The development of a Roebel cable based

1 MVA HTS transformer

Neil Glasson

11 October 2011

Mike Staines1, Mohinder Pannu2, N. J. Long1, Rod Badcock1, Nathan Allpress1, Logan Ward1

1 Industrial Research Limited, New Zealand2 Wilson Transformer Pty Ltd, Melbourne

Outline

• Introduction

• Design overview

• Roebel cable

• Short circuit fault

• High voltage insulation

• Heat transfer

• Cryostat design

• Summary

Project Partners

www.hts110.co.nz

www.wtc.com.au

www.eteltransformers.co.nz

www.fabrum.co.nz

www.pbworld.com

www.vectorelectricity.co.nz

www.gcsuperconductors.com

www.weltec.ac.nz

www.aut.ac.nz

Design parametersParameter Value

Primary Voltage 11,000 V

Secondary Voltage 415 V

Maximum Operating Temperature 70 K, liquid nitrogen cooling

Target Rating 1 MVA

Primary Connection Delta

Secondary Connection Wye

LV Winding20 turns 15/5 Roebel cable per phase

(20 turn single layer solenoid winding)

LV Rated current 1390 A rms

HV Winding918 turns of 4 mm YBCO wire per phase

(24 double pancakes of 38.25 turns each)

HV Rated current 30 A rms

HV winding

• Uses 4 mm Superpower tape

• I/Ic ~ 25%

• Polyimide wrap insulation

• 24 double pancakes

• No encapsulation due to

concerns of

– Poorer heat transfer

– Voids lowering withstand voltage

– Potential Ic degradation

LV winding

• Single layer solenoid on fibre glass composite former

• Liquid nitrogen in contact with cable surface

• YBCO Roebel Cable

– L = 20 m

– 15/5 (15 strands, 5 mm width)

– Self field Ic ~ 1400 A @ 77 K

• No strand or cable insulation

• Flux deflectors will be added

First transformer cable configured for end-to-end Ic test

Core

• Warm core with

cruciform, 6-stepped

section of high grade

core steel

• No-load loss of about

750 W

– At 20:1 cooling penalty,

cold bore is not feasible

YBCO Roebel Cable

Roebel cable or Continuously Transposed Cable (CTC) is useful for

Forming a high current capacity conductor

100’s to 1000’s of Amps (even 10,000s at low T)

Reducing AC losses

rule of thumb – magnetisation losses scale with strand width

Strands wound together and geometric parameters

Roebel strand

Cables are labelled with the convention

# of strands / strand width

We are making two designs 15/5 and 10/2

Roebel Cable ManufacturePunch tool and

frame

Tape de-spool

Tape re-spool

Control

systems

Set-up for automated

multi-strand Roebel

strand production.

(a)

(b)

(c)

(d)

Formation of Roebel punched strands in 40 mm and 12

mm wide feedstock material.

(a) 4 x 5 mm strands in 40 mm wide material,

(b) 1 x 5 mm wide strand in 12 mm wide material,

(c) 10 x 2 mm strands in 40 mm wide material,

(d) 3 x 2 mm wide strands in 12 mm wide material.

Punching

WindingAutomated planetary wind system for

15/5 cable

Capable of winding several hundred

continuous metres of cable.

Wire qualification

0 10 20 30 40 500.90

0.92

0.94

0.96

0.98

1.00

Co

rre

latio

n

Position (m)

Wire 2

22 )()(

))((),(

yyxx

yyxxYXCorrel

For Roebel we require 2D uniformity

- Scan wire magnetically (penetrated

or remnant field)

- Quantify uniformity using statistical

correlation with an ideal magnetic

profile

Correlation along a

length of YBCO wire, a

minimum Correl can be

specified for input wire

Where |Correl | ≤ 1 X is a dataset representing calculated field

Y{y1…yj} is magnetic data across tape

0.000 T

200mm

(a)(a)

0.023 T

(b)

We use continuous scanning of the Remnant

magnetic field to assess tape quality (a) tape

with a known defect, and (b) tape with only

small scale variability.

0 2 4 6 8 10 120.010

0.015

0.020

0.025

0.030

0.035

0.040 cor_0.99

cor_0.90

cor_0.75

Fie

ld (

T)

Position (mm)

Example Profiles

Some wire is extremely good !

Roebel cable performance

• Measured end-to-end Ic of

cable is 1400 A DC @ 77 K.

• Design calls for 1390 A rms

= 1970 A peak.

• Ic tests on strand samples at

70 K and subsequent

analysis predicts 2500 A

cable Ic. Ipeak/Ic = 0.79

• This is yet to be confirmed

by experiment on a cable at

70 K.

Use of load line method to predict cable critical current

Described by Staines et al, The development of a Roebel cable

based 1 MVA HTS transformer, to be published in Supercond. Sci. Technol. 24 (2011)

Short circuit fault handling

• 2 second short circuit withstand is a common requirement of conventional transformer standards.

• What is a realistic target for HTS?– HTS lacks the resistive conductor cross sectional area to carry a

short circuit for 2 seconds. What is a realistic short circuit duration for an HTS transformer?

• Determine fault current limiting performance.– HTS under short circuit exhibits highly non-linear response – how will

this limit fault current?

• Cable strand has exposed cut edges.– Roebel strand is punched – exposing edges of conductor. Does this

pose a risk of conductor damage due to ingress of LN2 and then subsequent heating in a short circuit?

Short circuit simulation

• Adiabatic model – assumes no heat transfer from the

conductor into liquid nitrogen.

• Assuming instantaneous shift of current into 40 micron

thickness of copper (20 micron each side,

Ω= 12.5 x 10-3 / m per strand @ 77 K).

• Simulation incorporates temperature dependence of

both resistance and thermal capacity of the conductor.

Simulation output

• LV winding will reach

350 K at t = 200 ms

• Impedance doubles

from 0.05 pu to 0.1 pu

initially then up to 0.5 pu

over 200ms (due to change

in resistance as

temperature increases).

-0.05 0 0.05 0.1 0.15 0.2 0.25 0.350

100

150

200

250

300

350

400

450

Tp, T

s vs. Time

Time (s)T

p,

Ts (

K)

TP

Ts

Tp=Temperature of primary windingTs= Temperature of secondary winding

Simulation output

• Fault limited peak

current = 14 kA

(about 35% of the peak

that would occur if

limited only by the 0.05

pu leakage reactance).

pu = per unit, actual value divided by base value.

Cable will offer significant

current limiting during a

short circuit fault.

V2F

V2E

V2D

V2C

V2B

V2A

+ve Terminal

-ve Terminal

Current Shunt

V1

Short circuit strand testing

• 200 ms DC pulse, peaking

at >15 V/m in strand

(transformer short circuit

voltage = 12 V/m).

• T increased to 300 K

in 60 ms

• There was no damage

(no change in Ic).

• Repeat test after 3 week

soak in LN2.

Conclusion – 200 ms duration short circuit is OK and

there is no evidence of damage due to nitrogen ingress

into the exposed cut edge of the cable strands.

Impulse response in HV winding

From VOLNAProprietary modelling software – modelling response to standard 95 kV impulse

21 kV peak voltage

High voltage insulation

* Guide for the statistical analysis of

electrical insulation breakdown data

• 95 kV impulse response modelling

– HV winding must withstand 550 V from turn-to-turn.

(21 kV / 38 turns per double pancake)

– Commercial 25 micron polyimide wrapped insulation was

tested to confirm suitability.

• Two wrapped strands held

together and stressed

until breakdown @ 77 K.

• 20 tests analysed according

to IEC 62539*

• Insulation will survive 2 kV.

Heat transfer experiment

• Fix power dissipation in strand and measured

temperature rise of cable winding in LN2

• Used conductor with Tc < 65 K. Copper layer used as

both resistive heater and temperature sensor.

• Tested at 68 K and 77 K, both at atmospheric pressure.

Typical heat transfer regimes

Van SciverHelium Cryogenics

Note: hysteresis in transitions from one regime to the other and the shape of transition from convection to nucleate.

Heat transfer resultsHeat Transfer - Dependence on Bath Temperature

0.001

0.01

0.1

1

10

0.01 0.1 1 10

Conductor Temperature Rise, dT [K]

Po

we

r D

issip

ate

d [

W/m

]

77K Outer

77K Middle

77K Inner

Subcooled Outer

Subcooled Middle

Subcooled Inner

Assume 1 W/m of strand heating due to AC loss.

Sub cooling necessary to avoid nucleate boiling.

Convective cooling keeps T < 3 K.

Location of strand has minimal impact.

3

Convectiveheat transfer

Nucleate boiling

Sub cooled results are at 68 K and atmospheric pressure.Inner, middle and outer refer to position in Roebel cable strand stack –inner being against the winding former, outer being directly exposed to liquid nitrogen.

Desired conditions

inside the transformer cryostat

• The windings must be held in the temperature range

65K to 70K in order to achieve the required

performance of the superconducting cable.

• It is desirable to ensure that the evolution of gas in the

windings is minimized. Gas bubbles degrade the

dielectric performance of the LN2 as well as risking an

accumulation of gas that might thermally insulate

regions of the windings and give rise to hot spots.

S.M.Baek et al, Electrical Breakdown Properties of Liquid Nitrogen for Electrical Insulation Design of Pancake Coil Type HTS Transformer. IEEE Trans. Appl. Supercond. Vol 13, No.2, June 2003

Nitrogen phase diagram

Temperaturerange

Max. pressure

Vapour and liquid only co-exist on the saturation line

Operation at elevated pressure allows significant temperature increase before phase change.

Atmospheric pressure

Operation heremakes pressurevessel design easier

Cryostat layout• Common cryostat for

all three phases is best

– Smaller footprint and reduced heat load due to

fewer electrical bushings and reduced shell area.

– Simplified nitrogen circulation system – only one

inlet and one outlet required.

• Three individual cryostats - easiest to

manufacture

– Cryostat pressure design is easier

– Less risk with a simple cylindrical structure

• Compromise with three individual

eccentric cryostats

– Simple cylindrical shells

– Reduced transformer length and nitrogen volume

X

1.3 X

1.5 X

Cryostat lid arrangement

Gas in equilibrium due to temperature profile through foam

Liquid return

Electrical bushing

Bulk liquid at 65K – 70K

Foam insulation plug

Cryostat work in progress

Cryostat cold shell test vessel

Single phase cryostat base cold shell and mould

www.fabrum.co.nz

Summary

• Roebel cable described – high current capacity/low AC loss. Tests and analysis predict cable Ic of 2500 A at 70 K.

• Short circuit fault across the Roebel cable winding can be tolerated if disconnected within 200 ms.

• Standard copper stabilizer provides significant fault current limiting - Simulation predicts short circuit fault limited to 35% of the prospective fault current

• Standard 25 micron spiral wrapped polyimide insulation will withstand 2 kV turn-to-turn.

• 1 W/m continuous dissipation in sub-cooled LN2 will be below nucleate boiling regime.

• Cable temperature will be up to 3 K warmer than bulk LN2.

Summary cont...

• Cryostat design discussed. The target normal operating

temperature in the cryostat is in the range 65 K to 70 K.

• There is benefit in operating cryostats at elevated pressure.

• A foam plug beneath the lid will allow operation at elevated

pressure while maintaining a gas/liquid transition at some

point between the lid and the bulk liquid region.