From 7f075cb00e3729af3c6201d7621391fa131d1bdd Mon Sep 17 00:00:00 2001 From: Delchini Marco Date: Sat, 28 Dec 2024 13:01:53 -0500 Subject: [PATCH] update paper.md --- paper.bib | 24 ++++++++++++++++++++++++ paper.md | 22 +++++++++++++++------- 2 files changed, 39 insertions(+), 7 deletions(-) diff --git a/paper.bib b/paper.bib index 435b145..c6569c4 100644 --- a/paper.bib +++ b/paper.bib @@ -19,3 +19,27 @@ @article{Clausen2013 doi = {10.1103/PhysRevE.87.013309}, url = {https://link.aps.org/doi/10.1103/PhysRevE.87.013309} } + +@misc{olcf-web, + title = {OLCF}, + note = {https://www.olcf.ornl.gov/}, + url = {https://www.olcf.ornl.gov/}, + year = {2024} +} + +@article{nicoud:hal-00910373, + TITLE = {{Subgrid-scale stress modelling based on the square of the velocity gradient tensor}}, + AUTHOR = {Nicoud, Franck and Ducros, Fr{\'e}d{\'e}ric}, + URL = {https://hal.science/hal-00910373}, + JOURNAL = {{Flow, Turbulence and Combustion}}, + PUBLISHER = {{Springer Verlag}}, + VOLUME = {62}, + NUMBER = {3}, + PAGES = {183-200}, + YEAR = {1999}, + DOI = {10.1023/A:1009995426001}, + KEYWORDS = {large eddy simulations ; wall-bounded flow ; unstructured mesh ; transition ; LARGE-EDDY SIMULATION ; NUMERICAL-SIMULATION ; TURBULENCE ; FLOW}, + PDF = {https://hal.science/hal-00910373v1/file/paper.pdf}, + HAL_ID = {hal-00910373}, + HAL_VERSION = {v1}, +} \ No newline at end of file diff --git a/paper.md b/paper.md index f73a91b..c936f28 100644 --- a/paper.md +++ b/paper.md @@ -16,6 +16,10 @@ authors: affiliation: 2 - name: Kalyan Gottiparthi affiliation: 3 + - name: Ryan Glasby + affiliation: 4 + - name: Franklin Stuart + affiliation: 5 affiliations: - name: Nuclear Energy and Fuel Cycle Division, Oak Ridge National Laboratory index: 1 @@ -23,6 +27,10 @@ affiliations: index: 2 - name: National Center for Computation Science Division, Oak Ridge National Laboratory index: 3 + - name: Computational Science and Engineering Division, Oak Ridge National Laboratory + index: 4 + - name: Enrichment Science and Engineering Division, Oak Ridge National Laboratory + index: 5 date: 19 December 2024 bibliography: paper.bib --- @@ -33,26 +41,26 @@ The demand for high-performance computational fluid dynamics and multiphysics so # Statement of need -The core work of the cross-cutting VERTEX Laboratory Directed Research and Development (LDRD) initiative aims to create a new multiphysics simulation framework supporting physical phenomena key to Oak Ridge National Laboratory (ORNL) mission-critical challenges. ORNL has clearly demonstrated needs in modeling and simulation of gas dynamics, rarefied flow, plasma-surface interaction, electromagnetics, magneto-hydrodynamics (MHD), and thermal hydraulics for conducting fluids, collisionless and collisional plasma, and structural mechanics. The project is organized into four technical areas: VERTEX-CORE, VERTEX-MAXWELL, VERTEX-CFD, and VERTEX-CLOSURE. Each area focuses on a specific physics. +The core work of the cross-cutting VERTEX Laboratory Directed Research and Development (LDRD) initiative aims to create a new multiphysics simulation framework supporting physical phenomena key to Oak Ridge National Laboratory (ORNL) mission-critical challenges. ORNL has clearly demonstrated needs in modeling and simulation of gas dynamics, rarefied flow, plasma-surface interaction, electromagnetics, magneto-hydrodynamics (MHD), and thermal hydraulics for conducting fluids, collisionless and collisional plasma, and structural mechanics. As part of the VERTEX initiative, the primary mission of the VERTEX-CFD team is to develop modeling and simulation capabilities to accurately model the physics in fusion blanket design. It thus requires a multiphysics solver to implement the incompressible Navier-Stokes (NS) equation to conjugate a heat transfer model and an MHD solver. Solvers, finite element methods, and other relevant tools are provided by the [Trilinos package](https://trilinos.github.io/) [@trilinos-website]. The VERTEX-CFD solver is designed to scale on HPC platforms by leveraging Kokkos programming language to ensure compatibility with various CPU and GPU architectures. -# Current capabilities and development worflow +# Current capabilities and development workflow -VERTEX-CFD solver is still under active development and currently implement the following capabilities: incompressible Navier-Stokes equations [@Clausen2013], temperature equation, induction-less and full-induction MHD models, RANS turbulence models and WALE (LES) turbulence model. Each new physics is implemented in closure models with unit tests. Physical models and coupling between equations were verified and validated against bechmark problems taken from the published literature: isothermal flows, heated flows, transient and steady-state cases, turbulent cases, and MHD flows. VERTEX-CFD solver has demonstrated second-order temporal and spatial accuracy. Scaling of the VERTEX-CFD solver was assessed on CPUs and GPUs architecture. It was found that strong and weak scaling were comparable to other CFD solvers alike NekRS. (ADD FIGURE). +VERTEX-CFD solver is still under active development and currently implements the following capabilities: incompressible Navier-Stokes equations [@Clausen2013], temperature equation, induction-less and full-induction MHD models, RANS turbulence models and WALE (LES) [@nicoud:hal-00910373] turbulence model. Each new physics is implemented in closure models with unit tests. Physical models and coupling between equations were verified and validated against bechmark problems taken from the published literature: isothermal flows, heated flows, transient and steady-state cases, turbulent cases, and MHD flows. VERTEX-CFD solver has demonstrated second-order temporal and spatial accuracy. Scaling of the VERTEX-CFD solver was assessed on CPUs and GPUs architecture. It was found that strong and weak scaling were comparable to other CFD solvers alike NekRS. (ADD FIGURE). -The long term objectives of the VERTEX initiative is to faciliate the additon of new physical models by relying on a plug-and-play architecture, and also guarentee the correctness of the implemented model over time. New physics and equations are easily added to the global tree and allow for quick deployment of new physical model on HPC platforms. Such approach can only be made possible by setting clear requirements and review process for all developers contributing to the project code: any changes and aditions to the source code is reviewed and tested before being merged. VERTEX-CFD solver is tested daily on a continuous integration (CI) workflow that is hosted on ORNL network. +The long term objectives of the VERTEX initiative is to facilitate the addition of new physical models by relying on a plug-and-play architecture, and also guarantee the correctness of the implemented model over time. New physics and equations are easily added to the global tree and allow for quick deployment of new physical model on HPC platforms. Such approach can only be made possible by setting clear requirements and review process for all developers contributing to the project code: any changes and additions to the source code is reviewed and tested before being merged. VERTEX-CFD solver is tested daily on a continuous integration (CI) workflow that is hosted on ORNL network. # Conclusions -VERTEX-CFD is an open-source CFD solver that relies on a finite element discretization method to solve for the incompressible Navier-Stokes equations coupled to a temperature equation and MHD equation. Reynolds Averaged Navier-Stokes (RANS) turbulence models and large eddy simulation model are also available. The code relies on the Trilinos package and offers a wide range of temporal integrators, solvers and preconditioners to run on CPU- and GPU-enabled platforms. The code was verified and validated for steady and unsteady incompressible flows with benchmark cases taken from the published literature: natural convection, viscous heating, laminar flow over a circle, and turbulent channels. It was also demonstrated that VERTEX-CFD solver scales on CPUs (Perlmutter) and GPUs (Perlmutter and Summit) architectures. +VERTEX-CFD is an open-source CFD solver that relies on a finite element discretization method to solve for the incompressible Navier-Stokes equations coupled to a temperature equation and MHD equation. Reynolds Averaged Navier-Stokes (RANS) turbulence models and large eddy simulation model are also available. The code relies on the Trilinos package and offers a wide range of temporal integrators, solvers and preconditioners to run on CPU- and GPU-enabled platforms. VERTEX-CFD solver was verified and validated for steady and unsteady incompressible flows with benchmark cases taken from the published literature: natural convection, viscous heating, laminar flow over a circle, and turbulent channels. It was also demonstrated that VERTEX-CFD solver scales on CPUs (Perlmutter) and GPUs (Perlmutter and Summit [@olcf-web]) architectures. Future development will focus on implementing wall functions for RANS models, and the addition of conjugate heat transfer capabilities. -# Acknolegements +# Acknowledgements -This work was funded by the Laboratory Directed Research and Development (LDRD) program at Oak Ridge National Laboratory, and the Scientific Discovery through Advanced Computing (SciDac) programm. +This work was funded by the Laboratory Directed Research and Development (LDRD) program at Oak Ridge National Laboratory, and the Scientific Discovery through Advanced Computing (SciDac) program. # Disclaimer