Design of laminate composite layups using genetic algorithms

This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 90.244.24.105

This content was downloaded on 19/04/2015 at 10:05

Please note that terms and conditions apply.

Investigations of current limiting properties of the MgB2 wires subjected to pulse overcurrents

in the benchtop tester

View the table of contents for this issue, or go to the journal homepage for more

2007 Supercond. Sci. Technol. 20 320

(http://iopscience.iop.org/0953-2048/20/4/004)

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING SUPERCONDUCTOR SCIENCE AND TECHNOLOGY

Supercond. Sci. Technol. 20 (2007) 320–326 doi:10.1088/0953-2048/20/4/004

Investigations of current limitingproperties of the MgB2 wires subjected topulse overcurrents in the benchtop tester*Lin Ye1,2, M Majoros3, A M Campbell1, T Coombs1, S Harrison4,P Sargent5, M Haslett5 and M Husband6

1 Interdisciplinary Research Center (IRC) in Superconductivity, Department of Engineering,University of Cambridge, Madingley Road, Cambridge CB3 0HE, UK2 Department of Electrical Engineering, CAU, PO Box 210, Beijing 100083,People’s Republic of China3 Laboratories for Applied Superconductivity and Magnetism, Ohio State University,Columbus, OH 43210, USA4 Scientific Magnetics, Culham Science Centre, Culham, Abingdon, Oxfordshire OX14 3DB,UK5 Diboride Conductors Ltd, Cambridge CB1 3QJ, UK6 Strategic Research Center (SRC)–Electrical Engineering, Rolls-Royce Plc., Derby DE248BJ, UK

E-mail: [email protected]

Received 25 December 2006, in final form 4 February 2007Published 26 February 2007Online at stacks.iop.org/SUST/20/320

AbstractA laboratory scale desktop test system including a cryogenic system, an ACpulse generation system and a real time data acquisition program inLabView/DAQmx, has been developed to evaluate the quench properties ofMgB2 wires as an element in a superconducting fault current limiter underpulse overcurrents at 25 K in self-field conditions. The MgB2 samples startedfrom a superconducting state and demonstrated good current limitingproperties characterized by a fast transition to the normal state during the firsthalf of the cycle and a continuously limiting effect in the subsequent cycleswithout burnouts. The experimental and numerical simulation results on thequench behaviour indicate the feasibility of using MgB2 for futuresuperconducting fault current limiter (SFCL) applications.

(Some figures in this article are in colour only in the electronic version)

1. Introduction

The need for fault current limiters (FCLs) is associatedwith the continuous growth and interconnection of modernpower systems and the increase in decentralized embeddedgeneration, which result in a progressive increase in shortcircuit levels far beyond their original design capacity.Several fault current limiting devices are being pursued byresearch institutions, utilities and manufacturers around theworld to protect electric networks from high fault currents.With the development of superconducting materials, the

* This work is supported by Rolls-Royce Plc and the UK Department of Trade& Industry (DTI).

superconducting fault current limiter (SFCL) is consideredto be one of the most promising solutions to controllingfault-current levels on utility distribution and transmissionnetworks [1, 2].

The high temperature SFCL can be fabricated using hightemperature superconductor (HTS) materials such as BSCCObulks, YBCO thin films or 2-G coated conductors, etc. Hightemperature superconductors can be cooled by liquid nitrogen,which is cheap at 77 K. However, their manufacturing processrequires a great deal of silver, a costly material. As candidatesfor the SFCLs, HTS materials still remain very expensive.

Since the discovery of magnesium diboride (MgB2) in2001 [3], remarkable progress has already been achieved interms of manufacturing MgB2 into useful long lengths at a

0953-2048/07/040320+07$30.00 © 2007 IOP Publishing Ltd Printed in the UK 320

Investigations of current limiting properties of the MgB2 wires subjected to pulse overcurrents in the benchtop tester

(a) (b)

Baffles

Current lead

Displacer

Sample mount

Figure 1. The desktop cryostat system: (a) cryostat, (b) test insert.

low cost [4]. Furthermore, MgB2, which has a transitiontemperature of 39 K, has very attractive properties whenchilled to between 20 and 30 K, within the range of astandard commercial cryocooler, which should encourage thedevelopment of applications for the material [5]. While MgB2

has been considered for MRI, it is becoming a potentialcandidate material as well for use in power applications ofsuperconductivity such as fault current limiters, transformersand motors [6].

It is noted that the maximum short circuit current appearswithin the first half-cycle if the fault occurs at the zero-voltagecrossing point. So the onset of SFCLs must be available tolimit the maximum fault currents within milliseconds. Fromthe point view of power systems, the superconducting elementof the SFCL is required to generate a certain value of resistanceinserted into the circuit to limit the peak value of the faultcurrents within the first half-cycle. Hence, it is very importantto understand the quench and current limiting behaviours of theMgB2 sample when the overcurrent flows.

In the work described in this paper, a desktop test systemwas set up and the MgB2 samples supplied with severallimited cycles of AC pulse currents were experimentallyinvestigated to assess the feasibility of using MgB2 forpractical distribution-scale SFCL devices.

2. Desktop test cryostat system

2.1. Desktop test cryostat

The desktop test cryostat has been designed and manufacturedby the UK Scientific Magnetics (former Space CryomagneticsLtd) for testing samples at temperatures as low as 25 K [7].Figure 1 shows the desktop test cryostat system.

The desktop test cryostat has five major subcomponents:vacuum insulated bucket Dewar, test insert, cryocooler,temperature controller and monitor, vacuum pump set andisolated box.

2.2. Vacuum vessel and test insert

The Dewar consists of an aluminium vacuum vessel(figure 1(a)) surrounding a cylindrical copper bucket. Thebucket is connected to the top flange by a fibreglass neck tube.

(a)

(b) (c)

Figure 2. The cryocooler (a), cold head (b) and vacuum pumpset (c).

Figure 3. Cooling-down profile of the cryogenic system.

The test insert with the main cryogenic interfaces mounted in itis shown in figure 1(b). The test insert is used for mounting thesuperconducting samples, and is then suspended from the topplate inside the bucket Dewar with a bolted O-ring seal. Thetemperature monitor and controller powers the heaters on thebucket Dewar to maintain the preset operating temperature.

2.3. Cryocooler and vacuum pump set

The cryocooler cold head and vacuum pump set are illustratedin figures 2(a)–(c).

A single stage Gifford–McMahon cryocooler manufac-tured in the US by Cryomech is mounted with the cold headconnected to the bottom of the bucket. This allows the bucketto be chilled as it is made from high conductivity copper. Thecold head is inside the bucket Dewar and is not accessible with-out dismantling the vacuum vessel. The working volume forthe cryostat is within the bucket. The vacuum pump, whichconsists of a molecular drag pump with a set of oil-free back-ing pumps together with a controller, is used for evacuatingthe vacuum space in the bucket Dewar. The isolation box con-taining a step-down transformer provides the main switch for

321

L Ye et al

Figure 4. Warming-up profile of the cryogenic system.

the power supply to the compressor. The compressor systemgenerates the high purity helium flow for the cryocooler.

2.4. Temperature controller and monitor

The temperature controller and monitor can be used tocontrol the temperature of the cryostat to a user-defined setpoint required for a given temperature. The temperature ofthe system is measured using the Omega monitor module(nonlinear thermocouple) and shown on the panel. Two type-T thermocouples are mounted on the outside of the bucket:one is on the bottom plate and the other is halfway up theoutside of the copper cylinder. Wiring for these sensors passesthrough the bottom plate of the vacuum vessel for connectionto the temperature controller. The cryogenic heaters, wound atthe cold head of the cryocooler and the bottom of the vesselrespectively, can be PID controlled with feedback from themonitored temperature.

3. Cryogenics system operations

For the present work, the procedure is to fill the bucket withliquid nitrogen around the MgB2 sample, then freeze thenitrogen. A heater, controlled by a feedback loop, can be used

to dial up the required temperature. Before cooling belowambient temperature and any cryogenic testing, the vacuumspace in the bucket Dewar must be evacuated to a pressurebelow 5×10−4 mbar with the pumps, because the performanceof the cryocooler is strongly influenced by the quality of thevacuum. The cooling-down and warming-up profiles of thecryogenic system are shown in figures 3 and 4 respectively.

The MgB2 element was mounted on a sample holder inthe cryostat and cooled down from 77 K using the cryocooler.The cryostat is cooled down from the temperature of liquidnitrogen. Liquid nitrogen is decanted into the sample spacethrough a funnel. About 30 l of liquid nitrogen are requiredto fill the space with the test insert in place. It takes about45 min to cool the system down to 77 K. The cryocooler isthen powered to begin the process of freezing the nitrogen,taking nearly 12 h to reach 24 K from 77 K. After about2 h the nitrogen reaches the freezing point of 65 K, and thenitrogen temperature reaches a minimum of 33 K after thecryocooler was operated for a further 5 h. From the evolutionof the temperature and voltage drop with time (figure 4), wecan observe the phase changes at 63 K (freezing) and 35 K(α/β solid phase change), respectively. When the operatingtemperature reaches 36 K, the sample starts to go into thesuperconducting state with the voltage dropping to zero.

The cryostat can be warmed up from the cryogenictemperature by simply turning off the cryocooler. The nitrogenwill melt then start to boil. It takes just 4 h to reach 77 K from24 K (figure 4) if the heater is turned on to speed up the warm-up procedure.

4. Sample testing set-up under AC pulse overcurrents

4.1. Experimental rig

A typical fault in an AC power system would have a veryfast rise time and would require a response by the SFCL in afraction of a cycle. To test the response of the MgB2 conductorto very short pulses that greatly exceed the materials’ criticalcurrents, an AC pulse generation system and a real-time dataacquisition system in LabView/DAQ were developed. Figure 5is a schematic diagram of the overcurrent test circuit which isused to evaluate the quench and current limitation behaviourof the selected MgB2 samples when AC pulsed currents are

Figure 5. The schematic diagram of the MgB2 sample test circuit.

322

Investigations of current limiting properties of the MgB2 wires subjected to pulse overcurrents in the benchtop tester

Figure 6. Current-electric field versus time under AC pulse currentsat 25 K.

applied. Compared with the frequently used discharge of acapacitance bank, the up-to-date experimental rig can createa 50 Hz AC pulse current with a controllable amplitude. Thisproved to be reliable for measuring the E–J characteristics andthe limiting behaviour. The pulse currents have the advantageof reducing the joule energy dissipation and heating at thecurrent lead contacts as well as avoiding an unstable burn-outwhich occurs with a steady DC current.

The experimental apparatus can be used to generate a largeAC current up to 2000 Apeak over a few cycles to evaluate theshort pulse response of the MgB2 samples. The primary sideof the variable autotransformer (variac) is connected via anelectronic switch to the 220 V utility source, and the tunablesecondary side of the variac is connected to the primary side ofa transformer with a winding ratio of 250/30. The secondaryside of this transformer is connected to the MgB2 sample. TheAC pulse current is applied to the sample and can be adjustedthrough varying taps on the variac. An electronic switchdetects the zero-crossing point of the current and can be set byLabView to switch off the feed with a fixed number of cyclesof AC pulses. Soldered voltage contacts on the sample areused to monitor the voltage drop at various points in the MgB2

sample. A 60 mV/250 A shunt resistor is used to measurethe current passing through the sample. The measured voltagefrom the voltage contacts can be amplified and also fed into thecomputer.

The voltage and current values are collected simultane-ously by using a multichannel real-time data acquisition card(hardware) via USB. The DAQ Assistant in LabView was usedinteractively to access the functionality of the data card, config-ure I/O controls, test panels, scale channels, triggering and pro-gram measurement tasks [8]. The associated graphical sourcecode can acquire data into analysis routines for display of theresults in engineering units on a graph and save them as anEXCEL file for post-processing.

4.2. MgB2 sample preparation

As sample conductors, we used MgB2 wires from DiborideConductors Ltd. The sample is stainless steel tube in Cusheath MgB2 monocore wire with heat treatment at 850 ◦C for

Figure 7. Enlarged current-electric field versus time under AC pulsecurrents at 25 K.

7 min in a flow of argon. The sample diameter and lengthare 1 mm and 10 cm, respectively. The critical current (Ic)of the specimen is 141 A at 25 K in self-field. The samplesare mounted on the outside of the sample mount, which is a250 mm diameter cylinder of fibreglass. The straight MgB2

wire is soldered to copper segments at each end (the currentcontacts) both for minimization of contact resistance heatingand for maximum thermal stability. Voltage taps are fixedon the sample at 15 mm intervals so that the voltage dropsgenerated in adjacent regions of the sample can be detectedsimultaneously.

5. AC pulse experiments results and discussions

5.1. Quench properties under AC pulse overcurrents

Distribution level circuit breakers typically open in two cyclesafter receiving the order to open. This plus one cycle delayand sensing time makes the fastest expected distribution levelopening time three cycles. In order to simulate the realsituation in power systems, the number of pulses was chosento be six. The MgB2 sample was mounted on the sampleholder in the cryostat and cooled down to around 25 Kusing the cryocooler. During the experiments, we variedthe amplitude of the AC pulse currents applied to the MgB2

sample by changing the tap of the variac. Voltage taps werefixed on the sample at 15 mm intervals so that the voltagein three adjacent regions of the sample could be detectedsimultaneously. Figures 6 and 7 show the example of measuredwaveforms of the current and electric field (voltage per unitlength) versus time when six pulses at 50 Hz were applied at25 K in self-field.

The MgB2 sample was in the superconducting state whencooled to 25 K and the electric field is zero. AC overcurrentswere suddenly supplied to the MgB2 sample for six cycles. Forvarious peak values of the overcurrent, the transport currentand voltage across each section could be measured. Whenthe applied AC pulse current amplitude reached 450 Apeak,the sample started to generate resistance in the first half-cycle due to the quench, and then the current was limited to300 Apeak in the second cycle. In the subsequent cycles the

323

L Ye et al

Table 1. Measured quench currents and critical currents at differentoperating temperatures.

4.2 K 25 K 29 K

Quench current (A) Iq 798 280 198Critical current (A) Ic (1 μV cm−1) 292 141 76Iq/Ic ∼2.7 ∼2 ∼2.6

voltages continue to develop, causing the current to decreasefurther due to the limiting effect. It was observed that allthe channels quenched uniformly in the first half-cycle andthe quench current of the MgB2 sample at 25 K is 280 Apeak

(figure 7). There was a sharp transient into the normal state inthe first half-cycle because voltages across the sample startedto develop due to the heating effects. A purely resistivevoltage appeared at the second half-cycle, for which thermalrunaway occurred. It was also observed that the electric fieldin each channel was in phase with the damping current. Theovercurrent was limited to 300 Apeak in the second cycle,and in the sixth cycle, it was further reduced to 200 Apeak

(figure 6). The maximum electric field under overcurrentreached 0.96 V cm−1 without destruction of the sample. Weobtained the DC critical current (Ic) 141 A at an electric fieldof 1 μV cm−1 at 25 K in self-field.

We have also carried out DC experiments at 4.2 and 28 K.Table 1 presents the measured DC critical currents based on the1 μV cm−1 criterion and quench currents at different operatingtemperatures in self-field.

It was observed that the difference between the AC quenchcurrent and the DC critical current measured varied by a factorof 2–3 for this case. The reason behind the difference hasbeen shown to be due to the thermal effect by simulations (seesection 6).

5.2. Bending tests

Electromechanical properties are one of the key issues whichneed to be considered in superconductor applications. Thecurrent-carrying capability of the superconducting wire isinfluenced by stress. Bending of the superconducting wirescauses complex stress patterns in the superconductors becauseit introduces tensile and compressive stresses to the wires.Thus, additional bending tests were also performed on theMgB2 wires at 4.2 and 25 K in self-field. In all cases, thewires were bent smoothly round a wooden former at roomtemperature, followed by measurements of quench currentsusing the same overcurrent pulsed system (figure 5).

The bending radius was assumed to be the radius of theround former used to bend the wire. The bending radii were10, 20 and 30 cm. The quench current versus bending radiusof the MgB2 was shown in figure 8. Before bending, thequench currents of the straight sample were 600 A @ 4.2 K and280 A @ 25 K. However, after being bent, the quench currentswere observed to decrease to 400 A @ 4.2 K and 150 A @ 25 K(bending radius = 30 cm), 350 A @ 4.2 K and 105 A @ 25 K(bending radius = 20 cm) and 260 A @ 4.2 K and 66 A @ 25 K(bending radius = 10 cm), respectively. Compared with thestraight (unbent) samples, the quench currents of bent sampleswere found to decrease about 30%–70%, corresponding tothe decreasing bending radii. That means the bending stressbecomes higher as the bending radius decreases.

Figure 8. Quench current versus bending radius of the MgB2

monocore wire.

The results in figure 8 indicate that bending results in asignificant degradation of the quench current with decreasingradius. This gives evidence of the importance of controlling themechanical properties of MgB2 wires during fabrication. It istherefore recommended that the MgB2 wires be wound beforeannealing to minimize the risk of a winding process inducedfailure.

6. Modelling and simulation

The difference between the AC quench currents and theDC critical currents was experimentally investigated. Aspreviously observed the difference between the AC quenchcurrents and the DC critical currents varied by a factor of 2–3. Therefore, modelling and simulation under the AC andDC conditions will be used to examine the reason behind thedifference.

The quench was modelled using single lumped materialand adiabatic conditions, i.e. a simple first-order nonlineardifferential equation. This is the variation of resistancewith current assumed. The parameters are chosen to give1 μV cm−1 at Ic (150 A at 27 K) and 0.999 of normal resistanceat 3Ic. The problem with modelling the quench is that 95%of quench behaviour takes place in an unstable region whereno measurements of the E–J characteristics are possible.However, a plausible approximation to the likely curve canbe used and the parameters fitted to the results. The Fermifunction for the occupation of electronic states is a suitable ‘S’shape with an exponential beginning. We assume a resistanceof the form as follows in equation (1):

R = Rn

1 + exp[k(Ic − I + I0)] . (1)

This has the required characteristics of a low value atlow currents and Rn at high currents. The point of rapidchange is determined by the fitting parameter I0, where k isthe other fitting parameter. The two fitting parameters werechosen so that at the measured DC critical current the fieldwas 1 μV cm−1, and at three times Ic the resistance was0.999 of the normal value. To follow the transition it wasassumed that Ic went linearly to zero between 25 K and the

324

Investigations of current limiting properties of the MgB2 wires subjected to pulse overcurrents in the benchtop tester

0 100 200 300 400 500 6000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Current

Res

ista

nce

R 27KR 38K

Figure 9. The resistance as a function of current. At 27 K Ic is 150 A(1 μV cm−1) and the resistance is 0.999 of the normal state at threetimes Ic.

Figure 10. The voltage plotted against time at various currents.

critical temperature. The curves are shown below for twotemperatures; then we can predict how the sample quenchesat various applied currents (figure 9).

Figure 10 shows the quench process at various constantcurrents assuming adiabatic heating. It can be seen that forcurrents not far above Ic it takes some time for the quench tooccur. This is because the resistivity is very low initially, soheating is slow. However, at some point the positive feedbackbetween the heating and consequent reduction in Ic leads toan extremely rapid quench to the normal state in about 20 μs.The closer the current to Ic the longer it takes for the quench tooccur.

The critical current is 150 A but if a quench is to occurwithin 8 ms we need to apply 210 A (figure 10). This explainsqualitatively the difference between the DC critical current,which is probably the true value, and the quench current.Quench times are shorter here because we are applying afixed maximum current rather than a sine wave, and in theexperiments the quench current was measured to be roughlytwice the DC critical current. The fact that this is very time-dependent needs to be taken into account in any systemsanalysis.

Figure 11. Voltage waveforms at various currents.

Figure 12. The time to switch against current.

Figure 11 shows the quench if the current is sinusoidal.The DC critical current is 150 A but it needs a current of 250 Ato switch in the first quarter-cycle. Currents less than 230 Atake a long time to switch, and when conduction to the outsideis included, will not switch at all. Figure 12 shows clearly howthe switching time depends on the peak current.

Modelling and simulation has confirmed the near factorof two difference between the AC quench current and the DCcritical current. The modelling and simulations have shownthat this is mainly due to a thermal effect.

7. Conclusion

A desktop tester system was developed to investigate thequench behaviour of the MgB2 samples under AC pulsecurrents at 50 Hz in self-field. The MgB2 wires startedfrom a superconducting state and demonstrated fault currentlimiting properties. Operation of the desktop tester withMgB2 samples demonstrated that the MgB2 wires have currentlimiting properties under pulse overcurrents. Therefore, MgB2

is a possible material option for future distribution-level SFCL

325

L Ye et al

applications. Also the results of this work are generallyapplicable for better understanding and further development oflarge-scale application devices.

Acknowledgments

The authors would like to express their gratitude to ProfessorJan Evetts, Dr A Kursumovic, Dr D Duvic and Mr D Astillat the University of Cambridge for their valuable help duringthe experiments. Thanks are also due to Dr P Malkin, Mr.B. Simmers and Dr S Weeks at Rolls-Royce Plc, UK. Weacknowledge the UK Scientific Magnetics, Oxfordshire for thedesign and manufacture of the cryogenic system and DiborideConductors Ltd at Cambridge for providing the MgB2 samples.We are especially grateful for support from Rolls-Royce plcand the UK Department of Trade & Industry (DTI) undercontract CHBL/C/019/00027–Superconducting Fault CurrentLimiter for Electrical Marine Propulsion (SuFCLEMP).

References

[1] Ye L et al 2002 Application studies of superconducting faultcurrent limiters in electric power systems IEEE Trans. Appl.Supercond. 12 900–3

[2] Ye L et al 2006 Behavior investigations of superconducting faultcurrent limiters in electric power systems IEEE Trans. Appl.Supercond. 16 662–5

[3] Nagamatsu J, Nakagawa N, Muranaka T, Zenitani Y andAkimitsu J 2001 The origin of multiple superconducting gapsin MgB2 Nature 410 63–4

[4] Grasso G 2005 Cost considerations for MgB2 conductorsSCENET-WG Workshop (Enschede, March 2005)

[5] Museniche R et al 2006 The behaviour of cryogen-free MgB2

react and wind coils Supercond. Sci. Technol. 19 126–31[6] Komarek P 2005 Application prospects for MgB2 conductors

SCENET-WG Workshop (Enschede, March 2005)[7] Harrison S 2005 User Manual for the Desktop Test Cryostat

(UK: Scientific Magnetics)[8] 2006 LabView/DAQ Manual version 8 (USA: National

Instruments)

326