diff --git a/software/feelpp/WP1/WP1.tex b/software/feelpp/WP1/WP1.tex index 2e371e0..0084d87 100644 --- a/software/feelpp/WP1/WP1.tex +++ b/software/feelpp/WP1/WP1.tex @@ -611,6 +611,26 @@ \subsubsection{Benchmark \#2: Linear elasticity : NAFEMS LE10} \paragraph{Input/Output Dataset Description} +\begin{itemize} +\item \textbf{Input Data:} + \begin{itemize} + \item Meshes: We have generated three levels of mesh called M2, M3 + and M4. These meshes are stored in GMSH format. The statistics can be found in + \Cref{tab:wp1:feelpp:nafems-le10:discr_stat}. We have also prepared for + each mesh level a collection of partitioned mesh. + The format used is an in-house mesh format of \Feelpp based on + JSON+HDF5 file type. + The Gmsh meshes and the partitioned meshes can be found on our Girder + database management, in the \Feelpp collections. + \item Setup: Use standard setup of \Feelpp toolboxes. It corresponds to a cfg + file and JSON file. These config files are present in the Github of feelpp. + \item Sif image: feelpp:v0.111.0-preview.10-noble-sif (stored in the Github registry of \Feelpp) + \end{itemize} +\item \textbf{Output Data:} The output includes the computed values of + validation measure in CSV files format, export visualization files (mesh, + partitioning, displacement, ...), and the time taken to perform each simulation step. +\end{itemize} + \begin{table}[!ht] \centering { \setlength{\parindent}{0pt} @@ -855,49 +875,12 @@ \subsubsection{Benchmark \#3: Thermo-Electric Coupling} This benchmark models the temperature field and electric current distribution in an high field resistive magnet of the Laboratoire National des Champs Magnétiques Intenses. The magnet consist in a set of 14 copper alloys cylindrical tubes connected 2 by 2 into series by rings. In each tube, the current path is defined by 2 helical cuts of 0.2 mm width. The rings are machined to let water flow in between each tube in a channel of 0.8 mm. The magnet is operated at 12 MW with an imosed total current of 31 kA. The water flow in the magnet is about 140 l/s. The water cooling of the magnet is modelling by using Robin boundary conditions with parameters derived from classical correlation in thermo-hydraulics. -%% TODO add ref to Cecile or Romain Phd %% -A more detailled version of the full model is available in Daversion2016. The model is run with \texttt{thermoelectric} \Feelpp toolbox. - -\paragraph{Benchmarking Tools Used} -The benchmark was performed on \textbf{Gaya} supercomputer (see \Cref{sec:arch:gaya}) and \textbf{Discoverer} supercomputer (see -\Cref{sec:arch:eurohpc-ju}). -The performance tools integrated into the \Feelpp-toolboxes framework were used to measure -the execution time. -Moreover, we need to say that we have used several \Feelpp installations -\begin{itemize} -\item \textbf{Gaya} : native application from Ubuntu packages of Jammy OS. -\item \textbf{Discoverer} : Apptainer with \Feelpp SIF image based on Ubuntu - Noble OS. -\end{itemize} -Note: the \Feelpp version is identical but the dependencies (like Petsc) -which are of course more recent with Noble. - -The metrics measured are the execution time of the main components of the simulation. We enumerate these parts in the following: -\begin{itemize} -\item \textbf{Init}: load mesh from filesystem and initialize solid toolbox (finite element context and algebraic data structure) -\item \textbf{Assembly}: calculate and assemble the matrix and rhs values obtained using the finite element method -\item \textbf{Solve}: the linear system by using a preconditioned GMRES. Results - are presented in \Cref{sec:WP3:Feelpp:benchmark:hl-31}. -\item \textbf{PostProcess}: Export on the filesystem a visualization format (EnsighGold) of the - solution and other fields of interest such as current density and electric field. -\end{itemize} +A more detailled version of the full model is available in \cite{daver2016,Hild2020}. The +model is run with \texttt{thermoelectric} \Feelpp toolbox. - -\paragraph{Input/Output Dataset Description} - -\begin{itemize} - \item \textbf{Input Data:} The input dataset consists in a 3D tetrahedral mesh with about 50 millions of tetras along with the configuration files necessary to run the simulations. - \item \textbf{Output Data:} The output includes the computed temperature, current distribution stored in HDF5 format, as well as some integral quantities such a the total power dissipated by the magnet. - \item \textbf{Data Repository:} All input and output datasets are available in a unistra girder repository (collection HiFiMagnet, HL-31). -\end{itemize} - - -\paragraph{Description} - -This benchmark models the temperature field and electric current distribution in an high field resistive magnet of the Laboratoire National des Champs Magnétiques Intenses. The magnet consist in a set of 14 copper alloys cylindrical tubes connected 2 by 2 into series by rings. In each tube, the current path is defined by 2 helical cuts of 0.2 mm width. The rings are machined to let water flow in between each tube -in a channel of 0.8 mm. The magnet is operated at 12 MW with an imosed total current of 31 kA. The water flow in the magnet is about 140 l/s. The water cooling of the magnet is modelling by using Robin boundary conditions with parameters derived from classical correlation in thermo-hydraulics. -%% TODO add ref to Cecile or Romain Phd %% -A more detailled version of the full model is available in Daversion2016. The model is run with \texttt{thermoelectric} \Feelpp toolbox. +The geometry used in this benchmark performance is illustrated in +\Cref{fig:wp1:feelpp:hl-31:visualization-geometry}. This is a complex domain +composed of a large number of components, with some very thin parts. \paragraph{Benchmarking Tools Used} The benchmark was performed on \textbf{Gaya} supercomputer (see \Cref{sec:arch:gaya}) and \textbf{Discoverer} supercomputer (see @@ -923,16 +906,6 @@ \subsubsection{Benchmark \#3: Thermo-Electric Coupling} solution and other fields of interest such as current density and electric field. \end{itemize} - -\paragraph{Input/Output Dataset Description} - -\begin{itemize} - \item \textbf{Input Data:} The input dataset consists in a 3D tetrahedral mesh with about 50 millions of tetras along with the configuration files necessary to run the simulations. - \item \textbf{Output Data:} The output includes the computed temperature, current distribution stored in HDF5 format, as well as some integral quantities such a the total power dissipated by the magnet. - \item \textbf{Data Repository:} All input and output datasets are available in a unistra girder repository (collection HiFiMagnet, HL-31). -\end{itemize} - - \begin{figure}[!ht] \centering \begin{subfigure}[c]{0.49\textwidth} @@ -949,8 +922,37 @@ \subsubsection{Benchmark \#3: Thermo-Electric Coupling} \end{figure} + \paragraph{Input/Output Dataset Description} +\begin{itemize} +\item \textbf{Input Data:} + \begin{itemize} + \item Meshes: We have generated three levels of mesh called M1, M2 + and M3. These meshes are stored in GMSH format. The statistics can be found in + \Cref{tab:wp1:feelpp:thermal_bridges:discr_stat}. We have also prepared for + each mesh level a collection of partitioned mesh. + The format used is an in-house mesh format of \Feelpp based on + JSON+HDF5 file type. + The Gmsh meshes and the partitioned meshes can be found on our Girder + database management, in the \Feelpp collections. + \item Setup: Use standard setup of \Feelpp toolboxes. It corresponds to a cfg + file and JSON file. These config files are present in the Github of feelpp. + \item[]{\Feelpp distributions} + \begin{itemize} + \item SIF image (Apptainer): feelpp:v0.111.0-preview.10-noble-sif (stored in the Github registry of \Feelpp) + \item Ubuntu package Jammy (Native) feelpp:v0.111.0-preview.10 + \end{itemize} + \end{itemize} +\item \textbf{Output Data:} The output includes export visualization files (mesh, + partitioning, temperature, elecric potential, current density, electric + field, ...), the time taken to perform each simulation step and + some integral quantities such a the total power dissipated by the magnet. +\item \textbf{Data Repository:} All inputdatasets are available in a unistra girder repository (collection HiFiMagnet, HL-31). +\end{itemize} + + + \begin{table}[h!] \centering { \setlength{\parindent}{0pt}