Crosswind stability of rail vehicles has been a research area for several decades,mainly motivated by vehicle overturning accidents and higher speeds, but in recenttimes also by issues of lower energy consumption and track maintenance costsdemanding lower vehicle weights. During everyday operation, rail vehicles are subjectedto large lateral influences from track irregularities, track curves and crosswind,leading to large suspension deflections and increased crosswind sensitivity.Also unsteady crosswind like gusts calls for attention. Simulations of possible vehicleoverturning are necessary, but need to take large deflections and high shear inthe suspension into account. If delivering reasonable results, simulations representan important tool for overturning prediction of the vehicle. In the present work, multibody simulations of a high-speed vehicle at large lateralinfluences from track curves and track irregularities have been carried out, using ahalf-vehicle model in 2D and a model of a whole vehicle in 3D. The vehicle modelsalso include different suspension models. Corresponding field measurements ofthe relative lateral and vertical deflections in the secondary suspension have beenperformed on a fast train and used for validation of the multibody simulations,resulting in good agreement between measurements and simulations. The 3D vehicle model was further used to study the vehicle response to unsteadycrosswind during curve negotiation where aerodynamic loads obtained by unsteadyComputational Fluid Dynamics, namely Detached Eddy Simulations, representingthree types of gusts were used. In addition, the method of Quasi Transient GustModelling was evaluated in terms of overturning risk. Strong lateral and roll responsesof the vehicle and influences of the gust duration and the relative differencebetween mean and maximum wind speed were observed. Further, variations of suspensionand mass properties of the vehicle were performed to study the influenceon crosswind sensitivity. The position of the centre of mass of the carbody and thelateral bumpstop clearance showed significant influence on the crosswind stability.

Michelin develops and manufactures tires. To check their tire, different tires tests are used. During the testing process of agricultural tires, some endurance tests are performed on a circular test track where a convoy, made of a truck pulling two single-axle trailers without suspension system, is driven at a constant speed. Depending on the tested tires, it has been observed that the convoy may become severely instable. This instability is a large problem for the test team since the tires undertake too much overload which makes test results unexploitable. This is time consuming and expensive for the company. The aim of this thesis is two-folded. The first aim is to identify the phenomenon causing the instability from experimental data and also the parameters which influence the phenomenon. The second aim is to model the phenomenon from technical data in order to predict instability behaviors in advance. Models have been developed and the behavior have been analyzed and compared with experiments. The main result of this work is that the instable behavior is due to the excitation of natural modes of the convoy by the tires frequencies. Natural modes are identified as the natural bouncing mode of the trailers. The bouncing natural mode of a trailer depends on tires stiffnesses and the load on the trailer. Tires excitation frequencies are related to the test speed, the trailer track width and the rolling circumference of the tires. To get a good prediction tires stiffness and tires rolling circumference under operation require to be better characterized. Thanks to this work the test teams have a better understanding of the phenomenon and a tool which can be used to give a rough indication about problematic configurations. This tool cannot totally predict instable behaviors with the technical data at our disposal since some more parameters, which are not quantified for now, might influence the phenomenon. Nevertheless, these parameters have been highlighted by this study and if explored by the test team, a working predicting tool could be achieved.

No description available.

Planning vessels can reach high operational speeds thanks to their hull design. Some hull forms will develop large volumes of spray attached to the hull surface. The whisker spray is a thin layer consisting in droplets of water which can ac-count for a large proportion of the total resistance. A new concept to redirect the spray, called de°ectors, has been developed by the Swedish company Petestep. These de°ectors indicate higher e˝ciency than the time-proven spray rails tech-nology by removing a bigger portion of the spray area. The spray is re°ected back-wards rather than to the sides, which allows kinetic energy contained in the spray sheet to be converted to additional forward thrust. However, there have only been a few studies conducted on the e˛ects of de°ectors and there is no precise method to analyze their e˝ciency over the full range of operating speeds. For the above-mentioned reason, experimental testing is needed to have a more complete under-standing of the phenomenon in calm water and waves. In this study, model scale tests of a modular planning hull are carried out at the Davidson Laboratory towing tank. The goal is to verify the beneÿts of the spray de°ectors by direct comparison with the bare hull conÿguration at the same trim angle.

Hydrodynamic analysis of high-speed craft is a complex problem. For years naval architects have tried improving the eÿciency of the hull, changing its form, the number of hulls, adding spray rails or steps. Spray deflectors, an innovative modification of commonly used spray rails by a Swedish company called Petestep, might be the next advancement in that process. Experimental validation of this concept is the topic of this study. Previous years of towing tank tests showed compelling results, but were unable to draw definite conclusions, mainly due to inability to compare the results of bare hull and deflector hull tests which was a result of unequal trim achieved for equal input parameters (such as weight and speed).With amended test speed range and deflector design modifications, this project aimed to yield unambigu-ous results. Three spray deflectors were designed for di˙erent design speeds, tested and compared with bare hull tests for drag and trim. The results showed that deflectors did not change the running position of the hull compared to bare hull, allowing for an unequivocal drag comparison. Deflectors reduced drag by up to 4.9%.A number of design improvements were proposed as next steps in this research. A further successful proof of concept (increased stability, lower resistance and fuel consumption) could provide a substantial improvement in high-speed craft design resulting in a cheaper and a more comfortable ride with a lower environmental impact.

The trend of increasingly automated tightening stations in the motor vehicle powertrain production industry, is driving the tooling industry to adapt its of-fering. This thesis aims to answer the question if collaborative robots (cobots) are suitable for tightening applications in motor vehicle engine assembly lines. A collaborative robot is in principle similar to a conventional industrial robotic arm. The part that differentiates it from an industrial robot is its ability to detect and in some cases prevent a collision with a human of close proximity. In order to answer this question three case studies are conducted where the deployment of collaborative robots is assessed. To specifically answer the case of a motor vehicle engine line, two lines are mapped station by station. Then these stations are recreated with a collaborative robot and a tightening tool and a few applications are tested on an actual motor. The rest is simulated with equations, dictating how much time is needed to complete a sequence of tightenings.As a result, most stations on a motor vehicle assembly line could in theory be done by using tightening tools together with one or multiple collaborative robots. However, these kinds of lines are often highly automated. For this rea-son large industrial robots and multi-tool fixtures are preferred today. The large volumes of these lines, mostly mitigate the expected financial gains of replacing a human operator with a cobot. In many cases, the cobot would not have the time to complete all bolts of an application on an engine within the give takt time. Multi-cylinder engines (which is the absolute norm for passenger cars) have too many bolts for one cobot and tool to complete. If multiple cobots are to be used, the cost of the station further increases. Furthermore, re-balancing a line simply to accommodate a cobot would also not be beneficial.Probably a much better application for collaborative robots and tightening tools are smaller, more operator dependent lines. Here the value of a cobot can be utilized to a greater extent. These manufacturers have greater product variance and re-balance their lines more often. As a suggestion for further research, this segment could be invesigated further. / Den allt ökande trenden av mer automatiserade åtdragningsstationer i fordonsindustrins tillverkningslina av motorer, driver verktygsleverantörer att se över sin produktportfölj. Detta examensarbete har som syfte att besvara frågan om kollaborativa robotar (cobots) är lämpliga att nyttjas för åtdragande applikationer i produktionslinor av motorer inom bilindustrin. En kollaborativ robot är i sin princip lik en industriell robot arm. Det som skiljer sig åt är att en kollaborativ robot har förmågan att upptäcka och i vissa fall förebygga en eventuell kollision med en människa som befinner sig i närheten. För att besvara detta har tre fallstudier utförts. För att specifikt kunna besvara frågan gällande motorproduktionslinor har två linor kartlagts station för station. Sedan har några av dessa stationer återskapats fysiskt med en motor och en kollaborativ robot samt åtdragande verktyg. Andra stationer har simulerats med ekvationer som bestämmer hur mycket tid som krävs för att slutföra en sekvens av åtdragningar. Resultatet visar att de flesta stationerna i en motorproduktionslina inom bilindustrin kan utföras av åtdragande verktyg monterade på en eller flera kollaborativa robotar. Men, dessa typer av produktionslinor är ofta väldigt automatiserade. Ofta används stora industriella robotar eller fixturer som innehåller många åtdragande verktyg. De stora volymerna som sådana produktionslinor producerar, dämpar vinsterna av en förväntad kostnadsbesparing. Stora och dyra installationer är inget problem när produktionsomsättningen är så stor. I många fall kan en cobot inte ersätta en människa eller fixtur inom takttiden på en motorproduktionslina där motorn har flera cylindrar, viket är den absoluta normen för personbilar. Om flera cobotar ska användas för att hinna med takttiden är kostnaden snabbt uppe i nivåer som är jämförbara med en mer traditionell installation. Om en cobot skulle appliceras effektivt måste då hela produktionslinan balanseras om enbart för detta, vilket blir ett mycket dyrare projekt. Förmodligen en mycket bättre applikation för kollaborativa robotar är mindre och mer operatörsberoende produktionslinor. Här kan värdet med cobots bättre nyttjas. Dessa tillverkare har större produktvarians och måste balansera om deras produktionslinjer mer frekvent. Som ett förslag för framtida studier kring kollaborativa robotar i produktionslinor är detta ett segment som kan undersökas ytterligare.

Water covers about 70 % of Earth’s surface. Most of the water is located in the ocean’s, which have an average depth of almost 3700 meters. This implies that there is a multitude of phenomena to investigate and discover. But due to human limitations other methods and tools need to be considered in order to investigate the depths. With help of underwater robotics it can be possible to reach further and deeper. In Sweden a research center named SMaRC (Swedish Maritime Robotics Centre) was established. Mainly to maintain the knowledge Sweden posses in underwater technology. The goal is to develop the next generation robots which are smarter, offers longer reach and higher endurance. Currently one of the projects at SMaRC is dealing with underwater communication, where a five element SONAR array has been developed.This thesis deals with characterising the five element SONAR array developed in SMaRC. This is done using two different methods, one theoretical and one experimental. The theoretical method involves modelling the directivity and beam steering the SONAR array has, using theoretical formulas. While the experimental method involves a field test where the directivity and beam steering is measured and analyzed. These two methods are then compared with each other and the result is analyzed.Both the methods are showing similar results for the directivity, where the main lobe is located at 0°. Beam steering of 45° and 60° also gives similar results. For a steering of 45° the main lobe is closer to the input angle in the theoretical method than in the experimental method. In beam steering of 60° the locations of the main lobes are shifted and the experimental method lies closer to the input angle. The side lobes also differs slightly between these two methods and no explanation was found. The Source Level (SL) for different beam steering are recorded in the experimental method to: Phase of 0° gives a SL of 131 dB re. μPA/V at 1 m Phase of 45° gives a SL of 129 dB re. μPA/V at 1 m Phase of 60° gives a SL of 128 dB re. μPA/V at 1 m / Vatten täcker omkring 70 % av jordens yta. Den största delen av vattnet hittas i haven, vilkethar ett medeldjup på nästan 3700 meter. Detta innebär att det finns en mängd fenomen attundersöka och upptäcka. Men på grund av människans begränsning måste andra metoderoch verktyg övervägas för att undersöka dessa djup. Med hjälp av undervattensrobotik är detmöjligt att nå längre och djupare. Ett forskningscenter vid namn SMaRC (Swedish MaritimeRobotics Center) inrättades i Sverige. Främst för att upprätthålla den kunskap som Sverigebesitter inom undervattensteknik. Målet för SMaRC är att utveckla nästa generation av robotarsom är smartare, erbjuder längre räckvidd och längre uthållighet. För närvarande handlarett av projekten på SMaRC om undervattenskommunikation, där en SONAR array med femelement utvecklats.Denna avhandling handlar om att karakterisera en SONAR array bestående av fem elementsom utvecklats hos SMaRC. Detta är gjort genom att använda sig av två olika metoder,en teoretisk och en experimentell. Den teoretiska metoden involverar beräkningar avdirektiviteten och möjliga fasförskjutningar hos SONAR arryen genom teoretiska formler.Medan i den experimentella metoden involveras av ett fälttest där direktiviteten och möjligaphasförskjutningar är inspelade och analyserade. Dessa två metoder jämförs sedan medvarandra och resultatet analyseras.Båda metoderna visar liknade resultat för direktiviteten, där huvudloben är belägen vid 0°.Fasförskjutningen för 45° och 60° ger också liknande resultat. En fasförskjutning av 45° liggerhuvudloben för den teoretiska metoden närmare den givna vinkeln än för den experimenellametoden. En fasförskjutning av 60° är positionen för huvudloben flyttad och den experimenellametoden ligger nu närmare till den givna vinkel. Sidloberna skiljer sig också en aning mellanmetoderna och det är svårt att säga något om dem. Signalnivå (SL) för olika fasförskjutningarär detekterad i den experimetella metoden med följande resultat: En fasförskjutning på 0° ger en SL på 131 dB re. μPA/V , 1 m En fasförskjutning på 45° ger en SL på 129 dB re. μPA/V , 1 m En fasförskjutning på 60° ger en SL på 128 dB re. μPA/V , 1 m

En semi-empirisk modell som beskriver vitvattets sönderfall i den turbulenta vaken har härletts utifrån optiska studier av Stridsbåt 90H och Buster Magnum. Bakgrunden är ett ökat hot från rymdburna system mot den svenska Marinen. Fokus har varit Amfibiekårens båtpark. Vitvattenspår syns tydligt i det optiska spektrumet och avges ofrånkomligen från både skrov och propulsor i planande farter. Vid tillräckligt minskad fart försvinner skrovens vaksignatur i rymdburna sensorer men jetaggregaten från amfibiefartygen (Stridsbåt och lätt Trossbåt) avger fortsatt stora mängder vitvatten. Både en högre vitvattenfri marschfart och kortare vitvattenspår skulle minska spårbarheten. Detta leder uppmärksamheten till valet av propulsor. Idag existerar inga lämpliga metoder för jämförelser och därmed inga fastställda krav avseende deras vitvattensignatur.  Beräkningar med modellen görs med initial tomrumsfraktion på den turbulenta vakens yta och vattnets salinitet som invärden. Ett mätförslag som bygger på lokal mätning med konduktivitetsmätare presenteras i rapporten. Förutsatt att dessa variabler är kända är det enklare att bedriva jämförande studier än med t.ex. hydrodynamiska simuleringar och aktiv akustik. Modellen tar hänsyn till signifikanta operativa, miljö- och designvariabler och är således bättre anpassad för planande fartyg i ett näringsrikt och salthaltsvarierande Östersjön än de semi-empiriska vitvattenmodeller som sedan tidigare existerar. / A semi-empirical model for calculating the whitewater signature in the turbulent wake was derived based on optical studies of the Combat Boat 90H and Buster Magnum. The background to the study is an increased threat from Satellite remote sensing towards the swedish marine military forces. A demarcation is made to the vessels of the Amphibious Corps. White-water traces are clearly visible in the optical spectrum and are inevitably emitted from both hull and propulsor at planning speed. At sufficiently reduced speed, the sea surface signature from the hull disappear in satellite sensors, but the white water from the jet-powered amphibious vessels (Combat boat 90 and Fast Supply Vessel) continues to emit large amounts of white water. Both a higher white-water-free marching speed and shorter white-water tracks reduces traceability with space-borne sensors. This brings attention to the choice of propulsor. Today there are no established methods for comparison and thus no established requirements for their whitewater signature. Input variables to the model are local void fraction measured in the surface of the turbulent wake and the salinity of the water. A measuring proposal based on the conductivity probe method is presented in the report. Given these input variables are known, the model is more convenient for comparative studies than hydrodynamic simulations or active acoustics. Significant operational, environmental and design variables are taken into account which makes the model better suited for planning vessels operating in greatly varying water conditions (such as the Baltic Sea) than previous existing semi-empirical equivalents.

The use of composite materials has been common in small craft boat building for a long time. In recent years, there has been a huge push in the development of different types of appendages such as hydrofoils. These hydrofoils are commonly manufactured in carbon fibre composites, due to high requirements in weight and stiffness. These appendages can be difficult to develop and complex to manufacture since manufacturing methods for composites are complex. KTH Royal Institute of Technology is developing a hydrofoil concept for a small autonomous vessel. The hydrofoil is designed to be built in carbon fibre composite. It requires to have control surfaces in order to maintain a stable flight and the electrical propulsion is located on it as well. This makes this hydrofoil one of a kind and the parts that build up the hydrofoil have more specifications than just to be designed from a hydrodynamic and structural point of view like a conventional hydrofoil. This thesis investigated what manufacturing methods should be used when building a hydrofoil like this. Existing manufacturing methods such as vacuum infusion and different types of prepreg moulding have been reviewed and are presented early in the report. The methods have been analyzed from the perspective of the components of the hydrofoil, resulting in an initial manufacturing strategy for the different components. The strategy includes everything from a 3D-model of the part to a finished product, including sheet design, mould manufacturing and moulding of the part. Several tests were conducted before a component was successfully manufactured. Each test was evaluated and presented in such a manner that the reader can understand what is needed to be improved and why. The conclusions of each test lead to an improvement of the manufacturing technique and a new test until the final result was acquired. The tests were examined with a microscope to verify the quality of the part. Then a weight fraction analysis was made on these parts. The final conclusions of the thesis gave successful methods to manufacture the different parts of the hydrofoil. A fast manufacturing method for product development of complex parts was achieved. The resulting parts from the tests show good quality from the analysis.

The purpose of this work is to find out if an artificial neural network can be useful purpose of this work is to find out if an artificial neural network can be useful in order to detect rail squats with the existing Quiet Track Measurement System (QTMS). Squats are surface-initiated rail defects which arise due to rolling contact fatigue. The monitoring system, installed on seven trains running on the green line in the Stockholm underground, aims to improve the maintenance process. The early detection and surveillance of defects helps to extend the service life of the tracks and reduce operating costs. An artificial neural network is used to analyse the the continuously recorded measurements, which consist of vertical bogie acceleration and surrounding noise, each sampled with a frequency of 22 kHz.In particular, the power spectral density as input for multi-layer Fully-connected Neural Network (FNN) has proven to be promising for accurate squat predictions. The supervised learning was carried out according to the one-vs-all principle, i.e. squats versus all other events. A two-hidden-layer FNN has finally been chosen to complement the QTMS. The usage of the full available frequency range from almost DC up to 11kHz, but minimum 7 kHz, allows good prediction with only low false prediction rates. When concatenating all six measurement channels to a single classifier input, an accuracy of over 96% for the squat class and up to 99.98% can in total be achieved. The chosen network type also showed high stability despite quite strong parameter variations and a massive under-representation of squat observations in the measurement data.However, since limited maintenance information about actual squats is available for labelling and testing, more evaluation is needed. The correct identification of mis-labelled squats indicates the high potentials of artificial neural networks. / Squats uppkommer på rälsytor på grund av rullkontaktutmattning och kan ställa till stora problem om de inte upptäcks i tid. Att upptäcka fel i ett tidigt skede kan minska underhållskostnaderna. Syftet med det här examensarbetet är att studera om man med hjälp av ett neuralt nätverk kan detektera så kallade räls-squats med hjälp ett mätsystem som heter Quiet Track Measurement System (QTMS). Mätsystemet är installerat hos sju tåg på den gröna linjen i Stockholms tunnelbana. Systemet kan användas för att övervaka rälsslitage och kan därför effektivisera underhållet. Den här rapporten presenterar ett artificiellt neuralt nätverk för att kontinuerligt kunna analysera mätdata. Mätdata består av vertikal boggiacceleration och ljudmätningar, båda samplas med frekvensen 22 kHz.Frekvensdomänrepresentationen av uppmätta data i form av spektraltätheten i kombination med flerskikts- och helt anslutna neurala nätverk (FNN) visade sig vara lovande för korrekta förutsägelser. övervakad inlärning tillämpades enligt the one-verus-all principen, dvs antingen har man squats eller så har man inte squats. En artificiell neuron-modell med två dolda lager valdes till slut för att komplettera existerande mätsystem. Användningen av ett brett frekvens-område från nästan DC upp till 7 kHz möjliggör god förutsägelse med bara låga förutsägelser av falska squats. När man sammanlänkar alla sex mätkanaler till en enskild klassificeringsinput kan man uppnå en noggrannhet över 96%, som högst kan man uppnå 99.98%. Nätverket visade också hög stabilitet trots ganska starka parametervariationer och en obalans mellan tillgängliga data för de två klasserna.Eftersom bara några få underhållsprotokoll fanns tillgängliga krävs mer utvärdering, men korrekt identifiering av felklassificerade squats indikerar att den verkliga noggrannheten kan bli ännu bättre. Detta ger också förutsättningar för att snabbare kunna prediktera squats.