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

Interlocking CSEB buildings Nepal Performance and failure behaviour investigated by masonry prism and triplet shear testing
(2026) Lilja, Kajsa; Edström, Kalle
Nepal is situated in-between the Indian and Eurasian tectonic plates resulting in a long standing history of seismic activity. The company Build up Nepal, was founded as a response to the major 2015 earthquake in Nepal and works with interlocking Compressed Stabilised Earth Blocks (CSEBs) which are made up from soil, sand, cement and water which are compressed and cured. Build up Nepal works with engaging communities, creating jobs and helping people build their own house as well as providing engineering services. Modern interlocking CSEBs present a viable alternative in reconstructing seismic resilient houses in a economical way, yet its complex structural behaviour lacks universally established building codes. Build up Nepal seeks to make this construction method standardised in Nepal and get further government approval for the system. In order to achieve approvals, more reliant data on the performance of their system is needed. Based on the system implemented by Build up Nepal, this thesis investigates methods of testing and possible improvements for interlocking CSEB masonry. By exploring how testing can be done in a simple way, which parts of the system contribute to overall strength, and what improvements are feasible in rural Nepal, this thesis aims to support Build up Nepal’s goal of standardising this construction method. The standard Determination of initial shear strength was adapted for the equipment available in the workshop connected to Build up Nepal’s office in Kathmandu and a testing plan was developed. Together with experimental testing a field trip to constructions sites in western Nepal was conducted to understand the circumstances and context. During testing, several limitations of the setup were acknowledged. The precision of certain testing equipment was low, indicating that a future step for the company is to move testing to a laboratory environment. A key issue with the test setup was the unintended increase of pre-compression loads; these are supposed to remain constant, which proved impossible with the current equipment. The results from the experimental testing demonstrate that increasing the cement content in grout and adding bed mortar improves overall system strength. However, insights from the site visit show that grout mix might be varying and application of bed mortar potentially is a challenge. The interlocking keys do not only support alignment during mason but also provides a 83% gain in shear strength compared to specimens without the keys. The recommendations are to continue using the interlocking keys and to further investigating bed mortar before widely adopting it in construction. Moreover, the cement to sand ratio of one to three should be used and simple measuring equipment should be provided to ensure consistency in batch size and mix proportions.
Operationalizing Physics-Informed Deep Learning in Fire Safety: A Robust Design and Product Lifecycle Approach
(2026) Jakka, Sai Srinivasa Manideep; Rudesh, Likhith
Measuring fire heat release rate (HRR) is central to fire safety engineering. In tunnels and facades, however, conventional oxygen-consumption calorimetry is often impractical due to geometric constraints, harsh thermal environments, and sensor degradation caused by smoke and flame. Because surveillance cameras are already widely installed in tunnel infrastructure, this thesis develops VisionKoopmanNet, an image-based deep learning pipeline that estimates HRR directly from optical video and embeds these predictions within a structured operational framework for safety-critical evaluation. The technical pipeline is built in three phases. Phase 1 trains a physics-informed Koopman-LSTM HRR encoder on 756 Fire Dynamics Simulator (FDS) scenarios and reaches R2 = 0.9996 on 1.73 million held-out test samples. Phase 2 trains an EfficientNet-B7 image branch on the NIST Fire Calorimetry Database using an experiment-level stratified-by-peak-HRR train/val/test split (391/84/82 experiments) and a proportional frame-to-HRR index mapping that corrects a hardcoded fps=5.0 assumption that had silently misaligned every window with pre-ignition baseline samples in earlier work. Two complementary Phase 2 results are reported. The image-only HRR estimator, representing the operationally relevant capability for tunnel and facade deployment where calorimetry sensors are unavailable, predicts absolute HRR from camera video alone with no sensor input at inference and achieves held-out test MAE = 251 kW on 50,720 windows drawn from 82 stratified-held-out NIST experiments, with sub-150 kW MAE in the 0–200 kW regime (n = 34,177 windows, 67% of the test set) and documented failure modes on rare megawatt-scale fires (n = 361 at >5 MW). The sensor-fused forecaster, which takes the most recent calorimetry reading as an additive anchor in log-target space alongside image features, achieves test MAE = 8.93 kW but matches a multi-step persistence baseline (8.88 kW) within 0.5% across every per-bin slice tested; empirically, the decoder learns Δ = 0 for every input and the prediction collapses to copying the sensor reading. This residualanchor trap survives every architectural protection investigated (log-target regression, hard spectral-radius projection, end-to-end unfrozen encoder, stratified split, activefire weighted sampler, magnitude-weighted MAE) and is reported as a structural information-theoretic finding rather than a tuning failure. Phase 3 attempts to fuse the two encoders into a shared Koopman latent space; three alignment strategies (mean-squared-error on unpaired data, joint retraining, contrastive InfoNCE) all fail to produce per-sample cross-modal correspondence, with three documented failure modes (no improvement, catastrophic forgetting, and representation collapse). The Phase 3 negative result, traced to a data-level cause (namely that the FDS scenarios and NIST experiments represent different physical fires rather than the same fire observed across two modalities), is reported as a primary scientific contribution. v Alongside technical pipeline development, this thesis answers two research questions on safety-critical operationalization. The first investigates how Critical-to-Quality (CTQ) requirements, Operational Design Domains (ODD), and Stage-Gate lifecycle controls can be designed to ensure rigorous specification, workflow compatibility, and robustness for an offline fire prediction prototype. The second investigates how Failure Mode and Effects Analysis (FMEA), safety wrappers, and human-in-the-loop escalation rules can be structured to govern validation and risk management prior to operational deployment. The primary contributions of this thesis are fourfold: (1) A working camera-based image-only HRR estimator (MAE = 251 kW on held-out NIST testing, sub-150 kW for fires up to 200 kW) suitable for tunnel and facade scenarios where physical calorimetry cannot be installed; (2) A structural negative finding demonstrating the residual-anchor trap in sensor-fused HRR forecasting; (3) An empirical investigation into cross-modal latent alignment without paired multi-modal data, identifying three distinct failure modes in cross-modal learning for fire dynamics; and (4) An endto- end operationalization framework combining CTQ metrics, ODD delimitation, Stage-Gate quality gates, FMEA risk prioritization, and runtime safety wrappers to govern bounded prototype safety. A separate methodological contribution is the documentation of a silent hardcoded frame-rate alignment error that previously inflated Phase 2 metric scores, together with the multi-iteration rebuild that corrected it. The scope is strictly bounded to offline analysis of recorded NIST video sequences; real-time edge deployment, multi-camera fusion, hardware procurement, graphical user interfaces, and formal compliance certification remain out of scope for future work.
Reduced-Order Modeling in Nonlinear Driveline Systems
(2026) Enström, William; Johnsson, Alexander
This thesis presents and investigates different approaches for constructing reducedorder BEV driveline models in state-space form using a data-driven approach. Existing Multi-Body Dynamic (MBD) models of the full driveline are often computationally demanding, while reduced models of sub-parts may lack a clear connection to the full-scale model and to one another. In order to improve simulation times while maintaining consistency with the more general MBD model, local Linear Time-Invariant (LTI) models are identified around selected operating points using subspace system identification methods and data generated from the larger MBD model. However, system internal periodic disturbances related to driveline rotation complicate the system identification as it contaminates the externally forced response. Two methods addressing this are investigated: Removing the frequency components associated with these disturbances from the data, or augmenting the model with additional input signals with those frequencies. Both methods improve the model quality by reducing the effect of the periodic disturbances. The model constructed by including additional input signals is also shown to more accurately reproduce the amplitude and frequency of the periodic disturbance. In order to model gyroscopic effects, such as resonance frequencies depending on the operating speed, several local LTI models are identified at different operating points. The local models are combined by either linear interpolation of the matrix elements, fitting a basis function to the matrix elements or by interpolating the outputs of each model. All three methods result in models with similar input-output accuracy, but differ in their requirements when creating the local models. In particular, methods based on output interpolation do not require system internal consistency between local models, unlike methods that rely on the local models sharing a common state-space realization. The disturbance-modeling approach based on additional input signals is also extended to the Linear Parametric Variant (LPV) framework, enabling the resulting models to capture both the speed-dependent system dynamics and the periodic disturbances associated with driveline rotation.
Integrating Highly Conductive 2D Materials in Metal Matrix
(2026) Berntsson, Sanna
MXenes have attracted significant attention in recent due to their high electrical conductivity and potential for usage in composite material. This thesis investigated Ti3C2Tx MXene/Al composites with varying aluminum content to see whether the conductivity of the composites can overcome the one for pure aluminum. The MX ene was synthesized through an in-situ HF etching process, mixed with aluminium powder, pressed into coin-like pellets, and sintered in argon atmosphere at 550 ◦C. Contact conductivity and Rockwell hardness measurements where performed, as well as microstructure analysis through SEM and EDX. The results showed that high-Al composites showed measurable conductivity and im proved mechanical integrity, while lower-Al composites exhibited fragile behaviour and measurement challenges. Sintering influenced both conductivity and hardness, indicating strong dependence on densification and microstructure. This was sup ported by the microstructure analysis. The study demonstrates the importance of composition and thermal processing in determining the electrical and mechanical properties of MXene/Al composites. None of the composites overcame the electrical performance for pure aluminium, however there are many aspects of this project that could be investigated further before declaring MXene/Al composites not useful for electrical transport applications. These aspects include examining the residual aluminium in the pure MXene samples and comparing the results of other sinter ing methods better suited for powder composites, such as spark plasma sintering (SPS).
Towards Qualification of Female Human Body Model for Virtual Testing: Setup for 50th percentile female human body model and parametric study of the boundary conditions
(2026) Dhamane, Pushkar Mahesh
Road traffic injuries pose a major global safety challenge for vehicle occupants and vulnerable road users, such as pedestrians and cyclists. To prevent injuries, crash testing plays a vital role in automotive safety assessment. While early biomechanical evaluations relied on post-mortem human surrogates (PMHS), standard physical crash testing today is primarily conducted using Anthropomorphic Test Devices (ATDs). However, ATDs possess structural limitations: their rigid responses cannot fully replicate complex human soft-tissue and skeletal injury mechanisms. Additionally, physical testing using ATDs is constrained by high financial costs and testing complexity. To address this limitation, virtual testing is being proposed as a complement to physical testing for vehicle safety assessment, for example the HBM4VT framework under Euro NCAP has established qualification protocols for the 50th percentile male HBM (50M) through the Crash Protection 550 (CP550) protocols. This thesis evaluates the biofidelity scoring of a 50th percentile female HBM (50F) across key hub impact scenarios: Kroell (frontal thoracic), Viano (thoracic, abdominal, and pelvic), Compigne (shoulder), and Agnew (angled thoracic) and performs a parametric study on uncertainties that can influence the biofidelity scores. Simulation results were compared against experimental post-mortem human subject (PMHS) reference corridors generated using the ARCGen, with objective metric scores calculated using the Ellipse and Dynamic arc-length warping (DALW) scoring method and, ISO 18571. Evaluated parameters included physical setup variables (seating angle, HBM posture, and arm orientation) and ARCGen algorithmic parameters (cut_simulation with boolean values and number of warping points (nWarpCtrlPts) ranging from 0 to 5. Results demonstrate that scores exhibit higher sensitivity to physical setup variations than corridor algorithm configurations.