Click for next page ( 26


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 25
25 Test Methods and Precision Testing The validity of thresholds developed in this way is also de- pendent on the validity of the physical test methods (such as The recommended QC methods have been written in pull-out tests) used as the basis for measuring bond perfor- AASHTO/ASTM standard method format in Appendix C, mance. At the direction of the NCHRP supervisory panel, the where they are titled: transfer length testing originally planned for this test program 1. Test Method for the Determination of the Surface Tension as a basis for developing thresholds for the surface and chem- of Steel Strand by Contact Angle Measurement, ical test results was not conducted. Instead, the thresholds 2. Test Method for Weight Loss on Ignition (LOI) of Steel for the chemical and surface test methods were based on the Strand, acceptance limits for the mortar pull-out tests proposed by 3. Test Method for Change in Corrosion Potential of Steel Russell and adopted by NASPA. Strand, and The bond strength thresholds proposed by Russell are stated 4. Test Method for Identification and Quantification of in terms of the force at 0.1-in. slip measured by the NASPA Residue on Steel Strand by Extraction, Gravimetric, and mortar pull-out test procedure. They are based on a set of Spectroscopical Analyses. development length tests conducted in parallel with the development of the NASPA strand bond test (Russell 2001, Testing was conducted to provide the basis for a precision Russell 2006). The thresholds were derived using develop- statement accompanying the proposed test methods devel- ment length tests on four strand sources, (in what is referred oped for identifying strand bond performance. to as the NASPA Round III study [Russell 2001]), and they are The results for the four recommended test methods are defined in terms of acceptance criteria for the average force at given in Table 6. Note that each repeat test included testing of 0.1-in. slip from six pull-outs with a lower criterion for any three strand sections. The precision and bias statements to be single measurement of the six pull-outs. The Round III report added to the standard test methods take the form illustrated proposed thresholds of 7,300 and 5,500 lbs, for the minimum in the following example: permissible average and single test result, respectively for 1/2-in. diameter strand (Russell 2001). These minimum Test Method for Weight Loss on Ignition (LOI) of Strand thresholds have since been increased to 10,500 and 9,000 lbs, Single-Operator Precision--The single-operator standard but without additional testing (Russell 2006). For 0.6-in. deviation was found to be 0.014 mg/cm2. Therefore, results of diameter strand, the suggested thresholds are 12,600 and two properly conducted tests by the same operator on the same 10,800 lbs for the minimum permissible average and single source are not expected to differ from each other by more than test result, respectively (Russell 2006). No threshold has been 0.041 mg/cm2. (These numbers represent, respectively, the (1s) suggested for other sizes of strand. and (d2s) limits as described in ASTM C670 [ASTM 2003].) Despite the somewhat limited scope of the development Bias--Since there is no accepted reference material suitable for determining the bias in this test method, no statement on bias process used to establish these NASPA test thresholds, the is made. threshold determination effort for the surface and chemical testing conducted in this study was performed assuming that these thresholds were well-defined lower bounds for good Development of Thresholds bonding behavior. As has been done throughout this study, For the recommended surface and chemical test methods the thresholds were converted to bond stresses calculated to be useful in a QC setting, thresholds for acceptable bond as the force divided by the nominal surface area to support behavior are needed. The usefulness of acceptance/rejection comparisons among all of the tested strands. When converted thresholds for the surface and chemical test results is de- to a bond stress, the minimum threshold on the average of pendent on the correlation of these results with minimum six tests of 10,500 lbs is equal to 0.313 ksi. This value was used acceptable bond strengths established by physical test methods. as the basis for the threshold analysis. Table 6. Precision test results for recommended QC tests. Organic Residue Extraction Weight Loss on Ignition Average Contact Angle after Change in Corrosion Potential Test Method (mg/cm2) (LOI) (mg/cm2) Lime Dip () (V) after 6 h Number of Repeats 6 6 6 5 Average of Results 0.069 0.091 73 -0.334 Standard Deviation of Results 0.013 0.014 4 0.047

OCR for page 25
26 Table 7. Regression coefficients for single-predictor models. Threshold Corresponding Coefficient of Constant Coefficient to Mortar Predictor Determination (x0) ( ) Pull-Out (R2) Stress of 0.313 ksi Weight Loss on Ignition (mg/cm2) 0.445 -1.403 0.16 Not Possible Contact Angle after Lime Dip () 1.393 -0.012 0.57 73 Change in Corrosion Potential after 6 h (V)--as Received 0.766 1.741 0.68 -0.175 Extracted Organic Residue (mg/cm2) 0.453 -1.752 0.12 Not Possible Extracted Organic Residue (mg/cm2)--Stearate Only 0.436 -1.943 0.63 Not Possible Thresholds Based on Regression bond. This test must be run on recently manufactured strand with Single Predictor with no surface weathering or rust (i.e., bright strand). The initial efforts made to define thresholds for each of these Change in Corrosion Potential--The prediction interval recommended QC methods were based on single-predictor for Change in Corrosion Potential with a one-sided confi- linear regressions and are described below. The results are dence level of 90% is shown in Figure 6. As can be seen in this summarized in Table 7. figure, this prediction interval exceeds 0.313 ksi when the Weight Loss on Ignition (LOI)--The prediction interval Change in Corrosion Potential is less negative than -0.175 V. for LOI with a one-sided confidence level of 90% is shown in Therefore, based on the data and the NASPA-defined thresh- Figure 4. As can be seen in this figure, the prediction interval old on mortar pull-out at 0.1-in. slip stress, a Change in Cor- does not exceed 0.313 ksi anywhere over the range of test rosion Potential of -0.175 V or more (i.e., less negative) is results observed in this study. For that reason, no threshold recommended to give a good confidence of adequate bond. can be determined. Organic Residue Extraction--The prediction interval for Contact Angle Measurement after Lime Dip--The predic- organic residue extraction with a one-sided confidence level of tion interval for Contact Angle after Lime Dip with a one-sided 90% is shown in Figure 7. As can be seen in this figure, the pre- confidence level of 90% is shown in Figure 5. As can be seen diction interval does not exceed 0.313 ksi anywhere over the in this figure, this prediction interval exceeds 0.313 ksi when range of test results observed in this study. For that reason, the contact angle is less than 73. Therefore, based on the data no threshold can be determined. A similar analysis was and the NASPA-defined threshold on mortar pull-out stress attempted considering only those sources with organic residue at 0.1-in. slip, a Contact Angle after Lime Dip of 73 or lower that the FTIR analyses indicated was primarily stearate. This is recommended to give a good (90%) confidence of adequate was done to eliminate potentially confounding influences of 0.700 Mortar Pull-Out 0.1-in Slip Stress (ksi) 0.600 Loss on Ignition 0.500 (mg/cm2) Prediction Interval Lower Bound 0.400 0.300 0.200 y = -1.4031x + 0.4454 0.100 R2 = 0.1595 0.000 -0.100 -0.05 0 0.05 0.1 0.15 Loss on Ignition (mg/cm2) Figure 4. Prediction interval (confidence level 90%) for Loss on Ignition. Threshold not possible.

OCR for page 25
27 0.700 () Contact Angle After Lime () Mortar Pull-Out 0.1-in Slip Stress (ksi) 0.600 Prediction Interval Lower Bound Threshold on Contact Angle After Lime () 0.500 Threshold on Mortar Pull Out 0.400 0.300 y = -0.0123x + 1.3929 0.200 R2 = 0.5701 0.100 0.000 -0.100 60 70 80 90 100 110 Contact Angle after Lime () Figure 5. Prediction interval (confidence level 90%) for Contact Angle after Lime Dip. non-stearate-based lubricants and other surface contaminants. based on the adjusted coefficient of determination (R2 adj.), The prediction interval for this stearate residue with a one-sided were as follows: confidence level of 90% is shown in Figure 8. As can be seen in this figure, the R2 is higher, but the prediction interval still does Contact Angle Measurement after Lime Dip & Change in not exceed 0.313 ksi anywhere over the range of test results ob- Corrosion Potential, served in this study, and no threshold can be determined. Contact Angle Measurement after Lime Dip & Organic Residue Extraction (100% stearate only), and Thresholds Based on Regression Weight Loss on Ignition (LOI) & Contact Angle Measure- with Multiple Predictors ment after Lime Dip & Change in Corrosion Potential. An attempt also was made to determine if combinations of Note that for multiple-predictor regression, a larger number test results (e.g., a combination of contact angle and organic of variables will increase the R2. Therefore, the adjusted R2 residue extraction test results) correlated with bond perfor- statistic, which accounts for the number of degrees of freedom mance. Although numerous linear combinations were exam- in the dataset, was calculated as a means to compensate for ined, the three combinations that showed the best correlation, this potentially misleading effect. The regression coefficients 0.700 Mortar Pull-Out 0.1-in Slip Stress (ksi) 0.600 y = 1.7413x + 0.7656 R2 = 0.6818 0.500 0.400 0.300 Change in Corr. Pot. (V) 0.200 Prediction Interval Lower Bound Threshold on Change in Corr. Pot. (V) 0.100 Threshold on Mortar Pull Out 0.000 -0.350 -0.300 -0.250 -0.200 -0.150 -0.100 -0.050 0.000 Change in Corr. Pot. (V) Figure 6. Prediction interval (confidence level 90%) for Change in Corrosion Potential.

OCR for page 25
28 0.700 Mortar Pull-Out 0.1-in Slip Stress (ksi) 0.600 Extracted Organic Residue (mg/cm2) 0.500 Prediction Interval Lower Bound 0.400 0.300 0.200 y = -1.7515x + 0.453 0.100 2 R = 0.1241 0.000 -0.100 0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 Extracted Organic Residue (mg/cm2) Figure 7. Prediction interval (confidence level 90%) for Organic Residue. Threshold not possible. and the R2 adj. for these three models are given in Tables 8 by the contact angle and organic residue extraction measure- to 10. The R2 adj. values for these combinations were high and ment methods. Given the high level of correlation with the equal to 0.73, 0.98, and 0.76, respectively. multiple regression approach, this model may be particularly The regression indicated that the last combination of useful in a production setting where the lubricant in use is predictors listed above (Contact Angle Measurement after known. Lime Dip & Organic Residue Extraction) was a good predictor The prediction interval can not be shown in a two- of bond and was performed based only on those strand dimensional plot as was done with the single-variable models. sources that the FTIR analysis of the organic residue identi- This is because multiple combinations of variables can give the fied as being stearate only. This limited the number of data same output. For this reason, a separate prediction interval points used to develop the regression model to five, but was must be calculated for each combination of variables. To give done as a means of eliminating potentially confounding influ- a sense of how these multiple regression models might be used, ences of non-stearate-based lubricants on the results obtained tables have been prepared showing the predicted pull out, the 0.450 Extracted Organic Residue Mortar Pull-Out 0.1-in Slip Stress (ksi) 0.400 (mg/cm2) 0.350 Prediction Interval Lower Bound 0.300 0.250 0.200 y = -1.9426x + 0.4355 0.150 2 R = 0.6282 0.100 0.050 0.000 0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 Extracted Organic Residue (mg/cm2) Figure 8. Prediction interval (confidence level 90%) for Organic Residue when FTIR analysis indicates organic residue is primarily stearate. Threshold not possible.