curve fitting of stress-strain curve and rrt of ag/bi-2212 round wires
DESCRIPTION
Curve fitting of stress-strain curve and RRT of Ag/Bi-2212 round wires. Presented by H.W. Weijers Measurements performed by Bob Walsh and Dustin McRea Presented at Andong National University Andong, July 15 th 2009. Outline. Overview of recent testing Curve fitting RRT on Bi-2212 - PowerPoint PPT PresentationTRANSCRIPT
Curve fitting of stress-strain curve and RRT of Ag/Bi-2212 round wires
Presented by H.W. WeijersMeasurements performed by Bob Walsh and Dustin McRea
Presented at Andong National UniversityAndong, July 15th 2009
Outline• Overview of recent testing• Curve fitting• RRT on Bi-2212• Conclusions
Recent testing
• Student ran measurements with support• Hydraulic rig with calibrated load cell• Used both Shepic (19 g, not balanced) and
Nyilas type (4.2 g) extensometer• 90 mm between grips• Drill chucks for round wire• Plate clamps for tape• Grips:128 g (zeroed out), pins: 76.5g each
(lower pin not zeroed out)
Recent testinggrips and extensometers
Nyilas extensometer close up
Load cell
Bi-2223 3-ply Brass“raw” data
1,2 Shepic3,3 Nyilas
Bi-2223 3-ply stainless steel“raw” data
1,2 Shepic3,3 Nyilas
Bi2212 round wire“raw” data
1,2 Shepic3,4 NyilasReturn line slopes more useful than initial slope
Curve fitting of initial slope
• Purposes– Provide fit of data
• Section Annex 3
– Method to determine initial slope E0
• Section Annex 10
• Vary range to gain insight in quality of fit
Linear and 2nd order poly fit comparison
to 20 MPa to 40 MPa to 60 MPa to 80 MPa to 100 MPaMaterial linear polynomial linear polynomial linear polynomial linear polynomial linear polynomial
E
GPa R2E
GPa R2E
GPa R2E
GPa R2E
GPa R2E
GPa R2E
GPa RE
GPa RE
GPa RE
GPa R
3U19-1 105 0.9915 93 0.9927 100 0.9983 111 0.9990 89 0.9949 114 0.9997 82 0.9939 106 0.9996 3U19-2 103 0.9971 93 0.9978 100 0.9990 105 0.9992 98 0.9994 105 0.9998 96 0.9994 105 0.9999 93 0.9986 106 0.9999
3U19-3 95 0.9850 98 0.9850 96 0.9962 93 0.9963 96 0.9987 96 0.9987 94 0.9991 100 0.9994 93 0.9991 101 0.99973U19-4 104 0.9926 91 0.9940 101 0.9433 106 0.9984 98 0.9990 106 0.9994 95 0.9987 107 0.9998 92 0.9983 106 0.9999
3B15-1 93 0.9964 91 0.9964 94 0.9995 93 0.9995 94 0.9998 96 0.9998 92 0.9996 98 0.9999 91 0.9996 97 0.99993B15-2 90 0.9976 94 0.9980 90 0.9993 89 0.9993 91 0.9997 88 0.9997 91 0.9998 89 0.9999 91 0.9999 91 0.99993B15-3 96 0.9927 87 0.9939 100 0.9983 93 0.9987 99 0.9992 101 0.9993 97 0.9994 103 0.9996 95 0.9992 104 0.99983B15-4 152 0.9711 140 0.9716 152 0.9974 106 0.9975 145 0.9982 110 0.9993 140 0.9985 107 0.9996 92 0.9986 104 0.9997
Bi 2212-1 63 0.9943 84 0.9990 56 0.9939 69 0.9988 50 0.9917 65 0.9995 44 0.9870 62 0.9996 Bi 2212-2 49 0.9990 45 0.9994 51 0.9990 47 0.9997 51 0.9996 50 0.9996 48 0.9967 56 0.9989 Bi 2212-3 79 0.9978 88 0.9987 68 0.9937 83 0.9986 59 0.9876 80 0.9994 49 0.9788 74 0.9993 Bi 2212-4* 87 1.0000 87 1.0000 83 0.9998 82 0.9988 78 0.9974 91 0.9995
Poly fit has almost always higher R2 value, otherwise equal
Fitting of initial curvewith increasing stress range
• Linear fit: No convergence of slope and R2
• 2nd order poly: Convergence, ~ same Eo as linear fits extrapolated to zero
Curve fitting of initial slope• 2nd order polynomial fit data better than 1st order
(linear) fit– Not surprising, but both fit with R2 > 0.99– Clearly higher R2 values; linear fit trends down with
increasing range for 2nd order fit doesn’t.– Comparable scatter in E0?
• Propose ?– For reinforced conductor, linear and 2nd order poly fits
could be used, but “range” and convergence criteria need to de defined
Curve fitting over larger rangeR2 values fit to 0.3%
power poly
3U19-1
3U19-2 0.9992 0.9999
3U19-3 0.9954 0.9999
3U19-4 0.9972 0.9999
3B15-1 0.9957 1
3B15-2 0.9936 0.9999
3B15-3 0.9964 1
3B15-4 0.9941 0.9999
range 0.9936-0.9992 0.9999-1
Bi 2212-1 0.9765 0.9989
Bi 2212-2 0.9131 0.9916
Bi 2212-3 0.9933 0.9974
Bi 2212-4 0.9711 0.9862
Data for increasing strain only (return lines removed)
Curve fitting of data 0 to 0.3%• 2nd order polynomial fit data better than power fit
– Consistently higher R2
– Increasing lower bound of range above zero for power fit (as per Standard (A-10) does not necessarily improve either fit
– Trendline power fit in Excel sometimes fails to find proper fit: solver works better
• Propose for Standard– “Range” and convergence criteria need to de defined
• Upper bound of range of fit of 0.5% (A-10) is too high• 0.3% is more reasonable, or fraction (80-90%?) of strain where
curve “kinks” (Relasticmax, elasticmax)
– Skip “a(-b)n” from Standard and replace with poly
Conclusions• Data with Nyilas and Shepic extensometers very
comparable• Unreinforced wire sensitive to handling and/or sample-
sample variation– Chucks are suitable, require careful handling– Value of round-robin TBD
• 2nd order polynomial fit of initial slope– Fits better than linear or a(-b)n
– Not necessarily better predictor of Eo compared to linear– Discussion needed to define criterion,
• 2nd order polynomial fit of slope to 0.3%– Fits better than a(-b)n
– Clearly a better choice– 0 to 0.5% is too large a range for BSCCO
• Some proposals made to adapt Standard
Miscellaneous proposals
• Test report section 10.2, Optional results– When reporting % elongation to failure, add
location of failure (at grips, within extensometer) to report