Bayesian calibration of strength parameters using hydrocode simulations of symmetric impact shock experiments of Al-5083

Journal Article · 26 November 2018 · Journal of Applied Physics

DOI: https://doi.org/10.1063/1.5051442 · OSTI ID:1489952

Walters, David J.  ^[1]; Biswas, Ayan  ^[1]; Lawrence, Earl Christopher  ^[1]; Francom, Devin Craig  ^[1]; Luscher, Darby Jon  ^[1]; Fredenburg, David Anthony  ^[1]; Moran, Kelly Renee  ^[2]; Sweeney, Christine Marie  ^[1]; Sandberg, Richard L.  ^[1]; Ahrens, James Paul  ^[1]; Bolme, Cynthia Anne  ^[1]

---

1. Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
2. Duke Univ., Durham, NC (United States); Los Alamos National Lab. (LANL), Los Alamos, NM (United States)

We report that predictive modeling of materials requires accurately parameterized constitutive models. Parameterizing models that describe dynamic strength and plasticity require experimentally probing materials in a variety of strain rate regimes. Some experimental protocols (e.g., plate impact) probe the constitutive response of a material using indirect measures such as free surface velocimetry. Manual efforts to parameterize constitutive models using indirect experimental measures often lead to non-unique optimizations without quantification of parameter uncertainty. This study uses a Bayesian statistical approach to find model parameters and to quantify the uncertainty of the resulting parameters. The technique is demonstrated by parameterizing the Johnson-Cook strength model for aluminum alloy 5083 by coupling hydrocode simulations and velocimetry measurements of a series of plate impact experiments. Simulation inputs and outputs are used to calibrate an emulator that mimics the outputs of the computationally intensive simulations. Varying the amount of experimental data available for emulator calibration showed clear differences in the degree of uncertainty and uniqueness of the resulting optimized Johnson-Cook parameters for Al-5083. The results of the optimization provided a numerical evaluation of the degree of confidence in model parameters and model performance. Lastly, given an understanding of the physical effects of certain model parameters, individual parameter uncertainty can be leveraged to quickly identify gaps in the physical domains covered by completed experiments.

• Research Organization: Los Alamos National Laboratory (LANL)
• Sponsoring Organization: Laboratory Directed Research and Development (LDRD); USDOE
• Grant/Contract Number: 89233218CNA000001; AC52-06NA25396
• OSTI ID: 1489952
• Report Number(s): LA-UR-18-20884
• Journal Information: Journal of Applied Physics, Journal Name: Journal of Applied Physics Journal Issue: 20 Vol. 124; ISSN 0021-8979
• Publisher: American Institute of Physics (AIP)Copyright Statement
• Country of Publication: United States
• Language: English