Approximated Repair Methods for Outlier Strain Data from Hole-Drilling Residual Measurements

Abstract

The current ASTM E 837 standard gives the standardized procedure for the evaluation of uniform and non uniform residual stresses, that is, stresses that do not vary and vary significantly with depth from the specimen surface, respectively. For non uniform stresses, the standard states that many small increments should be done in order to have a stable calculation of the residual stress profile. In addition, it states that irregularities as well as outlier strain points should be investigated and if necessary, the hole-drilling test should be repeated. In some applications outside the laboratory, the availability to repeat a test with outlier points is not possible. In these cases, the standard does not show the more appropriated way which should be followed to use only valid measured data (without outliers). For this reason, a stress profile corresponding to a shot peening test was simulated and one hole step was included in different hole depths as outlier point in order to evaluate the feasibility of some proposed ways of computation. These ways were: (a) following the ASTM procedure but replacing outlier strain points by linear interpolation of neighboring good ones, (b) following the ASTM procedure but replacing outlier strain points by a parabola interpolation of neighboring good ones and (c) using cumulative relaxation strain functions and only good measured points. Statistical criteria were also introduced and developed in order to identify outlier points. Results show a practical procedure to detect outlier points in experimental strain data.