FAST Regressions

A FAST capability with applications beyond aircraft sizing.

Maxfield Arnson, Rawan Aljaber, Paul Mokotoff, Gökçin Çınar
software
PDF Code Video

Overview

Future Aircraft Sizing Tool (FAST) utilizes a large database of aircraft, the FAST Aerobase, in conjuction with the regression package to make predicitions about aircraft parameters before and during the sizing process. The regression package is available to use outside of FAST, and it can be used to explore relationships between aircraft parameters, make predictions about new designs, or to learn about probabilistic regressions. Additionally, the package can be used with custom datasets, assuming they are organized the same way as the FAST database. The IDEAS Lab recommends checking out the documentation in the regression package or referring to the videos on YouTube for easy use! The regression package is located in the github repository, and is included when forking.

Example regression output generated using FAST
Example regression output generated using FAST. This regression predicts fuselage length as a function of payload weight (which is used in the visualization package). It is the simplest of the regression package’s capabilities, predicting an output based on a single input.

Regression Performance

For a detailed explanation on the Gaussian Process Regressions used in this package, refer to our recent study:

  • Provisional Citation: Arnson, M. and Aljaber, R. and Cinar, G., “Predicting Conceptual Aircraft Design Parameters Using Gaussian Process Regressions on Historical Data,” in press, AIAA SciTech Forum, 2025.

This paper documents the data collection process, the regression methodology, and model tuning. Additionally, it compares the regressions’ performance to those in canonical aircraft design texts. The paper also establishes the relationship between the regression package and the FAST software, for example which modules utilize regressions and how they may be tuned.

FAST Results
Two FAST regressions on Operational Empty Weight (OEW). FAST 1 predicts outputs using data that was not used to train the model (test data) while FAST 2 predicts outputs using data that was used to train the model (recycles training data). Unsurprisingly, FAST 2 performs better than FAST 1.

Raymer Results
Regression on OEW using methodology from Daniel Raymer’s Aircraft Design: A conceptual Approach. This regression uses the same data set as the FAST regression in the above figure.

The two figures shown above provide examples of the analysis presented in our recent study. The full text explores additional regressions and compares performance to more expansive reference literature than shown here. Moreover, the paper provides statistical analysis of the error distributions for the aforementioned regressions and compares the distributions of various sources, including FAST.

Get Started with FAST

FAST is open-source and freely available under an Apache 2.0 license. To start using FAST, visit the GitHub repository here. The repository contains detailed documentation, including a comprehensive user guide to help you get started with the software.

If you use FAST in your research or projects, please cite our work as follows (note that these are provisional citations, awaiting final release in 2025):

  • Aircraft Sizing: Mokotoff, P., Arnson, M., & Cinar, G., “FAST: A Future Aircraft Sizing Tool for Electrified Aircraft Design,” in press, AIAA SciTech Forum, 2025.
  • Regressions: Arnson, M. and Aljaber, R. and Cinar, G., “Predicting Conceptual Aircraft Design Parameters Using Gaussian Process Regressions on Historical Data,” in press, AIAA SciTech Forum, 2025.
  • Visualization: Khailany, N., Mokotoff, P., & Cinar, G., “Aircraft Geometry and Propulsion Architecture Visualization for the Future Aircraft Sizing Tool (FAST)”, in press, AIAA SciTech Forum, 2025.
  • Historical Trends: Acar, H., Arnson, M., Tsai, M., & Cinar, G., ‘‘Historical Trends and Future Projections of Key Performance Parameters in Aircraft Design’’, under review, Journal of Aircraft, 2024.

FAST Video Tutorials Explore our comprehensive video tutorials to get the most out of the Future Aircraft Sizing Tool (FAST). Hosted by the IDEAS Lab researchers, these tutorials cover everything from installation to advanced features. Visit our YouTube channel for all videos.

Stay tuned for more tutorials and updates.

Watch the Playlist

Visit our YouTube channel for more tutorials.

Sign up for the FAST newsletter To stay informed about upcoming papers, new releases, and news about FAST, sign up for our newsletter here:

Funding and Acknowledgment: This work is sponsored by the NASA Aeronautics Research Mission Directorate and the Electrified Powertrain Flight Demonstration (EPFD) project, “Development of a Parametrically Driven Electrified Aircraft Design and Optimization Tool”. The IDEAS Lab would like to thank Ralph Jansen, Andrew Meade, Karin Bozak, Amy Chicatelli, Noah Listgarten, Dennis Rohn, and Gaudy Bezos-O’Connor from the NASA EPFD project for supporting this work and providing valuable technical input and feedback throughout the duration of the project.

Glenn Engineering and Research Support Contract (GEARS) Contract No. 80GRC020D0003

software fast
Maxfield Arnson
Maxfield Arnson
PhD Student and Graduate Research Assistant

Maxfield Arnson is a graduate student research assistant in the IDEAS Lab at the University of Michigan.

Rawan Aljaber
Rawan Aljaber
Undergraduate Research Assistant

Rawan is an undergraduate senior studying Aerospace Engineering at the University of Michigan. Rawan is working at the IDEAS Lab over the summer (2023) through the SURE (Summer Undergraduate Research in Engineering) program.

Paul Mokotoff
Paul Mokotoff
PhD Student and Graduate Research Assistant

Paul Mokotoff is a graduate student research assistant in the IDEAS Lab at the University of Michigan.

Gökçin Çınar
Gökçin Çınar
Assistant Professor of Aerospace Engineering