ASME STP-NU-019-1:2009 pdf free download

ASME STP-NU-019-1:2009 pdf free download.VERIFICATION OF ALLOWABLE STRESSES IN ASME SECTION III SUBSECTION NH FOR GRADE 91 STEEL.
7.3 Stress-Rupture, tR
As outlined in earlier section on aaiIabIe sources lbr creep-rupture data, the correlation of stressnipture data to predict the long-time strength of Or 91 steel has been an on-going activity at an international level for dcvadcs. The undertakings have been largely in support for the usc of Or 91 steel in ASME 131W Codes Section I and Viii. ASME Piping Codes 1331.1 and 1331.3 and corresponding overseas constnlction codes. The objective has been to estimate accurately the allowable stresses at the upper limit of the use temperature for (ir 91 steel. Many parametric procedures have been developed and compared but there remains no consensus as to which is best. Techniques to “improve” the accuracy of long time estimations include “censoring” data by not using data for times less than 3000 hours [241. region splitting by not using data produced at stresses above a fraction of the hot yield strength (25J and adding more parametric constants to the time- temperature-stress models [26j. However. it should be recognized that the criteria for setting S, in ill- NH are conservative relative to the criteria in ASME lI-D Table 1-100. so the onus to produce accurate estimates from the same database is not as demanding.
Data corresponding to rupture lives less than 100 hours were not used in the analyses. This left nearly 1600 data covering temperatures from 450 to 780’C (i40 to 1435’F). The Larson Miller global fit to these data is shown in Figure 16 (left) and lot-centered fit is shown in Figure 16 (right). One fit appeared to be as good as the other, although there was a fi.ur point difference in the optimized parameiric constant: -26 for the global lit and —30 for the lot centered lit. The SEE values were similar: 0.333 in log time for the global fit and 0.345 in log time for the lot-centered fit. The distribution of residuals for the two tits was similar, and information is shown in Figure 17 for the lot centered model. The plols show how the residuals were distributed about tero. The distributions with temperature and trwss are shown in Figure 18. These distributions show no strong bias (Figures l8a and l8b). When plotted against the observed rupture lives, the residuals tended to move from a negative bias to a positive bias with increasing life (Figure 18c).