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 33
33 separation of the ranges of dependent variables (QC test If the type(s) of pretreatment and lubricants used in the result) and independent variable (pull-out response) will wiredrawing process are known, method-specific correlations result in greater confidence in the model that is developed. and thresholds could be determined. For example, if only a Even for the same scatter on the individual test results, borax pretreatment and stearate-based drawing compounds if samples with a wider range of performance were used, were used to manufacture the strand in a certain plant, the a smaller prediction interval would result. effects of other wiredrawing compounds on the QC test results · A limited number of data points were available for the re- would not be present, and it is likely that a more consistent gression analysis--A greater number of data points would response would be achieved (resulting in higher R2 and smaller increase the confidence in the regression model estimates' prediction intervals). Based on the same predefined mortar ability to accurately predict performance. This would re- pull-out threshold, different QC test thresholds could be de- duce the prediction interval. veloped for that particular manufacturing process. This would require each manufacturer to maintain a record of QC test and pull-out test results from strand they produced. With the Future of Quality Control Program computational tool developed as part of this program, regres- and Thresholds sion analysis could be conducted. A QC program developed The discussion of the QC test methods and QC test program on this basis will be the most effective use of the recommended given here has focused on the sampled sources used in this QC test methods. test program. A number of thresholds have been proposed based on the relationships between the QC test and the mortar Computational Tool pull-out test results for this sample set. Although a signifi- cant amount of work and scientific rigor has gone into their As mentioned, a computational tool in the form of a development, these thresholds should not be considered Microsoft Excel-based spreadsheet has been developed. This absolute and immutable. Excel workbook was designed to predict whether prestressing Because of the issues discussed and the finite nature of strand will exhibit adequate bond properties based on results this test program, the number of samples included in the re- from surface and chemical QC test methods developed as gression analyses was limited. However, it is suggested that part of this project. The workbook performs this prediction the threshold development process could be an ongoing according to the procedures outlined in Appendix B of this process. If the recommended QC methods were conducted report. on the samples selected for inclusion in the quarterly mor- This is done by calculating the prediction intervals for single- tar pull-out test program currently underway by NASPA and multiple-predictor regressions and for determining the plants, that would provide nine additional data points for threshold on the QC test that corresponds to a predefined inclusion in the regression dataset each quarter. These data threshold using the mechanical test method. This tool is de- could be included in future regression analysis and used to signed to: (1) develop a regression model to predict mortar refine the existing thresholds or perhaps allow the definition pull-out stress from inputted surface and chemical QC test of future thresholds for those methods for which determi- and mechanical (mortar) pull-out test results, (2) establish the nation was not possible at this time. Even if these sources lower bound of the prediction interval for the regression model proved to be all of similar bond quality, the additional data for a desired level of confidence, (3) compare the lower bound would likely serve to improve confidence in the regression of this prediction interval with a predefined threshold for the model. mortar pull-out stress with the chosen level of confidence, and The proposed thresholds can be applied to new sources of (4) determine a pass/fail threshold for the QC test, if possible. strand, but this should be done with some caution. Strand There are individual worksheets for each of the recommended produced with a different pretreatment process or with a QC tests and for the recommended combinations of these lubricant with significantly different chemistry may not tests. The user has the ability to modify the following inputs: respond similarly to strand produced with a borax or zinc desired level of confidence (default: 90%), threshold for ac- phosphate pretreatment and largely calcium or sodium ceptability (default: 0.313 ksi), and the QC test result for a stearate lubricants. A main objective behind the inclusion of new strand source. the FTIR test in the organic residue extraction test method The acceptability of a given source is determined by com- is to confirm that the organic component of the lubricant at paring the lower bound on the prediction interval to a prede- least is consistent. It is likely that the existing thresholds will fined pass/fail threshold for the mortar pull-out test. Based on reject a larger number of sources than the mortar pull-out this comparison, a judgment is made as to whether the source test alone. exhibiting the QC test results is expected to exceed that
OCR for page 34
34 threshold with the chosen confidence level. Indication is then A universally applicable set of thresholds on the QC test re- given if the source passes (lower bound on prediction interval sults can not be generated, and the prediction interval cannot above threshold) or fails (lower bound on prediction interval be simply plotted. Instead, the prediction interval must be cal- below threshold). For the combined, multiple-predictor re- culated separately for each combination of QC test results. This gressions, the prediction interval is different for each set of tool accomplishes that calculation task. The multiple-predictor input QC results. analysis is interpreted in terms of the pass or fail statements.