Abstract Construction projects worldwide suffer from misaligned risk allocation, poorly chosen payment schemes, and opaque claims and dispute processes, leading to chronic cost overruns and litigation. Building on comparative analyses of widely used international and national standard forms of contract, empirical public works dispute case studies, and principal–agent theory of payment methods, this research develops a data-driven framework for designing incentive-compatible construction contracts. First, we code and compare more than 60 distinct risk events and 40 claims and dispute procedures across several representative standard contract regimes for large infrastructure projects. Second, we construct a dataset of approximately 80 real dispute or major change-order cases (including highway, bridge, and building projects) and classify them by triggering risk, procurement mode, and payment scheme. Third, we calibrate a structural principal–agent model of contractor effort, cost uncertainty, and risk aversion to this dataset, comparing fixed-price, unit-price, cost-plus, and target-price contracts. The calibrated model is then used to simulate how alternative combinations of risk allocation clauses and payment mechanisms affect expected owner cost, contractor risk exposure, and the probability of disputes. Simulation results are used to quantify potential performance gains from better-aligned contract design; illustrative scenarios target a 15%–25% reduction in expected total project cost, a 30%–40% reduction in formal disputes, and a 40% reduction in claims processing time when more balanced risk-sharing clauses and incentive-compatible payment schemes are adopted. The research is useful as a guideline for international contractors and public agencies involved in mega infrastructure projects.
Conventional potential field (PF) methods for unmanned aerial vehicle (UAV) navigation often fail in dynamic environments, particularly when obstacles and targets move with acceleration. In such cases, the repulsive forces generated by conventional PF approaches are insufficient, leading to oscillations, delayed responses, or even collisions. This paper presents an acceleration-based potential field (AB-PF) method designed to address these limitations by incorporating acceleration terms of both the UAV and the obstacles directly into the repulsive potential formulation. The resulting controller anticipates obstacle motion more effectively, enabling smoother avoidance and faster convergence to the desired path. The proposed method is validated experimentally using multiple UAV scenarios, including circular, infinity-shaped, and helical trajectories, with both static and accelerated obstacle drones. Performance is evaluated using key metrics: trajectory tracking error; root mean square error (RMSE), minimum and average separation distances, convergence time, and peak acceleration. Results demonstrate that AB-PF consistently outperforms conventional PF and Fuzzy-PF approaches. On average, AB-PF reduces RMSE by 25%–40%, increases separation distances by 15%–30%, and decreases convergence time by 20%–35%, while requiring up to 25% lower peak acceleration. These findings confirm that AB-PF improves both safety and dynamic smoothness in UAV motion planning, making it a promising strategy for real-time navigation in complex environments with accelerated obstacles.
This paper proposes an adaptive Overcurrent (OC) relay protection scheme that combines load forecasting and clustering-based operating state identification to dynamically adjust relay settings. A Long Short-Term Memory (LSTM) model predicts feeder current, allowing the system to proactively select appropriate relay setting groups for different load conditions. The method is validated on the IEEE 33-bus distribution network. Results show that the forecasting model achieves an Mean Absolute Percentage Error (MAPE) of 2.94 %, while the relay settings derived from predicted loads closely match those obtained from actual data. The difference in operating time is negligible between forecast-based and actual coordination and a safety protection mechanism is employed to deal with the forecast or classification errors by switching to a safe backup protection group. These results demonstrate the effectiveness of the proposed approach in maintaining reliable protection performance under changing operating conditions through SCADA, and are activated using IEC 61,850 protocols and GOOSE-based group switching mechanisms. The time-domain analysis shows a worst-case timing deviation below 41 ms, well within the 300 ms Coordination Time Interval (CTI) requirement.
Metal complexes constitute a key component of modern chemistry due to their structural diversity, adjustable electronic properties, and broad applicability in catalysis, materials science, and biomedicine. In recent years, artificial intelligence (AI) has increasingly contributed to accelerating the discovery and development of metal complexes by supporting several stages of research, including molecular design, synthesis, characterization, and functional optimization. These advances have improved the efficiency of discovery processes while also enhancing sustainability and the reliability of predictive models. This review first considers traditional coordination chemistry approaches alongside recently developed environmentally sustainable methods for the synthesis of metal complexes. Particular attention is given to ongoing challenges associated with reaction optimization, scalability, and reproducibility. In addition to experimental methodologies, machine learning methods are increasingly employed to complement conventional strategies by enabling a rapid estimate of physicochemical properties and catalytic activity. In the biomedical field, particular focus is placed on platinum- and ruthenium-based anticancer complexes. In this area, AI-assisted drug discovery strategies, computational molecular design, and predictive modeling have supported the development of next-generation metal-based therapeutics characterized by improved selectivity and reduced toxicity. The review also examines catalytic applications of metal complexes, including cross-coupling reactions, hydrogenation, transfer hydrogenation, electrocatalysis, photoredox catalysis, carbon dioxide activation, and asymmetric catalysis. Recent developments in hybrid photoelectrodes, supported catalytic systems, redox-active ligands, and rational molecular catalyst design are also considered. Furthermore, emerging AI-driven approaches for catalyst discovery, such as generative models, inverse design strategies, closed-loop automated experimentation, and high-throughput virtual screening, are presented as effective tools for accelerating catalyst development. Finally, the review discusses current limitations associated with AI-based methodologies, particularly those related to data availability, model interpretability, and generalizability. Future directions emphasize the importance of integrating explainable AI with mechanistic understanding in coordination chemistry in order to improve the reliability and interpretability of catalyst and drug discovery processes.
Phthalic acid esters (PAEs) are persistent, toxic endocrine disruptors commonly found in plastics and bottled water, necessitating sensitive analytical monitoring. This study developed a magnetic solid-phase extraction (MSPE) method combined with HPLC-DAD, using treated cypress wood (CW) as a sorbent to extract four PAEs: dimethyl phthalate (DMP), diethyl phthalate (DEP), benzyl butyl phthalate (BBP), and diisobutyl phthalate (DIBP). The wood underwent organic solvent washing, pyrolysis at various temperatures, magnetite loading, and different treatment sequences. The optimal adsorbent, MCW-400 °C-Pet.ether(1:7), was prepared by pyrolysis at 400 °C, washing with petroleum ether, and magnetization at a 1:7 magnetite-to-wood mass ratio, which enhanced porosity, exposed active sites, and improved magnetic separability. The MSPE-HPLC-DAD method was optimized for pH, adsorbent mass, and contact time, achieving low detection limits (0.027–0.098 mg L⁻¹), and high correlation coefficients (R² > 0.986). When applied to bottled water samples, the method produced recovery rates of 67.2–99.0
