Syftet: Utvärdera produktionstolerans av objekt som producerats genom additiv framställningsteknik (AF) för användning inom tandvård, samt att jämföra denna teknik med subtraktiv framställningsteknik (SF) genom reverse engineering.Material och metod: Tio exemplar av två olika geometriska objekt framställdes från fem olika AF maskiner och en SF maskin. Objekt A efterliknar ett inlay, medan objekt B återspeglar en modell av en fyrledsbro. Alla objekt delades in i olika mätled; X, Y och Z. Mätningarna utfördes med validerade och kalibrerade instrument. Linjära avstånd mättes med ett digitalt skjutmått och hörnradie samt vinklar mättes med ett digitalt mikroskop.Resultat: Vare sig additiv eller subtraktiv framställning uppvisade en perfekt matchning till CAD-filen med hänsyn till de parametrar som utvärderades i denna studie. Standardavvikelsen gällande linjära mätningar för subtraktiv framställning uppvisade konsekventa resultat i alla led, med undantag för X- och Y-led för objektet A och i Y-led för objekt B. Samtliga additiva tillverkningsgrupper hade en konsekvent standardavvikelse i X- och Y-led, men inte i Z-led. Med avseende på hörnradiemätningar, hade SF gruppen i överlag bättre produktionsnoggrannhet för både objekt A och B medan AM grupperna var mindre noggranna.Konklusion: Med hänsyn till begränsningarna med denna in vitro studie, stödjer resultat hypotesen, med hänsyn till att AF hade en bättre förmåga att återskapa komplexa och små geometrier jämfört med SF. Samtidigt identifierades en bättre reproducerbarhet hos SF gällande enkla geometrier och linjära avstånd. Vidare studier krävs för att bekräfta dessa resultat. / Purpose: To evaluate the production tolerance of objects produced by additive manufacturing systems (AM) for usage in dentistry and to compare with subtractive manufacturing system (SM) through reverse engineering. Materials and methods: Ten specimens of two geometrical objects were produced by five different AM machines and one SM machine. Object A mimics an inlay-shaped object, meanwhile object B reflects a four-unit bridge model. All the objects were divided into different measuring-axis; X, Y and Z. Measurements were performed with validated and calibrated equipment. Linear distances were measured with a digital calliper while corner radius and angle were measured with a digital microscope. Results: None of the additive manufacturing or subtractive manufacturing groups presented a perfect match to the CAD-file regarding all parameters included in present study. Considering linear measurements, the standard deviation for subtractive manufacturing group were consistent in all axis, except for X- and Y-axis in object A and Y-axis for object B. Meanwhile additive manufacturing groups had a consistent standard deviation in X- and Y- axis but not in Z-axis. Regarding corner radius measurements, SM group overall had the best accuracy for both object A and B comparing to AM groups. Conclusion: Within the limitations of this in vitro study, results support the hypothesis, considering AM had preferable capability to re-create complex and small geometry compare to SM. Meanwhile, SM were superior producing simple geometry and linear distances. Further studies are required to confirm these results.

With the roll out of new technologies and the worldbecoming more connected, there is a rising demand for higherbandwidth and new frequency bands. To meet the demand,higher frequencies are used in new communication systems.Higher frequencies come with the need for new antenna designsand one promising type of antenna is the lens antenna. In thispaper, a modulated geodesic Luneburg lens with a novel feedingmethod is proposed for use between 8-10 GHz. Furthermore, themanufacturing of the lens explores the possibility of 3D printingas a method of producing cheap antennas.The paper verifies the viability of using a patch antenna andhorn as a feeding method for a parallel-plate waveguide lens.First the lens is modeled and simulated in CST Microwave Studioand is then 3D-printed in PLA plastic and taped with coppertape. The antenna achieves -5 dB S11-parameter between 8-10GHz. The antenna also achieves 60 scanning in the azimuthplane. The antenna achieves a HPBW of 15. / Med utvecklingen av nya tekniker och envärld som blir allt mer digital är efterfrågan på större bandbreddoch nya frekvensband hög. För att möta efterfrågananvänds högre frekvenser i nya kommunikationssystem. Medanvändningen av högre frekvenser behövs nya antenndesigneroch en lovande typ av antenn är linsantennen. I den härartikeln föreslås en modulerad geodesic Luneburg lins med enny typ av matningsmetod för användning mellan 8-10 GHz. Förtillverkningen av linsen utforskas 3D-printning som en billig ochenkel metod.Artikeln verifierar användningen av en patch-antenn och etthorn som matningsmetod för en lins av parallella metallplattor.Först simuleras linsen i CST Microwave Studio och 3Dprintassedan i PLA-plast och tejpas med koppartejp. Antennenåstakommer -5 dB i S11-parameter mellan 8-10 GHz. Antennenhar en skanning av 60 i azimut-planet och har en HPBW av15. / Kandidatexjobb i elektroteknik 2021, KTH, Stockholm

With the rise of 5G and the increasing number ofdevices, novel antenna designs are needed to meet the demandof the future. In this report, the authors present a design andexperimental verification of a 3D-printed Geodesic ModulatedReflective Luneburg lens antenna working at the X-Band, 8-12GHz. The lens profile is calculated from the refractive index of aflat system using transformation optics. Furthermore, the lens ismodulated to minimize the height and chamfers are implementedto reduce reflections. A sliding waveguide connected to a coaxialcable is used to excite the lens while the transmitted signal isradiated from a sinusoidal flare. A copper-lined PLA substrateconstitutes the 3D-printed lens. The authors achieved a S11 below-10 dB across the spectrum and a realized gain exceeding 10 dBacross the sweeping angles at 12 GHz, showcasing the usabilityas a directed antenna. / Med det nya 5G nätverket och den ökandemängden enheter behövs nya antenner för att möta framtidensefterfrågan. I denna rapport presenterar författarna en designoch experimentell verifiering av en 3D-printad geodesisk moduleradreflekterande Luneburg linsantenn i X-bandet, 8-12 GHz.Linsprofilen beräknas från brytningsindexet för ett platt systemmed transformationsoptik. Dessutom är linsen modulerad föratt minimera höjden och kantavfasningar implementeras föratt minska reflektioner. En glidande vågledare ansluten till enkoaxialkabel används för att excitera linsen medan den sända signalenutstrålas från en vågledare med sinusformad avrundning.Ett kopparfodrat PLA-substrat utgör den 3D-printade linsen.Författarna uppnådde en S11 under -10 dB över spektrumet ochen realiserad förstärkning överstigande 10 dB över svepvinklarnavid 12 GHz, vilket visar linsens användbarhet som riktad antenn. / Kandidatexjobb i elektroteknik 2021, KTH, Stockholm

This project was initiated by Dr. Sasan Dadbaksh upon listening to the requirements I presented for my master thesis. My requirements were to do a master thesis project in the field of additive manufacturing specifically fused deposition modeling that should not only involve the research work but should also present an opportunity to develop hardware and should involve experimental testing. Then Sasan came up with the idea of developing a system capable to perform 3D printing with the extruder fixed in one position and the motion required for 3D printing will be provided by the robotic arm. The title of developing green build strategies for robot assisted plastic 3D printing came into being. The main concept behind the title of developing robot assisted plastic 3D printing platform is to develop such a system that can offer additive manufacturing services, specifically of fused deposition modeling 3D printing, as an inbound process during the manufacturing of any part through subtractive processes with the help of a robotic arm along with the repair of any kind of parts with the assistance of fused deposition modeling 3D printing. The main objectives of the master thesis include building a stationary filament extrusion module to interact with a robot hand and establishing a strategy for a robot hand to move the part to appropriate locations to complete building a part on a preform without support structures. The targets that were achieved with the completion of this thesis project includes the development of the complete hardware that consists of a mechanical structure with the option of mounting the components required to run the extrusion setup, learning the basic working of the software that are able to simulate the 3D printing process with the robotic arm (Robot Studio and Robo DK), creation of the simulation of the whole process, achieving communication between the robotic arm and the microcontroller of the extruder and finally the printing of a simple part for the demonstration. The components needed to be installed on the structure includes the motor, extruder, hot end, nozzle, filament. The structure also accumulated the required electronics that includes power supply, microcontroller, and an LCD to monitor the extrusion parameters. The developed machine runs on the state-of-the-art components that belong to the few of the best manufacturers of the technology.

O objetivo deste estudo foi o desenvolvimento de compósitos poliméricos flexíveis e altamente condutores elétricos preparados por moldagem por compressão e por fabricação de filamentos fundidos (FFF) para possíveis aplicações como materiais piezoresistivos ou piezoelétricos para sensores de compressão. Compósitos baseados em misturas de poli(fluoreto de vinilideno)/poliuretano termoplástico (PVDF/TPU) como matriz e contendo várias frações de negro de fumo-polipirrol (CB-PPy) como aditivo condutor foram preparados. Diversas técnicas de caracterização foram realizadas para avaliar as propriedades mecânicas, térmicas, químicas e elétricas, morfologia e printabilidade dos materiais investigados. Primeiro, blendas de PVDF/TPU com diferentes composições foram produzidas por mistura por fusão seguida de moldagem por compressão. Os resultados mostraram que a flexibilidade desejada para os materiais foi melhorada com a adição de TPU aos compósitos de PVDF. As imagens SEM evidenciaram a obtenção de uma blenda co-contínua com 50/50 vol% de PVDF/TPU. As blendas compostas de PVDF/TPU 38/62 vol% e a blenda co-contínua de PVDF/TPU 50/50 vol% foram selecionadas como matrizes para a preparação de compósitos moldados por compressão e impressos em 3D a fim de alcançar uma ótima combinação entre condutividade, propriedades mecânicas e printabilidade. Várias quantidades de negro de fumo-polipirrol, de 0 a 15%, foram adicionadas às blendas selecionadas para aumentar a condutividade elétrica dos compósitos e possivelmente atuar como agente nucleante para a fase cristalina do PVDF a fim de aumentar sua resposta piezoelétrica. A adição de CB-PPy aumentou a condutividade elétrica de todos os compósitos. No entanto, a condutividade elétrica dos compósitos baseados em blendas co-contínuas PVDF/TPU 50/50 vol% foi maior do que as encontradas para os compósitos de PVDF/TPU 38/62 vol% com mesma concentração de aditivo. De fato, o limiar de percolação elétrico dos compósitos com blenda co-contínua foi de 2%, enquanto o limiar de percolação elétrico dos compósitos compostos da blenda não contínua foi de 5%. Com relação às propriedades mecânicas, a incorporação do aditivo condutor nas blendas resultou em materiais mais rígidos com maior módulo de elasticidade, menor alongamento na ruptura e maior módulo de armazenamento. O módulo de armazenamento (G') e a viscosidade complexa (η*) dos compósitos aumentaram com a adição de CB-PPy. O limiar de percolação reológico foi de 3% para PVDF/TPU/CB-PPy 38/62 vol% e 1% para PVDF/TPU/CB-PPy 50/50 vol%, indicando que uma quantidade maior de carga poderia comprometer a processabilidade dos compósitos. A adição de CB-PPy também resultou na redução dos valores de Tg e Tm dos compósitos devido à redução da mobilidade das cadeias poliméricas. Com base na condutividade elétrica e no comportamento mecânico dos compósitos, três composições diferentes foram selecionadas para a extrusão de filamentos para serem posteriormente utilizados no processo de impressão 3D. No geral, as peças impressas em 3D apresentaram propriedades mecânicas e elétricas inferiores devido à presença de vazios, defeitos e camadas sobrepostas que podem dificultar o fluxo de elétrons. Os valores de condutividade elétrica dos compósitos impressos em 3D de PVDF/TPU/CB-PPy 38/62 vol% contendo 5% e 6% de CB-PPy são de uma a sete ordens de grandeza menores do que os encontrados para os compósitos com a mesma composição moldados por compressão. Mesmo que o valor da condutividade elétrica para o compósito PVDF/TPU 38/62 vol% com 6% de CB-PPy moldado por compressão foi de 1,94x10-1 S•m-1, o compósito impresso em 3D com a mesma composição mostrou um valor muito baixo de condutividade elétrica de 6,01x10-8 S•m-1. Por outro lado, o compósito co-contínuo de PVDF/TPU 50/50 vol% com 10% de aditivo impresso em 3D apresentou um alto valor de condutividade elétrica de 4,14×100 S•m-1 mesmo após o processo de impressão. Além disso, as respostas piezoresistivas dos compósitos foram investigadas. Para os compósitos PVDF/TPU/CB-PPy 38/62 vol%, as amostras moldadas por compressão e impressas em 3D com 5% e 6% de CB-PPy exibiram boa resposta piezoresistiva. No entanto, apenas os compósitos com 6% de aditivo apresentaram valores elevados de sensibilidade e gauge factor, atuação em ampla faixa de pressão e respostas piezoresistivas reprodutíveis durante a aplicação de 100 ciclos de compressão/descompressão para ambos os métodos de fabricação. Por outro lado, para os compósitos co-contínuos de PVDF/TPU/CB-PPy apenas a amostra moldada por compressão com 5% de CB-PPy apresentou respostas piezorresistivas boas e reprodutíveis. A cristalinidade e o teor de fase β do PVDF foram investigados para os compósitos. Embora o grau de cristalinidade das amostras tenha diminuído com a adição de CB-PPy, a porcentagem de fase β no PVDF aumentou. O coeficiente piezoelétrico d33 das amostras aumentou com a porcentagem de fase β. A adição de 6% ou mais de CB-PPy foi necessária para aumentar significativamente o coeficiente piezoelétrico (d33) dos compósitos. O conteúdo de fase β e as respostas piezoelétricas do PVDF foram menores para as amostras preparadas por FFF. Por fim, como pesquisa colateral, a eficiência de blindagem contra interferência eletromagnética (EMI-SE) foi medida para todos os compósitos. Compósitos com maior condutividade elétrica apresentaram melhor blindagem da radiação eletromagnética. Além disso, os compósitos baseados na blenda co-contínua apresentaram maior eficiência de blindagem contra EMI do que os compósitos de PVDF/TPU 38/62 vol%. O principal mecanismo de blindagem foi a absorção para todos os compósitos. As amostras preparadas por FFF apresentaram respostas de EMI-SE menores quando comparadas às amostras moldadas por compressão. / The aim of this study was the development of flexible and highly electrically conductive polymer composites via compression molding and fused filament fabrication for possible applications as piezoresistive or piezoelectric materials for pressure sensors. Composites based on blends of poly(vinylidene fluoride)/thermoplastic polyurethane (PVDF/TPU) as matrix and containing various fractions of carbon black-polypyrrole (CB-PPy) as conductive filler were prepared. Several characterization techniques were performed in order to evaluate the mechanical, thermal, chemical and electrical properties, morphology and printability of the investigated materials. First, PVDF/TPU blends with different compositions were prepared by melt compounding followed by compression molding. The results showed that the flexibility aimed for the final materials was improved with the addition of TPU to PVDF composites. SEM images evidenced the achievement of a co-continuous blend comprising 50/50 vol% of PVDF/TPU. The blends composed of PVDF/TPU 38/62 vol% and the co-continuous blend of PVDF/TPU 50/50 vol% were selected as matrices for the preparation of compression molded and 3D printed composites in order to achieve an optimal compromise between electrical conductivity, mechanical properties and printability. Various amounts of carbon black-polypyrrole, from 0 up to 15%, were added to the selected blends in order to rise the electrical conductivity of the composites and to possible act as nucleating filler for the β crystalline phase of PVDF in order to increase its piezoelectric response. The addition of CB-PPy increased the electrical conductivity of all composites. However, the electrical conductivity of composites based on PVDF/TPU 50/50 vol% co-continuous blends was higher than those found for PVDF/TPU 38/62 vol% composites at the same filler content. Indeed, the electrical percolation threshold of the conductive co-continuous composite blends was 2%, while the electrical percolation threshold of the composites with the nonco-continuous composite blends was 5%. With respect to the mechanical properties, the incorporation of the filler into the blends leaded to more rigid materials with higher elastic modulus, lower elongation at break and higher storage modulus. The storage modulus (G’) and complex viscosity (η*) of the composites increased with the addition of CB-PPy. The rheological percolation threshold was found to be 3% for PVDF/TPU/CB-PPy 38/62 vol% and 1% for PVDF/TPU/CB-PPy 50/50 vol%, indicating that higher amount of filler could compromise the processability of the composites. The addition of CB-PPy also resulted in a reduction on the Tg and Tm values of the composites due to the reduction of the mobility of the polymeric chains. Based on the electrical conductivity and mechanical behavior of the composites, three different compositions were selected for the extrusion of filaments to be used in a 3D printing process. Overall, the 3D printed parts presented lower mechanical and electrical properties because of the presence of voids, defects and overlapping layers that can hinder the flow of electrons. The electrical conductivity values of PVDF/TPU/CB-PPy 38/62 vol% composites containing 5% and 6 wt% of CB-PPy 3D printed samples are one to seven orders of magnitude lower than those found for compression molded composites with the same composition. Even if the electrical conductivity value for PVDF/TPU 38/62 vol% compression molded composite with 6% of CB-PPy was as high as 1.94x10-1 S•m-1, the 3D printed composite with same composition showed a very low electrical conductivity of 6.01x10-8 S•m-1. On the other hand, the 3D printed co-continuous composite PVDF/TPU 50/50 vol% with 10% of filler displayed a high value of electrical conductivity of 4.14×100 S•m-1 even after the printing process. Moreover, the piezoresistive responses of the composites were investigated. For PVDF/TPU/CB-PPy 38/62 vol% composites, the compression molded and 3D printed samples with 5% and 6% of CB-PPy exhibited good piezoresistive response. However, only the composites with 6% displayed high sensitivity and gauge factor values, large pressure range and reproducible piezoresistive responses under 100 cycles for both methods. On the other hand, for PVDF/TPU/CB-PPy co-continuous composites only the compression molded sample with 5% of CB-PPy presented good and reproducible piezoresistive responses. The crystallinity and β phase content of PVDF were investigated for the composites. Althought the degree of crystallinity of the samples decreased with the addition of CB-PPy, the percentage of β phase in PVDF was increased. The piezoelectric coefficient d33 of the samples increased with the percentage of β phase. The addition of 6% or more of CB-PPy was necessary to increase significatively the piezoelectric coefficient (d33) of the composites. The β phase content and piezoelectric responses of PVDF were lower for samples prepared by FFF. Finally, as a collateral research, the electromagnetic interference shielding effectiveness (EMI-SE) were measured for all composites. Composites with higher electrical conductivity showed better shielding of the electromagnetic radiation. In addition, composites based on the co-continuous blend displayed higher EMI shielding efficiency than 38/62 vol% composites. The main mechanism of shielding was absorption for all composites. Specimens prepared by FFF displayed diminished EMI-SE responses when compared to compression molded samples. / Lo scopo di questo studio è lo sviluppo di compositi polimerici flessibili e ad elevata conducibilità elettrica tramite stampaggio a compressione e manifattura additiva (fused filament fabrication) per possibili applicazioni come materiali piezoresistivi o piezoelettrici in sensori di pressione. In particolare, sono stati preparati compositi a base di miscele di poli(vinilidene fluoruro)/poliuretano termoplastico (PVDF/TPU) come matrice e contenenti varie frazioni di nerofumo-polipirrolo (CB-PPy) come riempitivo conduttivo. Sono state utilizzate diverse tecniche di caratterizzazione al fine di valutare le proprietà meccaniche, termiche, chimiche ed elettriche, la morfologia e la stampabilità dei materiali ottenuti. In primo luogo, miscele PVDF/TPU con diverse composizioni sono state preparate mediante mescolatura allo stato fuso seguita da stampaggio a compressione. I risultati hanno mostrato che la flessibilità del PVDF viene notevolemente migliorata dall’aggiunta di TPU. Le immagini SEM hanno evidenziato il raggiungimento di una miscela co-continua per una composizione 50/50% in volume di PVDF/TPU. Le miscele composte da PVDF/TPU 38/62 vol% e la miscela co-continua di PVDF/TPU 50/50 vol% sono state selezionate come matrici per la preparazione di compositi per stampaggio a compressione e manifattura additiva al fine di ottenere un compromesso ottimale tra conducibilità, proprietà meccaniche e stampabilità. Alle miscele selezionate sono state aggiunte varie quantità di nerofumo-polipirrolo, dallo 0 al 15%, per aumentare la conducibilità elettrica dei compositi ed eventualmente fungere da additivo nucleante per la fase β cristallina del PVDF al fine di aumentarne la risposta piezoelettrica. L'aggiunta di CB-PPy ha aumentato la conduttività elettrica di tutti i compositi. Tuttavia, la conduttività elettrica dei compositi basati su miscele co-continue di PVDF/TPU 50/50% in volume era superiore a quella trovata per compositi PVDF/TPU 38/62% in volume con lo stesso contenuto di riempitivo. Infatti, la soglia di percolazione elettrica delle miscele conduttive era del 2%, mentre la soglia di percolazione elettrica dei compositi con miscele composite non continue era del 5%. Per quanto riguarda le proprietà meccaniche, l'incorporazione del riempitivo nelle mescole ha portato a materiali più rigidi con modulo elastico più elevato, allungamento a rottura inferiore e modulo conservativo più elevato. Il modulo conservativo (G') e la viscosità complessa (η*) dei compositi sono aumentate con l'aggiunta di CB-PPy. La soglia di percolazione reologica è risultata essere del 3% per PVDF/TPU/CB-PPy 38/62 vol% e dell'1% per PVDF/TPU/CB-PPy 50/50 vol%, indicando che una maggiore quantità di riempitivo potrebbe compromettere la processabilità dei compositi. L'aggiunta di CB-PPy ha comportato anche una riduzione dei valori di Tg e Tm dei compositi a causa della riduzione della mobilità delle catene polimeriche. Sulla base della conduttività elettrica e del comportamento meccanico dei compositi, sono state selezionate tre diverse composizioni per l'estrusione di filamenti da utilizzare in un processo di stampa 3D. Nel complesso, le parti stampate in 3D presentavano proprietà meccaniche ed elettriche inferiori a causa della presenza di vuoti, difetti e strati sovrapposti che possono ostacolare il flusso di elettroni. I valori di conducibilità elettrica dei compositi PVDF/TPU/CB-PPy 38/62 vol% contenenti il 5% e il 6% di CB-PPy di campioni stampati in 3D sono da uno a sette ordini di grandezza inferiori a quelli trovati per i compositi stampati a compressione con la stessa composizione. Anche se il valore di conducibilità elettrica per il composito stampato a compressione PVDF/TPU 38/62 vol% con il 6% di CB-PPy era pari a 1,94x10-1 S•m-1, il composito stampato in 3D con la stessa composizione ha mostrato un valore molto basso di conducibilità elettrica, pari a 6,01x10-8 S•m-1. D'altra parte, il composito PVDF/TPU 50/50 vol% stampato in 3D con il 10% di riempitivo ha mostrato un elevato valore di conducibilità elettrica, pari a 4,14 × 100 S•m-1, anche dopo il processo di stampa. Inoltre, sono state studiate le risposte piezoresistive dei compositi. Per i compositi PVDF/TPU/CB-PPy 38/62 vol%, i campioni stampati a compressione e stampati in 3D con il 5% e il 6% di CB-PPy hanno mostrato una buona risposta piezoresistiva. Tuttavia, solo i compositi con il 6% hanno mostrato valori di sensibilità e gauge factor elevati, ampio intervallo di pressione e risposte piezoresistive riproducibili in 100 cicli per entrambi i metodi. D'altra parte, per i compositi co-continui PVDF/TPU/CB-PPy solo il campione stampato a compressione con il 5% di CB-PPy ha presentato risposte piezoresistive adeguate e riproducibili. La cristallinità e il contenuto di fase β del PVDF sono stati studiati per i compositi. Sebbene il grado di cristallinità dei campioni diminuisca con l'aggiunta di CB-PPy, la percentuale di fase β in PVDF risulta aumentata. Il coefficiente piezoelettrico d33 dei campioni aumenta anch’esso con la percentuale di fase β. L'aggiunta del 6% o più di CB-PPy è stata necessaria per aumentare significativamente il coefficiente piezoelettrico (d33) dei compositi. Il contenuto di fase β e le risposte piezoelettriche del PVDF sono inferiori per i campioni ottenuti mediante stampa 3D. Infine, come ricerca collaterale, è stata misurata l'efficacia della schermatura contro le interferenze elettromagnetiche (EMI-SE) per tutti i compositi. I compositi con una maggiore conduttività elettrica hanno mostrato una migliore schermatura della radiazione elettromagnetica. Inoltre, i compositi basati sulla miscela co-continua hanno mostrato un'efficienza di schermatura EMI maggiore rispetto ai compositi a 38/62% in volume. Per tutti i compositi, il principale meccanismo di schermatura è l'assorbimento. I campioni preparati mediante manifattura additiva hanno mostrato risposte EMI-SE inferiori rispetto ai campioni stampati a compressione.

Im Rahmen der additiven Fertigungsplanung bietet die geometriebasierte individuelle Fertigungsparametrisierung das Potential die Fertigungsgeschwindigkeit und -qualität weiter zu steigern. Dafür ist eine automatische Erkennung der zu fertigenden Geometriennotwendig, weshalb diese Arbeit die Eignung von faltungsbasierten künstlichen neuronalen Netzen zur Detektion von Regelgeometrien in Voxeldaten untersucht. Hierfür wird ein faltungsbasierter Detektionsalgorithmus entworfen und implementiert, welcher in der Lage ist, zylindrische, sphärische und ebene Flächen zu identifizieren. Dieser Algorithmus erreicht bei der Analyse von realen CAD-Daten eine Intersection over Union von 77,99% bzw. einen Dice-Score von 87,63 %. Dies entspricht den von vergleichbaren Algorithmen erreichten Genauigkeitswerten und bestätigt die Eignung des gewählten Ansatzes.:Kurzfassung III Abstract IV Abkürzungsverzeichnis VII Formelzeichen und Symbole VIII 1 Einleitung 1 1.1 Motivation 1 1.2 Zielstellung 2 1.3 Struktur der Arbeit 2 2 Stand der Forschung 4 2.1 Überblick CAM-Prozess 4 2.2 Geometrierepräsentationen 5 2.3 Deep Learning und faltungsbasierte neuronale Netze 9 2.3.1 Grundlagen des maschinellen Lernens 9 2.3.2 Künstliche Neuronale Netze 12 2.3.3 Faltungsbasierte neuronale Netze 15 3 Anforderungen und Konzeption 24 3.1 Vorgehensmodell und Anforderungen 24 3.2 Lösungskonzept 27 3.2.1 Struktur des Algorithmus 27 3.2.2 Versuchsplanung 31 4 Datengenerierung 32 4.1 Trainings- und Validierungsdaten 32 4.2 Testdaten 37 5 Auswahl, Training und Optimierung eines CNN-Algorithmus 39 5.1 Auswahl und Vorstellung des CNN-Algorithmus 39 5.2 Training des CNN-Algorithmus 41 5.3 Optimierung des CNN-Algorithmus 46 6 Transferlernen und Postprocessing 50 6.1 Transferlernen 50 6.1.1 Generierung von Trainings- und Validierungsdaten 51 6.1.2 Auswirkungen des Transferlernens 53 6.2 Postprocessing 55 6.2.1 Ersetzen von unsicheren Klassifizierungsergebnissen 55 6.2.2 Ersetzen von Ausreißern 57 7 Diskussion der Ergebnisse 58 7.1 Erreichte Detektionsleistung 58 7.2 Beschränkungen der Implementierung 61 8 Zusammenfassung 65 8.1 Zusammenfassung 65 8.2 Ausblick 66 Literaturverzeichnis 68 Abbildungsverzeichnis 75 Tabellenverzeichnis 77 Anlagenverzeichnis 78 Selbstständigkeitserklärung 79 / In the context of additive manufacturing planning, geometry-based individual manufacturing parameterization offers the potential to further increase manufacturing speed and quality. For this purpose, an automatic detection of the geometries to be manufactured is necessary, which is why this thesis investigates the suitability of convolution-based artificial neural networks for the detection of regular geometries in voxel data. Therefore, a convolutional-based detection algorithm is designed and implemented, which is able to identify cylindrical, spherical and planar surfaces. This algorithm achieves an intersection over union detection accuracy of 77.99% and a dice score of 87.63 %, respectively, when analyzing real CAD data. This is in line with the accuracy values achieved by comparable algorithms and confirms the suitability of the chosen approach.:Kurzfassung III Abstract IV Abkürzungsverzeichnis VII Formelzeichen und Symbole VIII 1 Einleitung 1 1.1 Motivation 1 1.2 Zielstellung 2 1.3 Struktur der Arbeit 2 2 Stand der Forschung 4 2.1 Überblick CAM-Prozess 4 2.2 Geometrierepräsentationen 5 2.3 Deep Learning und faltungsbasierte neuronale Netze 9 2.3.1 Grundlagen des maschinellen Lernens 9 2.3.2 Künstliche Neuronale Netze 12 2.3.3 Faltungsbasierte neuronale Netze 15 3 Anforderungen und Konzeption 24 3.1 Vorgehensmodell und Anforderungen 24 3.2 Lösungskonzept 27 3.2.1 Struktur des Algorithmus 27 3.2.2 Versuchsplanung 31 4 Datengenerierung 32 4.1 Trainings- und Validierungsdaten 32 4.2 Testdaten 37 5 Auswahl, Training und Optimierung eines CNN-Algorithmus 39 5.1 Auswahl und Vorstellung des CNN-Algorithmus 39 5.2 Training des CNN-Algorithmus 41 5.3 Optimierung des CNN-Algorithmus 46 6 Transferlernen und Postprocessing 50 6.1 Transferlernen 50 6.1.1 Generierung von Trainings- und Validierungsdaten 51 6.1.2 Auswirkungen des Transferlernens 53 6.2 Postprocessing 55 6.2.1 Ersetzen von unsicheren Klassifizierungsergebnissen 55 6.2.2 Ersetzen von Ausreißern 57 7 Diskussion der Ergebnisse 58 7.1 Erreichte Detektionsleistung 58 7.2 Beschränkungen der Implementierung 61 8 Zusammenfassung 65 8.1 Zusammenfassung 65 8.2 Ausblick 66 Literaturverzeichnis 68 Abbildungsverzeichnis 75 Tabellenverzeichnis 77 Anlagenverzeichnis 78 Selbstständigkeitserklärung 79

This project is in collaboration with Linnaeus University and Combitech AB, and part of the research arena WARA-PS, focuses on enhancing autonomous vehicles and their societal integration. By incorporating advanced autonomous systems, sensors, and communication tools across land, air, and sea, the aim is to reduce human intervention in hazardous environments. Current watercrafts are technically equipped but suffer from low availability and high manufacturing costs. The project proposes a modular and adaptable design, allowing 3D-printing of watercraft parts to customize for specific needs. The goal is to develop a 3D-printed watercraft applicable in research, rescue missions, and societal uses. The resulting nearly full-scale watercraft features a functioning drivetrain and detailed manufacturing documentation, highlighting waterproofing and customizable technical components. Extensive testing ensured its functionality and performance under various conditions.

Although nonflexible, scaled molecular models like Pauling-Corey's and its descendants have made significant contributions in structural biology research and pedagogy, recent technical advances in 3D printing and electronics make it possible to go one step further in designing physical models of biomacromolecules: to make them conformationally dynamic. We report the design, construction, and validation of a flexible, scaled, physical model of the polypeptide chain, which accurately reproduces the bond rotational degrees-of-freedom in the peptide backbone. The coarse-grained backbone model consists of repeating amide and alpha-carbon units, connected by mechanical bonds (corresponding to phi and psi angles) that include realistic barriers to rotation that closely approximate those found at the molecular scale. Longer-range hydrogen-bonding interactions are also incorporated, allowing the chain to easily fold into stable secondary structures. This physical model can serve as the basis for linking tangible bio-macromolecular models directly to the vast array of existing computational tools to provide an enhanced and interactive human-computer interface. We have explored the boundaries of this direction at the interface of computational tools and physical models of biological macromolecules at the nano-scale. Using a CAD-biocomputational framework, we have provided a methodology to design and build physical protein models focusing on shape and dynamics. We have also developed a workflow and an interface implemented for such bio-modeling tools. This physical-digital interface paradigm, at the intersection of native state proteins (P), computational models (C) and physical models (P), provides new opportunities for building an interactive computational modeling tool for protein folding and drug design. Furthermore, this model is easily constructed with readily obtainable parts and promises to be a tremendous educational aid to the intuitive understanding of chain folding as the basis for macromolecular structure. / Ph. D.

Widespread use of solid oxide fuel cells is hindered by a lack of long-term durability of seals between metallic and ceramic components caused by thermal expansion mismatch induced cracking. A novel gas seal design incorporating an engineered thermal expansion gradient in a SrO-La₂O₃-A₂O₃-B₂O₃-SiO₂ glass matrix with a TiNiHf shape memory alloy mesh for active stress relief and crack healing is being developed. Coefficient of thermal expansion (CTE) measurements of the seal and fuel cell components shows the possibility for a thermal expansion gradient. Differential scanning calorimetry and microscopy have shown that the TiNiHf alloy has a shape memory transition in the desired range of 200-250ºC. The oxide glass partially crystallizes during thermal cycling which has been observed through X-ray diffraction and dilatometry. The CTE decreases from 9.3Ã 10-6/°C to 6.6Ã 10-6/°C after thermal cycling. Neutron diffraction data from TiNiHf /glass composite samples reveals that the TiNiHf alloy has the ability of absorbing residual stresses from a glass matrix during martensitic phase transition. There is evidence from microscopy that the glass composition is important in determining if reaction will occur with the TiNiHf alloy. The TiNiHf alloy mesh structures can be created using the 3D printing process. This process has been adapted to allow for printing of very thin wire mesh structures of Ni and NiTi powders with a more suitable binder solution. A bi-layer test fixture has been developed which will be useful for assessing leak rate through seal materials. / Master of Science

The work herein is, in part, motivated by the idea of creating optimized, actuating structures using additive manufacturing processes (AM). By developing a consistent, repeatable method for designing and manufacturing multi-material compliant mechanisms, significant performance improvements can be seen in application, such as increased mechanism deflection. There are three distinct categories of research that contribute to this overall motivating idea: 1) investigation of an appropriate multi-material topology optimization process for multi-material jetting, 2) understanding the role that manufacturing constraints play in the fabrication of complex, optimized structures, and 3) investigation of an appropriate process for embedding actuating elements within material jetted parts. PolyJet material jetting is the focus of this dissertation research as it is one of the only AM processes capable of utilizing multiple material phases (e.g., stiff and flexible) within a single build, making it uniquely qualified for manufacturing complex, multi-material compliant mechanisms. However, there are two limitations with the PolyJet process within this context: 1) there is currently a dearth of understanding regarding both single and multi-material manufacturing constraints in the PolyJet process and 2) there is no robust embedding methodology for the in-situ embedding of foreign actuating elements within the PolyJet process. These two gaps (and how they relate to the field of compliant mechanism design) will be discussed in detail in this dissertation. Specific manufacturing constraints investigated include 1) "design for embedding" considerations, 2) removal of support material from printed parts, 3) self-supporting angle of surfaces, 4) post-process survivability of fine features, 5) minimum manufacturable feature size, and 6) material properties of digital materials with relation to feature size. The key manufacturing process and geometric design factors that influence each of these constraints are experimentally determined, as well as the quantitative limitations that each constraint imposes on design. / Ph. D.