TWI Industrial Member Report Summary 993/2011
By D Panggabean and C Carpentier
This report describes progress to date on a validation programme being carried out in TWI for the evaluation of the performance of finite element analysis (FEA) for the modelling of inspection of small fatigue cracks using ultrasonic phased array techniques.
Fatigue cracks may grow under service conditions; they occur in structural welds and can cause serious damage if not detected in a timely fashion. Over time, various non-destructive testing (NDT) methods have been developed for the detection of this type of flaw. Detection and characterisation of fatigue cracks present difficult challenges due to variations in crack orientation, roughness and size. One of the most important techniques for the detection of fatigue cracks has proven to be ultrasonic testing (UT), especially for cracks at a surface that is inaccessible for inspection (e.g. at rivet holes in aircraft). There is a need to be able to detect and to evaluate small cracks if integrity of fatigue sensitive structures is to be assured. Today, phased array technology is recognised to be a potential technique for the detection and sizing of very small cracks because of its ability to generate a sector of beam angle with a better ultrasonic beam resolution.
Theoretical models for ultrasonic inspection have been developed over a number of years. Modelling is a tool used to support the optimisation, the validation and a faster understanding of ultrasonic procedures and without manufacturing of expensive mock-ups.
TWI has gained over the years, considerable experience in the understanding and the use of semi-analytical software such as CIVA, developed by the CEA (French nuclear commission), and the in-house software developed by British Energy (BE). Large validation campaigns versus experimental measurement and other theories have been carried out to establish the limits of validity of these software packages (Carpentier et al, 2009). The two main theories used in semi-analytical models are Kirchhoff theory and the Geometrical Theory of Diffraction (GTD); both these theories have well documented limitations (Schneider and Chapman, 1997).
FEA promises to provide a more accurate approach for the simulation of the following realistic, but more complex, situations:
- Cracks with complex geometry (branched stress corrosion cracks and rough fatigue cracks).
- Cracks grouped in clusters (eg hydrogen cracks).
- Cracks that are small compared to the ultrasonic wavelength.
This project investigated the capability of the FEA method for the modelling of small smooth cracks. It was decided to limit this initial investigation of the FEA tools to smooth cracks because they are easier to model. Despite this restriction, the study has still a practical value since smooth defects are a reasonable representation of fatigue cracks and lack of fusion.
One potential application of a model allowing the accurate prediction of signals from small cracks (ie smaller than two ultrasonic wavelengths) is in justifying the capability of UT performed to a manufacturing standard such as Code Case 2235-9 in the ASME boiler and pressure vessel code. This Code Case specifies the flaw acceptance criteria for different weld thicknesses and flaw aspect ratios. It can be noted that the acceptance criteria for configurations with low wall thicknesses containing long flaws tend to require the rejection of flaws of small height and some of these cases cannot be reliably modelled by semi-analytical models. This limitation will be important for austenitic coarse grained materials where low frequency is required.
This report summarises the preliminary works for the evaluation and the choice of the FEA packages to be used to conduct the study on small smooth fatigue crack. In this preliminary work, the FEA package PZFlex developed by the Weidlinger Associates was tested on known benchmark cases for the ultrasonic beam computation and the response on large smooth crack (ie greater than two wavelengths). The performances of PZFlex were compared against experimental data and results from semi analytical and FEA models.
Objectives of the preliminary work
- Establish the influential model parameters that need to be considered for FE modelling of ultrasonic testing.
- Compare the performance of the FE modelling package PZFlex for ultrasonic beam computation and smooth crack detection with previous validation evidence.