Engineering Science Letter

Development of a Non-Invasive Method for Monitoring of HV Circuit Breaker Switching Time

2026-08-24T02:10:40+00:00

Modern HV circuit breakers may be vulnerable to catastrophic failure as they are designed for higher stress than the earlier CB designs with multiple interrupters per pole. With the advantage of controlled switching, improved performance is obtained for dielectrically well-designed interrupters, which achieve a re-ignition-free window during opening of the CB, in turn minimizing the risk of nozzle puncture. On occasion, asset owners may wish to check whether the CB is performing satisfactorily and whether the controllers are providing reliable and repeatable stress control. Monitoring of voltage waveforms during switching using well-established offline diagnostic methods will provide information about small re-ignitions and re-strikes. However, waveform measurement at moderately high signal frequency would require a CB outage to connect specialized equipment. A non-invasive measurement technique devising re-striking voltage sensors has been developed by the authors to measure high-frequency voltage waveforms occurring during switching operations without the need for an outage. Results of tests performed in the laboratory and 245 kV substation illustrating the capability of this new method to detect re-ignitions are presented in this paper. The proposed diagnostic approach relies on parameters such as operating times, pre-strike characteristics, and restrike detection. Transient electromagnetic emissions have been identified as a promising means to evaluate the above parameters non-intrusively.

Subject Area Classification of Journal Articles Based on Metadata Using Bag of Words and Naïve Bayes

2026-08-24T02:10:42+00:00

The rapid growth of scientific publications poses challenges in grouping journal articles based on subject area, especially when using metadata such as titles, abstracts, and keywords. However, differences in feature representation and classification algorithms often result in varying performance, requiring comparative studies to determine the optimal model combination. This study compares four combinations of subject area classification models, namely TF-IDF + Naïve Bayes, TF-IDF + Support Vector Machine, Bag-of-Words + Support Vector Machine, and Bag-of-Words + Naïve Bayes. The research process included text preprocessing, feature extraction, and testing using an 80% training and 20% testing data split scheme in five scenarios. The evaluation was performed using confusion matrices, accuracy, precision, recall, and F1-score. The experimental results showed variations in performance between models, with an average F1-score of 0.8103 for TF-IDF + Naïve Bayes, 0.8494 for TF-IDF + Support Vector Machine, 0.8297 for Bag-of-Words + Support Vector Machine, and 0.8335 for Bag-of-Words + Naïve Bayes as the best performance. These findings indicate that a word frequency-based approach combined with Naïve Bayes is effective for classifying journal article subject areas based on metadata, although challenges remain in subject areas with semantic proximity.

Ensemble Machine Learning Models for Accurate Prediction of the Carbon Footprint of SCM-Blended Concrete

2026-08-24T02:10:37+00:00

Concrete contributes approximately 8% of global CO₂ emissions. The incorporation of Supplementary Cementitious Materials (SCMs) as partial cement replacements is widely recognized as an effective strategy to reduce the carbon footprint of concrete. However, accurately quantifying the relationship between mix composition and carbon emissions remains challenging. This study develops a machine learning model to predict the carbon footprint of SCM-based concrete using material composition data. A global dataset comprising 1,456 mix designs collected from 136 publications across 27 countries was compiled, resulting in 1,294 valid samples after preprocessing. Four regression algorithms were evaluated: Support Vector Regression (SVR), Random Forest Regression (RFR), Decision Tree Regression (DTR), and Gradient Boosting Regression (GBR), with hyperparameter tuning using 5-fold cross-validation. All models achieved high predictive accuracy (R² > 0.998), with GBR demonstrating the best performance (R² = 0.9996; RMSE = 1.7452 kg CO₂/m³; MAE = 1.2779 kg CO₂/m³). Feature importance analysis identified cement as the dominant contributor (>99.8%) to emissions. Sensitivity analysis confirmed a strong linear relationship between cement content and CO₂ emissions (~0.82 kg CO₂ per kg cement). These findings support emission-reduction strategies in sustainable concrete design.

Scrum-Driven Task Management for University IT Governance: A Multi-Role Usability Evaluation

2026-08-24T02:10:35+00:00

Task management is an essential aspect of organizational operations that involves coordinating various work activities and facilitating collaboration among team members. However, in many organizations, task management processes are still carried out through informal communication channels such as direct conversations, email, or instant messaging applications, which complicate task monitoring and documentation. This study aims to develop a web-based task management system that supports more structured task management within the Directorate of Information and Communication Technology at Universitas Sebelas Maret. The system was developed through four iterative Scrum sprints over 10 weeks. Functional testing employed the Black-Box Testing method, and usability was assessed using the System Usability Scale (SUS) with 20 purposively selected respondents across four user roles: administrators, sub-directorate heads, section heads, and staff members. The results indicate that all system functions operated successfully, and the system achieved an average SUS score of 75.25 — categorized as Good and exceeding the standard usability benchmark of 68, with scores ranging from 71.0 (section heads) to 81.0 (administrators). These findings suggest that Agile Scrum-based development produces task management systems that are adaptive to organizational needs and well-accepted across hierarchical user roles, with implications for IT governance practice in higher education institutions.

Trade-Off Analysis of Moving Average Filter in Light Sensors

2026-08-24T02:10:33+00:00

Light intensity measurements based on sensors often experience signal quality degradation due to noise interference. Although filtering methods are commonly used to reduce noise, improvements in signal stability are frequently accompanied by changes in dynamic response, resulting in a trade-off between accuracy and response speed that has not been extensively analyzed. This study aims to quantitatively evaluate the effect of Moving Average Filter (MAF) parameters on this trade-off in light sensor systems. The proposed method employs a first-order system simulation with a step input signal contaminated by White Gaussian Noise, which is subsequently processed using the MAF with various window sizes ( = 3, 5, 10, 20, and 40). The evaluation is conducted using Root Mean Square Error (RMSE) and rise time as indicators of estimation accuracy and response speed, respectively. The results demonstrate that increasing the window size significantly reduces the RMSE, decreasing from 0.0156 to 0.0072 under step-up conditions and from 0.0132 to 0.0066 under step-down conditions. Optimal performance is observed within the range of = 10–20. However, this improvement in accuracy is accompanied by an increase in rise time, from 0.1027 s to 0.1175 s for step-up conditions and from 0.1036 s to 0.1191 s for step-down conditions, indicating a slower dynamic response. These findings confirm the existence of a trade-off between signal accuracy and response speed. Therefore, this study provides a quantitative basis for determining optimal filter parameters by considering the balance between measurement accuracy and system responsiveness in light intensity sensing applications.

Taguchi-Based Optimization of Mix Design and Mechanical Properties of Epoxy Polymer Concrete with Recycled Coarse Aggregate

2026-08-24T02:10:29+00:00

This study examines the mix design and mechanical properties of epoxy polymer concrete utilizing fully recycled coarse aggregate and fly ash as a mineral filler through a Taguchi-based optimization framework. An L9 orthogonal array including four control factors and three levels was employed to incorporate epoxy resin content, resin–hardener ratio, coarse-to-fine aggregate ratio, and class C fly ash proportion. All mixtures were designed using the absolute-volume method so that the total volume of the constituent materials exactly matched the target specimen volume. The research test object was a 50 × 50 × 50 mm cube that follows the experimental design in method B of the ASTM C 579-96 standard for polymer concrete. Cube specimens tested at seven days for compressive strength and static modulus of elasticity based on the stress–strain response. The compressive strengths varied from around 11.14 to 57.20 MPa, whereas the average static modulus ranged from about 0.58 GPa to nearly 2.95 GPa. Taguchi analysis and ANOVAs conducted on both mean strength and S/N ratios consistently revealed epoxy content, resin–hardener ratio, and coarse-to-fine aggregate ratio as the dominant factors, while fly ash served as a secondary modifier. The optimal combination comprises 25% polymer matrix of the total specimen volume, a resin–hardener ratio of 1.5:1, a balanced coarse-to-fine aggregate ratio of 1:1, and an quantity of fly ash around 30% of the total volume of fine aggregate. According to Taguchi analysis, this optimal combination leads to a predicted compressive strength of approximately 52.89 to 61.23 MPa. An independent confirmation mixture prepared at this optimal combination achieved an average strength of about 57.46 MPa, aligning well with Taguchi predictions. The linear relationship between compressive strength and static modulus of elasticity shows a positive correlation. This linear relationship enables a straightforward empirical formula to determine the stiffness of epoxy polymer concrete with recycled coarse material, predicated on its compressive strength.

Multi-Modal Deep Learning Approach for Waste Management: Integrating Image Classification and Text Mining for Environmental Awareness

2026-08-24T02:10:31+00:00

Environmental degradation caused by inefficient waste management remains a major global challenge, largely due to the limitations of conventional systems that rely on manual waste sorting and limited utilization of heterogeneous data sources. This study proposes a novel multi-modal deep learning framework that integrates visual and textual information to enhance waste classification performance while simultaneously providing insights into environmental awareness. The proposed framework combines convolutional neural networks (CNNs) for waste image classification and a recurrent neural network with long short-term memory (LSTM) architecture for text analysis. Visual and textual feature representations are integrated through a feature-level fusion strategy using vector concatenation before final classification. The image dataset consists of six waste categories, cardboard, glass, metal, paper, plastic, and trash, while the textual dataset contains waste management descriptions, community feedback, and environmental discourse collected from public and field sources. Environmental awareness was assessed through text mining by identifying dominant themes related to recycling practices, waste sorting behavior, environmental responsibility, and public concern regarding pollution and sustainability issues. Experimental results demonstrate that the proposed multimodal framework achieves an accuracy of 88.9% and an F1-score of 0.89, outperforming image-only and text-only models with accuracies of 78.4% and 81.2%, respectively. This corresponds to absolute performance improvements of 10.5% over the image-based model and 7.7% over the text-based model, while reducing the classification error rate by 40.96%. Furthermore, the multimodal model exhibits superior robustness under degraded data conditions, with only a 4.7% reduction in accuracy compared to larger performance declines observed in unimodal approaches. The main contribution of this study lies in the integration of waste image recognition and environmental-awareness extraction within a unified multimodal learning framework, enabling not only accurate waste categorization but also the generation of behavioral and sustainability-related insights that support more intelligent and sustainable waste management systems.

Ride-Hailing Quality Gaps and Improvement Priorities: SERVQUAL-Kano Study in Mid-Sized Indonesian City

2026-08-24T02:10:27+00:00

Digitalization has accelerated ride-hailing growth by offering urban commuters convenience and flexibility. In Indonesia, rising usage has intensified competition among platforms such as Grab and its rivals. Despite its extensive reach, Grab continues to face negative perceptions concerning pricing, inconsistent service, and suboptimal user experiences, which erode customer satisfaction and loyalty. Malang City was selected for its high mobility and rapidly growing user base. As a mid-size city with moderate density and growing transport demand, Malang is suitable for studying ride-hailing beyond megacities. This study integrates SERVQUAL and the Kano Model to identify, classify, and prioritize service attributes influencing GrabBike user satisfaction. A purposive sample of 385 respondents completed an online survey (Google Forms) between December 2025 and January 2026. The overall quality ratio (Q) was 0.85, and Tangibles was the lowest performing dimension. RL4 (fare suitability) recorded the largest gap (-1.65), a critical priority. Kano classification yielded 7 Must-Be, 3 One-Dimensional, 3 Attractive, 1 Indifferent, and 1 Reverse attributes. This integrated approach constitutes a novel, data-driven framework for prioritizing service enhancements. The findings provide guidance to remedy dissatisfiers and invest in delight-enhancing attributes, while establishing local quality benchmarks for policymakers.

Optimization of Concrete Mix Composition Containing Fly Ash and Slag Using a Machine Learning Algorithm for Compressive Strength Prediction

2026-08-24T02:10:25+00:00

The modern construction industry faces significant challenges in developing sustainable concrete materials while maintaining structural quality requirements. Conventional trial-and-error methods for concrete mix design are time-consuming, costly, and often result in high variability in concrete quality. This study presents an integrated framework that combines machine learning techniques for concrete compressive strength prediction with genetic algorithm optimization to determine optimal mix compositions containing fly ash and blast furnace slag. Two predictive models were developed using the UCI Machine Learning Repository concrete dataset comprising 1,030 samples: Artificial Neural Network (ANN) Ensemble and Support Vector Regression (SVR). The ANN model demonstrated superior performance, achieving R² values ranging from 0.7475 to 0.8372, RMSE values between 6.11 and 7.94 MPa, and classification accuracy of 86.92% for concrete quality categorization across three classes (Class I: <20 MPa, Class II: 20-35 MPa, Class III: >35 MPa). In comparison, the SVR model achieved competitive but slightly lower performance with R² values of 0.7491-0.8378 and classification accuracy of 80.37%. The stability and generalizability of both models were confirmed through five-fold cross-validation. Subsequently, genetic algorithm optimization was applied to determine optimal mix compositions for each quality class while ensuring compliance with Indonesian National Standards (SNI 2847:2019, SNI 2461:2011, and SNI 8297:2016). The optimization process successfully produced concrete mix designs that achieved target compressive strengths of 14.95 MPa for Class I, 27.48 MPa for Class II, and 59.99 MPa for Class III. This framework demonstrates significant potential for developing sustainable concrete with optimal performance while meeting applicable technical standards, thereby contributing to a reduced carbon footprint in the construction industry through strategic utilization of supplementary cementitious materials.

Advantages and Challenges of VR in Engineering Education

2026-08-24T02:10:22+00:00

Virtual reality (VR) technology has the potential to revolutionise engineering education by providing students with immersive and interactive learning experiences not possible with conventional methods. Virtual reality can be used to simulate real-world environments and equipment, teach engineering concepts and principles, and prepare students for industry practices. However, the use of Virtual Reality in engineering education presents challenges, such as the limited availability and expense of VR equipment, the need for technical support and training for faculty and students, and the requirement for the creation and maintenance of Virtual Reality content. This issue was developed based on a narrative review and a literature review. The narrative review was conducted after the author studied and gathered the latest information. To effectively implement VR in engineering education, it is necessary to weigh the advantages and disadvantages of the technology and allocate resources accordingly. The benefit of VR in engineering education is increased accessibility to simulations and hands-on learning experiences, increased student engagement and motivation, the ability to visualize and manipulate 3D objects and concepts in a realistic environment, and the chance to safely test and experiment with equipment and processes. Future use of augmented reality (AR) and mixed reality (MR) technology in engineering education might provide students with even more immersive and interactive learning experiences. The use of Virtual Reality in engineering education has the potential to transform how engineering is taught and learnt by offering students more interesting and interactive learning experiences that better prepare them for success in the field.

A Systematic Approach to Reducing Compressor Downtime by 95.3% Within Six Months Toward Industry 4.0 Implementation

2026-08-24T02:10:18+00:00

Workshop is one of the sections in the company that is tasked with supporting the productivity process, from Welding fabrication work, motor rewinding, machine tool work, to punch procurement (Mold). Supported by various machines used to support the production process, both for punch repair and fabrication. For this reason, the workshop always tries to provide the best, namely by creating improvements to speed up the process of procuring punches, both lower and upper, as well as master motifs and plain, while the improvements made are making improvements in the Compressor area. Thus, this compressor machine is a vital supporter of the machine's running process and product water usage services to achieve the target results of punch repair in the workshop, the purpose of the research is to add an automatic remote control at the nearest location & can be operated with Wi-Fi, create an auto evaporator cleaning program, install a compressor alarm notification, create an auto drain program for the compressor tube, create a safe compressor room. By implementing several improvements, the results were achieved by reducing downtime by 95.3% to 96.7%. Downtime decreased from 2784 minutes per 3 months to 92 minutes per 3 months. The potential profit was Rp 159.962.422 per year. Therefore, the research results can be applied to all workshop and production machines.

Sustainable Inventory Optimization: A Multi-Item EOQ Model with All-Unit Discount, Capacity Constraint, and Carbon Emission

2026-08-24T02:10:14+00:00

Inventory is one of the operational components that contributes to carbon emissions through warehouse electricity consumption; however, this aspect has not yet been integrated into conventional inventory models, despite growing attention to green supply chain management. This study develops a multi-item Economic Order Quantity model that considers all-unit discounts, warehouse capacity constraints, and carbon emissions as cost components and examines the effect of carbon emission variables on the mathematical model structure. The development was carried out by adding a carbon cost variable, calculated as the product of average warehouse energy consumption, the carbon emission factor, and the carbon tax, to the storage cost equation and the formula for the optimal reorder interval. Solutions were obtained through an iterative approach to determine a valid discount price interval. Proportional adjustments were then applied to satisfy the warehouse capacity constraint. A numerical example involving three product items yielded an optimal cycle time t* = 0.144 periods, a reorder interval after capacity adjustment G = 0.108 periods, and total inventory costs of IDR 2,086,099,971 per planning period. The addition of carbon emission variables increases the holding cost component and reduces the optimal reorder interval compared to the reference model, resulting in a total inventory cost IDR 1,488,996 higher than the reference model. Sensitivity analysis results indicate that warehouse capacity has the greatest influence on total costs compared to carbon emission parameters. This study contributes to the integration of environmental aspects into a deterministic multi-item inventory model, thereby providing a more comprehensive decision-making framework for retail managers.

Investigation of Temperature Field Evolution in Pangasius Fillets during Freezing

2026-08-24T02:10:16+00:00

This study investigates the spatial and temporal evolution of temperature within Pangasius fillets during freezing using a finite element model of two-dimensional nonlinear transient heat conduction with phase change and convective boundary conditions. The model was implemented in ANSYS for a representative fillet cross-section and simulated under an air temperature of −40 °C and air velocity of 10 m/s. Temperature histories at 25 representative nodes were analyzed to characterize local freezing behavior and identify differences between surface and interior regions. The results reveal three characteristic stages: rapid precooling toward 0 °C, a phase-change period dominated by latent heat release, and subsequent sensible cooling after freezing is completed. Surface and corner regions cooled substantially faster than interior locations because of stronger convective heat transfer, while the geometric center exhibited the longest freezing delay. The maximum thermal delay between the corner and center occurred near −3 °C, reaching approximately 800 s, and decreased as the temperature approached the fully frozen state. Below approximately −18 to −20 °C, temperature-time curves exhibited increasingly similar slopes, indicating completion of phase change. These findings demonstrate that temperature-field analysis can support reliable freezing-time prediction and provide a quantitative basis for optimizing operating conditions, energy efficiency, and product quality.

EV Battery Recycling Adoption: Insights from Dealers and Workshops in Yogyakarta, Indonesia

2026-08-24T02:10:20+00:00

The rapid growth of electric vehicles in Indonesia has intensified battery waste management challenges, while the participation of informal repair businesses in battery recycling services remains limited. Existing studies have largely focused on consumers' recycling behavior, providing little understanding of how informal actors develop the intention to provide recycling services. This study develops a TPB-based behavioral framework by incorporating environmental, social, economic, and technological dimensions as contextual determinants of intention. Survey data from informal repair businesses in Yogyakarta were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The results show that the environmental and technological dimensions significantly influence intention, whereas the social and economic dimensions do not. Intention, in turn, significantly increases willingness to provide battery recycling services. These findings indicate that the underlying behavioral mechanisms of recycling service providers are driven primarily by environmental awareness and technological readiness. This study contributes by elucidating the mechanisms that shape the intentions of recycling service providers, an underexplored stakeholder group in developing countries, and offers practical suggestions for designing interventions that strengthen battery recycling services through environmental education, the provision of appropriate technology, and clear institutional support to encourage participation.

The Effect of Work Stress on the Driving Behavior of Online Motorcycle Taxi Drivers

2026-08-24T02:10:11+00:00

Working as an online motorcycle taxi driver is a job many people do nowadays. The development of digital technology has made working as an online motorcycle taxi driver a new source of hope for job seekers. Various digital services, such as these, are widely available, making this job increasingly popular. Amidst busy activities, online services greatly facilitate the work of many people. Many online motorcycle taxi drivers currently offer delivery and pickup services. With more and more people doing this job, pressure has ultimately built up on each driver. Due to heavy workloads, some drivers feel stressed. The purpose of this study is to determine whether there is a relationship between workload, work environment, and traffic violations on driving Behavior, moderated by the variable of work stress. Data collection was conducted by distributing questionnaires directly for each variable. The respondents who participated in the study were 40 online motorcycle taxi drivers. The results show that there is an influence between workload, work environment, and traffic violations on driving behavior. However, after conducting a moderation test, it turns out that work stress does not moderate the influence of workload on driving behavior.