Quality Assurance of Measured Response Intended for Fatigue Life Prediction

Abstract

By in situ measurements, the real distribution of stress ranges caused by service loads can be recorded. Inherent effects in the measured response as disturbance can, however, create puzzles in the interpretation of the result. In large monitoring campaigns, it is not possible to examine the result from every gauge over the whole measured period by visual control. An approach for a quality assurance of the measured response is presented here based on established statistical methods. The stress range spectra, the product of the monitoring program intended for fatigue assessment, are analyzed. The aim is to find deviant spectra and identify corrupt gauges. An additional aspect is the length of the monitoring period, which is the required duration for obtaining a stable result. A case study of a monitored Swedish steel railway bridge is incorporated to exemplify the approaches. Some statistical distributions for the monitored stress ranges are also presented and incorporated in a fatigue assessment.