Power Electronics for Fuel Cell Based Power Generation Systems

Allen HefnerNIST

The Semiconductor Electronics Division

(NIST/DOE Interagency Agreement)

Outline

I. IntroductionII. Technology Impact AnalysisIII. Component TechnologiesIV. Power Converter ArchitecturesV. Cost Estimates and Simulation

I. IntroductionObjective:• High-Megawatt Power Conditioning Systems (PCS) are

required to convert: – from power produced by Fuel Cells (FC) in future power plants – to very high voltage and power required for delivery to the grid

Motivation:• DoE SECA cost goals:

– FC generator plant $400/kW – including $40-100/kW for PCS

• Today’s PCS cost (Fuel Cell Energy Inc.):– FC generator plant $3,000/kW – including $260/kW for power converter (to 18 kV AC)

Fuel Cell Plant PCS

Power Conditioning

System(PCS)

Fuel Cell Stack60 Hz Step-upTransformer Power Grid

$40-$100 / kW for PCS is a difficult stretch goal !

$40-$100 / kW

• NIST/DOE Inter-Agency Agreement– NIST lead effort to determine expected impact of advanced

technologies on future FC power plant PCS

• Inter-Agency Power Group (IAPG)– Form an interagency task group for high-megawatt power

converter technologies under IAPG

• Industry Roadmap– Initiate a roadmap process to offer guidance for further

development of high-megawatt PCS technology

• National Science Foundation (NSF)– Establish power electronics curriculums and fundamental

research programs for energy systems technology

Federal and IndustryPCS Program Coordination

Outline

I. IntroductionII. Technology Impact AnalysisIII. Component TechnologiesIV. Power Converter ArchitecturesV. Cost Estimates and Simulation

II. Technology Impact Analysis• Perform Independent Analysis of technologies that

may reduce cost of PCS for future FC Power Plants

• Methodology for impact study:– Include input from broad power electronics, power

component, and power engineering communities– Classify power converter architectures, topologies, and

component technologies that may reduce cost – Perform tabular calculations of cost for each option using

estimated advantages of new technologies– Use component modeling, and circuit and system

simulations to verify and refine calculations

Methodology for Impact Analysis:

NIST Coordination

Architectures•ABB•Fuel Cell Energy•General Electric•Infineon•Siemens

Topologies•VA Tech FEEC•Satcon•Oak Ridge NL•Univ. Illinois

Semiconductors•CREE•NIST•Powerex

Power Converters•VA Tech FEEC•Los Alamos NL•Texas A&M•Oak Ridge NL

Cost Estimate Simulation

Passive Components•General Atomics•Los Alamos NL

High Megawatt Converter Workshop: http://www.high-megawatt.nist.gov/workshop-1-24-07/

Parameters of Impact Analysis• Boundary conditions and performance parameters:

– FC Stack: center tap ~700 V DC, 0.6 MW– Require individual FC stack current control with low ripple

• Power electronics and/or transformer to 18 kV AC, and transformer from 18 kV AC to 300 kV AC transmission

• Converter cost components:– Semiconductors– Module Packaging and Interconnects– Cooling System– Magnetics: Filter Inductors and

HF voltage isolation transformers– 60 Hz Transformer up to 18 kV– Breakers and Switchgear

SemiconductorsPackaging and Interconnects

HF transformersFilter Inductors and Capacitors

Cooling System60 Hz Transformer up to 18 kV

Breakers and Switchgear

Ripple < 2%Stack Voltage Range~700 to 1000 VFuture: 2X voltage ?

$40-$100 / kW

300 MW PCS

IEEE – 519IEEE – 1547

Harmonic DistortionFuture: HVDC transmission ?

18 kVAC

300 kVAC

Approx. 500Fuel Cells

~700 VDC

~700 VDC

Outline

I. IntroductionII. Technology Impact StudyIII. Component TechnologiesIV. Power Converter ArchitecturesV. Cost Estimates and Simulation

A. Low-Voltage Semiconductors

• Baseline: 1200 V silicon IGBT with silicon PiN diode– 1200 V is sweet spot for silicon IGBTs at 15 – 20 kHz switching

due to speed voltage trade-off, high volume, and maturity– Mature technology; cost may not continue to decrease

• 1200 V silicon IGBT switch with SiC Schottky diode– More efficient at 20 kHz less loss, lower heat removal cost

lower temperature and longer life– Higher frequency less filter inductor cost– Emerging commercial technology; cost continues to decrease

– What is cost break-point for 1200 V SiC Schottky diode?

B. High-Frequency Transformer• Baseline: Traditional Ferrite Material

– Expensive processing for high-power, high frequency (HF)– Mature technology; cost may NOT decrease

• HF magnetic materials: Nano-crystalline or Metglas®

– Size and cost decrease inversely proportional to frequency– Emerging commercial technology; cost continues to decrease

– What is cost break-point for HF magnetic material?

10 100 103 104 105

10-3

10-2

10-1

1

10

Frequency (Hz)

Cor

e Si

ze

10-4

10-5

10-6

60 Hz 20 kHz10 100 103 104 105

10-3

10-2

10-1

1

10

Frequency (Hz)

Cor

e Si

ze

10-4

10-5

10-6

60 Hz 20 kHz

C. High-Voltage Semiconductors • Baseline: High-Voltage (HV) Silicon devices (IGBT, IGCT)

– Typically ~6.5 kV blocking voltage maximum– Requires multi-level inverter for 4160 V AC – Low switching frequency (< 500 Hz) requires larger filter

• High-Voltage, High-Frequency SiC Switch and Diodes12 kV, 20 kHz SiC MOSFET switch and SiC Schottky diode:

– Less inverter levels due to higher voltage – Less loss, lower heat removal cost– Less filter inductance required due to higher frequency

15 kV, 5 kHz SiC IGBT switch and SiC PiN diode:– Higher current per die than SiC MOSFET, therefore lower cost

– What is cost break point for HV-HF SiC power semiconductors?

DARPA HPE MOSFET High Speed at High Voltage

-5

0

5

10

15

20

Dra

in C

urre

nt (A

)

-1500

0

1500

3000

4500

6000

Dra

in-S

ourc

e Vo

ltage

(V)

Area = 0.125 cm2

T = 25o C

Vd

Id

SiC MOSFET: 10 kV, 30 ns Silicon IGBT: 4.5 kV, 2us

1us /div

3000 V

15 ns /div

0 V

Area= 0.15 cm2

A. Hefner, et.al. “Recent Advances in High-Voltage, High-Frequency Silicon-Carbide Power Devices,” IEEE IAS Annual Meeting, October 2006, pp. 330-337.

Outline

I. IntroductionII. Technology Impact StudyIII. Component TechnologiesIV. Power Converter ArchitecturesV. Cost Estimates and Simulation

IV. Power Converter ArchitecturesA. Low-Voltage Inverters (460 V AC):

– Require high current for each 0.6 MW FC– and large number of Inverters for 300 MW Plant

B. Medium-Voltage Inverters (4160 V AC):– Lower inverter current for each 0.6 MW FC– Combine multiple FCs with single high power inverter

C. High-Voltage Inverters (18 kV AC):– Replaces 60 Hz transformer with isolation from HF transformer – Cascade enables: 18 kV AC inverter by series connection, and

interleaved switching decreases losses and filter requirement

A. Low-Voltage Inverters480 V AC Inverter,

60 Hz Transformer to 18 kV AC

1) First Generation: ~350 V FC, DC-DC to 750 V, 480 V AC Inverter

2) Baseline: ~700 V FC (center tap), DC-DC to 750 V,480 V AC inverter

– 1200 V is “Sweet spot” for silicon semiconductors due to market size and speed voltage trade-off

3) Present Generation: ~700 V FC, 480 V AC Inverter– Fewer semiconductors due to single stage converter

Lo

480 V AC

~350 V DC

DCDC~

ACDC

750 V DC

18 kV AC

Lo

480 V AC

~700 V DC

DCDC~

ACDC

750 V DC

18 kV AC

Lo2

480 V AC

~700 V DC

ACDC~

18 kV AC

1) First Generation 2) Baseline 3) Present Generation

0.3 MW 0.6 MW 0.6 MW

B. Medium Voltage Inverters4160 VAC Inverter

60 Hz Transformer to 18 kV AC

MV DC Common Bus:

LV-to-MV DC-DC converters:– Requires high voltage-gain DC-DC converter, HF transformer– HV SiC diode rectifiers substantially reduce loss !

Common MV inverter:– Silicon semiconductors require multiple levels for MV– HV SiC semiconductors provide lower switching loss

Lo

4160 V AC

~700 V DC

DCDC~

ACDC

MVDC Common Bus

18 kV AC

1) MV DC Common Bus: LV-to-MV DC-DC, Common Inverter

~700 V DC

DCDC~

DCDC~

… ~700 V DC

6750 V DC

High Voltage Gain DCDC Converter

0.6 MW

#1 #2 #8

MV Inverter4.6 MW

C. High Voltage Inverters18 kV AC Inverter,

No 60 Hz Transformer !

High-Voltage Cascade:

Series connected, voltage-isolated LV-to-MV DC-DC– Requires HV-HF isolation and output for each phase

Medium-Frequency, phase-interleaved invertersY-Configuration:

– 6.5 kV Silicon semiconductors require 6 levels– 12 kV SiC semiconductors require 3 levels

Lo

18 kV AC

DCDC~

ACDC

Series Connected, Phase Interleaved

HV-Cascade: Isolated MV-DCDCs, Interleaved MV Inverters

DCDC~

DCDC~

~700 V DC

0.6 MW/Phase

#3

5 kV DC, 3 outputHV&HF Isolated

HV, HF Isolated DC-DC

0.6 MW

#1 #2

ACDC

ACDC

5 kV pp ACMF, MVInverter0.2 MW

Phase 1Phase 3Phase 2

Neutral

Outline

I. IntroductionII. Technology Impact AnalysisIII. Component TechnologiesIV. Power Converter ArchitecturesV. Cost Estimates and Simulation

V. Cost Estimates and SimulationA. Cost Estimate: Tabular calculation of cost for each

PCS option using estimated advantages of newtechnologies

B. Simulation: Component modeling, and circuit andsystem simulations to verify and refine calculations

• Physics based models for component technologies• Simulation schematics for power converter topologies

Cost Estimate Simulation

A. Tabular Calculations of Cost• Future, high-volume costs: 5 to 10 years, 1 GW/yr

• Advanced Technology Goals and Cost Break Points– 1.2 kV Schottky diodes: $0.2/A– 12 kV Schottky diodes: $1/A– 12 kV Half-bridge SiC-MOSFET/SiC-Schottky: $10/A– 15 kV SiC-PiN: $0.4/A– 15 kV SiC-IGBT/SiC-PiN Module: $3.3/A– Nano-crystalline transformer: $2/kW– Power Electronics DC-DC, DC-AC: 150 % overhead– 60Hz Transformer and Switchgear: 50 % overhead

$0$20$40$60$80

$100$120$140$160$180$200Transformer &

SwitchgearOther PE

Semiconductor

Cooling

Magnetics

Inverter Voltage Low Low Low Low LowConverter Stages One One Two Two TwoLV-SiC Schottky yes yes

yes

yes yesHF Transformer Ferrite Nano60 Hz Transformer yes yes yes yes

Preliminary $/kW: LV Inverter

Risk Level: HighConsiderableModerateLow

$0$20$40$60$80

$100$120$140$160$180$200Transformer &

SwitchgearOther PE

Semiconductor

Cooling

Magnetics

Inverter Voltage Medium Medium High High HighHV-SiC Diode Schottky Schottky Schottky PiNHV-SiC Switch MOSFET MOSFET IGBTHF Transformer Nano Nano Nano Nano Nano60 Hz Transformer yes yes

Preliminary $/kW: MV & HV Inverter

Risk Level: HighConsiderableModerateLow

loss

loss

B. Component Modeling, Circuit and System Simulations• Physics based models developed for advanced

component technologies– 12 kV SiC Junction Barrier Schottky Diode (JBS)– 12 kV SiC Power MOSFET– 1200 V SiC Junction Barrier Schottky Diode (JBS)

• Models validated for static and switching characteristics

• Simulation schematics for power converter topologies

Model Validation for 12 kV, 100 ASiC Junction Barrier Schottky Diode

-40.0

-20.0

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

2.0E-07 4.0E-07 6.0E-07 8.0E-07 1.0E-06 1.2E-06 1.4E-06 1.6E-06Time [s]

Ano

de C

urre

nt [A

]

-8000

-7000

-6000

-5000

-4000

-3000

-2000

-1000

0

1000

Ano

de-C

atho

de V

olta

ge [V

]

Measured 15 A/us Simulated 15 A/us Measured 32 A/us

Simulated 32 A/us Measured 67 A/us Simulated 67 A/us

T = 25 ºC

Voltages

Currents

Forward Characteristics Reverse Recovery Switching

Model Validation for 12 kV, 100 A SiC Power MOSFET

Forward Output Characteristics Inductive Load Switching

Neutral

…Lo

ACDC

HF Transformer versus 60 Hz Transformer

ACDC

ACDC

HV Cascade Inverter Phase 1

3 Output DC-DC #1

Optional DC-DC

Lo

DCDC~

ACDC

18 k VAC60 Hz Transformer

…

LowVoltage Inverter

18 k VAC

HF Transformer

Rectifier RectifierRectifier

Inverter

Low Voltage Inverter Schematic

cba

480 Vl-l277 Vl-n

Vdc

ean

ebn

ecn

700 Vto

1 kV

ic

ia

ib

18 kVl-l10.4 kVl-n

FuelCell0.6 MW n

LineSwitchgear

TransformerSwitchgear

Plant DistributionLine

Lower Voltage Inverter Simulations

Output currents

Inverter output line-line voltage

Output phase-neutral voltage

Inverter output phase-neutral voltage

Time

75ms 80ms 85ms 90ms 95ms 100msI(La2) I(Lb2) I(Lc2)

-1.2KA-0.6KA

0A0.6KA1.2KA

V(Ea:+)-V(Ea:-) V(Eb:+)-V(Eb:-) V(Ec:+)-V(Ec:-)-400V

0V200V400V

SEL>>

V(VaL)-V(VbL)-1.0KV-0.5KV

0V0.5KV1.0KV

V(VaL)-V(vn)-500V-250V

0V250V500V

Time (ms)2015105 25

ia icib

van

ean ebn ecn

1.2 kA

1 kV

–1.2 kA

0

–500 V

500 V

–1 kV

–400 V

400 V

0

0

0

0

vab

20 kHz

750 V

High-Voltage Y-Connected Cascade Inverter Schematic

+–

van vbn vcn

ean

ic

ebn

+–

+–

ia

ib

+–

+–

+–

+–

+–

+–

Plant DistributionLine

ecn

LineSwitchgear

High-Voltage Y-Connected Cascade Inverter Simulation

total inverter output phase voltage

3 levels

1 level

Time

75ms 80ms 85ms 90ms 95ms 100msI(Rao) I(Rbo) -I(Rao)-I(Rbo)

-400A-200A

0A200A400A

V(Vo+)-V(Vao4) V(Vbo+)-V(Vbo4) V(Vao4)+V(Vbo4)-V(Vo+)-V(Vbo+)-1.5KV

-0.5KV

0.5KV

1.5KVV(Vao1)-V(Vao4)

-1.5KV

0.5KV

1.5KV

SEL>>

V(Vao1)-V(Vao2)-500V-250V

0V250V500V

Time (ms)2015105 25

ia icib

ean ebn ecn

400 A

0

15 kV

–400 A

0

–5 kV

5 kV

–15 kV

–15 kV

15 kV

0

0

(1 kHz switching)0

(eq. 3 kHz switching)

inverter 1 output voltage

utility phase voltage

output currents

van

va1

Conclusion• The SECA goal of $40-$100 / kW for the Fuel Cell

Plant PCS is a difficult stretch goal.• Federal and Industry efforts have been initiated to

address High-Megawatt PCS requirements. • Analysis of technologies that may reduce cost of

PCS for fuel cell power plants:– high-frequency magnetic materials– high-voltage high-frequency SiC semiconductors– 1200 V SiC Schottky diodes

• High voltage inverter may permit galvanic isolation via low cost HF transformer instead of costly 60 Hz transformer

AcknowledgmentsI would like to acknowledge Sam Biondo (DOE-Fossil Energy), Don Collins (DOE - NETL), Wayne Surdoval (DOE-NETL), Frank Holcomb (ERDC-CERL), Ron Wolk(Consultant), Heather Quedenfeld (DOE - NETL), and Jason Lai (VT-FEEC) for organizing activities to address PCS needs for fuel cell power plants, and in addition the participants ofthe Workshop on High Megawatt Converters for providing technical guidance.

I would also like to acknowledge Jason Lai (VT-FEEC), HaoQian (VT-FEEC), Angel Rivera-López (NIST), Tam Duong (NIST), José M. Ortiz-Rodríguez (NIST), Colleen Hood (NIST), and Robert Michelet (NIST) for technical contributions to the presented PCS cost analysis effort.