Today, design of wind energy support structures is to a large extent a manual process. It requires a lot of experience, and the design tools are often based on simplified methods. As larger structures are being developed and installations move to larger water-depths, the need for efficient and accurate design tools increases. Simulation-based design is a promising technique that can help automate this process. In this study, Spall’s simultaneous perturbation stochastic approximation (SPSA) method was implemented to automatically optimize thickness and diameter of the members in offshore lattice tower support structures. The method utilizes a pseudo-gradient based on only two function evaluations per iteration, which allows for a computationally efficient process. Each evaluation of the design consists of time-domain simulations of the complete wind turbine in FEDEM Windpower, subsequent rainflow counting and calculation of joint lifetimes with stress concentration factors. The utilization of both ultimate and fatigue limit states is reported for each joint. Tower weight was chosen as an indicator of cost, and an objective function comprising variables for weight and joint lifetimes was defined. Joint lifetime was ignored whenever its value was above the design lifetime of the tower, allowing the algorithm to search solely for the lightest design, as long as the design lifetime constraint was sustained. The method has shown promising results, and is able to successfully find viable designs, even when starting from highly unacceptable starting points. Some of the major challenges when using SPSA for lattice support structures are to find a good objective function, as well as appropriate values for the parameters controlling perturbation and step size. Existing guidelines were followed when doing this calibration, but for an efficient search the parameters had to be adapted. Results for both appropriate parameters and the optimization itself are reported for the 10MW NOWITECH reference turbine on a full-height lattice tower. These results show that superior results can be achieved, but at a high cost in terms of computational time. Recommendation is given to use alternative methods to come up with a partially optimized staring point, from which the SPSA method can optimize further.

Brittle fracture is a large problem for steel structures in the arctic region. It is thus important to qualify materials and welds so they do not behave in a brittle manner. Since fracture testing of the heat affected zone (HAZ) around a weld gives a lot of scatter, doing weld simulated testing is proposed as an alternative method. In this thesis cracks in weld simulated HAZ specimens are compared to cracks in real welds, by use of finite element simulations. A weld simulated specimen is usually more brittle than a real weld. The goal of this thesis is thus to find a general rule for how much more brittle a weld simulated test specimen is, compared to a real weld on a structure. It would then be possible to establish how brittle a real weld is based on the result from the weld simulated testing. As a fracture criterion the Weibull stress is used, which is a statistical criterion. Crack tip opening displacement (CTOD) is used as a measure on how brittle a specimen is. To compare weld simulated specimens with real welds, two 2D modified boundary layer (MBL) models are used. One homogeneous model to represent weld simulated specimens, and one with three different materials to represent a real weld. The three materials in the weld model are base material, welded material and heat affected zone. On the two models a large parameter study is performed. The variables investigated are: - Position of the crack relative to the HAZ. - Size of the HAZ. - Geometry constraint. - Mismatch in WM. - Mismatch in HAZ. - Hardening. - The Weibull exponent m. There have also been made 3D models to investigate the size effect on the weld simulated specimen. This is because a weld simulated specimen is limited to a cross-section of $10$x$10$~mm. The parameter study concludes that it is mainly the size of the HAZ, the yield stress mismatch and the geometry constraint, that make weld simulated specimens more brittle than welds. The 3D simulations are however concluding that the geometry constraint effect can not be included, due to the size of the small test specimen. Based on these results a general relationship is proposed between the critical CTOD for a weld simulated specimen, and the critical CTOD for a real weld. There are three requirements for this relationship to be valid: - At least 10% overmatch in HAZ compared to base material. - No more than 10% undermatch in HAZ compared to welded material. - Maximal brittle HAZ thickness of 0.5 mm.

The transport of oxygenated blood from the placenta to the human fetus has received great attention in clinical Doppler velocimetry studies, especially the ductus venosus. The ductus venosus is connecting the intra abdominal portion of the umbilical vein and the inferior vena cava at the inlet of the right atrium and is therefore significant when examining the fetus state of health. In this thesis, the distribution and pulsations of flow waves in the ductus venosus, umbilical vein and left portal vein are investigated. Furthermore, important model parameters of the fetus circulation are determined with stochastic simulations and sensitivity analysis. The sensitivity analysis was put forward in order to interpret the uncertainty of the many variables in the fetus circulation and how the flow waves are affected. The deterministic blood flow was modelled with a 1D mathematical model. A combination of the 1D mathematical model with a generalized Polynomial Chaos method was applied for the stochastic simulations. All the simulations in this thesis are performed in context of fetal diseases, motivated by typical detrimental signs in fluid mechanics related to the umbilical vein/ductus venosus bifurcation. Furthermore, reversed flow in ductus venosus during atrial contraction, often observed in chromosomally abnormal fetuses was demonstrated. It was found that the reason was high pressure difference between ductus venosus and left portal vein. Pulsations transmitted into the umbilical vein is another detrimental sign in fetus circulation and the parameter providing such pulsations is the diameter of ductus venosus. Further investigations related to the model parameters as well as boundary conditions are suggested in order to expand knowledge and confidence establishing the fetus well being.

The thesis starts with a simple description of the system and some brief theory needed for the following analysis. Further a factory acceptance test is performed, this implies measurement of the dynamic response due to load generated by the system itself under normal operation and making sure the dynamic behaviour are within the acceptance criteria. The data from the test is used to point out some weaknesses in the design from a dynamical point of view and some potential ideas of improvement are presented. To introduce additional damping of some of the rotational equipment is the idea which is proceeded with. A great simplification in this thesis is that only transversal vibrations (the y-component of the vibrations) are considered after the factory acceptance test is performed. This results in less reliable quantities although the trends and the final conclusion remains correct. A comprehensive finite element analysis is carried out to study the effect of a passive damper solution. It turns out that passive dampers can limit the vibrations remarkably, but if the foundation gets too soft the vibrations will increase. The low frequency vibration however, is not affected much by the additional damping. Some important eigenfrequencies are also studied and put into consideration. A sensitivity analysis of the constant damping ratio in the finite element solver is also performed. In addition a more trivial analysis is performed to study the effect of an active damper solution. This solution seems to be less effective and does not influence the low frequencies considerably neither. It also introduces additional risk and weight to the system, which is highly undesirable. The thesis concludes that passive damping of the hydraulic power unit has favourable effects on the system. If the main goal is to reduce low frequency vibration it might not be an entirely satisfying solution.

En løsning for to hoved lager og en bunnplate for NOWITECH 10MW referanse turbinen er foreslått basert på dimensjonering i henhold til ekstremlaster. Ekstremlastene er basert på ekstrapolerte laster under vanlig strøm produksjon for normale og ekstreme turbulens forhold spesifisert i IEC61400-3:2008 - Design requirements for offshore wind turbines [1]. Bunnplaten består av to komponenter med en total vekt på 82.4 tonn, bunnplaten har hittil ingen yaw-løsning for overføring av laster til tårnet. Styrkeanalyser er utført basert med ekstremlastene og den høyeste von Mises spenningen ble funnet til å være 208,3MPa. En konfigurasjon av et kulelager og et dobbelt konisk rullelager ble valgt ut ifra SKF sin produktkalatog. Tilslutt er videre arbeid for bunnplaten diskutert.

This thesis is based on a project from the previous semester (fall 2011) and initiated by Sigve Gjerstad at Frank Mohn Flato y AS. The thesis introduces the FWP system and different aspects regarding a static analysis of it's main component; The FWP Module. It takes a brief look at different meshing techniques and other choices that needs to be made in the early stages of an analysis. A series of simple analyses are carried out to show how shell elements are the best representation for a plate structure subjected to pressure. A series of simplified blast load analyses are presented where different choices within the Finite Element Method are compared. It's concluded that it is sufficient to consider only one of the two load steps to get the maximum values of stress and deformation. This saves us computation time with no loss of accuracy. The analyses also conclude that a blast load analysis is dependant on a non-linear material model to get reasonable result. A linear material model assumes stress is proportional to strain, even beyond the yield strength. This results in unrealistic high stresses. Implicit solver versus explicit solver are compared in the case of blast loading, which is a problem that requires short time increments. It is clear that the results are similar, however the computational cost is much higher for the implicit solver. It is also shown that stainless steel is more beneficial than structural steel in blast load scenarios. Finally, model simplification is studied as yet another way to decrease the computation time. This implies simplifying solid models with a mid-surface features, representing the model with shells

Filament winding is a fabrication method for composite material structures, in which fibres are wound around a rotating mandrel. It is a versatile and dexterous process especially well-suited for creating and optimizing parts with a linear rotational axis. Products like pressure tanks, golf clubs or violin bows are commonly created using this technique. The winding itself is done through software solutions that generate a CNC program for the part in question. There are several such software solutions commercially available, all with different modes of operation and functionalities. However, they are also proprietary and offer little to no access into their inner logic. To optimise a part before production Finite Element Analysis software is often used. The part in question is modelled; material, forces and constraints are applied; and an analysis is run. Currently (June 2012), there are few options available for analysing filament wound products. Modelling a part with accurate filament winding layup generally has to be done manually, in a very time-consuming process. In this thesis, the author has performed a pilot study into the development of filament winding software. Software has been developed, capable of integrating both with a filament winding machine and with Finite Element Analysis software, and operating as a link between the two. The software has functionalities to extract geometrical variables from an Abaqus mandrel model; to write G codes and create a CNC program file; simulate a filament winding process in the Abaqus viewport; and, using a CNC program file, add accurate and corresponding layup to an Abaqus part. The main goal of this thesis, however, has been to create something that will serve as a basis from which others can continue development. The intention being that the software will be open source, so that anyone and everyone using it may change, improve and add on to it.

Kaldnes HYBAS (Hybrid Biofilm Activated-Sludge) avløpsvann-renseprosess består av to prosesser som er relativt godt forstått: Aktivslam (AS) og Moving-Bed Biofilm Reactor (MBBR), biofilm med bevegelige bærere. Det finnes en stor del litteratur, men spesielt for Aktivslam, som beskriver virkemåten og dimensjoneringsprosedyrer for de to systemene. For HYBAS er det viktig å få kunnskap om hvordan biofilm interakterer med aktivslammet og å danne en metode for dimensjonering av en fullstendig HYBAS prosess. Professor Hallvard Ødegaard ved NTNU har foreslått en modell for HYBAS, formulert som en 15-stegs utregnings-prosedyre, delvis basert på erfaringsdata, men med vekt på å være logisk og intuitiv. Denne modellen er analysert for overenstemmelse med litteraturen. Små modifikasjoner som forbedrer nøyaktigheten og generaliserer modellen er foreslått. Modellen er implementert i et Excel regneark for å muliggjøre eksperimentering og videre utvikling av modellen. En analyse av enkelte komponenter av modellen mot data fra et pilot-anlegg er forsøkt, men med betydelige problemer. Noen gode overenstemmelser med modellen blir vist, men også sterke uoverenstemmelser med modellen blir identifisert, med forbehold.

Hensikten med denne oppgaven var å kartlegge spillvarmestrømmene på et aluminiumsverk, samt diskutere deres utnyttelsespotensialer. Det ble ansett som mest hensiktsmessig å ta utgangspunkt i Hydro Aluminium Sunndal. Hydro var villige til å samarbeide om dette med kriterie om at oppgaven må kunne være matnyttig for et NTNU-Hydro-prosjekt som omhandler utvikling av en rågassvarmeveksler. Å undersøke partikkelkonsentrasjon og partikkelstørrelsesfordeling i rågassen fra elektrolysen på Hydro Sunndal ble dermed en viktig del av oppgaven. Spillvarmestrømmene på Hydro Sunndal ble kvantifisert til 3,4 TWh per år. Av denne energien forsvinner 304 GWh/år med rågassen fra SU3 ved en gasstemperatur på omtrent 90C, og 673 GWh/år med rågassen fra SU4 ved en gasstemperatur på omtrent 110C. 977 GWh/år med spillvarme har derfor et middels stort utnyttelsespotensial. Energien er meget godt egnet for distribusjon med fjernvarme, men markedet for dette i nærliggende områder er meget lite. Elektrisitetsproduksjon kan bli aktuelt i fremtiden. Spillvarmestrømmene er hovedsakelig kartlagt med dataer fra ulike kilder på Hydro Sunndal, blant annet fra HERE-databasen og målinger gjort av laboratorieteamet på verket, samt et måleprogram for varmetapet gjennom en elektrolysecelles overflater, utført fra 28. til 30. april 2008 på Hydro Sunndal, med tilhørende beregningsmodell. Et måleprogram for rågassen fra elektrolyseceller ble utført på Hydro Sunndal i perioden 10. - 14. mars 2008. Støvkonsentrasjonen i rågassen fra 56 elektrolyseceller i SU3 ble målt til å være 363 mg/Nm3 med et standardavvik på 12 %. Dette avviker sterkt fra resultater fra målinger gjort av Hydro Sunndals laboratorieteam i perioden februar 2007 til januar 2008, hvor støvkonsentrasjonen er målt til 848 mg/Nm3 med et standardavvik på 15 %. Avviket antas i hovedsak å skyldes naturlige variasjoner i støvkonsentrasjonen i rågassen. Resultatene for partikkelstørrelsesfordelingen forelå ikke da oppgaven gikk i trykken. Anbefalt videre arbeid er analyse av partikkelstørrelsesfordeling i rågassen så snart de nødvendige data foreligger, et omfattende måleprogram for å kartlegge energibalansen til en elektrolysecelle, samt et måleprogram for å kartlegge partikkelkonsentrasjon og -størrelsesfordeling for rågassen fra Hydro SU4.

Spyling (Flushing) av gass med en væske kolonne er viktig i forbindelse med trykk testing av rørledninger. For å vurdere multifase simulatoren OLGA 6.0 sin evne til å predikere spyling av en rørledning har simuleringer i OLGA blitt sammenlignet med små skala forsøk. En test rigg har blitt satt opp med konfigurasjonen av en bølgeformet rørledning. Den viktigste variabelen var høyden på vannet i reservoaret. Forsøkene ble filmet med et video kamera. Slutt tilstanded ble logget ved å måle den vertikale høyden av væske kolonnene i de ulike rør seksjonene. Dette ble sammenlignet med slutt tilstanden i OLGA simuleringene. Et Matlab skripe ble utviklet for å gjøre bilde analyse av filmen. Bilde analysen ble brukt til å sammenligne det transiente forløpet av eksperimentene med simuleringene i OLGA. Slutt tilstanden i forsøkene hvor røret ikke ble spylt var i god overensstemmelse med simuleringene i OLGA. Det transiente forløpet var mye raskere i OLGA. Forholdet mellom tiden det tok væsken å nå utløpet i eksperimentet hvor røret ble spylt og simuleringen i OLGA var 2.5. Dette forholdet avtok med innløpstrykket. Grunnen til denne uoverensstemmelsen er vurdert å komme av at det ikke er noen modell for overflatespenning mellom fluid og vegg i OLGA. For å finne minste løftehøyde for at OLGA skulle predikere spyling av røret, ble en parameterstudie av innløpstrykket utført. Faktoren mellom løftehøyden som var nødvendig for å spyle røret i eksperimentene og OLGA simuleringen var 0.84. Dette var overraskende siden OLGA predikerte et mye raskere transient forløp med større hastighet og bevegelsesmengde. Grunnen til over prediksjonen av den nødvendige løftehøyden antaes å komme av at OLGA til en liten grad tar høyde for strømnings historikk. Effekten av dette er at væskeplugger forsvinner i overgangen mellom et oppover rør og et nedover rør.