T. Macdonald

SUAVE: An Open-Source Environment for Multi-Fidelity Conceptual Vehicle Design

SUAVE, a conceptual level aircraft design environment, incorporates multiple information sources to analyze unconventional configurations. Developing the capability of producing credible conceptual level design conclusions for futuristic aircraft with advanced technologies is a primary directive. Many software tools for aircraft conceptual design rely upon empirical correlations and other handbook approximations. SUAVE proposes a way to design aircraft featuring advanced technologies by augmenting relevant correlations with physics-based methods. SUAVE is constructed as a modular set of analysis tools written compactly and evaluated with minimal programming effort. Additional capabilities can be incorporated using extensible interfaces and prototyped with a top-level script. The flexibility of the environment allows the creation of arbitrary mission profiles, unconventional propulsion networks, and right-fidelity at right-time discipline analyses. This article will first explain how SUAVE’s analysis capabilities are organized to enable flexibility. Then, it will summarize the analysis strategies for the various disciplines required to evaluate a mission. Of particular interest will be the construction of unconventional energy networks necessary to evaluate configurations such as hybrid-electric commercial transports and solar-electric unmanned aerial vehicles (UAVs). Finally, verification and validation studies will be presented to demonstrate the capabilities of SUAVE, including cases for conventional and unconventional vehicles. While some of these cases will be optimized results, discussion of SUAVE’s interface with optimization will be reserved for a future publication.

SUAVE: An Open-Source Environment for Conceptual Vehicle Design and Optimization

SUAVE, a conceptual level aircraft design environment, incorporates multiple information sources to analyze unconventional configurations. Developing the capability to produce credible conceptual level design conclusions for futuristic aircraft with advanced technologies is a primary directive. This work builds upon previous work where SUAVE analyzed aircraft to show how SUAVE may be integrated into external packages to optimize aerospace vehicles. In the context of optimization, SUAVE operates as a “black-box” function with multiple inputs and multiple outputs. Several convenient functions are provided to enable connecting the optimization packages to SUAVE more easily. Assuming an optimization algorithm is minimizing an objective subject to constraints by iteratively modifying input variables, SUAVE’s code structure is general enough to be driven from a variety of optimization packages. To this point, connections to PyOpt and SciPy have been integrated into SUAVE. We present results for a multi-mission regional aircraft, a family of UAVs and a tradeoff between noise and fuel burn on a large single-aisle aircraft. These designs show the immense amount of flexibility and diversity that SUAVE can handle. This includes various levels of fidelity. While SUAVE is setup from the beginning to handle multi-fidelity analysis, further study is necessary to integrate multiple fidelity levels into a single vehicle optimization.

SUAVE: An Open-Source Environment Enabling Unconventional Vehicle Designs through Higher Fidelity

SUAVE is a conceptual level aircraft design environment that incorporates multiple information sources to analyze unconventional configurations. This work incorporates higherfidelity tools to build upon previous efforts where SUAVE analyzed and optimized several types of aircraft using low-fidelity methods. This is done in an automated way that incorporates three external programs. The first is OpenVSP, which is used for geometry creation, area calculation, and surface meshing. The second is Gmsh, which uses these surface meshes to create volume meshes. The third is SU2, which is used to run Euler CFD simulations. Wetted areas from OpenVSP and lift from SU2 is used to enhance SUAVE’s aerodynamic analyses. We present results for a verification case with the Onera M6 wing, then present mission results with a conventional narrow-body airliner, a supersonic jet, and a blended wing body.

SUAVE: An Open-Source Environment Enabling Multi-Fidelity Vehicle Optimization

SUAVE is a conceptual level aerospace vehicle design environment that allows a user to incorporate new methods and information sources to analyze both conventional and unconventional configurations. This paper builds upon previous works where SUAVE analyzed and optimized several types of aircraft using low-fidelity methods, and also incorporated links with higher-fidelity tools. Here we demonstrate SUAVE’s use in the multi-fidelity optimization of unconventional designs. The objective of this type of framework is to enable high performance while working with constrained computational resources. This capability will be demonstrated through the use of additive correction surrogates and trust region model management, with SUAVE managing the levels of fidelity according to these methods. Two different test cases are optimized here. We present results for a supersonic transport with varying wing area and aspect ratio, and a blended-wing-body aircraft with varying planform values subject to stability constraints. Both are analyzed at the cruise condition with two levels of analysis fidelity.

Sizing Methods for Aircraft of Variable Propulsion System Complexity

This paper compares the computational cost and robustness of a number of different algorithms for the sizing and optimization of aircraft. In particular, three classes of aircraft will be investigated, each with a different level of propulsion system complexity. The first is a conventional turbofan-powered aircraft. A lithium-air battery-powered aircraft, incorporating both energy and power constraints is second. Third is an aluminum-air/lithium-ion battery powered aircraft, which switches to different energy systems depending on the overall propulsion system power requirements. Several different aircraft design problem formulations will be considered for each of these aircraft. Sizing loops as well as determining constraints such as fuel margin by exposing them to the optimizer will both be explored. Results suggest that integrating surrogate models to inform the sizing loop leads to substantially fewer function evaluations than a simple successive substitution, while maintaining a relative insensitivity to a poorly-informed initial guess that optimizer-sizing algorithms lack.

Increasing Conceptual Design Fidelity with Prebuilt Airfoil Databases

Due to time and monetary constraints, the conceptual design process for new aircraft is typically limited to inexpensive low-fidelity methods. Unfortunately, these methods generally only provide the fidelity necessary to confidently go forward with a new design when they have been fitted to previous aircraft of the same type. The conceptual design process has difficulty producing accurate estimates of performance in unconventional aircraft. To remedy this, new methods must be developed that can produce higher fidelity results without a significant increase in cost. Here we present a method to do this in the aerodynamic analysis. First, we generate a large database of airfoils that are optimized using CFD to minimize drag in a variety of flight conditions that are likely to be seen on the wing of a subsonic transport aircraft. We then present methods to integrate these airfoils along an arbitrary wing to find the minimum possible drag. This process is meant to mimic the ”expert designer” often pointed to in current correlation equations for determining the drag of a wing. Since the airfoils are generated before hand in a manner that is easy to parallelize, this process is not substantially more expensive than other methods for determining the aerodynamic properties of an aircraft at the conceptual level.