- Email: [email protected]
Thermo hydraulic and quench propagation characteristics of SST-1 TF coil
Fusion Engineering and Design 89 (2014) 115–121
Contents lists available at ScienceDirect
Fusion Engineering and Design journal homepage: www.elsevier.com/locate/fusengdes
Thermo hydraulic and quench propagation characteristics of SST-1 TF coil A.N. Sharma a,∗ , S. Pradhan a , J.L. Duchateau b , Y. Khristi a , U. Prasad a , K. Doshi a , P. Varmora a , D. Patel a , V.L. Tanna a a b
Institute for Plasma Research, Gandhinagar, India CEA Cadarache, 13108 St Paul lez Durance Cedex, France
h i g h l i g h t s • • • • •
Details of SST-1 TF coils, CICC. Details of SST-1 TF coil cold test. Quench analysis of TF magnet. Flow changes following quench. Predictive analysis of assembled magnet system.
a r t i c l e
i n f o
Article history: Received 20 August 2013 Received in revised form 12 October 2013 Accepted 23 December 2013 Available online 25 January 2014 Keywords: SST-1 TF coil Quench Normal zone Gandalf Two phase and supercritical helium
a b s t r a c t SST-1 toroidal field (TF) magnet system is comprising of sixteen superconducting modified ‘D’ shaped TF coils. During single coil test campaigns spanning from June 10, 2010 till January 24, 2011; the electromagnetic, thermal hydraulic and mechanical performances of each TF magnet have been qualified at its respective nominal operating current of 10,000 A in either two-phase or supercritical helium cooling conditions. During the current charging experiments, few quenches have initiated either as a consequence of irrecoverable normal zones or being induced in some of the TF magnets. Quench evolution in the TF coils have been analyzed in detail in order to understand the thermal hydraulic and quench propagation characteristics of the SST-1 TF magnets. The same were also simulated using 1D code Gandalf. This paper elaborates the details of the analyses and the quench simulation results. A predictive quench propagation analysis of 16 assembled TF magnets system has also been reported in this paper. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Steady-state superconducting tokamak (SST-1) is designed for steady state operation, in both the single null and the double null configuration [1]. SST-1 has superconducting toroidal field (TF) and poloidal field (PF) magnet system. As a mandate of SST-1 mission, each TF coil is necessarily to be cold tested to its full operating conditions. This decision was reasonable as in a superconducting steady state tokamak configuration, the access to TF magnets for repair is very limited. The replacement of an assembled TF magnet may involve partial or full dismantling of the machine. Thus, each SST-1 TF magnet was consciously tested for its electromagnetic,
∗ Corresponding author. Tel.: +91 79 23962164; fax: +91 79 23962277. E-mail addresses: [email protected], [email protected] (A.N. Sharma). 0920-3796/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.fusengdes.2013.12.058
thermal hydraulic and mechanical performances at its full operating current in either two-phase or supercritical cooling conditions and has been qualified prior to its assembly on SST-1 machine shell. Quenches observed during these single coil test campaigns were analyzed in detail in order to understand its thermal hydraulic and quench propagation characteristics. Some of the quenches have also been also simulated using 1-D code Gandalf [2]. 2. SST-1 TF magnet system SST-1 TF magnet system consists of 16 superconducting, modified ‘D’ shaped coils arranged symmetrically around the major axis spaced 22.5◦ apart. It is designed to give magnetic flux density of 3.0 T at plasma axis with ripple <2% within the plasma volume at its nominal operating current of 10,000 A. Each TF coil is made up of 6 double pancakes (DP) with each pancake having nine turns. The base conductor for these magnets is the cable-in-conduit
116
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
Table 1 Technical details of SST-1 TF magnet system. Number of coils Shape Turn per coil Double pancakes per coil Rated current Field at plasma axis Maximum field Maximum field ripple Bore dimensions (radial) Bore dimensions (vertical)
16 Modified D 108 6 10 kA 3.0 T 5.1 T <2% 1190 mm 1746 mm
conductor (CICC) described in next section [3]. The main parameters of TF coils are summarized in Table 1. The winding pack is shrunk fitted into a SS316 L casing. All coils are connected in series and are protected against quench by suitable quench detection and protection system [4–6]. The straight legs of TF coils are wedged to form the inner vault to support the centering forces. The outer vault is formed by connecting outer inter coil structures (OICS) between the TF coils to resist the overturning torque experienced by the TF system from interaction with PF coils and plasma. The magnet winding pack is cooled by two phase helium or Supercritical Helium at 0.4 MPa and supply of 4.5 K. The nominal mass flow rate for each TF coil is 16 g/s. Cold helium is fed from the high field region of the magnet. For this purpose, inlet stubs have been welded to the double pancakes from the inner side of the winding pack. Originally, it was envisaged to cool the TF casings with cold helium flowing in SS tubes soldered on the coil casings. However this idea was abandoned due to multiple leaks observed in these tubes. Indigenously developed SS316L single sided bubble panel type supercritical helium cooled radiation shields were developed to introduce case cooling. These were welded to the inner ring of the TF coil case. These panels were also cold tested along with the TF magnets and were found to be helium leak tight offering very low pressure drops. Fig. 1 shows a pair of TF coil with OICS, joints, manifolds and the super critical helium cooled radiation shield.
Outer dimensions (radial) Outer dimensions (vertical) Average turn length Weight of one TF coil Centering force per coil Tension in the coil Total inductance Total stored energy Dump time constant Peak dump voltage
1560 mm 2120 mm 5500 mm 1920 kg 2.73 MN 90–110 MPa 1.12 H 56 MJ 12 s ±600 V
3. SST-1 CICC SST-1 TF and PF magnets are made using identical NbTi/Cu based CICC. Its main features are given in Table 2. Each strand of CICC has about 1224 ± 30 hexagonal NbTi filaments. Its twist directions, in all stages, are anti clockwise (Z) and the pitches may vary within ±10% of the specified nominal values. The last stage bundled cable was wrapped with 25 micron SS 304 foil with 50% overlap. The void fraction in the cable space, measured at random intervals along the length of the cable was 40 ± 2%. The choice of SS304L as the conduit material and its 1.5 mm thickness was motivated from the requirements that the conduit shall also be used as a load bearing structure inside the winding pack where the operating stresses may be as high as 300 MPa at cryogenic temperature. Structural analysis of TF coils under electromagnetic loads has been presented in [7]. Before mass production of SST-1 CICC by manufactures, a model coil (MC) test was done to validate the feasibility of using same type of NbTi based CICC for both TF and PF coil windings [8]. MC had double pancake type windings with hydraulic path lengths, flows and currents similar to SST-1 TF and PF magnets. Transient background fields were generated in longitudinal and transverse directions by separate solenoids. To validate CICC appropriateness for steady state operation of TF coils, MC was slowly charged up to 12,000 A corresponding to 6.2 T of maximum self field, with mass flow rate of 0.5 g/s. No quench was observed in this test. To demonstrate CICC suitability in all expected operational loads during different plasma operation stages and events like disruptions, background field solenoids were used to generate disturbances in parallel and transverse directions. Simultaneous ac parallel and transverse disturbances were applied to MC to simulate possible disturbances from feedback control coils during SST-1 operation. It was found that at 10 kA and 1.2 g/s CICC can withstand pulsed longitudinal disturbances up to 0.29 T as against the design value of 0.27 T corresponding to 330 kA plasma current is SST-1. CICC can withstand sinusoidal disturbances of amplitude more than 25 mT of 125 ms duration in any orientation. Fast ramp rate operational feasibility of CICC was demonstrated by charging MC with ramp rates corresponding to 2 T/s at mass flow rates of 0.8–0.9 g/s. Current sharing temperature was measured to be about 6 K at 10 kA (5.1 T) operation. Controlled quench tests using resistive heaters were done at different operating currents of 6–10 kA. The energy margin at 6 kA is about 0.75 MJ/m3 . Measured normal zone propagation speed was observed in between 1.3 and 2 m/s during these tests. Following the encouraging results as per the design values, SST-1 CICC mass production was allowed at the industry.
4. TF coils test program
Fig. 1. SST-1 TF coil pair.
In a tokamak configuration, access to TF magnets for repair is very limited and its replacement may involve partial or full dismantling of machine. So each TF magnet was tested for its electromagnetic, thermal hydraulic and mechanical performances at its
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
117
Table 2 SST-1 CICC and strand details. Operating current (A) Critical current at 5 T, 4.2 K (A) Cu: non Cu Cu RRR Number of strands Cabling pattern Cable twist pitch (mm) 1st Stage 2nd Stage 3rd Stage 4th Stage
10,000 35,000 ≥4.9 100 135 3×3×3×5 40 ± 4 75 ± 7 130 ± 13 290 ± 29
full operating current in either two-phase or supercritical cooling conditions. SST-1 TF coil test has following primary objectives: 1. To demonstrate the ability to carry transport current of 10,000 A in either supercritical helium cooling or two phase helium cooling modes. 2. Demonstration of joint resistances <5 n at 5 K and 10 kA of transport current for all inter-double pancake joints. 3. Demonstration of the helium leak tightness of integrated magnet with leak acceptance limit of less than 1 × 10−6 Pa m3 s−1 at 0.4 MPa, 4.5 K supercritical helium. The integrated magnet means magnet with 5 inter-pancake joints, 2 busbar joints, 12 electrical breaks of inlet and outlet manifolds and all the hydraulic connections that exist inside the experimental cryostat. 4. Demonstration of ground insulation resistance (between the winding pack and that of the grounded structure) greater than 1 M at 4.5 K with 1 kV DC voltage. 5. To study the normal zone related thermo hydraulic aspects in supercritical helium cooling or two phase helium cooling modes. All 16 TF magnets have been individual tested during June 10, 2010 to January 24, 2011 in a dedicated coil test cryostat [9,10]. All magnets have successfully demonstrated the above mentioned test objectives. During the current charging experiments of these coils few quenches have been observed at near nominal operating current of TF coils. Origin of these quenches was linked to the heat loads from coil casings and coil support structures inside the cryostat. No quenches were observed, if coil case temperature was maintained to less than 30 K during current charging. In some current charging experiments, quenches were observed in the +ve and −ve busbar sections. These quenches have also been associated with the heat loads from busbar supports as these supports were taken from coil casings.
Void fraction % Conduit material Outer dimension (mm2 ) Conduit thickness (mm) Strand diameter (mm) Filament diameter (m) Cu area in strand (mm2 ) Filament twist pitch (mm) Hysteresis loss at ±3 T, 4.2 K (mJ/cm3 ) for strand Index ‘n’
40 ± SS304L 14.8 × 14.8 (±0.05) 1.5 ± 0.08 0.86 ± 0.005 10 ≥0.482 12 ± 2 <100 >25
Calibrated cernox and carbon ceramic temperature sensors were installed at inlet and outlet headers, outlet of inter-double pancake joints, busbar joints, current lead section and TF thermal shield inlet [11]. Calibrated pressure transmitters were placed at inlet and outlet header and TF thermal shield inlet. Calibrated venturi flow meters were placed at main inlet, main outlet and outlet of −ve current lead section. Calibrated cryogenic hall probes were installed at inner straight section of TF coil to measure the magnetic field generated by the coil. For measurements of all these sensors, modular signal conditioning electronics with very high accuracy, stability and with isolation had been developed in-house [12]. Measurement accuracy for temperature sensors, hall probes, pressure sensors, flow meters was, respectively, ±0.1 K for 4.5 K to 80 K temperature range, 0.1%, 0.1% and ∼2%. Voltage taps were installed across each DP at the joint locations, across busbar joint locations and across the current lead sections. For quench detection logic, each DP was compared with its neighboring DP, +ve busbar was compared with −ve busbar and +ve current lead was compared with −ve current lead. If the difference voltage between these sections becomes greater than ±150 mV for more than 100 ms then quench trigger is generated by the quench detection hardware system. These threshold voltage and time levels were selected on the basis of maximum allowable hotspot temperature and the levels of expected electromagnetically coupled voltage levels on the voltage signals of voltage taps. The signals from all voltage taps, temperature sensors, mass flow meters, hall probes and pressure transmitters were continuously acquired during the current charging experiments to study the coil behavior under electromagnetic loads. PXI based data acquisition system was developed for coil test experiments. Its flexible GUI based application program allowed to monitor and store the data in slow rates during cool-down and warm up and at faster rates during current charging experiments [13]. 6. TF 15 quench details
5. TF coil hydraulics, sensors and diagnostics TF coil hydraulic scheme and sensor and diagnostics during the coil test are shown in Fig. 2. As shown, helium from inlet manifold enters middle of each double pancake (DP1 to DP6). Helium exits are on inter-double pancake joints (not shown in figure). This helium is then collected in outlet manifold (except half of DP1 and DP6). Helium of the first half of DP1 and the second half of DP6 flows through +ve busbar and current lead and −ve busbar and current leads respectively. Helium of these two sections is collected and fed to radiation shield of coil casing. Helium from outlet manifold and TF thermal shield outlet flows back to 1.3 kW at 4.5 K helium refrigerator/liquefier system via the outlet header. Electrical isolators are used at appropriate locations to avoid electrical shorting. So when the nominal flow of 16 g/s is given at inlet header, each DP will receive 2.5 g/s and each pancake will have 1.25 g/s flow. Since TF thermal shield is fed with outlet of 2 single pancakes, it will have mass flow rate of 2.5 g/s.
SST-1 TF coil 15 was cooled using supercritical helium with inlet temperature at 5.67 K and inlet pressure of 0.386 MPa. Total mass flow rate measured at Inlet of coil was 27 g/s. Outlet temperature was 6.032 K and pressure was 0.338 MPa. During the current charging shot, coil quenched during ramp up phase when current was about 8960 A. When voltage tap signals were analyzed it was observed that DP6 section of the coil has quenched. The voltage tap signals of all pancakes along with the coil current are shown in Fig. 3a. Changes in flow, pressure and temperature signals are shown in Fig. 3b–d, respectively. 7. Normal zone propagation speed estimation As the quench was generated in the coil during the rampup phase, measured voltage across DP6 has inductive component in addition to the normal zone voltage. This voltage will be
118
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
Fig. 2. TF coil test sensors and hydraulic layout.
proportional to self and mutual inductances of the double pancakes and the current ramp rate of the coil. Inductive voltage component across each DP is about 94 mV. The total voltage developed across DP6 is 556 mV. Correcting it for inductive voltage, actual normal zone voltage signal is calculated to be about 184 mV (as quench detection card has a gain of 2). This quench voltage has developed in time of 1.17 s. As the coil current ramp rate is only 20 A/s, the operating current of 8960 A is considered constant during the quench development time. SST-1 CICC resistance was calculated using copper resistivity of 0.23 n m, at temperature of about 15 K (RRR 100, magnetic field 2.2 T). Under these conditions, SST-1 CICC resistance is calculated to be 3.4 /m. Using the normal zone voltage and the
(a)
(b)
0.2 0.0 -0.2 -0.4 -0.6 Quench Trigger
672.8
673.2
673.6
Mian Inlet Flow (g/s)
0.4
2000
40
6
4 20 2 Quench Trigger
-0.8 0 674.0
674.4
0 672
679
6
(d)
PS Current Inlet Pressure DP6 Outlet Pressure
8000
Main Inlet Busbar Joint 2 Nr DP6 flow meter
24
20 5
4000 4 2000
Pressure (bar)
6000
Temperature (K)
(c)
686
Relative time (s)
Relative time (s)
Coil Current (A)
8
0.6
4000
672.4
Main Inlet Flow DP6 Outlet flow
16
Quench Trigger
12
8 Quench Trigger
0
3 670
672
674
676
678
680
682
684
686
688
690
692
4 680
Relative time (s)
720
Relative time (s) Fig. 3. Quench development in TF15 coil.
760
800
DP6 Outlet flow (g/s)
Coil Current (A)
A rather unexpected result of increase in inlet mass flow rate has been observed in TF 15 quench as shown in Fig. 3b. This is a new observation, as in general flow reduction following quench have been reported or the results of simulations have predicted it [14–17]. Following the quench, initially inlet flow decreased slightly, and then it increased substantially.
DP VOltage (V)
6000
8. Inlet flow increase following the quench
0.8
Current DP1 DP2 DP3 DP4 DP5 DP6
8000
CICC resistance, it is found that about 5.93 m length of the CICC has quenched in time of ∼1.17 s. This gives a normal zone propagation speed of ∼5.07 m/s.
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
rate to the CICC. ‘Jun 3’ and ‘Jun 4’ are check valves, which after a threshold pressure; open the connections to ‘Vol 3’, a large adiabatic volume representing a relief reservoir. ‘Vol 1’ and ‘Vol 2’ are small volumes representing inlet and outlet manifolds.
10000
DelP Current
8000
6000 1.2 4000
Current (A)
Differential Pressure (bar)
1.8
2000 0.6 0
19:10:08
19:10:18
19:10:28
119
19:10:38
time Fig. 4. Pressure drop across the coil following TF 15 coil quench.
We tried to find the reason for this. When data from the cryogenic plant was analyzed it was found that during quench propagation period, difference between inlet and outlet pressure of the coil initially decreased slightly, and then it increased substantially. Probably due to quench which occurred near outlet, as explained in later sections, the outlet pressure started increasing. In order to maintain the pressure difference across the load, plant also started increasing the inlet pressure. This response of the plant along with the thermal hydraulic effects of quench led to pressure fluctuations across the coil. As the pressure difference is closely linked to the mass flow rate in the coil, increased pressure drop as plotted in Fig. 4, led to increase in mass flow rate of coil initially and then it returned to its normal value after about 20 s. This behavior was observed in few other coils as well during the coil test. A database has been generated of the level of these fluctuations and the time taken by these fluctuations to reach the inlet/outlet sensors. It will be used to generate the threshold limits for secondary quench detection and will be reported in detail at a later stage. 9. Quench simulation TF 15 quench was simulated using code GANDALF 2.1, which is a one-dimensional finite elements code for simulation of the quench and stability of CICC based magnets. Gandalf code’s input parameters, which include CICC specifications and the operating conditions, were generated from the details provided earlier in this paper. This code has option of modifying its subroutines to take care of different CICC designs [18]. Coefficients of Gandalf subroutine exts.f which implements the relation between critical current, temperature and critical field for NbTi were adjusted to obtain the critical current density (Jc) of 2853 A/mm2 at 5 T, 4.2 K. Gandalf implemented Jc values are plotted in Fig. 5a along with the experimental measured Jc values [19] which show a good matching. Other subroutines of this code like exth.f which implements the heat transfer coefficients, extf.f which implements friction factor were also modified to adapt the code to use properties relevant for SST-1 CICC. Heat transfer coefficients and friction factor correlations were implemented as reported in model coil simulation report [20]. Magnetic field ‘B’ variation along the DP6 length was calculated using M’C code [21]. This spatial variation of magnetic field was implemented using ‘extb.f’ subroutine of Gandalf. Gandalf implemented ‘B’ is plotted along the DP6 length in Fig. 5b. A simplified model of cryogenic plant associated with the coil test was implemented using flower subroutine of the code and is shown in Fig. 5c. In this model ‘Jun 1’ is CICC which is simulated by Gandalf. ‘Jun 2’ is the volumetric pump to provide the required mass flow
10. Simulation results As during the experiment, major temperature rise was observed at the outlet of the coil as compared to inlet so it was expected that quench occurred near to outlet on the outermost turns of DP6. Also CICC of this turn and the innermost turn are in contact on 2 sides with the coil casings. Rests of the turns of last DP have only one side in contact with the coil casing. Since exact location of quench initiation was not known, a parametric study was done by initiating quench in different locations of last turn of DP6. As the quench was initiated from the heat load coming from coil casing, in Gandalf heat load was distributed over long lengths of about 1 m. The amount of heat load was taken as just the minimum required to initiate a non-recoverable normal zone. After the parametric study by varying amount of heat load, location of heat load and length of heated zone a good matching was found with the experimental results when quench was initiated by heating 1 m length between 43 m and 44 m distance from inlet. Linear heat flux of 1200 W/m with time duration of 50 ms was used in this trial. Other parameters like helium temperature, pressure, mass flow rate and operating current were as per the experimental conditions described earlier. Quench development time during the experiment was about 1.17 s after which the threshold voltage of quench detection circuit was reached and dump was initiated after 100 ms delay time. So in Gandalf also, current was maintained at 8960 A for 1.17 s and after which current is dumped with experimental dump time constant. To compare the experimental plot and Gandalf plots of normal zone voltage development, zero in the time line of Gandalf plot was shifted by 672.9 s. This is the time at which quench was initiated as per the experiment data. The comparison of normal zone voltage is shown in Fig. 6a. The figure shows a good matching till the point marked as ‘quench triggered’. This is the point at which in the experiment, quench dump was initiated. So the rapid rise in experimental quench voltage is due to generation of large dump voltage and it is not the rise of normal zone voltage. In the simulation this dump voltage is missing and current decay initiates smoothly as programmed in Gandalf input file. So Gandalf normal zone voltage shows a reducing trend after this point. The rise in temperature at DP6 inlet and outlet during the experiment and in Gandalf simulation shows similar peak values. This is an important result as it has given the information about probable the hot spot temperature inside the magnet. Since the temperature at the outlet and inlet could be simulated within ±1 K, peak value at hotspot temperature inside is also expected to match within similar error margin. As actual hotspot temperature cannot be measured in big size actual magnets like TF coil of a Tokamak, we can make use of these simulation results to predict it. This is discussed in next section. The observed differences in simulation results with experiment result are also because heating due to ac losses are not included in simulation. 11. Predictive quench simulation for assembled TF magnet system The code modified for TF15 simulation has been used to do the predictive quench simulation for 16 TF magnets assembled in series. The main difference between a stand-alone coil and assemble TF magnet system will be in the background magnetic field. In single coil test, at the operating current of 10,000 A, the expected
120
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
Fig. 5. Implementation of SST-1 CICC and coil test parameters in Gandalf.
peak magnetic field is 2.2 T where as in assembled magnet system it will be 5.1 T at the same operating current. This is due to superposition of magnetic field from neighboring TF coils. Also dump time constant will be 12 s against the 1.7 s used for single coil test. So keeping all other input parameters of Gandalf same, the magnetic field distribution was changed and hot spot temperature was simulated with a dump time constant of 12 s. The maximum hot spot temperature is found to be about 118 K. This is under the design temperature limit of 150 K. So SST-1 TF coil quench detection system design parameters like threshold voltage and delay time are also acceptable.
0.4
Another predictive simulation was done for the hypothetical case of missed quench by fast diagnostic of voltage measurement method. This quench then will be detected by some slow diagnostics like flow, pressure or temperature sensor. As observed from different coil test data, the maximum time delay of about 1.8 s is present before quench induced pressure, flow or temperature is recorded. So a worst case of delay time of 3 s was considered for this study. For this case operating current of 10,000 A, background magnetic field of 5.1 T and dump time constant of 12 s were selected. Quench was generated in the middle of one of the DP. Dump was initiated after a delay time of 3 s in Gandalf subroutine. The
Experimental Gandalf
16
Voltage (V)
0.2
Temperature difference (K)
14
Quench Triggered
Expt Inlet Expt Outlet Gandalf inlet Gandalf outlet
12 10 8 6 4 2 0
0.0 673
674
time
676
680
time
Fig. 6. (a) Normal zone voltage comparison. (b) Inlet and outlet temperature comparison.
684
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
simulated hot spot temperature for this case is found to be 180 K. This is 30 K higher than SST-1 design limit of 150 K but still under the generally considered safe limit of 250 K for fusion grade magnets. 12. Conclusion All SST-1 TF magnets have been individually cold tested at its nominal operating current is two phase or supercritical helium flow conditions successfully. All design parameters of SST-1 CICC as tested in MC test and now as wound TF magnets have been verified. Unconventional results of mass flow increase at inlet following quench was observed in many coil tests. It was found to be linked to the helium plant response to the quench events. Quench observed in TF15 coil was analyzed to estimate the normal zone propagation speed. This quench was simulated using 1-D code, after suitably modifying its subroutines for SST-1 CICC. Simulation results have given detailed information about the hotspot temperature and the normal zone propagation speed. Reasonably good agreement with the experimental data has been found in the main parameters of normal zone propagation speed and the temperature rise. These results were used to make predictive analysis for assembled TF magnet. A case of missed quench by voltage tap signal was also simulated. It was found that even with the delay of 3 s, the hot spot temperature is about 180 K. So TF magnets are safe even for this case when quench is detected by a slow response diagnostic of flow or pressure. Results of these analyses confirm the suitability of the quench detection algorithms of SST-1 TF magnets. References [1] S. Pradhan, SST-1 Mission Team, Status of SST-1 refurbishment, Journal of Fusion Research Series 9 (2010) 650–657. [2] L. Bottura, A numerical model for the simulation of quench in the ITER magnets, Journal of Computational Physics 124 (1) (1996). [3] S. Pradhan, Y.C. Saxena, O.P. Subrat Das, D.P. Anashkin, V.E. Ivanov, Keilin, et al., Superconducting cable-in-conduit-conductor for SST-1 magnets, in: 2nd IAEA Technical Committee Meeting on Steady State Operation of Magnetic Fusion Devices, vol. 2, Fukuoka, Japan, 1999, p. 482. [4] A.N. Sharma, C.J. Hansalia, Y. Yeole, G. Bansal, S. Pradhan, A. Moitra, et al., Quench detection and data acquisition system for SST-1 superconducting magnets, Fusion Engineering and Design 74 (November (1–4)) (2005) 819–823.
121
[5] A.N. Sharma, Y. Yeole, D. Sonara, A. Moitra, S. Pradhan, B. Sarkar, et al., Quench detection and protection system for superconducting magnet of SST-1 tokamak, in: 4th IAEA TCM on Steady State Operation of magnetic Fusion Devices, February 1–5, Ahmedabad, 2005. [6] Y. Khristi, A.N. Sharma, K. Doshi, U. Prasad, P. Varmora, S. Kedia, et al., Quench detection system for TF coil-test campaigns of SST-1, IEEE Transactions on Applied Superconductivity 22 (April (2)) (2012) 4200108. [7] http://www.comsol.com/paper/magneto-structural-analysis-of-fusion-gradesuperconducting-toroidal-field-coils-7362 [8] O. Anashkin, D.P. Ivanov, V.E. Keilin, N.Ph. Kopeikin, I.A. Kovalev, A.V. Krivykh, et al., Results of the model coil tests for the cable-in-conduit conductor for SST-1 tokamak, IEEE Transactions on Applied Superconductivity 12 (March (1)) (2002), pp. 567, 570. [9] S. Pradhan, SST-1 toroidal field magnet tests: some results and lessons learnt, IEEE Transactions on Applied Superconductivity 22 (June (3)) (2012), p. 9501804. [10] S. Kedia, S. Pradhan, Y. Khristi, U. Prasad, K. Doshi, A.N. Sharma, Measurement of electromagnetic and thermal stresses on conduction-cooled joints of the SST-1 spare TF coil, IEEE Transactions on Applied Superconductivity 20 (December (6)) (2010). [11] A.N. Dipak Patel, U. Sharma, P. Prasad, Y. Varmora, K. Khristi, Doshi, et al., Thermal anchoring of wires in large scale superconducting coil test experiment, Fusion Engineering and Design 88 (Ju2ne (5)) (2013) 374–379. [12] K. Doshi, Y. Khristi, A.N. Sharma, P. Varmora, S. Kedia, U. Prasad, et al., Precision signal conditioning electronics for cryogenic temperature and magnetic field measurements in SST-1, IEEE Transactions on Plasma Science 40 (2012) 641–645. [13] P. Varmora, A.N. Sharma, U. Prasad, D. Patel, K. Doshi, Y. Khristi, et al., PXI based data acquisition system for SST-1 TF test program, Indian Journal of Cryogenics 38 (January) (2013) 104– 109. [14] M. Sugimoto, T. Isono, K. Yoshida, H. Tsuji, I. Takano, T. Hamajima, et al., An evaluation of the inlet flow reduction for a cable-in-conduit conductor in pulsed operation, Cryogenics 38 (October (10)) (1998) 989–994. [15] M. Sugimoto, T. Kato, K. Kawano, T. Hiyama, Y. Kamiyauchi, H. Ishida, et al., Test results of the DPC-TJ: thermal and hydraulic performance, Cryogenics 33 (6) (1993) 597–602. [16] K. Yoshida, H. Takigami, H. Kubo, Analytical studies on the hotspot temperature of cable-in-conduit conductors, Cryogenics 41 (August (8)) (2001) 583–594. [17] S. Nicollet, B. Lacroix, D. Bessette, R. Copetti, J.L. Duchateau, M.C. Gouachet, et al., Thermal-hydraulic behaviour of the ITER TF system during a quench development, Fusion Engineering and Design 86 (October (6–8)) (2011) 1497–1500. [18] User Manual, Gandalf version 2.1. [19] Performance of Cable-in-Conduit-Conductors for SST-1 at IPR, HITACHI CABLE REVIEW No. 20 (August 2001). [20] A. Sharma, J.L. Duchateau, Simulation of SST-1 model coil behaviour using Gandalf + Flower, AIM/NTT-2006.036. [21] L. Bottura, Analytical Calculation of Magnetic Field in an Isoparametric Brick, CryoSoft Internal Note CRYO-97-2, 1997.
Contents lists available at ScienceDirect
Fusion Engineering and Design journal homepage: www.elsevier.com/locate/fusengdes
Thermo hydraulic and quench propagation characteristics of SST-1 TF coil A.N. Sharma a,∗ , S. Pradhan a , J.L. Duchateau b , Y. Khristi a , U. Prasad a , K. Doshi a , P. Varmora a , D. Patel a , V.L. Tanna a a b
Institute for Plasma Research, Gandhinagar, India CEA Cadarache, 13108 St Paul lez Durance Cedex, France
h i g h l i g h t s • • • • •
Details of SST-1 TF coils, CICC. Details of SST-1 TF coil cold test. Quench analysis of TF magnet. Flow changes following quench. Predictive analysis of assembled magnet system.
a r t i c l e
i n f o
Article history: Received 20 August 2013 Received in revised form 12 October 2013 Accepted 23 December 2013 Available online 25 January 2014 Keywords: SST-1 TF coil Quench Normal zone Gandalf Two phase and supercritical helium
a b s t r a c t SST-1 toroidal field (TF) magnet system is comprising of sixteen superconducting modified ‘D’ shaped TF coils. During single coil test campaigns spanning from June 10, 2010 till January 24, 2011; the electromagnetic, thermal hydraulic and mechanical performances of each TF magnet have been qualified at its respective nominal operating current of 10,000 A in either two-phase or supercritical helium cooling conditions. During the current charging experiments, few quenches have initiated either as a consequence of irrecoverable normal zones or being induced in some of the TF magnets. Quench evolution in the TF coils have been analyzed in detail in order to understand the thermal hydraulic and quench propagation characteristics of the SST-1 TF magnets. The same were also simulated using 1D code Gandalf. This paper elaborates the details of the analyses and the quench simulation results. A predictive quench propagation analysis of 16 assembled TF magnets system has also been reported in this paper. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Steady-state superconducting tokamak (SST-1) is designed for steady state operation, in both the single null and the double null configuration [1]. SST-1 has superconducting toroidal field (TF) and poloidal field (PF) magnet system. As a mandate of SST-1 mission, each TF coil is necessarily to be cold tested to its full operating conditions. This decision was reasonable as in a superconducting steady state tokamak configuration, the access to TF magnets for repair is very limited. The replacement of an assembled TF magnet may involve partial or full dismantling of the machine. Thus, each SST-1 TF magnet was consciously tested for its electromagnetic,
∗ Corresponding author. Tel.: +91 79 23962164; fax: +91 79 23962277. E-mail addresses: [email protected], [email protected] (A.N. Sharma). 0920-3796/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.fusengdes.2013.12.058
thermal hydraulic and mechanical performances at its full operating current in either two-phase or supercritical cooling conditions and has been qualified prior to its assembly on SST-1 machine shell. Quenches observed during these single coil test campaigns were analyzed in detail in order to understand its thermal hydraulic and quench propagation characteristics. Some of the quenches have also been also simulated using 1-D code Gandalf [2]. 2. SST-1 TF magnet system SST-1 TF magnet system consists of 16 superconducting, modified ‘D’ shaped coils arranged symmetrically around the major axis spaced 22.5◦ apart. It is designed to give magnetic flux density of 3.0 T at plasma axis with ripple <2% within the plasma volume at its nominal operating current of 10,000 A. Each TF coil is made up of 6 double pancakes (DP) with each pancake having nine turns. The base conductor for these magnets is the cable-in-conduit
116
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
Table 1 Technical details of SST-1 TF magnet system. Number of coils Shape Turn per coil Double pancakes per coil Rated current Field at plasma axis Maximum field Maximum field ripple Bore dimensions (radial) Bore dimensions (vertical)
16 Modified D 108 6 10 kA 3.0 T 5.1 T <2% 1190 mm 1746 mm
conductor (CICC) described in next section [3]. The main parameters of TF coils are summarized in Table 1. The winding pack is shrunk fitted into a SS316 L casing. All coils are connected in series and are protected against quench by suitable quench detection and protection system [4–6]. The straight legs of TF coils are wedged to form the inner vault to support the centering forces. The outer vault is formed by connecting outer inter coil structures (OICS) between the TF coils to resist the overturning torque experienced by the TF system from interaction with PF coils and plasma. The magnet winding pack is cooled by two phase helium or Supercritical Helium at 0.4 MPa and supply of 4.5 K. The nominal mass flow rate for each TF coil is 16 g/s. Cold helium is fed from the high field region of the magnet. For this purpose, inlet stubs have been welded to the double pancakes from the inner side of the winding pack. Originally, it was envisaged to cool the TF casings with cold helium flowing in SS tubes soldered on the coil casings. However this idea was abandoned due to multiple leaks observed in these tubes. Indigenously developed SS316L single sided bubble panel type supercritical helium cooled radiation shields were developed to introduce case cooling. These were welded to the inner ring of the TF coil case. These panels were also cold tested along with the TF magnets and were found to be helium leak tight offering very low pressure drops. Fig. 1 shows a pair of TF coil with OICS, joints, manifolds and the super critical helium cooled radiation shield.
Outer dimensions (radial) Outer dimensions (vertical) Average turn length Weight of one TF coil Centering force per coil Tension in the coil Total inductance Total stored energy Dump time constant Peak dump voltage
1560 mm 2120 mm 5500 mm 1920 kg 2.73 MN 90–110 MPa 1.12 H 56 MJ 12 s ±600 V
3. SST-1 CICC SST-1 TF and PF magnets are made using identical NbTi/Cu based CICC. Its main features are given in Table 2. Each strand of CICC has about 1224 ± 30 hexagonal NbTi filaments. Its twist directions, in all stages, are anti clockwise (Z) and the pitches may vary within ±10% of the specified nominal values. The last stage bundled cable was wrapped with 25 micron SS 304 foil with 50% overlap. The void fraction in the cable space, measured at random intervals along the length of the cable was 40 ± 2%. The choice of SS304L as the conduit material and its 1.5 mm thickness was motivated from the requirements that the conduit shall also be used as a load bearing structure inside the winding pack where the operating stresses may be as high as 300 MPa at cryogenic temperature. Structural analysis of TF coils under electromagnetic loads has been presented in [7]. Before mass production of SST-1 CICC by manufactures, a model coil (MC) test was done to validate the feasibility of using same type of NbTi based CICC for both TF and PF coil windings [8]. MC had double pancake type windings with hydraulic path lengths, flows and currents similar to SST-1 TF and PF magnets. Transient background fields were generated in longitudinal and transverse directions by separate solenoids. To validate CICC appropriateness for steady state operation of TF coils, MC was slowly charged up to 12,000 A corresponding to 6.2 T of maximum self field, with mass flow rate of 0.5 g/s. No quench was observed in this test. To demonstrate CICC suitability in all expected operational loads during different plasma operation stages and events like disruptions, background field solenoids were used to generate disturbances in parallel and transverse directions. Simultaneous ac parallel and transverse disturbances were applied to MC to simulate possible disturbances from feedback control coils during SST-1 operation. It was found that at 10 kA and 1.2 g/s CICC can withstand pulsed longitudinal disturbances up to 0.29 T as against the design value of 0.27 T corresponding to 330 kA plasma current is SST-1. CICC can withstand sinusoidal disturbances of amplitude more than 25 mT of 125 ms duration in any orientation. Fast ramp rate operational feasibility of CICC was demonstrated by charging MC with ramp rates corresponding to 2 T/s at mass flow rates of 0.8–0.9 g/s. Current sharing temperature was measured to be about 6 K at 10 kA (5.1 T) operation. Controlled quench tests using resistive heaters were done at different operating currents of 6–10 kA. The energy margin at 6 kA is about 0.75 MJ/m3 . Measured normal zone propagation speed was observed in between 1.3 and 2 m/s during these tests. Following the encouraging results as per the design values, SST-1 CICC mass production was allowed at the industry.
4. TF coils test program
Fig. 1. SST-1 TF coil pair.
In a tokamak configuration, access to TF magnets for repair is very limited and its replacement may involve partial or full dismantling of machine. So each TF magnet was tested for its electromagnetic, thermal hydraulic and mechanical performances at its
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
117
Table 2 SST-1 CICC and strand details. Operating current (A) Critical current at 5 T, 4.2 K (A) Cu: non Cu Cu RRR Number of strands Cabling pattern Cable twist pitch (mm) 1st Stage 2nd Stage 3rd Stage 4th Stage
10,000 35,000 ≥4.9 100 135 3×3×3×5 40 ± 4 75 ± 7 130 ± 13 290 ± 29
full operating current in either two-phase or supercritical cooling conditions. SST-1 TF coil test has following primary objectives: 1. To demonstrate the ability to carry transport current of 10,000 A in either supercritical helium cooling or two phase helium cooling modes. 2. Demonstration of joint resistances <5 n at 5 K and 10 kA of transport current for all inter-double pancake joints. 3. Demonstration of the helium leak tightness of integrated magnet with leak acceptance limit of less than 1 × 10−6 Pa m3 s−1 at 0.4 MPa, 4.5 K supercritical helium. The integrated magnet means magnet with 5 inter-pancake joints, 2 busbar joints, 12 electrical breaks of inlet and outlet manifolds and all the hydraulic connections that exist inside the experimental cryostat. 4. Demonstration of ground insulation resistance (between the winding pack and that of the grounded structure) greater than 1 M at 4.5 K with 1 kV DC voltage. 5. To study the normal zone related thermo hydraulic aspects in supercritical helium cooling or two phase helium cooling modes. All 16 TF magnets have been individual tested during June 10, 2010 to January 24, 2011 in a dedicated coil test cryostat [9,10]. All magnets have successfully demonstrated the above mentioned test objectives. During the current charging experiments of these coils few quenches have been observed at near nominal operating current of TF coils. Origin of these quenches was linked to the heat loads from coil casings and coil support structures inside the cryostat. No quenches were observed, if coil case temperature was maintained to less than 30 K during current charging. In some current charging experiments, quenches were observed in the +ve and −ve busbar sections. These quenches have also been associated with the heat loads from busbar supports as these supports were taken from coil casings.
Void fraction % Conduit material Outer dimension (mm2 ) Conduit thickness (mm) Strand diameter (mm) Filament diameter (m) Cu area in strand (mm2 ) Filament twist pitch (mm) Hysteresis loss at ±3 T, 4.2 K (mJ/cm3 ) for strand Index ‘n’
40 ± SS304L 14.8 × 14.8 (±0.05) 1.5 ± 0.08 0.86 ± 0.005 10 ≥0.482 12 ± 2 <100 >25
Calibrated cernox and carbon ceramic temperature sensors were installed at inlet and outlet headers, outlet of inter-double pancake joints, busbar joints, current lead section and TF thermal shield inlet [11]. Calibrated pressure transmitters were placed at inlet and outlet header and TF thermal shield inlet. Calibrated venturi flow meters were placed at main inlet, main outlet and outlet of −ve current lead section. Calibrated cryogenic hall probes were installed at inner straight section of TF coil to measure the magnetic field generated by the coil. For measurements of all these sensors, modular signal conditioning electronics with very high accuracy, stability and with isolation had been developed in-house [12]. Measurement accuracy for temperature sensors, hall probes, pressure sensors, flow meters was, respectively, ±0.1 K for 4.5 K to 80 K temperature range, 0.1%, 0.1% and ∼2%. Voltage taps were installed across each DP at the joint locations, across busbar joint locations and across the current lead sections. For quench detection logic, each DP was compared with its neighboring DP, +ve busbar was compared with −ve busbar and +ve current lead was compared with −ve current lead. If the difference voltage between these sections becomes greater than ±150 mV for more than 100 ms then quench trigger is generated by the quench detection hardware system. These threshold voltage and time levels were selected on the basis of maximum allowable hotspot temperature and the levels of expected electromagnetically coupled voltage levels on the voltage signals of voltage taps. The signals from all voltage taps, temperature sensors, mass flow meters, hall probes and pressure transmitters were continuously acquired during the current charging experiments to study the coil behavior under electromagnetic loads. PXI based data acquisition system was developed for coil test experiments. Its flexible GUI based application program allowed to monitor and store the data in slow rates during cool-down and warm up and at faster rates during current charging experiments [13]. 6. TF 15 quench details
5. TF coil hydraulics, sensors and diagnostics TF coil hydraulic scheme and sensor and diagnostics during the coil test are shown in Fig. 2. As shown, helium from inlet manifold enters middle of each double pancake (DP1 to DP6). Helium exits are on inter-double pancake joints (not shown in figure). This helium is then collected in outlet manifold (except half of DP1 and DP6). Helium of the first half of DP1 and the second half of DP6 flows through +ve busbar and current lead and −ve busbar and current leads respectively. Helium of these two sections is collected and fed to radiation shield of coil casing. Helium from outlet manifold and TF thermal shield outlet flows back to 1.3 kW at 4.5 K helium refrigerator/liquefier system via the outlet header. Electrical isolators are used at appropriate locations to avoid electrical shorting. So when the nominal flow of 16 g/s is given at inlet header, each DP will receive 2.5 g/s and each pancake will have 1.25 g/s flow. Since TF thermal shield is fed with outlet of 2 single pancakes, it will have mass flow rate of 2.5 g/s.
SST-1 TF coil 15 was cooled using supercritical helium with inlet temperature at 5.67 K and inlet pressure of 0.386 MPa. Total mass flow rate measured at Inlet of coil was 27 g/s. Outlet temperature was 6.032 K and pressure was 0.338 MPa. During the current charging shot, coil quenched during ramp up phase when current was about 8960 A. When voltage tap signals were analyzed it was observed that DP6 section of the coil has quenched. The voltage tap signals of all pancakes along with the coil current are shown in Fig. 3a. Changes in flow, pressure and temperature signals are shown in Fig. 3b–d, respectively. 7. Normal zone propagation speed estimation As the quench was generated in the coil during the rampup phase, measured voltage across DP6 has inductive component in addition to the normal zone voltage. This voltage will be
118
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
Fig. 2. TF coil test sensors and hydraulic layout.
proportional to self and mutual inductances of the double pancakes and the current ramp rate of the coil. Inductive voltage component across each DP is about 94 mV. The total voltage developed across DP6 is 556 mV. Correcting it for inductive voltage, actual normal zone voltage signal is calculated to be about 184 mV (as quench detection card has a gain of 2). This quench voltage has developed in time of 1.17 s. As the coil current ramp rate is only 20 A/s, the operating current of 8960 A is considered constant during the quench development time. SST-1 CICC resistance was calculated using copper resistivity of 0.23 n m, at temperature of about 15 K (RRR 100, magnetic field 2.2 T). Under these conditions, SST-1 CICC resistance is calculated to be 3.4 /m. Using the normal zone voltage and the
(a)
(b)
0.2 0.0 -0.2 -0.4 -0.6 Quench Trigger
672.8
673.2
673.6
Mian Inlet Flow (g/s)
0.4
2000
40
6
4 20 2 Quench Trigger
-0.8 0 674.0
674.4
0 672
679
6
(d)
PS Current Inlet Pressure DP6 Outlet Pressure
8000
Main Inlet Busbar Joint 2 Nr DP6 flow meter
24
20 5
4000 4 2000
Pressure (bar)
6000
Temperature (K)
(c)
686
Relative time (s)
Relative time (s)
Coil Current (A)
8
0.6
4000
672.4
Main Inlet Flow DP6 Outlet flow
16
Quench Trigger
12
8 Quench Trigger
0
3 670
672
674
676
678
680
682
684
686
688
690
692
4 680
Relative time (s)
720
Relative time (s) Fig. 3. Quench development in TF15 coil.
760
800
DP6 Outlet flow (g/s)
Coil Current (A)
A rather unexpected result of increase in inlet mass flow rate has been observed in TF 15 quench as shown in Fig. 3b. This is a new observation, as in general flow reduction following quench have been reported or the results of simulations have predicted it [14–17]. Following the quench, initially inlet flow decreased slightly, and then it increased substantially.
DP VOltage (V)
6000
8. Inlet flow increase following the quench
0.8
Current DP1 DP2 DP3 DP4 DP5 DP6
8000
CICC resistance, it is found that about 5.93 m length of the CICC has quenched in time of ∼1.17 s. This gives a normal zone propagation speed of ∼5.07 m/s.
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
rate to the CICC. ‘Jun 3’ and ‘Jun 4’ are check valves, which after a threshold pressure; open the connections to ‘Vol 3’, a large adiabatic volume representing a relief reservoir. ‘Vol 1’ and ‘Vol 2’ are small volumes representing inlet and outlet manifolds.
10000
DelP Current
8000
6000 1.2 4000
Current (A)
Differential Pressure (bar)
1.8
2000 0.6 0
19:10:08
19:10:18
19:10:28
119
19:10:38
time Fig. 4. Pressure drop across the coil following TF 15 coil quench.
We tried to find the reason for this. When data from the cryogenic plant was analyzed it was found that during quench propagation period, difference between inlet and outlet pressure of the coil initially decreased slightly, and then it increased substantially. Probably due to quench which occurred near outlet, as explained in later sections, the outlet pressure started increasing. In order to maintain the pressure difference across the load, plant also started increasing the inlet pressure. This response of the plant along with the thermal hydraulic effects of quench led to pressure fluctuations across the coil. As the pressure difference is closely linked to the mass flow rate in the coil, increased pressure drop as plotted in Fig. 4, led to increase in mass flow rate of coil initially and then it returned to its normal value after about 20 s. This behavior was observed in few other coils as well during the coil test. A database has been generated of the level of these fluctuations and the time taken by these fluctuations to reach the inlet/outlet sensors. It will be used to generate the threshold limits for secondary quench detection and will be reported in detail at a later stage. 9. Quench simulation TF 15 quench was simulated using code GANDALF 2.1, which is a one-dimensional finite elements code for simulation of the quench and stability of CICC based magnets. Gandalf code’s input parameters, which include CICC specifications and the operating conditions, were generated from the details provided earlier in this paper. This code has option of modifying its subroutines to take care of different CICC designs [18]. Coefficients of Gandalf subroutine exts.f which implements the relation between critical current, temperature and critical field for NbTi were adjusted to obtain the critical current density (Jc) of 2853 A/mm2 at 5 T, 4.2 K. Gandalf implemented Jc values are plotted in Fig. 5a along with the experimental measured Jc values [19] which show a good matching. Other subroutines of this code like exth.f which implements the heat transfer coefficients, extf.f which implements friction factor were also modified to adapt the code to use properties relevant for SST-1 CICC. Heat transfer coefficients and friction factor correlations were implemented as reported in model coil simulation report [20]. Magnetic field ‘B’ variation along the DP6 length was calculated using M’C code [21]. This spatial variation of magnetic field was implemented using ‘extb.f’ subroutine of Gandalf. Gandalf implemented ‘B’ is plotted along the DP6 length in Fig. 5b. A simplified model of cryogenic plant associated with the coil test was implemented using flower subroutine of the code and is shown in Fig. 5c. In this model ‘Jun 1’ is CICC which is simulated by Gandalf. ‘Jun 2’ is the volumetric pump to provide the required mass flow
10. Simulation results As during the experiment, major temperature rise was observed at the outlet of the coil as compared to inlet so it was expected that quench occurred near to outlet on the outermost turns of DP6. Also CICC of this turn and the innermost turn are in contact on 2 sides with the coil casings. Rests of the turns of last DP have only one side in contact with the coil casing. Since exact location of quench initiation was not known, a parametric study was done by initiating quench in different locations of last turn of DP6. As the quench was initiated from the heat load coming from coil casing, in Gandalf heat load was distributed over long lengths of about 1 m. The amount of heat load was taken as just the minimum required to initiate a non-recoverable normal zone. After the parametric study by varying amount of heat load, location of heat load and length of heated zone a good matching was found with the experimental results when quench was initiated by heating 1 m length between 43 m and 44 m distance from inlet. Linear heat flux of 1200 W/m with time duration of 50 ms was used in this trial. Other parameters like helium temperature, pressure, mass flow rate and operating current were as per the experimental conditions described earlier. Quench development time during the experiment was about 1.17 s after which the threshold voltage of quench detection circuit was reached and dump was initiated after 100 ms delay time. So in Gandalf also, current was maintained at 8960 A for 1.17 s and after which current is dumped with experimental dump time constant. To compare the experimental plot and Gandalf plots of normal zone voltage development, zero in the time line of Gandalf plot was shifted by 672.9 s. This is the time at which quench was initiated as per the experiment data. The comparison of normal zone voltage is shown in Fig. 6a. The figure shows a good matching till the point marked as ‘quench triggered’. This is the point at which in the experiment, quench dump was initiated. So the rapid rise in experimental quench voltage is due to generation of large dump voltage and it is not the rise of normal zone voltage. In the simulation this dump voltage is missing and current decay initiates smoothly as programmed in Gandalf input file. So Gandalf normal zone voltage shows a reducing trend after this point. The rise in temperature at DP6 inlet and outlet during the experiment and in Gandalf simulation shows similar peak values. This is an important result as it has given the information about probable the hot spot temperature inside the magnet. Since the temperature at the outlet and inlet could be simulated within ±1 K, peak value at hotspot temperature inside is also expected to match within similar error margin. As actual hotspot temperature cannot be measured in big size actual magnets like TF coil of a Tokamak, we can make use of these simulation results to predict it. This is discussed in next section. The observed differences in simulation results with experiment result are also because heating due to ac losses are not included in simulation. 11. Predictive quench simulation for assembled TF magnet system The code modified for TF15 simulation has been used to do the predictive quench simulation for 16 TF magnets assembled in series. The main difference between a stand-alone coil and assemble TF magnet system will be in the background magnetic field. In single coil test, at the operating current of 10,000 A, the expected
120
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
Fig. 5. Implementation of SST-1 CICC and coil test parameters in Gandalf.
peak magnetic field is 2.2 T where as in assembled magnet system it will be 5.1 T at the same operating current. This is due to superposition of magnetic field from neighboring TF coils. Also dump time constant will be 12 s against the 1.7 s used for single coil test. So keeping all other input parameters of Gandalf same, the magnetic field distribution was changed and hot spot temperature was simulated with a dump time constant of 12 s. The maximum hot spot temperature is found to be about 118 K. This is under the design temperature limit of 150 K. So SST-1 TF coil quench detection system design parameters like threshold voltage and delay time are also acceptable.
0.4
Another predictive simulation was done for the hypothetical case of missed quench by fast diagnostic of voltage measurement method. This quench then will be detected by some slow diagnostics like flow, pressure or temperature sensor. As observed from different coil test data, the maximum time delay of about 1.8 s is present before quench induced pressure, flow or temperature is recorded. So a worst case of delay time of 3 s was considered for this study. For this case operating current of 10,000 A, background magnetic field of 5.1 T and dump time constant of 12 s were selected. Quench was generated in the middle of one of the DP. Dump was initiated after a delay time of 3 s in Gandalf subroutine. The
Experimental Gandalf
16
Voltage (V)
0.2
Temperature difference (K)
14
Quench Triggered
Expt Inlet Expt Outlet Gandalf inlet Gandalf outlet
12 10 8 6 4 2 0
0.0 673
674
time
676
680
time
Fig. 6. (a) Normal zone voltage comparison. (b) Inlet and outlet temperature comparison.
684
A.N. Sharma et al. / Fusion Engineering and Design 89 (2014) 115–121
simulated hot spot temperature for this case is found to be 180 K. This is 30 K higher than SST-1 design limit of 150 K but still under the generally considered safe limit of 250 K for fusion grade magnets. 12. Conclusion All SST-1 TF magnets have been individually cold tested at its nominal operating current is two phase or supercritical helium flow conditions successfully. All design parameters of SST-1 CICC as tested in MC test and now as wound TF magnets have been verified. Unconventional results of mass flow increase at inlet following quench was observed in many coil tests. It was found to be linked to the helium plant response to the quench events. Quench observed in TF15 coil was analyzed to estimate the normal zone propagation speed. This quench was simulated using 1-D code, after suitably modifying its subroutines for SST-1 CICC. Simulation results have given detailed information about the hotspot temperature and the normal zone propagation speed. Reasonably good agreement with the experimental data has been found in the main parameters of normal zone propagation speed and the temperature rise. These results were used to make predictive analysis for assembled TF magnet. A case of missed quench by voltage tap signal was also simulated. It was found that even with the delay of 3 s, the hot spot temperature is about 180 K. So TF magnets are safe even for this case when quench is detected by a slow response diagnostic of flow or pressure. Results of these analyses confirm the suitability of the quench detection algorithms of SST-1 TF magnets. References [1] S. Pradhan, SST-1 Mission Team, Status of SST-1 refurbishment, Journal of Fusion Research Series 9 (2010) 650–657. [2] L. Bottura, A numerical model for the simulation of quench in the ITER magnets, Journal of Computational Physics 124 (1) (1996). [3] S. Pradhan, Y.C. Saxena, O.P. Subrat Das, D.P. Anashkin, V.E. Ivanov, Keilin, et al., Superconducting cable-in-conduit-conductor for SST-1 magnets, in: 2nd IAEA Technical Committee Meeting on Steady State Operation of Magnetic Fusion Devices, vol. 2, Fukuoka, Japan, 1999, p. 482. [4] A.N. Sharma, C.J. Hansalia, Y. Yeole, G. Bansal, S. Pradhan, A. Moitra, et al., Quench detection and data acquisition system for SST-1 superconducting magnets, Fusion Engineering and Design 74 (November (1–4)) (2005) 819–823.
121
[5] A.N. Sharma, Y. Yeole, D. Sonara, A. Moitra, S. Pradhan, B. Sarkar, et al., Quench detection and protection system for superconducting magnet of SST-1 tokamak, in: 4th IAEA TCM on Steady State Operation of magnetic Fusion Devices, February 1–5, Ahmedabad, 2005. [6] Y. Khristi, A.N. Sharma, K. Doshi, U. Prasad, P. Varmora, S. Kedia, et al., Quench detection system for TF coil-test campaigns of SST-1, IEEE Transactions on Applied Superconductivity 22 (April (2)) (2012) 4200108. [7] http://www.comsol.com/paper/magneto-structural-analysis-of-fusion-gradesuperconducting-toroidal-field-coils-7362 [8] O. Anashkin, D.P. Ivanov, V.E. Keilin, N.Ph. Kopeikin, I.A. Kovalev, A.V. Krivykh, et al., Results of the model coil tests for the cable-in-conduit conductor for SST-1 tokamak, IEEE Transactions on Applied Superconductivity 12 (March (1)) (2002), pp. 567, 570. [9] S. Pradhan, SST-1 toroidal field magnet tests: some results and lessons learnt, IEEE Transactions on Applied Superconductivity 22 (June (3)) (2012), p. 9501804. [10] S. Kedia, S. Pradhan, Y. Khristi, U. Prasad, K. Doshi, A.N. Sharma, Measurement of electromagnetic and thermal stresses on conduction-cooled joints of the SST-1 spare TF coil, IEEE Transactions on Applied Superconductivity 20 (December (6)) (2010). [11] A.N. Dipak Patel, U. Sharma, P. Prasad, Y. Varmora, K. Khristi, Doshi, et al., Thermal anchoring of wires in large scale superconducting coil test experiment, Fusion Engineering and Design 88 (Ju2ne (5)) (2013) 374–379. [12] K. Doshi, Y. Khristi, A.N. Sharma, P. Varmora, S. Kedia, U. Prasad, et al., Precision signal conditioning electronics for cryogenic temperature and magnetic field measurements in SST-1, IEEE Transactions on Plasma Science 40 (2012) 641–645. [13] P. Varmora, A.N. Sharma, U. Prasad, D. Patel, K. Doshi, Y. Khristi, et al., PXI based data acquisition system for SST-1 TF test program, Indian Journal of Cryogenics 38 (January) (2013) 104– 109. [14] M. Sugimoto, T. Isono, K. Yoshida, H. Tsuji, I. Takano, T. Hamajima, et al., An evaluation of the inlet flow reduction for a cable-in-conduit conductor in pulsed operation, Cryogenics 38 (October (10)) (1998) 989–994. [15] M. Sugimoto, T. Kato, K. Kawano, T. Hiyama, Y. Kamiyauchi, H. Ishida, et al., Test results of the DPC-TJ: thermal and hydraulic performance, Cryogenics 33 (6) (1993) 597–602. [16] K. Yoshida, H. Takigami, H. Kubo, Analytical studies on the hotspot temperature of cable-in-conduit conductors, Cryogenics 41 (August (8)) (2001) 583–594. [17] S. Nicollet, B. Lacroix, D. Bessette, R. Copetti, J.L. Duchateau, M.C. Gouachet, et al., Thermal-hydraulic behaviour of the ITER TF system during a quench development, Fusion Engineering and Design 86 (October (6–8)) (2011) 1497–1500. [18] User Manual, Gandalf version 2.1. [19] Performance of Cable-in-Conduit-Conductors for SST-1 at IPR, HITACHI CABLE REVIEW No. 20 (August 2001). [20] A. Sharma, J.L. Duchateau, Simulation of SST-1 model coil behaviour using Gandalf + Flower, AIM/NTT-2006.036. [21] L. Bottura, Analytical Calculation of Magnetic Field in an Isoparametric Brick, CryoSoft Internal Note CRYO-97-2, 1997.