high-efficiency low-cost dc-dc converter for sofc · a high-efficiency low-cost dc-dc converter for...
TRANSCRIPT
JSL1
A High-Efficiency Low-Cost DC-DC Converter for SOFC
Performance and Control of V6 Converter
Presented byDr. Jih-Sheng (Jason) Lai
Virginia Polytechnic Institute and State UniversityFuture Energy Electronics Center
May 11 – May 13, 2004
SECA Core Technology Program Review Meeting
Boston, Massachusetts
JSL2
Acknowledgement
• Technical support and encouragement of Don Collins of DOE NETL are greatly appreciated.
• Also quoted by Don Collins that a 1% increase in efficiency is worth $75/kWe given a $6.50/mbtu gas cost for a SOFC power plant of the size about 150kW – A motivation to high-efficiency power converter design
JSL3
Outline
1. Fuel Cell Power Plant and DC/DC Converter Topology Options
2. Features of V6 Converter 3. Calorimeter Setup and Test 4. Testing with PEM Fuel Cells5. Conclusion
JSL4
1. Block Diagram of the SOFC Power Plant
• Fuel cell output or converter input is low-voltage DC with a wide-range variation
• Plant output is high-voltage ac• Multiple-stage power conversions including
isolation are needed
DC-ACLF-HF
AC-ACLV-HV
AC-DCHV-HV
DC-ACHV-HV
DC-AC Inverter
+Filter
SOFCAC-DC
Rectifier+
FilterVin
+
–
+
–
BridgeConverter
HFXformer
SECA DC/DC converter22 to 50 V
400 V 120/240V
JSL5
DC/DC Converter Topology Options
• Single-phase – Half-bridge converter• Two-phase – Full-bridge converter• Three-phase – Three-phase bridge
converter• Six-phase – The proposed V6 converter
JSL6
Single-Phase Half-Bridge Converter
1 : nHF-AC
Half-bridge converter
Xformer
Rectifier+LC filter
Act
ive
Load
20V>250A
Sol
id-O
xide
Fue
l Cel
l
400V5 kW
JSL7
Two-Phase Full-Bridge Converter
1 : n
HF ACXformer
Rectifier+LC filter
Act
ive
Load
Sol
id-O
xide
Fue
l Cel
l
Full-bridge converter
20V>250A
400V5 kW
JSL8
Full-Bridge Converter with Paralleled Devices to Achieve the Desired Efficiency
Load6x 6x
• With 6 devices in parallel, the two-leg converter can barely achieve 97% efficiency
• Problems are additional losses in parasitic components, voltage clamp, interconnects, filter inductor, transformer, diodes, etc.
Voltage clampSol
id-O
xide
Fue
l Cel
l
20V250A
JSL9
A Three-Phase Bridge Converter
vin
L
Cf
+
–
Co vo
+
–
1 : nS1 S5
S4 S2
a
c
HF ACXformer Rectifier+LC filter
D1
D4
D3
D6
iA ia
iL
Act
ive
Load
Fuel
Cel
l or
othe
r vo
ltage
sour
ce
3-phase bridge inverter
S3
S6
b
D5
D2
A
B
Cn
–iC
• Hard switching • With 4 devices in parallel per switch• Efficiency ≈ 95%
JSL10
Circuit Diagram of the Proposed V6 Converter
Sol
id-O
xide
Fue
l Cel
l
Act
ive
Load
Rectifier+LC filterSix-phase bridge converter
HF ACXformer
JSL11
Major Issues Associated with the DC/DC Converter
• Cost• Efficiency• Reliability• Ripple current• Transient response along with auxiliary
energy storage requirement• Communication with fuel cell controller• Electromagnetic interference (EMI) emission
JSL12
2. Key Features of the V6 Converter• Double output voltage reduce turns ratio and
associated leakage inductance• No overshoot and ringing on primary side device
voltage• DC link inductor current ripple elimination cost and
size reduction on inductor• Secondary voltage overshoot reduction cost and size
reduction with elimination of voltage clamping• Significant EMI reduction cost reduction on EMI filter• Soft switching over a wide load range• High efficiency ~97% • Low device temperature High reliability
JSL13
Photograph of the V6 Converter
Transformers
Rectifiers
Output Filter
To Load
DC Input
Input Bulk Cap
Phase-shift Controller
Transformers
Power MOSFET
HF Cap
JSL14
Voltage Conversion Ratio
0 60 120 180
α(º)
n⋅Vdc
2n⋅Vdc
<Vo>
DC/DCMode
dcoo VnV ⋅⋅>=<60α
°<<° 600 .1 α
°<<° 12060 .2 α
dco VnV ⋅⋅>=< 2°<<° 180120 .3 α
0
2n⋅Vdc
6swT
oswT
360⋅α
Vd
6swT
oswT
360⋅α
Vd2n⋅Vdc
2n⋅Vdc
6swT
Vd
0
3Phase xfmr voltage doubling effect
dcoo VnV ⋅⋅>=<60α
JSL15
Comparison between Full-Bridge and V6 Converters
Full Bridge Converter V6 Converter
iL
iLvd
vd
• Secondary inductor current is ripple-less; and in principle, no dc link inductor is needed
• Secondary voltage swing is eliminated with <40% voltage overshoot as compared to 250%
JSL16
Significant DC link Inductor Size Reduction
With V6 converter, an effective 10x reduction in DC link filter inductor in terms of cost, size and weight
Single Phase500W60Hz
Single Phase5 kW50kHz
Three Phase5 kW50kHz
Single Phase500W60Hz
Single Phase5 kW50kHz
Three Phase5 kW50kHz
Single Phase500W60Hz
Single Phase5 kW50kHz
Three Phase5 kW50kHz
JSL17
Input and Output Voltages and Currents at 1kW Output Condition
Vin
VoVoIin
Vin
Iin
IoIo
(a) Full bridge converter (b) V6 Converter
Significant improvement with V6 converterLess EMIBetter efficiency (97% versus 87% after calibration)
(97%)(87%)
JSL18
Efficiency Measurement Results
80%82%84%86%88%90%92%94%96%98%
100%
0 1000 2000 3000 4000Output power (W)
Experimental data and trend line
• Measurement error: within 1%• Heat sink temperature rise:
<20°C at 2kW with natural convection
JSL19
Where are the Losses?
• Switch conduction • Diode conduction • Transformer • Output rectifier • Output filter inductor and capacitor• Input capacitor• Parasitics
– Copper traces– Interconnects
JSL20
3. Calorimetry for Accurate Loss Measurement
• A 50-liter calorimeter• Calibration of the 50-liter calorimeter • A 160-liter calorimeter• Calibration of the 160-liter calorimeter
JSL21
Basic Calorimeter Principle
( )
2
where
: heat flow (W)
: thermal conductivity of the barrier 0.029 W/m K for styrofoam: surface area (m ): temperature ( C): thickness of barrier
hot coldk A T TQt d
Qtkk
ATd
⋅ ⋅ −=
⋅
o
OutsideAmbient Air
InternalAmbient Air ≅
JSL22
Calorimeter Setup Diagram
Heatsink
Power
Resist
ors
Fan
Fan
Fan EndMiddle
Lid End
Bottom
To DataLogger
JSL23
A 50-liter Calorimeter
JSL24
Calibration Results of the 50-Liter Calorimeter
y = 0.7659x + 3.7749
0
10
20
30
40
50
60
70
0 10 20 30 40 50 60 70 80 90Power Loss (W)
Tem
pera
ture
(deg
C)
Temperature RiseLinear (Temperature Rise)
JSL25
A 160-Liter Calorimeter Setup
Calibration with resistor bankfans
JSL26
Test the 160-Liter Calorimeter at 120-W Condition
0102030405060708090
0 100 200 300 400 500 600Time (min)
Tem
pera
ture
(deg
C)
Average probe temperature
Temperature rise
JSL27
Temperature Rise Versus Power Loss
y = 0.4086x + 2.9771
1520253035404550556065
30 40 50 60 70 80 90 100 110 120 130Power Loss (W)
Tem
pera
ture
(deg
C)
TempRise
Linear (TempRise)
JSL28
4. Test DC/DC Converter with a PEM Fuel Cell
• Fuel cell polarization curve • Dynamic fuel cell voltage and current output• Dynamic fuel cell model in electrical circuit• Model verification • Comparison with slow time constant power
supply test
JSL29
Polarization Curves with Different Compressor Speeds
I_Ifc
0A 10A 20A 30A 40A 50A 60AV(fc)
20V
25V
30V
35V
40V
45V
2
3
1
1. Fuel cell runs at parasitic load condition Compressor is running at low speed
2. Fuel cell is fully loaded Compressor is not immediately
responding to load step voltage dips
3. Compressor speeds up, fuel cell voltage picks up
JSL30
Fuel Cell Dynamic Characteristics
0102030405060
0 2 4 6 8 10 12t (sec)
0
500
1000
1500
2000
0 2 4 6 8 10 12t (sec)
vFC(V)
iFC(A)
pFC(W)Step load: 1.47kWParasitic power: 70W
voltage undershoot (2.5V) due to compressor delay
150W dip
27.2V
300W power overshoot
43V
1st time constant
2nd time constant
JSL31
Fuel Cell Dynamic Circuit Model
1st time constant
+−
+ −
X
+
+−
2nd time constant
+
• High load current high voltage drop• Low output voltage low voltage drop
Multiplying ratio
parasitic load
Transient load
VocFuel cell
High current branch
Low current branch
JSL32
Verification of Fuel Cell Model with Resistive Load Transient
Fuel Cell Output Voltage (10V/div)
10µs/div
Fuel Cell Output Current (5A/div)
Fuel Cell Output Power (200W/div)
0s 50us 100us0
10
20
30
40
501
0W
0.2KW
0.4KW
0.6KW
0.8KW
1.0KW2
>>
Vfc
PfcIfc
(b)Simulation
results
(a) Experimental
results
Vfc
PfcIfc
<1µs time response on Vfc
JSL33
Test Load Transient with a Slow Power Supply
Voltage maintains constant after transient
Voltage dip
Load current
inductor current
Voltage returns to original level
JSL34
Fuel Cell Voltage Dynamic with V6 Converter Load Transient
Input Voltage from Fuel Cell (5V/div)
100ms/div
SimulatedExperimental
Significantly slower time constant (≈50ms) due to 30 mF V6-converter input capacitor and a long cable
JSL35
V6-Converter Output under Load Step
Load Current (2A/div)
Input Current (20A/div)Output Voltage (50V/div)
20ms/div
With voltage control loop bandwidth designed at 20 Hz, settling time is about 40ms under load step
JSL36
V6-Converter Output under Load Dump
Load Current (2A/Div)
Output Voltage (50V/Div)
10ms/Div
Input Current (20A/Div)
With voltage control loop bandwidth design at 20 Hz, settling time is about 40ms under load dump
JSL37
Findings of Fuel Cell Modeling and Converter Test Results
• Fuel cell stack shows very fast dynamic, nearly instantly without time constant
• Perception of slow fuel cell time constant is related to ancillary system not fuel cell stack
• Output voltage dynamic is dominated by the converter interface capacitor and cable inductor
• Output current dynamic is dominated by the load
JSL38
Summary
• Static and dynamic performances of a new V6 converter has been presented
• Two calorimeters were built and calibrated• PEM fuel cell has been tested under
dynamic conditions for controller design purpose
• A circuit model has been verified for fuel cell dynamic study
JSL39
Future Work
• Test V6 converter with Calorimeter• Incorporate DC/AC inverter for current
ripple evaluation• Work with PNNL for SOFC dynamic
modeling in electrical circuits and test V6 converter with SOFC
• Develop energy balancing and control strategies with SOFC
• Test EMI performance at EPRI-PEAC