LAUTECH Journal of Engineering and Technology

Explainable ensemble deep learning model for predicting diabetic retinopathy based on APTOS 2019 eye pack dataset

2025-02-13T21:16:28+00:00

Detection of diabetic retinopathy (DR) as early as possible is vital in mitigating the complicated issues associated with the disease. Recent advances in artificial intelligence (AI), particularly deep learning (DL) techniques, have led to appreciable increase in the accuracy of predicting various disease classes. However, the challenge of AI models is the difficulty in providing insights into how and why a model arrives in attaining decision-making to facilitate trust and adoption in clinical settings. Therefore, this study aimed to enhance the detection rate of DR and explain the significant regions on the image for the model's overall performance. This study utilised Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, Simple Recurrent Neural Networks (SRNN), and XGBoost in an ensemble model (EM). Specifically, Shapley Additive exPlanations (SHAP), a popular Explainable Artificial Intelligence (XAI) technique was utilised to identify and provide insights to which parts of the images features that contribute to the model's overall performance. After a series of experiments using the APTOS 2019 eye pack dataset collected from the Kaggle repository to evaluate the performance of CNN, LSTM, SRNN, and XGBoost. The EM outperformed all the other models with 95.63% accuracy, 97.79% precision, 93.64% recall rate, 98.79% F1-score and 97.75% AUC score. Also, SHAP analysis revealed significant regions on the image that influenced predictions, thus showing how important interpretability was for the model. The results imply that the ensemble DL, particularly with XGBoost, enhances the detection of DR, thereby improving the efficiency of screening tests and supporting personalised treatment plans in clinical practice through integrating these advanced models with XAI tools, creating trust towards automated diagnostic systems.

Influence of penetration angles on global-thermo-hydraulic-performance of shell and tube heat exchangers with multi-cross sectional tube configurations

2025-02-17T10:49:48+00:00

Shell-and-Tube Heat Exchanger (STHE), a vital component for efficient energy management when used with Straight-Tube Geometries (STG) is associated with low Global-Thermo-Hydraulic-Performance (GTHP). These contribute to the high energy demand of processing plants. The recently developed STHEs with modified tube configurations have not adequately addressed these limitations and necessitated a continuous search for tubes with improved performance. Multi-cross sectional tube geometrical (MSTG) configurations are known to improve GTHP along flow lines. This process has not been thoroughly investigated. Therefore, this study was designed to investigate the influence of penetration angles (PAs) on the STHEs’ performances using MSTG configurations. ; The numerical analysis was evaluated in terms of GTHP indices on STHE with Convergent-Divergent-Tube-Geometry (CDTG) and Divergent-Convergent-Tube-Geometry (DCTG) configurations of varying penetration angles (PAs), 5,10,15,…,90°. Numerical GTHP for STHE with STG was 1.0, while that obtained for STHE with CDTG configurations for all PAs fall between 1.50 to 1.625 with the highest at PA indicating a 50% minimum improvement in GTHP over STHE-STG. For DCTG, GTHP were between 1.43 and 1.585 for all PAs with the highest at PA indicating a 43% minimum improvement in GTHP over STHE-STG. Replacing STHE-STG with STHE-MSTG can improved their GTHPs in processing plant.

Influence of ethanol on the spectral properties of natural dyes from microbes: implications for dye-sensitized solar cell performance

2025-02-17T10:41:10+00:00

The increasing need for renewable energy sources that are both sustainable and less costly has spurred growing research into natural dyes as eco-friendly sensitizers for dye-sensitized solar cells (DSSCs). This study entails the extraction and characterization of natural pigments derived from Chlamydomonas starii and Coelastrella sp. focusing on solvent selection and thermal treatment. Ethanol proved much better than ethyl acetate and had the most remarkable effects on retaining functional groups, especially hydroxyl (O-H) and carbonyl (C=O), which would be important in terms of dye adsorption for improved photon absorption and electron injection. Spectroscopic analyses indicated that the ethanol-extracted dyes, especially non-heated and moderately heated ones, would enable a broad absorption of light within the visible spectrum. These findings hence have demonstrated that natural dyes extracted with ethanol were better than those extracted with ethyl acetate, which makes ethanol more efficient, scalable, and greener solvent for extraction of natural dyes compared to ethyl acetate. Future research is to focus on mixed solvent systems and stabilization techniques to improve dye performance even further. This will therefore be the greatest scion towards high-efficiency DSSCs while further supporting the global transition efforts toward sustainable energy technologies.

The Assessment of Groundwater Quality for Drinking and Construction uses in Odo-Oba Town,Southwestern Nigeria.

2025-02-13T21:23:34+00:00

Assessment of Groundwater Quality for Drinking and Construction uses in Odo-Oba Town, Southwestern Nigeria. 1Temilade Oladeji and 2Olufunmilayo Adewoye : helenetemi@gmail.com, aoadewoye55@lautech.edu.ng Abstract Groundwater is a crucial natural resource for human survival, serving as the primary source of freshwater worldwide. However, a lack of data on water quality can have detrimental effects on human health and resource management. This study aims investigates the chemical composition of groundwater in Odo-Oba, Oyo State, Nigeria, to assess its quality for drinking and construction. The objective is to evaluate the potential for possible contamination, to determine the water facies type and assess its suitabilityfor sustainable development using water quality indices, plotsand statistical analysis for potable water supply and construction purposes.Twenty groundwater samples were analyzed for physical and chemical parameters and were compared to international standards. Results showed slightly acidic water with moderate temperatures and elevated mineral content. Potassium was the most abundant cation, while chloride was the dominant anion. Most parameters were within acceptable limits, but some samples were unsuitable for drinking. Rock-water interactions significantly influence the water chemistry. While groundwater in Odo-Oba is unsuitable for drinking without treatment, it is suitable for construction. However, further microbiological analysis and proper waste management are crucial to protect groundwater resources. Testing concrete samples for pH changes is recommended to assess potential impacts on structural integrity. Keywords: Assessment, Groundwater, Quality, Construction, Drinking, Suitability

Unsteady aerodynamic performance of spiked and unspiked aloe vera-inspired aerofoils at low reynolds number

2025-02-13T21:32:39+00:00

The increasing demand for Micro Aerial Vehicles (MAVs) and other low-speed aerial platforms has intensified the focus on optimizing aerofoil designs for low Reynolds number applications. This study presents a comparative analysis of the unsteady aerodynamic performance of aloe vera-inspired aerofoils, both spiked and non-spiked, at Reynolds number 1.5 × 10⁵. By employing the Unsteady Reynolds-Averaged Navier-Stokes (URANS) equations alongside the Transition SST (four-equation) turbulence model, the research accurately characterizes the flow field and aerodynamic loads under dynamic conditions. The results indicate that the novel spiked aerofoil exhibits superior performance at all angles of attack, demonstrating delayed flow separation and enhanced lift coefficient and lift-to-drag ratios compared to the non-spiked aloe vera aerofoil.

Renewable energy potential of bio-Oil from pyrolysis of gmelina arborea seeds cultivated in Nigeria

2025-02-19T20:47:01+00:00

The increasing demand for renewable energy sources has intensified the search for sustainable alternatives to fossil fuels. Gmelina arborea seeds, an underutilized biomass abundant in Nigeria, hold potential for bio-oil production. This study aims to explore the feasibility of utilizing Gmelina arborea seeds for energy generation through pyrolysis, contributing to cleaner energy production and environmental conservation. Seeds were collected from LAUTECH, Ogbomoso, and subjected to air and oven drying to reduce moisture content before being ground into powder. Pyrolysis was performed in a bench-scale screw reactor at temperatures ranging from 485 to 596 °C. The bio-oil produced was characterized using standard ASTM methods, including proximate and ultimate analyses, Fourier Transform Infrared Spectroscopy (FTIR), and Gas Chromatography-Mass Spectrometry (GC-MS), to determine its chemical composition and energy potential. The bio-oil yield ranged from 21.3 to 25.3 wt.%, with the highest yield of 25.3 wt.% achieved at 596 °C. Characterization revealed favorable energy properties, including a Higher Heating Value (HHV) of 40.13 MJ/kg, kinematic viscosity, density, and flash point within practical application ranges. FTIR analysis identified functional groups such as alkenes, carboxylic acids, alcohols, ethers, and ketones, while GC-MS detected hydrocarbons like alkanes, alkenes, phenols, and naphthalene. The low nitrogen content (2.58-2.80 wt.%) indicates minimal environmental impact. This study highlights the viability of Gmelina arborea seeds as a renewable bioenergy feedstock, offering a cleaner, sustainable alternative to conventional fuels.

Functional, physicochemical and pasting properties of pupuru and pupuru analogue made from cassava-orange fleshed sweet potato

2025-02-20T20:52:14+00:00

Blends of cassava (cas) and orange-fleshed sweet potato (ofsp) at the ratio of 75:25%, 50:50%, 25:75%, and 0:100% (cas: ofsp) respectively, and 100% cassava as the control were processed to pupuru flour. Samples were analyzed for functional properties, physicochemical properties and pasting properties. The packed bulk density and WAC ranged from 0.71-0.87, 0.79-0.93, 0.84-1.02, 0.95-1.06 g/ml; 2.01-4.05, 2.07-4.07, 2.04-3.12 and 2.07-3.15 g/ml at 75:25%, 50:50%, 25:75% and 0:100%, respectively. The pH ranged from 4.00-5.15, 3.63-4.12, 3.53-4.00 and 4.35-5.01 at 75:25%, 50:50%, 25:75%, and 0:100%, respectively. The cyanide contents ranged from 3.70-0.50, 3.40-0.42, 3.00-0.31, 0.27-2.97 and 1.31-4.40 at 75:25%, 50:50%, 25:75%, and 0:100%, respectively. The pasting time ranged from 4.60 -5.67 min, 4.20 -5.40 min, 4.30-5.50 min and 4.20 -5.50 min for the blended sample at 75:25%, 50:50%, 25:75%, and 0:100% (cas: ofsp), respectively. The pasting temperature ranged between 72.6-87.2⁰C, 66.4-85.5⁰C, 69.5-85.5⁰ C and 65.5-81.5⁰C for 75:25%, 50:50%, 25:75%, and 0:100% (cas: ofsp), respectively. The final viscosity ranged from 1060-2645 RVU, 1368-2361 RVU, and 1268-2281, 1028-2081 for 75:25%, 50:50%, 25:75%, and 0:100% (cas: ofsp), respectively. **This study has established the potentials of ofsp in the production of an acceptable nutrient densed pupuru to enhance food and nutrition security.**

Soft computing based load forecasting using artificial neural networks: a case study of Lagos, Nigeria

2025-02-19T13:07:08+00:00

This study introduces a soft computing approach using Artificial Neural Networks (ANN) for load forecasting, specifically focusing on predicting the minimum and maximum load power. The goal is to efficiently allocate the expected power to suitable load centers. The analysis utilizes a 3-year historical dataset of load consumption in Lagos, a city in Western Nigeria. A Multi-layered Perceptron (MLP) network is employed to generate short-term load forecasts for the area. The inputs for the network include monthly data, while the output parameters are load data obtained from the energy company, which are used to predict power needs in the geographical area. The ANN training employs supervised learning and the back-propagation algorithm, implemented through MATLAB & SIMULINK. The input and target data are preprocessed and normalized within the range of -1 and 1. The network is continuously trained until desirable regression values and a disparity graph are achieved. The study demonstrates significant success with regression values of 0.96, 0.97, and 0.97 obtained over three consecutive years (2021/2022, 2022/2023 and 2023/2024) which indicate that the model accurately predict the load of year 2024. The developed model holds promise for independent power companies in Nigeria to enhance load allocation planning and forecast expected revenue.

Performance evaluation of Osprey optimization algorithm-based proportional integral derivative controller for speed control of a brushless direct current motor

2025-02-19T13:15:37+00:00

ABSTRACT
Owing to diversity of application, speed regulation of Brushless Direct Current (BLDC) motor is essential in order to achieve best performance of the motor. In this paper, an appropriately tuned controller such as Proportional Integral Derivative (PID) is employed to achieve effective speed control of the motor. In tuning the parameters of PID controller, conventional techniques often pose great difficulties due to non-linearity often exhibited by DC motors. As a solution, metaheuristic optimization techniques are adopted to optimally tune the PID controller parameters for optimal performance of the BLDC motor in terms of speed. Thus, Osprey Optimization Algorithm (OOA) tuned PID controller (OOA-PID) was used to achieve better performance of BLDC motor speed. Kirchoff’s Voltage Law and Newton’s second law of motion were employed to derive the BLDC motor mathematical model. The PID mathematical equation was also described and an optimization model was formulated using the Integral of Time Multiplied Absolute Error (ITAE) and optimized using OOA. The performance of the OOA-PID controller with BLDC motor was evaluated using performance metrics such as rise time, settling time, overshoot and steady state error. Simulations were done using MATLAB (R2021b). Simulation result shows that an OOA-PID controller gave better response when compared with existing ziegler Nichols PID (ZN-PID) used for the same purpose.

Utilization of local gum Arabic as an adhesive for the production of particle board from maize cobs and coconut shells

2025-02-19T13:25:31+00:00

The increase in the cost of building and furniture materials like particle board, ceiling and roofing sheets in Nigeria worsens by the day. The development of an alternative to curb this menace using agricultural waste such as maize cobs and coconut shells are required. In this study, maize cob, coconut shell and gum Arabic as adhesive were used in the development of particle board. The particle board was produced by mixing maize cob and coconut shell in different proportion. The physical and mechanical properties of the particle board such as density, water absorption, impact strength. compressive and tensile tests were investigated. The density ranged from 518.6 -843.3 Kg/m³, water absorption was between 11.05 and 70.2% by varying time of immersion at 30min, 1hr and 2hr. The compressive strength, tensile strength and impact strength fell within the range of 15.9 - 20.15MPa, 18 - 62KPa and 175 - 274.06KJ/m² respectively. The results showed that the maize cob, coconut shell and gum Arabic are good candidates for building and furniture applications.

Energy Theft and Fault Detection in Smart Energy Meter Using Fuzzy Logic

2025-02-19T13:30:46+00:00

Energy theft and fault is a significant global issue in the energy sector, which accounts for non-technical loss in power utilities. Specifically, in Nigerian context, energy theft arises from meter tampering to reduce energy consumption readings, unauthorized or illegal connections, and meter swapping. The nexus of Internet of Things and Fuzzy logic learning techniques applied in smart metering system could mitigate this endemic problem. While there are efforts in literature that address this problem. However, there is still need for an optimal and smart metering system with an increased intelligence and accuracy to address energy theft and fault detection problem in distribution network. Hence, this work presents a solution to energy theft and fault detection in the distribution network by designing a smart energy metering system that applies simple fuzzy logic rules to analyze data thereby detecting theft and fault in the system. The system compares the measurements from a pole-based smart meter and consumer unit meter for detection of fault and theft. The smart energy meter is designed with a microcontroller and uses Wi-Fi communication for data transmission from pole-based metering point and consumer unit metering point to the central server for analysis and processing. The result shows that Fuzzy logic is useful in solving energy theft and fault detection problems. The accuracy and precision are 95% and 95% respectively for theft detection as well as 91% and 95% for fault detection.

Comparative design of flow reactors for the production of 100,000 tons per year of cumene from the catalytic alkylation of propylene and benzene

2025-02-19T20:59:10+00:00

This research is driven by the need to ensure effective, economic and sustainable processes for cumene production from the catalytic alkylation of propylene and benzene in flow reactors. The flow reactors are the continuous stirred tank reactor (CSTR) and the plug flow reactor (PFR) where the alkylation reaction occurred. The reactors were designed by exploring the conservation principle of mass and energy over the reactors. The performance model of the reactors were simulated using MATLAB at the same initial feed and operating temperature of 481.1k and 483k with fractional conversion changes within the range of X_A at an interval of 0.05. The comparative analysis of the flow reactors design was based on the target product yield (cumene yield) and the energy efficiency of the process. The cumene yield is dependent on the reactor volume while the energy efficiency of the process depends on the quantity of heat generated per unit volume of the reactor. At maximum fractional conversion of 0.95, the volume of the CSTR and the PFR design were 52.296m³ and 19.771m³ with a percentage difference of 22.6% while the quantity of heat generated per unit volume of the CSTR and PFR were 0.013j/sm³ and 0.035j/sm³ with a percentage difference of 22.9%. The above comparative design analysis showed that in terms of cumene yield, the CSTR displayed a better performance characteristics as indicated by the reactor volume while in terms of energy efficiency, the PFR showed a better performance characteristics as indicated by the quantity of heat generated per unit volume of the reactor. This article has shown that both the CSTR and the PFR are suitable for cumene production and the choice of reactor depends on the designer’s primary need.

This research is driven by the need to ensure effective, economic and sustainable processes for cumene production from the catalytic alkylation of propylene and benzene in flow reactors. The flow reactors are the continuous stirred tank reactor (CSTR) and the plug flow reactor (PFR) where the alkylation reaction occurred. The reactors were designed by exploring the conservation principle of mass and energy over the reactors. The performance model of the reactors were simulated using MATLAB at the same initial feed and operating temperature of 481.1k and 483k with fractional conversion changes within the range of X_A at an interval of 0.05. The comparative analysis of the flow reactors design was based on the target product yield (cumene yield) and the energy efficiency of the process. The cumene yield is dependent on the reactor volume while the energy efficiency of the process depends on the quantity of heat generated per unit volume of the reactor. At maximum fractional conversion of 0.95, the volume of the CSTR and the PFR design were 52.296m³ and 19.771m³ with a percentage difference of 22.6% while the quantity of heat generated per unit volume of the CSTR and PFR were 0.013j/sm³ and 0.035j/sm³ with a percentage difference of 22.9%. The above comparative design analysis showed that in terms of cumene yield, the CSTR displayed better performance characteristics as indicated by the reactor volume while in terms of energy efficiency, the PFR showed better performance characteristics as indicated by the quantity of heat generated per unit volume of the reactor. This article has shown that both the CSTR and the PFR are suitable for cumene production and the choice of reactor depends on the designer’s primary need.