Chalmers Open Digital Repository

Välkommen till Chalmers öppna digitala arkiv!

Här hittar du:

Studentarbeten utgivna på lärosätet, såväl kandidatarbeten som examensarbeten på grund- och masternivå
Digitala specialsamlingar, som t ex Chalmers modellkammare
Utvalda projektrapporter

Forskningspublikationer, rapporter och avhandlingar hittar du i research.chalmers.se

Enheter i Chalmers ODR

Välj en enhet för att se alla samlingar.

Visar 1 - 2 av 2

Senast publicerade

Development of Sustainable Battery Materials from Steel Waste
(2026) Smith, Erik Charles
The global transition away from fossil fuels has driven surging demand for battery production and the associated materials such as lithium iron phosphate (LFP). Yet the LFP supply chain remains heavily concentrated in China, posing significant sup ply chain vulnerabilities through geopolitical tensions. In order to onshore battery manufacturing in the European and North American markets new feedstocks must be developed. This thesis investigates the feasibility of producing LFP from steel machining swarf through the lens of particle size optimization utilizing high energy ball milling. Surfactant assisted ball milling with 5 wt% myristic acid addition was identified as the optimal condition for sufficient particle size reduction of the iron phosphate precursor, reducing the median particle size from approximately 3 µm to 0.5 µm. The resulting LFP cathode achieved a specific capacity of 102 mAhg−1 at 0.5C with approximately 85% capacity retention after 90 cycles, demonstration the viability of converting steel waste into LFP cathode material while necessitating further optimization to become competitive with commercial LFP
Iterative Clustering for Memory-Assisted Reinforcement Learning in De Novo Drug Design Language Models
(2026) Winkler, Jan Marten
Reinforcement learning for de novo molecular design often suffers from mode collapse, where the policy rapidly concentrates on narrow regions of chemical space at the expense of structural diversity. This thesis presents a modular approach for combining experience replay and clustering approaches to enhance diversity within REINVENT. Across three targets, an implementation based on BitBIRCH clustering generally improved fingerprint-based diversity and produced more diverse high scoring molecules than baseline and scaffold-based methods. The results also reveal a trade-off between diversity and score optimization. Experience replay accelerates early learning but may restrict exploration, while combining it with BitBIRCH partially mitigates this effect and can improve diversity further. Overall, the proposed method is effective but target dependent, emphasizing the need to evaluate molecular quality and multiple forms of diversity together.
Predictive Resource Tuning of Apache Spark Jobs in Kubernetes - A Machine Learning driven approach, with uncertainty aware predictions
(2026) Ndirangu, Leah Wanja; Petersson, Alexander
Apache Spark is a widely used distributed data processing engine for large-scale analytics and machine learning workloads, typically executed in multi-node cluster environments. Kubernetes is increasingly used to deploy and manage these workloads in container orchestration, resource scheduling, and scalable cluster management. Despite this infrastructure, Spark workloads require users to specify CPU and memory allocations in advance. In practice, these resource requests are often overestimated to prevent job failures under uncertain workload requirements. This leads to inefficient resource utilization in shared clusters, reduced overall throughput, and increased operational cost in multi-tenant environments. This thesis addresses the problem of static and overly conservative resource provisioning for Spark jobs running on Kubernetes. This thesis proposes a machine learning-based resource tuning system that learns from historical execution behavior to predict more accurate resource allocations for defined workload types. In addition, the system incorporates uncertainty awareness to ensure that resource adjustments are applied more conservatively when prediction confidence is low. The proposed approach aims to reduce resource overprovisioning while maintaining execution stability, thereby improving cluster efficiency in distributed Spark environments. The results demonstrate that the tuning approach reduced memory from 64 GB to 8 GB and cores from 4 to 1 per executor for a designed test job. The total savings was 616 GB of memory and 30 cores per such job after tuning for 40 iterations. This increased resource utilization, although execution time increased by approximately 55%. Uncertainty-aware Wilson intervals were successfully incorporated into memory tuning but not core tuning due to its continuous cost-based optimization objective.
Assessment of sustainable mass management strategies in infrastructure projects - Application of the SCORE method in a case study of Lundbyleden
(2026) Said, Kamal
Large infrastructure projects often result in large amounts of excavated material. This raises important questions about handling, transport and opportunities for reuse. This thesis applies the SCORE (Sustainable Choice of Remediation) tool to the Lundbyleden project in Gothenburg to help guide sustainable management of these materials. Three mass management alternatives were compared: i) sending all the material to landfill, ii) sieving to allow for some reuse of masses, and iii) a combination of sieving and washing, increasing the reuse rate even more. Each alternative was assessed for its environmental, social and economic impacts to identify which best balanced the different aspects of sustainability. Further, three scenarios were developed to investigate the impact of different perspectives (sustainability vs. circular economy) and different references (“no action” or “landfill disposal”). The results, which were supported by Monte Carlo simulations, showed that treating the material to allow for reuse of masses generally gave better sustainability results than sending it directly to landfill. The results indicate that the most preferred alternative depends on the evaluation approach. From an environmental standpoint, sieving combined with washing had the best environmental performance, due to reduced landfill use, lower transportation impacts and greater reuse potential. On the other hand, from a circular economy perspective that considers both environmental and economic factors, sieving alone was more favourable, as it delivered environmental benefits at lower cost compared to washing. Including all sustainability dimensions resulted in considerable uncertainty, especially since the social dimension could only be assessed with limited data. The findings should therefore be interpreted as an assessment of circular mass management, rather than a full sustainability assessment. Although SCORE effectively structured the comparison, for infrastructure projects, referring to landfill disposal provides a more practical reference than ‘no action,’ as it mirrors actual decision making.
Resilience-Based Optimization of Rehabilitation Planning under Budget Constraints: Developing a Risk-Oriented Decision Model considering Supply Security
(2026) Raible, Marvin
Ensuring a reliable, long-term supply of drinking water is a significant challenge for municipal water utilities, particularly given strict budget constraints and challenging topographic conditions. This thesis aims to develop a transferable, risk-oriented optimization model for rehabilitating municipal water supply networks. The model evaluates renewal needs and financial sustainability. Through lifespan simulations, the model prioritizes specific pipeline sections using a risk-based, multi-criteria framework. This model is based on the concept of localized risk, which is defined as a function of structural failure probability and the resulting socioeconomic and physical consequences. The relevant evaluation criteria, including asset age, material generations, empirical damage rates, operating pressure, and traffic loads, are defined based on the technical guidelines of (DVGW e.V. W 403, 2025) and scientific literature. The Analytic Hierarchy Process (AHP) is systematically applied to determine the objective weighting of these conflicting criteria. Furthermore, this systematic approach establishes a mathematically reliable ratio of conditions to consequences, thereby eliminating qualitative biases. Based on the evaluated criteria, a risk index is derived to enable detailed prioritization of pipeline sections. Additionally, the model incorporates resilience indicators and consumer criticality metrics to account for the impact of potential pipeline failures on populations and infrastructure that are most vulnerable. The developed evaluation model was applied and validated using a real pipeline network in a Stadtwerke Freudenstadt case study. This application demonstrates that combining multi-criteria, risk-based prioritization with strategic rehabilitation planning under fixed financial constraints establishes a consistent, transparent, and structured basis for municipal decision-making. Furthermore, the model transforms local boundary conditions into a methodology that can be applied to other water utilities. The model provides a robust foundation for short-, medium-, and long-term rehabilitation planning.