Publikationer från KTH
Not a member yet
57933 research outputs found
Sort by
Regional climate projections using a deep-learning–based model-ranking and downscaling framework: application to European climate zones
No full text Accurate regional climate projection calls for high-resolution downscaling of Global Climate Models (GCMs). This work presents a deep-learning-based multi-model evaluation and downscaling framework ranking 32 Coupled Model Intercomparison Project Phase 6 (CMIP6) models using a Deep Learning-TOPSIS (DL-TOPSIS) mechanism and refines outputs using advanced deep-learning models. Using nine performance criteria, five Köppen-Geiger climate zones—Tropical, Arid, Temperate, Continental, and Polar—are investigated over four seasons. While TaiESM1 and CMCC-CM2-SR5 show notable biases, ranking results show that NorESM2-LM, GISS-E2-1-G, and HadGEM3-GC31-LL outperform other models. Four models contribute to downscaling the top-ranked GCMs to 0.1o resolution (Vision Transformer (ViT), Geospatial Spatiotemporal Transformer with Attention and Imbalance-Aware Network (GeoSTANet), CNN-LSTM, CNN-Long Short-Term Memory (ConvLSTM)). Effectively capturing temperature extremes (TXx, TNn), GeoSTANet achieves the highest accuracy (Root Mean Square Error (RMSE) = 1.57oC, Kling-Gupta Efficiency (KGE) = 0.89, Nash-Sutcliffe Efficiency (NSE) = 0.85, Correlation (r) = 0.92), so reducing RMSE by 20% over ConvLSTM. CNN-LSTM and ConvLSTM do well in Continental and Temperate zones; ViT finds fine-scale temperature fluctuations difficult. These results confirm that multi-criteria ranking improves GCM selection for regional climate studies and transformer-based downscaling exceeds conventional deep-learning methods. This framework offers a scalable method to enhance high-resolution climate projections, benefiting impact assessments and adaptation plans. The present study has the following limitations, which will be addressed in future works: (i) temperature-only focus, (ii) unquantified scenario uncertainty, and (iii) higher computational cost of transformer models.QC 20250825E-CONTRAI
Improving Circularity via Chemical Recycling to all Rings
No full textAliphatic polyesters synthesized via ring-opening polymerization (ROP) have properties competitive to incumbent plastic (PE, PP), while simultaneously opening up for chemical recycling to monomer (CRM). However, not all aliphatic polyesters are prone to undergo CRM, and the ability to shift the equilibrium between polymer and monomer is tightly associated with the initial monomer structure. The standard strategy to measure CRM is to evaluate the change in free energy during polymerization (triangle GROP). However, triangle GROP is only one-dimensional by assessing the equilibrium between initial monomer and polymer. But under active catalytic conditions, the depolymerization of polymers can lead to formation of larger rings, such as dimers, trimers, tetramers, and so on, via the ring-chain equilibrium (RCE), meaning that the real thermodynamic recycling landscape is multi-dimensional. This work introduces a multi-dimensional chemical recycling to all rings (CRR) via a highly active catalytic system to reach RCE. Thermodynamically triangle GRCE is completely different from triangle GROP. Using triangle GRCE instead of triangle GROP allows us to achieve CRR for polymers notoriously difficult to achieve CRM for, as exemplified within by CRR for poly(epsilon-caprolactone), poly(pentadecalactone), and mixed polymer systems. Overall, this work provides a new general concept of closing the material loop.QC 20250428</p
Modelleringsstudier av diffusion av gasformiga fissionsprodukter i UN
No full textUranium Nitride (UN) has attracted considerable interest as a candidate for advanced nuclear fuels due to its high fissile density, elevated thermal conductivity, and improved performance over Uranium Dioxide (UO2). Despite these advantages, its behavior under irradiation is not yet fully understood, particularly concerning defect energetics and the retention and release of volatile fission products such as Xenon (Xe) in the presence of Oxygen (O), resulting from the oxidation of UN.Using Density Functional Theory (DFT), later improved with DFT+U parameters, within the VASP framework, a UN supercell was built in order to evaluate defect formation energies, solute-vacancy binding, and migration barriers via the Climbing Image Nudged Elastic Band (CINEB) method.The results show that DFT+U provides a more accurate description of the magnetic state of UN supercells and yields formation energies of uranium and nitrogen vacancies (5–6 eV and 4–5 eV, respectively) in closer agreement with literature values than standard DFT.Incorporation energies show that Xe is energetically costly to insert into the lattice, while O incorporates spontaneously with a strong driving force.Binding energies reveal the largest discrepancies between DFT and DFT+U, reflecting their different ground-state energetics. Migration energy barriers, obtained with CINEB, demonstrate that oxygen in the vicinity of the migration path increases the barrier for both Vacancy (V) and xenon migration. In particular, vacancy–solute exchange energies for Xe–V were found to be significantly lower than those for U–V, suggesting that Xe–V exchanges can occur frequently, whereas U–V jumps, though less frequent, are the key contributors to long-range diffusion.Uraniumnitrid (UN) har väckt stort intresse som ett alternativt avancerat kärnbränsle på grund av dess höga klyvbarhet, goda värmeledningsförmåga och förbättrade prestanda jämfört med urandioxid (UO2). Trots dessa fördelar är dess beteende under bestrålningännu inte fullt ut förstått, särskilt vad gäller defektenergier samt retention och frisättning avflyktiga fissionsprodukter såsom xenon (Xe) i närvaro av syre (O), som uppstår vid oxidation av UN.Med hjälp av densitetsfunktionalteori (DFT), senare förbättrad med DFT+U-parametrar inom VASP-ramverket, byggdes en UN-supercell för att utvärdera defektbildningsenergier, lösningsmedel-vakansbindning och migrationsbarriärer via Climbing Image Nudged Elastic Band (CI-NEB)-metoden.Resultaten visar att DFT+U ger en mer korrekt beskrivning av det magnetiska tillståndet i UN-superceller och ger bildningsenergier för uran- respektive kvävevakanser (5–6 eV respektive 4–5 eV) som stämmer bättre överens med litteraturvärden än standard-DFT. Inkorporeringsen- ergierna visar att det är energimässigt ogynnsamt att infoga Xe i gitterstrukturen, medan O inkorporeras spontant med en stark drivkraft.Bindningsenergierna avslöjar de största skillnaderna mellan DFT och DFT+U, vilket återspeglar deras olika grundtillståndsenergier. Migrationsenergibarriärer, erhållna med CI-NEB, visar att syre i närheten av migrationsvägen höjer barriären för både vakans- och xenonmigration. Särskilt visade sig utbytesenergierna mellan vakanser och lösta ämnen vara betydligt lägre för Xe–V än för U–V, vilket indikerar att Xe–V-utbyten kan förekomma ofta, medan U–V-hopp, även om de är mindre frekventa, är de viktigaste bidragande faktorerna till långväga diffusion
Further Development and Testing of a Concentrator for Studies on the Health Impacts of Nanoparticles
No full textSom en del av ett EU-finansierat projekt, nPETS (GA 954377), bedrivs vid Stockholms universitet forskning på hur nanopartiklar från transportsektorn påverkar lungceller. Men eftersom koncentrationen av dessa nanopartiklar normalt är för låg och för att kunna studera effekterna krävs en ökad partikelkoncentration, vilket görs via en koncentrator baserad på ett tillväxtrör, med tre temperaturzoner, och en virtuell impaktor. En sådan koncentrator har tidigare utvecklats av studenter vid KTH, men denna var vid projektets start inte verifierad och krävde vidare utveckling. Denna rapport sammanfattar tidigare gjorda val, beskriver arbetsprocessen för att identifiera problem och de åtgärder som gjorts för att uppnå en fungerande koncentrator med koncentrationsfaktor på minst fem (5x), samtidigt som den ställda kravspecifikationen efterföljdes. Rapporten beskriver hur man kan säkerställa en fungerande kylning för tillväxtröret och beskriver efterföljande tester som skedde på respektive delsystem och koncentratorn som helhet. Vid testningen av delsystemen användes en nebulisator för att generera partiklar och en Optical Particle Counter (OPC) för att räkna dessa, medan helsystemstestet använde aerosoler från omgivande luft. Testerna av delsystemen visade impaktorns effektiva verkningsområde som 0.25-4 µm och tillväxtrörets maximala förstoringsökning, med temperaturerna 5 °C, 45 °C och 15 °C, vid intervallet > 0.35 till > 0.4 µm. Vid den slutliga verifieringen av hela systemet kunde den totala koncentrationsfaktorn uppmätas till 1.98 ± 0.115, vilket innebär att den efterfrågade koncentrationsfaktorn på 5x inte uppnås med rådande inställningar. Under projektet bestämdes inte heller den färdiga koncentratorns verkliga verkningsområde, och man vet därför inte vilken effekt koncentratorn har på de eftersökta nanopartiklarna.As part of an EU-financed project, nPETS (GA 954377), research is carried out at Stockholm University on how nanoparticles from the transport sector affect lung cells. Because the concentration of those nanoparticles is usually too low to study, a higher concentration is needed. The higher concentration is attained through a concentrator, based on a growth tube with three different temperature zones, and a virtual impactor. Such a concentrator was developed by students at KTH but was unverified at the start of this project and required further development. This report summarizes previous choices and describes the work processes employed to identify and resolve issues, aiming to achieve a functional concentrator with a concentration factor of at least five (5x), while adhering to the requirements specification. The report describes how you could ensure a working cooling system for the growth tube, and describes the following tests for each subsystem, and the final end-to-end test. When testing the subsystems, a nebulizer was used to generate particles and an Optical Particle Counter (OPC) to count said particle, while the final test used the surrounding aerosols when conducted. The tests on the subsystems showed an effective range of the virtual impactor, between 0.25-0.4 µm, and the maximal magnification increase for the growth tube, with the temperatures 5 °C, 45 °C, and 15 °C, for the interval >0.35 to >0.4 µm. At the final verification of the whole system the concentration factor was measured to be 1.98 ± 0.115, which mean that it did not fulfil the requested concentration factor of 5x with the current settings. During the project, the working range of the final concentrator was not determined, and therefor, one does not know what effect the concentrator has on the sought nanoparticles
Optimization of Assembly Flow and Design as a foundation for Future Automation
No full textDetta examensarbete genomfördes under våren 2025 i samarbete med start-up-företaget Currentium. Arbetet fokuserade på att förbättra monteringsprocessen för företagets nya powerbankmodell, Currentium Genesis. Produkten har utvecklats för att möta ett växande behov av bärbar energiförsörjning, både i friluftssammanhang och i samhällen med begränsad tillgång till elektricitet. I nuläget monteras Currentium Genesis helt manuellt, med en cykeltid på cirka 90 minuter per enhet. För att möjliggöra en framtida skalbar produktion och kortare monteringstider undersöktes hur produktionsflödet kan optimeras och vilka moment som lämpar sig för automatisering. Arbetet baserades på processkartläggning, Lean-principer och Design for Assembly & Manufacturing (DFAM), samt stödmetoder som DYNAMO och USA-metoden. Fyra förbättringsområden identifierades: strukturerad omarbetning av monteringsflödet, designjusteringar av komponenter, införande av monteringsjiggar, samt val av lämplig automatiseringslösning för lödningsmomentet. Resultatet blev ett förenklat och effektiviserat monteringsflöde som reducerar antalet steg och förbereder processen för automatisering. Specifikt identifierades lödningen som en flaskhals där automation ger störst potential. Genom att ta fram nya monteringshjälpmedel och anpassa produktdesignen skapades också bättre förutsättningar för standardiserat arbete enligt 5S-principerna. Sammanfattningsvis har examensarbetet lagt grunden för en framtida uppskalning av Currentiums produktion, med förkortade monteringstider och ökad möjlighet till automation.This thesis project was carried out in the spring of 2025 in collaboration with the start-up company Currentium. The work focused on improving the assembly process for the company's new power bank model, Currentium Genesis. The product has been developed to meet a growing need for portable energy supply, both in outdoor contexts and in communities with limited access to electricity. Currently, the Currentium Genesis is assembled entirely through manual means, with a cycle time of approximately 90 minutes per unit. To enable future scalable production and shorter assembly times, it was investigated how the production flow can be optimized, and which steps are suitable for automation. The work was based on process mapping, Lean principles and Design for Assembly & Manufacturing (DFAM), as well as supporting methods such as DYNAMO and the USA method. Four key areas of improvement were identified: structured reworking of the assembly flow, design adjustments of components, introduction of assembly jigs, and selection of a suitable automation solution for the soldering step. The result was a simplified and streamlined assembly flow that reduces the number of steps and prepares the process for automation. Specifically, soldering was identified as a bottleneck where automation offers the greatest potential. By developing new assembly aids and adapting the product design, better conditions were also created for standardized work according to the 5S principles. In conclusion, the thesis has established a foundation for future scaling of Currentium's production, with shortened assembly times and greater possibilities for automation
T in Forest
No full textThis project deals with the never opened subway station Kymlinge on the blue subway line number eleven, situated between Hallonbergen and Kista in the northwest part of Stockholm municipality. The station is located on the border of an area of non-developed land called Norra Kymlinge and the nature reserve Igelbäcken. Nothing has been built in the area altough the exploitation of the greater area of Järva fältet, where Kymlinge is situated, was a part of the Swedish Million Program. A rich history of political discord has kept the area in an ”in-between”, non-developed state. The project presented is a design proposal for a final opening of the Kymlinge station. But not as a direct instigator for further development in the area, but rather the opposite. By building sparse on existing structures and serving existing activities in the nature reserve it is a proposal for a subway station in the middle of the forest. With the forest being kept as it is. The station is a seed of concrete that never grew out of the ground. The result of an idea that never flourished and bloomed. Here the seed is cracked open
The Tutte Polynomial of a Graph: With Applications to Knot Theory and Matroid Polytopes
No full textThe aim of this thesis is to explore the Tutte polynomial and how it encodes combinatorial, topological and geometric information through its connections to graphs, knots and matroids. The thesis begins by outlining essential graph theory in Section 2 and continues to define the Tutte polynomial in Section 3. The section continues with some of its properties such as multiplicativity and duality. Then the universality of the Tutte polynomial among deletion-contraction invariants is illustrated via examples. In Section 4 an introduction is given to knot theory and how it is connected to the Jones polynomial, another deletion-contraction invariant. Here concepts as knot invariants and knot equivalence are discussed. Section 5 explores matroid theory and shows an inequality in the Tutte polynomial on the weak -map from a graphic matroid polytope to one of its faces
Extending Loftgångshus - A Study of Deck Access Housing in Sweden
No full textThis thesis explores how the typology of loftgångshus can be extended, both in concept and in practice. While often caught between utopian ideals and stereotypical defamation, this project moves beyond visions and stigma to study how the typology functions in action. Combining theoretical research with design investigation, the thesis reconsiders the spatial and social potential of the loftgång, identifying scale, threshold, and resident ownership as key factors in the quality of shared access space. These themes are tested through a design proposal in Hjulsta, Stockholm, addressing contemporary housing challenges in Million Program suburbs. By extending an existing loftgångshus at Edingekroken 15, a new experimental strategy for densification is proposed—one that allows new and old spaces to coexist within the same property. Income from added apartments can help fund renovations of the existing building, offering a model of co-financing that benefits current residents. The proposal treats loftgångshus as a social, economic, and structural system. Though situated in Hjulsta, the strategies explored are applicable across Stockholm. Rather than repeating past mistakes or discarding the typology, this project proposes building on its existing logic—by extending the understanding, lifetime, and space of loftgångshus
Cliques and Clique Relaxations in the Synaptic Network of C. Elegans
No full textThis project explores the problem of identifying cohesive subgroups in networks by leveraging graph motifs such as cliques, s-plexes, s-cliques and s-clubs. Each motif is paired with three objective functions, maximizing either cardinality, degree or edge weight. Additionally, we formulate a fourth, fractional, objective and linearize this. The resulting optimization problems are solved with mixed integer linear programming on a real-life network, the synaptic network of the nematode worm C. elegans. This analysis uncovers meaningful properties of the network, such as the role of different cells and types of neurons. It also highlights key differences in the motifs and objective functions themselves, such as trade-offs between subgroup size and internal connectivity. The framework for analysis developed in this project provides a powerful tool for exploring cohesion across the field of network analysis, with potential applications beyond the context of neural network
Pursuit Evasion Strategies in Complex Dynamic Terrains Using Optimisation and Game Theory
No full textThis paper investigates the validity of combining Game Theory with optimisation to solve a pursuit evasion problem on a complex dynamic terrain. By giving the players different strategies the impact of these strategies are analysed when they are weighed differently. The paper simulates one pursuer and one evader and focuses on minimising the capture time of the evader by optimising the pursuers strategy. The results showed that letting different strategies dominate depending on the environment yields the shortest capture times. Concluding that Game Theory can be used to aid in solving a pursuit evasion problem