Mechanics and Maritime Sciences

<h4>Temperature Prediction Evaluation of Electric Heater Performance in Exhaust Aftertreatment Systems</h4>
<p>Examiner: Jonas Sjöblom<br>Supervisor: Jonas Sjöblom, Swathy Suresh<br>Opponent: Abner Lawrence, Simran Joy</p>
<p>Abstract of thesis:<br>Increasingly stringent vehicle emission standards call for improved performance of exhaust aftertreatment systems. A common problem is emissions during cold starts when the exhaust temperature is not sufficient to heat up the catalysts to operating conditions. Traditionally, this challenge is met by supplying additional exhaust heat using engine-based thermal management techniques by injecting additional fuel or decreasing engine efficiency. However, these techniques come with penalties such as increased fuel consumption and noise. An alternate solution is to use an electric heater to heat up the exhaust gases flowing into the catalysts. This can provide a large amount of heat close to the catalyst to enable them to reach operating temperatures quickly.&nbsp;<br>This study investigates the performance of an electric heater installed upstream of the silencer in a 380hp 13-liter heavy-duty diesel engine. Tests in an air flow rig, an engine test cell were complemented with 1-D simulations to evaluate catalyst heating performance, reduction of nitrogen oxides and energy consumed by the heater in steady and transient cycles. A major finding in this study was the effect of flow maldistribution limiting the effectiveness of the heater and the aftertreatment system. 1-D models were found to be useful in testing different test cases but are challenged by flow maldistribution into the aftertreatment system.&nbsp;<br>The results show that a 10kW heater works to improve nitrogen oxide emissions but need to be used in combination with engine modes that have low nitrogen oxide emissions. There is potential for lower emissions and energy consumption with improvements to the heater and urea dosing strategies. The performance of the system was also found to be significantly affected by the installation by heat losses in the exhaust flow.</p>
<p>Date and time: 18h of August 2025, 11:00-12:00<br>Location: Förbränningsrummet</p>

<h4>Temperature Prediction In Car Lamps Using Sun Load Data</h4>
<p>Examiner: Lars Davidson<br>Supervisor: Kevin Ståhl and Britta Mattsson<br>Opponent: Yogesh Peruvemba Krishnan</p>
<p>Date and time: 18h of August 2025, 13:00-14:00<br>Location: M2 Room Navier</p>

<h4>Investigation of Disc Cavity Sealing Efficiency using CFD</h4>
<p>Examiner: Tomas Grönstedt<br>Supervisor: Navid Mikaillian<br>Opponent: Bilal Mohammed Sajjad Siddiqui</p>
<p>Date and time: 28h of August 2025, 13:15-14:15<br>Location: M2 Room Navier</p>

<h4>Exploring Domain-Adapted LLMs for Crash Narrative Information Extraction: From Stated Facts to Inferred Insights</h4>
<p>Examiner: Robert Thomson<br>Supervisor: Jordanka Kovaceva<br>Opponent: Pasquale Bianco</p>
<p>Abstract of thesis<br>Free-text crash narratives recorded in real-world crash databases have been shown to play a significant role in improving traffic safety. But they remain hard to large-scale analysis due to unstructured writing, heterogeneous terminology, and uneven detail. The development of LLMs offers a promising way to automatically extract information from narratives by asking questions. However, crash narratives remain hard for LLMs to analyze because of a lack of traffic safety domain knowledge. Moreover, relying on closed-source Large Language Models (LLMs) through external APIs poses privacy risks for crash data and often underperforms due to limited traffic knowledge. Out of the above concerns, we study whether smaller open-source LLMs can support traffic related information extraction from crash narratives, targeting three challenging objectives: travel direction of the vehicles involved in the crash, identifying the manner of collision, and classifying crash type in multi-vehicle scenarios that require accurate per-vehicle prediction. In the first phase, we addressed a task where vehicle travel directions are explicitly described in text, using a manually labeled subset of the CISS dataset. The goal was to test whether models trained on general corpora could match or surpass domain-specific baselines. Results showed that fine-tuned transformers (e.g., BERT) achieved the highest accuracy, while advanced prompting strategies, especially Chain of Thought, allowed some 8B-scale LLMs to reach about 90%, making them strong alternatives. For the second and third tasks, we focus on LLMs’ ability to extract implicit information from text, which requires deeper reasoning. To bridge domain gaps, we apply Low-Rank Adaptation (LoRA) to inject traffic-specific knowledge. On the CISS dataset, our LoRA-fine-tuned 3B models outperform GPT-4o while requiring minimal training resources. Further analysis of LLM-annotated data shows that LLMs can both compensate for and correct limitations in manual annotations while preserving key distributional characteristics. These results indicate that advanced prompting and fine-tuned open-source models are effective for large-scale traffic safety studies.</p>
<p>Date and time: 19h of September 2025, 15:00-16:00<br>Location: M2 room Newton</p>

Architecture and Civil Engineering

Bachelor’s thesis Chemistry and Chemical Engineering

Chemistry and Chemical Engineering

Communication and Learning in Science

Computer Science and Engineering

Electrical Engineering

Industrial and Materials Science

Life Sciences

Mathematical Sciences

Microtechnology and Nanoscience

Physics

Space, Earth and Environment

Technology Management and Economics

Presentation of master thesis

Choose subject

Course syllabus for

Subject

Each academic year contains different courses

See programme page at universityadmissions.se

Educational area

Permissible substitutions of compulsory courses

To see your programme syllabus – select the academic year in which you started the programme.

[en: page.studyprogrammepage.syllabusesview.coursesdescription]

Presentation of master thesis

Master thesis presentations at Mechanics and Maritime Sciences

18 August 2025

11:00-12:00 Atlin Roy, MPMOB

13:00-14:00 Bilal Mohammed Sajjad Siddiqui, MPAME

28 August 2025

13:15-14:15 Yogesh Peruvemba Krishnan, Applied Mechanics

19 September 2025

15:00-16:00 Xixi Wang and Mattia Carlino