2017 Volume 58 Issue 4 Pages 679-687
Influences of crack size-distribution and specimen length on the critical current and n-value of heterogeneously cracked superconducting tapes were studied by a Monte Carlo simulation method combined with a model of crack-induced current shunting. In simulation, model specimens, constituted of a series of sections having cracks with different size to each other, were used. It was shown by the present simulation that (i) both of the critical current and n-value decrease with increasing width of crack size distribution, (ii) n-value decreases more sensitively to the increase in width of crack size distribution in comparison with the critical current, and (iii) the features (i) and (ii) stated above are enhanced with increasing specimen length. Also, the experimentally observed feature that the width of distribution of critical current decreases with increasing length in heterogeneously cracked superconducting tapes was realized by the present simulation. This result was in good accordance with the reported feature that local information of critical current values in a specimen is diluted when the voltage tap distance is large.
In fabrication and in operation, RE(Y, Sm, Dy, Gd,....)Ba2Cu3O7−δ coated superconducting tapes (hereafter noted as REBCO tapes), as well as filamentary Bi2Sr2Ca2Cu3O10+x (BSCCO), Nb3Sn, Nb3Al and MgB2 tapes, are subjected to thermal, mechanical and electromagnetic stresses/strains. When the superconducting layers/filaments are cracked by such stresses, the critical current Ic and n-value of both coated1–9) and filamentary10–19) tapes are reduced. Cracking of the coated layer/filament takes place heterogeneously. Accordingly, even at a given applied stress/strain, the Ic- and n-values are different from location to location within a specimen9,14), they are different from specimen to specimen7,9,13,14) and they are dependent on specimen length9,14).
In our recent work20), the extreme case where one crack exists in a REBCO-coated specimen was studied for various specimen lengths by a modeling analysis. The experimentally observed feature that the decrease in n-value with decreasing Ic becomes sharper for longer specimen under local crack-evolution9) was reproduced. Then, we investigated the Ic and n-value of a REBCO-coated model specimen constituted of a series of three sections having cracks with different size to each other21). It was shown that, (a) in both cases of large and small differences in crack size among sections, the section with the largest crack plays a dominant role in determination of the voltage (V) –current (I) curve of the specimen and acts a dominant role to reduce Ic and n-value of the specimen from the original Ic and n-value in the non-cracked state, and (b) under a given size of the largest crack, the Ic becomes higher but the n-value becomes lower with increasing difference in crack size among the sections than the Ic and n-value in the case of no difference in crack size. The features (a) and (b) have been observed experimentally in both REBCO-coated- and BSCCO-filamentary specimens9,14).
As shown above, the influence of difference in crack size among the sections has been studied to some extent but not systematically. It is required to develop a method to obtain systematic information. In the present work, a Monte Carlo simulation method was employed, with which the influences of crack size distribution and specimen length on Ic and n-value were investigated.
The configuration of the model specimen employed in the present work is shown in Fig. 1(a). The specimen is constituted of a series of N sections (S(1), S(2),...S(N)) with a length L0 = 1.5 cm in each. Each section has one crack and the crack size is different among sections. In simulation, N was varied from 1 to 20, namely the length of the specimen L = NL0 was varied from 1.5 to 30 cm, due to the following reasons. (a) It has been known that local information of Ic-values in a specimen is diluted when the voltage tap distance is large22). Preliminary simulation study indicated that this feature can be detected in the specimen length-range of L = 1.5 to 30 cm. (b) For materials characterization, experiments to measure Ic and n-value under applied stress in laboratory are usually conducted in the specimen length range less than several tens of centimeters.
Schematic representation of (a) the model specimen with length L (3~30 cm), composed of a series of sections with a length L0 (=1.5 cm in this work) and (b) current path in a section with a partial crack.
For description of the V(voltage)–I(current) relation of sections under an existent partial crack, where partial crack means the crack that exists in a part of transverse cross-section of the superconducting phase, the model of Fang et al.10) was employed in a modified form3,5,7,9,13,14,19,21). Noting the Ic and n-value of sections in the non-cracked state as Ic0 and n0, respectively, and the critical electric field for determination of critical current as Ec (=1 μV/cm in this work), the V–I curve of a non-cracked section with a length L0 is expressed as
\[V = E_{\rm c} L_0 \left( \frac{I}{I_{\rm c0}} \right)^{n_0}\] | (1) |
\[V = E_{\rm c} L_0 \left( \frac{I}{I_{\rm c0}} \right)^{n_0} + V_{\rm RE}\] | (2) |
\[I = I_{\rm RE} + I_{\rm S} = I_{\rm c0} (1-f) \left( \frac{L_0}{s} \right)^{1/n_0} \left[ \frac{V_{\rm RE}}{E_{\rm c}L_0} \right]^{1/n_0} + \frac{V_{\rm RE}}{R_{\rm t}}\] | (3) |
The ligament parameter ${\rm Lp}=(1-f)(L_0/s)^{1/n_0}$ has a physical meaning as Ic/Ic0 (ratio of the critical current in cracked state (Ic) to the critical current in non-cracked state (Ic0)) in the virtual case where the voltage VRE developed at the crack is equal to the whole voltage V and current shunting is negligible20). In practice, current shunting occurs and also voltage is developed not only at the crack but also at the non-cracked region away from the crack, and hence the Ic/Ic0 is not equal to the ligament parameter. However, it is noted that the ligament parameter still gives a fairly good approximation for Ic/Ic0 in both of REBCO-coated tape3,5,7,9,19,21) and BSCCO filamentary tape13,14).
In the present work, the value of $L_{\rm p}=(1-f)(L_0/s)^{1/n_0}$ was used directly in calculation of eqs. (2) and (3). It has been shown that, in experimental works, the value of $L_{\rm p}=(1-f)(L_0/s)^{1/n_0}$ can be obtained for each cracked specimen by analysis of the measured V–I curve3,5,7,9,13). Also, in the simulation study as in the present work, it can be used as a parameter to obtain Ic and n-value of cracked specimen20,21,23). By the approach that uses the value of $(1-f)(L_0/s)^{1/n_0}$ directly in calculation, the specimen length-dependence of critical current and n-value in the specimens with one crack20), and the competition among cracks with different size in contribution to synthesize the V–I curve of the specimen with plural cracks21) were revealed. Also, the experimental results of the correlation between n-value and Ic could be described through the calculation using this value23). This approach has the following advantages. (i) It can be used even when the values of s and f are unknown. (ii) It is simple and can be a useful tool in practice. However, these advantages, in turn, are disadvantage on the point that the influence of each of s and f on current shunting phenomenon, n-value and Ic is not specified directly. For specification, the values of s and f and their correlation shall be obtained in advance. In the present work, by using $L_{\rm p}=(1-f)(L_0/s)^{1/n_0}$ in calculation, the aim to reveal the influences of crack size-distribution and specimen length on the critical current and n-value of heterogeneously cracked superconducting tapes was achieved satisfactorily, as shown later in Section 3.
2.3 Application of Monte Carlo method to give ligament parameter-value and calculation of V–I curve of sections with a length L0Since wide distribution of ligament parameter $L_{\rm p}(=(1-f)(L_0/s)^{1/n_0})$ corresponds to wide distribution of f (crack size) and since the standard deviation of 1 − f is the same as that of f, the standard deviation of Lp, ΔLp, was used as a monitor of the width of crack size distribution; the larger the ΔLp, the wider the width of crack size distribution. For formulation of distribution of Lp of cracked sections with a length L0, the normal distribution function was used. Noting the average of Lp as Lp,ave, the cumulative probability F(Lp) was expressed by
\[F(L_{\rm P}) = \frac{1}{2} \left\{1 + {\rm erf} \left( \frac{L_{\rm p} - L_{\rm p,ave}}{\sqrt 2 \Delta L_{\rm p}} \right) \right\}\] | (4) |
The V–I curve of each cracked section with a length L0 = 1.5 cm was calculated by substituting the Lp-value, given by the Monte Carlo method stated above, and the values of Rt = 2 μΩ, Ic0 = 200 A and n0 = 40 into eqs. (2) and (3). Here, the values of Rt, Ic0 and n0 were taken from the experimental result7) of a copper-stabilized DyBCO tape test-pieces, whose V–I curves were measured at 77 K in a self-magnetic field for the voltage probe distance L0 = 1.5 cm. The thickness of the superconducting DyBCO layer of the tape was 2.5 µm. The critical current, Ic0, and n-value, n0, in the non-cracked state were taken from each average value that were obtained by applying the same criterion as in this work to the V–I curves measured under no applied stress. The value of Rt (2 μΩ) was taken from an average Rt-value obtained by analysis of the V–I curves of the test-pieces in which the superconducting DyBCO layer was cracked under applied tensile stress.
The simulation procedure based on the Monte Carlo method stated above was repeated for 120 times and, as a result, 120 sets of (V–I curve, Ic-value, n-value) for the sections were obtained for each ΔLp value.
2.4 Calculation of V–I curve, and estimation of critical current and n-value of specimensAs the specimen is constituted of a series electric circuit of sections (Fig. 1(a)), the current of the specimen is the same as that of sections;
\[I = I_{{\rm S}(i)}(i = 1\ {\rm to} \ N)\] | (5) |
\[V = \sum \limits_{i = 1}^N V_{{\rm S}(i)} \] | (6) |
Figure 2 shows six examples of the simulated V–I curves of specimens with a length L = 9 cm and sections with a length L = L0 = 1.5 cm, obtained for the random values generated in the computer R = 0.063, 0.514, 0.667, 0.203, 0.951, 0.641 for the section S1 to S6, respectively. To obtain Lp-values of sections, these random values were used commonly for ΔLp = (a) 0.01, (b) 0.025, (c) 0.05, (d) 0.1, (e) 0.15 and (f) 0.2. The obtained Lp-values of the sections in the examples in Fig. 2 had the following features. (i) The average values were 0.675~0.680 in all examples, (ii) the values of the standard deviation ΔLp,Ex of the obtained ligament parameter values for the examples shown in Fig. 2(a) to (f) were 0.011, 0.028, 0.056, 0.111, 0.165 and 0.215, respectively, and accordingly, (iii) the examples (a) to (f) had almost the same average crack size but the width of crack size distribution was different from example to example; it was small in the example (a) and it became larger in the order of the examples (a) to (f).
Examples of the V–I curves of the specimens (L = 9 cm) and sections (L = L0 = 1.5 cm), obtained for the computer-generated random values R = 0.063, 0.514, 0.667, 0203, 0.951 and 0.641 for the sections S1 to S6, respectively. With these random values, the average ligament parameter of sections was almost common in all examples but the standard deviation of the ligament parameter of sections ΔLp,Ex, was varied from 0.011 (a) to 0.215 (f).
From the V–I curves in Fig. 2, the Ic- and n-values of sections and specimens were obtained with the criteria stated in Subsection 2.4. The average critical current, Ic,ave, and average n-value, nave, of the sections, and Ic and n-value of the specimen are shown for each example in Fig. 2, for comparison.
Figure 3(a) shows the obtained Ic- and Ic,ave-values of the sections and Ic-values of the specimens for the 6 examples, plotted against ΔLp,Ex–value that refers the difference in crack size among the sections. The changes in Ic,ave of the sections and Ic of specimen with increasing ΔLp,Ex are shown with the broken and solid curves, respectively. Figure 3(b) shows the n-values of the sections and specimens, and nave-values of the sections, plotted against ΔLp,Ex–value, together with the changes in nave-value of the sections (broken curve) and n-value of specimen (solid curve) with increasing ΔLp,Ex.
Critical current Ic and n-value of sections and specimens obtained from the V–I curves of the examples (a) to (f) in Fig. 2, plotted against the standard deviation of ligament parameter ΔLp,Ex of sections in the examples. (a) Change in Ic-values of the sections and specimens, and the average critical current (Ic,ave) of sections with ΔLp,Ex. (b) Change in n-value of sections and specimens, and the average n-value (nave) of sections with ΔLp,Ex. (c) Comparison of the change in Ic/Ic(ΔLp = 0) and n/n(ΔLp = 0) with ΔLp,Ex, where Ic/Ic(ΔLp = 0) and n/n(ΔLp = 0) refer to the retention of critical current Ic and n-value of the specimen in comparison with the critical current, Ic(ΔLp = 0), and n-value, n(ΔLp = 0) in the case of ΔLp = 0 where the ligament parameter (crack size) is the same in all sections.
If all sections have a same ligament parameter (same crack size), namely if ΔLp = 0, all sections and the specimen have a same Ic and a same n-value. When the Ic and n-value of specimen in a case of ΔLp = 0 are noted as Ic(ΔLp = 0) and n(ΔLp = 0), respectively, the values of Ic/Ic(ΔLp = 0) and n/n(ΔLp = 0) show the retention of critical current and n-value in comparison with those in the case of ΔLp = 0, respectively. Figure 3(c) shows the change in Ic/Ic(ΔLp = 0) (solid curve) and n/n(ΔLp = 0) (broken curve) with increasing ΔLp,Ex.
The following features are read from the results in Figs. 2 and 3.
Two examples, showing the influence of crack size distribution of sections on the n-value of the specimen, are presented in Fig. 4. These examples were taken up from many simulation results as to show that n-value is not uniquely determined by the critical current value; namely, plural n-values can exists for one critical current value in heterogeneously cracked specimens. In the example shown in Fig. 4(a), the V–I curves of sections exist rather densely, corresponding to the case where the difference in crack size among sections is relatively small. On the other hand, in the example in (b), the V–I curves of sections exist rather coarsely, corresponding the case where the difference in crack size among sections is large. The Ic values of the specimen in (a) and (b) were the same (122 A) but n-values in (a) and (b) were different; 18.6 and 13.6, respectively. In this way, even when the specimens have a same Ic-value, plural n-values can exist, depending on the difference in crack size among the sections. This result shows that Ic and n-value are not in one to one relation under the occurrence of heterogeneous cracking of superconducting layer.
Examples showing that n-value of the specimen is different (n = 18.6 in the example in (a) but 13.6 in the example in (b)), while the critical current is the same (122 A in both examples). In comparison with the example in (b), the example in (a) is characterized by the smaller distance of the V–I curves of sections (namely smaller difference in crack size among sections), and hence characterized by more number of sections that contribute to synthesize the voltage of the specimen, acing to raise n-value of the specimen.
Figure 5 shows an example of the plot of n-value against Ic-value. In this example, the results of the 18 specimens with a length L = 9 cm simulated for ΔLp = 0.01, 0.025, 0.05, 0.1, 0.15 and 0.2 are presented. The (Ic, n) value for each specimen is shown with open symbols. The closed symbols show the average values (Ic,ave, nave) of 18 specimens, for each value of ΔLp. The tendency that the decrease in n-value with decreasing Ic is clearly found. However, as indicated above, different n-values (Ic-values) can exist for a given Ic-value (n-value). The results in Fig. 5 clearly show that, as the extent of difference in crack size among the sections is different from specimen to specimen, the relation of n-value to Ic in specimens is not uniquely determined. For description of the experimentally measured correlation between Ic and n-value of cracked superconducting tape, an approach to surround the n vs Ic plot by the upper bound and the lower bound23) may be useful.
Plot of n-value against critical current value of the specimens with a length L = 9 cm simulated for ΔLp = 0.01 to 0.2. The (Ic, n) value for each specimen is shown with open symbol. The closed symbols show the average values of 18 specimens for each value of ΔLp.
Figures 6 and 7 show the obtained Ic- and n-values, plotted against the standard deviation of ligament parameter, ΔLp, for representative specimen length L = (a) 1.5 cm, (b) 3 cm, (c) 6 cm, (d) 9 cm and (e) 15 cm. The results for L = 1.5 cm were taken from the results for the sections with L = L0 = 1.5 cm. Figures 8 and 9 show the average critical current, Ic,ave and average n-value, nave, plotted against ΔLp and L, respectively. The following features are read from Figs. 6 to 9.
Simulated critical current (Ic)-values for specimen length L = (a) 1.5 cm, (b) 3.0 cm, (c) 6 cm, (d) 9 cm and (e) 15 cm, plotted against the standard deviation of the ligament parameter ΔLp.
Simulated n-values for specimen length L = (a) 1.5 cm, (b) 3.0 cm, (c) 6 cm, (d) 9 cm and (e) 15 cm, plotted against the standard deviation of the ligament parameter ΔLp.
Change in (a) average critical current, Ic,ave and (b) average n-value, nave, with increasing standard deviation of ligament parameter ΔLp for L = 1.5~30 cm.
Change in (a) average critical current, Ic,ave and (b) average n-value, nave, with increasing specimen length L for standard deviation of ligament parameter ΔLp = 0.01 to 0.2.
As has been shown in Fig. 5, the relation of n-value to Ic of each specimen was complex, depending on the extent of difference in crack size among sections. However, despite the complexity, the tendency that the n-value decreases with decreasing Ic was clearly detected by plotting the average n-value, nave, against average Ic, Ic,ave. In the example in Fig. 5, the specimen length was limited to 9 cm. As a next step, to reveal the influences of specimen length on the change in nave with Ic,ave, the values of (Ic,ave, nave) were read from the Ic,ave-L and nave-L relations in Fig. 9, and nave was plotted against Ic,ave for L = 3~30 cm, as shown in Fig. 10.
Influences of specimen length L on the change in average n-value, nave, with decreasing average critical current, Ic,ave in the range of ΔLp = 0~0.2 and L = 3~30 cm.
When ΔLp = 0, values of (Ic,ave, nave) do not change with L. Both Ic,ave and nave decrease with increase in ΔLp (Figs. 8 and 10) and this phenomenon is enhanced in longer specimens (Fig. 9). It is important that the level of nave of nave–Ic,ave relation becomes lower with increasing L (Fig. 10), reflecting the feature that n-value is reduced more than the critical current by cracks (Fig. 3(c)). It has been reported that, when the values of (Ic, n) of the coated- and filamentary-tape with stress-induced cracks were measured for short and long specimens, the level of nave of the nave-Ic,ave correlation diagram of longer specimen was lower than that of shorter specimen9,14). This experimentally observed feature of the length dependence of the nave-Ic,ave relation is reproduced by the present simulation method, as in Fig. 10, from a viewpoint of heterogeneous cracking.
3.5 Variation of critical current value with position along the length, affected by voltage tap distanceIt has been shown experimentally that the variations in the local critical current (Ic) along the specimen length depend strongly on the voltage tap distance; larger tap distance can give the appearance of higher uniformity for Ic than is actually present22). This phenomenon was reproduced well by the present approach as follows. The section- and specimen- length L (1.5 to 30 cm) used in the simulation study was regarded as the voltage tap distance. Taking the total length of a specimen as 180 cm, and arranging the simulation results of Ic-values to position order along the specimen length, we had Fig. 11, in which the variation of Ic with position along the length for voltage tap distance L = (a) 1.5, (b) 3, (c) 6, (d) 15 and (e) 30 cm under a condition of ΔLp = 0.1 were representatively presented. The Ic,ave and ΔIc for each L refer to the average and standard deviation of Ic-values, respectively. In the case where the Ic values were estimated in a step of 1.5 cm, the ΔIc was high as 21.0 A. When the Ic values were estimated in step of 3, 6, 15 and 30 cm, ΔIc decreased to 17.2, 16.0, 10.3 and 5.73 A, respectively. The reduction in ΔIc with increasing voltage tap distance observed in the present work reproduces well the reported important warning that local information of critical current values in a specimen is diluted when the tap distance is large22). Average and standard deviation of Ic values tend to decrease with increasing L under the distributed crack size, due to the higher probability to contain larger crack in longer specimens (larger voltage tap distances).
Variation of critical current Ic with position in 180 cm-specimen, obtained for the voltage tap distance of L = (a) 1.5 cm, (b) 3 cm, (c) 6 cm, (d) 15 cm and (e) 30 cm, simulated under the condition of Lp,ave = 0.67 and ΔLp = 0.1. The Ic,ave and ΔIc refer to the average and standard deviation of Ic-values. The lower bound of Ic for each L, corresponding to the lowest Ic among 1.5 cm-sections contained in the voltage tap distance L, is shown with rectangles, for reference.
As has been shown in Fig. 2 and Fig. 3(a), the largest crack-section (lowest Ic section) plays a dominant role in determination of Ic of specimens. If all sections have the same crack size as the largest one, the V of the V–I curve of specimen increases most rapidly. Accordingly, the lower bound of the critical current of the specimen is given by this situation where the Ic value of the specimen is the same as that of the lowest Ic section. Thus the lower bound of Ic of specimen is given by the lowest Ic value among the sections. In the cases of 3, 6, 15 and 30 cm- voltage tap distances, the lower bounds of Ic are given by the lowest Ic value among 2, 4, 10 and 20 sections existing between the taps. The rectangle shows the lower bound. The shape of the simulated variation of Ic-value with position is well described qualitatively, confirming that the lowest Ic section with the largest crack plays a significant role in determination of Ic of the specimen as an assemble of sections.
Influences of crack size-distribution and specimen length on critical current and n-value of heterogeneously cracked superconducting tape were studied by means of a Monte Carlo simulation in combination with a model of crack-induced current shunting. In simulation, model specimen that is constituted of a series of sections having different size-crack to each other was used. The crack size-distribution was monitored by the ligament parameter-distribution. The standard deviation of ligament parameter was varied from 0.01, corresponding to very narrow crack size-distribution, to 0.2, corresponding to very wide one. Specimen length was varied in the range of 1.5~30 cm. Main results are summarized as follows.