Accurate Flow Research Articles

End-to-end wind turbine wake modelling with deep graph representation learning

Wind turbine wake modelling is of crucial importance to accurate resource assessment, to layout optimisation, and to the operational control of wind farms. This work proposes a surrogate model for the representation of wind turbine wakes based on a state-of-the-art graph representation learning method termed a graph neural network. The proposed end-to-end deep learning model operates directly on unstructured meshes and has been validated against high-fidelity data, demonstrating its ability to rapidly make accurate 3D flow field predictions for various inlet conditions and turbine yaw angles. The specific graph neural network model employed here is shown to generalise well to unseen data and is less sensitive to over-smoothing compared to common graph neural networks. A case study based upon a real world wind farm further demonstrates the capability of the proposed approach to predict farm scale power generation. Moreover, the proposed graph neural network framework is flexible and highly generic and as formulated here can be applied to any steady state computational fluid dynamics simulations on unstructured meshes.

On citizens' right to information: Justification and analysis of the democratic right to be well informed

On citizens' right to information: Justification and analysis of the democratic right to be well informed

Evaluation of Surface Skimmer Flow Rates and Size Selection

A floating surface skimmer is a device used to dewater a sediment basin as it fills. The skimmer floats on the surface, draining the least turbid water as sediment falls out of suspension. An adjustable orifice on the skimmer regulates the filling and draining rate of the basin. After significant runoff events, skimmers slowly drain the basin over several days to maximize settling, while draining less turbid water from the top of the water column. Manufacturers have published data for products that customers can use to decide on the skimmer type, size, and orifice opening, but these design parameters tend to be very rough estimates with numerous assumptions. This study details a methodology for testing skimmers, including the materials, data collection process, and data analysis approach required to obtain the most accurate skimmer flow rate data. Testing was performed on a 15.2 cm (6.0 in.) post-construction stormwater skimmer prototype provided by J.W. Faircloth & Son, Inc., in an approximately 30 m3 (1000 ft3) evaluation tank. This skimmer used an adjustable sluice gate to control flow rate and had two barrel lengths. Six sluice gate opening sizes were tested three times for both barrel lengths, resulting in 36 tests performed. Experiments revealed that the skimmer had a capacity ranging from 0.03 m3/s (0.5 ft3/s) with a 2.54 cm (1.0 in.) opening and a capacity of 0.071 to 0.085 m3/s (2.5–3.0 ft3/s) with an opening of 15.2 cm (6.0 in.). Experimental results were then used to create a user-interactive skimmer sizing tool to provide more accurate information on sediment basin storage and drawdown times based on skimmer selection.

Using pressure-driven flow systems to evaluate laser speckle contrast imaging.

Microfluidic flow phantom studies are commonly used for characterizing the performance of laser speckle contrast imaging (LSCI) instruments. The selection of the flow control system is critical for the reliable generation of flow during testing. The majority of recent LSCI studies using microfluidics used syringe pumps for flow control. We quantified the uncertainty in flow generation for a syringe pump and a pressure-regulated flow system. We then assessed the performance of both LSCI and multi-exposure speckle imaging (MESI) using the pressure-regulated flow system across a range of flow speeds. The syringe pump and pressure-regulated flow systems were evaluated during stepped flow profile experiments in a microfluidic device using an inline flow sensor. The uncertainty associated with each flow system was calculated and used to determine the reliability for instrument testing. The pressure-regulated flow system was then used to characterize the relative performance of LSCI and MESI during stepped flow profile experiments while using the inline flow sensor as reference. The pressure-regulated flow system produced much more stable and reproducible flow outputs compared to the syringe pump. The expanded uncertainty for the syringe pump was 8 to higher than that of the pressure-regulated flow system across the tested flow speeds. Using the pressure-regulated flow system, MESI outperformed single-exposure LSCI at all flow speeds and closely mirrored the flow sensor measurements, with average errors of and , respectively. Pressure-regulated flow systems should be used instead of syringe pumps when assessing the performance of flow measurement techniques with microfluidic studies. MESI offers more accurate relative flow measurements than traditional LSCI across a wide range of flow speeds.

Research on the flow characteristics identification of steam turbine valve based on FCM-LSSVM

Due to aging and deformation of the through-flow path and system modifications, the flow characteristics of the turbine inlet valve often deviate from the design value, which affects the unit load control accuracy and operational stability. In order to obtain the actual valve flow characteristics of the turbine and thus improve the FM performance, an FCMLSSVM model is proposed in this paper to identify the valve flow characteristics. First, FCM clustering is proposed to classify the historical operating data of the plant and obtain a wide range of variable operating conditions. Then, using least squares support vector machine (LSSVM), the relationship between turbine input and output variables was modeled in each condition cluster, with integrated valve position command, speed, and real power generated as input variables and actual steam inlet flow as output variables. Using a 330 MW turbine unit as an application example, the established FCM-LSSVM model was validated for the valve flow characteristics of the turbine. The results show that the model can obtain accurate valve flow characteristics without conducting tests on the turbine. The method can save a lot of labor and material resources in doing the characteristic test, and after comparison, the proposed method can identify the flow characteristics more accurately among the existing neural network identification methods, which can provide technical support to improve the unit frequency regulation characteristics and improve the accuracy of valve operation.

An automatic and integrated self-diagnosing system for the silting disease of drainage pipelines based on SSAE-TSNE and MS-LSTM

The regular detection and diagnosis mechanism for the silting disease of drainage pipelines (SDP) is critical for making dredging decisions and flood forecasting. Simultaneously, there is a complex coupling relationship between the pipeline siltation and the dynamic changes of the inlet and outlet flows. However, the traditional silting detection frameworks of drainage pipelines face difficulties in two aspects, accurate flow rate measurements and efficient and intelligent siltation diagnoses. In this paper, an automatic and integrated self-diagnosing system for the silting disease of drainage pipelines is proposed. First, a gradient penalty generative adversarial network (GPGAN), a target recognition algorithm called guided faster R-CNN (GF R-CNN) and an image segmentation method called parallel multiscale U-Net (PMSU-Net) are introduced to dynamically measure the flow of water in the drainage pipelines. Second, an automatic siltation detection and diagnosis algorithm of SDP is constructed based on a t-SNE stack sparse autoencoder (SSAE-TSNE) and multiscale long short-term memory (MS-LSTM). Then, a full-scale prototype verifies the feasibility of the system. This improved flow measurement algorithm achieves the highest accuracy of 90.32%, and an F1 score of 0.963 in comparison with the other algorithms, and the precision-recall curve was the closest to the top right corner. The error rate of the proposed self-diagnosing system is controlled within 8%, which is far better than the other algorithms. Finally, the system algorithms are embedded in an intelligent platform consisting of an unmanned pipeline measuring device (UPM), an unmanned aerial vehicle (UAV) and a server diagnostic centre. The practical application and test show that the accuracy, precision and response speed of the proposed integrated self-diagnosing system have obvious advantages in the function of siltation detection and diagnosis in drainage pipelines.

Short-term traffic flow prediction model based on a shared weight gate recurrent unit neural network

Accurate traffic flow prediction is critical for enhancing traffic network operational efficiency. With the continuous expansion of traffic networks, providing reliable and efficient multi-step traffic flow prediction for large-scale traffic networks with a large number of sensors deployed has become a challenging issue. In this paper, we propose a multi-step many-to-many traffic prediction model for large-scale traffic networks, called spatio-temporal Shared GRU (STSGRU), which receive inputs from multiple sensors and provides predictions for all sensors simultaneously. First, we model the weekly pattern of traffic flow, using periodicity to explore long-term temporal features and provide smooth traffic flow to reduce the impact of data volatility. Second, different from existing models, we propose a shared weight mechanism to achieve many-to-many prediction without mapping traffic networks to images or graph structures. The proposed model strikes a delicate balance between complexity and accuracy. We validate the effectiveness of the proposed method on the Caltrans Performance Measurement System (PeMS) dataset. The results show that our model achieves similar prediction performance with advanced graph neural networks and has higher flexibility.

Study of air compressibility effects on the aerodynamic performance of the IEA-15 MW offshore wind turbine

With the gradual upscaling of offshore wind turbines, the blade size becomes large and the blade tip velocity reaches almost a Mach number of 0.3, the limit of the incompressible flow assumption.To simulate wind turbine flows, an incompressible solver was often a good choice as it gave good predictions and the Mach number of the wind turbine flows were small. When the Mach number is increased to be close to 0.3, the validity of incompressible flow may be questionable and thus needs to be checked. In this paper, an accurate wind turbine flow solution method is proposed to study the compressibility effects. First, the National Renewable Energy Laboratory (NREL) Phase VI wind turbine is used as an example to verify the method at a small Mach number. Through a comparative analysis between the compressible, the incompressible numerical results, and experimental data, the proposed method is found with a good simulation accuracy. Next, the flow fields of the large wind turbine IEA-15 MW, under different wind conditions are simulated. The local Mach number, density and pressure coefficient at different blade cross-sections of the compressible results are analyzed and compared to its incompressible ones, and the influences of compressibility on the thrust, torque and power of the wind turbine are quantified. The results show that at the rated wind speed, the thrust of the compressible flow simulation is 1.40 % higher than that of the incompressible one, and the torque of the compressible flow simulation is nearly 11.07 % higher than that of the incompressible one. As a result, the compressibility effects are recommended to be added in future studies of large wind turbines.

A novel data-driven vanadium redox flow battery modelling approach using the convolutional neural network

This paper proposes a highly accurate data-driven vanadium redox flow battery (VRB) modelling approach for power engineering studies. The proposed approach overcomes the common problem of high model dependency that is encountered by the existing electrochemical principle or equivalent circuit based VRB modelling methods. It directly learns the behavioural relationship between VRB current, flow rate, state-of-charge and voltage through experimentally trained convolutional neural networks (CNN) and thus, avoids the usage of complicated equations for power engineering studies. This contributes to greatly simplify the studies of electrical systems that integrate VRB with improved accuracy. The validity of the proposed approach is verified by experimental results. Noticeably, the performances of both two-dimensional CNN (2D-CNN) and one-dimensional CNN (1D-CNN) on VRB modelling are compared and analysed.

Modeling the Combined Effect of Travelers’ Contrarian Behavior, Learning and Inertia on the Day-to-Day Dynamics of Route Choice

Understanding the many facets of repeated route choice behavior in traffic networks is essential for obtaining accurate flow forecasts and enhancing the effectiveness of traffic management measures. This paper presents a model of the day-to-day evolution of route choices incorporating travelers’ contrarian behavior, learning and inertia. The model is formulated as a discrete-time nonlinear dynamical system, and its properties are investigated analytically and numerically with a focus on the effect of the fraction of individuals adopting a contrarian route choice behavior. The findings of the study indicate that the extent of contrarian behavior may have significant impacts on the attractiveness and stability of network equilibria as well as on global system performance. We show that a properly balanced combination of direct and contrarian subjects can protect the system from instabilities triggered by other behavioral and network features. Our results also suggest that the fixed point stability range may depend to a considerable extent on travelers’ inertia and memory of previous experiences, as well as on the form of the travel cost functions used in the model. The occurrence of contrarian behavior should be explicitly taken into account in the design of traffic management schemes involving the deployment of Advanced Traveler Information Systems (ATISs), as it may act as a mitigating factor against the concentration of choices on the recommended routes. The analytical framework proposed in this paper represents a novel contribution, since contrarian behavior in repeated route choice has been investigated mainly by means of empirical or simulation approaches thus far.

Case Study: Remote Monitoring of a Solids-Management System in the UKCS

_ A well in the Southern North Sea was shut in due to clay and proppant production. Before it was shut in, it had been producing 1,199 B/D. The challenge presented by the operator was to clean out the well and reestablish production. The operator wanted to boost sustainability metrics and reduce emissions as much as possible. Adding further complexity to the project, the platform in use was a normally unmanned installation (NUI). This type of platform meant no crew could stay overnight and resulted in several new challenges being considered upfront. 1. Remote monitoring abilities were crucial. 2. A reliable, accurate information flow was key for planning workflow during the day. 3. The work window was reduced to 12 hours per day, so maintenance and the emptying of solids had to be carried out during the day. 4. Limited crane capacity was available for the rig-up and assembly. The above considerations, coupled with the ability to remove personnel on board (POB), the efficiency of solids handling, and compact footprint, meant the DualFlow 5K PSI desander was selected as the equipment of choice by the operator. Advance Planning A 7-week wellbore cleanup was planned to reinstate production while also removing the clay/proppant solids. During this time, the FourPhase team also monitored the well performance. Understanding the space constraints offshore was key. To kick off the project, a feasibility report was developed to capture basic well data, which included flow rates, oil/gas properties, pressure, temperature, expected solids production, and expected erosional rates. A site visit was arranged to identify laydown areas, pipework routing, connections, utilities, and rig-up possibilities. Simulations were also carried out to assess the flow and erosional performance and to ensure the planned work was within the operational window of the desander. By creating a 3D rig-up simulation in advance, it was determined that an efficient rig-up could be achieved. This stage also created logistical efficiencies, as accurate measurements for equipment and piping ensured that minimum excess material was shipped offshore. This approach further boosted the ESG metrics of the projects. Implementation The desander contains two patented, single-liner cyclones in one unit. As one vessel is isolated and flushed, production continues through the other, delivering a proven separation efficiency of 99.8% and a dynamic range of 20-μm to 10-mm solids up to 20,000 B/D and 20 MMscf/D. The result is continuous production and better asset management because the data allow early stages of erosion to be managed. Due to its design, it was uniquely able to tackle the specific challenges posed by this project. The system was monitored by the operator’s nearby control platform, where the well could be shut in if there was an emergency. The Operations and Technology Centre in Bergen, along with the FourPhase Aberdeen facilities, remotely monitored the system’s key operating parameters. From a POB perspective, the solution removed traditionally labor-intensive manual operations such as valve manipulation and flushing operations. These operations were run automatically or with a click of a button remotely. Removal of such operations directly reduced POB and associated emissions.

Multiphase Flowrate Measurement With Multimodal Sensors and Temporal Convolutional Network

Accurate multiphase flow measurement is vital in monitoring and optimizing various production processes. Deep learning has as of late arose as a promising approach for assessing multiphase flowrate dependent on various customary flow meters. In this paper, we propose a multi-modal sensor and Temporal Convolution Network (TCN) based method to predict the volumetric flowrate of oil/gas two-phase flows. The volumetric flowrates of the liquid and gas phase vary from 0.96 - 6.13 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\text{m}^{{3}}$ </tex-math></inline-formula> /h and 5.5 - 121.2 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\text{m}^{{3}}$ </tex-math></inline-formula> /h, respectively. The multi-modal sequential sensing data are simultaneously collected from a Venturi tube and a dual-plane Electrical Capacitance Tomography (ECT) sensor in a pilot-scale multiphase phase flow facility. The reference data are derived from the single-phase flowmeters. Z-score and First-Difference (FD) data pre-processing methods are employed to manipulate the collected instantaneous time series multi-modal sensing data. The pre-processed data are utilized for training the TCN model. Experimental results reveal that the TCN model can effectively predict the multiphase flowrate based on the multi-modal sensing data. The results provide guidance on data pre-processing methods for multiphase flowrate estimation and demonstrate the effectiveness of combining multi-modal sensors and TCN for multiphase flowrate prediction under complex flow conditions.

Enhancing Network Performance Tomography in Software-Defined Cloud Network

For cloud network performance profiling, network tomography based on end-to-end measurement is often used in deducing the network performance for its efficiency. However, most tomography problems are under-constrained, which require additional assumptions or probing monitors planted among network switches, which are often unavailable in software-defined networking (SDN) environment. On the other hand, SDN-based flow mirroring could provide accurate flow information, but the cost of both gathering and analysing the packet traces is tremendous that it is impossible to cover the whole network. We propose ScoutFlow, a method combining SDN flow measurement and end-to-end performance tomography, to achieve accurate performance profiling for cloud network while keeping low monitoring overhead. We evaluate ScoutFlow in our campus data center cloud, and the experiment shows good scalability and accuracy.

CFD Analysis of Pressure Variation inside Iron-Ore Slurry Pipelines

Abstract: Slurries, basically regarded as two – phase (solid – liquid) mixtures of ore and the carrier medium, generally water, are transported through underground pipelines. The rheological properties of the slurry also play an important role in determining the salient parameters of hydraulic transport such as head loss and pressure drop in commercial slurry pipelines. CFD has provided an upper hand in mathematical modelling and flow simulation, as it provides accurate and dependable flow simulation results. This article primarily focuses on determining the pressure drop across the ends of the pipeline, as slurry flows through it, and to establish a graphical depiction of the various simulations made using different set of conditions. The pressure drop for R500 radius of curvature of bends was found to be lying between 4.27% to 6.53%, as the bend angles increased from 3° to 18°, while the same was found to be lying between 3.49% to 6.2% for R2500 radius of curvature of bends. The pressure drop for R500 radius of curvature of bends was found to be lying between 4.91% to 8.3%, as the inlet velocity of slurry increased from 0.5 m/s to 3m/s, while the same was found to be lying between 5.05% to 8.63% for R2500 radius of curvature of bends.

Particle mass flow determination in dust-laden supersonic flows by means of simultaneous application of optical measurement techniques

Particle mass flow rate and particle mass concentration are key parameters for describing two-phase flows, especially for particle-induced heating augmentation analysis. This work addresses the question of how accurate particle mass flow rate can be determined with three non-intrusive measurement approaches, based on shadowgraphy, particle tracking velocimetry (PTV), and scattered light intensity, in supersonic flows. In terms of shadowgraphy and PTV, the particle mass flow rate was determined by measuring individual particle characteristics, namely particle size, velocity, and density, as well as the measurement volume. The presented shadowgraphy procedure is based on the commercial LaVision DaVis software and additional shadowgraphy corrections. Multiple tests were conducted in the experimental test facility GBK of DLR with varying flow conditions, at a Mach number of 2.1, unit Reynolds number (Re∞) ranging from 5e7 1/m to 1.5e8 1/m, total temperature (T0) ranging from 303 to 544 K, and particle materials, namely Al2O3, MgO, and SiO2, in the size range of 1 to 60 µm. Particle size distributions of Al2O3 and MgO particles could be reproduced with shadowgraphy quite well, while the PTV procedure resulted in non-similar distributions. Pycnometer measurements indicated MgO particle density to be significantly lower than reference values. A DaVis parameter variation analysis resulted in a particle mass flow rate uncertainty of shadowgraphy of up to 30%. The particle mass flow rate uncertainty of PTV is approx. 76%, and the respective uncertainty of scaled PTV and scattered light intensity approach is 28%. The particle mass flow rate, measured with shadowgraphy, is 58% higher than those of the semi-axisymmetric scattered light intensity approach, which can be explained by a higher particle concentration at the injection plane.

A separate modelling approach for short-term bus passenger flow prediction based on behavioural patterns: A hybrid decision tree method

Accurate short-term passenger flow prediction plays an important role in transit planning and operation. Existing research is mostly based on a joint modelling approach in which transit demand is predicted in an aggregated manner taking the overall passenger flow as input. A critical problem for the joint modelling approach is that the complexity of passenger flow composition and the distinct behavioural response to influential factors are missing out. To address this challenge, this paper proposes a separate modelling approach for passenger flow prediction based on behavioural patterns. To this end, we develop a novel hybrid decision tree (HDT) model coupled with a decision tree model and time series model. The upper layer is a decision tree model, in which the dataset is divided according to passenger types and influential factors, while the lower layer is the time series model achieved by the recurrent neural network. Particularly, this research first undertakes passenger classification using smartcard data through cluster analysis, from which the correlation between the classified passenger flow and influential factors is obtained. The proposed method is tested in a real-life bus route in Guangzhou, China. We also investigate the impact of passenger classification schemes and the minimum amount of data contained by leaf nodes on the performance of the HDT model. Based on this, we recommend the best classification scheme and the optimal value of the minimum amount of data contained by leaf nodes. Comparisons show that our method outperforms other traditional methods in terms of both prediction accuracy and stability. In addition, our method could also provide the prediction of passenger flow composition, which provides more references for customized bus service design.

CFD as a Decision Tool for Pumped Storage Hydropower Plant Flow Measurement Method

Suitable and accurate flow measurement in pumped storage hydropower plants (PSP) is a challenging task due to the entirely different hydraulic behaviour of the penstock. This study presents a novel approach to choosing a suitable flow measurement method and position. The focus is on the flow measurement in a specific short penstock of the largest peak-load hydropower plant, Orlík, after its transformation to a PSP. Our approach is based on three main pillars: numerical modelling of fluid flow (ANSYS CFX), standards, and scientific literature. First, the steady-state numerical model output for the current state is compared to historical measurements of point velocities using current meters and measured hydraulic losses in the penstock. Subsequently, for the planned conversion to the reversible Francis turbine, including shape modifications of the flow paths, a steady numerical simulation of the flow in the penstock was performed in both turbine and pump modes. By analysing the resulting pressure and velocity fields and comparing them to standards and scientific literature, the values of the uncertainty in the flow measurement were calculated. The outcome is a straightforward evaluation and comparison of three main flow measurement methods: current meter, pressure–time, and ultrasonic transit time.

Practical approach for determining material parameters when predicting grain size after static recrystallization

The static recrystallization (SRX) behavior of 42CrMo steel was investigated by a non-isothermal hot compression test. The conventional empirical equation for SRX was applied to a finite element (FE) simulator using a nonconventional manner of parameter characterization. Parameter characterization was conducted by applying the FE-coupled optimization method to cylinder compression, which is an axisymmetric upsetting process. An accurate and practical flow stress model was used to precisely determine the critical strain value and define the instant when SRX occurred. The results revealed that the microstructural evolution during SRX was substantially influenced by temperature, strain rate, degree of deformation, and initial austenite grain size. The predicted recrystallized SRX grain sizes exhibited good agreement with the measured values when the model's parameters for the calculation of grain size were obtained using a FE-coupled optimization scheme.

Flow Curve of Superalloy 718 under Hot Forming in a Region of &lt;i&gt;γ&lt;/i&gt;” Precipitation

This study aims to formulate a constitutive equation to accurately determine the flow stress of superalloy 718 for effective production of gas turbine disks. A hot-compression test with superalloy 718 at temperatures ranging from 900 to 1000°C, a reduction rate of 67%, and strain rates of 0.1, 1, and 10 s−1, respectively, was conducted to analyze the flow stress in the dynamic precipitation region. An accurate flow stress curve was obtained for each strain rate and initial temperature. The flow curves obtained at a deformation temperature of 900°C and strain rates of 0.1 and 1 s−1, represent a combination of work-hardening and dynamic recovery. Dynamic recrystallization (DRX) behavior was observed under other deformation conditions. At a deformation temperature of 950°C and each strain rate, the strain at the onset of DRX (εc) decreases, and DRX tends to occur rapidly. In addition, the steady-state stress at a strain rate of 1 s−1 was greater than that at a higher strain rate of 10 s−1. The lowest steady-state stress among all the experimental conditions was observed at a strain rate of 10 s−1. This may be attributed to the role of nucleation sites, precipitation hardening caused by dynamically precipitated γ” phases at approximately 950°C and a strain rate of 1 s−1, and dynamic softening effects due to significant heat generated by deformation at a strain rate of 10 s−1. A new constitutive equation for the generalized flow curve of superalloy 718 was obtained by considering these metallurgical phenomena.

A deep-learning approach for reconstructing 3D turbulent flows from 2D observation data

Turbulence is a complex phenomenon that has a chaotic nature with multiple spatio-temporal scales, making predictions of turbulent flows a challenging topic. Nowadays, an abundance of high-fidelity databases can be generated by experimental measurements and numerical simulations, but obtaining such accurate data in full-scale applications is currently not possible. This motivates utilising deep learning on subsets of the available data to reduce the required cost of reconstructing the full flow in such full-scale applications. Here, we develop a generative-adversarial-network (GAN)-based model to reconstruct the three-dimensional velocity fields from flow data represented by a cross-plane of unpaired two-dimensional velocity observations. The model could successfully reconstruct the flow fields with accurate flow structures, statistics and spectra. The results indicate that our model can be successfully utilised for reconstructing three-dimensional flows from two-dimensional experimental measurements. Consequently, a remarkable reduction in the complexity of the experimental setup and the storage cost can be achieved.