Probabilistic Assessment of UPV Test Result of a Degasser Unit

The novelty of this paper is to address a fundamental characteristic of ultrasonic pulse velocity (UPV) test result called distribution pattern and development of a methodology to simplify non-normal distribution patterns of UPV into normal distribution domain for easy applicability of quality assurance criteria for practicing engineers. In this paper probabilistic assessment methodology of UPV test results collected from a de-gasser structure of a refinery in India is discussed. Detailed analysis is performed to identify the distribution characteristics of UPV. No conventional distribution type including normal distribution fitted for UPV test results. A further attempt is made to derive simplified normal distributions with two different modes from the empirical cumulative distribution function of in situ test results. This creates scope for application of existing normal distribution based quality assurance criteria to non-normal domain of UPV. Criteria of characteristic strength instead of mean value is proposed as a better parameter for estimation of the quality of the material. Detailed physical interpretation and justification for the methodology is discussed with mathematical reasoning. The method is discussed in reference to existing literature. It is shown that the method is applicable to derive quality assurance criteria where simple assumption of normal distribution is not valid for UPV. Thus the method has the potential to make UPV a more useful test based on statistical reasoning for application on full scale structures.