Land drainage and flood-relief pumping stations are crucial infrastructure to safeguard lives, property, and agriculture from the devastating threat of flooding. However, these pumping stations create unwanted barriers to fish migration and can pose health risks, particularly for anguillid eels, which primarily live in freshwater but migrate to the ocean for reproduction. Fish-safe pumps have the potential to mitigate fish injury and mortality, but assessments of conditions during fish passage are lacking. This case study features a state-of-the-art full-scale live-fish pump testing facility, equipped with an axial-flow pump with a fish-safe design, to inform fish-focused pump design and optimize hydraulic conditions prior to live-fish testing. The potential for damage to passing fish was numerically investigated at optimal operating conditions using volume-based criteria, analyzing predictions from Computational Fluid Dynamics (CFD). Specifically, a damage index ( ξ ) was obtained for five hydraulic stressors that exceeded injury thresholds within rotating parts of the pumping equipment. In terms of hydrodynamic damage, the rate of pressure change ( d p / d t ; 3.2%) had the highest negative effect on fish, followed by damage caused by wall shear stress ( τ 0 ; 0.8%), and spatial velocity gradient ( d u / d y ; 0.07%), and thus should be minimized in future designs. The hydrodynamic damage due to rapid pressure changes (RPC) or low absolute pressure ( p abs ) within the impeller was negligible. Consequently, for the studied pump, the main priority should be optimizing rotor blade geometry while maintaining pumping efficiency, followed by reducing the risks of rapid pressure change and shear damage. Overall, this numerical assessment of the axial-flow pump provided invaluable insights into mechanisms of fish damage, informing future further improved designs without the ethical concerns and logistical challenges associated with live fish testing. • Eel passage conditions through a fish-safe axial-flow pump were assessed using a CFD model at the best efficiency point. • Several hydraulic stressors, relating to fish/eel damage, were numerically evaluated. • Predicted damage index values for each hydraulic stressor were in the range 0–0.032. • Rate of pressure change, ratio of pressure changes, and wall shear stress were the most dominant.

Development of weighted efficiency for photovoltaic inverters: A Brazilian case study

Abstract Rotordynamic forces acting on impellers and turbines in multistage turbomachines can significantly influence rotor critical speeds. In axial-flow pumps with unshrouded impellers, fluid-induced lateral forces are impacted by the unsteady tip leakage flow across the blade tips. This study investigates rotordynamic forces in a helico-axial compression cell, in a multistage configuration. A transient computational fluid dynamics (CFD) approach, employing frame change models, is used to determine fluid-induced forces on the whirling rotor. Whirl angular frequencies from −0.75 to 1.10 times the rotor angular frequency were simulated for two operating conditions: the best efficiency point (BEP) and 75% of BEP. The simulations revealed destabilizing forces for negative whirl frequencies and generally stabilizing forces for positive whirl frequencies. Overall, the forces were larger at part-load conditions, relative to those at BEP. Rotordynamic coefficients were derived, showing positive direct stiffness, but near-zero values at part load. Negative tangential rotordynamic forces at small whirl frequencies resulted in significant negative cross-coupled stiffness coefficients, also in BEP. Variations in impeller inlet flow angles, linked to the local rate of backflow, affected each impeller channel's pressure increase, varying by approximately ±3% at 75% of BEP. This results in a net lateral force, ultimately responsible for the negative cross-coupled stiffness coefficients. The study demonstrates how significant rotordynamic forces can arise from hydrodynamic effects linked to impeller backflow.

DyReMamba for efficient point cloud analysis with dynamic reordering and bidirectional state space modeling.

In response to the increasing complexity of industrial robot assembly lines under the intelligent manufacturing paradigm, as well as the industry pain points of low efficiency, high costs, and significant safety risks associated with traditional physical debugging, this research applies Digital Twin (DT) technology to the entire lifecycle of assembly line development. Based on the five-dimensional DT model theory, a high-fidelity virtual twin environment is constructed by integrating geometric, physical behavior, and PLC control logic modeling. Utilizing methodologies such as literature analysis, kinematic modeling, and cyber-physical mapping via the OPC UA protocol, simulation analyses—including virtual commissioning and collision detection—are conducted. These processes achieve pre-verification of control logic and precise optimization of production cycles. Experimental results demonstrate that the developed DT system accurately identifies control logic defects during the design phase, reducing the commissioning cycle by over 30% and significantly enhancing robotic operational efficiency without the need for physical hardware. These findings provide theoretical support and a practical engineering reference for the digital transformation of automated production lines.

Bio-invasion and decadal changes in the trophic dynamics of a temporarily closed estuary: An Ecopath model from Veli-Akkulam Estuary, Kerala, India.

Corrigendum to “An improved probability efficient point method based on outer approximation for unit commitment of multi-area power systems” [Results in Engineering, Volume 28, (2025), Article Number 108347

In this paper, we mainly investigate the (proper) efficiency in nonconvex multi-objective optimization. We observe that the asymptotic cone of a set is closely linked to both the existence and boundedness of the efficient points of the set. First, we introduce several new concepts: (C-) asymptotically regular sets and (C-) asymptotically convex sets, which generalize the regularity and convexity, respectively. Secondly, we demonstrate that if a closed set is C-asymptotically regular (convex), then the existence of (properly) efficient points, domination property, C-boundedness, C-semicompactness and C-compactness of the set are equivalent. Furthermore, by analyzing the relations between the asymptotic cone of a set and the ordering cone, we establish the boundedness (or unboundedness) of the efficient point set. Finally, we apply these results to the piecewise convex multi-objective optimization problem.

Periodontal clinical trials commonly use surrogate end points (e.g., probing pocket depth [PPD], clinical attachment level [CAL], bleeding on probing [BoP], and composite scores such as the Periodontal Risk Score [PRS]) for efficiency, yet tooth loss, treatment burden/cost, and re-treatment are the definitive patient-relevant measures of long-term success. This mixed-methods study evaluated the predictive performance of three surrogate definitions and contextualized the findings with a 50-year narrative review of longitudinal evidence linking PPD, CAL, and BoP to tooth retention. Additionally, a retrospective cohort (2001-2024) of 160 patients (919 teeth) treated at the University of Michigan was studied. Each participant had baseline PPD ≥6 mm after initial therapy (sites nonresponsive to initial nonsurgical therapy) and received up to 20 years of supportive periodontal therapy (SPT) following active periodontal therapy (APT). Three surrogate definitions: TEP A (no pockets ≥6 mm), TEP B (pockets ≤4 mm without BoP), and the PRS were compared against two true end points: tooth loss due to periodontitis (TLP) and need for additional therapy (re-treatment). Logistic regression with generalized estimating equations was used to estimate sensitivity, specificity, positive/negative predictive values, and area under the curve (AUC). Overall, tooth loss was 28.4%, with 18.9% due to periodontitis. Additional treatment was required by 91.5% of patients. For TLP, TEP-B provided the highest sensitivity, whereas PRS = 3 provided the highest specificity and positive predictive value (AUC = 0.556, p = 0.021). PRS ≥2 showed the best overall discrimination/accuracy for TLP (AUC = 0.637, p < 0.001). TEP-A and TEP-B demonstrated modest discrimination (AUC = 0.567, p < 0.007 and AUC = 0.549, p = 0.047, respectively). For re-treatment, TEP-B yielded 82.2% sensitivity, and PRS = 3 demonstrated 96.6% specificity (PPV = 96.8%); PRS ≥2 alone achieved AUC = 0.625 (p = 0.033), while no significant differences were observed among the remaining models. For practical application, baseline risk stratification can be performed using PRS ≥2 to support early screening for periodontitis-related tooth loss, while PRS = 3 can be reserved for high-confidence confirmation when minimizing false positives is critical. During postsurgical follow-up after APT and throughout SPT, TEP-B (PPD ≤4 mm with no BoP) can be used as the primary monitoring target to flag likely instability and identify teeth likely to require re-treatment, whereas PRS = 3 can be applied to guide final treatment decisions or resource-intensive interventions.

Purpose This study aims to deliver a high-fidelity, open-literature Computational Fluid Dynamics (CFD) dataset and validated aerodynamic performance curves for the commercial DJI 9450 self-tightening propeller in hover across the practical quadcopter operating range (2,000–3,500 RPM). By combining a refined SST k-ω MRF methodology (GCI &amp;lt; 0.9%) with rigorous validation against NASA Ames, UIUC, TUM and DJI experimental benchmarks (errors &amp;lt; 1.5%) establishes the first comprehensive thrust, torque, power, and efficiency characteristics for this exact geometry. The work identifies the optimum efficiency point (η_max = 76.4% at 3,100 RPM) and quantifies performance degradation mechanisms at higher RPM, providing immediate design guidelines for commercial multirotor UAVs. Design/methodology/approach A high-fidelity steady-state CFD framework was developed using ANSYS Fluent 2024 R2. The exact DJI 9450 self-tightening propeller geometry (239 mm diameter) was modelled. A cylindrical domain with Multiple Reference Frame (MRF) was employed. A hybrid hexa-dominant mesh (6.96 × 106 cells, 15 prism layers, y+ &amp;lt; 0.95) was generated and grid convergence verified (GCI = 0.82% for thrust). The SST k-ω Turbulence model with low-Re corrections and second-order discretization was used. Simulations were performed at 2,000, 2,500, 3,000 and 3,500 RPM. Results were rigorously validated against NASA/CR-2017-219428 (Nowicki, 2017), Brandt and Selig (2017), Theile (2016) and DJI official reports (mean errors &amp;lt; 1.5%). Findings CFD simulations reveal that thrust rises quadratically from 1.57 N (2,000 RPM) to 5.91 N (3,500 RPM), while required power increases from 28.4 W to 148.6 W. Propeller efficiency peaks at 76.4% at 3,100 RPM and drops sharply beyond 3,200 RPM due to intensified tip vortices and local Mach number reaching 0.41. Turbulence kinetic energy at blade tips exceeds 87 m2/s2 at 3,500 RPM, causing significant energy dissipation. Validation against NASA/CR-2017-219428 (Nowicki, 2017), UIUC, TUM and DJI datasets confirms thrust and torque predictions within 1.5% error, establishing 3,000–3,200 RPM as the optimum operating window for the DJI 9450 propeller in hover. Originality/value This study presents the first open-literature, high-fidelity CFD dataset and validated performance curves (thrust, torque, power, η) for the exact DJI 9450 self-tightening propeller across the realistic quadcopter hover range of 2,000–3,500 RPM. Unlike previous generic low-Re or high-RPM studies, it achieves grid-converged solutions (GCI &amp;lt; 0.9%) and rigorous validation against four independent experimental benchmarks, including NASA/CR-2017–219428 (Nowicki, 2017), with errors below 1.5%. The work identifies the precise efficiency peak (η = 76.4% at 31,00 RPM) and quantifies tip-vortex-induced losses, delivering immediately usable design data for millions of commercial Phantom-series and similar multirotor UAVs worldwide.

To meet the strict non-contact measurement requirements for the assembly of aircraft engine pipelines and to overcome the limitations of the traditional three-dimensional laser scanning workflow, this study proposes an automated pipeline point cloud processing and deviation analysis framework. Through a standardized three-dimensional laser scanning procedure, high-resolution pipeline point clouds are obtained and preprocessed. Based on the geometric characteristics of the pipeline, automated algorithms for point cloud feature segmentation, axis extraction, and model registration are developed. Particularly, the three-dimensional extended Douglas-Peucker (DP) algorithm is introduced to achieve efficient point cloud downsampling while retaining necessary geometric and structural features. These algorithms are fully integrated into a unified software platform, supporting one-click operation, and can automatically analyze and obtain five key types of pipeline deviations: angular deviation, radial deviation, axial deviation, roundness error, and diameter error. The platform also provides intuitive visualization effects and comprehensive report generation functions to facilitate quantitative inspection and analysis. Test results show that the proposed method significantly improves the processing efficiency and measurement reliability of complex pipeline systems. The developed framework provides a powerful practical solution for the automated geometric inspection of aircraft engine pipelines and lays a solid foundation for subsequent quality assessment tasks.

This paper presents a model-free Best Efficiency Point (BEP) tracking method for centrifugal pumps without head measurement or manufacturer-provided characteristic curves. The proposed approach combines a discrete finite-difference extremum-seeking control (ESC) scheme with an efficiency approximation proxy derived from measurable variables—namely, flow rate and electrical power. Under constant head conditions, the proxy function is analytically shown to be proportional to the true pump efficiency, enabling real-time BEP localization using only motor feedback signals. The ESC algorithm employs a sign-based gradient rule with adaptive step-size reduction to achieve rapid and stable convergence without mathematical models. A Python-based simulation using a Schneider SUB 15-0.5cv pump demonstrates that the method can track the BEP with negligible steady-state error (less than 0.1% efficiency deviation). The proposed framework offers a cost-effective solution for efficient optimization for mobile pumping applications in large water resources where installing head sensors is impractical.

Spatially-enhanced Spiking neural network for efficient point cloud analysis.

Abstract In this paper, the influence of different back vanes and back channels on fiber entry into the back shroud cavity is experimentally investigated. Vane and channel contours correspond to backward curved blades of a semi-open two-channel impeller. The configurations feature varying axial gap widths between back shroud and rear housing wall. Vane height and channel depth are also changed throughout the tests. Vane and channel width as well as number are constant. The axial gap width between rear housing wall and back vane top or channel ground is constant as well. For each configuration, the operating points part load, best efficiency point and overload are examined. Tested configurations are compared to unmodified reference configurations regarding fiber dry mass accumulated in a defined area within the back shroud cavity after a test duration of 60 min. The experiments demonstrate that selected back vane configurations reduce fiber entry by up to 100%, causing minor decreases in operating point-averaged overall efficiency of 0.8% and less. The back channel configuration investigated at minimum axial gap width eliminates the minor fiber entry of the reference configuration, decreasing operating point-averaged overall efficiency by 0.9%. In contrast, mean fiber entry is significantly increased by the back channels at larger gap widths by up to 0.679 g. Except for one back channel configuration with minor fiber entry of up to 0.046g, none of the presented back vane or back channel configurations shows fiber entry at part load.

Large language models (LLMs) require considerable computation and energy resources during training and deployment. While scaling laws for training have guided much recent progress, inference costs represent a significant and growing component of the overall resource burden, particularly for reasoning models. Existing compute-optimality characterizations that consider model size, dataset size and inference tokens in isolation or fixed combinations may overlook more efficient operating points. We introduce directed stochastic skill search (DS3), a general framework that represents inference as stochastic traversal over a learnt skill graph. From a simplified yet expressive instantiation, we derive closed-form expressions for task success and compute cost across a wide range of inference strategies-including chain-of-thought (CoT) and tree-of-thought (ToT)-enabling comparisons by task difficulty and model capability. We extend a prior graph framework of LLM training to include inference and bridge DS3 with empirical scaling laws. We theoretically recover observed patterns, including linear accuracy scaling with log-compute, variation in preferred inference strategies by task and capability, emergent behaviour elicited by reasoning despite parameter plateaus and both best-of-N and majority voting (MV) captured within one framework. By characterizing training-inference interdependencies, our framework deepens theoretical understanding and supports principled algorithmic design and resource allocation. This article is part of the discussion meeting issue 'Bits, neurons and qubits for sustainable AI'.

Automated detection of surface defects on three-dimensional (3D) parts is vital for ensuring product quality and safety in manufacturing. However, three key challenges hinder reliable detection: geometric context ambiguity across complex part shapes, domain mismatch between generic pretrained features and industrial scans (with their unique noise and reflectivity), and the scarcity of diverse defect examples for training. To overcome these issues, we propose a novel single-forward-pass framework for point cloud anomaly detection, comprising three new modules: (1) Spatial Context Aggregation, which grounds each local patch in a set of learned global prototypes via an optimal-transport alignment to resolve context ambiguity; (2) Feature Adaptor, a lightweight two-layer multilayer perceptron (MLP) that fine-tunes self-supervised Point-MAE embeddings to the specific characteristics of industrial scans; and (3) Selective Anomalous Feature Generator, which synthesizes realistic hard negatives by corrupting random subsets of feature tokens, thus mitigating the need for extensive defect labels. An attention-based discriminator trained with patch-wise supervision learns to distinguish these hard negatives from genuine defect-free patterns. At inference, our pipeline delivers dense per-point anomaly scores in a single pass at up to 13.5 frames per second (FPS). On the Real3D-AD benchmark, we observe point-level improvements of 2.8% in area under the receiver operating characteristic curve (AUROC) and 5.7% in area under the precision-recall curve (AUPR), with object-level gains of 3.0% (AUROC) and 3.5% (AUPR). Evaluated on our newly released Industrial3D-AD dataset, which captures realistic sensor noise and reflective materials, we see similar enhancements (2.9%/5.3% point-level, 2.8%/3.3% object-level).

Abstract Transmitting 3D point cloud data through wireless networks is challenging, as it entails balancing the demand for precise reconstruction with energy efficiency and consistent performance under changing channel conditions. These issues stem from the high‐dimensional nature of the data and the dynamic wireless communication environment. To address these challenges, we introduce latent space power control (LSPC), a system that features a novel loss function designed to optimize three main objectives: maintaining reconstruction accuracy, reducing energy consumption, and achieving the target signal‐to‐noise ratio (SNR). To preserve the geometric structure of the data, the system employs PointNet++ and a dynamic graph convolutional neural network for feature extraction during compression and reconstruction. Experimental results show that LSPC provides better reconstruction quality than existing methods and uses power more efficiently across various SNR levels. It also performs reliably under adverse wireless conditions, making it a valuable solution for 3D point cloud communication in applications such as autonomous vehicles, augmented reality, and environmental monitoring.

The increasing digitalization of retail operations has intensified the need for efficient and reliable Point of Sale (POS) and inventory management systems. Small and medium enterprises (SMEs) often struggle with separate and unintegrated systems that lead to data inconsistencies, manual errors, and operational inefficiencies. This paper presents POSventory, an integrated POS and inventory management platform designed to synchronize sales, billing, and stock data in real time. Built using Spring Boot and MySQL, the system integrates authentication via JWT and Spring Security, along with secure payment gateways such as Razorpay and Stripe. The system offers modules for sales processing, inventory tracking, and analytics visualization, making it scalable and cost-effective for SMEs. The architecture, workflow, and algorithmic models have been analyzed for performance and feasibility. Experimental results demonstrate a significant improvement in synchronization speed and operational transparency.

Abstract. The complex geometry of Javanese architecture poses significant challenges for 3D semantic segmentation in cultural heritage documentation. This study evaluates state-of-the-art Point Transformers, i.e., PTv1, PTv2, PTv3, and LitePT, on the Sewu temple dataset, focusing on robustness and efficiency. While PTv1 and PTv2 achieve the highest Intersection-over-Union (mIoU 0.71), they incur high computational costs. Conversely, LitePT provides an optimal balance, delivering competitive results (0.69 mIoU) while being drastically faster. Furthermore, experiments with limited data reveal the significant benefits of transfer learning from European heritage datasets. We conclude that efficient Point Transformer architectures are promising for the automated understanding of complex non-European monuments.

We implement proximal point type algorithms for finding an efficient point for nonconvex multiobjective optimization problems in which the objective functions are quasiconvex and satisfy a prox-convexity assumption. Our proposed algorithms combine the proximal point method for the minimization problem with the infeasible projection method for variational inequalities and their generate iterative sequences that converges to efficient solution points of the multiobjective optimization problem under mild assumptions. Furthermore, we also propose accelerated versions of the proposed algorithms by adding an inertial term and we established the nonasymptotic O ( 1 k ) convergence rate, too. Numerical illustrations shows the practical usability of the proposed algorithms.