Institutional-Repository, University of Moratuwa
Welcome to the University of Moratuwa Digital Repository, which houses postgraduate theses and dissertations, research articles presented at conferences by faculties and departments, university-published journal articles and research publications authored by academic staff. This online repository stores, preserves and distributes the University's scholarly work. This service allows University members to share their research with a larger audience.
 Research Publications |  Thesis & Dissertation |  E- Books |
 UoM Journal Publications |  UoM Conference Proceedings |  Articles published in Scimago's Q1 journals |  UoM Research Reports |  Other Articles authored by UoM staff |
Recent Submissions
item: Thesis-Abstract
Development of a thermoplastic vulcanizate from natural rubber and polyethylene as a competitive material for roofing sheets
(2026) Wickramaarachchi, WVWH; Walpalage, S; Egodage, SM
In Sri Lanka, roofing materials are commonly manufactured from clay, metal, plastic, wood, and asbestos, each associated with specific limitations. Despite well-documented health risks, asbestos roofing remains widely used due to its low cost. However, extensive scientific evidence linking asbestos exposure to severe health hazards has prompted the Government of Sri Lanka to initiate measures toward banning asbestos-based roofing materials. This has created an urgent need for safer, durable, and cost-effective alternatives. Thermoplastic vulcanizates (TPVs), produced through dynamic vulcanization in which a rubber phase is selectively crosslinked and dispersed within a thermoplastic matrix, offer a promising solution. In this context, the present study focuses on developing a sustainable TPV roofing material using natural rubber (NR) and polyethylene (PE).
Initially, NR/PE blends were prepared via melt blending using a twin-screw extruder with three polyethylene grades, namely low-density polyethylene (LDPE), linear low-density polyethylene (LLDPE), and high-density polyethylene (HDPE), at varying compositions to identify the most suitable PE grade for roofing applications. Based on mechanical, physical, and water absorption performance, HDPE was identified as the optimal thermoplastic component. Subsequently, TPVs from NR and HDPE were further optimized by varying blend composition and incorporating particulate fillers to improve performance and reduce cost. Six fillers, including dolomite, calcium carbonate, barium sulfate (BaSO₄), snobrite clay, talc, and kaolin, were evaluated, with BaSO₄ showing the best overall compatibility and reinforcement. Further optimization involved adjusting the loadings of dicumyl peroxide (DCP) as the vulcanizing agent, BaSO₄ filler, and a zirconate coupling agent to enhance interfacial adhesion. Mechanical, thermal, and physical properties were evaluated in accordance with international standards, and morphology was examined using scanning electron microscopy. Prototype roofing sheets (40 cm × 40 cm × 0.6 cm) were fabricated by compression moulding and assessed against relevant PVC-based roofing specifications.
NR/HDPE-based TPVs showed higher tensile strength, tear strength, and hardness than those made with LDPE and LLDPE. Although none reached the 25 MPa tensile strength of commercial PVC-based roofing sheets, all exceeded the 3 MPa minimum required for roofing, with impact strength surpassing that of PVC-based roofing sheets. NR/HDPE-based TPV exhibited the lowest water absorption (0.04%), nearly matching PVC-based roofing sheets and significantly better than asbestos sheets. Based on overall performance, HDPE was identified as the most suitable PE grade for NR/PE roofing applications. The 20/80 (NR/HDPE) composition offers the best balance of mechanical strength and impact resistance, making it ideal for roofing applications. The TPV prepared with BaSO₄ filler demonstrated the highest overall performance, with optimal results observed at a filler loading of 50 phr and 1 phr of DCP. A zirconate loading of 1.5 phr resulted in optimum mechanical properties and refined phase morphology, indicating enhanced interfacial compatibility between the NR and HDPE phases in the TPV. The prototype sheet prepared using the optimized formulation showed comparable impact strength, elongation at break, density, and thermal conductivity to that of conventional PVC-based roofing sheets
item: Thesis-Full-text
Intersection waypoint identification for vision based indoor navigation of nano-scale drones
(2026) Kodithuwakku, HKAYD; Hettiarachchi, C; Sooriyaarachchi, S
Nano-scale drones represent a promising platform for indoor inspection and search operations in GPS-denied, constrained environments, as their compact form factor enables access to narrow spaces. However, they require lightweight, resource-efficient autonomous navigation algorithms. Vision-based navigation offers an infrastructurefree substitute to GPS and external positioning systems. Simple vision-based navigational algorithms such as ring attractor, optical flow-based approaches, suffer from drift and cumulative error over time. Introducing semantic waypoints provides a mechanism to mitigate such drift. This work proposes an egocentric, vision-based waypoint navigation framework that formulates waypoint detection as an intersection understanding problem, cast as a concept learning task. We first formulate the intersection identification problem as a multi-class classification. To ensure a model can be trained with limited data and still preserve the conceptual understanding, each intersection is decomposed into its fundamental directional components: left, right, and forward. This novel multi-label classification is advantageous because it can extend to zero-shot inference and complex junction types. Further, we augment the limited real data with synthetic visual data collected from a safer digital twin environment which replicates the real-world scenes. We design five experimental setups in which model weights are initialized from Kaiming He, ImageNet or synthetic pre-trained weights obtained using the Digital Twin dataset. We show that, (i) Intersections can be used as waypoints for the navigation of nano-drones. (ii) Multi-label classification outperforms multi-class formulation in intersection identification. (iii) Model trained using digital twin data alone can yield an F1 score of 0.6 in real-world predictions, compared to 0.5 from a random classifier. (iv) Introducing limited real-world training data can bring the model up to near-perfect accuracy with faster convergence. (v)The multi-label classification of junctions provides insightful visual explanations ensuring reliability and generalizability of the conceptual framework. Finally, the proposed framework is effective in environments such as libraries and supermarkets where structured aisle naturally exists and GPS is unavailable. This method offers an infrastructure and drift-resistant solution for the navigation of nano-scale drones by leveraging intersections as semantic waypoints.
item: Thesis-Abstract
Fire performance of masonry walls : experimental investigation of wall bonding patterns
(2026) Wanasinghe, WMCS; Weerasinghe, TGPL
Masonry walls are widely used as load bearing and compartmentation elements due to their inherent fire resistance; however, their structural performance under fire exposure is governed by complex thermo-mechanical interactions that depend on masonry unit geometry, mortar configuration, and bonding arrangement. Despite extensive international research, experimental data on the fire performance of masonry constructed using locally available materials remain limited, particularly for Sri Lankan construction context. This study experimentally investigates the fire performance of load-bearing masonry walls using medium-scale wallette testing. Eight masonry wallettes (including the trial specimen) were constructed using three commonly used masonry unit types—smart bricks, polished bricks, and standard solid clay bricks—incorporating different mortar joint thicknesses and bond patterns. Specimens were exposed to one-sided heating following a time–temperature curve based on ISO 834 using a medium-scale laboratory furnace for a duration of 60 minutes. The experimental program evaluated thermal response, vertical and lateral deformation behavior, visual damage characteristics, and residual compressive strength after cooling. The results showed that all masonry wallettes maintained unexposed surface temperatures below commonly referenced insulation limits under the applied heating conditions. A consistent trend was observed indicating that increased mortar joint thickness was associated with higher internal temperatures and greater vertical and lateral displacements. Bond pattern exhibited a secondary influence, with Flemish bond demonstrating comparatively improved deformation control relative to stretcher and English bonds. Smart brick masonry showed comparatively greater strength degradation after fire exposure, while polished and standard brick masonry retained higher residual compressive capacity, reflecting the influence of unit geometry and brick–mortar interaction. Overall, the findings indicate that masonry fire performance is governed by the combined effects of unit type, mortar configuration, and bonding arrangement. The study provides experimental evidence relevant to locally available masonry systems and demonstrates the suitability of medium-scale testing for comparative performance-based fire evaluation.
item: Thesis-Abstract
A Responsible technological transformation framework for micromobility manufacturing industry in Sri Lanka
(2026) Iresha, WADC; Perera, HN
The manufacturing sector is experiencing rapid technological disruption driven by Industry 4.0 and the transition toward Industry 5.0. While these advancements promise efficiency and innovation, they also raise critical concerns regarding ethics, sustainability, and workforce displacement. This study examines Responsible Technological Transformation (RTT) as a framework for balancing technological adoption with social, environmental, and organizational considerations. Drawing on Dynamic Capability Theory, Socio-Technical Theory, and Resource Dependence Theory, the research develops and empirically tests a model linking Human–Machine Collaboration (HMC) and Human-Centered Artificial Intelligence (HCAI) to Technology Readiness at both technical (TRL) and organizational (OTRL) levels, with RTT mediating their effects on supply chain capabilities and manufacturing performance. A survey of two hundred respondents from Sri Lanka’s micro-mobility manufacturing sector was analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). Sri Lanka serves as a representative case of developing economies navigating responsible transformation, offering insights for other emerging manufacturing contexts. Unlike prior studies focused narrowly on Responsible AI or Industry 4.0/5.0, this research integrates HMC, HCAI, TRL/OTRL, and RTT into a unified pathway, uncovering how responsible transformation translates human- centered technologies into resilience, agility, and firm performance. Results reveal that HMC and HCAI significantly enhance both TRL and OTRL, with organizational readiness exerting a stronger influence. RTT emerges as a critical enabler of Supply Chain Resilience (SCR) and Supply Chain Agility (SCA), which in turn positively impact Manufacturing Firm Performance (MFP). The study concludes that sustainable competitiveness in manufacturing depends on embedding responsibility alongside technological adoption.
item: Thesis-Full-text
Customer segmentation and cross-selling recommendation engine for the telecom industry : a machine learning approach
(2025) Nanayakkara, JD; Dias, D
In the telecommunication industry, personalized service delivery and effective crossselling strategies are vital for enhancing customer engagement and driving revenue growth. This research presents a data-driven approach to customer segmentation and personalized recommendation, aiming to promote the transition from Double-Play (Voice and Internet) to Triple-Play (Voice, Internet and IPTV) service adoption. The study applies unsupervised machine learning techniques, namely Gaussian Mixture Models (GMM) and K-Means clustering to segment customers based on demographic, billing, and service usage attributes. Multiple internal evaluation metrics, including the Silhouette Score, Calinski-Harabasz Index, Davies-Bouldin Index and the Elbow method are employed to determine the optimal clustering structure and validate the segmentation quality. To generate personalized service recommendations, a collaborative filtering approach based on Singular Value Decomposition (SVD) is utilized. The model predicts the most suitable IPTV package for Double-Play customers by learning from existing customer interaction patterns. The performance of the recommendation system is evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). The implementation leverages Python-based tools and libraries for data preprocessing, modeling and visualization. The research offers a practical and scalable framework for targeted marketing and intelligent service personalization, contributing to the advancement of data-driven strategies in the telecommunications domain.