Description:

Abstract: Computational modeling of dynamic fracture is crucial for ensuring structural safety and optimizing material performance under rapid loading conditions, such as impacts and explosions, in applications ranging from aerospace and automotive to defense and geophysics. It underpins designs for tougher materials with improved energy absorption, as well as understanding microstructural effects on crack propagation. This talk will focus on an in-depth assessment and understanding of continuum damage modeling of dynamic fracture with an emphasis on model accuracy and regularization with respect to mesh size. We will first discuss dynamic fracture predictions with a local, isotropic continuum damage model based on a frame-invariant effective strain. We demonstrate that overall, various experimentally observed aspects of dynamic cracks are reproduced by this simple model, including acceleration of cracks to a steady state velocity, increased micro-branching and macro-branching with increased strain rates, crack velocity dependence of energy dissipation and fracture surface area, and the behavior of dynamic cracks at material interfaces. However, phenomena such as fragmentation cannot be explicitly predicted. Comparisons with phase-field models further underscore the effectiveness of this approach. Next, we will discuss model regularization with respect to mesh size. Since dynamic fracture is typically characterized by extensive crack branching, it usually implies the suppression of damage localization. Yet, continuum damage models must be employed along with localization limiters such as the crack band model (CBM) and rate dependent damage (RDD). A detailed analysis into model regularization reveals that CBM ensures mesh-independent energy dissipation at lower loading rates, where localized or minimally branched cracks dominate, but becomes ineffective at higher rates due to increased branching and diffusion of fractures. Conversely, RDD introduces an internal length scale that regularizes the solution at higher rates but struggles at lower ones. Combining CBM and RDD does not universally resolve mesh dependency issues, particularly at low rates, highlighting the absence of a single “go-to” strategy for achieving mesh objectivity across all conditions. These findings offer critical insights for researchers and engineers working to refine finite element modeling for dynamic fracture applications, as well as outline directions for future work.

Bio: Prof. Kedar Kirane is an associate professor of Mechanical Engineering at Stony Brook University, New York. His research focuses on characterizing, understanding, and predicting the fracturing and scaling behavior of various conventional and advanced heterogeneous materials. These include brittle materials, fiber reinforced composites, polymers, nanocomposites, geological and cementitious materials, and soft materials. His research combines experimental, computational, and theoretical approaches. The overarching goal is to develop reliable predictive capabilities and sound scientific bases for safe designs in various engineering applications. Prof. Kirane obtained his Ph.D. in 2014 from Northwestern University and joined the Mechanical Engineering faculty at Stony Brook University in Sept 2017. He also holds an M.S. degree from the Ohio State University (2007) and a B.S from the University of Pune, India (2004), both in mechanical engineering. Prior to joining Stony Brook, Prof. Kirane worked in industry, as a finite element analyst at Goodyear Tire & Rubber Co and later as a senior research engineer at ExxonMobil Corp. His research is supported by DOD ARO, DOD ONR, NSF NRT, and ASME. He is the recipient of the 2020 Orr Early Career Award by ASME’s Materials Division, the 2019 DOD ARO Young Investigator Award, and the 2018 Haythornthwaite Research Initiation Grant by ASME’s Applied Mechanics Division.