were selected as a switching device for the front-end DC-DC converter
since they should operate under the low voltage and high current condition. The peak
voltage of a MOSFET is 41V which is the maximum fuel cell voltage. Considering a safety
margin due to voltage spikes originated from the leakage inductance of the high frequency
transformer, the device voltage rating over 70V should be acceptable. By multiplying the av
and peak value of the diode by the turns ratio of the transformer, the av and peak values of
the MOSFET can be calculated be 125A and 275A, respectively. Considering safety
margins, MOSFETs with ratings of 70V, 340A were selected from a manufacture.
Transformer Design
As mentioned at the above, two 2.5KW high frequency transformers are employed and the
transformer design for the 2.5KW transformer is described in the following.
a. Core material
Ferrite is an ideal core material for transformers and inductors in the frequency range 20 KHz
to 3MHz, due to the combination of low core cost and core losses. Ferrite core is chosen as a
material of a high frequency transformer.
b. Core size
The power handling capacity of a transformer core can be determined by its area product
WaAc, where Wa is the available core window area, and Ac is the effective core cross-sectional
area. The SNUT team would follow the procedure for transformer core size selection provided
by Magnetics Co [8].
The area product is given by,
Wa Ac =
Pdc C 108
4 e B fs K
(9)
where Pdc is output power, C is current capacity, e is transformer efficiency, B is flux density,
18
Minimum
Area
Materials
Combination
2300
K
(min.)
2500
R
(min.)
3000
P
25%
5000
F
(min.)
Ie
Ae
(cm)
(cm2)
2,440
2,650
4,240
23.1
3.39
EER
Ve
SET
NOM.
Wt.
(cm2)
(cm3)
(gms)
3.14
78.6
396
WaAc
(cm4)
34
fs is switching frequency and K is winding factor. The core type of choice is EER core, then
C = 5.07 103 cm2 / Amp
(10)
The transformer efficiency is assumed to be 90%. The winding factor is K = 0.3 (primary side
only). The flux density B is assumed be 4500(gauss). Then, the WaAc product is calculated as,
Wa Ac = 23.47cm 4
(11)
Using the core selection table by area product distribution (refer to Appendix C), the core of
47054-EC was selected. Magnetic data for the selected core is shown in Table 4.1.
c. Number of turns
Once a core is chosen, the calculation of primary and secondary turns and their wire sizes are
readily accomplished. The number of primary turns is given by [8],
Np =
VP 108
41108
=
= 6.02turns
4 B A f s 4 2000 3.39 25000
(12)
Here, Vp is the peak primary voltage is and A is the cross-sectional area of the core, which are
given in Table 4.1. Considering duty loss of 20% at the secondary winding of the transformer
originated from the leakage inductance, the final number of turns for primary and secondary
windings are determined to be,
Np : Ns = 6 : 90
d. Wire size
The wire size of the transformer windings is calculated based on the rms value of the winding
current. Since the winding current is 2 times larger than the switch current at each side of the
transformer, the rms vales of the primary and secondary winding currents can be calculated to
19
be 187A and 12.5A, respectively. Then, at 500 circular mils per rms ampere the required number
of circular mils is obtained by [9],
Primary circular mil requirement
Current
Ref.
Voltage
Controller
Current
Controller
Fuel Cell
DC Power Input
Front end
DC-DC Converter
Current limiter
DC link Voltage
Fig. 4.4 Control block diagram for front end DC-DC converter
20
To Inv
SW3
C1
SW1
Lf
I SA
Ia
Lf
400VDC
SW4
Ib
Vab
A
B
SW2
Vb
C2
Cf
Va
Cf
N
21
The inverter system consists of two half-bridge inverters, utilizing center tapped dc link
capacitors followed by output filters. The front-end DC-DC converter maintains equal 200V
on the dc-link capacitors, and two inverter legs are operated to generate a split single-phase
120/240Vac, 60Hz output.
A low cost DSP is implemented to provide the control scheme for the inverter system. A
digital PI compensator is employed to regulate the output voltage under varying load condition.
4.2.1
Inverter design
In this section the ratings of the power components in the inverter system are determined. A
detailed list of the inverter requirement for our inverter design is,
- 5kW continuous @ displacement factor 0.7 leading or lagging, 10kW overload for 1 min.
- Output voltage : 120/240V(split-phase).
- Output frequency : 60Hz0.1Hz.
- Output voltage THD : less than 5% when supplying a stand nonlinear test load.
- Output voltage regulation quality : output voltage tolerance no wider than 6%.
DC Link Capacitors
For a worst case, 10KW overload for 1 minute at displacement factor of 0.7 is considered,
then the output VA becomes,
VAout =
10000
= 14280VA
0.7
(13)
I a , rms =
14280
= 59.5 A
2 120
(14)
For the sake of simplicity, the output current ia is assumed to consist of only fundamental (Ia,1)
and third harmonic (Ia,3). Further, assuming Ia,3 = 0.7Ia,1 since this is a typical case of a single
phase rectifier type nonlinear load [5],
2
I a . rms I a ,1 + I a ,3 = 1.22 I a ,1
(15)
Therefore, the fundamental rms value of each phase output current becomes,
22
I a ,1 =
59.5
= 48.77 A
1.22
(16)
The most dominant component of the DC-link capacitor current ic1 is the fundamental
frequency current, the rms value of which equals,
I c1,1
1
I a ,1 = 24.3 A
2
(17)
For a permissible voltage ripple Vc1 less than 10% or 20V, capacitance can be obtained by,
I c1,1
24.3
=
= 3222 F
Vc1 2 60 20
C1 =
(18)
1
1
Vc1, peak = VDC + Vc1 = 210V
2
2
(19)
Based on these designed values, an actual device of 250V 3300uF was selected from a
manufacture.
Inverter switches
The av current rating of an inverter switch can be obtained by,
I SA,av = 28 A
(20)
The peak voltage rating of each IGBT is 420V which is the peak dc link voltage.
Based on these designed values, IGBTs with 600V 50A rating were selected from a manufacture.
Output filter design
Assuming the switching frequency fs to be 20kHz, the frequency ratio is,
n=
fs
= 333.3
f1
(21)
An equivalent circuit for output filter design is shown in Fig. 4.6. The transfer function Hn for
the equivalent circuit can be obtained by,
Hn =
Va , n
jX c Z L , n
=
Vin nX L X c + jZ L , n (n 2 X L X c )
23
(22)
jnX L
Vin
Ia
-jXc
n
Va,h
ZL,n
X L X c
(23)
Hn =
Xc
1
=
n X L X c n2 X L 1
Xc
(24)
Therefore, to satisfy THD requirement of less than 5%, only the switching frequency
component is considered as [5],
1
0.045
2 XL
1
n
Xc
(25)
To limit the ripple voltage across the filter capacitor generated from the third harmonic load
current, an equivalent circuit is considered as shown in Fig. 4.7. The current flowing through the
filter capacitor is,
24
jnX L
Ic
-jXc
n
Va,h
Ia,h
Ic =
jhX L
jX
C + jhX L
h
I a,h
(26)
Then, the voltage across the filter capacitor at a harmonic frequency becomes,
jhX L
I a,h
2 XL
1 h
XC
(27)
Va , h hX L I a , h
(28)
XL
1
XC
if h 2
(29)
3 X L I a ,3
Va ,1
(30)
The capacitor ripple voltage at the third harmonic frequency is limited to 3% of the fundamental
output voltage. Then, the impedance of the filter inductor can be determined by,
XL
0.03 Va ,1
3 I a ,3
0.03 120
= 0.035
3 (34.14)
25
(31)
Lf =
XL
0.035
=
= 92.84H
2f1 2 60
(32)
From eqns. (25) and (31) the impedance of the capacitor can be obtained by,
XC = 167.46
(33)
Cf =
1
= 16 F
2 f1 X c
(34)
Based on these designed values, inductors with 100H and capacitor with 20F were selected
from a manufacture.
4.2.2
The control for the entire SNUT inverter system is done with the Texas Instrument
TMS320LF2407 DSP [7]. The DSP is a 40MIPS, fixed-point processor. This chip has 16 PWM
signals, 41 general purpose digital I/O pins, 16 high-speed A/D converter inputs, and a serial
communication port. By implementing the control via DSP, the proposed inverter system will
offer increased flexibility and will minimize component cost. The goal of the DSP control is as
follows: 1) the PWM gating signals for IGBTs in the inverter stage are generated according to
the modulation index 2) all of the sensing parameters are sent back to the DSP and are
monitored for control and for fault conditions 3) output voltage regulation is implemented to
meet THD specifications under varying load conditions 4) communication between the DSP in
the inverter system and the fuel cell controller are provided through RS485 and two TTL signals
5) The current reference for the bi-directional DC-DC converter is calculated by comparing the
fuel cell current command to the output real power, and the resultant reference with
charge/discharge mode is sent to the bi-directional DC-DC converter.
4.2.3
Output voltage tolerance should not be wider than 6% over the full line voltage and
temperature range, from no-load to full-load. To meet the output voltage tolerance requirement
26
the AC output voltage is sensed and a closed-loop control is implemented with a digital PI
compensator in the DSP.
The AC output voltage sensing circuit consists of a potential transformer (PT), a gain stage, an
offset stage and a filtering stage. The PT is low in cost and has an isolation. The output of the
gain stage is sent to the offset stage, which is necessary because the A/D converter in DSP is
unipolar. The last stage is a high frequency noise filter which is the unity gain, non-inverting 2th
order Butterworth with cutoff frequency of 5kHz. The output voltage from the sensing circuit is
fed to the DSP and is subtracted from a sine wave reference. This error signal is applied to a PI
compensator and then the resultant signal is compared to a triangular wave of 20KHz. A
sinusoidal PWM signal is generated and sent to the gate drive circuit for IGBTs. All the PI
compensation and the sinusoidal PWM generation are implemented within the DSP. Thus, the
DSP will adjust the modulation index to keep the output voltage regulated under unbalanced
load condition and from no-load to full-load. The circuit diagram for the DSP board and the gate
driver are shown in the appendix A.3 and A.4.
4.2.4
Unbalance in dc-link capacitor voltages causes generation of even harmonics in the inverter
output voltages. A control method of balancing the capacitor voltages is shown in Fig. 4.8.
Va*
Vb*
Va.dc*= 0
+
Va
PWM.A
INV
PWM.B
Phase.A offset
Va.dc
LPF
27
Vb
Suppose output voltage Va has a positive dc offset, which means that the upper capacitor voltage
is greater than the lower capacitor voltage. The output voltage Va is sensed and passed through a
low pass filter to obtain a dc component of voltage Va. This causes addition of a positive value
to the reference output voltage Vb* resulting in a decrease in upper capacitor voltage and an
increase in lower capacitor voltage.
4.2.5
The SNUT DC-to-AC inverter provides the protection capability of over-current, short circuit,
and over temperature in the circuit to prevent damage to the front-end DC-DC converter stage,
fuel cell and inverter itself. Over-current protection is implement by using a current transformer
followed by several op-amp stages. If the output current measured ranges between 100% and
110% of full load current over 1 minute, a signal is sent to the gate drive for shutdown and to
the DSP to light up the over current fault LED. Also, the DC-to-AC inverter can be protected
from output short circuit. If the output current measured exceeds a threshold of 110% of full
load current, a signal is sent to the gate drive for immediate shutdown and to the DSP to light up
the short circuit fault LED. Temperature protection is implemented by using a bimetal as a
temperature sensor that is mounted on the heat sink of the inverter. If the temperature of the
sensor rises over 60C a fan on the heat sink starts to operate. If the temperature of the sensor
rises over 80C a signal is sent to the gate drive for immediate shutdown and to the DSP to light
up the over temperature fault LED. The inverter also shuts down for safe operation if the DC
link voltage goes over 500V or under 300V. The schematic diagram for the sensing circuit is
shown in the Appendix A.2.
28
Fig. 4.10
Then, the bi-directional DC-DC converter is operated to charge or discharge the battery
according to the current reference and the mode of operation determined by the DSP. The two
PWM controllers, UCC3825 and UCC3895, are employed for charge and discharge modes of
operation, respectively. The control block diagram for the bi-directional DC-DC converter is
shown in Fig. 4.10. The PI control is implemented for current control of both converters. The
inductor current is sensed and compared to a reference determined by the DSP, and then the
error is applied to a PI compensator.
The PWM controller generates the gating signal for switching devices of a converter. Only
one of the two converters should be in operation while the other is in idle state. If the current
drawn from the battery exceeds 120A which is the maximum discharging current, the chip will
shut down the bi-directional DC-DC converter by disabling all the gate signals. The battery
voltage is fed to the DSP and is used for the calculation of the charging current reference. When
the battery is in deep discharge (< 42V) or over charge (>56.7V) states, the DSP will shut down
the bi-directional DC-DC converter to protect the battery from being damaged.
Fig. 4.11 shows the inductor voltage and current waveforms for charge and discharge modes,
respectively. Let us define the turns ratio n2 of the high frequency transformer T2 to be,
n2 =
Ns
Np
(35)
30
Vdc
1
) DTS = Vbatt ( D)TS
2
n2
(36)
which gives
Vbatt 2 D
=
Vdc
n2
(37)
Vdc
1
) ( Dd )TS
n2
2
(38)
which gives
Vdc
n2
=
Vbatt (1 2 Dd )
(39)
Vdc
n2
=
Vbatt 2(1 D)
(40)
31
Vdc
<0
n2
(41)
For the above equation to be always satisfied, the turns ratio is,
n2 < 6.597
(42)
The turns ratio should be selected as large as possible due to the fact that larger turns ratio gives
smaller inductance L, smaller voltage rating of MOSFETs and smaller current rating of IGBTs.
Therefore, the turns ratio is determined to be n2 = 6.
Inductor design
When the battery is being charged, ripple component in the charging current should be
restricted by inductor L on the battery side. The magnitude of the charging current depends on
the capacity of the battery. The maximum charging current is assumed to be 45A since the
capacity of the battery in the FCT fuel cell system is known to be 155Ah. We allow the ripple
current to be 20% of the maximum charging current, that is, iL = 9A
During the charge mode, it can be seen from eqn (37) that the duty ratio D lies between
0.33 < D < 0.5
(43)
When the switches S3 and S6 are turned on during the charge mode we have (See Fig. 4.14(a)),
(Vdc / n2 ) Vbatt
i
= L
L
DTS
(44)
Then, combining eqns. (37) and (44) the inductance is obtained by,
L =
(1 2 D) D
Vdc
n2 iL f s
(45)
32
directional DC-DC converter should be able to supply full power of 5KW at a maximum load of
10KW. The maximum dc current at the output can be calculated as,
I dc =
Pdc
= 12.5 A
Vdc
(46)
Ignoring power loss in the bi-directional DC-DC converter, average power on the battery side is
the same as average power on the dc link side. Then, the average inductor current become,
IL =
n2 I dc
2(1 D)
(47)
(48)
Then, combining eqn. (48) and D=0.5+Dd the ripple current can be determined by,
i L =
( D 0.5)
Vbatt
L fs
(49)
From eqns. (47) and (49), the rms current rating of the inductor becomes
I L rms = 113 A
(50)
Therefore, we choose an inductor with the inductance of 39uH and the rms inductor current
rating of 113A.
Switch ratings
Since the maximum battery discharge current is much larger than the maximum battery
charge current, the switch ratings of the bi-directional DC-DC converter should also be
determined based on the discharge mode of operation at full load (5KW, 1min.).
The voltage and current ratings at the worst case are listed in Table 4.2. Actual devices have
been selected from a manufacture.
33
Table 4.2 Voltage and current ratings of the switch based on maximum discharge operation
Devices
Current rating
MOSFETs
Ipeak = 119.6A
S1,S2
Iav = 60A
IGBTs
Ipeak = 19.93A
S3~S6
Iav = 9.5A
Voltage rating
Vpeak = 140V
Vpeak = 420V
200V. 100A
MOSFETs
600V. 20A
IGBTs
Transformer Design
In order to reduce the leakage inductances of the high frequency transformer two 2.5KW
transformers are connected in parallel to form a 5KW transformer. The reduced transformer
leakage inductance s results in reducing the rating of the snubber mounted on the primary side.
The 2.5KW transformer design is described in the following.
a. Transformer core material
Ferrite core is chosen as a material of a high frequency transformer due to its low cost
and low losses characteristics for transformers and inductors in the frequency range 20 KHz
to 3MHz.
b. Transformer core size
The SNUT team would follow the procedure for transformer core size selection provided by
Magnetic Inc. [8]. The core type of choice is EER core, then
C = 5.07 10 3 cm 2 / Amp
(51)
The transformer efficiency is assumed to be 90%. The winding factor is K = 0.3(primary side
only). The flux density B is assumed be 2000(gauss) and fs = 20KHz.
Then, from eqn(9) the WaAc product is given by,
Wa Ac =
2500 (5.07 10 3 ) 10 8
= 29.34cm 4
4 0.9 2000 20000 0.3
(52)
Using the core selection table by area product distribution, a core of 47054-EC(MAGNETICS
Inc.) was selected. Magnetic data for the selected core is shown in Table 4.1.
c. number of turns
The number of primary turns is given by [8],
34
Np =
VP 108
140 108
=
= 25.81turns
4 BAf s
4 2000 3.39 20000
(53)
From the turns ratio which has been obtained by n2=6, the final number of turns for primary and
secondary windings are determined to be,
Np : Ns = 26: 156
d. wire size
At a discharge mode of operation at the full load (5KW, 1min.), the rms vales of the primary
and secondary winding currents are 39.2A and 8.4A, respectively. Then, at 500 circular mils per
rms ampere the required number of circular mils is obtained by [9],
Primary circular mil requirement = 500 I rms = 19,600
Secondary circular mil requirement = 500 I rms = 4,200
Hence, the wire sizes AWG 7 and AWG 13 are selected from AWG table for the primary and
secondary wires, respectively.
35
Low Battery Fault (lower than 48v) : Low Battery Fault LED
(when the battery voltage is under than 48V)
making the size of the heat sink smaller. In this section the design of the heat sink for the
SNUT inverter system is detailed. The operating parameters such as total power dissipation
are defined and thermal circuits of the switching devices mounted on the heat sink are
established and analyzed. The first step is to calculate the power dissipation of a switching
device according to the equations below.
MOSFETs
Switching loss Psw =
1
I ds ( rms ) Vdc ( peak ) f s [t on + t off ]
2
ton
ts
(54)
(55)
(56)
IGBTs
Switching loss Psw =
1
Vdc ( peak ) I CE ( rms ) f s [ton + toff ]
2
ton
ts
(57)
(58)
(59)
Then, a thermal equivalent circuit for analyzing thermal characteristic of the heat sink is defined
as shown in Fig. 4.13 when two kinds of power devices are mounted on a heat sink.
Given power loss Pl (where, l = 1 or 2) of a switching device, junction to case thermal
resistance Rjc,l case to heat sink thermal resistance Rch,l ambient temperature Ta, and juntion
temperature Tj,l heat sink to ambient thermal resistance Rha can be obtained in the following
procedure.
The case temperature Tc,l can be given as,
Tc ,l = T j ,l Pl R jc ,l
(60)
38
Th ,l = Tc ,l Pl Rch ,l
(61)
Th = Th ,1 + Th , 2
(62)
Rha =
Th Ta
P1 + P2
(63)
39
calculated or the heat sink can directly be selected from a manufacture by the heat sink to
ambient thermal resistance obtained. Fig.4.14 to Fig.4.15 shows the heat sinks of each section
selected based on the heat sink to ambient thermal resistance Rha. A safety margin was
considered for actual heat sink selection from the manufacture.
Table 4.3 Power dissipation in the device used
Section
Device
Front-end
DC-DC
MOSFET
16.7
32.2
48.9
Diode
0.012
13.2
13.21
Inverter
IGBT
70.56
50.4
120.96
Bi-directional
DC-DC
MOSFET
48.3
59.4
107.7
IGBT
34.9
39.9
Device
Q`ty
Rjc(C/W)
Tj(C)
Tc(C)
Front-end
DC-DC
MOSFET
12
0.21
70.26
60
Diode
1.25
65.85
49.34
Inverter
IGBT
0.22
112.5
85.9
Bi-directional
DC-DC
MOSFET
0.18
103.3
80.92
IGBT
0.31
67.54
55.17
Th(C)
Rha(C/W)
45.38
0.11
49.7
0.08
40
5. Simulation
In this section the whole inverter system designed in the previous section is simulated using
PSIM to validate the proposed design concept. The simulation was done with realistic
parameters of the selected device if possible. Fig. 5.1 shows the simulated waveforms for a
worst case of load transient from 5KW to 10KW and back to 5KW again. Fig. 5.1(a) shows the
inverter current on the dc side demonstrating an increase and a decrease in average value
according to the load change. Fig. 5.1(b) shows the output current of the front-end DC-DC
converter. The average value of the current did not change during the transients even if the
ripple was increased due to the operation of bi-directional DC-DC converter. The dc link
voltage was well regulated at 400V, as shown in Fig. 5.1(c). Fig. 5.1(d) and (e) shows the
current waveforms of the dc link side and the battery side of the bi-directional DC-DC converter,
respectively. This illustrates that for a sudden load change the bi-directional DC-DC converter
draws an amount of power from the battery, which is difference in power between the fuel cell
and the load. The output phase voltage was well regulated during the load transient while the
output current shows an increase and a decrease according to the load change.
41
(a)
(b)
(c)
(d)
(e)
(f)
(g)
Fig. 5.1 Simulated waveforms (5KW 10KW 5KW) (a) inverter current on the dc side,
(b) output current of the front-end DC-DC converter, (c) dc link voltage, (d) current
waveforms on the dc link side of the bi-directional DC-DC converter, (e) current
waveforms of the battery side of the bi-directional DC-DC converter, (f) output phase
voltage, (g) output phase current
42
6. Experimental Result
A 10KW prototype inverter has been built in the laboratory of SNUT, and experimental
waveforms are presented in this section. A programmable DC power source capable of
supplying 5KW was used instead of the fuel cell. Four 12V, 80AH batteries are connected in
series to form a 48V battery.
4.4KW load level are shown in Fig. 6.1. Fig. 6.1(a) shows the output phase voltages VAN and
VBN, respectively. Fig. 6.1(b) shows the output voltage VAB and current with a linear load.
The experimental waveforms were obtained for a transient discharge mode of operation, that
is, a load increase from 2kW to 2.7kW. The upper trace in Fig. 6.2(a) shows the output current
of the front end DC-DC converter whose average value did not change after the transient even if
the ripple was slightly increased due to operation of bi-directional DC-DC converter. The lower
trace in Fig. 6.2(a) shows the dc link voltage, which undergoes an overshoot and is stabilized.
Fig. 6.2(b) shows the inverter current on the dc side indicating an increase in average value
according to the load increase. Fig.6.2(c) and (d) show the PWM current waveform on the dc
link side of the bi-directional DC-DC converter and its extended waveform in time scale,
respectively. Fig. 6.2(e) and (f) show the current waveform on the battery side of the bidirectional DC-DC converter and its extended waveform in time scale, respectively. This
demonstrates that for a sudden load increase the bi-directional DC-DC converter would quickly
draw an amount of power from the battery, which is difference in power between the fuel cell
and the load. The upper trace in Fig. 6.2(g) shows the output phase voltage which is well
regulated during the load increase. The lower trace in Fig. 6.2(g) shows a magnitude increase of
the output phase current indicating load increase. Fig. 6.2(h) shows the extended waveforms in
time scale for Fig. 6.2(g).
Photograph of the SNUT fuel cell inverter system is shown in Fig. 6.3..
43
44
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
Fig. 6.2 Experimental waveforms ( 2KW 2.7KW ), (a) upper : output current of the front
end converter ; lower: dc link voltage, (b) dc side inverter current, (c) output current of the
bi-directional converter, (d) extended waveform of (c), (e) battery current, (f) extended
waveform of (e), (g) output phase voltage and current, (h) extended waveforms of (g)
45
7. Performance evaluation
The 10kW prototype inverter system was tested from no load to 4.4kW load. Experimental
performances of some important design items have been obtained and compared to minimum
target requirement of the inverter system as shown in Table 7.1. The SNUT prototype inverter
met the minimum target requirements for most of the design items such as frequency regulation,
THD of the output voltage, output voltage regulation and input current ripple. The SNUT
prototype demonstrated a good performance, especially in THD of output voltage and output
voltage regulation. The efficiency of the front end DC-DC converter section was 90% and that
of the inverter section was 97% resulting in total system efficiency of 88%.
The SNUT team is trying to increase the efficiency by optimizing design and selection of the
devices.
46
2003 FEC
Specification performance
SNUT team
Experimental performance
Frequency
60Hz 0.1Hz
59.95Hz ~ 60.09Hz
5%
Regulation
6%
-2.4% ~ +0.2%
3%
Total 88%
(DC-DC:90%, INV:97%)
8. Bill of Materials
A detailed bill of materials for the front end DC-DC converter, the DC-AC inverter, and the
bi-directional DC-DC converter sections is listed in Table 8.1. We could not find some electrical
parts such as power switching devices and transformer cores in Korea. Therefore, we sometimes
had to use parts which has much larger ratings than the designed value.
9. Cost analysis
The SNUT team has been placing great emphasis on cost through the whole design process.
There are many factors to purchasing electrical parts. The cost analysis is based on the
spreadsheets evaluation forms provided in the 2003 FEC workshop. The result of the cost
analysis for the front end DC-DC converter, the DC-AC inverter and the bi-directional DC-DC
converter are shown in Table 9.1 to 9.3, respectively. The cost of the front end DC-DC converter
was $233.23. The cost of the DC-AC inverter was $150.06. The cost of the bi-directional DCDC converter was $121.1. The total cost of the 10kW SNUT fuel cell inverter system was
$504.39. The values in the table indicate only preliminary, relative cost estimates, not dollars. A
detailed bill of material will be developed and provided in the final report for evaluation of
actual cost of product.
47
Qty.
IXYS (IXFN180N10)
IXYS (IXFN180N20)
Fuji (2MBI200N-060)
Toshiba (MG50J2YS50)
IXYS (DESI 2X31-10B)
12
2
2
2
5
Chang-Sung (CH270125E)
Chang-Sung (CH572060E)
Magnetics (K5528B026)
2
2
1
Agilent (HCPL-316J)
2
2
14
TI (TMS320LF2407)
Unitrode (UCC3895)
Unitrode (UC3825)
1
2
1
Burr-Brown (ISO122)
Burr-Brown (INA126)
Burr-Brown (INA117)
National (LF353)
Fulltech (UF15P-12H)
2
2
1
20
3
1
3
Power switch
MOSFETs (100V, 180A)
MOSFETs (200V, 180A )
IGBTs (600V, 200A )
IGBTs (600V, 50A)
Diodes (1000V, 2X30A )
Inductor
MPP core (100uH, 15A)
MPP core (93uH, 60A)
MPP core (40uH, 120A)
Transformer
Ferrite EE core (2.5kW)
Capacitor
Electrolytic (400V, 1000uF)
Electrolytic (400V, 3300uF)
Opto-isolated gate driver IC
Control
DSP controller
PWM controller
Amplifier
Isolation amplifier
Instrumentation amplifier
Differential amplifier
Op-amp
Fans (120Vac, 33W)
Sensor
Hall current (400A)
Hall current (200A)
Table 9.1 Cost spread sheet for front end DC-DC converter
DEVICE
QTY DESIG
UNIT MEASURE
VOLT
(Vpk)
VOLT
(Vrms)
CUR
(Avg)
CUR
(Arms)
UNIT EXTENDED
COST
COST
DIODE
8 D1~D8
410
6.25
2.29
18.34
MOSFET
4 S1~S4
41
125
6.56
26.24
CAP (ALUM)
2 C1,2
19.78
39.55
TRANSFORMER
1 T1
389.8
26.57
26.57
CHOKE
2 L1, 2
12.6
41.40
82.79
34.73
34.73
LOSSES
3222 uF
210
47.3
100 UH
416.8 W
CONTROL
5.00
TOTAL
233.23
48
QTY DESIG
UNIT MEASURE
IGBT
2 S1~S4
CAP (ALUM)
2 C3,4
16 uF
CHOKE
2 L3,4
93 UH
LOSSES
VOLT
(Vpk)
VOLT
(Vrms)
420
CUR
(Avg)
CUR
(Arms)
28
170
60
134.4 W
UNIT EXTENDED
COST
COST
4.77
9.53
0.17
0.33
59.50
119.00
11.20
11.20
CONTROL
10.00
TOTAL
150.06
QTY DESIG
UNIT MEASURE
VOLT
(Vpk)
VOLT
(Vrms)
CUR
(Avg)
CUR
(Arms)
UNIT EXTENDED
COST
COST
IGBT
2 S11~S14
420
9.5
1.69
3.38
MOSFET
2 S9, S10
140
60
8.32
16.65
CHOKE
1 L5
113
69.49
69.49
TRANSFORMER
1 T2
73.3
10.95
10.95
12.63
12.63
LOSSES
40 UH
58
151.5 W
CONTROL
8.00
TOTAL
121.10
10. Conclusion
The objective of this project is to develop a low cost, high efficiency 10kW inverter system
for a SOFC system. In this report a power circuit topology for the inverter system has been
chosen after evaluating two possible topologies in a practical way proposed. All the component
ratings were designed along with thorough analysis on the chosen topology. A hardware
prototype capable of supplying 10kW load was built and tested at a laboratory of Seoul National
University of Technology. The SNUT prototype inverter met the minimum target requirements
and demonstrated a good performance in most of the design items. The SNUT team has been
trying to increase the efficiency and to decrease the cost by optimizing design and selection of
the devices. The SNUT team strongly believes the final prototype meet the efficiency and cost
requirements.
49
11. Reference
[1] R. Anahara, S. Yokokawa and M. Sakurai, Present Status and Future Prospects for Fuel Cell Power
Systems, Proceedings of the IEEE, vol. 81, no. 3, March 1993, pp.399-407
[2] A. Emadi and S. Williamson, Status Review of Power Electronic Converters for Fuel Cell
Applications, Journal of Power Electronics, vol. 1, no. 2, Oct. 2001, pp.133-144
[3] N. Azli, A. Yatim, DSP-based Online Optimal PWM Multilevel Control for Fuel Cell Power
Conditioning Systems, IEEE IECON conf. rec, 2001, pp.921-926
[4] Final Reports from the 2001 Future Energy Challenge, Available:
http://www.energychallenge.org
50
Appendices
Vin
200k
Vout(+)
0.1u
2
GND
<NO.2>DC-DC(+15V)
Vout(-)
<NO.2>DC-DC(-15V)
GND
SMPS(-15V)
<NO.2>DC-DC converter(15v)
SMPS(+15V)
0.1u
0.1u
10k
1M
10n
Vout DC 400V
2
0.1u
V+
Vin+
10n
Vin-
1k
Vo
V-
GND
10
+Vs2
GND
-Vs2
Vout
8
7
V_Sensing
7
6
15
10k
16
Vin
GND
0.1u
-Vs1
<NO.2>DC-DC(-15V)
+Vs1
<NO.2>DC-DC(+15V)
ISO122
INA 126
10k
0.1u
0.1u
0.1u
0.1u
200k
+15
U11
VR1
+5
10k
1
D2
U2A
LF353
1
10k
C18
47uF
Ian_rms
+
DIODE
11
1
+5
0.1uF
10K
2
3
4
5
6
7
8
9
0.1uF
1
3
G3
LE
OE
4081
U6C
14
G2
8
10
G1
-15
10k
0.1uF
4
Ibn
10k
U1B
LF353
7
C18
47uF
Ibn_rms
+
+15
+5
8
0.1uF
10k
OV_Ian
D3
4081
U7B
4081
D5
20k
-15
U7A
1
2
R3
20k
14
10k
OV_I SET 3
U8A
LF353
1
14
1N4148
Ian_rms
+5
D5
4
+15
+15
ON : OV_I 1N4148
10k
0.1uF
R2
2
3
470
6
5
U3B
LF353
7
4148
+15
10k
C18
4.7uF
+5
Van_rms
10k
D5
4
D2
U3A
LF353
1
6
OV_I SET 5
+15
1k
10k
OV_Ibn
Ibn_rms
0.1uF
R3
TRAN_HM31
1N4148
U8B
LF353
7
D3
20k
ON : OV_I 1N4148
R2
300
10k
3
5
470
6
5
4148
U4B
LF353
7
R1
10k
10k
Vbn_rms
TRAN_HM31
4081
10
D3
0.1uF
SW1
SD_ON
ON : SD
OFF :
R3
1k
U7C
8
9
C18
4.7uF
D2
U4A
LF353
1
14
4
10k
Vbn_6V
360
T1
Vbn
+5
0.1uF
+15
51
D5
20k
-15
5
6
R3
8
10k
Van_6V
360
T1
300
Van
11
13
10k
100
0.1uF
D2
DIODE
0.1uF
10K
U2A
LF353
1
101
4
-
-15
EL25P1
U1A
LF353
1
10K
ON :
R2
300
SD
U6D
12
0.1uF
-15
201
SG2
4081
14
10k
0.1uF
VR1
SG3
+15
50k
+15
U11
SG4
4081
U6B
74HC573
+15
U6A
14
G4
19
18
17
16
15
14
13
12
Q0
Q1
Q2
Q3
Q4
Q5
Q6
Q7
D0
D1
D2
D3
D4
D5
D6
D7
U1B
LF353
7
U8
PWM4
PWM3
PWM2
PWM1
14
10k
8
101
100
0.1uF
10k
0.1uF
4
Ian
4
3
-15
EL25P1
U1A
LF353
1
10K
-15
0.1uF
-15
201
0.1uF
4081
SG1
52
+5
D2
100
100
SG1
SG3
1
3.3K
2
3
0.1uF
4
5
FAULT 3
6
7
330pF
Vin+
VE
Vin-
V LED2+
Vcc1
GND1
DESAT
Vcc2
RESET
FAULT
Vc
Vout
V LED1+
V EE1
V LED1-
V EE2
16
15
DIODE
+15
0.1uF
1
3.3K
100pF
14
0.1uF
13
Q6
12
11
BDW93C
10
10 1W
G3
FAULT 1
10
Q8
BDW94C
330pF
D1
1N4746
HCPL-316J
0.1uF
GND_1 GND_1GND_1
1
2
3
4
5
6
7
330pF
GND1
V LED2+
DESAT
Vcc2
RESET
Vc
FAULT
Vout
V LED1+
V LED1-
V EE1
V EE2
14
+15
0.1uF
100pF
13
Q6
12
11
BDW93C
10
10 1W
G1
10
Q8
BDW94C
HCPL-316J
D1
1N4746
47K
D1
1N4746
S3
0.1uF
0.1uF
S1
-15
+5
D2
100
Vin+
VE
Vin-
V LED2+
Vcc1
GND1
DESAT
Vcc2
RESET
Vc
FAULT
Vout
V LED1+
V EE1
V LED1-
V EE2
16
15
DIODE
+15
0.1uF
3.3K
2
3
0.1uF
Q6
12
11
100pF
14
13
BDW93C
10
5
10 1W
G4
FAULT 2
6
7
10
Q8
BDW94C
330pF
D1
1N4746
HCPL-316J
Vin+
VE
0.1uF
DC LINK
Vin-
V LED2+
Vcc1
GND1
DESAT
Vcc2
RESET
Vc
FAULT
Vout
V LED1+
V EE1
V LED1-
V EE2
DIODE
16
15
14
+15
0.1uF
100pF
13
Q6
12
11
BDW93C
10
10 1W
Q8
BDW94C
HCPL-316J
D1
1N4746
47K
D1
1N4746
S4
0.1uF
-15
53
G2
10
47K
D1
1N4746
0.1uF
D2
DC LINK
SG2
FAULT 4
Vcc1
15
GND_1
100
0.1uF
DC LINK
DIODE
16
-15
+5
3.3K
Vin-
VE
SG4
Vin+
47K
D1
1N4746
0.1uF
D2
DC LINK
0.1uF
-15
S2
54
MAGNETICS Inc.]
55