Dri

vin

g ec

on

om

ic g

row

th, i

nn

ova

tio

n,

and

wo

rkfo

rce

dev

elo

pm

ent

4th International Workshop on

Grid Simulator Testing of Wind Turbine DrivetrainsNREL ESIF April 4th, 2017

J. Curtiss Fox

» Johnson Controls BESS

» ComRent Inductive Load Bank

» Transformer Saturation

» High Speed Motor and VFD

» Other Projects

» 2.3 Wind Turbine

» 3.2 Wind Turbine

1 MW, 500 kWh

Static Losses

» The efficiency of a BESS is highly dependent upon the power level

» Efficiency curves offer better insight into operational efficiency across the power spectrum

» Can be used to model system performance and enable feed-forward control schemes

» Requires highly calibrated instrumentation

Charge Discharge

Charge Discharge

Zoom in on low power

levels

» Compounding value streams require coordination of the system efficiency and the desired state of charge

» While frequency regulation may have high power demands, much of the time is spent passing through zero power

» Estimates for system state of charge can be modeled given system efficiencies» Models do not need to be complex if they are data driven» Very useful in setting up energy testing profiles» Predict performance in energy applications to refine initial conditions

» 13.8 kV Inductive Load Bank» Shipped to San Francisco for Trans Bay HVDC testing

L1,L2, L3

MPBB Configuration

Power Converter Module

Power Converter Module Schematic

AC

DC

DCAC

AC

DC

DCAC

AC

DC

DCAC

1av

vSvRv TSv

1bv

1cv

av

bv

cv

MPBB Configuration Schematic

Output power 1 MW

Motor speed 15,000 rpm

Motor voltage 4.16 kV

Drive input voltage 13.8 kV

Drive topology Series H-bridge

Switching device 1.7 kV SiC MOSFETSiC in existing drive design: reduce losses, higher fsw

» 1MW, 4P, 15,000 rpm

» Core length = 14”

» Stator OD = 22”

» Rotor OD = 10”

» Air gap (magnetic) = 0.093”

» 2 TEWAC coolers (water to air)˃ 2 turbo pressure blowers with

each blower circulating the air for ½ of the motor

TEWAC coolers

Turbo pressure blowers

• Integrating 1700V 400A SiCMOSFET Modules

• Based on TWMC’s modular VersaBridge design

• Three modules (“Slices”)

• 7-level with interleaving (36-pulse)

• 2-phase cooling for high power density

• 96.2 % minimum efficiency at full load (Si devices @ 60 Hz / 600 Hz switching)

At 800VDC: Protection Works At 850VDC: Protection Fails

At 850VDC: Protection Works At 900VDC: Protection Fails

MFG 1 SiC MOSFET

MFG 2SiC MOSFET

» Investigating the failure of a Yg-yg, 5 limb core transformer in the field

» Unique failure mechanism associated with loose leads on the dip pole allowing for a down-stream phase-to-phase connection

» This type of transformer is typical fused at 1.5 to 2.5 PU and is widely used in a power range of 100 kVA to 1500 kVA for three phase distribution

» The failure causes zero sequence voltage to be imposed on the transformer and saturate the zero sequence flux path

A

B

C

N

a

b

c

n

Transformer

Primary Fuses

200 % Rating

The Lead for Phase A

falls into the lead for

Phase B

24 kV Feeder

Circuit

Solidly Grounded

5 Limb Core Type

Transformer

Yyn0 Vector Group

Primary and

Secondary

Coils

Phase A

Primary and

Secondary

Coils

Phase B

Primary and

Secondary

Coils

Phase C

Primary and

Secondary

Coils

Phase A

Primary and

Secondary

Coils

Phase B

Primary and

Secondary

Coils

Phase C

Φ0A Φ1 Φ2 Φ3 Φ0B Φ1 Φ2 Φ3 Φ0

Lim

b 0

A

Lim

b 0

B

Lim

b 1

Lim

b 2

Lim

b 3

Lim

b 0

Lim

b 1

Lim

b 2

Lim

b 3

Actual Construction of a 5 limb core Modeled Construction of a 5 limb core

Limb 3 Magnetizing Branch

Limb 2 Magnetizing Branch

Limb 1 Magnetizing Branch

VA

N-M

AG

Leakage

Inductance

Winding

Resistance

VAN Saturation

Model

Imag(Flux)

1

S

Flux =

Integral of

Voltage

VAN-MAG

Imag Zero Seq

Leakage

Inductance

Winding

Resistance

VBN

VB

N-M

AG

Saturation

Model

Imag(Flux)

1

S

Flux =

Integral of

Voltage

VBN-MAG

Imag Zero Seq

Leakage

Inductance

Winding

Resistance

VCN

VC

N-M

AG

Saturation

Model

Imag(Flux)

1

S

Flux =

Integral of

Voltage

VCN-MAG

Imag Zero Seq

Limb 0 Magnetizing Branch

Imag Zero Seq

Saturation

Model

Imag(Flux)

1

S

Flux =

Integral of

Voltage

VAN-MAG

VAN-MAG

VAN-MAG

1

3

Scale for

Limb 0

Size

» Simulations demonstrate large inrush currents as misconnection is made from phase A to B

» These decay off quickly to relatively low steady state saturation current

» This saturation current can potentially be within the limits of the fusing

» Catastrophic failure is possible from this scenario

» Full scale testing provides data similar to simulation

» The steady state saturation current is demonstrated to be near the fuse ratings

» Future work will be to refine and validate the simulation model based upon captured data

» Calculate the power into the transformer and estimate the time to failure

2 PU

1 PU

Phase A Phase B Phase C Total

Active Power (P) 64 kW 63.5 kW 32.3 kW 159.7 kW

Reactive Power (Q) 437 kVAR 437 kVAR 6.9 kVAR 881.4 kVAR

Apparent Power (S) 441.8 kVA 441.8 kVA 33 kVA 895.798 kVA

VsNo Load Core Losses

Total

Power 0.963 kW

𝑃 = σ𝑛=1∞ 𝑉𝑛𝐼𝑛 cos(𝜃𝑉𝑛 −𝜃𝐼𝑛) 𝑄 = σ𝑛=1

∞ 𝑉𝑛𝐼𝑛 sin(𝜃𝑉𝑛 −𝜃𝐼𝑛)

Thermal modelling estimates 23 min for oil to reach 104 ͦC.

» Power Line Noise Study ˃ SRNL - LDRD

» Improving Distribution Transformer Efficiency and Lifetime˃ SRNL - GMLC

» Improving the Rol of Short-term Energy Storage and Large Motor Loads for Active Power Controls for Wind Power˃ NREL - GMLC

» Distributed Energy Resource Optimization˃ EPRI - NYSERDA

» Delivered March 2017

» Installation June 2017

» Voltage ride-through testing can become very complex

» Transformers, cabling and test equipment required for system interconnection introduce voltage drops

» Dynamic reactive current requirements of the smart inverter can create dynamic voltage disturbances

0.01 0.1 1 10 1000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

Time (s)V

olt

age

(pu

)

Nominal – indefinite operation

HV1 – within 0.16 s drop to < 10% current

HV2

LV1 – maintain > 80% current

LV2 – maintain > 80% current

LV3 – within 0.16 s drop to

<10% current

FIDVR Event with the WECC Fault Ride-

Through Boundaries CA Rule 21 VRT Boundaries

» Frequency ride-through testing is much easier than voltage ride-through˃ Everyone agrees how fast the system is rotating

» Rate of change in frequency (Hz/sec) should be maximized within control limits

Trip Level Test Time to Trip Test

4 6 8 10 12 1459.9

60

60.1

60.2

60.3

60.4

60.5

60.6

60.7

Fre

qu

ency

(H

z)

Time (s)

0

500

1000

1500

Cu

rren

t (A

)

Ref Freq

PCC Ph A (A)

PCC Ph B (A)

PCC Ph C (A)

1.5 1.55 1.6 1.65 1.7 1.75 1.859

59.5

60

60.5

61

61.5

62

Fre

qu

ency

(H

z)

Time (s)

1.5 1.55 1.6 1.65 1.7 1.75 1.8-2000

0

2000

Cu

rren

t (A

)

Ref Freq

PCC Ph A (A)

PCC Ph B (A)

PCC Ph C (A)

Trip Delay

Time = 133 ms