CFD Data Research Articles

Investigation on performance of particle swarm optimization (PSO) algorithm based fuzzy inference system (PSOFIS) in a combination of CFD modeling for prediction of fluid flow

Herein, a reactor of bubble column type with non-equilibrium thermal condition between air and water is mechanistically modeled and simulated by the CFD technique. Moreover, the combination of the adaptive network (AN) trainer with the fuzzy inference system (FIS) as the artificial intelligence method calling ANFIS has already shown potential in the optimization of CFD approach. Although the artificial intelligence method of particle swarm optimization (PSO) algorithm based fuzzy inference system (PSOFIS) has a good background for optimizing the other fields of research, there are not any investigations on the cooperation of this method with the CFD. The PSOFIS can reduce all the difficulties and simplify the investigation by elimination of the additional CFD simulations. In fact, after achieving the best intelligence, all the predictions can be done by the PSOFIS instead of the massive computational efforts needed for CFD modeling. The first aim of this study is to develop the PSOFIS for use in the CFD approach application. The second one is to make a comparison between the PSOFIS and ANFIS for the accurate prediction of the CFD results. In the present study, the CFD data are learned by the PSOFIS for prediction of the water velocity inside the bubble column. The values of input numbers, swarm sizes, and inertia weights are investigated for the best intelligence. Once the best intelligence is achieved, there is no need to mesh refinement in the CFD domain. The mesh density can be increased, and the newer predictions can be done in an easier way by the PSOFIS with much less computational efforts. For a strong verification, the results of the PSOFIS in the prediction of the liquid velocity are compared with those of the ANFIS. It was shown that for the same fuzzy set parameters, the PSOFIS predictions are closer to the CFD in comparison with the ANFIS. The regression number (R) of the PSOFIS (0.98) was a little more than that of the ANFIS (0.97). The PSOFIS showed a powerful potential in mesh density increment from 9477 to 774,468 and accurate predictions for the new nodes independent of the CFD modeling.

CFD and ANN approach to predict the flow pattern around the square and rectangular bluff body for high Reynolds number

The present numerical investigation consists of a detailed evaluation of flow over the bluff body to determine its behavior of boundary layer flow separation around the surfaces. In the present study, square and rectangle shape bluff body is investigated for higher Reynolds number ranging from 63,651 to 636,505 with three different cases of the angle of inclination 0°,10°, and 20°. Optimize k–∊ standard turbulent model with fine mesh grid is used to perform numerical investigation, and the results are validated from the literature in capturing flow separation from the surfaces. It is found that the flow separation regime behind the square and rectangular bluff body changes with Reynolds number and angle of inclination. It is also observed that the flow separation increases from the top side of the bluff body with an increase in the angle of inclination. The size and frequency of vortex generation increase with an increase in Reynolds number. A correlation is proposed for the vortex generation as a function of Reynolds number and angle of inclination. The prediction of vortices generation behind the square and rectangular bluff body is determined using the artificial neural network driven by CFD Data. A verification study is proposed in ANN approach based on neuron independent study to validate output field parameter of CFD.

A Multi-Stage Axial Compressor Characteristic Calibration Method based on One-Dimensional Average Streamline Method

Abstract As the core component of modern gas turbines, multistage axial compressors have a decisive influence on the performance of gas turbines. This paper proposes a multi-stage axial compressor characteristic calibration method based on one-dimensional average streamline method. Firstly, according to the physical laws followed by the multi-stage axial compressor, the aerodynamic analysis method of axial compressors is established which based on the one-dimensional average streamline method. Secondly, develop an automatic calibration method for the cascade performance model, use genetic algorithms to establish a complete blade wake Momentum thickness coefficient scalar database to realize the automatic calibration of compressor characteristics. This method has fast calculation speed and versatility. It can predict, encrypt and extrapolate compressor characteristics. It can be used to predict the characteristics of inlet guide vanes and stationary vanes when they are adjustable, or to calculate the overall performance without experimental data or CFD data.

Development and Validation of an Advanced Actuator Line Model for Wind Turbines

As wind turbine technology proceeds towards the development of more advanced and complex machines, modelling tools with fidelity higher than the ubiquitous Blade Element Momentum (BEM) method are needed. Among them, the Actuator Line Method (ALM) stands out in terms of accuracy and computational cost. Moving from this background, an advanced ALM method has been developed within the commercial solver CONVERGE®. As elements of novelty, this tool features a Lagrangian method for sampling the local inflow velocity and a piece-wise smearing function for the force projection process. Various sub-models for both Horizontal Axis Wind Turbines (HAWTs) (e.g. the Shen tip loss correction) and Vertical Axis Wind Turbines (VAWTs) (e.g. the MIT dynamic stall model) has also been included. Aim of the research is to address the new challenges posed by modern machines. HAWTs are in fact getting larger and larger, shifting the research focus on the interaction of increasingly deformable blades with the atmosphere at the micro- and mesoscale level. VAWTs on the other hand, whose popularity has arisen in the last years, thanks to their advantages in non-conventional applications, e.g. floating offshore installations, are extremely complex machines to study, due to their inherently unsteady aerodynamics. The approach has been validated on selected test cases, i.e. the DTU 10MW turbine and a real 2-blade H-rotor, for which both high-fidelity CFD and experimental data are available.

An Effective Hybrid Atrous Convolutional Network for Pixel-Level Crack Detection

Automated pixel-level crack detection is one of the essential tasks in the field of defect inspection. Deep convolutional neural networks, typically using encoder–decoder architectures, have been successfully applied to many crack detection scenes in recent works. However, encoder–decoder networks commonly rely on downsampling and upsampling operations and have a large number of parameters, which may influence the accuracy of crack prediction due to the cracks usually have long, narrow sizes, and the labeled training set is always limited. To address these issues, we propose a simple and effective hybrid atrous convolutional network (HACNet). HACNet maintains the same spatial resolution throughout the whole architecture. It can retain more spatial precision in prediction. HACNet uses atrous convolutions with the proper dilation rates to enlarge the receptive field and a hybrid approach connecting these convolutions to aggregate multiscale features. The resulting architecture can achieve accurate segmentation with relatively few parameters. Evaluations on the public CFD data set, CrackTree206 data set, Deepcrack data set (DCD), and Yang et al. Crack data set (YCD) demonstrate that our method can obtain promising results, compared with other recent approaches. Evaluation on self-collected images and SDNET2018 data set illustrates the good potential of HACNet for practical applications.

CFD study and experimental validation of low liquid-loading flow assurance in oil and gas transport: studying the effect of fluid properties and operating conditions on flow variables

Low liquid-loading flow frequently occurs during the transport of gas products in various industries, such as in the Oil & Gas, the Food, and the Pharmaceutical Industries. Even small amounts of liquid can have a significant effect on the flow conditions inside the pipeline, such as increased pressure loss, pipe wall stresses and corrosion, and liquid holdup along the pipeline. However, most studies that analyze this type of flow only use atmospheric pressures and horizontal 1-in or 2-in pipes, which do not accurately represent the range of operating conditions present in industrial applications. Therefore, this study focused on modeling low liquid-loading flow in medium-sized (6–10 in) pipes, using CFD simulations and experimental data from the University of Tulsa, and then applying it to real operating conditions from a Colombian gas pipeline. An acceptable difference was observed between experimental and CFD data, both for the liquid holdup (18%) and for the pressure drop (12%). Variables like pressure drop and wall shear stress increase with phase velocity, operating pressure, and pipe inclination. Liquid holdup increases with liquid velocity but decreases with all other factors. The relation of flow variables with phase velocities is of particular interest: Doubling the gas velocity decreased holdup 70% and increased pressure drop tenfold. On the other hand, the presence of the liquid phase seems to be more influential on process variables than its exact flowrate; the introduction of the liquid phase to a single-phase gas causes an increase in pressure loss by a factor of three, but doubling the liquid velocity only increases the pressure loss by a further 30%.

Investigation on heat transfer enhancement in a circular pipe with artificial roughness

This paper presents a numerical analysis of convective heat transfer enhancement of transverse ribs in circular tubes. Several CFD simulations are carried out for turbulent airflow to analyze heat transfer and pressure drop provided with transverse rectangular ribs. The rib height and pitch are widely varied along with the flow Reynolds number. The effect of each parameter is examined and discussed. To accurately predict major parameters (Nusselt number, friction factor, and thermal hydraulic performance parameter) a deep neural network is developed, trained, and tested by current CFD data. The result demonstrates that artificial neural network shows better performance compared to other methods of prediction (e.g. power-law approximation) and can offer an economical and powerful approach for modeling optimal heat enhancement parameters.

Application of machine learning methods for investigating the heat transfer enhancement performance in a circular tube with artificial roughness

This paper presents a hybrid approach for investigation of heat transfer enhancement performance using computational fluid dynamics and artificial neural network. More than 5,000 CFD simulations are carried out for turbulent flow in pipes provided with artificial roughness of transverse rectangular ribs to analyze heat transfer, pressure drop, and thermal hydraulic performance. The rib height and pitch are widely varied along with the flow Reynolds number, working fluid, and material of roughness elements. To accurately predict major parameters (Nusselt number, friction factor, and thermal hydraulic performance) a deep neural network is developed, trained, and tested by current CFD data. The ANN allowed finding optimal rib roughness parameters for the current problem and opened perspectives of industrial application due to low computational cost and prediction error of less than 1.5%.

On the reliability of the 3D steady RANS approach in predicting microscale wind conditions in seaport areas: The case of the IJmuiden sea lock

Coastal areas and seaport areas are exposed to high wind speeds which may involve risks for the ships and people working in the area. Therefore, knowledge of the microscale wind conditions is essential for safe maneuvering and mooring of ships and optimizing harbor design. In the present study, 3D steady RANS CFD simulations with the realizable k-ε turbulence model are performed for the new configuration of the “IJmuiden sea lock” in Amsterdam, the largest sea lock in the world at the time of writing this article. The computed wind speed and turbulence intensity amplification factors and the local wind directions are validated with on-site measurements for the old configuration of the sea lock. For the wind speed amplification factor and local wind direction, a satisfactory agreement is obtained with 90% of CFD data within ±30% from the measured data. Conversely, for the turbulence intensity amplification factor, less satisfactory agreement is found with 74% of CFD data within ±30% from the measured data. Overall, the 3D steady RANS approach shows a sufficiently high reliability for predicting the wind conditions in the seaport area under neutral atmospheric conditions.

Data Mining and Machine Learning Techniques for Aerodynamic Databases: Introduction, Methodology and Potential Benefits

Machine learning and data mining techniques are nowadays being used in many business sectors to exploit the data in order to detect trends, discover certain features and patters, or even predict the future. However, in the field of aerodynamics, the application of these techniques is still in the initial stages. This paper focuses on exploring the benefits that machine learning and data mining techniques can offer to aerodynamicists in order to extract knowledge from the CFD data and to make quick predictions of aerodynamic coefficients. For this purpose, three aerodynamic databases (NACA0012 airfoil, RAE2822 airfoil and 3D DPW wing) have been used and results show that machine-learning and data-mining techniques have a huge potential also in this field.

A 6-DoF maneuvering model for the rapid estimation of hydrodynamic actions in deep and shallow waters

We present a modular mathematical model and a reference technique for the rapid estimation of maneuvering trajectories and motion time histories of single- and twin-screw propulsion ships. Heave, roll and pitch radiation damping are estimated from a non-linear unified seakeeping/maneuvering time-domain tool using numerical decay tests and then implemented to a 6–DoF model in the form of critical damping and natural period. Short waves are idealised by numerical integration along the vessel's waterline profile and associated hydrodynamic actions are implemented in a response curve format. For the rapid assessment of hull in-plane hydrodynamic forces, derivatives are implemented via semi-empirical methods, CFD or model test data. Results are validated against experiments available for zig-zag and turning cycle trajectories of vessels with different hull forms and propulsion configurations. It is concluded that the approach presented is feasible for the prediction of maneuvering trajectories of existing or new-build vessels and for estimating the evasive velocity in way of contact before grounding.

Soot Emission Simulations of a Single Sector Model Combustor Using Incompletely Stirred Reactor Network Modeling

AbstractThe simulation of soot evolution is a problem of relevance for the development of low-emission aero-engine combustors. Apart from detailed CFD approaches, it is important to also develop models with modest computational cost so that a large number of geometries can be explored, especially in view of the need to predict engine-out soot particle size distributions (PSDs) to meet future regulations. This paper presents an approach based on Incompletely Stirred Reactor Network (ISRN) modeling that simplifies calculations, allowing for the use of very complex chemistry and soot models. The method relies on a network of incompletely stirred reactors (ISRs), which are inhomogeneous in terms of mixture fraction but characterized by homogeneous conditional averages, with the conditioning performed on the mixture fraction. The ISRN approach is demonstrated here for a single sector lean-burn model combustor operating on Jet-A1 fuel in pilot-only mode, for which detailed CFD and experimental data are available. Results show that reasonable accuracy is obtained at a significantly reduced computational cost. Real fuel chemistry and a detailed physicochemical sectional soot model are consequently employed to investigate the sensitivity of ISRN predictions to the chosen chemical mechanism and provide an estimate of the soot PSD at the combustor exit.

Application of generative deep learning to predict temperature, flow and species distributions using simulation data of a methane combustor

A data-driven surrogate modelling methodology of CFD simulation data is proposed.It uses convolutional variational autoencoders (VAE) and multi-layer perceptron (MLP) neural networks in an integrated manner to predict temperature, velocity and species mass fraction profiles on a cell-by-cell basis. The VAE performs feature extraction and data compression, while significantly reducing the number of network interconnections and ensuring physically realistic predictions.The MLP maps the CFD boundary condition values to the encodings generated with the VAE encoder network. The approach is demonstrated via application to a 2D axisymmetric methane-fired turbulent jet diffusion flame. The integrated network model produced average normalized mean absolute errors (NMAE) of 2.41% for the temperature predictions, 0.99% for velocity and 0.925% for species mass fractions. It is, therefore, possible to predict 2D CFD data fields with reasonable accuracy and generalizability, although high NMAE values were observed for certain cells confined to a small region near the centreline of the burner. The methodology lays the foundation for application to larger industrial problems to visualize processes inside equipment, deploy virtual sensors, perform quick what-if analysis, explore the design space, link to optimization routines to effectively control equipment, detect anomalies, or to form part of lower-dimensional system simulations.

Modeling the static and dynamic characteristics of a wind turbine airfoil and validation with experimental data

Dynamic stall modeling remains one of the most difficult topic in wind turbine aerodynamics despite its long research effort in the past. In the present studies, comprehensive investigations involving CFD, engineering models and experimental data were performed within the joint collaborative work DSWind. Both static and dynamic characteristics of a pitching wind turbine airfoil under a turbulent boundary layer flow were considered. The results indicate that optimized time-accurate CFD calculations can serve as reference data for verifying the engineering models, which can be potentially used for tuning the models under different air flow conditions.

Comparison of the free vortex wake and actuator line methods to study the loads of a wind turbine in imposed surge motion

To enable the development of floating offshore wind farms, it is important to have a clear understanding of the aerodyamic forces applied on a floating offshore wind turbine. The paper presents comparisons between a lifting line free vortex wake method and an actuator line method in the case of a wind turbine in surge with blade-resolved CFD data as a reference. Each model is compared to a quasi-steady estimation of the loads to understand the variations due to the surge movement. The near-wake flow field is investigated in order to give an insight into the flow features leading to the observed behavior. Both methods predict a higher axial velocity at the position of the rotor and one radius downstream of the rotor in surge conditions compared to the fixed case.

Formation of Vortices in Idealised Branching Vessels: A CFD Benchmark Study.

Atherosclerosis preferentially occurs near the junction of branching vessels, where blood recirculation tends to occur (Malek et al. in J Am Med Assoc 282(21):2035-2042, 1999, https://doi.org/10.1001/jama.282.21.2035 ). For decades, CFD has been used to predict flow patterns such as separation and recirculation zones in hemodynamic models, but those predictions have rarely been validated with experimental data. In the context of verification and validation (V&V), we first conduct a CFD benchmark calculation that reproduces the vortex detection experiments of Karino and Goldsmith (1980) with idealised branching blood vessels (Karino and Goldsmith in Trans. Am. Soc. Artif. Internal Organs 26:500-506, 1980). The critical conditions for the formation of recirculation vortices, the so-called critical Reynolds numbers, are the main parameters for comparison with the experimental data to demonstrate the credibility of the CFD workflow. We then characterise the wall shear stresses and develop a surrogate model for the size of formed vortices. An automated parametric study generating more than 12,000 CFD simulations was performed, sweeping the geometries and flow conditions found in the experiments by Karino and Goldsmith. The flow conditions were restricted to steady-state laminar flow, with a range of inflow Reynolds numbers up to 350, with various flow ratios between the main branch outlet and side branch outlet. The side branch diameter was scaled relative to the main branch diameter, ranging from 1.05/3 to 3/3; and the branching angles ranged in size from [Formula: see text] to [Formula: see text]. Recirculation vortices were detected by the inversion of the velocity vector at certain locations, as well as by the inversion of the wall shear stress (WSS) vector. The CFD simulations demonstrated good agreement with the experimental data on the critical Reynolds numbers. The spatial distributions of WSS on each branch were analysed to identify potential regions of disease. Once a vortex is formed, the size of the vortex increases by the square root of the Reynolds number. The CFD data was fitted to a surrogate model that accurately predicts the vortex size without the need to run computationally more expensive CFD simulations. This benchmark study validates the CFD simulation of vortex detection in idealised branching vessels under comprehensive flow conditions. This work also proposes a surrogate model for the size of the vortex, which could reduce the computational requirements in the studies related to branching vessels and complex vascular systems.

A Comparison Study between 3-D CFD and Experimental Data of Butterfly Valve Coefficients. (Dept. M. ( Power ) )

window.dataLayer = window.dataLayer || []; function gtag(){dataLayer.push(arguments);} gtag("js", new Date()); gtag("config", "UA-72322659-1");

Analysis of the explosion behaviour of wheat starch/pyrolysis gases hybrid mixtures through experimentation and CFD-DPM simulations

As a result of their intrinsic compl5exity and their severe consequences, the study of hybrid mixture explosions requires the consideration of several additional factors and parameters than the study of gas-only and dust-only explosions. The main objective of this work is to perform an exhaustive analysis of the explosion severity of wheat starch/pyrolysis gases hybrid mixtures in the context of the 20-L Sphere Test and to determine how severe hybrid mixture explosions can be when compared to one phase explosions. Several experimental tests were performed at different concentrations of both phases and at fixed operating conditions of ignition delay time (tv) and nozzle geometry. CFD simulations from a previously-validated model were run alongside in order to complement the experimental findings and expand the understanding of the hybrid system as a whole, especially in terms of the development and interaction of the phases’ chemical reactions. The experimental and CFD data were analysed using severity maps and adapted stoichiometric parameters proposed in the literature. It was found that the explosion regimes proposed in the literature are not universally-applicable to all types of hybrid mixtures, given that, for wheat starch/pyrolysis gases hybrid mixtures, each regime could not be uniquely characterized and differences in the explosion behaviour were observed within the same regime. In contrast, it was found that analysing the data in terms of parameters that take into account the nature of the chemical reactions of the phases is a superior methodology given that it provides a more rigorous comparative basis for assessing the explosion severity of different types of combustible materials, including hybrid mixtures.

Parallel computing in automation of decoupled fluid-thermostructural simulation approach

1. GLAENZEL J., IHLENFELDT S., NAUMANN C., PUTZ M., 2018, Efficient quantification of free and forced convection via the decoupling of thermo-mechanical and thermo-fluidic simulations of machine tools, Journal of Machine Engineering, 18/2, 41–53. Google Scholar

Investigation on thermal efficiency of LPG cooking burner using computational fluid dynamics

This study aims to propose a novel method in predicting the thermal efficiency of an LPG-energy saving burner (EB) using CFD data. The CFD data of the flow velocity and the combustion temperature distributions agree very well with the experimental data. The correlated data obtained from the experimental measurements and the numerical heat flux are then used to estimate the thermal efficiency of the EB. The effects of LPG released-pressure on its flow phenomena and thermal efficiency are investigated. Finally, the correlation equation to determine thermal efficiency is presented. It is found that the difference between the thermal efficiency obtained from this equation and one from the experimental data over the various range of LPG released-pressure is around ±2.41%. Besides, two improved versions of the EB burner, namely EB-W and EB-WT burners, are presented and evaluated numerically. The predicted thermal efficiencies obtained from both burners are 9.02% and 7.87%, respectively, higher than the original EB burner. These predicted thermal efficiencies agree well with the experimental results. Several validations confirm that the proposed method can be simply applied to evaluate and enhance the thermal efficiency of the burners.