IUT Institutional Repository

IUT Institutional Repository is a digital service that collects, preserves, and distributes digital material. Repositories are important tools for preserving an organization's legacy; they facilitate digital preservation and scholarly communication.

Communities in IUT Institutional Repository

Select a community to browse its collections.

Now showing 1 - 5 of 12

Recent Submissions

A Comparative Study of Supervised and Unsupervised Deep Learning Models for Fabric Defect Localization in Smart Textile Manufacturing
(Department of Mechanical and Production Engineering(MPE), Islamic University of Technology(IUT), Board Bazar, Gazipur-1704, Bangladesh, 2025-10-25) Bhuiyan , Md. Mahmudul Hasan; Khan, Nabila Islam
The increasing demand for automated visual inspection in textile manufacturing has intensified the demand for strong and efficient defect detection systems. Traditional manual inspection methods are often inconsistent and resource-intensive, while the inherent variability of fabric textures creates obstacles for conventional machine vision techniques. To overcome these constraints, this study presents a comparative analysis of supervised semantic segmentation and unsupervised anomaly detection models for fabric defect localization using the ZJU-Leaper benchmark dataset. A total of nine deep learning architectures were evaluated across three methodological categories: (i) CNN-based supervised models (U-Net, PSPNet, FPN, DeepLabV3, U-Net++, and DeepLabV3+), (ii) Transformer-based supervised models (UPerNet and SegFormer), and (iii) an unsupervised anomaly detection model (EfficientAD). All models were trained on standardized preprocessed images and optimized under identical hyperparameter settings to ensure fair benchmarking. The evaluation followed the multi-level protocol defined in the ZJU-Leaper paper, incorporating sample-, pixel-, and region-level metrics, alongside efficiency indicators such as FLOPs, parameter count, and FPS to assess deployment feasibility. The results show distinct trade-offs between accuracy and efficiency across the examined architectures. Transformer-based models demonstrated superior generalization on complex textures, achieving balanced performance between global context capture and computational cost. CNN-based models excelled in fine-grained localization, offering high Dice and IoU scores at moderate inference speeds. Conversely, the unsupervised EfficientAD model achieved competitive detection accuracy without relying on labeled data, underscoring its potential for defect-scarce industrial scenarios. Overall, this research establishes a unified evaluation framework for benchmarking supervised and unsupervised deep learning models in fabric inspection. The results show into the architectural and operational considerations necessary for developing scalable, real-time, and label-efficient quality control systems in smart textile manufacturing environments.
Design, Fabrication and Performance Study of a Solar Water Desalination (Single Slope Still)
(Department of Mechanical and Production Engineering(MPE), Islamic University of Technology(IUT), Board Bazar, Gazipur-1704, Bangladesh, 2025-10-30) Kamara, Aakifah Yainkain; Bah, Momodou; Jarjusey, Alimamy
Access to clean water remains a critical global challenge, particularly in arid and remote regions where conventional desalination systems are costly and energy-intensive. To address this, the present study aims to develop an affordable and sustainable solar desalination unit capable of producing potable water using only solar energy. It is hypothesized that a simple single-slope solar still, optimized through effective thermal insulation, shallow basin design, and proper cover inclination, can significantly improve freshwater yield and overall efficiency. A prototype was designed, fabricated, and experimentally tested at outdoor conditions using a tempered-glass cover, a shallow, black-painted basin, a thermal isolator, a brine drain, and a channel used to collect condensates. The performance was evaluated in terms of temperature distribution, solar energy absorption, and hourly distillate output. The design was to increase thermal retention and evaporation speed based on the literature suggestions regarding the most effective cover angles, shallow water depth, and better insulation. Solar radiation, ambient/basin/glass temperatures, relative humidity, and production of hourly distillate were measured during various test days in August at different levels of irradiance. The daily freshwater yield was found to be at a maximum 1.15 L/day during test days with a maximum thermal efficiency of 39.7% with the highest performance achieved under strong solar irradiance and minimal cloud cover. Incident radiation and operating temperatures were found to have a strong dependence on productivity, and cloud cover and rainfall adversely affected it; the use of an insulation layer and a shallow basin enhanced performance as compared to traditional untreated designs. The modified unit exhibited improved performance with higher efficiency and competitive yield, while preserving a simple design, minimal maintenance, and zero fuel consumption. The findings confirm that a simple, fuel-free, and low-maintenance solar still can effectively provide potable water in resource-limited environments, supporting sustainable development goals related to clean water and renewable energy.
Research and Development of Printed Circuit Heat exchanger
(Department of Mechanical and Production Engineering(MPE), Islamic University of Technology(IUT), Board Bazar, Gazipur-1704, Bangladesh, 2025-11-15) Ousmane, Seone; Selikouma, Traore; Rahman, Aliyu Shuaibu Abdul
The present report is a summary of modern, experimental, numerical, and analytical studies on the printed circuit board heat exchanger geometries and will include zigzag, serpentine, straight, wavy, and cellular channel designs to arrive at practical conclusions on the efficiency of high-performance compact recuperators with and without the use of He-Xe and supercritical 𝑠𝐶𝑂2 environments. Printed Circuit Heat Exchangers (PCHEs) are known to have a high ratio of heat-transfer area to volume and demonstrate desirable traits in respect of their pressure and temperature performance and have been proposed to be used in micro transport reactors, in 𝑠𝐶𝑂2 or Brayton systems. In this context, geometry is a critical variable that represents the trade-off between thermal effectiveness and pressure drop: zigzag and cellular zigzag designs can considerably improve the effectiveness and the heat-transfer coefficient compared to straight or serpentine channels, but at the price of a larger pressure drop; but with possible compensations across modified zigzag designs, with cellular designs or with channel designs made of straight channels, or simply three-dimensional wavy or sinusoidal geometries. The secondary effects of interest encompass: axial conduction through thin plates, fin performance, the sensitivity of working fluid properties (such as the He-Xe mole fraction or 𝑠𝐶𝑂2 operation just above pseudo-critical conditions), and mechanical/manufacturing constraints (including brazing, tolerances, and stresses), all of which exert a profound influence on the performance and durability that can be attained. It is recommended that multi objective geometry optimization (considering performance, pressure drop, and mass) be undertaken, experimental validation should occur under conditions relevant to the application (such as He-Xe mixtures and 𝑠𝐶𝑂2 transcritical operations), and enhancements in materials and joining techniques should be pursued to mitigate thermal stresses and leakage risks; furthermore, additional research on fouling, transient response, and manufacturability is warranted.
Toward Ergonomic and Productive Human-Machine Interaction: A Conceptual Study of Hybrid Gesture and Voice Interfaces.
(Department of Mechanical and Production Engineering(MPE), Islamic University of Technology(IUT), Board Bazar, Gazipur-1704, Bangladesh, 2025-10-25) Sheila, Mapon Mounde Mariama; Salahouddine , Abdoul Wadoud; Bacar, Moufadhul Said
Industrial automation increasingly relies on Human Machine Interfaces (HMIs), yet conventional unimodal controls such as buttons or gestures alone often lead to operator fatigue and reduced productivity. This thesis addresses this challenge by evaluating a Hybrid HMI that integrates gesture and voice control to enhance ergonomics, efficiency, and operator performance. Using a conceptual evaluation approach, mockups and workflow diagrams were employed to simulate hybrid control scenarios. Twenty-five participants assessed the system through two standardized methods: the System Usability Scale (SUS) for usability and NASA TLX for perceived workload. Results revealed a mean SUS score of 81, exceeding the benchmark of 78, and demonstrated excellent usability. The workload analysis indicated low frustration, moderate effort, and high satisfaction with performance, highlighting that hybrid interfaces alleviate strain while improving operator responsiveness. These findings underscore the potential of integrating gesture and voice control to outperform unimodal gesture-only or button-only systems. The study concludes that hybrid multimodal HMIs hold significant promise for advancing Industry 4.0 goals by fostering more ergonomic, resilient, and productive work environments. While conceptual, this research emphasizes the value of early-stage usability testing through mockups to inform design decisions before costly prototyping. Future work will focus on developing a physical prototype, testing under real industrial conditions.
A Hierarchically Architected Bio-inspired Impact Resistant Natural Fiber Reinforced Composite
(Department of Mechanical and Production Engineering(MPE), Islamic University of Technology(IUT), Board Bazar, Gazipur-1704, Bangladesh, 2025-10-30) Amin , Ahsanul; Hossain, Mohammad Abraar
The Bouligand structure, known for helicoidal fiber alignment in biological materials, offers unique mechanisms for enhancing impact resistance and damage tolerance. This study investigated the mechanical performance of a novel bioinspired composite based on the Bouligand architecture using densified wood (Gmelina Arborea) as fiber reinforcement. Composite laminates were fabricated through densification, angular slicing, and epoxy-based lamination to emulate helicoidal layering. The samples were prepared with varying stacking angles and subjected to a comprehensive set of mechanical tests to evaluate their flexural, compressive, impact, shear, and moisture absorption properties. Testing revealed that the composite with the Bouligand structure exhibited significant performance enhancements in multiple domains, whereas anisotropic testing confirmed greater strength in the radial loading directions. The impact strength was approximately 9 times greater than that of untreated wood, which was attributed to energy-dissipative mechanisms such as fiber pull-out and crack deflection. Flexural testing revealed an ultimate strength of 198.22 MPa and a modulus of 27.15 GPa, indicating substantial improvement over conventional stacking configurations. The interlaminar shear strength was highest for the Bouligand configurations, which suggests improved interfacial adhesion due to twisted layering. Microscopic analysis confirmed the role of fiber‒matrix interactions and ply orientation in promoting discontinuous crack propagation. Moisture absorption tests revealed enhancements in hydrophobic characteristics. In addition to being lightweight (0.738 g/cm 3), the renewability of fibers demonstrates strong potential for the composite in structural applications, especially in impact-prone or beam structures in transportation or protective panels. This work provides a scalable, bioinspired framework for engineering natural fiber composites with enhanced multifunctional performance, which is a promising alternative to traditional synthetic composites