Computational Fluid Dynamics Data Research Articles

Combining Computational Fluid Dynamics and Experimental Data to Understand Fish Schooling Behavior.

Understanding the flow physics behind fish schooling poses significant challenges due to the difficulties in directly measuring hydrodynamic performance and the three-dimensional, chaotic, and complex flow structures generated by collective moving organisms. Numerous previous simulations and experiments have utilized computational, mechanical, or robotic models to represent live fish. And existing studies of live fish schools have contributed significantly to dissecting the complexities of fish schooling. But the scarcity of combined approaches that include both computational and experimental studies, ideally of the same fish schools, has limited our ability to understand the physical factors that are involved in fish collective behavior. This underscores the necessity of developing new approaches to working directly with live fish schools. An integrated method that combines experiments on live fish schools with computational fluid dynamic (CFD) simulations represents an innovative method of studying the hydrodynamics of fish schooling. CFD techniques can deliver accurate performance measurements and high-fidelity flow characteristics for comprehensive analysis. Concurrently, experimental approaches can capture the precise locomotor kinematics of fish and offer additional flow information through particle image velocimetry (PIV) measurements, potentially enhancing the accuracy and efficiency of CFD studies via advanced data assimilation techniques. The flow patterns observed in PIV experiments with fish schools and the complex hydrodynamic interactions revealed by integrated analyses highlight the complexity of fish schooling, prompting a reevaluation of the classic Weihs model of school dynamics. The synergy between CFD models and experimental data grants us comprehensive insights into the flow dynamics of fish schools, facilitating the evaluation of their functional significance and enabling comparative studies of schooling behavior. In addition, we consider the challenges in developing integrated analytical methods and suggest promising directions for future research.

Planing Hull Hydrodynamic Performance Prediction Using LincoSim Virtual Towing Tank

This work shows the performance of LincoSim, a web-based virtual towing tank enabling automated and standardized calm water computational fluid dynamics (CFD) data sampling, extending previous published applications to the case of a high-speed hull. The calculations are performed for a 1:10 scale model of a 43 ft powerboat hull form in the Froude number range from 0.3 to 2.0. The counterpart physical model is the experimental fluid dynamics (EFD) campaign performed at the University of Naples Federico II, where the resistance, sinkage and trim data have been measured. The EFD/CFD data comparison is performed and shown with a discussion of the spotted differences. The average percentage differences between the EFD and CFD data for the whole speed range are 1.84, 6.87 and 6.94 for the resistance, dynamic trim, and sinkage, respectively. These results confirm the maturity of the standardized and automated CFD modeling for calm water hydrodynamic analysis included in LincoSim, even at very high Froude numbers. The wetted length of the keel and chine and the wetted surface are calculated from numerical data using the advanced post-processing. Finally, as a work in progress, we test a first comparison for the same hull of the EFD and CFD data, considering two seakeeping conditions for head waves at a given wavelength for two velocity conditions. Also, this kind of analysis confirms the tight correlation between the measured and computed outcomes. This synergic interplay of EFD and CFD can link the advantages of both methods to support hull design but also requires experiment planning and final data analysis to obtain physical parameters not easily measurable in laboratory, such as the wetted surface, wetted lengths, proper viscous contribution, and pressure distribution both in calm water and in waves.

Distributed Physics-Informed machine learning strategies for two-phase flows

This study delves into applying Distributed Learning Machines (DLM), a subset of Physics-Informed Neural Networks (PINN), in tackling benchmark challenges within two-phase flows. Specifically, two variants of DLM, namely Distributed Physics-Informed Neural Networks (DPINN) and Transfer Physics-Informed Neural Networks (TPINN), are studied. The DLM architecture strategically divides the global domain into distinct non-overlapping sub-domains, with interconnected solutions facilitated by interface conditions embedded in the loss function. Forward and inverse benchmark problems in two-phase flows are explored: (a) bubble in a reversing vortex and (b) Bubble Rising under Buoyancy. The Volume of Fluid (VOF) method handles interface transport in both scenarios, with the inverse problem incorporating interface-position data during the training phase. The forward problem highlights the effectiveness of DPINN in capturing the interface using a simple transport equation. The distinctive contribution of this work lies in its exploration of the inverse problem, offering insights into the scalability of distributed architectures when dealing with a system of governing equations. Following the validation of an initial PINN model against Computational Fluid Dynamics (CFD) data, the study extends to DPINN and TPINN. A parametric study optimizes network hyperparameters, emphasizing the regularization of loss terms within the DPINN loss function. A self-adaptive weighting strategy based on a Gaussian probabilistic model dynamically adjusts loss weights during training to overcome challenges associated with manual parameter tuning. The evaluation of accuracy against CFD data and published results underscore the efficacy of DLMs in addressing two-phase flow problems. Additionally, the computational efficiency of distributed networks is explored compared to traditional PINNs.

A comparative assessment of CFD based LSTM and GRU for hydrodynamic predictions of gas-solid fluidized bed

Numerical simulation of large-scale fluidized beds presents significant challenges for both scaling up the simulations and evaluating their performance. However, full-scale computational fluid dynamics (CFD) simulations of these reactors incur substantial computational time and cost. To overcome these challenges, we propose a novel approach that combines CFD simulations with data-driven deep-learning models to predict complex hydrodynamics. In this study, we trained two machine learning (ML) algorithms, namely the Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU), using spatiotemporal data obtained from CFD simulations. Impressively, the trained models accurately predicted future events at successive time steps, exhibiting excellent agreement for mean voidage, gas velocity, and particle velocity. Among the models utilized, LSTM outperformed the other. This demonstrated its superior ability to capture spatiotemporal from CFD data while requiring less computational time and power. The trained models provide a reliable means to extrapolate desirable mean particle velocity and voidage values within fluidized beds. These findings hold promising opportunities for enhancing the design and operation of gas-solid fluidized bed reactors, potentially leading to a significant reduction of up to 57–68% in simulation time.

Sensitivity of Post-TAVR Hemodynamics to the Distal Aortic Arch Anatomy: A High-Fidelity CFD Study.

Patient-specific simulations of transcatheter aortic valve (TAV) using computational fluid dynamics (CFD) often rely on assumptions regarding proximal and distal anatomy due to the limited availability of high-resolution imaging away from the TAV site and the primary research focus being near the TAV. However, the influence of these anatomical assumptions on computational efficiency and resulting flow characteristics remains uncertain. This study aimed to investigate the impact of different distal aortic arch anatomies-some of them commonly used in literature-on flow and hemodynamics in the vicinity of the TAV using large eddy simulations (LES). Three aortic root anatomical configurations with four representative distal aortic arch types were considered in this study. The arch types included a 90-degree bend, an idealized distal aortic arch anatomy, a clipped version of the idealized distal aortic arch, and an anatomy extruded along the normal of segmented anatomical boundary. Hemodynamic parameters both instantaneous and time-averaged such as Wall Shear Stress (WSS), and Oscillatory Shear Index (OSI) were derived and compared from high-fidelity CFD data. While there were minor differences in flow and hemodynamics across the configurations examined, they were generally not significant within our region of interest i.e., the aortic root. The choice of extension type had a modest impact on TAV hemodynamics, especially in the vicinity of the TAV with variations observed in local flow patterns and parameters near the TAV. However, these differences were not substantial enough to cause significant deviations in the overall flow and hemodynamic characteristics. The results suggest that under the given configuration and boundary conditions, the type of outflow extension had a modest impact on hemodynamics proximal to the TAV. The findings contribute to a better understanding of flow dynamics in TAV configurations, providing insights for future studies in TAV-related experiments as well as numerical simulations. Additionally, they help mitigate the uncertainties associated with patient-specific geometries, offering increased flexibility in computational modeling.

Analysis of aero engine plume potential core infrared signature

Purpose Aero-engine exhaust plume length can be more than the aircraft length, making it easier to detect and track by infrared seeker. Aim of this study is to analyze the effect of free stream Mach number (M∞) on length of potential core of plume. Also, change in infrared (IR) signature of plume and aircraft surface with variation in elevation angle (θ) is examined. Design/methodology/approach Convergent divergent (CD) nozzle is located outside the rear fuselage of the aircraft. A two dimensional axisymmetric computational fluid dynamics (CFD) study was carried out to study effect of M∞ on potential core. The CFD data with aircraft and plume was then used for IR signature analysis. The sensor position is changed with respect to aircraft from directly bottom towards frontal section of aircraft. The IR signature is studied in mid wave IR (MWIR) and long wave IR (LWIR) band. Findings The potential plume core length and width increases as M∞ increases. At higher altitudes, the potential core length increases for a fixed M∞. The plume emits radiation in the MWIR band, whereas the aerodynamically heated aircraft surface emits IR in the LWIR band. The IR signature in the MWIR band continuously decreases as the sensor position changes from directly bottom towards frontal. In the LWIR band the IR signature initially decreases as the sensor moves from the directly bottom to the frontal, as the sensor begins to see the wing leading edges and nose cone, the IR signature in the LWIR band slightly increases. Originality/value The novelty of this study comes from the data reported on the effect of free stream Mach number on the potential plume core and variation of the overall IR signature of aircraft with change in elevation angle from directly below towards frontal section of aircraft.

Prediction of instantaneous flow characteristics of hydrocyclone with long short-term memory network based on computational fluid dynamics data

Accelerating the prediction time of separation performance and flow field characteristics in industrial hydrocyclones holds paramount importance for real-time control. Machine learning methods exhibit significant advantages in this particular aspect. This study presents a network model that integrates a long short-term memory (LSTM) layer with a fully connected layer to accurately forecast the flow field and separation performance under various operational conditions, leveraging CFD data. The study evaluates the effect of factors such as the number of LSTM layers and neurons per layer on the model's performance, aiming to identify the optimal network structure. The results demonstrate that the predicted flow field and performance of the hydrocyclones using LSTM closely align with the outcomes from CFD simulations. The average absolute percentage error is constrained to within 6%, which significantly enhances the prediction efficiency. The findings contribute to a more efficient and automated separation process in industrial environments.

Modeling of UV-C survival of foodborne pathogens and predicting microbial inactivation on fresh-cut ‘Tommy Atkins’ mango using CFD

Abstract Shortwave ultraviolet light (UV-C) disinfection is an emerging technology used to enhance food safety by reducing the pathogen load. Computational fluid dynamics (CFD) served as a numerical simulation tool to calculate the average radiation intensity within a disinfection chamber. The resulting CFD data was employed to estimate the UV-C inactivation kinetic parameters for Escherichia coli O157:H7, Salmonella Typhimurium, and Listeria monocytogenes. Experimental procedures involved irradiating bacterial suspensions with UV-C doses ranging from 0 to 6.028 kJ/m2. The inactivation of S. Typhimurium was described using a log-linear equation, while UV-C survival curves for E. coli O157:H7 and L. monocytogenes were best fitted to Weibull model. Subsequently, the integration of CFD simulations and kinetic parameters enabled the estimation of UV-C doses approaching 6 kJ/m2 for the treatment of fresh-cut ‘Tommy Atkins’ mangoes inoculated with the mentioned microorganisms. This integrated approach partially predicted the inactivation of pathogens on the surface of mango spears.

Capturing mesoscale structures in computational fluid dynamics simulations of gas‐solid flows

AbstractMultiphase reactors' performance depends on the mesoscale structures formed due to multiphase hydrodynamics. Examples of mesoscale structures include gas bubbles in a fluidized bed and particle clusters in a riser. Experimental investigation of these mesoscale structures is challenging and expensive. To this end, computational fluid dynamics (CFD) simulations are extensively employed; however, post‐processing CFD data to capture mesoscale structures is challenging. This article develops a density‐based spatial clustering of applications with noise (DBSCAN)‐based methodology to capture and characterize mesoscale structures from multiphase CFD simulation data. DBSCAN is an unsupervised machine‐learning algorithm, which requires the value of two hyperparameters. A simple technique to calculate these hyperparameters is provided and the performance of DBSCAN is assessed on CFD‐DEM simulations of bubbling fluidized beds and particle clustering. We demonstrate the computational complexity of DBSCAN to be , lower than the existing techniques, by testing its scalability on highly resolved grids (up to 100 million grid points).

A semi-supervised framework for computational fluid dynamics prediction

Data-driven deep learning approach heavily relies on the diversity and quantity of data. Acquiring data in the computational fluid dynamics (CFD) domain is a time and computationally intensive process. This paper proposes a semi-supervised learning method called discriminative regression fitters (DRF) for aerodynamic prediction of airfoils. DRF utilizes neural networks’ memory property to dynamically divide pseudo-labeled data into easy and difficult subsets using a model of Gaussian distribution. The method classifies unlabeled data based on loss and updates the pseudo-labeled data, improving the model’s generalization capability. Experiments on airfoil regression task datasets show that DRF achieves similar or better prediction accuracy than fully supervised approaches. It reduces data acquisition time by 70%. Ablation studies and qualitative results verify the effectiveness of DRF. The surrogate model obtained from DRF is extended to airfoil optimization, demonstrating its practicality. DRF provides a promising direction for improving the regression task while reducing the reliance on large amounts of CFD data.

Modal analysis of vortex rope using dynamic mode decomposition

The decelerating swirling flow in the draft tube of hydraulic turbines at part load conditions often results in a self-induced instability known as vortex rope. This phenomenon is associated with detrimental pressure pulsations in the hydropower system that need to be mitigated. A deep understanding of such instability is essential for developing effective mitigation and control strategies. The current article exploits the dynamic mode decomposition (DMD) algorithm to perform an in-depth modal analysis of the physical aspects of the vortex rope. DMD can efficiently identify distinct coherent structures with isolated frequencies. The sparsity-promoting variant of DMD is exploited to extract the most influential modes. The computational fluid dynamics (CFD) data is generated via a resolved improved delayed detached eddy simulation using OpenFOAM. Frequency analysis of the CFD data uncovered peaks at the normalized frequencies of f/fn=0.56 and 0.63, whose origins seemed initially unclear. Nevertheless, the DMD modal analysis elucidates that these excitations are associated with the rotation of the reunited vortex and fluctuations of the separated boundary layer, respectively. The non-linear dynamics of the flow field are unveiled through a modal decomposition revealing distinctive coherent structures with isolated frequencies. These include rotational and plunging modes of the vortex rope, traveling wakes of the blades, boundary layer separation due to strong adverse pressure gradient, and a reunited vortex core. The flow field reconstruction through time dynamics of DMD modes highlights while it is possible to achieve a perfect flow field reconstruction considering all recovered modes, the model typically fails to predict future behavior with an acceptable level of accuracy. The chaotic nature of the resolved turbulent flow field presents a substantial challenge for predicting the future through a model built based on prior events. The current modal analysis not only provides a more comprehensive understanding of the physics underlying the vortex rope phenomenon but also lays the groundwork for potential future applications in controlling mechanisms.

Multi-objective optimization of a hypersonic airbreathing vehicle

Multi-objective optimization of a hypersonic airbreathing engine (scramjet technology) was carried out with the aim of maximizing thrust and minimizing drag while satisfying a series of design constraints, such as avoiding unstart (blockage of supersonic flow within the combustion chamber) by ensuring that the pressure ratio across the shock waves remains below the adverse pressure gradient given by the Korkegi limit, geometry correction to achieve shock on-lip condition, and temperature and pressure requirements at the inlet exit. Using the relations presented in the literature, pressure and viscous drag are estimated analytically. The analytical approach is verified against computational fluid dynamics data from Ansys Fluent to solve two-dimensional compressible Reynolds-averaged Navier–Stokes flow equations, with transition shear stress transport as the turbulence closure model. Comparing the total drag and the flow properties at the combustion chamber entrance shows the model's feasibility for the optimization approach. Three different approaches were conducted to formulate the multi-objective function to determine the one that can find the highest number of geometries satisfying the Korkegi limit with the highest net thrust. The best approach was the multi-objective function formulated with the uninstalled thrust, total pressure recovery, and pressure drag, concentrating the search in the region with greater uninstalled thrust and lower drag and nearly doubling the value of net thrust compared to the first formulation, which uses the uninstalled thrust, pressure drag, and viscous drag.

An axisymmetric characteristic model for enhancing prediction of centrifugal compressor characteristics

Rapid and accurate determination of compressor characteristic maps is essential for the initial design of centrifugal compressors in aircraft power systems. The accuracy of existing methodologies, which rely on combinations of loss models, varies significantly depending on the compressor’s geometry and operational range. This variance necessitates substantial experimental or Computational Fluid Dynamics (CFD) data for coefficient calibration. To address this challenge, this study presents an axisymmetric characteristic model for compressor performance assessment. This model incorporates the factors of blade angle, meridional passage area, and the radial deflection angle of meridional streamlines of the compressor. These factors are derived from fundamental aerodynamic equations encompassing mass, momentum, and energy conservation of the compressor. In contrast to conventional one-dimensional approaches, the proposed method reduces the number of loss coefficients and more effectively accounts for the impact of geometric alterations on centrifugal compressor properties. Furthermore, the model reduces dependence on experimental and CFD data. Efficacy of the model is validated using experimental data from four distinct types of centrifugal compressors. Correlation analysis reveals that the model’s coefficients can be expressed as functions of the ratio of the Reynolds number to the impeller tip speed. This ratio serves as a characteristic parameter for the design and optimization of centrifugal compressors. Consequently, the proposed method offers an efficient and accurate means for the quick computation of centrifugal compressor characteristics. This is of great significance for improving the efficiency of centrifugal compressors and reducing energy consumption.

Highly detailed structural integrity assessment of the reactor pressure vessel nozzle of Atucha-I during a pressurized thermal shock event

In pressurized water reactor (PWR) nuclear plants, reactor pressure vessel (RPV) nozzles are critical due to geometric stress concentrations that worsen with large temperature gradients. In the context of Atucha-I Long Term Operation (LTO) assessment, this work presents a high-fidelity thermo-mechanical simulation approach to verify the stress levels on the main nozzles of the vessel under a pressurized thermal shock event. Nucleoeléctrica Argentina S.A. (NA-SA) has provided the boundary conditions, including the thermal plume that results from computational fluid dynamics (CFD) simulations of the transient event. Then, using a one-way and non-conformal projection procedure, CFD data is coupled to a highly-detailed 3D finite element model (FEM), including the RPV, closure head, nuts and bolts, and moderator tank, all interconnected through non-linear contact condition that accounts for friction. Finally, a 3D FEM submodel assuming an elastoplastic material behavior with non-linear isotropic hardening is used to verify the structural integrity of the nozzle. Results indicate that the plastic strain and stresses on the nozzles are low enough to ensure the correct function of the reactor during the LTO period.

Spatially Resolved Modeling of the Nonlinear Dynamics of a Laminar Premixed Flame With a Multilayer Perceptron—Convolution Autoencoder Network

Abstract This work presents a multilayer perceptron-convolutional auto-encoder (MLP-CAE) neural network, which accurately predicts the two-dimensional flame dynamics of an acoustically excited premixed laminar flame. The architecture maps the acoustic perturbation time series into a heat release rate field, capturing flame lengths and shapes. This extends previous neural network models, which predicted only the field-integrated value. The MLP-CAE comprises two submodels: an MLP and a CAE. The idea behind the CAE network is to find a lower dimensional latent space of the heat release rate field. The MLP is responsible for modeling the flame dynamics by transforming the acoustic forcing signal into this latent space, enabling the decoder to produce the flow field distributions. To train the MLP-CAE, computational fluid dynamics (CFD) flame simulations with a broadband acoustic forcing were used. Its normalized amplitude was set to 0.5 and 1.0, ensuring a nonlinear flame response. The network was found to accurately predict the perturbed flame shapes. Additionally, it conserved the correct frequency response as verified by the global and local flame describing functions. The MLP-CAE provides a building block toward a potential shift away from a “0D” flame analysis with the acoustic compactness assumption. Combined with an acoustic network, the generated flame fields could provide more physical insight into the thermoacoustic dynamics. Those capabilities do not come at an additional significant computational cost, as even previous nonspatial flame models had to train on the CFD data, which included field distributions.

A Data-Driven Machine Learning Approach for Turbulent Flow Field Prediction Based on Direct Computational Fluid Dynamics Database

A novel approach is presented for predicting compressible turbulent flow fields using a neural network-based data-driven method. Accurate prediction in turbulent regions heavily relies on the resolution of available data. Traditional methods, employing image-based techniques by mapping scattered computational fluid dynamics (CFD) data onto Cartesian grids, encounter data scarcity in critical areas such as the boundary layer and wake. Recently, convolutional neural networks (CNN) have gained prominence as the most widely referenced technique in fluid dynamics, utilizing flow field images as datasets for flow field prediction. However, CNN requires datasets with a high pixel density to enhance training accuracy in crucial regions, thereby increasing the input data volume and machine training time. To address this challenge, our proposed method deviates from using flow field images and instead generates datasets directly from the flow field properties of CFD grid points. By employing this approach, several advantages are realized. Firstly, the network benefits from the favorable characteristics of unstructured grids, such as varying point spacing near the object surface and in the far field, which effectively reduces the amount of input data and consequently the machine training cost. Secondly, the construction of the training dataset eliminates the need for interpolation or extrapolation, thereby preserving the accuracy of CFD data. In this case, a simple multilayer perceptron can be trained using the proposed dataset. Various flow field properties, including static pressure, turbulent kinetic energy, and velocity components, can be predicted with high accuracy within a few seconds.

Establishment and validation of a relationship model between nozzle experiments and CFD results based on convolutional neural network

The acquisition of experimental data in a supersonic wind tunnel often faces challenges of complexity and high costs. Furthermore, there are limitations in the control of experimental conditions and measurement techniques. In this study, experiments were conducted on the start-up and shutdown processes of a single expansion ramp nozzle (SERN), and a dataset was established in conjunction with computational fluid dynamics (CFD). By utilizing experimental data and CFD simulation results, we employed a deep learning-based approach, utilizing one-dimensional convolutional neural networks (CNN), to establish a relationship model between the experimental and CFD results, with the aim of using CFD results to predict experimental outcomes. We provide detailed descriptions of the experimental methods, numerical simulation methods, and CNN model, and conduct training and testing of the model. The results demonstrate that the CNN model is capable of accurately predicting the wall pressure distribution of the SERN, with low prediction errors. In interpolation and extrapolation cases, the model exhibits good accuracy and generalization ability, with coefficients of determination above 96%. Compared to CFD calculations, the CNN model proves to be more reliable and accurate in predicting SERN performance. Additionally, we establish a comparative model that does not utilize CFD data to verify the superiority of the proposed CNN model. The results show that the CNN model with CFD data improves performance on average by 14%. This approach reduces reliance on experiments to a certain extent, lowering experimental costs and complexity. It not only saves the resources and time required for experiments but also compensates for the limitations in controlling experimental conditions and measurement techniques. Therefore, this research provides a viable solution to overcome the challenges and limitations of acquiring experimental data.

Nonlinear multispectral tomographic absorption deflection spectroscopy based on Bayesian estimation for spatially resolved multiparameter measurement in methane flame exhaust

In aerospace propulsion systems with high temperature, pressure, and distortion, inhomogeneous thermophysical gradients can cause line-of-sight deflection and spectral integral deviation. This introduces nonlinearity and measurement error into spectral models, impacting multiparameter field measurements. To address this, we developed a nonlinear multispectral tomographic absorption deflection spectroscopy (MTADS) technique, based on Bayesian inference, for temperature and gas concentration prediction. Our forward model considers inhomogeneous refractive index gradient and laser deflection, synthesising multispectral laser absorption signal through a gas mixing equation and multipoint ray tracing. Simulation results reveal inhomogeneous parameter distribution can cause non-Gaussian and non-zero mean deviations of the spectral absorption signal. The inverse model increases spectral lines to expand measurement signals under sparse optical arrangements. Combining the measured laser absorption signal with Bayesian statistical information of the estimated parameter field improves measurement robustness and accuracy. The maximum a posteriori (MAP) estimation introduces priors such as experiment samples, computational fluid dynamics data, and diffusive transport-induced spatial smoothing. A proof-of-concept based on methane diffusion flame data demonstrates the effectiveness of MTADS, enabling quantitative flame topology and parameter distribution, more accurate 2D temperature, and concentration imaging. These measurements are crucial for revealing chemical reaction dynamics and monitoring combustion systems.

Heat transfer enhancement of serpentine channels with twisted tape insert by computational fluid dynamics and artificial intelligence

AbstractImproving the thermal–hydraulic performance of heat exchangers is a critical challenge. In this study, we investigated the impact of twisted tape insert geometrical variables, namely pitch length (P) and diameter (W), on the Nusselt number (Nu) and friction factor (f) within serpentine channels. To accomplish this, we employed computational fluid dynamics (CFD) as our simulation technique, with water as the working fluid. The validity of the CFD data was confirmed, leading to the development of artificial intelligence (AI) subsets: artificial neural network (ANN) and genetic algorithm (GA). In the prediction models, the following input variables were considered: Reynolds number (Re), (P/Dh), and (W/Dh). The ANN and GA models achieved a mean relative error (MRE) of 0.125% and 1.326% for Nu, and 0.139% and 4.104% for f, respectively. These results demonstrate the high accuracy of the correlations, with the ANN model showcasing superior performance.

Is well-mixed model of an indoor space with ceiling fans valid for studying pathogen transmission?

Airborne transmission is one of the main modes for the transmission of highly infectious diseases such as COVID-19. Pathogen laden aerosols from an infected person can be transported by air to a susceptible population. A widely used model for airborne transmission considers the indoor space to be well-mixed such that the pathogen concentration is spatially homogeneous. Other models that employ computational fluid dynamics (CFD) allow tracking the spatiotemporal variation of infection probability in indoor spaces but are computationally expensive. Here, we compare the predictions of a well-mixed continuously stirred tank reactor (CSTR) model for indoor transmission with CFD for airflow, along with the Lagrangian tracking of aerosol particles. Of particular interest is the ventilation using ceiling fans, which are common in South East Asia. It is found that the behavior of particles at the walls plays an important role. Two limiting cases are studied: all particles reaching the wall get trapped vs all particles being reflected from the solid boundaries. We propose a modification to the CSTR equation to include the wall effect, and it matches the CFD data closely.