Airfoil Optimizer

Tool to optimize an airfoil for minimum drag at a given design lift coefficient using a genetic algorithm.

Airfoil Specification

An airfoil is specified by a mean chord line, thickness distribution, and trailing edge thickness. The mean chord line and thickness distribution are parameterized using the Class-Shape Transformation (CST) technique. The number of CST coefficients used to parameterize the mean chord line and thickness distributions can be chosen by the user.

Usage

Optimization cases are specified in the Runfile. Each line in this file is should to be a comma-separated list of positional argument values (*args) and followed by /or key-value pairs (**kwargs) and corresponds to a single optimization problem to be solved. Acceptable key-value pairs correspond to the inputs taken by the run function specified in the runner.py script:

def run(cl, n_c, n_t, b_c=8, b_t=8, b_te=8, gen=100,
        fix_te=True,
        constrain_thickness=True, constrain_area=True, constrain_moment=True,
        cm_ref=None, seed=None, n_proc=28,
        report=False,
        results_output_folder=None):
    """
    Solve the specified optimization problem and handle reporting of results.

    Parameters
    ----------
    cl : float
        Design lift coefficient
    n_c, n_t : int
        Number of CST coefficients for the chord line and thickness distribution, respectively
    b_c, b_t, b_te : int, optional
        Number of bits to encode each of the CST coefficients of the chord line/thickness distribution, and TE thickness
        8 bits each by default.
    gen : int, optional
        Number of generations to use for the genetic algorithm. 100 by default
    fix_te : bool, optional
        True if the trailing edge thickness should be fixed. True by default
    constrain_thickness, constrain_area, constrain_moment : bool, optional
        True if the thickness, area, and/or moment coefficient should be constrained, respectively. All True by default
    cm_ref : float, optional
        If constrain_moment is True, this will be the maximum (absolute) moment coefficient. If None, initial Cm is used
    seed : int, optional
        Seed to use for the random number generator which creates an initial population for the genetic algorithm
    n_proc : int, optional
        Number of processors to use to evaluate functions in parallel using MPI. 28 by default
    report : bool, optional
        True if the results should be reported via email. False by default
    results_output_folder : str, optional
        Name of the shared folder in which to store results. By default, an ISO formatted UTC timestamp will be used.
    """
    ...

For example, a line in the Runfile could be:

1.0, 3, 3, 8, 8, gen=100, fix_te=True, constrain_moment=False, n_proc=16

This defines an airfoil optimization at a design lift coefficient of 1.0, using 3 CST coefficients for the chord line and thickness distribution, 8 bits to encode each of the CST coefficients, and 100 generations for the genetic algorithm , while keeping the TE thickness fixed and not constraining the moment coefficient, using 16 processors.

After specifying optimization problems in the Runfile, run the command

docker-compose up --build

from the root directory of the repository to solve all specified problems in order.

Results

The results of each optimization run are also written to persistent files, available on the host machine in the share directory relative to the root directory of this repository. By default, each optimization problem will write to its own sub-folder under the share folder with a name based on the ISO formatted UTC date/time at the time the optimization finished.

Additionally, if report=True is specified, the results of an optimization run will be send to an email address of the user's choosing. For this to work, the user should specify an STMP server hostname and port, username, password, and receiver address in stmp_settings.conf.