PSI - Issue 52

A.D. Cummings et al. / Procedia Structural Integrity 52 (2024) 762–784 A. Cummings / Structural Integrity Procedia 00 (2023) 000–000

769

8

Post manufacture, the body was subject to ultrasonic inspections for volumetric flaws. Flaws greater than or equal to 25mm in length or an area of 100mm 2 were rejected. In this study initial embedded elliptical flaws of 5mm x 25mm (2a x 2c) have been considered. In cases where this size may be too onerous, a reduced size of 2 x 7mm has been considered, with provision for the possible re-inspection of regions of the package body, see flaw sizing guidance in Annex T of BS7910 (2019). The package body was also subjected to surface flaw Non-Destructive Testing (NDT) using Magnetic Particle Inspection (MPI), no detected flaws were permitted, and any found were dressed out providing the loss of thickness was acceptable. According to records, no repairs were made i.e. no flaws were found. Guidance in Annex T of BS7910 (2019) suggests MPI can reliably detect semi-elliptical surface breaking flaws of 1 x 5mm (a x 2c). This size of surface flaw has been used in the following assessment.

4. Overview of approach

Fig. 2 presents an overview of the assessment approach. The initial drop orientation study is performed to determine the case(s) that may challenge the avoidance of brittle fracture. During this stage elastic-plastic material properties for all components are applied, varying the angle of drop orientation from 0 ◦ (lid down) to 180 ◦ (base down) and a fixed drop height of 1.2m. The models are run with the explicit Finite Element code LS DYNA (2022). During this stage the contour plot (d3plot) states are requested every 0.5ms. Stress time histories are written to the binary output files at selected elements within the model at anticipated locations of high stress. For example, elements were selected at the centre of the base inner and outer wall and output requested to the binary file (binout) every 10 − 6 seconds. A su ffi cient number of elements are selected throughout the package to adequately assess the package body, lid bolts and lid - these elements can be considered as virtual strain gauges. This approach reduces the size of output files and expediates post processing. This stage of the assessment is automated with bash, TCL and python scripting. Plots of maximum principal stress vs drop orientation angle are produced for each of the selected elements. Once the worst case drop orientations have been determined (there can be several), the models are each re-run twice. The first run is performed with elastic material properties in the region of interest to provide elastic stresses that are compatible with the solutions for stress intensity and reference stress provided in BS7910 (2019). In any load transfer controlling regions (i.e. the shock absorber or lifting trunnions) the material modelling is elastic-plastic including the e ff ects of strain rate sensitivity which increases the load into the region of interest. From the stress time histories the time of peak stress occurrences are obtained, and the second pass run is used to request contour plot output at each peak time. Again, this approach minimizes output file size. The duration of the stress pulse is also obtained, this is taken as an estimate of t f , and is used for deriving dynamic fracture toughness, K mat . The second pass runs produce contour plots of the maximum principal stresses which are then used to derive crack driving force from the uncracked model. This is achieved by postulating cracks perpendicular to the principal stress direction. Assigning a Stress Classification Line (SCL) across the section enables plotting of the stress distribution across the section thickness (perpendicular to the flaw and the maximum principal stress direction). Stress linearization techniques are then used to obtain the crack driving force following guidance material i.e. BS7910 (2019) or ASME FFS-1 / API579-1 (2016). Often this results in membrane and bending stresses from which values of stress intensity, K I and σ ref , can be calculated either by cracked body FEA modelling or the use of compendia solutions BS7910 (2019) or ASME FFS-1 / API579-1 (2016). At the end of this stage, reducing the number of cases to consider is possible by comparing the characterised stresses across similar sections. The next stage is to classify the stresses as either primary or secondary - for this the original elastic-plastic model is used but the yield stress of the component / region of interest is reduced by a factor F y = 0.99 → 0.05. until either (1) the whole section yields (considered as plastic collapse) or (2) the stresses redistribute and yielding never fully develops through the section. For outcome (1) the stresses are classified as primary stress and the uncracked plastic collapse ratio L r , u = F y . For outcome (2) the stresses are classified as secondary stresses and will not contribute to plastic collapse. This method was developed as a dynamic assessment equivalent to say, increasing the pressure in a pressure vessel to observe whether the stresses contribute to plastic collapse.