Session: DA-08-03 Developments in FFS Assessment

Paper Number: 123393

123393 - The Impact of Fea Modeling Techniques for Level 3 Dent Engineering Critical Assessment: Shell vs. Solid Elements 

Engineering Critical Assessments (ECA) is crucial to evaluate the fitness for service of pipeline dents to comply with regulatory standards such as US CFR 49 part 192.712(c) and Canada CSA Z662 clause 10.10.4. For dents requiring a level 3 assessment, finite element analysis (FEA) is often employed to evaluate complex dent features categorized from in-line inspection (ILI) or direct examination of the pipeline. Various analysis techniques and element types exist within FEA for performing level 3 dent assessments. The appropriate analysis technique and corresponding element types (i.e., shell or solid elements) are often chosen by balancing the dent ECA's complexity with computational accuracy and time. Shell elements are more computationally efficient than solid elements, however, the through-thickness stress distribution is more accurate with solid elements. If computational time is of little concern, then solid elements are typically employed to obtain an accurate representation of the stress/strain distributions in and around the dent feature for further engineering assessment once FEA is completed. For instance, when the dent interacts with other complex deformation features such as metal loss, shell elements account for the local metal loss feature to achieve the most accurate results. However, the nature of dent ECA rarely affords an assessment schedule with such a flexible time constraints. As such, this study will compare the differences in using shell vs. solid elements for a dent ECA, both plain dents (i.e. interacting with no other features) and dents interacting with metal loss, using the same indenter. Variations in the stress/strain results and overall differences in the calculated remaining life of the feature will be assessed as a function of the solver time. The goal of the study is to provide guidance on how and to what extent the element type can affect the overall conclusions of two different element types of dent ECA.  

Presenting Author: Alex Brust DNV

Presenting Author Biography: Dr. Alex Brust is an engineer in the Hydrogen Services and Model group at DNV under the Materials Advisory Services Section in Dublin, Ohio. He graduated from The Ohio State University with a doctorate in computational materials science and engineering under funding acquired from the Air Force Research Laboratory (AFRL). The work involved the use of a clustering algorithm known as graph cutting to inverse problems in materials science, including image segmentation, cluster identification from noisy data sets, and the probabilistic reconstruction of the parent phase from EBSD-indexed child phase in ferrous and titanium alloys. The latter work utilized a Bayesian framework to computationally measure the austenite-to-martensite orientation relationship (OR) in a variety of steels and ferrous alloys before probabilistically reconstructing the prior austenite phase. The manuscript detailing the reconstruction process was recently selected by the publisher, Microscopy and Microanalysis, for further exposure in Microscopy Today.

He was then hired for a post-doctoral position at AFRL in the Novel Additive Manufacturing group. Here, he developed several algorithms related to the registration and fusion of data sets representing additively manufactured builds using a variety of different machine learning, image processing, and computational geometry techniques. His projects included the semi-automated registration of dense in-situ spectral sensor point cloud data to sparse scan vector intent data, the automated registration of a stack of x-Ray Chromatography (xCT) images of a single build part to the surface geometric mesh of the same build part, the automated 3D segmentation of the xCT images to identify pores within the build part, and the fusion of in-situ spectral sensor data to the abovementioned segmented pores.

His current work involves building tools involving advanced computational techniques including machine learning and probabilistic analysis for simulated pipeline defect assessments according to API 1183 standards. He continues working on microstructural analysis and is looking for correlations between plastic strain accumulations and crack growth rates. He is also working in areas involving data science and predictive programming related to decommissioning and decarburization of pipelines.

Authors:

Alex Brust DNV
Luyao Xu DNV
David Kemp DNV