We study the evolution of interface instabilities between Newtonian and shear-thinning fluids in a planar Hele-Shaw cell. This work builds upon a two-dimensional gap-averaged model previously developed for flow simulations involving power-law fluids in such geometries. The focus is on interfacial instability, viscous fingering dynamics, and quantifying the growth and suppression mechanisms governing perturbation evolution, specifically the growth rate of an initially perturbed fluid-fluid interface. We conduct gap-averaged simulations using the CFD software OpenFOAM and track the spatiotemporal evolution of an initial sinusoidal fluid-fluid interface. We compare the growth or decay rate of the initial perturbation amplitude to available analytical (linear stability) result for generalized Newtonian fluids. The results demonstrate good agreement between the 2D simulations and the theoretical linear stability analysis for Newtonian fluids, while showing notable deviations for power-law fluids, reflecting the added complexity of non-Newtonian behavior. The influence of key parameters on the stability and dynamics of viscous fingering is systematically investigated, including rheological properties of power-law fluids such as the consistency index k and flow behavior index n, interfacial tension, and effective friction pressure gradients at the interface. For power-law fluids, interface stability is strongly influenced by the interplay between k and n. Stronger shear-thinning behavior, associated with lower n, enhances stability by increasing the effective viscosity at low shear rates, while lower k values reduce flow resistance, promoting instability and the development of viscous fingers. The effective friction pressure gradient at the interface also plays a critical role in driving instability, where a higher positive gradient promotes the development of viscous fingers, particularly under conditions of varying fluid rheology.

Aerodynamic drag dominates the resistive forces in many sports at racing speeds, and small athlete posture changes can produce practically meaningful changes in performance. Standard approaches like field testing, wind-tunnel testing and computational fluid dynamics (CFD) can provide accurate results, but are time-consuming. Recent advances in scientific machine learning have enabled surrogate models that predict flow quantities at a fraction of the computational cost of a full CFD simulation. However, most applications currently focus on industrial geometries, such as cars or aircraft, where the surface is morphed to generate large datasets for training. It remains unclear how well these methods transfer to athlete geometries where variability is dominated by articulated pose changes rather than smooth shape morphing. In this work, a dataset for cyclist aerodynamics was generated by combining 12 scanned athlete geometries with 20 postures per athlete. CFD simulations were performed for these 240 geometries, which were then used to train a state-of-the-art surrogate model. The generalization of the model to an unseen geometry was investigated, along with the balance between number of positions and number of unique geometries in the dataset. The surrogate model predicts drag area with a mean absolute percentage error of < 3

Urban planning authorities often require pedestrian-level wind (PLW) assessments using computational fluid dynamics (CFD) for permitting. While PLW is typically conducted with Reynolds-averaged Navier–Stokes (RANS) simulations, there is ongoing debate among practitioners about using costlier Large Eddy Simulations (LES). This can be further complicated by an inconsistent definition of mean wind speed. This study compares RANS and LES across three benchmark cases: an isolated 1:1:2 block (A), the Michelstadt array (B), and the Niigata district (C). The results were validated against wind tunnel data using two wind speed metrics: magnitude of time-averaged velocity (smv) and time-average of instantaneous speed (sms). Results showed that LES performed better in predicting wind speeds, with higher Pearson correlations: Case A had LES at 0.97 and RANS at 0.86, while Case B showed LES at 0.93 and RANS at 0.72. For Case C, LES with sms achieved the highest correlation (over 0.82) and the lowest average relative deviation across eight wind directions. However, LES-smv underpredicted low-wind-speed areas similarly to RANS. Thus, RANS and LES validation discrepancies can arise from wind speed definitions. In terms of flow physics, RANS produces less mixing in wakes and prolonged wind deficits, resulting in milder Lawson comfort map activity classes.

High-speed wind turbine bearings play a critical role in ensuring operational reliability and efficiency, making accurate remaining useful life (RUL) prediction is essential for predictive maintenance. Several remaining useful life prediction algorithms have been discussed by many researchers, but there is a lack of degradation trend extraction based on deep learning algorithms. However, there are also several forecasting algorithms available, and it is important to study their accuracy and effectiveness to select the most effective one. The purpose of this study is to analyse the RUL prediction accuracy for different Algorithms: Facebook Prophet, Auto-regressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU). A long-short-term memory encoder (LSTM-AE) is first used to extract the degradation trend from vibration signals and then feed it to the forecasting algorithms. The utilized dataset in this study contains open-source data about the progressive failure of the high-speed bearing of the wind turbine due to an inner race fault. The results show that the proposed combination of LSTM-AE and Prophet forecasting is the most effective with an accuracy of 93.18

In wind engineering simulations, Large Eddy SimulationsLarge Eddy Simulation (LES) (LES) can generate more accurate results than simulations based on Reynolds-averaged Navier-Stokes (RANS) models. However, inflow boundary conditions for LESLarge Eddy Simulation (LES) are more challenging than for RANS. This study applies LESLarge Eddy Simulation (LES) to flow around an isolated building to investigate the effect of various inflow conditionsInflow conditions. Four methods are used to produce inflows at the boundary inlet: precursor simulation, digital filter method (DFM), Prescribed Wavevector Random Flow Generator (PRFG ^3 ), and turbulenceTurbulence-free mean profile. Overall, all methods with turbulenceTurbulence generation at the inflow produced accurate results for mean flow and Reynolds stresses. The turbulenceTurbulence-free inflow case was accurate downstream of the building, indicating that turbulenceTurbulence generated by the building is sufficient for accurate wake predictions. The DFM inflow increased computational time by 172 ^3 increased the time by 20

Wind turbine blades are critical components, and their structural integrity is essential for uninterrupted operation and minimizing downtime. Although various methods are used to monitor the health of wind turbine blades, several research challenges persist, such as the reliance on manual feature engineering and the limited availability of large amounts of labeled data. In this study, a novel approach is proposed that will overcome the limitations of manual feature extraction and label data challenges. In the proposed work, time series vibration signals from the blade are first converted into spectrograms and passed through a CNN-based autoencoder that is trained solely on healthy data to learn a compact latent representation. Anomalies are then flagged in three complementary ways: (i) by thresholding the autoencoder’s reconstruction error, (ii) by applying an Isolation Forest to the latent features, and (iii) by evaluating the same features with a One-Class SVM. The outputs of these detectors are subsequently benchmarked, providing a systematic comparison of their ability to discriminate between vibration-induced faults, such as cracks and erosion, and normal operation on a controlled test-rig dataset, the autoencoder achieves 97.2 % accuracy, outperforming the Isolation Forest and One-Class SVM by 8%–27%. These results demonstrate that zero-label, deep-feature pipelines can deliver reliable and scalable blade-fault detection, paving the way for more cost-effective predictive maintenance in wind farms.

Computational and experimental investigations of flow over athletes are key elements of sports fluid dynamics to analyse performance and equipment design. However, the use of athlete-specific models often limits reproducibility, collaboration, and data sharing due to ethical and competitive constraints. This paper introduces and defines Generic Athlete Models as openly available, standardized geometries accompanied by benchmark flow datasets. The case is made that such models are essential to improve the reliability, comparability, and transparency of fluid dynamics research in sports. While only a few generic athlete models currently exist, this paper outlines clear directions for their further development and broader adoption.

This article addresses the challenges of assessing pedestrian-level wind conditions in urban environments using a deep learning approach. The influence of large buildings on urban wind patterns has significant implications for thermal comfort, pollutant transport, pedestrian safety, and energy usage. Traditional methods, such as wind tunnel testing, are time-consuming and costly, leading to a growing interest in computational methods like computational fluid dynamics (CFD) simulations. However, CFD still requires a significant time investment for such studies, limiting the available time for design modification prior to lockdown. This study proposes a deep learning surrogate model based on a MLP-mixer architecture to predict mean flow conditions for complex arrays of buildings. The model is trained on a diverse dataset of synthetic geometries and corresponding CFD simulations, demonstrating its effectiveness in capturing intricate wind dynamics. The article discusses the model architecture and data preparation and evaluates its performance qualitatively and quantitatively. Results show promising capabilities in replicating key wind features with a mean error of 0.3 m/s and rarely exceeding 0.75 m/s, making the proposed model a valuable tool for early-stage urban wind modelling.

We study the entrance length in eccentric annular geometries, where the axes of the inner and outer pipes forming the annulus are offset from each other. Such geometries are considered relevant for several industrial applications, such as drilling of wells for hydrocarbon production or geothermal energy recovery, as well as biomedical research involving annular vessels. Previous studies have shown that eccentricity increases the entrance length in annular geometries, and this has been attributed to azimuthal redistribution of fluid between the wide and narrow sides of the annulus. The present computational study aims to increase the understanding of entrance lengths in eccentric annuli for laminar flow with Reynolds numbers spanning the range from the creep limit and up to intermediate laminar values, below the regime of annular gap instabilities. We find that the entrance lengths in eccentric annuli generally increase for narrower annular gaps (higher aspect ratios). Moreover, the entrance lengths in the annuli with higher aspect ratios are also more sensitive to the degree of eccentricity compared to wider annuli (lower aspect ratios). We report empirical correlations for the dimensionless entrance length using the same form as earlier studies ( Le=[C0n+(C1Re)n]1/n). We find that eccentricity, which breaks the axial symmetry of the annulus and causes azimuthal flow in the entrance region, impacts the value of the exponent n, which has been assigned the value of 1.6 in previous studies of axisymmetric conduits.

This work presents a physics-based two-dimensional model for simulating displacement flows of power-law fluids in Hele-Shaw cells. The model is derived by approximating fully developed velocity profiles across the gap-wise direction and averaging the mass and momentum conservation equations, resulting in a two-dimensional formulation that efficiently captures complex fluid dynamics. Implemented in OpenFOAM, this approach achieves computational speeds over 200 times faster than comparable 3D simulations, while preserving the accuracy of displacement dynamics. Validated against 3D DNS results and experimental data, this 2D model accurately replicates observed flow phenomena. Simulations of over 70 cases examined the effect of the ratio of friction pressure gradients (RFG) between fluid pairs on interface stability. Results show that RFGs below unity maintain a flat interface, while higher values induce viscous fingering. In cases with RFG closer to unity, a longer duct or extended displacement time is required for significant finger growth.

There is an increase in reliance on large-eddy simulations (LES) over traditional Reynolds-Averaged Navier- Stokes (RANS) simulations for conducting urban wind studies because of their potential to capture detailed flow characteristics and unsteady flow phenomena. Validation remains a crucial aspect of computational fluid dynamics (CFD) analysis. Yet, LES validation often relies on traditional RANS-based metrics that focus on mean quantities, despite LES providing richer flow details. With adequate LES validation guidelines lacking in the computational wind engineering literature, this paper introduces anew validation metric tailored for LES in urban wind studies. This study uses the "Michelstadt" test case, a semi-idealized model of a generic European city, to demonstrate the metric's evaluation. It begins by assessing the importance of mesh sensitivity and inflow generation techniques in achieving high-fidelity LES results. Then, the proposed metric, called the overall area metric (OAM), improves the evaluation of LES results by quantitatively comparing the cumulative density functions (CDFs) of the velocity time series of LES with experiments. The LES results for mean velocity and Reynolds stresses align well with the experimental data based on traditional hit rate and factor of two metrics both within and above the urban canopy layer (UCL). The OAM reveals poor results above the building compared to the results within the UCL for the mean streamwise velocity. Therefore, the OAM metric accurately represents velocity distributions, allowing validation of a wider range of wind speeds, unlike previous metrics. This is important in recent LES studies on rare high-wind events, such as gusts.

This study investigates fluid forces and viscous torque on an inner cylinder that simultaneously rotates about its own axis and orbits within an outer cylinder. This problem is particularly relevant for industrial applications such as drill string dynamics and journal bearings, where understanding fluid-structure interactions is crucial for predicting and controlling system behavior. We combine analytical solutions valid in the weakly inertial regime with computational fluid dynamics simulations to examine how these forces depend on the annulus geometry, eccentricity, and the relative magnitudes of rotational and orbital motion. Our results show that available analytical solutions provide fair predictions of the radial and tangential force components, as well as the viscous torque when inertial effects are weak. For stronger inertial effects, we develop approximate expressions that extend beyond the weakly inertial regime. We identify a geometry-dependent orbital speed at which the tangential force component vanishes, and show that the stability of orbital motion depends on the aspect ratio. High aspect ratios (narrow annulus) lead to significant pressure build-up and net centralizing forces, while lower aspect ratios (wider annulus) may permit stable orbital configurations. The developed expressions are particularly relevant for vibration modeling of drill strings or journal bearings where the structure is coupled with the fluid forces.

Evaluating the pedestrian-level wind environment (PLWE) during urban design and planning is important due to its impact on air quality, human comfort, and safety. This study uses datasets from large eddy simulation (LES) and wind tunnel experiment (WTE) to predict low-occurring wind speed (LOWS) events in a complex European urban layout known as the "Michelstadt" case. The study first compares LES results of high-order statistics with those of the WTE and then investigates the relationship of wind statistics with gust factor (GF) and peak factor (PF). Positive skewness values are found at all measured locations for wind speed, while negative skewness values are reported for the wind velocity components on the street canyons perpendicular to the incoming wind direction at pedestrian height. The skewness of both speed and velocity components showed the highest correlation with PF for all exceedance probability values compared to the coefficient of variation, standard deviation, and kurtosis. The integral timescale showed a marginal correlation with the PF and GF. The relationships of statistics with GF and PF are then compared with the theoretical curves based on the Weibull and GramCharlier series (GCS) for estimating low-occurring events of wind speed and velocity components. The findings indicate that the Weibull distribution accurately represents the wind speed and related PF and GF but shows limitations for velocity components where GCS provides better results. The study emphasizes the need for improved statistical models to predict wind velocity components and calls for more experimental studies to compare higher-order statistics with LES.

Pedestrian-level wind environments are strongly influenced by urban morphology, with large and tall buildings playing a significant role. City authorities increasingly mandate assessments of pedestrian wind conditions before approving new construction. Computational fluid dynamics (CFD) models can provide a detailed understanding of the aerodynamic environment; however, in early design stages, urban morphologies are subject to change, requiring multiple simulations, adding substantial financial and time burdens to projects. To address this challenge, we develop a deep learning approach for the rapid inference of pedestrian-level wind conditions using a multi-layer perceptron (MLP)-mixer architecture. By embedding 3D structural details into the training data, our model can infer wind conditions around complex structures such as lift-up designs and skyways while maintaining inference times on the order of fractions of a second. This extends the capabilities of deep learning models that typically reduce the problem to a 2D image-to-image translation task, omitting crucial structural details. We conduct an extensive evaluation of our model and compare its performance to the widely adopted UNet architecture, demonstrating that the MLP-mixer outperforms UNet across all evaluation metrics. Notably, the MLP-Mixer achieves a mean squared error approximately 2.6 times lower, a peak signal-to-noise ratio 3.7 dB higher and the highest recorded structural similarity index of 0.991. These results indicate improved agreement with the reference CFD data. We anticipate that the MLP-mixer model will serve as a valuable tool in early-stage urban design workflows, enabling faster and more efficient wind assessments.

In this study, we use a two-dimensional multiple relaxation time (MRT) approach for simulating polymeric fluids. A correction term is introduced into the source term to remove non-physical terms and improve numerical accuracy of the simulations. The correction term preserves the locality of the collision process and ensures numerical stability across a range of Weissenberg numbers when coupled with non-linear constitutive equations. This approach is applied to the Phan-Thien-Tanner (PTT) model and the Oldroyd-B model, where the first exhibits viscoelastic and shear-thinning behavior while the second is purely viscoelastic. To evaluate the numerical accuracy and stability of the proposed MRT-LBM approach, we apply it to planar Poiseuille flow as well as simplified four-roll mill benchmarks. In the case of the four-roll mill, we specifically examine the effects of shear-thinning and viscoelasticity in steady elongational flows and their transitions to oscillatory and chaotic or turbulent behaviors, known as elastic instability. Our results indicate that the non-linearity in the stress-strain rate relationship and the microstructural dynamics of polymer chains, as described by non-linear constitutive models, make the standard BGK-LBM approach incapable to accurately capture the complex behavior of polymers without introducing numerical artifacts. On the other hand the MRT-LBM method maintains numerical stability and accuracy across a broad range of Weissenberg (up to Wi = 20) and should therefore be the method of choice when simulating these types of flows.

Pedestrian wind comfort and safety are critical in urban design, especially as cities densify and climate change impacts intensify. While Computational Fluid Dynamics (CFD) simulations are increasingly used in these assessments, a universally agreed-upon number of wind directions for accurate results has yet to be established. This study addresses this gap by exploring the influence of urban morphology, characterized by different degrees of urbanization and building layouts, on the required number of wind directions for accurate assessments. Extensive CFD simulations were conducted in five geometrically diverse locations: Bryne, Oslo, London, Singapore, and New York. Pedestrian wind comfort maps were generated, and velocity amplification factor (VAF) dynamics were analyzed. The results suggest that urban complexity does not significantly affect the required number of wind directions for reliable assessments. The study provides practical guidance for selecting the number of wind directions based on the study’s focus: a minimum of 8 for basic assessments, 24 for high-accuracy assessments, and at least 36 for safety-focused assessments. This research significantly contributes to urban planning and design, empowering stakeholders with valuable insights for shaping resilient and comfortable urban environments.

Reverse-circulation cementing is an alternative strategy for well cementing where the cementing fluids are injected directly into the annulus from the surface. This cementing strategy can reduce downhole circulation pressures compared to conventional circulation cementing and potentially eliminate the need for retarders in the cement slurry. In reverse-circulation operations, the fluid hierarchy will normally involve density-unstable combinations along the annulus. Since the annular geometry prevents the mechanical separation of fluids, reverse-circulation cementing is associated with a risk of slurry contamination and mixing during placement. Although reverse-circulation cementing has been known for several decades and is used for cementing of both onshore and offshore wells, it remains unclear whether conventional circulation job design guidelines apply to reverse-cementing or indeed how fluid properties should be optimized for such operations. The purpose of the current study is to contribute to the understanding of buoyant annular displacements, with a particular focus on the role of viscosity hierarchy on the annular displacement in vertical and near-vertical annuli. We present a combined experimental and numerical study of density-unstable downward displacements in a downscaled, narrow concentric annulus. A transparent annulus flow loop was used to conduct downward displacements. A high-speed camera and a mirror arrangement were used to track the displacement. Numerical simulations of the experiments and selected other cases were performed using the open-source OpenFOAM computation framework. We study Newtonian and mildly shear-thinning fluids, and our study aims to determine whether it is more efficient to use a displacing fluid with higher viscosity or lower viscosity than the displaced fluid while maintaining a constant average viscosity for the fluid pair. The experimental and numerical results, which are in good qualitative agreement, demonstrate that the viscosity hierarchy of the fluids significantly affects the displacement flow features. Our results show that a more viscous displaced fluid leads to faster growth of the instabilities and, as a result, less efficient displacement. Oppositely, we observe less tendency for finger growth and a more diffusive mixing region for more viscous displacing fluids. The effect of the viscosity hierarchy can get stronger by increasing the inclination of the annulus and the viscosity difference between the fluids from 0.006 to about 0.02 Pa & sdot;s. The findings can assist in the selection of fluid properties for future reverse-circulation displacement operations.

Construction of wells for oil production, geothermal energy recovery or geological storage of carbon dioxide is performed in stages by drilling the rock formation to a certain vertical depth, and then isolating the drilled section by running and cementing a casing string in the hole. The conventional cementing strategy relies on displacing the annular space behind the casing from the bottom and toward the surface. The conventional circulation direction therefore places the denser cementing fluids below the original annular fluid. An alternative cement placement strategy is to displace the annular space from the surface and downward by injecting cementing fluids directly into the annulus. Such reverse circulation operations have the benefit of lower circulation pressures and a reduced risk of fracturing the formation during placement, but also leads to density-unstable displacement conditions and increased risk of fluid contamination. To better understand how the design of cementing fluids and their placement rate affect reverse circulation displacements, we performed a series of computational simulations of density-unstable displacements using a realistic three-dimensional geometrical model of a wellbore annulus. We address the effects of wellbore inclination and inner pipe eccentricity, with a particular focus on impacts of the fluid viscosity hierarchy on the annular displacement efficiency. Our results show that increasing the displaced fluid viscosity will act to suppress the tendency for buoyant backflow, while it can worsen displacement of the narrow side of the eccentric annulus. Transverse secondary flows, which are stronger in cases where the displaced fluid is less viscous, contribute to the displacement of the narrow, low side of the annulus. We also observe that increasing the imposed axial velocity will tend to stabilize the annular displacement and suppress backflow, in agreement with previous work for buoyant pipe displacements. The present computational study is a step toward understanding how buoyant, inertial, and viscous stresses affect density-unstable displacement flows for reverse circulation cementing.

Tidal stream turbines present an appealing approach for consistent renewable energy generation. Capitalizing on the periodic nature of tidal streams, these turbines can be designed to function bidirectionally, eliminating the need for a yaw control system. Although airfoils and hydrofoils in general have been extensively studied, there is less work available on bidirectional designs. Through the use of optimization methods, this work introduces bidirectional hydrofoils of varying thicknesses with improved performance compared to elliptic profiles. The optimization was performed using the hydrodynamic efficiency as the objective function, with simulations performed using 2D computational fluid dynamics. The optimized hydrofoil was then used in a turbine design, which is simulated using blade element momentum theory. The hydrofoils showed an improvement in hydrodynamic efficiency of up to 10 %. In the turbine design, this resulted in an improvement of 5.3 % in power coefficient and a reduction in thrust coefficient of 13.2 %. The new line of hydrofoils are promising for use in cost-effective and reliable tidal stream turbine designs.