An efficient surrogate model for emulation and physics extraction of large eddy simulations

Reproducibility materials for An efficient surrogate model for emulation and physics extraction of large eddy simulations by Simon Mak, Chih-Li Sung, Xingjian Wang, Shiang-Ting Yeh, Yu-Hung Chang, V. Roshan Joseph, Vigor Yang & C. F. Jeff Wu

Author Contributions Checklist Form

Data

Abstract (Mandatory)

The full data consists of flow simulations for the x-, y- and z-velocities, pressure, density and temperature of a swirl injector system. The overall memory requirement for the full dataset is over 100Gb, which is too large to submit alongside our manuscript. As requested by the AE, we have included a smaller, intermediate dataset consisting of extracted CPOD coefficients, which can be used to train the proposed emulator model.

Availability (Mandatory)

The full simulation data is too large to accompany the manuscript, so we have included an intermediate dataset with extracted physics coefficients. We are working towards making the full simulation data available in an online repository.

Description (Mandatory if data available)

• Permissions: Dataset simulated using internal code
• Licensing information: -----
• Link to data: https://bitbucket.org/chihli/injectoremulation
• Data provenance, including identifier or link to original data if different than above: -----
• File format: .RData
• Metadata (including data dictionary): -----
• Version information: -----

Optional Information (complete as necessary)

Code

Abstract (Mandatory)

The submitted manuscript is accompanied by a .zip file containing the intermediate data, a package source file “InjectorEmulationHelper.tar.gz”, and the code used for generating results. This code is written in R and C++, with Rcpp providing the interface between the two languages.

Description (Mandatory)

• How delivered: R package and code
• Licensing information: MIT License
• Link to code/repository: https://bitbucket.org/chihli/injectoremulation
• Version information: master+c72b3fb

Optional Information (complete as necessary)

• Hardware requirements: 10 nodes, with 20 cores requested in each node; NFS shared file system; 250 GB of storage space; 128 GB of RAM
• Supporting software requirements (e.g., libraries and dependencies, including version numbers)
  • Tecplot 360 2016
  • R (>= 3.2.4):
    • Platform: x86_64-pc-linux-gnu (64-bit)
    • Running under: Red Hat Enterprise Linux Client release 6.7 (Santiago)
    • Packages: parallel, stats, graphics, grDevices, utils, datasets, methods, base, InjectorEmulationHelper (>= 0.1.0), doMPI (>= 0.2.2), Rmpi (>= 0.6-5), shape (>= 1.4.2), colorRamps (>= 2.3), SLHD (>= 2.1-1), huge (>= 1.2.7), MASS (>= 7.3-45), igraph (>= 1.0.1), lattice (>= 0.20-33), Matrix (>= 1.2-4), nloptr (>= 1.0.4), Rarpack (>= 0.10-0), biganalytics (>= 1.1.12), biglm (>= 0.9-1), DBI (>= 0.3.1), bigmemory (>= 4.5.8), bigmemory.sri (>= 0.1.3), snowfall (>= 1.84-6.1), snow (>= 0.4-1), doParallel (>= 1.0.10), iterators (>= 1.0.8), foreach (>= 1.4.3), Rcpp (>= 0.12.3), magrittr (>= 1.5), grid (>= 3.2.4), codetools (>= 0.2-14), compiler (>= 3.2.4)

Instructions for Use

Reproducibility (Mandatory)

• What is to be reproduced (e.g., "All tables and figure from paper", "Tables 1-4”, etc.) Figures 6, 7, 11 and 14.
• How to reproduce analyses (e.g., workflow information, makefile, wrapper scripts) The above figures can be reproduced by running “example2.R”. The R package “InjectorEmulationHelper” needs to be installed from the source package file “InjectorEmulationHelper.tar.gz”. The reproduced results for Figures 6, 7 and 11 are outputted to the directories “output2/POD_expansion/6”, “output2/validation/comparison” and “output2/validation/CIwidth”, while the results for Figure 14 are printed on the console. These results are then fed into Tecplot (a post-processing software for visualizing engineering simulations) to generate the figures in the paper.

Replication (Optional)

• How to use software in other settings (or links to such information, e.g., R package vignettes, demos or other examples)

• Because the full simulation data is too large to accompany the manuscript, we have provided a simple example in “example1.R”, which uses a simulated toy dataset to illustrate the full emulation procedure. The raw data can be changed by following the same format in the directory “data1/rawdata folder”: first three columns correspond to the x-, y - and z-coordinates, and other columns are the responses. The experimental design can also be changed from the file “data1/DoE.csv”. The geometry and raw data for testing validation cases can also be modified in the file “data1/newX.csv” and the folder “data1/testdata”, respectively.

Notes

As requested by the AE, we have included an intermediate dataset (the extracted physics coefficients) for reproducing numerical results in the paper. We are working towards making the full simulation data (over 100Gb in storage) publicly available in an online repository.