Computational Certification Under Limited Experimental Data
Abstract
We will develop and validate a novel computational certification framework that relies onlimited experimental data. The proposed computational certification builds on the PI~s prior work1which showed that the high fidelity multiscale software2 calibrated to limited experimental data iscapable of making reliable blind predictions, i.e., predictions matching experimental data withoutseeing it beforehand. The proposed computational certification framework consist of five majorelements: (i) a high fidelity model that can be calibrated to limited experimental database withrelative ease; (ii) a digital database that can be created using the high fidelity model; (iii) a lowerfidelity, but computationally superior reduced order model (ROM), that can be calibrated to theintegrated experimental-digital database and subsequently used for component analysis andcertification; (iv) a particle-based manufacturing simulation engine capable of predicting defectstructure and residual stresses arising from the resin transfer molding (RTM) and thus furtherreducing reliance on physical testing; and (v) a subscale (unresolved scale) simulation engine thatcalibrates a subscale model possessing defect structure to a single-scale defect-freephenomenological model.All phases of model calibration and digital database development will be fully automatedand implemented in: (i) FOOF, which is a three-dimensional nonlinear multiscale-multiphysicsfinite element code written in object oriented Fortran developed by the PI~s group, (ii) MultiSPH,which is a multiscale SPH code for process modeling currently under development by the PI~sgroup, and (iii) the Seg3D-Cleaver, which are the BSD-style open source codes for creatingmicrostructural finite element models directly from micrographs.We will validate the proposed computational certification framework on the carbon fiberpolymer matrix composites material system. The AGATE3 and NCAMP4 databases will be usedto validate the computational certification framework. The database from the recent DOEprogram entitled ~Integrated Computational Materials Engineering (ICME): Development ofCarbon Fiber Composites for Lightweight Vehicles~ will be used validate the influence ofmanufacturing on material performance. The Sikorsky CH-53K tail rotor mini-flexbeam will beused for demonstration of capabilities developed.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 04, 2017
- Source ID
- N000141712038
Entities
People
- Jacob Fish
Organizations
- Office of Naval Research
- Trustees of Columbia University in the City of New York
- United States Navy