A comparative study of uncertainty quantification methods in gust response analysis of a Lift-Plus-Cruise eVTOL aircraft wing

Bingran Wang, Michael Warner, Aoran Tian, Luca Scotzniovsky and John T. Hwang · 2025

Read A comparative study of uncertainty quantification methods in gust response analysis of a Lift-Plus-Cruise eVTOL aircraft wing by Bingran Wang, Michael Warner, Aoran Tian, Luca Scotzniovsky and John T. Hwang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-03964.txt.

Introduction

Bingran Wang and four co-authors study how random wind gusts affect the wing of an electric aircraft design called Lift-Plus-Cruise. Their work appeared in twenty twenty-five. Gusts and flight conditions are never exact, so the team asks how much that doubt changes the wing's response.

The authors couple an air flow solver with a shell model of the wing structure. They then compare five ways of measuring risk, among them Monte Carlo sampling and polynomial chaos. Each method estimates tip movement and strain energy. Results show wide spread in the outputs even when inputs vary only a little. Which method works best depends on which risk measure a reader cares about.

Catalog record

CREATORBingran Wang, Michael Warner, Aoran Tian, Luca Scotzniovsky and John T. Hwang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-03964

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A comparative study of uncertainty quantification methods in gust response analysis of a Lift-Plus-Cruise eVTOL aircraft wing Bingran Wang, Michael Warner, Aoran Tian, Luca Scotzniovsky and John T. Hwang Abstract Wind gusts, being inherently stochastic, can significantly influence the safety and performance of aircraft. This study investigates a three-dimensional uncertainty quantification (UQ) problem to explore how uncertainties in gust and flight conditions affect the structural response of a Lift-Plus-Cruise eVTOL aircraft wing. The analysis employs an unsteady aeroelastic model with a one-way coupling between a panel method aerodynamic solver and a shell analysis structural solver to predict the wing’s response under varying conditions. Additionally, this paper presents a comparative evaluation of commonly used non-intrusive UQ methods, including non-intrusive polynomial chaos, kriging, Monte Carlo, univariate dimension reduction, and gradient-enhanced univariate dimension reduction. These methods are assessed based on their effectiveness in estimating various risk measures—mean, standard deviation, and 95th percentile—of critical structural response outputs such as maximum tip displacement and average strain energy. The numerical results reveal significant variability in the structural response outputs, even under relatively small ranges of uncertain inputs. This highlights the sensitivity of the system to uncertainties in gust and flight conditions. Furthermore, the performance of the implemented UQ methods varies significantly depending on the specific risk measures and the quantity of interest being analyzed. 1 Introduction The accurate prediction of gust loads is critical for aircraft design and certification, as gust-induced aeroelastic responses can significantly impact the safety and performance of an aircraft. Gust loads arise from atmospheric turbulence and are particularly challenging to model due to their dynamic and stochastic nature. These loads must be carefully analyzed to ensure compliance with regulatory requirements and to maintain the safety margins necessary for aircraft certification. Traditional deterministic approaches often fail to capture the inherent uncertainties in gust characteristics and flight conditions. This underscores the importance of incorporating uncertainty quantification (UQ) into aeroelastic analysis to better understand and predict the variability in structural responses [ 1 ] . UQ is a crucial tool in engineering, which enables engineers to evaluate how input uncertainties influence the system’s outputs. Its applications span diverse fields, including weather forecasting [ 2 , 3 ] , machine learning [ 4 ] , structural analysis [ 5 , 6 ] , and aircraft design [ 7 , 8 , 9 ] . Common non-intrusive UQ methods include polynomial chaos, kriging, and Monte Carlo. Kriging, also known as Gaussian process regression, constructs a surrogate response surface from input-output data, enabling efficient model evaluations for reliability analysis [ 10 , 11 ] . The polynomial chaos method represents quantities of interest (QoIs) using orthogonal polynomials tailored to the input distributions, leveraging smoothness in the random space for rapid convergence with either a sampling or an integration approach [ 12 , 13 , 14 ] . Both kriging and polynomial chaos are effective for low- to medium-dimensional problems but face scalability challenges as dimensionality increases, limiting their direct application in high-dimensional UQ problems. Conversely, the Monte Carlo method uses random sampling to estimate the risk measures of the QoIs. This method is not only easy to implement but also well-suited for high-dimensional problems due to their convergence rate being independent of input dimensionality. Several approximation methods are also widely used to reduce computational costs without significant loss of accuracy. These include active subspace [ 15 ] , first-order Taylor expansion [ 16 ] , univariate dimension reduction (UDR) [ 15 ] , and gradient-enhanced dimension reduction [ 17 ] methods. Such methods aim to strike a balance between computational efficiency and predictive accuracy by reducing the problem dimension or creating an approximation function for the model, making them practical for large-scale or high-fidelity engineering problems. Each UQ method has its strengths and limitations, with performance varying depending on the specific problem setup and characteristics. Consequently, selecting the most appropriate method requires careful consideration of the problem’s dimensionality, complexity, and desired accuracy. In previous work on aircraft gust response analysis, Pettit et al. [ 18 ] conducted UQ on an aerodynamic model with uncertain gust velocity spectra using the NIPC method. Similarly, Cook et al. [ 19 ] applied UQ to aeroelastic analysis, accounting for structural and flight condition uncertainties, also leveraging NIPC. Xiang’s thesis [ 20 ] extended this research by performing time-domain structural optimization of eVTOL wings under gust loads. However, UQ problems involving unsteady aeroelastic analysis remain underexplored, particularly when considering uncertainties in gust conditions. This work focuses on a Lift-Plus-Cruise (LPC) eVTOL aircraft design. The development of eVTOL aircraft has gained significant attention in academia due to their potential to transform next-generation transportation systems. Recent studies on eVTOL design have emphasized multidisciplinary design optimization [ 21 ] , trajectory optimization [ 22 ] , and acoustic modeling [ 23 ] . Despite these advancements, the impact of uncertain gust and flight conditions on eVTOL performance remains an open question. In this paper, we formulate a three-dimensional UQ problem that considers uncertainties in gust and flight conditions. An unsteady aeroelastic simulation model is used to predict the structural response of the wing under varying conditions. Specifically, the aeroelastic simulation integrates a mid-fidelity panel method aerodynamic solver with a high-fidelity shell-based structural solver using a one-way coupling approach. The computational model is implemented on a graph-based modeling framework, Computational System Design Language (CSDL) [ 24 ] . CSDL enables automatic differentiation [ 25 ] and computational graph transformation [ 26 , 27 ] to reduce the computational cost for UQ and optimization. The QoIs in this UQ problem are the maximum tip displacement and average strain energy. To estimate the three risk measures—mean, standard deviation, and 95th percentile—of the QoIs, we implement five UQ methods: non-intrusive polynomial chaos, kriging, Monte Carlo, univariate dimension reduction, and gradient-enhanced univariate dimension reduction. The numerical results reveal significant variability in both QoIs—maximum tip displacement and average strain energy—even under small variations in the uncertain gust and flight conditions. This highlights the sensitivity of the structural response to these uncertainties and underscores the importance of incorporating UQ in gust response analysis for aircraft design. Among the UQ methods compared, kriging demonstrated superior performance in estimating risk measures for maximum tip displacement, as its interpolation-based approach effectively handles highly nonlinear response surfaces. In contrast, NIPC excelled in estimating risk measures for average strain energy, as its underlying function exhibits smoother behaviour. In the meantime, UDR and GUDR are cheap alternative methods for estimating various risk measures, while Monte Calro underperformed due to the low dimensionality of the UQ problem. This paper is organized as follows. Section 2 presents detailed description on the aeroelastic simulation model and the UQ problem. Section 3 shows the numerical results of the implemented UQ methods. Section 4 summarizes the work and offers concluding thoughts. 2 Problem set-up In this paper, we address an uncertainty quantification problem related to the structural response of a Lift-Plus-Cruise eVTOL aircraft wing under stochastic flight and gust conditions. The Lift-Plus-Cruise (LPC) concept is a well-established design for eVTOL aircraft, characterized by dedicated rotors for vertical takeoff and landing, as well as fixed wings and a rear pusher propeller for cruise flight. The Lift-Plus-Cruise concept is illustrated in Fig. 1 . To investigate

Read the complete text