CCES Unicamp

High-Performance Computing and Inertial Confinement Fusion (ICF)

Dr. Britton Olson

Data: 19/02/2019

Horário: 10:00h

Local: FEM – Auditório KD2

Abstract:

Lawrence Livermore National Laboratory (LLNL) in California is home to the largest laser facility (NIF) in the world and the 2nd fastest supercomputer, Sierra.  Indeed, this has been the hallmark and legacy of the Lab; being pioneers in fusion/nuclear energy research and in High-Performance Computing. In this talk, I will focus on a select few physics and computational issues we are researching at LLNL.

Inertial Confinement Fusion (ICF) at NIF is plagued by inefficiencies due to hydrodynamic instability growth and turbulent mixing. The fidelity of simple models has been insufficient to inform ICF design and therefore, higher fidelity approaches have been an active areas of research. We present results from a large-eddy simulation (LES) study on phenomenological mixing configurations which occur in ICF capsules. The data suggest that secondary flow features (not captured by low fidelity models) have a significant impact on the mean flow and mixing of the capsule.

High-Performance Computing (HPC) is approaching the Exascale at LLNL, enabling massive (~100 billions grid points) calculations of ICF mixing to be conducted. LLNL has many HPC tools and proxy applications that are being developed to facilitate simulation code development on these large machines. A brief overview of these tools and efforts will be given, with a focus on open source projects and GPU enabling technologies.

Finally, a finite difference research code, Pyranda, is introduced. Pyranda is the open source proxy application of the Miranda code and is framework for solving systems of PDEs on a structured, curvilinear mesh on massively parallel machines. The core solver is a Fortran library that has been wrapped (via f2py) in python which allows for rapid model development and greater solver flexibility. A handful of simple test problems and Pyranda features are presented.

An informal workshop for interested students and faculty will be held on Tuesday and Wednesday at the School of Mechanical Engineering, Seminar DEMM seminar room, third floor of builindg – H.

Bio:

Dr. Britton Olson is a computational physicist at LLNL and joined the lab in 2012, spending two years as a postdoc.  His research has focused on hydrodynamics, turbulence mixing, and large eddy simulation. He also directs a summer internship program of roughly 50 students each year. Dr. Olson completed a PhD and Master’s degree at Stanford University with Sanjiva Lele and before that a degree in Mechanical Engineering from BYU. Dr. Olson has hosted two UNICAMP students at LLNL over the past 3 years as a co-PI with Professor William Wolf on FAPESP grants.

Related posts

Multiscale Analysis of Heterogeneous Materials

cces cces

Unveiling the Transformer Architecture: Fundamentals, Applications, and Future Perspectives

cces cces

Open Science, Open Data – Challenges in Data Sharing across Scientific Domains

cces cces