News at the Faculty of Informatics

Speaker: Joseph E. Bishop
  Sandia National Laboratories, USA
Date: Monday, July 3, 2017
Place: USI Lugano Campus, room SI-003, Informatics building (Via G. Buffi 13)
Time: 10:30-11:30



Two fundamental sources of error and uncertainty in macroscale solid-mechanics modeling are (1) the use of an approximate material model that represents in a mean sense the complex nonlinear processes occurring at the microscale, and (2) the use of homogenization theory with implicit approximations such as a separation-of-scales and the existence of a well-defined representative volume element. Macroscopic material models are typically in error when exercised outside of the calibration regime, and the assumptions in homogenization theory are typically violated in welded regions of a structure and for additively-manufactured (AM) metallic structures.

Previous work in understanding and quantifying these errors for polycrystalline materials used Direct Numerical Simulations (DNS) in which polycrystalline microstructures were embedded directly within engineering-scale structures. Crystal-plasticity models were used to represent the grain-scale physics. These brute-force DNS simulations, modeling up to 50000 grains, were compared with conventional macroscale simulations that use material properties obtained from homogenization theory. Even with the promise of future exascale computing, the use of DNS simulations in typical engineering calculations is infeasible. A more practical approach to realistic multiscale uncertainty quantification in solid mechanics is to adopt a posteriori model-form error-estimation techniques proposed by Oden and coworkers for composite materials (1996-2002). In this framework, a simple constitutive model is maintained at the macroscale, such as isotropic elasticity or von Mises plasticity, and the errors in engineering quantities of interest are assessed in a post-processing step using a localization process and error bounds. This approach is inherently scalable to arbitrarily large systems. Also, for a given loading scenario, the apparent macroscale properties can be adapted to minimize the goal-oriented error and uncertainty. This framework is demonstrated for laser-welded and additively-manufactured structures using a process-structure-performance modeling paradigm.



Joe Bishop received his Ph.D. in Aerospace Engineering from Texas A&M University in 1996. His graduate research was in the general areas of the mechanics of composite materials and the mechanisms and mechanics of material damping. From 1997 to 2004 he worked in the Synthesis & Analysis Department of the Powertrain Division of General Motors Corporation, performing thermal-structural analysis of internal combustion engines with a focus on predicting high-cycle fatigue performance of the base engine. He joined Sandia National Laboratories in 2004 in the Engineering Sciences Center. He has worked on diverse research topics such as impact and penetration, pervasive fracture and fragmentation modeling, polyhedral finite elements, geologic CO2 sequestration, and metal-additive manufacturing. His current research interests include multiscale modeling in support of metal-additive manufacturing, experimental methods and computational techniques for determining residual-stress fields, polyhedral finite element formulations and applications, model-form error estimation techniques in finite-element analysis, meshfree technologies for shock-physics applications, and second-generation wavelets.


Host: Prof. Kai Hormann