Computational Fluid Dynamics Solver Research Articles

Development of conjugate heat- and moisture-transfer model for pineapple drying using OpenFOAM

The kinetic model of drying is important for food preservation, particularly for pineapples, which have a high moisture content and are highly perishable. This paper presents a novel approach utilizing OpenFOAM, which is an open-source computational fluid dynamics (CFD) software, to model the hot-air drying kinetics of pineapple slices. The developed conjugate heat- and moisture-transfer model integrates forced-convection phenomena with moisture-diffusion dynamics. Owing to enhancements to the original OpenFOAM solver, C++ programming facilitates the appropriate representation of heat and moisture transfer based on the finite-volume method. Experimental data are utilized to derive a moisture diffusivity model, which achieved high statistical quality with an average R2 of 0.9443. Additionally, the SST k–ω turbulence model accurately describes the airflow around the pineapple-slice model. The developed model agrees well with the experimental results, with an average error of less than 1.29 %. Further analysis using the model provides insights into the effects of variations in the pineapple-slice geometry, where increased inner diameter and reduced thickness result in enhancement indices of 1.002 and 1.005, respectively. This novel CFD solver is promising for simulating the drying processes of various agricultural products, including pineapples, thus contributing to improvements in process optimization and efficiency.

Enhancing CFD solver with Machine Learning techniques

This study addresses the computational challenges in fluid flow simulations arising from demanding computational grids, required to capture the temporal and length scales involved. Our approach focuses on the pressure solver, as this is a resource-intensive component in Computational Fluid Dynamics (CFD) solvers. We achieve this by integrating a Machine Learning (ML) surrogate model with an incompressible fluid flow solver. We created two variants of an ML-enhanced CFD solver which were able to reduce the number of iterations required by the CFD pressure solver during unsteady flow simulations. Consequently, the simulations yielded comparable drag coefficients and Strouhal numbers, accompanied by an eightfold decrease in execution time. The performance enhancements are attributed to reduced computational effort per temporal iteration and early-stage forcing on the simulation dynamical behavior when using the ML-based surrogate models. This research introduces an approach to enhance the computational efficiency of fluid flow analyses by incorporating surrogate models to aid the pressure solver in CFD simulations. We propose a Hybrid CFD solver, ie. a physics-informed solver enhanced with data-driven surrogate models.

Graph and convolutional neural network coupling with a high-performance large-eddy simulation solver

Computational Fluid Dynamics (CFD) traditionally relies on long-standing numerical simulation strategies for the Navier–Stokes equations. Recently, interest in data-driven hybrid CFD solvers has spiked, leveraging pre-computed datasets to enhance various weak links inside existing solvers, such as closure models, under-resolved physics, or even to guide numerical resolution strategies. Running these hybrid solvers, notably in High Performance Computing (HPC) environments, presents specific challenges. In particular, context-aware deep learning (e.g. Convolutional (CNN) and Graph (GNN) Neural Networks) is promising for this task, but requires passing data representations between the physics solver and the neural network. In relevant industrial configurations, CFD meshes can be Cartesian but highly irregular, or unstructured, both of which do not match the pixel/voxel structure needed to run CNNs. In addition, discrepancies in programming language and libraries are common between CFD and machine learning applications. This work explores the many challenges of running a parallel hybrid solver in an HPC context, through the coupling of the AVBP CFD solver with neural networks in turbulent combustion and wall friction modeling applications. The knowledge gained is showcased in this article, as well as assembled in an actionable open-source library.

A one-dimensional computational model for blood flow in an elastic blood vessel with a rigid catheter.

Strokes are one of the leading causes of death in the United States. Stroke treatment involves removal or dissolution of the obstruction (usually a clot) in the blocked artery by catheter insertion. A computer simulation to systematically plan such patient-specific treatments needs a network of about 105 blood vessels including collaterals. The existing computational fluid dynamic (CFD) solvers are not employed for stroke treatment planning as they are incapable of providing solutions for such big arterial trees in a reasonable amount of time. This work presents a novel one-dimensional mathematical formulation for blood flow modeling in an elastic blood vessel with a centrally placed rigid catheter. The governing equations are first-order hyperbolic partial differential equations, and the hypergeometric function needs to be computed to obtain the characteristic system of these hyperbolic equations. We employed the Discontinuous Galerkin method to solve the hyperbolic system and validated the implementation by comparing it against a well-established 3D CFD solver using idealized vessels and a realistic truncated arterial network. The results showed clinically insignificant differences in steady flow cases, with overall variations between 1D and 3D models remaining below 10%. Additionally, the solver accurately captured wave reflection phenomena at domain discontinuities in unsteady cases. A primary advantage of this model over 3D solvers is its ease in obtaining a discretized geometry of complex vasculatures with multiple arterial branches. Thus, the 1D computational model offers good accuracy and applicability in simulating complex vasculatures, demonstrating promising potential for investigating patient-specific endovascular interventions in strokes.

Fast simulation of airfoil flow field via deep neural network

Computational Fluid Dynamics (CFD) has become an indispensable tool in the optimization design, and evaluation of aircraft aerodynamics. However, solving the Navier-Stokes (NS) equations is a time-consuming, memory demanding and computationally expensive task. Artificial intelligence offers a promising avenue for flow field solving. In this work, we propose a novel deep learning framework for rapidly reconstructing airfoil flow fields. Channel attention and spatial attention modules are utilized in the downsampling stage of the UNet to enhance the feature learning capabilities of the deep learning model. Additionally, Embedding the predicted values of the deep learning model as initial values into the CFD solver to accelerate its iterative convergence. The NACA series airfoils were used to validate the prediction accuracy and generalization of the deep learning model. The experimental results represent the deep learning model achieving flow field prediction speed three orders of magnitude faster than CFD solver. Furthermore, the CFD solver integrated with deep learning model demonstrates a threefold acceleration compared to CFD solver. By extensively mining historical flow field data, an efficient solution is derived for the rapid simulation of aircraft flow fields.

A deep learning computational fluid dynamics solver for simulating liquid hydrogen jets

Modeling and simulating the sudden depressurization of liquids inside nozzles is a significant challenge because of the plethora of the associated complex phenomena. This pressure drop together with the rapid phase change of the liquid is important characteristics of flash boiling. Computational fluid dynamics (CFD) multiscale simulations of flashing jets usually deploy additional models for modeling heat and mass transfer with long computational times. Intermediate steps such as volumetric meshing in mesh-based methods can also significantly increase the computational cost. This paper aims at providing academia and industry with a modeling tool to simulate and investigate the complex multi-facet phenomenon of flash-boiling atomization deploying a machine-learning method that could save thousand Central Processing Unit hours offering instantaneous CFD predictions. The presented machine-learning CFD method completely replaces the traditional CFD simulations workflow and requires little simulation expertise from the end-user. Notably, this is a novel model that couples for the first time the thermodynamic non-equilibrium with convolutional neural networks to simulate flashing liquid hydrogen jets thousand times faster than the standalone CFD solver. The accuracy of the novel approach is evaluated, demonstrating adequate accuracy compared to different unseen simulations and experiments. This work offers the groundwork for further accelerating CFD predictions in multiphase flow problems and could significantly improve testing flash-boiling scenarios in various industrial settings.

A hybrid atomistic-continuum framework for multiscale simulations of boiling

Boiling is a multiscale physics process where the nucleation of vapour bubbles occurs due to molecular-scale interactions between the fluid and a heated wall, but it also depends on the larger-scale hydrodynamics and thermal boundary layers determined by the outer system boundary conditions. Modelling boiling from the nanometre up to the millimetre scales at which bubble departure occurs is not possible via state-of-the-art simulation methods: Molecular Dynamics (MD) simulations can capture nucleation from first principles but are limited to nanometre scales due to their computational cost, whereas computational fluid dynamics (CFD) simulations based on the continuum Navier-Stokes equations cannot capture nucleation. Here, we present a novel multiscale simulation method which merges MD and CFD descriptions into a single modelling framework, where MD resolves the near-wall region where molecular interactions are important, and a CFD solver resolves the bulk flow. We model the progressive heating of a Lennard-Jones fluid via contact with a solid wall until a vapour bubble nucleates in the MD region of the domain and grows by entering in the CFD domain. Our results show that an incompressible CFD flow model based on the Volume Of Fluid method with interphase mass transfer calculated via the Hertz-Knudsen-Schrage equation is sufficient to obtain seamless coupling of phase fraction, velocity and temperature fields, with the hybrid MD-CFD framework yielding bubble dynamics closely matching those of MD alone.

CPU–GPU heterogeneous code acceleration of a finite volume Computational Fluid Dynamics solver

This research focuses on accelerating the finite-volume Computational Fluid Dynamics (CFD) solver, SENSEI, through concurrent CPU–GPU heterogeneous computing, leveraging multiple CPUs and GPUs. An overview of SENSEI, its discretization, and the heterogeneous computing workflow utilizing MPI and OpenACC are provided. A performance model for CPU–GPU heterogeneous computing, incorporating ghost cell exchange usually applied in stencil computations, is introduced to estimate the performance. This work explores the scaling performance of CPU–GPU heterogeneous computing in comparison to a pure multi-CPU/GPU setup using a supersonic inlet test case. The results emphasize the advantages of leveraging both CPU and GPU computational power concurrently. Highlighting the collaborative synergy of CPUs and GPUs as co-workers, this work shows improved performance over exclusive use of pure CPUs or GPUs, with the proposed performance model ensuring a fair estimation of these advantages. This work concludes by offering suggestions for application users interested in accelerating scientific computing simulations and providing feedback for hardware architects designing improved CPU–GPU heterogeneous systems for heterogeneous computing.

Numerical investigation of aerodynamic interactions for the coaxial rotor system in low-speed forward flight

This study is conducted to investigate the aerodynamic characteristics of a coaxial rotor in low-speed forward flight using a high-fidelity computational fluid dynamics (CFD) solver. Numerical simulations are performed on the coaxial rotor and its isolated upper and lower rotors at two small advance ratios, 0.12 and 0.24. Interaction events in the flowfield, especially blade-vortex-interaction (BVI), are studied in detail. As the forward speed increases, the airload distributions become concentrated on the front part of the disk. Fluctuations in the loading indicate self-BVIs within each rotor, and blade-crossover events and rotor-to-rotor BVIs between the coaxial upper and lower rotors. Principal BVI events are identified in the well-captured wake structures by the 8th-order TENO reconstruction scheme. Self-BVIs on all rotors reduce in strength as the forward speed increases, while rotor-to-rotor BVIs on the coaxial lower rotor increase in strength with the increasing forward speed. In the coaxial configuration, the wake of the upper rotor is induced to move backwards and downwards due to the exertion of the lower rotor wake. The wake of the coaxial lower rotor features the most complex flow nature among all rotors. At both advance ratios, the coaxial lower rotor experiences a parallel rotor-to-rotor BVI, causing highly impulsive loading fluctuations along the blade span.

Sediment transport analysis under combined action of waves and current using a novel semi-coupled computational fluid dynamics solver

Sediment transport analysis under combined action of waves and current using a novel semi-coupled computational fluid dynamics solver

GPU-accelerated CFD-DEM modeling of gas-solid flow with complex geometry and an application to raceway dynamics in industry-scale blast furnaces

The coupling of Computational Fluid Dynamics (CFD) and Discrete Element Model (DEM) is a powerful tool for simulating dense particulate systems, yet the conventional CFD-DEM has limits for systems with large particle numbers and complex geometry. This paper reports a novel GPU-based CFD-DEM model to simulate the gas–solid flow with large particle numbers and complex geometry. A novel coupling strategy between the CFD solver and DEM solver is developed, featuring high efficiency and stability. The developed model is validated against the experimental measurements, and its efficiency is compared to the previous CFD-DEM simulations. Then, for demonstration, the model is employed to simulate the dynamic behavior of gas–solid flow in the raceway in ironmaking blast furnaces by considering complex tuyere structure details and the huge particle numbers involved. This model allows to study the effect of the tuyere angle in terms of raceway formulation and tuyere erosion. The results show that the largest and most stable raceway volume can be reached at 5° downward tuyere, although the −5° tuyere nose experiences more wear than 10° downward tuyere. The model provides a cost-effective tool to overcome the longstanding challenge of simulating dense fluid-particle systems with huge particle numbers and complex geometry.

Low-Speed Space Launch System Computational Fluid Dynamics: A Comprehensive Overview

In this review paper, low-speed computational work from NASA Langley Research Center in support of the Space Launch System (SLS) is discussed. This information includes both historical and present efforts with the Kestrel computational fluid dynamics solver. The low-speed aerodynamics of SLS is highly complex, and the analysis of the unsteady flowfield requires significant computational efforts. The SLS mission profile varies from the vehicle static on the launch pad through high-speed ascent. This study focuses on the prelaunch as well as liftoff and transition portions of the flight both in proximity to the launch tower and in isolation. High-alpha conditions, as large as 90 deg, result in a flowfield dominated by massive, large-scale flow separation and asymmetric vortices. High-fidelity solutions require an unsteady computational formulation to accurately capture the aerodynamics of the vehicle. A detailed discussion of the computational approach is presented, followed by key efforts to support the program, both historic and present, including information that has been either published previously or that has never before been published external to NASA.

Simulating chemical mixing and molten pool shape in dissimilar welds using thermal fluid dynamics

Estimating the properties of dissimilar metal welded joints made using modern high-energy beam techniques presents significant challenges due to the complex nature of the process. These complexities arise from the different thermal and mechanical properties of the materials involved. The properties of a welded joint are primarily determined by the element mixing and final chemical composition when two dissimilar metals are fused together, or when a dissimilar filler material is used. Hence, a representative model capable of tracking elemental distribution and temperature is essential for accurately estimating the micro-scale and macro-scale mechanical properties of the joint, thus optimising the joining process. In this work, a high-fidelity thermal fluid flow model in the multiComponentlaserbeamFoam computational fluid dynamics (CFD) solver is used to simulate the process of laser beam welding. For this purpose, the joining process of AISI 304 stainless steel with a low alloy steel filler material (G3Si1) is simulated. This framework is rigorously validated with experimental results showing the capabilities of such modelling approach. Different strategies to model the filler and material properties are explored: a stationary filler model, applied through the parent material's thickness, was found to accurately mirror experimental results. This model is then used to study the flow properties, thermocapillary effect, and diffusion on the intermixing of composition, aspects that have not been previously examined in depth. The thermocapillary forces were concluded to be the most significant, dictating the peak composition value, the shape and topology of the molten pool. Diffusion fluxes were found to be negligible and not responsible for the majority of the mixing.

Vortex-induced vibrations of wind turbines: From single blade to full rotor simulations

Vortex-induced vibrations (VIVs) of wind turbine blades are a phenomenon that cannot be predicted sufficiently accurate by low fidelity methods. High fidelity methods, such as computational fluid dynamics (CFD) coupled to a structural solver come at a high computational cost. Most studies of the phenomenon have so far focused on analysis of single blades. For these cases, the so-called forced-motion method of prescribing the blade motion according to structural modeshapes has proven effective.In this article, VIVs of the IEA 10 MW reference turbine is studied in a fluid–structure interaction setup, where the structural dynamics of the full turbine are represented in the structural solver and the full rotor aerodynamics are simulated in the CFD solver. Due to the high computational cost this study is limited to only a single case with the rotor in bunny configuration, 80 degrees pitch and 90 degrees yaw. Also the purely structural power dissipation of the full turbine is studied to enable full rotor forced-motion in future work. A blade tip amplitude of roughly 17 meters is reached in the full turbine simulations, which is a significantly larger limit cycle amplitude than what would be reached in a corresponding isolated single blade simulation. The analysis also shows that forced-motion simulations combining three single blades vibrating according to the structural modeshape would be feasible. It is further observed, that the power injection or extraction close to the blade tip may change drastically at very large amplitudes compared to lower amplitudes.

A GPU-accelerated computational fluid dynamics solver for assessing shear-driven indoor airflow and virus transmission by scale-resolved simulations

We explore the applicability of MATLAB for 3D computational fluid dynamics (CFD) of shear-driven indoor airflows. A new scale-resolving, large-eddy simulation (LES) solver titled DNSLABIB is proposed for MATLAB utilizing graphics processing units (GPUs). In DNSLABIB, the finite difference method is applied for the convection and diffusion terms while a Poisson equation solver based on the fast Fourier transform (FFT) is employed for the pressure. The immersed boundary method (IBM) for Cartesian grids is proposed to model solid walls and objects, doorways, and air ducts by binary masking of the solid/fluid domains. The solver is validated in two canonical reference cases and against experimental data. Then, we demonstrate the validity of DNSLABIB in a room geometry by comparing the results against another CFD software (OpenFOAM). Next, we demonstrate the solver performance in several isothermal indoor ventilation configurations and the implications of the results are discussed in the context of airborne transmission of COVID-19. The novel numerical findings using the new CFD solver are as follows. First, a linear scaling of DNSLABIB is demonstrated and a speed-up by a factor of 3-4 is also noted in comparison to similar OpenFOAM simulations. Second, ventilation in three different indoor geometries are studied at both low (0.1 m/s) and high (1 m/s) airflow rates corresponding to Re=5000 and Re=50000. An analysis of the indoor CO2 concentration is carried out as the room is emptied from stale, high CO2 content air. We estimate the air changes per hour (ACH) values for three different room geometries and show that the numerical estimates from 3D CFD simulations differ by 80%–150% (Re=50000) and 75%–140% (Re=5000) from the theoretical ACH value based on the perfect mixing assumption. Third, the analysis of the CO2 probability distributions (PDFs) indicates a relatively non-uniform distribution of fresh air indoors. Fourth, utilizing a time-dependent Wells-Riley analysis, an example is provided on the growth of the cumulative infection risk which is shown to reduce rapidly after the ventilation is started. The average infection risk is shown to reduce by a factor of 2 for lower ventilation rates (ACH=3.4-6.3) and 10 for the higher ventilation rates (ACH=37-64). Finally, we utilize the new solver to comment on respiratory particle transport indoors. A key contribution of the paper is to provide an efficient, GPU compatible CFD solver environment enabling scale-resolved simulations (LES/DNS) of airflow in large indoor geometries on a single GPU designed for high-performance computing. The demonstrated efficacy of MATLAB for GPU computing indicates a high potential of DNSLABIB for various future developments on airflow prediction.

Numerical simulation of rime ice accretion on airfoil using rigid sphere model

A 2D/3D numerical model for rime ice shape prediction is developed by combining the rigid sphere model (RSM) and computational fluid dynamics (CFD), and its characteristics are investigated. In the RSM, the ice accretion is modeled based on first principles with little physical modeling. The ice particle on an airfoil is described as a rigid sphere, and the ice shape is represented as a collection of rigid spheres. We use a non-body-fitted hierarchical Cartesian grid-based CFD solver so that complex ice shape represented by RSM can be handled. The tree data of the hierarchical Cartesian grid is also used for fast search of droplet positions. In addition, multiple ice particles are modeled as one large particle to reduce the number of trajectory calculations. The sensitivity of several parameters in the calculation is studied in 2D. Subsequently, 3D ice accretion analysis on NACA0012 airfoil using the RSM is performed for the first time. The simulation results are compared with those of the wind-tunnel experiments conducted at NASA. Although the RSM is a simple ice accretion model, reliable results are obtained in both 2D and 3D calculations.

Utilizing Artificial intelligence to identify an Optimal Machine learning model for predicting fuel consumption in Diesel engines

This paper describes the utilization of artificial intelligence (AI) techniques to identify an optimal machine learning (ML) model for predicting dodecane fuel consumption in diesel combustion. The study incorporates sensitivity analysis to assess the impact levels of various parameters on fuel consumption, thereby highlighting the most influential factors. In addition, this study addresses the impact of noise and implements data cleaning techniques to ensure the reliability of the obtained results. To validate the accuracy of the predictions, the study performs several metrics and validation process, including comparisons with computational fluid dynamics (CFD) results and experimental data. Comprehensive comparisons are made among neural networks (NN), random forest regression (RFR), and Gaussian process regression (GPR) models, taking into account the complexity associated with fuel consumption predictions. The findings demonstrate that the GPR model outperforms the others in terms of accuracy, as evidenced by metrics such as mean absolute error (MAE), mean squared error (MSE), Pearson coefficient (PC), and R-squared (R2). The GPR model exhibits superior predictive ability, accurately detecting and predicting even individual data points that deviate from the overall trend. The significantly lower absolute error values also consistently indicate its higher accuracy compared with the NN and RFR models. Furthermore, the GPR model shows a remarkable speedup, approximately 1.7 times faster than traditional CFD solvers, and physically captures the momentum and thermal characteristics in a surface field prediction. Finally, the target optimization is assessed using the Euclidean distance as a fitness function, ensuring the reliability of predicted data.

Numerical analysis of aero-hydrodynamic wake flows of a floating offshore wind turbine subjected to atmospheric turbulence inflows

The influence of atmospheric turbulence inflow on coupled dynamic behaviors and wakes of floating offshore wind turbine (FOWT) is pronounced, especially as blade size increases. In this study, we explore the effects of atmospheric inflow on aero-hydrodynamics and wake characteristics of a spar-type FOWT. A well-validated in-house computational fluid dynamic (CFD) solver, FOWT-UALM-SJTU, is utilized to conduct the numerical simulations. To generate the realistic atmospheric inflow condition, the large eddy simulation (LES) with long-duration is employed. Two other cases involving uniform inflow and shear inflow are conducted to provide some comparable results. Compared to uniform inflow and shear inflow, the power and thrust of FOWT under atmospheric inflow display greater instability, with their power spectrums exhibiting higher intensity in high frequency region. The variation of yaw moment in atmospheric scenario is significantly drastic, leading to a remarkable response in platform yaw motion. Notably, the dominant frequency in this scenario is the blade passage frequency, as opposed to the incident wave frequency observed in the uniform and shear scenarios. The atmospheric inflow promotes the manifestation of wake breakdown and wake meandering, resulting in the faster wake recovery of FOWT. Besides, the meandering in the far wake is more significant compared to uniform and shear scenarios. The absence of vortex rings in atmospheric scenario is observed, along with the presence of more complex and smaller vortices in far wake. Our findings highlight the remarkable impacts of atmospheric inflow on dynamic responses and wake evolution of FOWT, and it is recommended that the atmospheric inflow be taken into account when assessing wake interactions between multiple FOWTs.

Research on Structural Collapse of a Containership under Combined Bending–Torsion by Oblique Waves

Large waves cause a great number of collapsed-ship accidents, resulting in the loss of many lives and properties. It has been found that most of these collapses are caused by encountering oblique waves. As a result, the ship structure experiences a complex collapse under combined bending and torsion. This paper utilizes a numerical hydroelasto-plastic approach, coupling CFD (Computational Fluid Dynamics) with the nonlinear FEM (Finite-Element Method), to study the structural collapse of a containership in oblique waves. First, a 4600 TEU containership was selected to study its collapse mechanism under oblique waves. Second, a hydroelasto-plastic numerical coupling of CFD and nonlinear FEM is used to co-calculate the wave loads and structural collapse of containership. The hydrodynamic model is constructed and used to solve wave loads in the CFD solver, and a nonlinear FEM model of containership with finer meshes is also modeled to solve the structural collapses, including plasticity and buckling. Third, several oblique-wave cases involving heading angles of 120°, 135°, 150°, and 180° are determined and calculated. Typical cases are discussed for time-domain stress histories and collapsed courses. Finally, the influence of oblique-wave parameters on structural collapse is discussed, and the collapse mechanism of containerships under the action of oblique waves is obtained, which provides a new understanding of ship structure design.

Investigation of the thermal performance of a spiral fin-and-tube heat exchanger by numerical methods

Spiral fin-and-tube heat exchangers (SPF HEX) are broadly employed heat transfer (HT) equipment for heating and cooling applications. Therefore, developing a model that predicts the HT characteristics could provide an advantage when selecting the required device considering operation conditions. The efficacy of the HT characteristics of a three-dimensional SPF HEX was examined under various boundary conditions with the help of a commercial computational fluid dynamics (CFD) solver and predicted by employing machine learning (ML) approaches as a rarely studied novel work. Three-dimensional, steady, and incompressible Reynolds-Averaged Navier-Stokes (RANS) formulas, including continuity, momentum and energy were solved to obtain numerical data. k-ε Realizable turbulence model, which has two additional transport equations, was employed. The outlet temperature of water was calculated at various water inlet velocity, air inlet velocity, and air inlet temperature with a constant water inlet temperature. First, the methods of CFD were utilized to obtain sixty cases, then the outlet temperatures of water were obtained at various conditions. The current CFD technique was verified with a work in literature. After collecting a labeled dataset, this data is fed into four recognized ML algorithms: Support Vector Machine (SVM), Linear Regression (LR), Decision Tree Regression (DT), and Random Forest (RF). In the presented results, the SVM algorithm achieved the highest accuracy rate with values of 0.1765 mean squared error (MSE), 0.3294 mean absolute error (MAE), and 0.9940 R-squared (R2). Furthermore, these results indicate that ML algorithms can deliver high prediction performance, potentially reducing the need for many experimental and numerical studies.