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

Acoustic emission monitoring of blending in continuous direct compression
(2026) Jagadeesan Ashok Kumar, Varshaa
Continuous direct compression (CDC) is increasingly used in pharmaceutical manufacturing for the continuous production of oral solid dosage forms. Within a CDC line, continuous blending is an important processing step because the amount of material retained in the blender affects material transport and residence behaviour. This thesis investigates whether acoustic measurements recorded during continuous blending contain information that can be related to residence mass. Acoustic signals covering different frequency ranges were recorded during six blender experiments performed at different rotational speeds and throughputs. Time-frequency analysis, singular value decomposition, and multivariate modelling were used to characterise the signals and assess the influence of operating conditions. Partial least squares discriminant analysis achieved 97.8% classification accuracy using frequency band features and 100% using a higher-resolution spectral representation. Multioutput partial least squares regression further showed that both rotational speed and throughput were strongly represented in the acoustic measurements. Residence mass was modelled using gravimetric measurements from each experimental run and evaluated using leave-one-run-out validation. High-dimensional spectral models produced strong training fits but poor prediction of omitted runs. Reducing the acoustic feature set and one PLS latent substantially improved cross-run generalisation. The best acoustic-only model achieved a training R2 of 0.940, a LORO R2 of 0.854, and a LORO RMSE of 266 g. Engineered process variables showed strong individual correlations with residence mass but did not improve LORO performance. The modelling procedure was also applied to separately acquired recordings using a different high-frequency measurement setup. Resampling reduced a systematic prediction offset, while a reference gravimetric measurement was required for bias correction. Overall, the results show that acoustic measurements contain information associated with both blender operating conditions and residence mass, while also highlighting the importance of model complexity, run-level validation, block selection, and measurement consistency when applying the approach to continuous pharmaceutical blending.
Innovation inom Glasögonförvaring: En studie i användarbeteenden och teknisk konstruktion
(2026) Sandkvist, Alice; Persson, Moa
More people are in need of glasses, while modern lifestyles place increasing demands on optimal vision, including near vision for mobile device use combined with active daily movement. As a result, the need for practical and reliable eyewear storage solutions has increased. Despite this, many users experience problems with existing solutions, which often leads to damaged glasses or difficulty accessing them when needed. This project focused on developing an innovative storage concept adapted for users who switch between multiple pairs of glasses and require convenient handling while on the move. The project was carried out as a bachelor’s thesis at the Department Industry and Materials Science at Chalmers University of Technology. The aim of the project was to develop a solution that simplifies everyday eyewear storage and thereby extends the lifespan of glasses. The project was commissioned by GOT Design, which also provided expertise in mechanical design and training in the CAD software CATIA V5. The work followed an iterative product development process that began with extensive user studies, including surveys, semi-structured interviews, observations and visits to opticians. To deepen the understanding of user needs, data collection continued throughout the project. The collected insights formed the foundation for the concept development, where both functionality and aesthetics were evaluated with users in several stages. The result was FlexCase, a modular storage solution that combines robust mechanical protection with high flexibility. The concept is characterized by a discreet aesthetic and integrated attachment points that enable quick and easy access for the user in different environments. Through its adaptable design, the solution facilitates eyewear storage for a broad group of eyewear users.
Aerodynamic Evaluation of Intermediate Compressor Ducts with Integrated Heat Exchangers
(2026) Johansson, Alexander
The aviation industry is actively investigating hydrogen as a future aircraft fuel due to its potential to significantly reduce greenhouse gas emissions. One promising application is the use of cryogenic hydrogen for engine heat management through intercooling, where integrated heat exchangers can be placed within intermediate compressor ducts (ICDs) between compressor stages. The aerodynamic performance of such ducts is influenced by complex flow phenomena, including diffuser flow separation, swirl, three-dimensional flow structures, adverse pressure gradients and strong curvature. This thesis investigates the aerodynamic performance of two ICD geometries with integrated heat exchangers using computational fluid dynamics (CFD). The study compares conventional 2D axisymmetric simulations, 2D axisymmetric simulations with swirl modelling, and full 3D simulations in order to evaluate the importance of three-dimensional flow effects. Two diffuser configurations with different lengths and curvatures were analysed using steady-state Reynolds-Averaged Navier-Stokes (RANS) simulations with the k − ω SST turbulence model. The results show that the shorter and more aggressive diffuser geometry produces around 11% higher total pressure losses than the longer geometry. Full 3D simulations consistently predict higher pressure losses than the corresponding 2D models, indicating that three-dimensional flow effects contribute to additional aerodynamic losses that are not fully captured by axisymmetric approaches. Inlet swirl was found to provide an improvement in diffuser pressure recovery, although its influence on overall pressure losses was limited. The integrated heat exchanger was found to have a significant stabilizing effect on the flow by suppressing diffuser separation, allowing aggressive diffuser geometries to achieve a pressure recovery comparable to less aggressive designs. One important observation is that 2D simulations underpredict both the pressure recovery and total pressure losses by up to 2.5% compared to the performed 3D simulations. The findings suggests that two-dimensional CFD models can provide useful preliminary assessments at a low computational cost, while full three-dimensional simulations are recommended when accurate prediction of flow structures and pressure-loss mechanisms is required.
QuadriSparse: RISC-V Sparse Matrix Accelerator and ISA Extension - A Tightly-Coupled Sparse-Dense Matrix Multiplication Accelerator for RISC-V
(2026) Erlandsson, Nik; Swärd, Oskar
Sparse dense matrix multiplication (SpMM) is an important operation in many applications such as inference and training of pruned large language models, graph analytics and scientific computing. These applications often operate on data that is inherently sparse. Applying dense matrix multiplication (GEMM) to sparse data wastes computations on zero-valued elements, which SpMM avoids by skipping them. However, accelerators designed for dense matrix multiplication do not necessarily support sparse matrix multiplication efficiently. This motivates extensions that can exploit sparsity while retaining computation capabilities for dense workloads. This thesis explores the prospect of extending a small dense matrix multiplication accelerator with additional hardware for SpMM, evaluating the the performance benefits against the hardware overhead. We introduce QuadriSparse, a SpMM extension for unstructured sparsity that adds a partly new datapath to the small and efficient RISC-V accelerator Quadrilatero. The accelerator includes three new instructions to load a tile of the sparse matrix (SPLD_W), load a tile of the dense matrix (DLD_W), and multiply the two using Gustavson’s algorithm (SPMAC_W). We evaluate the resulting accelerator in terms of execution time across different sparsity levels and matrix sizes and measure the FPGA resource utilization through synthesis. QuadriSparse achieves up to 6.6x lower execution time than dense execution on Quadrilatero at 99%sparsity and first outperforms the dense baseline at 95% sparsity. FPGA synthesis shows an increase in resource utilization of 4.6% in LUTs and 23% in DSP-blocks relative to the baseline design.
Cybersecurity in Decentralized Machine Learning for Battery Management Systems: Threats, Detection, and Defense
(2026) Afrem, Johny; Haj Ibrahim, Zaid
Federated Learning (FL) enables distributed devices to collaboratively train machine learning models without sharing raw data, making it suitable for Battery Management Systems (BMSs) that estimate battery State of Health (SOH). However, FL remains vulnerable to attacks such as poisoning attacks in which malicious participants manipulate local training or model updates to influence the learned model. This thesis investigates the security of decentralized battery health prediction systems implemented using the FEDn framework and an Adaptive Iterative Clustered Federated Learning (AICFL) architecture. A controlled experimental environment was developed to evaluate the impact of poisoning attacks on both traditional FL and clustered FL. Three attack categories were implemented and analyzed: model poisoning, stealth-oriented poisoning, and targeted backdoor attacks. To support attack analysis, a server-side suspicious-client risk scoring mechanism was developed to identify anomalous client behavior based on model update characteristics collected during training. Experimental results compare the effectiveness of attacks in FL and AICFL environments and examine how clustering influences attack propagation and model robustness. Experimental results show that poisoning attacks can significantly affect model behavior in both FL and AICFL environments. The impact varies across attack types, while the clustered architecture influences how malicious updates propagate through the federation. The proposed risk-scoring mechanism was able to identify clients exhibiting suspicious update patterns during training. The findings provide insights into the security challenges of clustered federated learning systems for battery management applications and contribute practical methods for analyzing malicious behavior in decentralized AI systems.