パラメトリック研究 (`foamlib.preprocessing`)

Functionality contributed by Henning Scheufler.

この機能により、ユーザーはOpenFOAMでパラメトリックスタディを設定・実行できます。これは複数のシミュレーションを異なるパラメータで実行し、結果への影響を分析する手法です。スタディでは、異なる実行間で特定のフィールドやソルバー設定を変更するように構成できます。

追加の依存関係のインストール

パラメトリックスタディ機能を使用するには、 foamlib.preprocessing モジュール用の追加依存関係をインストールする必要があります。以下のコマンドを実行することでインストールできます:

pip install foamlib[preprocessing]

Overview

The general workflow for a parametric study in OpenFOAM using foamlib is as follows:

        graph LR
  A[Template Case] --> B[Generate Case Variants]
  C[Modify Parameters] --> B
  B --> D[Run Simulation]
  D --> E[Post-process and Analyze]

A template case is created with the necessary configuration files. The user can then generate multiple case variants by modifying specific parameters, such as initial conditions, boundary conditions, or solver settings. Each variant is run independently, and the results are collected for post-processing and analysis.

The general concept of the parametric study is that template cases are copied to a new folder, and the parameters are modified in the copied case. This requires the definition of the following parameters:

template_case: The path to the template case that will be copied and modified.
output_folder: The folder where the modified cases will be stored.
case_name: The name of the case that will be created.
instructions: A list of the file and the key entry that needs to be modified in the template case.
value: the value for each instruction
case category: A category for the case, which can be used to group cases together for easier post-processing.

Multiple parammetric study generators are avaible and describe below in detail.

CSV Generator

The csv generator create the parametric study based on a CSV file that contains the parameters to be varied. This generator reads the CSV file, extracts the parameters, and generates multiple case variants by modifying the specified fields in the template case.

from foamlib.preprocessing.parameter_study import csv_generator

# Example usage
csv_generator(
    csv_file="path/to/your/parameters.csv",
    template_case="path/to/template/case",
    output_folder="path/to/output/folder"
).create_study()

This simple generator specifies the above requirements in the csv file, where the instruction (file and key name (here: NX,NY and someModel)) is defined in the system/simulationParameters file. The case_name and the category will be defined as additional columns in the CSV file.

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox           |
|  \\    /   O peration     | Version:  plus                                  |
|   \\  /    A nd           | Web:      www.OpenFOAM.com                      |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    location    "system";
    object      simulationParameters;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

NX 23;
NY 8;

someModel SomeModelName;


// ************************************************************************* //

The file can be included in every OpenFOAM dictionary by adding the following line and can be referenced with a dollar sign in the dictionary:

#include "system/simulationParameters"

blocks
(
    hex (0 1 2 3 4 5 6 7) ($NX $NY 1) simpleGrading (1 1 1)
);

The csv file needs to be contain a case_name column and the parameters to be varied. Additionally, columns can be specified to categorize the cases: Resolution, and Model. The generator will create a case for each row in the CSV file. This is necessary to simplify the post-processing. The csv file should look like this:

case_name	NX	NY	someModel	Resolution	Model
case_001	100	200	modelA	coarse	k-epsilon
case_002	150	300	modelB	fine	Spalart-Allmaras

Grid Parameter sweep

The grid parameter sweep generator allows users to define a set of parameters and creates all combinations of these parameters to generate multiple cases. In the example below, the grid resolution and initial height are varied across multiple cases.

The most important class for the grid generator is the GridParameter. This class encapsulates both the FoamDictInstruction and the CaseParameter. The FoamDictInstruction class is used to specify which files and keys in the OpenFOAM case should be modified.

from foamlib.preprocessing.grid_parameter_sweep import CaseParameter, GridParameter
from foamlib.preprocessing.of_dict import FoamDictInstruction

# helper function
def grid_parameters(scale) -> list[int]:
    return [
        int(23 * scale),
        int(8 * scale),
        int(19 * scale),
        int(42 * scale),
        int(4 * scale),
    ]


grid = GridParameter(
    parameter_name="grid",
    # generate 5 instructions in system/simulationParameters with the key1..5
    modify_dict=[
        FoamDictInstruction(
            file_name=Path("system/simulationParameters"),
            keys=[f"res{i}"],
        )
        for i in range(1, 6)
    ],
    parameters=[
        CaseParameter(
            name="coarse", values=grid_parameters(1) # return [23, 8, 19, 42, 4]
        ),
        CaseParameter(
            name="mid", values=grid_parameters(2) # return [46, 16, 38, 84, 8]
        ),
        CaseParameter(
            name="fine", values=grid_parameters(4) # return [92, 32, 76, 168, 16]
        )]
)

The modify_dict variable stores the instructions for the OpenFOAM dictionary that will be modified. The keys parameter specifies the keys in the dictionary that will be modified. The parameters variable stores the parameters that will be used to modify the keys in the dictionary. Each CaseParameter contains a name and a list of values that will be used to modify the keys in the dictionary.

from foamlib.preprocessing.grid_parameter_sweep import grid_generator

init_height = GridParameter(
    parameter_name="initHeight",
    modify_dict=[
        FoamDictInstruction(
            file_name=Path("system/simulationParameters"),
            keys=["initHeight"],
        )
    ],
    parameters=[
        CaseParameter(
            name="height_02", values=[0.2]
        ),
        CaseParameter(
            name="height_03", values=[0.3]
        ),
        CaseParameter(
            name="height_04", values=[0.4]
        )
    ],
)

study = grid_generator(
    parameters=[grid, init_height],
    template_case=template_case,
    output_folder=root / "Cases",
)

study.create_study(study_base_folder=root)

This code creates a parametric study that varies the grid resolution and initial height across multiple cases. The grid_generator function takes a list of GridParameter objects, which define the parameters to be varied, and generates all combinations of these parameters to create multiple cases. So, 3 times 3 cases will be generated, resulting in 9 cases in total. Each case will have a unique name based on the parameters used.

Convenience Functions for FoamDictInstruction

This section provides a reference to all functions available in the foamlib.preprocessing.system module.

foamlib.preprocessing.system.controlDict(keys: list[str]) → FoamDictInstruction

Alias for control_dict().

Deprecated since version 1.3.0: use control_dict() instead.

foamlib.preprocessing.system.control_dict(keys: list[str]) → FoamDictInstruction: Return the FoamDictInstruction for controlDict.

foamlib.preprocessing.system.fvSchemes(keys: list[str]) → FoamDictInstruction

Alias for fv_schemes().

Deprecated since version 1.3.0: use fv_schemes() instead.

foamlib.preprocessing.system.fvSolution(keys: list[str]) → FoamDictInstruction

Alias for fv_solution().

Deprecated since version 1.3.0: use fv_solution() instead.

foamlib.preprocessing.system.fv_schemes(keys: list[str]) → FoamDictInstruction: Return the FoamDictInstruction for fvSchemes.

foamlib.preprocessing.system.fv_solution(keys: list[str]) → FoamDictInstruction: Return the FoamDictInstruction for fvSolution.

foamlib.preprocessing.system.simulationParameters(keys: list[str]) → FoamDictInstruction

Alias for simulation_parameters().

Deprecated since version 1.3.0: use simulation_parameters() instead.

foamlib.preprocessing.system.simulation_parameters(keys: list[str]) → FoamDictInstruction: Return the FoamDictInstruction for simulationParameters.

パラメトリック研究 (foamlib.preprocessing)