KHU-MASLAB/RecurDynPython

RecurDyn automation using Python and ProcessNet

RecurDyn ProcessNet with Python

Automates RecurDyn operations with ProcessNet and Python.
For initial setups, please follow this link (Korean).
For official tutorials provided by FunctionBay Inc., refer to this link (Korean).

Detailed instructions are provided in Tutorial.ipynb

For useful tips, check out Tips.md (Korean)

Setup

Set rdSolverDir to "<YOUR_RECURDYN_INSTALL_DIR>\Bin\Solver\RDSolverRun.exe" in GlobalVariables.py.

Simulate using GUI solver

Call analysis.doe_gui.RunDOE_GUI with arguments.
You can modify DOE scenario by editing line 56~65 in analysis/doe_gui.py`
This method is not parallelizable.

Arguments

• ModelFileDir: str Absolute path of model file (*.rdyn).
• TopFolderName: str Folder name to create at ModelFileDir.
  • Each of simulation results will be saved in this folder.
• NumCPUCores: int Number of CPU threads to use per simulation.
  • Must be one of [0(Auto),1,2,4,8,16].
• EndTime: float Simulation end time.
• NumSteps: int Number of time steps.

RunDOE_GUI(
    ModelFileDir=f"{os.getcwd()}/SampleModel.rdyn",
    TopFolderName="TestDOE_GUI",
    NumCPUCores=8,
    EndTime=1,
    NumSteps=100,
)

Simulate using batch solver

This method is far more stable and parallelizable compared to GUI solvers.
It is highly recommended to run DOEs using batch solvers, especially you're handling large, complex model.

Call analysis.doe_batch.RunDOE_Batch with arguments.
You can control DOE scenario by editing line 59~67 in analysis/doe_batch.py
This method is parallelizable, but consumes corresponding number of RecurDyn licenses.

Arguments

• ModelFileDir: str Absolute path of model file (*.rdyn).
• TopFolderName: str Folder name to create at ModelFileDir.
  • Each of simulation results will be saved in this folder.
• NumCPUCores: int Number of CPU threads to use per simulation.
  • Must be one of [0(Auto),1,2,4,8,16].
• EndTime: float Simulation end time.
• NumSteps: int Number of time steps.
• NumParallelBatches: int Number of parallelized DOE runners (*.bat) to create.
  • The total number of simulations of your DOE will be splited by NumParallelBatches. For example, if you define DOE with 100 simulations and set this argument to 4, RunDOE_Batch will configure 4 parallelized DOE runners with each of them containing 25 simulations.
• NumBatRunsOnThisPC: int Number of runners to immediately execute on your current machine. Defaults to NumParallelBatches. Value should be within range of [0, NumParallelBatches].
  • This argument is configured to run DOE on multiple machines. Comprehensively, if you set NumParallelBatches to 10 and set NumBatRunsOnThisPC to 3, only the first 3 runners (*.bat) are executed immediately on current machine. You can transfer rest of the 7 runners with corresponding subfolders (which contains *.rmd and *.rss + $\alpha$ files) in ModelFileDir to other machines and execute them by hand. In this case, you need additional processing to modify RecurDyn solver path defined in runner files.

RunDOE_Batch(
    ModelFileDir=f"{os.getcwd()}/SampleModel.rdyn",
    TopFolderName="TestDOE_Batch",
    NumCPUCores=8,
    EndTime=1,
    NumSteps=100,
    NumParallelBatches=5,
)

Export data from results using analysis.export_data.rplt2csv

Numeric simulation results are stored in *.rplt format.
Variable names should be exactly the same to the ones in the *.rplt.
To explicitly check variable names, simply import *.rplt file on RecurDyn GUI. Variables to be exported are defined in GlobalVariables.GlobVar.DataExportTargets.

rplt2csv(f"{os.getcwd()}/TestDOE_Batch")

The function will recursively scan for all *.rplt files in the argument directory, and export variables in DataExportTargets in *.csv format.