Computational Fluid Dynamics Data Sets Research Articles

Deep neural network model for predicting thermal-hydraulic performance of a solar air heater with artificial roughness: Sensitivity, generalization capacity, and computational efficiency

A novel coupled computational fluid dynamics-deep neural network approach is proposed to accurately predict the performance of an artificially roughened solar air heater. Computational fluid dynamics data sets are used to develop an optimized neural network model for the prediction of thermal-hydraulic performance factor. The optimized neural network model with softmax transfer function in hidden layers has an architecture of 3‐37‐37‐1 and it predicts the thermal-hydraulic performance factor of the artificially roughened solar air heater with maximum, mean, and minimum error of 2.54 %, 0.245 %, and 0.0009 %, respectively. The concept of Shapley values is adopted to evaluate the global sensitivity analysis of the optimized deep neural network model and it is found that rib height is the most influential design parameter for the artificially roughened solar air heater. The generalization capacity of the neural network model is evaluated by comparing its predictions with experimental data, demonstrating superior predictive performance over traditional computational fluid dynamics simulations. The optimized artificial neural network model achieves a computational time gain of approximately 100 % against computational fluid dynamics.

Stacked Deep Learning Models for Fast Approximations of Steady-State Navier–Stokes Equations for Low Re Flow

Computational fluid dynamics (CFD) simulations are broadly used in many engineering and physics fields. CFD requires the solution of the Navier–Stokes (N-S) equations under complex flow and boundary conditions. However, applications of CFD simulations are computationally limited by the availability, speed, and parallelism of high-performance computing. To address this, machine learning techniques have been employed to create data-driven approximations for CFD to accelerate computational efficiency. Unfortunately, these methods predominantly depend on large labeled CFD datasets, which are costly to procure at the scale required for robust model development. In response, we introduce a weakly supervised approach that, through a multichannel input capturing boundary and geometric conditions, solves steady-state N-S equations. Our method achieves state-of-the-art results without relying on labeled simulation data, instead using a custom data-driven and physics-informed loss function and small-scale solutions to prime the model for solving the N-S equations. By training stacked models, we enhance resolution and predictability, yielding high-quality numerical solutions to N-S equations without hefty computational demands. Remarkably, our model, being highly adaptable, produces solutions on a 512 × 512 domain in a swift 7 ms, outpacing traditional CFD solvers by a factor of 1,000. This paves the way for real-time predictions on consumer hardware and Internet of Things devices, thereby boosting the scope, speed, and cost-efficiency of solving boundary-value fluid problems.

Development and evaluation of an immersive VR-CFD-Based tool for dust exposure mitigation in underground tunnelling operations

Long-term exposure to respirable coal and silica dust during underground tunnelling operations has gained increasing attention in recent years. The solution to effective mitigation of dust exposure depends not only on higher-order engineering controls, but also on administrative controls for frontline workers. However, there is a disconnect between knowledge gathered in the field of dust exposure monitoring and the frontline worker, resulting in important learnings being overlooked in underground tunnelling operations. To remedy this discrepancy, an immersive educational tool was developed to visualise computational fluid dynamics (CFD) modelling datasets of ventilation and respirable dust flow characteristics in the tunnel face in a virtual reality (VR) environment. An algorithm was developed for processing the large and complex CFD datasets into a form that can be processed and visualised using standalone VR headsets with limited processing power. The VR-CFD system was assessed by an industry expert and via many industry showcases, where regular feedback was received for making significant improvements in this education tool. This tool has been developed as a training platform to allow frontline workers to better understand the results of decisions made during tunnelling operations and the best practices for dust controls. A number of key technological achievements were made that can be used in the future to quickly translate real-world particulate readings collected from underground space into a VR visualiser. This VR-CFD digital technology can readily be extended to mining, construction and other tunnelling operations in the underground space, thus improving health and safety.

A physics-based machine learning technique rapidly reconstructs the wall-shear stress and pressure fields in coronary arteries.

With the global rise of cardiovascular disease including atherosclerosis, there is a high demand for accurate diagnostic tools that can be used during a short consultation. In view of pathology, abnormal blood flow patterns have been demonstrated to be strong predictors of atherosclerotic lesion incidence, location, progression, and rupture. Prediction of patient-specific blood flow patterns can hence enable fast clinical diagnosis. However, the current state of art for the technique is by employing 3D-imaging-based Computational Fluid Dynamics (CFD). The high computational cost renders these methods impractical. In this work, we present a novel method to expedite the reconstruction of 3D pressure and shear stress fields using a combination of a reduced-order CFD modelling technique together with non-linear regression tools from the Machine Learning (ML) paradigm. Specifically, we develop a proof-of-concept automated pipeline that uses randomised perturbations of an atherosclerotic pig coronary artery to produce a large dataset of unique mesh geometries with variable blood flow. A total of 1,407 geometries were generated from seven reference arteries and were used to simulate blood flow using the CFD solver Abaqus. This CFD dataset was then post-processed using the mesh-domain common-base Proper Orthogonal Decomposition (cPOD) method to obtain Eigen functions and principal coefficients, the latter of which is a product of the individual mesh flow solutions with the POD Eigenvectors. Being a data-reduction method, the POD enables the data to be represented using only the ten most significant modes, which captures cumulatively greater than 95% of variance of flow features due to mesh variations. Next, the node coordinate data of the meshes were embedded in a two-dimensional coordinate system using the t-distributed Stochastic Neighbor Embedding (-SNE) algorithm. The reduced dataset for -SNE coordinates and corresponding vector of POD coefficients were then used to train a Random Forest Regressor (RFR) model. The same methodology was applied to both the volumetric pressure solution and the wall shear stress. The predicted pattern of blood pressure, and shear stress in unseen arterial geometries were compared with the ground truth CFD solutions on "unseen" meshes. The new method was able to reliably reproduce the 3D coronary artery haemodynamics in less than 10 s.

Optimization of painting efficiency applying unique techniques of high-voltage conductors and nitrotherm spray: Developing deep learning models using computational fluid dynamics dataset

The impetus of the current three-dimensional Eulerian–Lagrangian work is to analyze the impact of simultaneously using the inventive high-voltage conductors and Nitrotherm spraying technique for maximizing the industrial painting process efficiency. This investigation employs high-fidelity computational fluid dynamics (CFD) results in deep learning models as an input dataset. The novel conductors are called high-voltage retractable blades (HVRB) and high-voltage adjustable control-ring (HVACR) mounted on the head of the electrostatic rotating bell sprayer. The influence of dominant operational parameters, such as temperature and velocity of injected nitrogen or air, droplets' electric charge values, and their size ranges, and electric field density are examined in the considered database for the Nitrotherm spraying methodology. This broad range of parametric investigation illustrates that the inclusion of shaping nitrogen flow, manipulated electric field density, and droplet charging weights significantly affect the spraying deposition rate. The pressurized clean heated nitrogen flow, which is injected from the nozzles of the atomizers, positively redirects and harmonizes the charged droplets that construct an optimized spray plume pattern with a smaller diameter. Using innovative HVRB and HVACR conductors is manipulated the electric fields and leads to denser distribution, intensifying the acting electric force on the droplets, resulting in higher spraying transfer efficiency (TE) and thicker film formation. Based on the results, employing the introduced conductors in combination with the heated nitrogen instead of air leads to higher TE, rare overspray occurrence, formation of an esthetic paint film, lower paint consumption, and application time. Also, the collected complete database is employed for machine learning investigation to predict flow with high accuracy, aiming to reduce computational time/cost. A convolutional auto-encoder is used to reduce the computational cost with just 10% of the initial CFD computations, with a mean error of 1% on the prediction of the deposited droplet areas of the spray. The analysis revealed that by employing recurrent convolutional layers, superior capturing of the input pattern is obtained, which significantly aids the final prediction.

Modeling Nonlinear Aeroelastic Forces for Bridge Decks with Various Leading Edges Using LSTM Networks

With the rapid increase in bridge spans, the mitigation of risk to flutter (aeroelastic instability) is of critical importance in the design of long-span bridges, especially considering the more frequent intense hurricanes under climate change. Although the strong nonlinearities of the aeroelastic (self-excited) forces in wind–bridge interactions can be well captured through either numerical simulations or experimental tests, both are expensive and time consuming. Hence, it is important to develop an efficient reduced-order model for the simulations of nonlinear aeroelastic forces on the bridge decks. This study proposes a reduced-order model based on the long short-term memory (LSTM) network to simulate the nonlinear aeroelastic forces on bridge decks with various leading edges, and thus rapidly predict the corresponding post-flutter behaviors of long-span bridges. To generate the training datasets, computational fluid dynamics (CFD) was employed to simulate the nonlinear aeroelasticities of bridge decks with a wide range of leading-edge configurations and wind speeds. Trained on the high-fidelity CFD datasets, the LSTM network takes the motion of a bridge deck, leading-edge angles and wind speeds as inputs and outputs the nonlinear aeroelastic forces on the bridge decks. A hybrid loss function utilizing the prediction errors of both aeroelastic forces simulated by the LSTM network and the bridge deck responses calculated by the Newmark-β algorithm was introduced into the training process to improve the network performance. The prediction results of the trained LSTM model were compared with the CFD simulations, which demonstrated that the nonlinear aeroelastic forces of the bridge deck with various leading edges can be accurately and efficiently acquired by the proposed LSTM model.

Metamodel-Interpreted Data Mining for Stability and Efficiency Enhancement of Multistage Axial-Flow Compressors

Abstract The tradeoff between higher efficiency and wider stability of performance map is still one of the bottlenecks to hamper the further research and development of advanced multistage axial-flow compressor. The recent rapid growth of computational resources and artificial intelligence has enabled data mining as one of the most effective and potential ways to gain a deep insight into the complex correlations between aerodynamic performance and three-dimensional geometry parameters. In the open literatures, however, few research works have been found on using the data mining that is independent of design optimization to extract priori design guidelines for multistage axial-flow compressor mainly due to the lack of proper data mining method focused on the interpretation of metamodel with full use of limited time-consuming computational fluid dynamics dataset. To tackle this issue, a metamodel-interpreted data mining framework is developed in which extreme gradient boosting (XGBoost) metamodel combined with Shapley additive explanation (SHAP) model are employed to locally interpret the feature importance of each sample in the computational fluid dynamics dataset and then extract the design guidelines in terms of the most influential geometry parameters and their beneficial variation directions. The developed method is applied to data mining of design guidelines for efficiency and stability enhancement of a front 3.5-stage transonic axial-flow compressor in ship-board gas turbine usage. The results show that the aerodynamic performance of the investigated multistage compressor is most sensitive to three-dimensional geometry parameters related to blade lean, blade twist, and variable stators. Specially, the variable stators mainly affect the stall margin at part speed. The blade lean mainly influences the adiabatic efficiency at design speed as well as the stall margin at both speeds, while the blade twist mainly influences the aerodynamic performance at design speed. New designs followed by the design guidelines are obtained and critical performance indicators related to the goals of the data mining task are verified. The stall margin at part speed is widened to 5.87% with adjustment of blade lean and twist and further to 23.31% with additional adjustment of variable stators. The peak adiabatic efficiency at design speed is improved by 0.06% in spite of extremely limited potential for efficiency enhancement of the original design. The present work is of scientific significance as well as industrial application value in the three-dimensional design optimization of advanced multistage axial-flow compressor at the affordable computational cost.

Spectral Bandedness in High-Fidelity Computational Fluid Dynamics Predicts Rupture Status in Intracranial Aneurysms.

Recent studies using high-fidelity computational fluid dynamics (CFD) have revealed high-frequency flow instabilities consistent with clinical reports of bruits and "musical murmurs", which have been speculated to contribute to aneurysm growth and rupture. We hypothesized that harmonic flow instabilities ("spectral bandedness") in aneurysm CFD data may be associated with rupture status. Before testing this hypothesis, we first present a novel method for quantifying and visualizing spectral bandedness in cardiovascular CFD datasets based on musical audio-processing tools. Motivated by previous studies of aneurysm hemodynamics, we also computed a selection of existing metrics that have demonstrated association with rupture in large studies. In a dataset of 50 bifurcation aneurysm geometries modeled using high-fidelity CFD, our spectral bandedness index (SBI) was the only metric significantly associated with rupture status (AUC = 0.76, p = 0.002), with a specificity of 79% (correctly predicting 19/24 unruptured cases) and sensitivity of 65% (correctly predicting 17/26 ruptured cases). Three-dimensional flow visualizations revealed coherent regions of high SBI to be associated with strong near-wall inflow jets and vortex-shedding/flutter phenomena in the aneurysm sac. We speculate that these intracycle, coherent flow instabilities may preferentially contribute to the progressive degradation of the aneurysm wall through flow-induced vibrational mechanisms, and that their absence in high-fidelity CFD may be useful for identifying intracranial aneurysms at lower risk of rupture.

Physics-Informed Transfer Learning Strategy to Accelerate Unsteady Fluid Flow Simulations

Since the derivation of the Navier Stokes equations, it has become possible to numerically solve real world viscous flow problems (computational fluid dynamics (CFD)). However, despite the rapid advancements in the performance of central processing units (CPUs), the computational cost of simulating transient flows with extremely small time/grid scale physics is still unrealistic. In recent years, machine learning (ML) technology has received significant attention across industries, and this big wave has propagated various interests in the fluid dynamics community. Recent ML CFD studies have revealed that completely suppressing the increase in error with the increase in interval between the training and prediction times in data driven methods is unrealistic. The development of a practical CFD acceleration methodology that applies ML is a remaining issue. Therefore, the objectives of this study were developing a realistic ML strategy based on a physics-informed transfer learning and validating the accuracy and acceleration performance of this strategy using an unsteady CFD dataset. This strategy can determine the timing of transfer learning while monitoring the residuals of the governing equations in a cross coupling computation framework. Consequently, our hypothesis that continuous fluid flow time series prediction is feasible was validated, as the intermediate CFD simulations periodically not only reduce the increased residuals but also update the network parameters. Notably, the cross coupling strategy with a grid based network model does not compromise the simulation accuracy for computational acceleration. The simulation was accelerated by 1.8 times in the laminar counterflow CFD dataset condition including the parameter updating time. Open source CFD software OpenFOAM and open-source ML software TensorFlow were used in this feasibility study.

Hemodynamic Study of a Patient-Specific Intracranial Aneurysm: Comparative Assessment of Tomographic PIV, Stereoscopic PIV, In Vivo MRI and Computational Fluid Dynamics

IntroductionWall shear stress (WSS) is associated with the growth and rupture of an intracranial aneurysm. To reveal their underlying connections, many image-based computational fluid dynamics (CFD) studies have been conducted. However, the methodological validations using both in vivo medical imaging and in vitro optical flow measurements were rarely accompanied in such studies.MethodsIn the present study, we performed a comparative assessment on the hemodynamics of a patient-specific intracranial saccular aneurysm using in vivo 4D Flow MRI, in silico CFD, in vitro stereoscopic and tomographic particle imaging velocimetry (Stereo-PIV and Tomo-PIV) techniques. PIV experiments and CFD were conducted under steady state corresponding to the peak systole of 4D Flow MRI.ResultsThe results showed that all modalities provided similar flow features and overall surface distribution of WSS. However, a large variation in the absolute WSS values was found. 4D Flow MRI estimated a 2- to 4-fold lower peak WSS (3.99 Pa) and a 1.6- to 2-fold lower mean WSS (0.94 Pa) than Tomo-PIV, Stereo-PIV, and CFD. Bland-Altman plots of WSS showed that the differences between PIV-/CFD-based WSS and 4D Flow MRI-based WSS increase with higher WSS magnitude. Such proportional trend was absent in the Bland-Altman comparison of velocity where the resolutions of PIV and CFD datasets were matched to 4D Flow MRI. We also found that because of superior resolution in the out-of-plane direction, WSS estimation by Tomo-PIV was higher than Stereo-PIV.ConclusionsOur results indicated that the differences in spatial resolution could be the main contributor to the discrepancies between each modality. The findings of this study suggest that with current techniques, care should be taken when using absolute WSS values to perform a quantitative risk analysis of aneurysm rupture.

Multi‐objective optimization of cane sugar continuous crystallization system design based on computational fluid dynamics

This work focused on improving circulation and mixing of the massecuite and reducing the energy loss in the cane sugar continuous crystallization system. The developed Computational Fluid Dynamics (CFD) model is based on the continuous crystallization system of a sugarcane mill in Guangxi, China and verified with the actual operation data. The calculated entropy production and pressure drop of the system are used as indices for assessing the performance of the system. Next, 350 CFD simulations are conducted in the parametric experiment platform with 9 parameters and produces a collection of data. Then based on the CFD data set, the data-driven model is employed for regression of the relations between key parameters and indices. The implemented data-driven model is used for Non-dominated Sorting Genetic Algorithm (NSGA-II) to obtain the optimal result of Pareto frontier. CFD simulation with the optimized parameters reduces entropy by 9.76% and reduces pressure drop by 11.52%. Novelty impact statement Multi-objective optimization of cane sugar continuous crystallization system is performed. The entropy is reduced by 9.76% and the pressure drop is reduced up to 11.52%. The proposed method show the good applicability.

Prediction model of flow field in an isolator over various operating conditions

The flow characteristics in a scramjet isolator are quite complicated, so it is difficult to achieve precise operating state detection on a scramjet isolator by wall pressure only. Flow field prediction using deep learning is a promising method to provide a rich source of information for isolator operating state detection. A data-driven model is proposed for the prediction of the flow field in an isolator by fusion convolutional neural networks using measurements of the pressure on the isolator. Numerical simulations of flow in an isolator at various Mach numbers and backpressures are carried out to establish the dataset. A convolutional neural network architecture composed of a fusion of paths is designed. The convolutional neural network is trained using the computational fluid dynamics dataset to establish the mapping relationship between the wall pressure on the isolator and the flow field in the isolator, including the Mach number field, the pressure field, and the temperature field. The trained model is then tested over various Mach numbers and backpressures. The relative error of the prediction results in most areas does not exceed 5%, and the correlation coefficients between prediction results and computational fluid dynamic results are all over 0.99 in the whole testing set. The predictions of this model are found to agree well with the computational fluid dynamics results, i.e., the data-driven model successfully learns the relationship between internal flow field and pressure experienced on the wall of an isolator.

A deep learning approach for the velocity field prediction in a scramjet isolator

The accurate parameter prediction of a flow field is of practical significance to promote the development of hypersonic flight. Velocity field prediction using deep learning is a promising method to provide an accurate velocity field in a scramjet isolator. A new approach for the velocity field prediction in a scramjet isolator is developed in this study. A data-driven model is proposed for the prediction of the velocity field in a scramjet isolator by convolutional neural networks (CNNs) using measurements of the pressure on the isolator. Numerical simulations of flow in a three-dimensional scramjet isolator at various Mach numbers and backpressures are carried out to establish the dataset capturing the flow mechanism over various operating conditions. A CNN architecture composed of multiple reconstruction modules and feature extraction modules is designed. The CNN is trained using the computational fluid dynamics dataset to establish the mapping relationship between the wall pressure on the isolator and the velocity field in the isolator. The trained model is then tested over various Mach numbers and backpressures. The data-driven model successfully learns the relationship between the velocity field and pressure experienced on the wall of the isolator, i.e., the trained CNN model successfully reconstructed the velocity field based on the wall pressure on the isolator with high accuracy.

Deep Regression Network-Assisted Efficient Streamline Generation Method

Streamlining is one of the most frequently utilized visualization methods to analyze the flow structure of computational fluid dynamics (CFD) data. However, it is challenging to find a set of streamlines showing the most prominent flow across the entire flow field due to the heavy computation time required to generate bundles of streamlines. In this paper, we propose an efficient streamline generation method that removes several seed candidates that are predicted as less important using a 3D U-net based regression model. We employ 3D line integral convolution (LIC) volumes that depict the entire flow field for training data of the proposed learning model and evaluate our method using a real-world CFD data set. We find using our model that we can obtain quality of visualization results comparable to that of the ground truth even when more than 90% of the seed candidates are truncated while operating 6.6~17.1 times faster than the competing method.

Influence of number of membership functions on prediction of membrane systems using adaptive network based fuzzy inference system (ANFIS)

In membrane separation technologies, membrane modules are used to separate chemical components. In membrane technology, understanding the behavior of fluids inside membrane module is challenging, and numerical methods are possible by using computational fluid dynamics (CFD). On the other hand, the optimization of membrane technology via CFD needs time and computational costs. Artificial Intelligence (AI) and CFD together can model a chemical process, including membrane technology and phase separation. This process can learn the process by learning the neural networks, and point by point learning of CFD mesh elements (computing nodes), and the fuzzy logic system can predict this process. In the current study, the adaptive neuro-fuzzy inference system (ANFIS) model and different parameters of ANFIS for learning a process based on membrane technology was used. The purpose behind using this model is to see how different tuning parameters of the ANFIS model can be used for increasing the exactness of the AI model and prediction of the membrane technology. These parameters were changed in this study, and the accuracy of the prediction was investigated. The results indicated that with low number of inputs, poor regression was obtained, less than 0.32 (R-value), but by increasing the number of inputs, the AI algorithm led to an increase in the prediction capability of the model. When the number of inputs increased to 4, the R-value was increased to 0.99, showing the high accuracy of model as well as its high capability in prediction of membrane process. The AI results were in good agreement with the CFD results. AI results were achieved in a limited time and with low computational costs. In terms of the categorization of CFD data-set, the AI framework plays a critical role in storing data in short memory, and the recovery mechanism can be very easy for users. Furthermore, the results were compared with Particle Swarm Optimization (PSOFIS), and Genetic Algorithm (GAFIS). The time for prediction and learning were compared to study the capability of the methods in prediction and their accuracy.

Optimization of submerged floating tunnel cross section based on parametric Bézier curves and hybrid backpropagation - genetic algorithm

The cross-section geometry of a submerged floating tunnel (SFT) has a large effect on hydrodynamic characteristics, structural behavior and service level, making the tunnel cross section the primary factor in optimizing efficiency. Minimizing the mean drag and the dynamic variability in the lift of the SFT cross section under bi-directional (i.e., tidal) flow has a dramatic impact on the reduction of structural displacements and mooring loads. Based on a parametric Bézier curve dynamically comprising the leading-edge radius, tunnel height and width to define the SFT geometry, a sensitivity analysis of the Bézier curve parameters for a fixed aspect ratio with prototype dimensions under uniform flow conditions was conducted by applying Computational Fluid Dynamics (CFD), and the pressure distribution around the SFT cross-section surface was analyzed. A theoretical method comprising the Kármán vortex street parameters was employed to verify the CFD simulation results. In order to determine the SFT cross section with optimal hydrodynamic properties, the mean drag and Root Mean Square (RMS) lift coefficients were selected as optimization objectives, and four Bézier curve parameters were the input variables, in a neural network and genetic algorithm optimization process (a hybrid BP-GA structure), which is less likely to become trapped in local minima. The results show the optimal tunnel cross section has a mean drag and a RMS lift coefficient reduced by 0.9% and 6.3%, respectively, compared to the original CFD dataset.

Feature representations using the reflected rectified linear unit (RReLU) activation

Deep Neural Networks (DNNs) have become the tool of choice for machine learning practitioners today. One important aspect of designing a neural network is the choice of the activation function to be used at the neurons of the different layers. In this work, we introduce a four-output activation function called the Reflected Rectified Linear Unit (RReLU) activation which considers both a feature and its negation during computation. Our activation function is “sparse”, in that only two of the four possible outputs are active at a given time. We test our activation function on the standard MNIST and CIFAR-10 datasets, which are classification problems, as well as on a novel Computational Fluid Dynamics (CFD) dataset which is posed as a regression problem. On the baseline network for the MNIST dataset, having two hidden layers, our activation function improves the validation accuracy from 0.09 to 0.97 compared to the well-known ReLU activation. For the CIFAR-10 dataset, we use a deep baseline network that achieves 0.78 validation accuracy with 20 epochs but overfits the data. Using the RReLU activation, we can achieve the same accuracy without overfitting the data. For the CFD dataset, we show that the RReLU activation can reduce the number of epochs from 100 (using ReLU) to 10 while obtaining the same levels of performance.

Key time steps selection for CFD data based on deep metric learning

As one of the main technologies of flow visualization, key time steps selection plays a key role in solving storage limit and has been intensively studied. In this paper, we introduce Deep Metric Learning (DML) into key time steps selection for Computational Fluid Dynamics (CFD) data and propose a local selection method based on DML. In specific, the proposed method samples small patches from CFD data, trains a Siamese deep neural network which has a symmetry structure with two Convolutional Neural Networks (CNN), and then selects the key time steps according to the similarities between consecutive time steps which are assessed by the networks. Compared with one of the existing local selection methods, the Myers’s method, our method has advantages in accuracy, precision and recall, and the selection results are better. Experimental results also demonstrate the good generalization of the proposed method on CFD datasets.

Development of an adaptive discharge coefficient to improve the accuracy of cross-ventilation airflow calculation in building energy simulation tools

Airflow network (AFN) model embedded in building energy simulation tools such as EnergyPlus is extensively used for prediction of cross-ventilation in buildings. The noticeable uncertainty in the measurement of the surface pressure, discharge coefficient, and simplifications applied to the orifice-based equation result in considerable discrepancies in the prediction of the cross-ventilation airflow rate. Computational Fluid Dynamics (CFD) provides more accurate results comparing to the orifice-based equations although with an excessive computational cost.The aim of this study is to improve the accuracy of the orifice-based model by development of an adaptive correlation for the discharge coefficient using CFD. Hence, a validated CFD model for the cross-ventilation of an unsheltered building is firstly developed using an experimental study. In the next step, by exploiting Latin hypercube sampling approaches, a large CFD dataset of 750 scenarios for different building geometries (i.e. square cube, cuboid and long corridor) is generated; the dataset is then coupled to the AFN cross-ventilation model to obtain an adaptive correlation for the discharge coefficient as a function of the openings' geometries and location using response surface and radial basis function models.Results show that the newly developed adaptive correlation successfully increases the accuracy of AFN model for the cross-ventilation modeling of unsheltered buildings as the relative errors for the airflow rate prediction of different building geometries are decreased up to 28% in comparison with the cases with constant discharge coefficient. Results also demonstrate the importance of considering the value of the local-surface wind pressure in the AFN model.

Methodology for Computational Fluid Dynamic Validation for Medical Use: Application to Intracranial Aneurysm.

Computational fluid dynamics (CFD) is a promising tool to aid in clinical diagnoses of cardiovascular diseases. However, it uses assumptions that simplify the complexities of the real cardiovascular flow. Due to high-stakes in the clinical setting, it is critical to calculate the effect of these assumptions in the CFD simulation results. However, existing CFD validation approaches do not quantify error in the simulation results due to the CFD solver's modeling assumptions. Instead, they directly compare CFD simulation results against validation data. Thus, to quantify the accuracy of a CFD solver, we developed a validation methodology that calculates the CFD model error (arising from modeling assumptions). Our methodology identifies independent error sources in CFD and validation experiments, and calculates the model error by parsing out other sources of error inherent in simulation and experiments. To demonstrate the method, we simulated the flow field of a patient-specific intracranial aneurysm (IA) in the commercial CFD software star-ccm+. Particle image velocimetry (PIV) provided validation datasets for the flow field on two orthogonal planes. The average model error in the star-ccm+ solver was 5.63 ± 5.49% along the intersecting validation line of the orthogonal planes. Furthermore, we demonstrated that our validation method is superior to existing validation approaches by applying three representative existing validation techniques to our CFD and experimental dataset, and comparing the validation results. Our validation methodology offers a streamlined workflow to extract the "true" accuracy of a CFD solver.