JOURNAL OF ENGINEERING AND COMPUTER SCIENCES

In this paper, the effect of the inclusion of natural pozzolan (NP) on the compressive strength of cement mortar is investigated. Prediction models of compressive strength of cement mortars having NP were developed utilizing response surface methodology (RSM). In addition, multi-objective optimization has been performed by maximizing the compressive strength with natural pozzolan replacing cement (NPC) in ranges between 0 and 40% by weight of cement and natural pozzolan replacing sand (NPS) in ranges between 0 and 40% by volume of sand. To reduce the negative impact of NP on the compressive strength, silica fume (SF) was used as a partial replacement for cement by weight with ratios of 5% and 10%. The experimental data comprises the 28-day compressive strength of twenty-three mortar mixes. The specimens have cubic shapes with dimensions of 50×50×50 mm. The experimental results were used as the database for developing prediction models to evaluate the impact of doses of NP and SF on NP mortar's compressive strength. Developed models were assessed and validated to check significance and suitability of these combinations in mortar. The involvement of every parameter was analyzed using ANOVA and other statistical measures. Optimal relationships were identified between compressive strength and NPC and/or NPS ratios. The ideal replacement levels of NP for enhancing the NP mortars compressive strength were determined by numerical optimization. Constructed models’ outcomes aligned well with experimental data. The predicting NP mortars compressive strength quadratic model outperforms the linear model. Models established in this study can predict compressive strength of cement mortars using NP and SF, the results of which will be highly useful for engineering community.

A rapid increase in electric-vehicle (EV) adoption is imposing new stresses on distribution feeders, yet quantitative evidence on how coordinated renewable generation and vehicle-to-grid (V2G) operation can mitigate these impacts remains limited, especially for radial networks with long laterals and evening-peaking demand. In this study, high-resolution (15-min) time-series power-flow simulations were performed on canonical 69-bus radial system to evaluate uncontrolled EV charging, managed charging with V2G triggers, and combined scenarios with feeder-embedded photovoltaics (PV) and wind turbine generation (WTG). A stochastic EV arrival model, battery state-of-charge tracking, and a simple charger model (2.3 kW per phase, pf≈0.95) were adopted; V2G discharging was activated when local voltage fell below 0.95 pu (V_on) and deactivated when voltage recovered above 0.97 pu (V_off) or when SoC reached the lower bound, to avoid chattering. PV/WTG outputs followed realistic diurnal profiles with lagging power factors representative of inverter limits. Backward/forward-sweep load flow yielded node voltages, branch currents, and total feeder losses across a 24-h horizon, and results were synthesized via heat maps, box plots, and daily power balance curves. It was found that coordinated V2G with feeder-level PV/WTG materially reduces peak source current, alleviates trunk and mid-feeder overloading, and lowers total active power losses, while lifting minimum voltages toward acceptable limits during critical evening windows. Benefits were strongest when renewable production temporally overlapped charging demand and where DERs were electrically proximate to stressed branches; residual constraints persisted during late-night peaks with weak renewable support. Overall, the results indicate that pragmatic V2G thresholds, modest feeder-sited renewables, and basic charging management can substantially improve hosting capacity without immediate network reinforcement.

There are several ways to forecast electric vehicle sales to help businesses make better decisions. One strategy is to use prescriptive and predictive analysis techniques to examine consumer preferences and buying behavior regarding electric vehicles. Additionally, models for predicting electric vehicle sales can be developed using logistic regression and decision trees. These models let businesses make well-informed judgments about their sales by offering patterns regarding customer mining, classification accuracy, and prediction effects. In this study, we collected survey data from approximately 452 respondents in Qassim State through an online questionnaire to analyze factors influencing electric vehicle (EV) adoption. The data consists of 23 conditional attributes that were evaluated during the analysis. This study suggests evaluating the features of the customer that may influence their decision to purchase electric vehicles. The clustering algorithm was implemented to segment customers based on these features. The current study's findings demonstrated the effectiveness of the model created using the K-means clustering algorithm derived from data mining in predicting the purchase of electric vehicles. The accuracy of the correctly classified instances was evaluated using three distinct classification techniques: BayesNet-D, NaiveBayes, and J48. According to the evaluation results from these algorithms, the overall accuracies in terms of correctly identified cases were 90.4867%, 88.0581%, and 91.8142%, respectively. Through a series of tests on real electric vehicle datasets gathered from individual users, the effectiveness of these classifier-based rule-based models is evaluated. The overall experimental findings and discussions can assist businesses of electric vehicle companies in making decisions on the development of intelligent systems for electric vehicle users. The data-mining algorithm's extraction of information in this   study improved our comprehension of the customer base's composition regarding purchasing electric vehicles and the company's assessment of the appropriate course of action for offering guidance, training, and expertise.

Ever since the emergence of Cryptocurrency in 2009, it has been thought of as the top alternative to fiat currencies. However, an increase in fraud has resulted in a significant challenge to its adoption and trust. The increase in the number of frauds is a critical concern for the industry, regulators, and users. New types of cryptocurrency fraud continue to emerge. However, the prevention strategies remain fragmented, with low consumer awareness. The proposed framework successfully mapped scam types to vulnerabilities and explained more than 90% of real-world scams using layered prevention strategies. While it highlights the gaps in the evolution of prevention and the need for layered awareness and controls, this study examines the effectiveness of fraud prevention strategies and identifies scams that exploit vulnerabilities across various layers. Starting with a chronological analysis of crypto frauds from 2009 to 2025 using the literature review of peer-reviewed papers, the study created a three-layer fraud framework comprising Infrastructure, Application, and UI. Using rule-based classification logic, such as keyword detection and conditional logic, we validated the framework against a dataset (9000 entries).