ABUAD Journal of Engineering Research and Development (AJERD)

Characterization and Comminution Energy Determination of Anka Manganese Ore for Steelmaking Applications

Yemisi Elizabeth Gbadamosi — 2025-09-09

The global demand for manganese continues to rise due to its critical role in steelmaking and emerging battery technologies. This study investigates the physicochemical properties and comminution energy requirements of manganese ore from the Anka deposit in Zamfara State, Nigeria. A comprehensive characterization was conducted using Energy Dispersive X-Ray Fluorescence (ED-XRFS), X-Ray Diffraction (XRD), and Scanning Electron Microscopy with Energy Dispersive Spectroscopy (SEM-EDS). Particle size analysis was carried out on the as-received sample, followed by a grindability test. The chemical analysis confirmed a high MnO content (52.112%), while mineralogical assessments identified pyrolusite and jacobsite as dominant phases. SEM-EDS analysis revealed the interlocking nature of manganese, iron, silicon, and aluminium within the ore matrix. Particle size analysis identified the optimal liberation size at -125+90 µm, assaying 53.973% MnO. Grindability tests using quartz and iron as reference ores yielded work index values of 11.34 kWh/ton and 14.62 kWh/ton, respectively. The average energy expended during grinding was 5.06 kWh. These findings provide a basis for designing energy-efficient beneficiation processes for Anka manganese ore, contributing to sustainable resource utilization and industrial cost reduction.

Evaluation of Optimized Carbonized Wheat Husk-Reinforced AA6061 Composite for Automotive Components Applications

Olawale Ibrahim Kolapo — 2025-10-02

In the pursuit of sustainable and lightweight materials for automotive applications, metal matrix composites (MMCs) have emerged as promising candidates due to their superior strength-to-weight ratios, enhanced wear resistance, and tailored mechanical properties. Aluminum-based composites, particularly those with AA6061 as the matrix, are widely recognized for their excellent corrosion resistance, weldability, and mechanical performance. However, there remains a need to improve the environmental sustainability, mechanical properties and lightweight properties of these materials through the incorporation of eco-friendly, sustainable, light and low-cost reinforcements. In this work, the reinforcement particulate, carbonized wheat husk (CWH) was gotten after pulverizing wheat husks to increase the surface area and charging it into a muffle furnace, subjected to a temperature of 900 ⁰C for 3 hours. Thereafter, AA6061 reinforced composites (AA6061-CWH) were produced using the stir casting method, optimized through the Taguchi's L9 orthogonal array. The composite developed at optimum parameters was then selected and compared with selected automotive components. The optimized AA6061-CWH composite offers a well-balanced mechanical profile. It delivers decent tensile strength, exceptional hardness, good impact resistance, and a lower density, making it an appealing material choice for a broad spectrum of automotive applications. Its application could support the automotive industry’s ongoing pursuit of improved performance, efficiency, and sustainability through the use of affordable and eco-friendly materials.

Optimising Logistics Operations using Geospatial Algorithms with Adaptive Route Evolution

Quadri Ayotunde Akinjole — 2025-10-02

Rapid urbanisation in cities like Lagos, Nigeria has impaired traffic congestion and route-planning, causing delivery delays stemming from logistic inefficiencies leading to operational costs; to overcome these challenges, an Adaptive Route Evolution (ARE) model that integrates A* for initial shortest-path generation, a Genetic Algorithm (GA) for offline route refinement, and Reinforcement Learning (RL) for real-time adjustments based on live traffic data is proposed. In simulations of two vehicles traversing standard delivery corridors with geospatial road-network data and real-time traffic APIs, Vehicle 1’s delivery time decreased from 78 min to 70 min after A* (–10.3%), 65 min after GA (–17%), and 61 min after RL (–21.8%), while Vehicle 2’s time fell from 120 min to 90 min after A* (–25%), 83 min after GA (–30.8%), and 79 min after RL (–34.2%), yielding an average 27.5% reduction in delivery time and an estimated 22% drop in fuel and driver costs. ARE provided a complete end-to-end optimisation framework by combining RL’s adaptability with A*’s shortest-path planning and GA’s sequence optimisation. It demonstrated scalability and effectiveness in dynamically optimising urban logistics, reducing both transit delays and carbon emissions, and enabling a more sustainable and cost-efficient transportation system in line with Sustainable Development Goals (SDGs) 7, 11, and 13.

Application of Artificial Intelligence in Chatbot and Social Media Community for a Mental Health Monitoring Web-Application

David Oluwaseun Kajewole — 2025-10-02

Mental health is a major challenge for modern society affecting people of working age and their families, jobs, and communities. To provide continuous emotional support, personalized assistance, and remote mental health tracking, chatbots and social media platforms are increasingly using Artificial Intelligence (AI). There is little or no previous literature that has documented a fully functional mental health monitoring web-app that uses AI to provide all these features: chatbot, e-community, video conferencing, and email automation. Chatbots are software systems that offer various interactive online services, including people with mental health needs and have been successfully integrated into the field of mental healthcare. Similarly, social media can help monitor the mental health situation by extracting information from posts for sentiment analysis as people often post their feelings on social media, so analysing these posts can reveal their mood, emotion, cognition, or mental state. This paper solves the problem sentiment by using the MERN stack to develop an interactive mental health monitoring application which features: social media community, chatbot named ‘Dave the HappyBot’, Email automation and video conferencing. The results showed that among the 13 users of the web-app from a performance indices class of A, B, C, D, and E, majority indicated that they felt ‘happier’ and ‘somewhat mentally relieved’ after interacting with the Chatbot. The A class recorded 18% responses, B class obtained 31%, the C class got 19%, the D class achieved 26%, and the E class gained 6%. The users were open to recommending the App to their peer and meeting their psychologist for constant mental health monitoring sessions. This paper can be further improved documenting the developments of web-app with more users who possess higher knowledge of software development and have a history of mental health struggle. Higher utility of the MERN stack would also be an improvement.

Modelling and Simulation of Spray Pyrolytic Graphene and Graphite Coatings on Polymeric Substrates for enhanced Oil and Gas Pipeline Repair Applications

Kenechi Ijeoma Ojemeni — 2025-10-02

The oil and gas industry faces significant challenges in pipeline corrosion management, with high corrosion resistance and electrical conductivity for effective cathodic protection (CP). This study focuses on the mathematical modelling of electrical conductivity in nanoparticle-sized graphene, graphite, and graphene-graphite coated polyvinyl chloride (PVC) and glass reinforced polymer (GRP) substrates, prepared via low-temperature spray pyrolysis (50–60 °C). Single- and double-layer coatings with graphene-graphite blend ratios of 1:0.5 and 1:1 were applied on 50 × 10 × 2 mm substrates to obtain coating thickness of 0 µm, 56.1 µm, 77.2 µm, 80.6 µm, 80.8 µm, 92.6 µm, and 97.9 µm respectively for the PVC samples while the GRP samples have coating thickness of 0 µm, 110.3 µm, 114.0 µm, 109.0 µm, 115.7 µm, 117.3 µm, 124.7 µm. Electrical conductivity was measured using an LCR meter, and polynomial models were developed to correlate conductivity with coating thickness. Cathodic protection simulations assessed the performance of coated composites in a 15 km pipeline, highlighting the impact of non-conductive sections and the efficacy of bypass wire designs. Results showed that single-layer graphene-coated GRP achieved the highest conductivity (1.8 × 10⁻⁶ S/m), while double-layer hybrid coatings (1-0.5D) offered optimal durability. The mathematical models accurately predicted conductivity trends, with GRP exhibiting superior performance compared to PVC due to better graphene integration. CP modelling revealed that non-conductive GRP sections cause localized underprotection, mitigated by optimized bypass designs. These findings demonstrate that graphene-graphite-coated GRP composites, supported by predictive conductivity models and CP simulations, are promising alternatives to steel for pipeline repairs, enhancing corrosion control and longevity in oil and gas applications.

Process-Composition Design of Hypoeutectic Aluminum-Silicon Alloy for High Performance Wear Resistance Application

Onyemaechi Melford Chima — 2025-10-12

This study investigates the process-composition design of hypoeutectic aluminium-silicon alloys aimed at improving the wear behaviour of the alloy for tribological application. Hypoeutectic Al-Si alloys with percentage composition of silicon ranging from 3 – 7.5% were cast at varying pouring temperatures of 700, 750 and 800⁰C. The impact of the process-composition parameters on wear rate and the microstructures of the alloy were determined using tribor testing apparatus and scanning electron microscopy/energy dispersive spectroscopy (SEM/EDS) respectively. The results obtained show that increasing silicon content from 3 – 7.5% significantly improved the wear rate of the alloy from 0.0360 – 0.0120 mg/m with optimum pouring temperature value at 700⁰C. The SEM micrographs indicate that higher percentage composition of silicon yielded the formation of more numbers of primary silicon phases with intermetallic phases that reduced material loss while optimum pouring temperature influenced solidification rate leading to a refined uniform homogeneous phases distributed in the microstructures. It was concluded that process parameter optimization carefully tailored through combination of silicon percentage composition and pouring temperature enhances the wear performance of Al-Si alloys for engineering applications in wear-critical environment.

Intelligent Web App for Flash Flood Prediction in Nigeria’s Coastal Regions

Ijegwa David Acheme — 2025-10-19

Many coastal cities in Nigeria and around the world are faced with the menace of flash floods and many times, it temporarily disrupts the socio-economic activities of residents. This project aims to address this challenge by developing a smart web application using machine learning to intelligently predict flash flood occurrence and offer recommendations. In order to achieve this, the random forest machine learning algorithm is utilized to analyze environmental data such as rainfall, river levels, and soil moisture necessary for the prediction of a flood which are captured in real-time using the OpenMeteo API. The machine learning model is then trained using these environmental variables and integrated into a web application for easy user interaction. The frontend of the web application is built with TypeScript, React.js, and Tree.js, providing an interactive and user-friendly interface for visualizing flood predictions, while the backend is built using MongoDB and python (FLASK framework). The goal is to offer accurate, real-time flood forecasts to help individuals prepare and respond effectively. This project demonstrates the integration of data science and web development to create a practical tool for disaster risk management. The random forest model was evaluated using the standard metrics for evaluating machine learning models and showed the following results; Accuracy of 96%, precision of 75% and recall of 91%. In addition, the model, showed a Real-time latency of less that one second, which is indicative of a fast response to changing environmental data input. Since flood conditions can change rapidly, this low real-time latency shows that the web is able to respond quickly to new sensor or satellite data input.

Detection of Bank Customer Churn Using Neural Network and Voting Classifier Ensembles

Abdulrahman Olajide — 2025-10-25

Customer churn is the loss of business clients to a competitor. Since keeping current clients is more economical than finding new ones, customer retention measures such as churn detection are now essential aspects of modern banking strategy. However, many existing studies rely heavily on conventional machine learning approaches such as Support Vector Machines, Logistic Regression , Random Forest, etc., often neglecting the deeper learning capabilities of neural networks. Also, the repeated use of the same small dataset by the banking studies may limit the improvement of the models’ generalisation. To address these gaps, this study presents a method that integrates deep learning for customer churn detection and a soft and hard voting classifier ensemble embedded with the best performing models over the years for results comparison, supported by a synthetic data augmentation method for model improvement. The study utilised a secondary banking churn dataset from Kaggle, which contained 10,000 unique customer records. To address the dataset limitations, a Conditional Tabular Generative Adversarial Network (CTGAN) model was used to generate an additional 10,000 records, expanding the dataset used for the study to 20,000 rows. Data preprocessing steps were done before training, including oversampling using Synthetic Minority Oversampling Technique (SMOTE). Model development and analysis processes were implemented using Python programming language with prominent libraries and frameworks on Google Colab. In this study, a Feedforward neural network and a soft and hard voting classifier were developed. The voting classifier ensembles integrated three prominent classifiers: Random Forest, XGBoost, and Logistic Regression. The performances were evaluated using Accuracy, F1 Score, and Area Under ROC Curve as metrics. Results show that while the Feedforward Neural Network achieved strong predictive performance with an accuracy of 88.23%, an F1 Score of 87.83% and an AUC of 94.73%, the ensemble approaches performed slightly better as the soft voting classifier delivered the best results, obtaining an accuracy of 89.46%, F1 Score of 88.92% and AUC of 95.40% showing the advantage of combining multiple models to leverage complementary strengths. After comparison with past studies, the proposed models did not surpass the very best outcomes. However, they remain highly competitive, achieving performance levels that are on par with or exceed many earlier works. The contribution of this work is to show how synthetic data augmentation, enhanced preprocessing, deep learning techniques, and machine learning ensembles can improve churn detection in banking studies. Banking institutions can utilise the results from this study to accurately detect churn, supporting proactive customer retention strategies, targeted marketing, and personalised financial services, thereby reducing revenue losses.

Effect of Locust Bean (Parkia Biglobosa) Pod Ash and Portland cement on Strength and Durability Properties of Compressed Earth Bricks

Justin Tapgun — 2025-10-25

The need for locally produced, durable and robust, readily available, inexpensive and environmentally friendly building materials has led to persistent bottlenecks in sustainable housing delivery. "The construction industry is exploring exciting new materials that are eco-friendly and perfect for construction projects." The major problem associated with compressed earth bricks is the high rate of water absorption and lack of durability properties because most soil in its natural condition needs more strength, dimensional stability, and durability, which are required for building materials. The effect of high rate of water absorption and other strength and durability issues make bricks to be soluble in water and limits its use and performance of the bricks. This experimental study assesses the effect of locust bean (Parkia biglobosa) pod ash and Portland cement on compressed earth bricks' strength and durability properties. Compressed earth bricks were tested for density, compressive strength, permeability, water absorption, shrinkage, sorptivity and abrasion resistance. The maximum compressive strength was achieved at 10%C:10%LBPA stabilization with a strength of 2.52 N/mm² and 2.80 N/mm²at 28 and 56 days, which shows a 50% and 53.21% increase in strength over the control brick samples, respectively. Bricks produced with cement and locust bean pod ash were less permeable and had high resistance to abrasion, less shrinkage, less porous and less sorptivity than 0% stabilization. In conclusion, cement and locust bean pod ash are good stabilizing agents in compressed earth bricks. The use of Portland cement and locust bean pod ash as a stabilizing material seems to be a feasible solution not only to the problem associated with compressed earth bricks but also helps in the adoption of Indigenous waste material of locust bean pod in the production of bricks which will help reduce the environmental problem. Therefore, this research recommends using cement and locust bean pod ash at 10%C:10%LBPA in compressed earth bricks, leading to robust, stabilized and durable bricks.

Lightweight CNN Architectures for Fault Diagnosis of Power Generator sets: A Comparative Study of MobileNet and AlexNet

Ekerette Bernard Ibanga — 2025-10-25

The use of power generator sets (3.5kVA – 5.5kVA) for domestic and commercial backup supply has become a mainstay in Nigeria due to the unstable electricity from the grid. To keep these backup supplies running, traditional diagnostic approaches that are reliant on manual inspections and physical measurements have been adopted. These are often time-consuming, reactive, and unsuitable for real-time monitoring. To address these challenges, a machine learning approach is presented by performing a comparative analysis of MobileNet and AlexNet convolutional neural networks for automated audio-based fault diagnosis in 5kVA generators. Fault signatures are obtained from acoustic data recorded from 25 generator units under five operational states—Normal, Caburetter, Exhaust, Valve, and Plug faults. Mel-Frequency Cepstral Coefficients (MFCC), Continuous Wavelet Transform (CWT), and Short-Time Fourier Transform (STFT) were employed to transform the raw audio signals into two-dimensional spectrograms that contain both temporal and spectral fault signatures. Using transfer learning, these spectrograms were utilized as input features to train versions of MobileNet and AlexNet, which were pre-trained on ImageNet weights. Their performance was evaluated using accuracy, precision, recall, F1-score, and ROC-AUC metrics. Results obtained from the evaluation metrics show that MobileNet significantly outperformed AlexNet across all feature transformations (MFCC, CWT, and SSTFT). It achieved a peak accuracy of 92% and an AUC of 0.99 with STFT spectrograms. In contrast, AlexNet achieved lower accuracies (54–59%), indicating lower discriminative power. The class-wise ROC-AUC analyses confirmed that MobileNet achieved near-perfect classification, particularly in distinguishing between Normal and any of the fault conditions, while AlexNet struggled with subtle classes, such as Plug and Valve faults. These findings indicate that STFT is the most discriminative spectrogram and MobileNet is the best-performing diagnostic framework. This makes it suitable for deployment in resource-constrained environments and edge devices. This research contributes to the advancement of intelligent, real-time condition monitoring of domestic generator sets, thereby reducing downtime and enhancing energy reliability in off-grid contexts.

Optimization of Biodiesel Production from corn and millet waste using Fenton nano catalyst

Ifiok Ekaette — 2025-10-25

This study investigates the biodiesel production from corn and millet waste oil blend using a Fenton nano-catalyst. The influence of catalyst concentration, reaction time, agitation speed, methanol-to-oil molar ratio, and temperature on biodiesel yield was analyzed. Response Surface Methodology (RSM) and a one-way analysis of variance (ANOVA) were employed to optimize the process parameters and understand the interactions between them. A 2-level, five-factor factorial design was used to analyse the effects of these parameters on the yield of biodiesel. The optimized conditions were a reaction time of 3 hours, a temperature of 50°C, an agitation speed of 300 rpm, a methanol-to-oil ratio of 9:1, and a catalyst concentration of 1.5 wt%. Under these conditions, the maximum biodiesel yield achieved was 81% (v/v). The significance of this research is that the utilization of waste agricultural products to produce alternative fuel presents an alternative to fossil fuel usage, is comparatively competitive in engine performance tests, and has very good emission control, with promising performance and emission reduction benefits for industrial applications.

Exploring the Biases in the Decision to Download mHealth Apps: A Discrete Choice Experimental Analysis of Nigerian Healthcare App Users

Ignatius Nwoyibe Ogbaga — 2025-10-25

Globally, mobile health (mHealth) tools are becoming more widely available. This provides great benefits for both patients and healthcare providers. However, mHealth adoption is still comparatively low. Furthermore, few studies have examined people's intentions to download mHealth apps. The purpose of this study is to investigate people's propensity to use mobile health apps. The discrete choice experiment method was employed to investigate healthcare application users in Nigeria. Two main features (attributes) of the mHealth applications, subscription cost and data protection, were presented to the 1600 participants in the study. Each feature had two levels: ‘free subscription’ and ‘$20 subscription’ for cost, and ‘data protection’ and ‘no data protection’ for data privacy. The pricing for data protection is $20, equivalent to N39, 058.76 (at the rate of $1 = N1, 502.26), while the free subscription option does not include data protection. A total of 1,600 participants were randomly selected through a web-based survey employing a proportionate stratified sampling technique. Data were analyzed using a conditional logistic regression model in R (Clogit), yielding 43,200 observations (1600 participants’× 18 available choices× 3 alternatives). The results revealed that cost and data protection significantly influenced users’ willingness to download mHealth apps. Specifically, the odds ratio (OR) for paid apps ($20) was 0.498 (95% CI: 0.468–0.530), indicating a reduced likelihood of download compared to free apps, while the OR for data protection was 0.25 (95% CI: 0.232–0.270), suggesting that participants strongly preferred apps that ensured privacy of health data. The model yielded a log-likelihood of –23,165.48 and an R² of 0.113, confirming a good model fit. The findings imply that users are significantly more inclined to adopt mHealth apps that are free and guarantee data protection. Reducing cost barriers and strengthening privacy measures are, therefore, essential strategies for enhancing mHealth adoption in Nigeria and similar contexts.