Fatigue properties of a commercial epoxy have been studied at RT, LNT and LHeT. The low temperature reactor irradiation effects have also been studied at LHeT. Repeated flexural stress with frequency of 2Hz was applied to the samples cured into cylindrical shapes of 4.0mm diameter and 30mm length. The samples were in contact with the coolant of liquid helium or liquid nitrogen during the fatigue experiments. Reactor irradiation was performed in Low Temperature Loop of Kyoto University Research Reactor with the fast neutron fluence of about 2.5×1016n·cm-2 and γ doses of 2.8×108rad. Normally the design parameters at cryogenic temperatures are apt to be based on the data at RT because the temperature decrease brings higher mechanical strength. However the fatigue strength of this epoxy at LHeT was about a half of that at LNT and was as low as that at RT data. For this reason the fatigue data at LHeT where superconducting magnets are used should be accumulated exactly. On this experimental result, a similar tendency of the fatigue properties between at LNT and at LHeT was observed in the stress-number of cycles curves represented by normalized stress on ordinate against fatigue life on abscissa with logarismic scale. It suggests the possibility of the substitution with fatigue data at LNT for that at LHeT. Low temperature irradiation brought on the epoxy used in this study the decrease of fatigue strength and the increase of scattering of fatigue strength. These results indicate that careful design should be performed in practical use of these organic materials to fusion reactor.
The dynamics of a superconducting inductor and a thyristor converter (I-C) unit have been studied. For the optimal control of I-C unit, the dynamics of the unit must be described. But difficulties arise in the deduction of dynamics of a superconducting magnet (SM) from the properties of conductors and the coil configurations. To avoid these difficulties, the identification of dynamics through input/output relations has been studied. For this study, the input/output data are acquired through the following procedures. The experimental system is the I-C unit composed of a 12 phase thyristor power converter and a 100kJ superconducting magnet. The input gate-signal of random waveform is given to the thyristor coverter and the response waveforms of currents and voltages of the I-C unit are sampled and memorized to form an input/output (I/O) sequence. The I/O sequence is identified by a multi-input/multi-output discrete time system (MI/MO DTS), and the dynamics of the I-C unit is described. The discrete time system which relates the input and output waveforms of the I-C unit is derived by being based on least square method such that the euclidean norm between the measured and simulated outputs is minimized. The simulated outputs of the derived MI/MO DTS have closely fitted to the measured waveforms. The derived systems revealed that one of the equivalent circuits of SM was a simple connection of R-L and the parameters were estimated so that the simulated waveforms of the equivalent circuit were closely fitted to the measured ones.
DDC (direct digital control) of superconducting magnet current by using the microcomputer has been studied. An inductor-converter (I-C) unit composed of a 6-pulse thyristor converter of Graetz bridge type and a superconducting magnet (SM) has been developed. A microcomputer based controller and its input/output interfaces are connected to the I-C unit and its current is feedback controlled. The closed loop system is formed via the following steps: detection of the SC current by using DCCT (direct current current transformer), A/D conversion of the SM current and input to the microcomputer, comparison with the reference input for generating the error signal, computation for compensation by PID, Dahlin or more sophisticated optimal control such as FTSC (finite time settling control), via FTSO (finite time settling observer) and generation of trigger pulses. These operations are performed by the program instructions of the microcomputer. The specifications of the developed system are: Inductor 3kJ, 93.5A at 5T, 0.585H, Converter 6 pulse Graetz bridge, 100V, 400A; Microcomputer TLCS 12A, 12 bits, RAM 2kW, PROM 0.5kW, 1.2MHz. By using this system, experimental studies have been performed. The control algorithms for digital control have been tested by experimental works as well as by computer simulations. The control methods PID and Dahlin have been implemented by the microcomputer and the accurate agreements between the results of the experiments and those of simulations have been found. These algorithms are simple and can be implemented by the fixed point operations. The computation time is less than 1/360sec. The optimal parameters can be obtained by the frequency domain method and by the simulations. The moderate oscillations have been found by PID compensation although the quite stable responses have been achieved by Dahlin algorithm. For the feedback control by FTSC via FTSO the floating operations are required, although the algorithm is quite simple. The simulated result reveals the closely fitted response to the reference waveform. If the control algorithm is implemented by using DEC LSI-11, the computation time is estimated 1/60sec or so.
SUS 304 and 316 austenitic stainless steel plates were welded by filler wires of Y308 and Y316 using TIG method, and the strength and the toughness of the weldments were evaluated at cryogenic temperatures (77K and 4.2K). The change in magnetic properties due to the strain-induced martensite (α′) were also examined. The yield strength of 304 (parent metals and weld metals) are greater than those of 316 at room temperature, but at 77K and 4.2K, the yield strength of 316 overcomes those of 304. The elongation of both weld metals are smaller than those of their parent metals at room temperature, but become equal to those of their parent metals at 77K and 4.2K. The tensile tests of the notched specimens of weld metals show that the notch yield strength ratios of 304/Y308 and 316/Y316 weld metals keep 2.4 and 1.9 at 4.2K respectively, and both fracture modes correspond to the ductile manner, although the notch sensitivities decrease with decreasing temperature. The relations of the volume pct of γ, ε′ and α′ to the amount of strain at 77K and 4.2K were obtained. The amount of martensite (α′) formed in the fractured specimens at 4.2K were 95% for 304/Y308 weld metal and 70% for 316/Y316 weld metal. An austenitic phase of 316/Y316 weld metal is more stable than that of 304/Y308 weld metal. It was found that the weld metals showed similar mechanical and magnetic properties to their parent metals at 77K and 4.2K because of the dominant effect of strain-induced martensitic transformation on those properties.