Deep Gaussian process (DGP) models are multi-layered hierarchical generalizations of the well-known Gaussian process (GP) models widely used to construct surrogate models of aerodynamic quantities of interest. Combining the desirable features of GP models and deep neural networks (DNN), DGP models are known to perform well when training data is scarce and the behavior of the system response is highly non-stationary. In this paper, the performance of DGP models is evaluated against GP models. Detailed comparisons are made and conclusions are drawn in terms of training time, data requirements, predictive error, and robustness to choice of training design of experiments, among other metrics. Additionally, sensitivity and scalability analyses are conducted for the DGP models to evaluate their usability. Finally, an adaptive construction of both models is presented, where the models are built sequentially by selecting points that maximize posterior variance. Several experiments are conducted with canonical test functions at varying input dimensions and a viscous transonic airfoil test case at 42 input dimensions. The experiments suggest that DGP models outperform traditional GP models in terms of accuracy but incur higher computational costs for training.
Ram Air Turbines (RATs) are small-scale wind powered turbines installed in commercial and military aircraft to generate power for in-flight hydraulics and electronics, in the event of total power failure. There is a need for efficient design of RATs that maximizes the power by minimizing weight and cost. The Lifting Line (LL) theory is applied towards the aerodynamic analysis of a 2-bladed RAT and the approach is validated against experimental data. The goal of this paper is to demonstrate the feasibility of the LL theory as an efficient tool for conceptual design of RATs. First, the methodology is validated by reproducing an existing RAT blade design for a known operating condition. Next, off-design conditions are explored for a design point. The validation study demonstrated that the LL theory can predict aerodynamic performance of RATs within 20% up to the transonic regime. Given the simple computational implementation, cheap computational cost and observed level of accuracy the LL theory appears to be an attractive approach towards the conceptual design and design space exploration of RATs. Thus, the paper also investigates the aerodynamic design capabilities of the LL theory, given some constraints and operating condition.
The need for more efficient engines to meet the next generation NASA N + 2 and N + 3 emission and fuel burn targets [1] for commercial transport vehicles can be addressed with Ultra High Bypass Ratio turbofan engines [2,3]. Bypass ratio for under-wing mounted engines is constrained by ground clearance and landing gear weight. An over-wing installation of the engine eliminates these constraints and comes with the added benefit of fan and/or jet noise shielding by the wing. However, wing-nacelle interference is exacerbated, resulting in an increased transonic drag penalty. On the other hand, past studies [4,5] have reported that upon identifying a suitable installation location and optimizing the wing and nacelle outer mold lines, superior aerodynamic performance can be obtained compared to the corresponding under-wing baseline. Therefore, aerodynamic shape optimization (ASO) of the engine installed wing is critical in order to realize the full potential of the Over-Wing Nacelle (OWN) aerodynamic performance.Key aspects of the OWN concept are (i) jet scrubbing on the wing upper surface leading to aerodynamic interference in the form of powered lift [6,7] for configurations where the engine is placed at the wing leading edge and (ii) engine inflow interference that can lead to losses in inlet pressure recovery for a trailing edge configuration. To account for such phenomena in the optimization problem, the engine has to be modeled under power-on conditions. This implies imposing thermodynamic boundary conditions on the engine fan, and bypass & core nozzle inlet faces in the CFD model. ASO with fixed powered boundary conditions might not account for the impact of the propulsion system on the airframe and vice versa. For instance, an optimized wing shape leads to improved airframe drag that lowers the required engine thrust. Modifying the engine thrust however may affect the flow-field over the wing and nacelle, modifying the airframe drag in return. A previous study [8] found statistically significant coupling between the airframe aerodynamics and the propulsion system, but the strength of the coupling, measured in terms of differences in the aerodynamic coefficients calculated with and without coupling, has not been quantified before. The goal of this paper is to address this gap, and in doing so hopes to answer the question: is Multidisciplinary Design Analysis and Optimization (MDAO) necessary for an OWN trailing edge configuration, or is a decoupled approach, where ASO is conducted with the engine at a fixed power setting, sufficient?This paper combines Reynolds Averaged Navier Stokes (RANS) analysis and a low-fidelity engine cycle model to perform a multidisciplinary analysis (MDA) of the OWN configuration at different angles of attack. A methodology to properly account for interdisciplinary coupling and solve the resulting MDA problem is presented. The Common Research Model (CRM) [9,10] wing and nacelle are used as the baseline geometry for this analysis. The CRM nacelle is modified by extracting a longitudinal...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.