High-fidelity Data Research Articles

Cost reduction for data acquisition based on data fusion: Reconstructing the surface temperature of a turbine blade

Turbine cooling is an effective way to improve the comprehensive performance and service life of gas turbines. In recent decades, there has been rapid growth in research into external cooling and internal cooling methods. As a result, there is a significant amount of experimental and numerical data. However, due to their multi-source nature, the datasets have different degrees of fidelity and different data structures, which hinder the effective use of the data. Besides, high-fidelity (HF) data often have high acquisition costs, which hinder their application in aerospace. A novel form of data fusion is introduced in this paper. We integrate multi-source data using special algorithms to produce more reliable data. A deep-learning neural network with the PointNet architecture is designed to establish two surrogate models: a high-fidelity model (HF model) trained by experimental data and a low-fidelity model (LF model) based on Reynolds-averaged Navier–Stokes simulation data. Both models predict results with less than 1% reference errors compared to their respective ground truth at most data points. In addition, we explore the role of transfer learning in multi-fidelity modeling. A fusion algorithm based on a Gaussian function and a weighted average strategy is proposed to combine the values from the HF model and the LF model. The presented results show that the fusion data are more accurate than computational fluid dynamics data, successfully meeting the goal of reducing the cost of data acquisition.

Real-Time Hybrid Testing of a Floating Offshore Wind Turbine Using a Surrogate-Based Aerodynamic Emulator

AbstractPhysical modeling of floating offshore wind turbines (FOWTs) is challenging due to the complexities associated with the simultaneous application of two different scaling laws, governing the hydrodynamic and aerodynamic loading on the structure. To avoid these issues, this paper presents a real-time hybrid testing (RTHT) strategy in which a feedback loop, consisting of an on-board fan and control algorithm, is utilized to emulate the aerodynamic forces acting on the FOWT system. Here, we apply this strategy to a 70th-scale IEA Wind 15 MW reference wind turbine mounted on a version of the VolturnUS-S platform. Unlike other similar methods, which directly simulate the aerodynamic loads for the fan’s control using an aerodynamic code running in parallel with the experiment, this example utilizes a surrogate model trained on numerical model data calculated in advance. This strategy enables high-fidelity numerical model data, or even physical data, to be included in the aerodynamic emulation, by removing the requirement for real-time simulation, and, therefore, potentially enables more accurate loading predictions to be used in the experiments. This paper documents the development of the real-time hybrid testing system in the Coastal Ocean And Sediment Transport (COAST) Laboratory at the University of Plymouth in the UK, including the hardware, software, and instrumentation setup, and demonstrates the power of the surrogate-based aerodynamic emulator based on numerical data calculated using OpenFAST.

The potential synergies between synthetic data and in silico trials in relation to generating representative virtual population cohorts

In silico trial methods promise to improve the path to market for both medicines and medical devices, targeting the development of products, reducing reliance on animal trials, and providing adjunct evidence to bolster regulatory submissions. In silico trials are only as good as the simulated data which underpins them, consequently, often the most difficult challenge when creating robust in silico models is the generation of simulated measurements or even virtual patients that are representative of real measurements and patients. This article digests the current state of the art for generating synthetic patient data outside the context of in silico trials and outlines potential synergies to unlock the potential of in silico trials using virtual populations, by exploiting synthetic patient data to model effects on a more diverse and representative population. Synthetic data could be defined as artificial data that mimic the properties and relationships in real data. Recent advances in synthetic data generation methodologies have allowed for the generation of high-fidelity synthetic data that are both statistically and clinically, indistinguishable from real patient data. Other experimental work has demonstrated that synthetic data generation methods can be used for selective sample boosting of underrepresented groups. This article will provide a brief outline of synthetic data generation approaches and discuss how evaluation frameworks developed to assess synthetic data fidelity and utility could be adapted to evaluate the similarity of virtual patients used for in silico trials, to real patients. The article will then discuss outstanding challenges and areas for further research that would advance both synthetic data generation methods and in silico trial methods. Finally, the article will also provide a perspective on what evidence will be required to facilitate wider acceptance of in silico trials for regulatory evaluation of medicines and medical devices, including implications for post marketing safety surveillance.

Transfer Learning of High-Fidelity Opacity Spectra in Autoencoders and Surrogate Models

Simulations of high energy density physics are expensive, largely in part for the need to produce nonlocal thermodynamic equilibrium opacities. High-fidelity spectra may reveal new physics in the simulations not seen with low-fidelity spectra, but the cost of these simulations also scales with the level of fidelity of the opacities being used. Neural networks are capable of reproducing these spectra, but neural networks need data to train them, which limits the level of fidelity of the training data. This article demonstrates that it is possible to reproduce high-fidelity spectra with median errors in the realm of 3%–4% using as few as 50 samples of high-fidelity Krypton data by performing transfer learning on a neural network trained on many times more low-fidelity data.

Evaluation of Key Spatiotemporal Learners for Print Track Anomaly Classification Using Melt Pool Image Streams

Recent applications of machine learning in metal additive manufacturing (MAM) have shown great potential to resolve critical barriers to MAM's widespread adoption. Recent research on the topic highlights the significance of using melt pool signatures to predict defects on-the-fly. While high-fidelity melt pool image data has the potential to enable accurate predictions, hardly any work exists on the use of state-of-the-art spatiotemporal models to leverage the information embedded in the transient and sequential nature of the additive process. This work introduces and implements some of the major deep spatiotemporal learners that can be adapted to classify melt pool image streams from different materials, systems, and applications. In this regard, two-stream networks with a spatial and temporal stream, a recurrent spatial network, and a factorized 3D convolutional neural network were investigated herein. The generalization abilities of these models to perturbations in the melt pool image data are tested using data perturbation techniques that are grounded in palpable process situations. The implemented architectures exhibit the ability to learn the spatiotemporal features of the melt pool image sequences. However, only the Kinetics400 pre-trained SlowFast network, which belongs to the two-stream category, showed strong generalisation abilities to data perturbations.

Elastic Seismic Imaging Enhancement of Sparse 4C Ocean-Bottom Node Data Using Deep Learning

The ocean bottom node (OBN) seismic acquisition system is designed to gather high-fidelity, wide-azimuth, and long-offset four-component (4C) data, which includes shear waves and enables the use of the elastic assumption in imaging and inversion. However, deploying geophysical instruments on the seafloor is difficult and costly, leading to the usual adoption of sparse node spacing. This can, however, lead to poor illumination and imaging challenges, especially in the shallow subsurface near the seafloor. To address these issues in the context of 4C elastic imaging, we propose a deep learning-based method using a multi-scale convolutional neural network (Ms-CNN) to improve the imaging quality of OBN surveys with sparse data acquisition. As an alternative to interpolating the sparse seismic data in the data domain, which can be a challenging task due to the limitations attributed to sampling theorem and the often larger amounts of data compared to the image, we train an Ms-CNN in a supervised fashion to map from sparse data images of PP and PS sections produced by 4C Gaussian beam migration to the equivalent dense data images, allowing for the direct processing of sparse data to improve imaging quality. Here, we combine the mean absolute error and multiscale structure similarity index measure in the loss function to optimize the network’s training process, and to help improve the performance. The effectiveness of the method is demonstrated through experiments on synthetic and field data, resulting in improved event continuity and reduced noise in migration results from sparse OBN acquisitions.

Multi-fidelity Surrogate Modelling of Wall Mounted Cubes

This paper focuses on the application of multi-fidelity surrogate modelling to characteristics of a flow as it changes with a parameter. This provides insight into the potential of combining multi-fidelity modelling approaches with varying fidelities of computational fluid dynamics methods to a parameter space exploration. A limited number of trusted high-fidelity large eddy simulation data points, in combination with an extended study using lower-fidelity Reynolds averaged Navier–Stokes modelling is used as the input for the surrogate model. Multi-fidelity surrogate models are implemented to bridge the low-fidelity and high-fidelity models providing an improved surrogate model over using a single fidelity alone. The flow around tandem wall mounted cubes at varying inlet yaw angle is used as an aerodynamic test case for this methodology. Results presented show that the multi-fidelity surrogate modelling provides a significant improvement over single fidelity modelling for the prediction of global flow properties. This methodology is then extended to combine multiple local flow features into the multi-fidelity model to build up fuller descriptions of the flow at angles not included in the training data for the model. The results of this are presented for both one-dimensional line plots at a range of locations along the center line of the flow and for two-dimensional slices of the velocity field. The multi-fidelity surrogate model produces results at locations in the parameter space away from the high fidelity training data that match closely to large eddy simulation results.

Multi-fidelity surrogate modeling using long short-term memory networks

When evaluating quantities of interest that depend on the solutions to differential equations, we inevitably face the trade-off between accuracy and efficiency. Especially for parametrized, time-dependent problems in engineering computations, it is often the case that acceptable computational budgets limit the availability of high-fidelity, accurate simulation data. Multi-fidelity surrogate modeling has emerged as an effective strategy to overcome this difficulty. Its key idea is to leverage many low-fidelity simulation data, less accurate but much faster to compute, to improve the approximations with limited high-fidelity data. In this work, we introduce a novel data-driven framework of multi-fidelity surrogate modeling for parametrized, time-dependent problems using long short-term memory (LSTM) networks, to enhance output predictions both for unseen parameter values and forward in time simultaneously — a task known to be particularly challenging for data-driven models. We demonstrate the wide applicability of the proposed approaches in a variety of engineering problems with high- and low-fidelity data generated through fine versus coarse meshes, small versus large time steps, or finite element full order versus deep learning reduced-order models. Numerical results show that the proposed multi-fidelity LSTM networks not only improve single-fidelity regression significantly, but also outperform the multi-fidelity models based on feed-forward neural networks.

Multifidelity Physics-Constrained Neural Networks With Minimax Architecture

Abstract Data sparsity is still the main challenge to apply machine learning models to solve complex scientific and engineering problems. The root cause is the “curse of dimensionality” in training these models. Training algorithms need to explore and exploit in a very high-dimensional parameter space to search the optimal parameters for complex models. In this study, a new scheme of multifidelity physics-constrained neural networks with minimax architecture is proposed to improve the data efficiency of training neural networks by incorporating physical knowledge as constraints and sampling data with various fidelities. In this new framework, fully connected neural networks with two levels of fidelities are combined to improve the prediction accuracy. The low-fidelity neural network is used to approximate the low-fidelity data, whereas the high-fidelity neural network is adopted to approximate the correlation function between the low-fidelity and high-fidelity data. To systematically search the optimal weights of various losses for reducing the training time, the Dual-Dimer algorithm is adopted to search high-order saddle points of the minimax optimization problem. The proposed framework is demonstrated with two-dimensional heat transfer, phase transition, and dendritic growth problems, which are fundamental in materials modeling. With the same set of training data, the prediction error of the multifidelity physics-constrained neural network with minimax architecture can be two orders of magnitude lower than that of the multifidelity neural network with minimax architecture.

Automated vehicle data pipeline for accident reconstruction: New insights from LiDAR, camera, and radar data

As automated vehicles are deployed across the world, it has become critically important to understand how these vehicles interact with each other, as well as with other conventional vehicles on the road. One such method to achieve a deeper understanding of the safety implications for Automated Vehicles (AVs) is to analyze instances where AVs were involved in crashes. Unfortunately, this poses a steep challenge to crash-scene investigators. It is virtually impossible to fully understand the factors that contributed to an AV involved crash without taking into account the vehicle’s perception and decision making. Furthermore, there is a tremendous amount of data that could provide insight into these crashes that is currently unused, as it also requires a deep understanding of the sensors and data management of the vehicle. To alleviate these problems, we propose a data pipeline that takes raw data from all on-board AV sensors such as LiDAR, radar, cameras, IMU’s, and GPS’s. We process this data into visual results that can be analyzed by crash scene investigators with no underlying knowledge of the vehicle’s perception system. To demonstrate the utility of this pipeline, we first analyze the latest information on AV crashes that have occurred in California and then select two crash scenarios that are analyzed in-depth using high-fidelity synthetic data generated from the automated vehicle simulator CARLA. The data visualization procedure is demonstrated on the real-world Kitti dataset by using the YOLO object detector and a monocular depth estimator called AdaBins. Depth from LIDAR is used as ground truth to calibrate and assess the effect of noise and errors in depth estimation. The visualization and data analysis from these scenarios clearly demonstrate the vast improvement in crash investigations that can be obtained from utilizing state-of-the-art sensing and perception systems used on AVs.

Offshore wind resource assessment based on scarce spatio-temporal measurements using matrix factorization

In the pre-construction of wind farms, wind resource assessment is of paramount importance. Measurements by lidars are a source of high-fidelity data. However, they are expensive and sparse in space and time. Contrarily, Weather Research and Forecasting models generate continuous data with relatively low fidelity. We propose a hybrid approach combining measurements and output from numerical simulations for the assessment of offshore wind. Firstly, the datasets were fed onto a matrix, with columns representing the spatial lidar and WRF points, and the rows representing the time steps. Entries of the matrix reflect the wind speed, empty entries represent unobserved data. Then, matrix factorization using Gaussian process was employed for filling the missing entries with statistically calculated estimates. The model was optimized with stochastic gradient descent to apply GP without approximation methods. To evaluate the method, wind speed data along the coast of Denmark were used. The proposed technique, evaluated using two experiments, resulted in 58% more accurate results than the industrial standard method with trivial increase of computational cost. The RMSE of the proposed method ranges between 0.35 and 0.52 m/s.

A comparative study of learning techniques for the compressible aerodynamics over a transonic RAE2822 airfoil

In this study, the modeling of the compressible pressure field on the RAE 2822 airfoil using deep learning (DL) is investigated. The objective is to generate, at low cost, the complete Mach envelope from a given aerodynamic database. A dataset with more than 2.000 RANS simulations at various angles of attack and Mach numbers has been created, and validated, to train fast and accurate surrogate models model of the pressure field. The procedure is inspired by the well-known analytical Prandtl–Glauert transformation, which maps solutions from one Mach number to another, but however fails in the transonic regime because of the strong non-linearities due to shock waves. Thus, instead of a classical direct generation of the pressure fields, DL is used here to perform a geometric transformation from an incompressible pressure field towards a compressible solution. The key question addressed in this study is: how accurate and sample efficient are DL techniques compared with classical surrogate modeling tools? To answer this question, two types of DL are investigated (CNN and GCNN) and compared with Proper Orthogonal Decomposition (POD) coupled to a Gaussian Process Regression (GPR). These methods are trained on various input resolutions and database sizes. Results show that DL models are able to accurately predict the intensity and location of the shock over a wide range of angles of attack and Mach numbers up to the transonic regime. Predictions of the lift coefficient reveal a relative error lower than 2% with respect to the high-fidelity data. Because of their inherent nonlinear nature, CNN and GCNN provide more accurate results even for small training datasets, whereas POD encounters difficulties to reconstruct properly shock waves. Since CNN requires inputs and outputs in a pixel-like format, it suffers from interpolation errors that can only be mitigated with fine resolutions, making the training more difficult. By performing convolutions directly on unstructured data, GCNN eliminates this interpolation error and provides the best accuracy and sample efficiency. This study shows that advanced DL techniques, such as CNN and GCNN, are capable of predicting complex flows, and outperform classical tools on such tasks.

Evaluation and quantification of compressor model predictive capabilities under modulation and extrapolation scenarios

Testing and evaluation of select semi-empirical and black-box compressor models is carried out to quantify performance in modulation (variable speed), extrapolation, and additionally, variable superheat scenarios. Three representative literature models and an artificial neural network (ANN) model are benchmarked against the industry standard AHRI model. A methodology quantifying model performance, compared against experimental data, in said scenarios is presented. High-fidelity test data taken from either a hot-gas bypass load stand or compressor calorimeter. Scroll, screw, reciprocating, and spool compressor technologies were collected with R410A, R1234ze(E), R134a, and R32 refrigerants totaling 434 experimental points. Data is divided into training, extrapolation, variable speed, and variable superheat data splits to examine model performance. Mean Absolute Percentage Error (MAPE) is computed for mass flow rate and power after training models with training data and evaluating them against the other data splits. Two literature models are true semi-empirical formulations while the other, the ANN, and AHRI model are more empirical in nature. Neither semi-empirical model predicted all compressors. When the compressor type is predicted, the semi-empirical models yield MAPE’s less than 8%, 5%, and 4% for mass flow rate and power prediction in extrapolation, modulation, and variable superheat scenarios, respectively. The exception is the Popovic and Shapiro model performing at 21% MAPE in variable superheat power prediction for the spool compressor with R1234ze(E). The ANN showed highest errors of 9.3%, 12%, and 17% in extrapolation, modulation, and variable superheat scenarios, respectively. All models outperformed the AHRI model by several orders of magnitude in these scenarios.

Utilizing high fidelity data into engineering model calculations for accurate wind turbine performance and load assessments under design load cases

AbstractWind turbines often have lower performance and experience higher loading in real operation compared to the original design performance. The reasons stem from the influences of complex atmospheric turbulence, blade contamination, surface imperfection and airfoil‐shape changes. Engineering models used for designing wind turbines are limited to information derived from blade sectional datasets, while details on the three‐dimensional blade characteristics are not captured. In these studies, a dedicated strategy to improve the prediction accuracy of engineering model calculations will be presented. The main aim is to present an elaborated effort to obtain a better estimate of the turbine loads in realistic operating conditions. The present studies are carried out by carefully utilizing data from high fidelity Computational Fluid Dynamics (CFD) computations into Blade Element Momentum (BEM) and Vortexline methods. The results are in a good agreement with detailed field measurement data of a 2.3 MW turbine. The studies are further extended to a large turbine having a rated power of 10 MW to provide an overview of its suitability for large turbines. Finally, calculations of the wind turbine under a realistic IEC design load case are demonstrated. The studies highlight important considerations for engineering modeling using BEM and Vortexline methods.

Reconstructing turbulent velocity information for arbitrarily gappy flow fields using the deep convolutional neural network

Detailed and reliable flow information is the basis for understanding and further mediating turbulent flows. Due to experimental limitations, such as the absence of seeding particles owing to an inhomogeneous tracer distribution or obstructed optical paths, gappy flow-field data frequently appear with diverse shapes. To resolve this problem, we propose herein the use of a convolutional neural network (CNN) model to reconstruct the velocity field with the missing information of wall-confined turbulent flows. We consider the example of a turbulent channel flow with a frictional Reynolds number Reτ=180 and use machine learning to attain the given objective. High-fidelity numerical data obtained by direct numerical simulation based on the lattice Boltzmann equation are used to generate the datasets required for network training, where data in randomly located square or rectangular regions are masked to provide a maximally realistic instantaneous gappy flow field. The results show that the missing information in gappy regions can be effectively reconstructed for both instantaneous and temporally continuous flow fields. Furthermore, the results are insensitive to the missing locations, even if the locations vary with time. The L2 relative error of the reconstructed instantaneous flow field is generally around 2%. Furthermore, an analysis based on the kinetic-energy spectrum and proper orthogonal decomposition verifies that the reconstructed data are physically consistent with the ground truth. The extracted dominating modes have a maximum relative error level of 10−3. The results obtained herein verify that the proposed CNN model provides complete and reliable data for gappy flows and are physically consistent with physical data.

High-fidelity surface flow data-driven aerodynamic solution strategy for non-smooth configurations: Study of compressor cascade with micro riblet surface

In this paper, a new aerodynamic solution strategy for non-smooth configurations is proposed based on the wall modification model by machine learning to perform numerical simulations, rather than directly describing the global flow field with massive grids. The aerodynamic effect of non-smooth configurations in the presence of pressure gradients is investigated utilizing the proposed method. Flow features of non-smooth surface are provided by high-fidelity surface flow data acquired through lattice Boltzmann method simulation. The wall modification model is constructed by Fruit fly Optimization Algorithm-Generalized Regression Neural Network (FOA-GRNN) to reproduce the behavior of microflow near the non-smooth surface. Typical flow features, e.g., velocity corrections induced by surface texture as the output of the FOA-GRNN model, are imposed on configuration boundaries, improving computational efficiency and wall resolution. The novel aerodynamic solution strategy is validated by comparing the results of the experiment. In addition, the performance analysis of compressor cascade with micro riblet surface utilizing the above method is conducted. The results indicate that the non-smooth surface structure decreases skin friction and turbulent intensity in the flow channel compared with smooth cascade, thus significantly reducing the total pressure loss. The paper shows a positive prospect of the data-driven strategy in evaluating the aerodynamic performances of non-smooth configurations and provides a reliable solution method for the subsequent design of micro-nano surfaces.

Assessment through high-fidelity simulations of a low-fidelity noise prediction tool for a vertical-axis wind turbine

Vertical-axis wind turbines have the potential to be installed nearby urban areas, where noise regulations are a constraint. Accurate modelling of the far-field noise with low-order fidelity methods is essential to account for noise early in the design phase. The challenge for the vertical-axis wind turbine is the unsteady azimuthal variation of the flow over the blades, which makes the prediction of the far-field noise complex with low-fidelity methods. In this paper, the state-of-the-art of low-fidelity methods are assessed against scale-resolving high-fidelity numerical simulations of a realistic vertical-axis wind turbine carried out with the lattice-Boltzmann very large eddy simulations method. High-fidelity numerical data are validated against experimental aerodynamics data of the same vertical-axis wind turbine. The low-fidelity method is based on the actuator cylinder model coupled with semi-empirical models for airfoil-self noise and turbulence-interaction noise. Results show a good agreement between the high-fidelity simulations and the low-fidelity model at low frequencies (i.e. between 2 × 101Hz and 1 × 102Hz), where turbulence-interaction noise is the dominant noise source. At higher frequencies, the airfoil-self noise dominates and existing methods, based on steady airfoils, do not correctly predict noise. This paper shows that the presented low-fidelity model predicts the aerodynamics and the aeroacoustics of the turbine with an acceptable accuracy for a design stage. However, improvements are needed to better predict the far-field noise for blades in an unsteady field.

Application of an overset grid method for the performance analysis of flapping airfoils

With the growing importance of renewable energy, the interest in energy-harvesting technologies based on flow-induced vibrations of airfoils has been rekindled in the past few years. Compared to conventional turbines, these devices are centrifugal stress-free and hence they are structurally more robust. They are also environmentally friendly due to the relatively low speeds, thus reducing the impact on flying animals. In order to numerically investigate these types of devices, mesh motion techniques must be used so that the computational grid can adapt to the time-varying shape of the domain and boundaries, thus preserving its robustness and quality. During the last years, many different dynamic mesh approaches have been developed to better predict the behaviour of a flow interacting with solid moving bodies. The aim of the present work is to apply and validate the overset grid method offered by the OpenFOAM software. In this type of mesh, one or more grid blocks are allowed to overlap with other sets of cells to solve the fluid domain with moving bodies, thus showing many potential advantages compared to the other existing dynamic mesh approaches. In the first part of this work, a numerical investigation of the flow over a stationary SD 7003 airfoil at low Reynolds number has been performed using an unsteady RANS approach. The computed results have then been compared to experimental and high-fidelity computational data available from the literature in order to validate the presented model in terms of numerical domain configuration, mesh refinement, and turbulence modelling. Subsequently, unsteady RANS simulations have been performed on the SD 7003 and NACA 0012 airfoils for two different amplitudes of flapping motion, i.e., 30° and 45°. Finally, the results obtained with the OpenFOAM overset grid solver have been compared with the numerical and experimental benchmarks presented by Kurtulus, showing an excellent agreement with both computed and measured data.

Multi-fidelity data-driven modelling of rate-dependent behaviour of soft clays

ABSTRACT Conventional phenomenological elasto-viscoplastic models include numerous parameters that need to be calibrated by case-specific experiments. Data-driven modelling has recently emerged and provided an alternative to constitutive modelling. This study proposes a modelling framework based on multi-fidelity data to model the rate-dependent behaviour of soft clays. In this framework, low-fidelity (LF) data generated by an elasto-viscoplastic model and high-fidelity (HF) data from experimental tests are necessary. Stress–strain-strain rate correlations behind LF and HF data can be captured by long short-term memory and feedforward neural networks, respectively, such that final predictions can be given by a multi-fidelity residual neural network (MR-NN). Such a framework with the same LF data is applied in Hong Kong marine deposits and Merville clay to investigate its feasibility and generalisation ability. In addition, the effect of LF data on the performance of MR-NN is discussed to verify the robustness of the framework. All results demonstrate that rate-dependent undrained shear strength and pore-water pressure can be accurately modelled through the framework, showing adaptive non-linear modelling capability, less demand for experimental data, and superior robustness. These characteristics indicate a considerable potential in modelling the rate-dependent behaviour of clays.

Surrogate Model Development for Optimized Blended-Wing-Body Aerodynamics

In the conceptual design phase of conventional-configuration aircraft, calibrated low-fidelity methods provide sufficiently accurate estimates of aerodynamic coefficients. It has been observed, however, that for blended-wing–body aircraft, important flow effects are not captured adequately with low-fidelity aerodynamic tools. Consequently, high-fidelity methods become necessary to study blended-wing–body aerodynamics. Since repeated function calls are needed in an optimization loop, high-fidelity analysis is prohibitively expensive in the conceptual design phase, where several optimization scenarios are considered. In this paper, the integration of high-fidelity data for blended-wing–body aircraft for a mission calculation module is presented. A surrogate model based on Gaussian processes (GPs) with acceptably low prediction error is sought as an alternative to Reynolds-averaged Navier–Stokes computational fluid dynamics. Three adaptations are considered: sparse GPs, mixtures of GP approximators, and need-based filtering for GP. The results provide benchmark values for this case and show that the combination of subsonic and transonic behaviors in the training set is problematic and that, for the considered datasets, sparse GP models suffer from oversmoothing, while mixtures of GP models suffer from overfitting. From the error levels, it is observed that a GP with an infinitely differentiable squared exponential kernel based on reduced data pertinent to mission analysis is the most effective option.