Automation of simulation supported POD computations with CIVA for power generation industry

摘要

Simulation of NDT is extensively used in different industries worldwide including the power industry. Used at the relevant stage, simulation allows reducing costs by decreasing the number of necessary mock-ups and measurements. For UT inspections of some components used in electric power plants, the testing performance evaluation requires the determination of a Probability of Detection value (PoD). In this case, the PoD value is based on the comparison between the ultrasonic inspection results expressed in ERS unit (Equivalent Reflector Size) and the real size of the flaw, considering the different transducers used for the inspection and accounting for the potential variability of influential parameters (type and orientation of flaws, uncertainties of transducer properties, etc.). As a PoD study relies on a lot of inspection results, this is particularly costly and time consuming, and this is a typical case where simulation helps to dramatically reduce the number of mock-ups and physical tests. Nowadays, CIVA is considered as the reference tool worldwide for simulating NDT and ALSTOM wished to use CIVA in order to support PoD studies with simulation. However, generating the required data is challenging especially because ERS values are not provided as default outputs from CIVA simulations. Then, conducting a PoD study requires running a lot of cases in order to cover the potential values of influential parameters for the uncertain data. Analyzing individual results and converting them manually in ERS unit would be prohibitive for a reasonable study. Finally, the PoD estimation requires accounting for different transducers which means repeating the same process for the different inspection cases and mixing-up the results in the relevant way. Once again, without automation, the process is really difficult to implement. Then, ALSTOM and EXTENDE worked together to develop a methodology associated with CIVA in order to make this kind of UT simulations more efficient. The tools developed by EXTENDE and allowing automation of the simulation process are presented in this paper.