A physics informed emulator for laser-driven radiating shock simulations
AuthorsMcClarren, Ryan G.
Paul Drake, R.
van der Holst, Bart
Paul Holloway, James
Kuranz, Carolyn C.
Torralva, Ben R.
KAUST Grant NumberKUS-C1-016-04
Permanent link to this recordhttp://hdl.handle.net/10754/597375
MetadataShow full item record
AbstractThis work discusses the uncertainty quantification aspect of quantification of margin and uncertainty (QMU) in the context of two linked computer codes. Specifically, we present a physics based reduction technique to deal with functional data from the first code and then develop an emulator for this reduced data. Our particular application deals with conditions created by laser deposition in a radiating shock experiment modeled using the Lagrangian, radiation-hydrodynamics code Hyades. Our goal is to construct an emulator and perform a sensitivity analysis of the functional output from Hyades to be used as an initial condition for a three-dimensional code that will compute the evolution of the radiating shock at later times. Initial attempts at purely statistical data reduction techniques, were not successful at reducing the number of parameters required to describe the Hyades output. We decided on an alternate approach using physical arguments to decide what features/locations of the output were relevant (e.g., the location of the shock front or the location of the maximum pressure) and then used a piecewise linear fit between these locations. This reduced the number of outputs needed from the emulator to 40, down from the O(1000) points in the Hyades output. Then, using Bayesian MARS and Gaussian process regression, we were able to build emulators for Hyades and study sensitivities to input parameters. © 2011 Elsevier Ltd. All rights reserved.
CitationMcClarren RG, Ryu D, Paul Drake R, Grosskopf M, Bingham D, et al. (2011) A physics informed emulator for laser-driven radiating shock simulations. Reliability Engineering & System Safety 96: 1194–1207. Available: http://dx.doi.org/10.1016/j.ress.2010.08.012.
SponsorsThis research was supported by the DOE NNSA/ASC under the Predictive Science Academic Alliance Program by Grant number DEFC52- 08NA28616. R.G. McClarren, D. Ryu, and B. Mallick's contributions were partially supported by Award no. KUS-C1-016-04, made by King Abdullah University of Science and Technology (KAUST).