Computational Fluid Dynamics Simulation Results Research Articles

Design and analysis of cooling jacket developed for vacuum power tubes by multiphase cooling

The vacuum tubes are used extensively in electronic industries. The MW level of vacuum tube are used in RF amplifiers and microwave sources in different sectors like defence, nuclear energy, satellites and medical diagnostics. The first step in the indigenous development of MW level of vacuum tube is the design of its cooling jacket. The cooling jacket of vacuum tubes use hypervapotron (HV) fins which are known for providing efficient cooling in compact space. The key objectives of this paper are to finalize the jacket’s construction material, evaluate the cooling jacket’s heat transfer performance at the MW level of RF (radiofrequency) power, and forecast safe operating conditions for given conditions of high heat flux and water flow rates. In this design, the vacuum tubes’ external dimensions are taken into consideration when designing the cooling jacket. ETP copper (Electrolytic Tough Pitch copper) and copper-chromium-zirconium (CuCrZr) are likely materials for the jacket’s construction. For the aforementioned design of the cooling jacket, FEA (Finite Element Analysis) simulations using a steady-state thermal module and computational fluid dynamics (CFD) simulations using an ANSYS CFX module are described for flow rates of 15, 30, and 60 lpm (liters per minute) and the heat flux 1, 2, 4, and 6 MW m−2. The results of CFD and FEA simulations are found to be in close agreement. A small deviation in results is seen after the start of nucleate boiling. 30 and 60 lpm are desirable flow rates for high heat flux values greater than 2 MW m−2. CuCrZr is the ideal material for a cooling jacket’s construction as it has safe working temperature limit of 350 °C at high heat flux values. Also, 15 lpm flow rate should be avoided while using this cooling jacket at heat flux values greater than or equal to 6 MW m−2.

An Analysis Model for Predicting Windage Power Loss of Aviation Spiral Bevel Gears Under Optimal Injection Jet Layout

With the increasing rotating speed of aviation spiral bevel gears, the load-independent windage power loss caused by hydrodynamic behavior has a greater impact on transmission efficiency. An analytical model is established to predict the windage power loss of spiral bevel gears with oil injection lubrication. The improved model regards the total windage losses as the sum of oil–gas dragging effects on the tooth flank, toe/heel, circular cone surface, circumference of the teeth, and tooth root. Then the numerical method and analytical method are combined to calculate windage losses. A multi-objective NSGA-II optimization algorithm is used to optimize the jet layout parameters. The rationality and accuracy of this model are verified by comparing the calculation results of an example with data sets of existing experimental findings and computational fluid dynamics (CFD) simulation results. The calculation results show that compared with the tooth profile assumption method in the existing literature, the improved calculation formula of windage dragging effect on tooth flank can better reflect the influence of oil injection lubrication layout parameters on windage loss and make the windage loss calculation more accurate. Finally, the influence of gear body parameters, working conditions parameters, geometric parameters, and injection lubrication layout parameters on the windage loss is examined. This research provides a theoretical basis and methodological guidance for optimization design of reducing windage power loss.

Design and Implementation of Temperature-sensitive Particle Temperature Measuring Structure in Microchannel

Any physical change or chemical reaction is closely related to temperature. Especially in the context of microscale continuous flow, the uniform temperature distribution affects the whole process of the reaction, so it is very important to detect the temperature in the microsystem. Temperature-sensitive particle temperature measurement is a kind of micro-scale non-contact temperature measurement. At present, the temperature-sensitive particle temperature measurement technology is mainly based on the direct design of experiments based on non-simulation. Starting from the microscopic point of view, this paper firstly analyzes the principle of fluorescence temperature measurement, establishes the consistency between the temperature-luminescence intensity relationship and the force-luminescence intensity relationship, and then uses the simulation software to build a microchannel temperature measurement model to analyze the temperature measurement effect of the selected temperature-sensitive particles. By comparing the Computational Fluid Dynamics (CFD) simulation results, the design of a non-contact particle temperature measuring structure inside the micro-channel is realized, which provides a feasible method in simulation.

A Parameter Correction method of CFD based on the Approximate Bayesian Computation technique

Numerical simulation and modeling techniques are becoming the primary research tools for aerodynamic analysis and design. However, various uncertainties in physical modeling and numerical simulation seriously affect the credibility of Computational Fluid Dynamics (CFD) simulation results. Therefore, CFD models need to be adjusted and modified with consideration of uncertainties to improve the prediction accuracy and confidence level of CFD numerical simulations. This paper presents a parameter correction method of CFD for aerodynamic analysis by making full use of the advantages of the Approximate Bayesian Computation (ABC) technique in dealing with the analysis and inference of complex statistical models, in which the parameters of turbulence models for CFD are inferenced. The proposed parameter correction method is applied to the aerodynamic prediction of the NACA0012 airfoil. The results show the feasibility and effectiveness of the proposed approach in improving CFD prediction accuracy.

Generating simplified ammonia reaction model using genetic algorithm and its integration into numerical combustion simulation of 1 MW test facility

A small reaction model for ammonia is necessary for the design and development of large-scale ammonia combustion systems using computational fluid dynamics (CFD) simulation. The present study generated a simplified reaction model for ammonia using a genetic algorithm (14 species and 45 reactions). Optimization was applied to reproduce the combustion properties of ignition delay times, mole fractions of ammonia and nitric oxide at the exit of a single perfectly stirred reactor (PSR), those of a two-staged PSR, and laminar flame speed. The generated simplified reaction model for ammonia was applied to combustion CFD simulation for a 1 MW boiler test facility. Results of CFD simulation reproduced the characteristics of two-staged combustion well. Nitrous oxide emissions predicted by the simulation agreed well with experimental results.

Exploring the influence of nasal vestibule structure on nasal obstruction using CFD and Machine Learning method

Motivated by clinical findings about the nasal vestibule, this study analyzes the aerodynamic characteristics of the nasal vestibule and attempt to determine anatomical features which have a large influence on airflow through a combination of Computational Fluid Dynamics (CFD) and machine learning method. Firstly, the aerodynamic characteristics of the nasal vestibule are detailedly analyzed using the CFD method. Based on CFD simulation results, we divide the nasal vestibule into two types with distinctly different airflow patterns, which is consistent with clinical findings. Secondly, we explore the relationship between anatomical features and aerodynamic characteristics by developing a novel machine learning model which could predict airflow patterns based on several anatomical features. Feature mining is performed to determine the anatomical feature which has the greatest impact on respiratory function. The method is developed and validated on 41 unilateral nasal vestibules from 26 patients with nasal obstruction. The correctness of the CFD analysis and the developed model is verified by comparing them with clinical findings.

The effect of natural ventilation on airborne transmission of the COVID-19 virus spread by sneezing in the classroom

Since school classrooms are of vital importance due to their impact on public health in COVID-19 and similar epidemics, it is imperative to develop new ventilation strategies to reduce the risk of transmission of the virus in the classroom. To be able to develop new ventilation strategies, the effect of local flow behaviors in the classroom on the airborne transmission of the virus under the most dramatic conditions must first be determined. In this study, the effect of natural ventilation on the airborne transmission of COVID-19-like viruses in the classroom in the case of sneezing by two infected students in a reference secondary school classroom was investigated in five scenarios. Firstly, experimental measurements were carried out in the reference class to validate the computational fluid dynamics (CFD) simulation results and determine the boundary conditions. Next, the effects of local flow behaviors on the airborne transmission of the virus were evaluated for five scenarios using the Eulerian-Lagrange method, a discrete phase model, and a temporary three-dimensional CFD model. In all scenarios, immediately after sneezing, between 57 and 60.2 % of the droplets containing the virus, mostly large and medium-sized (150 μm < d < 1000 μm) settled on the infected student's desk, while small droplets continued to move in the flow field. In addition, it was determined that the effect of natural ventilation in the classroom on the travel of virus droplets in the case of Redh < 8.04 × 104 (Reynolds number, Redh=Udh/νu, dh and are fluid velocity, hydraulic diameters of the door and window sections of the class and kinematic viscosity, respectively) was negligible.

Computational fluid dynamics validated by micro particle image velocimetry to estimate the risk of hemolysis in arteries with atherosclerotic lesions

We aimed to verify the relationship between blood vessel shape and hemolysis risk by using a blood rheological model that reflects the physiological processes related to blood flow. Blood rheology depends on many factors including the red blood cell concentration and local shear stress, which affect the hemolysis process. We introduced a rheology model suitable for modeling hemolysis flows observed in arteries with atherosclerotic lesions in vivo based on the population balance. The model predicts the concentration of single red blood cells and the concentration and size of red blood cell agglomerates, which affect blood rheology and hemolysis. Atherosclerotic lesion geometries were obtained based on image reconstructions from tomographic projections. Computational fluid dynamics (CFD) simulation results were compared with particle image velocimetry measurements of the geometries printed with a 3D printer. Based on the CFD simulation data, a correlation function was established by combining the essential parameters of vessel shape and flow rate with the maximum shear stress, which governs the hemolysis. The chemical engineering methodology was successfully applied to the analysis of biological systems. In future, the model can be implemented in image reconstruction software using tomographic projections to quickly analyze hemolysis risk in medical practice.

The effect of vortex generators on the hydrodynamic performance of a submarine at a high angle of attack using a multi-objective optimization and computational fluid dynamics

Having a suitable maneuverability is of a high importance in the design of submarines. They should have a proper stability during depth changes and emergency rise which requires patterns in their design. In present study, the optimal design of nature inspired Vortex Generators (VGs) attached to the leading edge of side rudders of DARPA sub-off submarine is studied considering an emergency rise condition. To understand the effect of ideal design of VGs on hydrodynamic performance of the submarine, Computational Fluid Dynamics (CFD) analysis are coupled with Multi-Objective Genetic Algorithm (MOGA). To perform the CFD-based optimizations at angle of attack equal to 30°, 4 input parameters based on geometry and positioning of VGs are considered and the drag and lift coefficients of submarine are set as the objective functions. The results of CFD simulations and the optimization shows that VGs optimal geometry have the properties of L = 0.021315 (m), α = 90°, t = 0.00625 (m) and d = 0.027 (m) which results in the CD and CL of submarine being equal to 0.081005 and 0.058114, respectively. Application of VGs causes an small increase in CD in comparison with main model due to enlargement of stagnation point at the leading-edge. VGs also divide the fluid-flow into two portions, both of which experience an enhancement in their turbulent kinetic energy at the leading-edge. Finally, a sensitivity analysis demonstrates the range of applicability of VGs for all the depth changes.

Computational fluid dynamics in root canal irrigation.

Root canal irrigation is an important step in root canal preparation procedures and has a great impact on the success rate of root canal treatment. Computational fluid dynamics (CFD) is a new method to study root canal irrigation. It can be used to simulate and visualize the process of root canal irrigation, and quantitatively evaluate the effect of root canal irrigation through parameters such as flow velocity and wall shear stress. In recent years, researchers have conducted extensive studies on the factors that influence root canal irrigation efficiency, such as the position of the irrigation needle, the size of root canal preparation, the types of irrigation needles, and so on. This article reviewed the development of root canal irrigation research methods, the steps of CFD simulation in root canal irrigation, and the application of CFD in root canal irrigation in recent years. It aimed to provide new research ideas for the application of CFD to root canal irrigation and to provide a reference for the clinical application of CFD simulation results.

Microclimatic control of a fourteenth-century Church in Italy: The design of systems by experiments and simulations

Heritage buildings represent part of a city’s identity, and their conservation, as well as the preservation of the works of art they house, is indispensable. This study gives a contribution to the topic of indoor microclimatic analysis and control in building heritage, focusing on the environmental monitoring of an ancient church in the center of Naples (Italy). The measurement campaign, performed during two typical winter and summer periods, was a necessary preliminary step to examine the current indoor microclimate and to calibrate a dynamic energy model of the church. The numerical model was employed for coupled analyses of the church by performing BES (Building Energy Simulations) and CFD (Computational fluid dynamic) simulations.The monitoring revealed the inadequacy of current indoor microclimatic conditions for the conservation of the artistic heritage, and for the comfort of the occupants, therefore a heating and cooling system was designed paying attention to the historical/artistic value of the church. The results of BES and CFD simulations demonstrate an indoor temperature increase of up to 4.4 °C in winter and a decrease of up to 3 °C during the summer, by assuring the preservation of artifacts (e.g., too cold or too hot conditions can induce cracks of wooden materials) and improving occupants’ thermal comfort.

Computational fluid dynamics modeling of a wafer etch temperature control system

Next-generation etching processes for semiconductor manufacturing exploit the potential of a variety of operating conditions, including cryogenic conditions at which high etch rates of silicon and very low etch rates of the photoresist are achieved. Thus, tight control of wafer temperature must be maintained. However, large and fast changes in the operating conditions make the wafer temperature control very challenging to be performed using typical etch cooling systems. The selection and evaluation of control tunings, material, and operating costs must be considered for next-generation etching processes under different operating strategies. These evaluations can be performed using digital twin environments (which we define in this paper to be a model that captures the major characteristics expected of a typical industrial process). Motivated by this, this project discusses the development of a computational fluid dynamics (CFD) model of a wafer temperature control (WTC) system that we will refer to as a “digital twin” due to its ability to capture major characteristics of typical wafer temperature control processes. The steps to develop the digital twin using the fluid simulation software ANSYS Fluent are described. Mesh and time independence tests are performed with a subsequent benchmark of the proposed ANSYS model with etch cooling system responses that meet expectations of a typical industrial cooling system. In addition, to quickly test different operating strategies, we propose a reduced-order model in Python based on ANSYS simulation data that is much faster to simulate than the ANSYS model itself. The reduced-order model captures the major features of the WTC system demonstrated in the CFD simulation results. Once the operating strategy is selected, this could be implemented in the digital twin using ANSYS to view flow and temperature profiles in depth.

A hybrid method for modelling wake flow of a wind turbine

A fast and accurate prediction of the wake of an upwind turbine is very important to quantify the performance of downwind turbines in offshore wind farms, which become larger and larger. The wake flow and dynamics may be quite accurately simulated by high-fidelity computational fluid dynamics (CFD) software but its computational costs are too high, in particular to simulate a long wake flow often required in engineering practice. Therefore, the wake is often modelled by simplified dynamic wake models in design practices. They are computationally efficient but could not catch all physics, depend on pre-specified empirical parameters, and are not suitable for flow near the turbine. This paper proposes a new hybrid method, in which the near wake flow is simulated by a CFD model based on Navier–Stokes equations with the turbine represented by actuator lines while the far wake flow is modelled by an improved simplified CFD-based dynamic wake model. The two models are two-way coupled at a section downwind the turbine. The newly formulated method is validated by the results of full CFD simulations in the whole domain. Its performances are investigated under different conditions. It will be demonstrated that the new method takes considerably less computational time than the full CFD tool to produce similar results.

CFD Simulation of Melt Pool Coolability in a Simulated Core Catcher Model

A severe accident involving core melt in a nuclear reactor is a major concern especially after Fukushima. Thus, to mitigate the effects of core melt accidents, an ex-vessel core catcher is being developed for Advanced Indian Nuclear Reactors. The core catcher design envisages using special refractory material. The cooling strategy of the core catcher is one of the key components in the design of the core catcher. Performing a full-scale prototypic experiment is extremely challenging and prohibitory due to the involvement of very high temperature and presence of radioactive materials. Therefore, a computational fluid dynamics (CFD) model capable of simulating the coolability of the melt pool is important to develop. In the present work, a two-dimensional (2D) CFD model was developed to understand the heat transfer phenomenon and solidification of the heat-generating simulant melt pool. The 2D symmetry geometry of the simulated core catcher vessel was used. The CFD model considers appropriate models for melting and solidification to understand crust formation in the melt pool and the k-ε turbulence model to resolve turbulence inside the melt pool. A decay heat of 1 MW/m3 was also considered inside the melt pool. The CFD simulation results were compared with the authors’ experimental results. The experiment involved a scaled-down ex-vessel core catcher model (CCM) employing electrical heaters to simulate decay heat. The experiment was carried out by melting about 25 L of sodium borosilicate glass using a cold crucible induction furnace at about 1200°C and cooling it in the scaled-down CCM. The scaled-down CCM was strategically cooled in three phases, namely, air cooled, indirect side cooling, and complete top flooding. To overcome the complexities of simulation of the initial melt pour condition, the CFD simulation was initialized with the temperatures just after the melt pour was completed in the experiment. Similar to the experimental conditions, the CFD simulations were carried out in three phases by changing the boundary condition. Comparison of the temperatures of the melt pool by the CFD simulations and experiments at different locations gave reasonable agreement. The evolution of crust formation, melt pool temperatures, core catcher inner wall temperatures, and heat flux distribution were investigated in detail using the CFD model.

Bionic liquid cooling plate thermal management system based on flow resistance-thermal resistance model

Since the computational fluid dynamics (CFD) method is cumbersome and computationally intensive in optimizing the battery thermal management system (BTMS), a comprehensive computational method of total thermal resistance and pressure loss of liquid-cooled battery thermal management system with coupled flow resistance model (FRM) and thermal resistance model (TRM) was proposed, so that the total thermal resistance (TTR) and pressure drop (ΔP) of liquid-cooled battery thermal management system under various operating conditions can be calculated concisely and quickly, and then the optimization of the structural parameters of liquid-cooled plates can be achieved quickly. First, the reliability of the CFD simulation was verified by liquid-cooled plate (LCP) experiments, and the reliability of the flow resistance-thermal resistance model was further verified by CFD simulation results. Then, the sensitivity of different mass flow rate and liquid cooling channel structure parameters were analyzed. The results showed that the thermal performance and system energy consumption of the cold plate were mainly influenced by the coolant mass flow rate (m) and the channel structure factors such as the bifurcation channel angle (α), channel width (b), channel height (h), and distance between bifurcation channels (d1, d4) of the cold plate. Finally, the Neighborhood Cultivation Genetic Algorithm (NCGA) was used to optimize the total thermal resistance (TTR) and pressure drop (ΔP) of the liquid-cooled plate with the above six parameters as design variables and the total thermal resistance (TTR) and pressure drop (ΔP) as objective functions. Compared to the model before optimization, the total thermal resistance (TTR) and pressure drop (ΔP) of the optimized cold plate were reduced by 0.2409 K/W and 8.7371 Pa, respectively.

Predicting Ventilation Rate in a Naturally Ventilated Dairy Barn in Wind-Forced Conditions Using Machine Learning Techniques

Precise ventilation rate estimation of a naturally ventilated livestock building can benefit the control of the indoor environment. Machine learning has become a useful technique in many research fields and might be applied to ventilation rate prediction. This paper developed a machine-learning model for ventilation rate prediction from batch computational fluid dynamics (CFD) simulation results. By comparing deep neural networks (DNN), support vector regression (SVR), and random forest (RF), the best machine learning algorithm was selected. By comparing the modeling scheme of direct single-output (ventilation rate) and indirect multiple-output (predict averaged air velocities normal to the openings, then calculate the ventilation rate), the performances of the machine learning models widely applied in ventilation rate prediction were evaluated. In addition, this paper further evaluated the impact of adding indoor air velocity measurement in ventilation rate prediction. The results showed that the modeling performance of the DNN algorithm (Mean Absolute Percentage Error (MAPE) = 20.1%) was better than those of the SVR (MAPE = 23.2%) and RF algorithm (MAPE = 21.0%). The scheme of multiple-output performed better (MAPE &lt; 8%) than the single-output scheme (MAPE = 20.1%), where MAPE was the mean absolute percentage error. Additionally, the comparison of modeling schemes with different inputs showed that the predictive accuracy could be improved by adding indoor velocities to the inputs. The MAPE decreased from 7.7% in the scheme without indoor velocity to 4.4% in the scheme with one indoor velocity, and 3.1% in the scheme with two indoor velocities. The location of the additional air velocity affected the accuracy of the predictive model, with the ones at the bottom layer performing better in the prediction than those at the top layer. This study enables a real-time and accurate prediction of the ventilation rate of a barn and provides a recommendation for optimal indoor sensor placement.

Changes in nasolabial angle may alter nasal valve morphology and airflow: a computational fluid dynamics study

Aim: Nasal valve (NV) dysfunctions are a significant cause of nasal obstruction. Changes in the nasolabial angle (NLA) may also cause changes in NV morphology. The effect of changes in the 3D structure of the nasal valve region (NVR) on nasal airflow has yet to be studied sufficiently. The accuracy of computational fluid dynamics (CFD) simulation results of nasal airflow has been confirmed by in vitro tests. Therefore, this study aimed to evaluate the effect of changes in NV structure and volume on nasal airflow based on the CFD method.&#x0D; Material and Method: We used CT images to create a 3D structural model of the NVR. First, CT images were transferred to MIMICS® software, and the nasal air passage was modeled. A solid reference model of the NVR was then created using SolidWorks software. Five different solid 3D nasal valve models were created with nasolabial angles of 85˚ in Model 1, 90˚ in Model 2, 95˚ in Model 3, 100˚ in Model 4, and 105˚ in Model 5. To simulate breathing during rest and exercise using the CFD method, the unilateral nasal airflow rates were set at 150 ml/s and 500 ml/s, respectively. The CFD method was then used to calculate each model’s airflow properties. Finally, the volumes of the models, pressure at the NV outlet, and airflow velocity were evaluated and calculated to investigate each model’s NV airflow characteristics. &#x0D; Results: Our study found a significant correlation between the nasolabial angle (NLA) and NVR volume (r=-0.998, p=0.000), flow rate and velocity (r=0.984, p=0.000), velocity and maximum pressure (r=0.920, p=0.000), velocity and minimum pressure (r=-0.969, p=0.000), flow rate and maximum pressure (r=0.974, p=0.000), and flow rate and minimum pressure (r=-0.950, p=0.000). There was no correlation between NLA increase and nasal airflow velocity. We determined that the highest pressure and lowest airflow velocity values were in the upper angle region and that the lowest pressure and highest airflow velocity values were at the bottom of the NVR in all models.&#x0D; Conclusion: Using the CFD method, we found a decrease in NVR volume and an increase in airflow velocity with an increase in NLA. In addition, we found that the pressure values in the NVR did not change significantly with the increase in NLA.

Determining flow stasis zones in the intracranial aneurysms and the relation between these zones and aneurysms' aspect ratios after flow diversions.

Flow diverter stents (FDSs) are widely used to treat aneurysms in the clinic. However, even the same flow diverter (FD) use on different patients' aneurysm sites can cause unexpected hemodynamics at the aneurysm region yielding low success rates for the overall treatment. Therefore, the present study aims to unfold why FDs do not work as they are supposed to for some patients and propose empirical correlation along with a contingency table analysis to estimate the flow stasis zones in the aneurysm sacs. The present work numerically evaluated the use of FRED4518 FDS on six patients' intracranial aneurysms based on patient-specific aneurysm geometries. Computational fluid dynamics (CFD) simulation results were further processed to identify the time evolution of weightless blood particles for six patients' aneurysms. Stagnation zone formation, incoming and outgoing blood flow at the aneurysm neck, and statistical analysis of six patients indicated that FRED4518 showed a large flow stasis zone for an aspect ratio larger than 0.75. However, FRED4518, used for aneurysms with an aspect ratio of less than 0.65, caused small stagnant flow zones based on the number of blood particles that stayed in the aneurysm sac. A patient-specific empirical equation is derived considering aneurysms' morphological characteristics to determine the amount of stagnated fluid flow zones and magnitude of the mean aneurysm velocity in the aneurysm sac for FRED4518 based on weightless fluid particle results for the first time in the literature. As a result, numerical simulation results and patient data-driven equation can help perceive stagnated fluid zone amount before FRED4518 placement by shedding light on neuro-interventional surgeons and radiologists.

Dynamic characteristic modeling of left ventricular assist devices based on hysteresis effects

ObjectiveThe purpose of this study is to develop a new model for the dynamic characteristics of left ventricular assist devices (LVADs) interacting with the cardiovascular system under constant-speed modes. MethodsA new hysteresis model is established on the basis of the hysteresis effect and turbomachinery principles. The simulation results from the hysteresis model were compared with the inertia model. The in-vitro experiment results of a centrifugal pump (from literature) and the unsteady computational fluid dynamics (CFD) simulation results of an axial pump were used as the benchmarks. ResultsCompared with the inertia model, at the partial support mode, the relative estimation error of the time to the maximum and minimum pump flow (Q) in the hysteresis model decreased at least 16.3% cardiac cycle (Tc) in the centrifugal pump and at least 1.9% Tc in the axial pump, indicating its ability to simulate more realistic Q fluctuations. Moreover, the hysteresis model could predict an accurate time distribution of different Q. ConclusionThe hysteresis model provides a general calculation method for simulating the dynamic characteristics of constant-speed LVADs under interaction with the cardiovascular system. It is more accurate than the inertia model. SignificanceThe hysteresis model is helpful for the rapid estimation of unsteady dynamic characteristics in absence of a physical pump prototype at the preliminary design stage.

Characterization of Non-Neutral Urban Canopy Wind Profile Using CFD Simulations—A Data-Driven Approach

Horizontally averaged wind profiles inside the urban canopy are used in many studies and numerical models. The existing analytical models are only applicable to a small range of aspect ratios and mostly neutral atmospheric conditions due to their underlying assumptions. In this study, a surrogate model for predicting horizontally averaged wind profiles in the street canyons of an idealized urban canopy for a wide range of urban morphologies and thermal forcing scenarios is developed with the help of machine learning techniques and computational fluid dynamics (CFD) simulation data. The influence of urban morphological parameters, atmospheric stability and wind conditions on the urban canopy wind flow is modeled using machine learning algorithms applied on CFD simulation results. The numerical model is validated using wind-tunnel data. Steady-state Reynolds averaged Navier–Stokes (RANS) simulations with a standard k-ϵ turbulence model for 252 different simulation conditions are performed on an idealized building geometry that consists of a regular array of 4 × 4 cubes. The simulation results are averaged horizontally to obtain the mean velocity and temperature profiles. Surrogate models are developed using the simulation outputs as training examples and the best model is chosen by comparing the performance of different machine learning models. The surrogate artificial neural network (ANN) model of this study outperforms the current state-of-the-art models in the prediction of horizontally averaged mean wind profile inside the urban canopy. The mean error (ME) and root-mean-square error (RMSE) of the discrete point prediction ANN model of this study are 0.016 m/s and 0.060 m/s, respectively, which is significantly lower compared to the best of the legacy models for which the errors are 0.048 m/s and 0.387 m/s, respectively.