optvl 1.0.0

A modified version of Mark Drela and Harold Youngren's famous AVL code with a direct Python interface and gradient based optimization support.

AVL + python + optimization = OptVL

Documentation

OptVL is a modified version of Mark Drela and Harold Youngren's famous AVL code with a python-wrapper and AD derivative routines for gradient-based optimization. The python wrapper allows one to easily conduct large parameter sweeps with a vortex lattice method or to include a vortex lattice method into a larger model. Together with the additional derivative routines, OptVL can be added to an optimzation loop for design refinement by a gradient-free or gradient-based optimizer. On top of this OptVL also has the ability to view the geometry and pressure distribution in python, Paraview, or Tecplot, which provides portable post-processing.

key features

• specify and modify the geometry from python
• compute derivatives of total forces with respect to geometric variables
• derivatives of stability derivatives
  • needed for neutral point constraint

Installation

The best way to get OptVL is to install it through pip

pip install optvl

Windows, macOS, and Linux are all supported!

Building locally

If you want to make OptVL locally then you have to clone the repository and use the following process.

In the root directory run

pip install . 

Basic usage

The API of OptVL was made to mirror the usage of AVL through its text interface. The user loads in a geometry file, adds constraints, and then executes analysis runs.

The AVL wrapper is implemented in the OVLSolver class. To use this wrapper, first one must initialize the OVLSolver object with a geometry file and optionally a mass file. After, the user can add constraints and then execute the run to generate data. Below is a basic example of this workflow.

from optvl import OVLSolver
import numpy as np

ovl = OVLSolver(geo_file="aircraft.avl")
ovl.set_constraint("alpha", 0.00)

# control surface names from geometry file
ovl.set_constraint("Elevator", 0.00, con_var="Cm pitch moment")
ovl.set_constraint("Rudder", 0.00, con_var="Cn yaw moment")

ovl.set_parameter("Mach", 0.3)

# This is the method that acutally runs the analysis
ovl.execute_run()

print("----------------- alpha sweep ----------------")
print("   Angle        Cl           Cd          Cdi          Cdv          Cm")
for alpha in range(10):
    ovl.set_constraint("alpha", alpha)
    ovl.execute_run()
    run_data = ovl.get_total_forces()
    print(
        f' {alpha:10.6f}   {run_data["CL"]:10.6f}   {run_data["CD"]:10.6f}   {run_data["CDi"]:10.6f}   {run_data["CDv"]:10.6f}   {run_data["CM"]:10.6f}'
    )

print("----------------- CL sweep ----------------")
print("   Angle        Cl           Cd          Cdff          Cdv          Cm")
for cl in np.arange(0.6,1.6,0.1):
    ovl.set_trim_condition("CL", cl)
    ovl.execute_run()
    run_data = ovl.get_total_forces()
    alpha = ovl.get_parameter("alpha")
    print(
        f' {alpha:10.6f}   {run_data["CL"]:10.6f}   {run_data["CD"]:10.6f}   {run_data["CDi"]:10.6f}   {run_data["CDv"]:10.6f}   {run_data["CM"]:10.6f}'
    )

Parameter sweep example

taper ratio sweep

Optimization example

optimize twist distribution

License

As stated, OptVL includes a modified version of the source code of AVL. AVL is distributed under the GLP version 2 license. As a consequence, OptVL is also distributed under a GLP license via the copyleft principle of the original license. Although, here we update the GLP license to version 3 as allowed by the original license.

This software includes portions of NetLib's LAPACK library. LAPACK is a software package provided by University of Tennessee, University of California Berkeley, University of Colorado Denver and NAG Ltd under the modified BSD license. See the license file in src/lapack/LICENSE.txt for more details.