A prediction model [3] of macrosegregation in continuously cast steel slabs developed by Barber et al from Corus Teesside Technology Centre, UK as an extension of the work of Miyazawa and Schwerdtfeger [4] is presented in this thesis. Improvements have been made in the model to include the effect of many bulges between the supporting rolls induced by the ferrostatic pressure exerted on the solidifying shell by the liquid phase, which were approximated following a deflection of an elastic beam model. The system involved three alloying elements that are 0.1 wt% C, 1.05 wt% Mn, and 0.25 wt% Si as initial concentrations. The iterative numerical methods used to solve for velocity and concentration were the Gauss-Seidel method and Upwind Differencing, respectively.Through a mesh sensitivity analysis, a horizontal growth rate of 1.25 was decided while the grid is equally spaced in the vertical direction over a two-dimensional mid-width plane of the CC22 slab caster configuration at Tata Steel Netherlands in Ijmuiden. Good results were obtained from a validation run despite the assumptions considered. The predicted macrosegregation patterns for C and Mn were to a great extent compatible with the measured ones. The maximum concentrations were predicted at the strand centerline that are 0.13 wt% C, 1.10 wt% Mn, and 0.26 wt% Si. With respect to measured values, it was slightly higher for C by 2.23% error but lower for Mn at 2.28%error, which could be seen negligible. The main factors concluded to influence the strand centerline macrosegregation were initial melt temperature, casting speed, cooling rate and supporting-rolls location when neglecting the mutual interactions between elements. Further improvements can include modifications of the assumptions followed in this model.

No description available.

Power Transformers are an important and complex part of the electrical transmission technology. One of the main producers of these transformers worldwide is Hitachi Energy, for which this work has been done.This work proposes to create a simulation model for the transport of moisture in transformer insulation material using Finite Element Method implemented in COMSOL 6.1. The transport of moisture in cellulose material is a complex process but the model assumes that diffusion is the dominant transport mechanism. For that, a model based on Fick’s second law is used. The value of the coefficient is defined experimentally at different temperatures. But some of the tools used for their measurements are inaccurate, for example at higher temperatures or because they only give an average moisture content at a given time. Then a difficulty is to assert the simulation’s validity by comparison with the experimental data. In the present work, care is given to validate the model by comparison with precedent works.Using experimental data coming from internal sources at the company (Hitachi Energy), curves are then drawn to see the process of ingress and egress of moisture in pressboard material under different boundary conditions to simulate several steps of the production process. Possible future research directions include the use of data from future experiments and the creation of a more complex model to simulate a more realistic geometry.

The large environmental impact related to the use of fossil fuel has driven the shift toward renewable sourced alternatives. Fossil diesel can be nowadays replaced by biodiesel, obtained from vegetable oils and fats, mostly used as biodiesel blends. However, some drawbacks are related to the use of this bio-fuel, among which the formation of deposit in injectors and filters causing a reduction of engine performances or engine failure. This thesis project focuses on the analysis of the mechanism of formation of soft particles in biodiesel deposits. These particles are constituted mostly of carboxylic metal soaps, and were found in biodiesel engines after using aged biofuel with contaminants, such as engine oil. The role of short chain fatty acids (SCFAs) has been investigated, together with the use of three different calcium sources, to analyse the formation mechanism of calcium soaps. Artificial ageing of B10 and B100 test fuels was performed, and in some cases an inert gas was bubbled to remove the formed SCFAs. Calcium sources, namely calcium oxide, calcium carbonate and engine oil, were added to investigatethe formation of soft particles. Ion chromatography, pH measurements, NMR spectroscopy and oxidation stability tests have been performed on the liquid test fuel to verify the presence and effect of SCFAs, while FTIR spectroscopy and GC/MS analyses were used to verify the presence of calcium soaps in the deposit and in solution. In contrast to the expectation, it was found that the presence of SCFAs in the fuel not is fundamental for the formation of carboxylic soap. Moreover, the different calcium sources result in different amounts and textures of metal soaps in deposits and in solution. / Den stora miljöpåverkan relaterad till användningen av fossilt bränsle har drivit övergången till förnybara alternativ. Fossil diesel kan numera ersättas av biodiesel, härledd av vegetabiliska oljor och fetter, oftast används som biodieselblandningar. Vissa nackdelar är emellertid relaterade till användningen av detta biobränsle, bland annat bildandet av avlagring i injektorer och filter som orsakar en minskning av motorprestanda eller motorfel. Detta projekt fokuserar på analysen av mekanismen för bildning av mjuka partiklar i biodieselavlagring. Dessa partiklar består huvudsakligen av karboxylmetall tvålar och hittades i biodieselmotorer efter användning av åldrad biobränsle med föroreningar, såsom motorolja. Rollen för kortkedjiga fettsyror (SCFA) har undersökts, tillsammans med användning av tre olika kalciumkomponenter, för att analysera bildningsmekanismen för kalciumtvålar. Experimentellt åldring av testbränslen B10 och B100 på labbet utfördes, och i vissa fall bubblades en inert gas för att avlägsna de bildade SCFA. Kalcium komponenter, nämligen kalciumoxid, kalciumkarbonat och motorolja, tillsattes för att undersöka bildningen av mjuka partiklar. Jonkromatografi, pH-mätningar, NMR- spektroskopi och oxidations stabilitetstester har utförts på det flytande testbränslet för att verifiera närvaron och effekten av SCFA, medan FTIR-spektroskopi och GC/MS-analyser användes för att verifiera närvaron av kalciumtvålar i sedimentet och i lösning. I motsats till förväntningarna visades det att närvaron av SCFA i bränslet inte är grundläggande för bildandet av karboxyltvål. Dessutom resulterar de olika kalciumkomponenterna i olika mängder och strukturer av metalltvålar i sediment och i lösning.

Lignin has the potential to be used as green material to inhibit the corrosion of metallic substrates. Mussel adhesive protein is used to modify lignin due to its great adhesive and film forming abilities. Electrochemical impedance spectroscopy (EIS) has been applied to in-situ measure the corrosion resistance of the formed surface composite films in the corrosive environment. The equivalent circuit is used to fit the EIS data to obtain the quantitative results of the surface films. The results show that MAP modified lignin composite film can provide enhanced corrosion protection to the carbon steel substrate and presents self-healing property. / Lignin har potential att användas som grönt material för att hämma korrosion av metalliska underlag. Mussellhäftande protein används för att modifiera lignin på grund av dess stora lim- och filmbildningsförmåga. Elektrokemisk impedansspektroskopi (EIS) har applicerats för att mäta korrosionsbeständigheten hos de bildade ytkompositfilmerna i den korrosiva miljön. Ekvivalentkretsen används för att passa EIS-data för att erhålla kvantitativa resultat från ytfilmerna. Resultaten visar att MAP-modifierad ligninkompositfilm kan ge förbättrat korrosionsskydd för kolfastsubstratet och uppvisar självhelande egenskaper.

Techniques to improve the speed at which materials are researched and developed has been conducted by investigating the machine learning methodology. These techniques offer solutions to connect the length scales of material prop- erties from atomistic and chemical features using materials databases generated from collected data. In this assessment, two material informatics methodologies are used to predict material properties in steels and nickel based superalloys using this approach. Martensite start temperature and sigma phase amount as a function of input composition has been modelled with the use of machine learning algorithms. The experimental methodology had a collection of over 2000 unique experimental martensite start temperature points. This yielded important information on higher order interactions for the martensite start temperature, and a root mean square error (rmse) of 29 Kelvin using ensemble tree based algorithms. The metamodel was designed using an artificial neural network from TensorFlow’s library to predict sigma phase fraction and its composition. The methodology for building, calculating, and using data from TC-Python will be laid out. This generates a model that would generalize sigma phase fraction 97.9 % of Thermo-Calc’s equilibrium model in 7.1 seconds compared to 227 hours neded in the simulation to calculate the same amount of material property data. / Tekniker för att förbättra hastigheten med material som forskas och utvecklas har genomförts genom att undersöka metodik för maskininlärning. Dessa tekniker erbjuder lösningar för att ansluta längdskalorna för materialegenskaper från atomistiska och kemiska egenskaper med hjälp av materialdatabaser genererade från insamlade data. I denna bedömning används två materialinformatikmetoder för att förutsäga materialegenskaper i stål och nickelbaserade superlegeringar med denna metod. Martensite-starttemperatur och sigmafasmängd som en funktion av ingångssammansättningen har modellerats med användning av maskininlärningsalgoritmer. Den experimentella metoden hade en samling av över 2000 unika experimentella starttemperaturpunkter för martensit. Detta gav viktig information om interaktioner med högre ordning för martensit-starttemperaturen och ett root-medelvärde-kvadratfel (rmse) på 29 Kelvin med användning av ensemble-trädbaserade algoritmer. Metamodellen designades med hjälp av ett artificiellt neuralt nätverk från TensorFlows bibliotek för att förutsäga sigma-fasfraktion och dess sammansättning. Metoden för att bygga, beräkna och använda data från TC-Python kommer att anges. Detta genererar en modell som skulle generalisera sigma-fasfraktion 97,9 % av Thermo-Calcs jämviktsmodell på 7,1 sekunder jämfört med 227 timmar som behövs i simuleringen för att beräkna samma mängd materialegenskapsdata.

<p>Confidential</p>

Swedish Nuclear Fuel and Waste Management Company, SKB, main mission is to manage and take care of the Swedish nuclear waste. The method chosen is to store the waste far down in the ground in capsules made of copper, which needs to be intact for 100 thousands years. Friction stir welding is used to seal the capsules, which causes oxidation due to heat, but is avoided by using gas protection. If this protection fails, SKB needs to be able to measure the thickness of the oxide to be able to determine if there is too much oxygen in the capsules, which can cause problems in the long run by making the capsules disintegrate and leaking nuclear waste. In this work different methods that SKB can use to measure the thickness of the copperoxide are examined. To find a method for thickness measurements several methods were examined through literature (Spectral-reflectance, Terahertz Time-Domain Spectroscopy, Laser absorption and usage of color). One of these was tested in practice, which was spectral-reflectance. Reference samples were manufactured with known oxide thickness to be used for callibration and teseting of the chosen instrument. The spectral-reflectance measurements were performed on the samples from the work which gave results that agreed with ellipsometry measurements that were made to verify the thickness of the oxides. Since spectral-reflectance worked on the samples and it is a non-contact and non-destructive method, it was the chosen method to be suggested to SKB.

This work investigates the development in microstructure and mechanical properties of roll leveled martensitic high strength structural steel by examinations of roll leveled strip sample and by the employment of roll leveling simulations. The aim of the thesis work is to generate inputs towards finite element method simulations and is performed at a cut to length line utilizing a pre-leveler and a roll leveler at SSAB. Roll leveled strip sample is extracted from a halted roll leveler with 13 roller setup. Mechanical properties are examined on the macroscopic scale by tensile testing and with Instrumented indentation testing on the microscopic scale. Microstructure examinations are conducted with electron backscatter diffraction and roll leveling simulations are performed with a micro bending machine. Tensile testing indicates a noticeable reduction in yield strength from pre-leveled strip portions and a negligible reduction in yield strength from roll leveling on a macroscopic scale in comparison to virgin strip material. Hardness testing in the thickness direction shows decreasing hardness on a microscopic scale. Microstructural examinations reveal a typical texture effect that can be related to hot rolling effect which remains unaffected even after roll leveling. Constant grain size in the top and bottom layers but larger grain size in the mid layer, and a shift in crystallographic orientation on the cross section is revealed by inverse pole figures. Roll leveling simulations with the bending machine on virgin and pre-leveled samples shows the Bauschinger effect as a result of alternating bending tests. Euler angle maps on these bend samples reveal that a large number of grains rotate at a certain degree with the first bend and grains retaining and maintaining their prior orientations with consecutive bends. There is an approximately 1:1 ratio in Euler angle change found for tensile and compression deformation modes. Generation of dislocations during bending simulations is found to differ between tensile and compressive deformation modes. Hardness examination on post bending samples is seen to decrease as a result of the alternating bending process and is discussed to be related to grain relaxation. Several future studies can be possible from the findings of this work which is proposed and discussed as further work. / Detta arbete undersöker utvecklingen i mikrostruktur och mekaniska egenskaper hos rullriktat martensitiskt höghållfast konstruktionsstål genom undersökning av rullriktat bandprov och genom rullriktningsimulering. Syftet med arbetet är att generera input till materialmodeller för finita element-simuleringar och utförs vid formatsträcka 4 hos SSAB i Borlänge. I detta arbete undersöks oriktat, förriktat och riktat band. Ett rullriktat prov extraheras från ett stoppat riktverk med 13 valsar. Mekaniska egenskaper på makronivå undersöks genom dragprovning och på mikronivå med nano-hårdhetsprovning. Mikrostrukturen undersöks med EBSD och rullriktningsimulering utförs med mikrobockningsmaskin. Dragprovning indikerar en märkbar minskning i sträckgräns på förriktade band och försumbar minskning i sträckgränsen på riktade band. Nano-hårdhetsprovning i tvärsnittet visar minskande hårdhet i mikroskopisk skala mellan oriktade, förriktade och riktade band. Mikrostrukturundersökningar avslöjar en varmvalsningseffekt som förblir opåverkad av rullriktningsprocessen och inkluderar en konstant kornstorlek i topp- och bottenlager men större kornstorlek i mellanskiktet samt förskjutning ikristallografisk orientering i bandens tvärsnitt. Rullriktningsimulering med bockningsmaskin på oriktade och förriktade böjprover visar Bauschinger-effekt som ett resultat av alternerande böjning. Euler-vinkelkartor på dessa böjprover avslöjar att en stor andel korn roterar en viss vinkel samt att korn återfår sin ursprungliga vinkel under efterföljande böjningar. Det råder ett nära 1: 1-förhållande i Euler-vinklar för drag- och kompression under bockning medan mängden felorienteringar visar sig skilja sig mellan drag- och kompressionsdeformationslägen. Generering av dislokationer under rullriktningssimulering visar sig skilja mellan drag och kompressionsdeformationslägen. Hårdhet på simuleringsprover ses minska som ett resultat av kontinuerlig bockningsprocess och föreslås uppstå som följd av en avslappningseffekt i korn. Flera framtida studier är möjliga från dessa fynd vilket diskuteras och föreslås som ytterligare arbete.

No description available.