Die Reduzierung von Ausschuss und unnötiger Nacharbeit ist ein elementares Ziel der Fertigungsindustrie. Mit der zunehmenden Datenverfügbarkeit und den Entwicklungen auf dem Gebiet der künstlichen Intelligenz (KI) für industrielle Anwendungen, wird auch im Bereich des Radial-Axial Ringwalzens (RAW) der Einsatz des maschinellen Lernens (ML) eruiert. Die Anwendungen hier sind beispielsweise die Prozessauslegung oder die Vorhersage der Ringqualität [1]. Allerdings ist die Genauigkeit dieser Vorhersagen derzeit noch durch die Menge und Qualität der Daten begrenzt [2]. Um das überwachte Lernen zur Vorhersage der Ringqualität anzuwenden, muss eine umfangreiche Datenbasis für Gut- und Ausschussteile erzeugt werden. Eine Möglichkeit, bestehende Datenbasen zu erweitern, besteht in der Nutzung von Prozesssimulationen zur Generierung synthetischer Daten. Im Bereich des Warmringwalzens gibt es jedoch derzeit keine schnelle Simulationsmethode, mit der eine ausreichend große synthetische Datenbank von gewalzten Teilen mit Form- oder Prozessfehlern generiert werden kann. Die Forschung zum Transferlernen zwischen verschiedenen Walzwerken und Datensätzen hat die neuartige Idee hervorgebracht, das Kaltringwalzen als Untersuchungsgegenstand heranzuziehen [2]. Im Folgenden wird untersucht, inwieweit das Kaltringwalzen, als ähnlicher Prozess, für die zukünftige Übertragung von Modellen und Ergebnissen auf das RAW verwendet werden kann. Im Vergleich zum RAW wird die Umformung beim Kaltringwalzen nur durch zwei Radialwalzen erreicht und der Prozess wird bei Raumtemperatur durchgeführt. Diese vereinfachte Verfahrensweise erlaubt es, ein halb-analytisches Modell zu entwickeln, das im Vergleich zu herkömmlichen FEM-Ansätzen, bei akzeptabler Genauigkeit, viel weniger Berechnungszeit erfordert. Zudem ermöglichen die geringere Ringgröße und der einfachere Walzprozess die Durchführung umfangreicher Forschungswalzungen zur Überprüfung der Qualität der synthetischen Daten.

Reducing scrap products and unnecessary rework has always been a goal of the manufacturing industry. With the increasing data availability and the developments in the field of artificial intelligence (AI) for industrial applications, machine learning (ML) has been applied to radial-axial ring rolling (RARR) to predict product quality [1]. However, the accuracy of these predictions is currently still limited by the quantity and quality of the data [2]. In order to apply supervised learning to predict part quality and possible scrap parts, there must be plenty of datasets logged for both good and scrap parts. One suitable way to increase the number of datasets is to utilize simulation strategies to generate synthetic datasets. However, in the hot ring rolling field, there is no fast simulation method that can be used to generate a sufficiently large synthetic database of rolled parts with form or process errors. The research on transfer learning between different mills and datasets has offered a new idea of taking a cold ring rolling process as the object of study [2]. Next it will investigate the extent to which the cold ring rolling can be used as a similar process for future transfer of models and results to radial-axial ring rolling. Compared to RARR, the cold ring rolling is a process under room temperature and contains complete radial forming instead of simultaneous forming in the radial and axial directions. The simpler forming mechanism makes it possible to build a semi-analytical model, which takes much less time compared to conventional FEMapproaches under acceptable accuracies. Furthermore, the smaller ring geometry, simplified rolling process and reduced energy consumption mean that in-house experiments can be conducted to verify the quality of the synthetic data based on confidence intervals.

The reliability of a mechanical system containing electronic packages is highly affectedby the environment the system is stationed in. The difference and fluctuationsbetween the ambient temperature and the operating temperature of the electronicpackage cause accumulation of inelastic strains in the package components thusdecreasing the service life. The most common failure modes of an electronic packagehas been identified from inspection of malfunctioning machines as cracks in the solderjoint and delamination between the glue and the die. Knowledge regarding therelationships between parameters affecting these failure modes, which are importantand which are not, is of high interest when developing new and existing products. SAAB AB would like to develop a methodology using design exploration to allow forevaluation of electronic packages using nonlinear finite element methods. A surrogate model was created and parameterized with HyperMorph to be used forthree linear static variations of design of experiments, where both the performance ofthe methods themselves and the relative importance of the parameters were ofinterest. A connectivity condition was also implemented to allow for relativemovement between components while keeping the mesh intact. The designexploration was executed using a Taguchi design, a Modified extensive latticesequence design and a fractional factorial design where the three methods werecompared as well as the parameter significance analysed. An optimization was thenperformed to find the optimal parameter settings within the allowed bounds to beused where a nominal model and an optimized model are evaluated with animplemented creep law. The fatigue life of the two models were then estimated. / Tillförlitligheten hos ett mekaniskt system med elektroniska kretsar påverkas starkt av miljön systemet används i. Skillnader och fluktuationer mellan omgivningens temperatur och arbetstemperaturen för de elektroniska kretsarna orsakar ackumulering av inelastiska töjningar, därmed förkortas det mekaniska systemets livstid. Dem vanligaste fel-moderna för en elektronisk krets har identifierats genom inspektion av felande maskiner som sprickbildning i lödfogarna och delaminering mellan processorn och dess lim. Kunskap hur förhållandet mellan parametrar som påverkar dessa fel-moder, vilka som är viktiga och vilka som inte är viktiga är av högt intresse vid utveckling av nya och redan existerande produkter. SAAB AB vill utveckla en metodik som utnyttjar statistisk försöksplanering för analyserande av elektroniska kretsar med hjälp av olinjära finita element metoder för att kunna spegla dess beteende på ett realistiskt sätt. En surrogatmodell skapades och parametriserades med hjälp av HyperMorph för att användas inom tre statiskt linjära varianter av statistisk försöksplanering, där både metodens prestanda och den relativa påverkan från parametrarna var av intresse. Ett kontaktvillkor implementerades för att tillåta relativ rörelse mellan komponenter samtidigt som nätet av finita element hölls intakt. Försöksplaneringsimuleringar utfördes med en Taguchi design, en Modified extensive lattice sequence design och en fractional factorial design, där de tre metoderna jämfördes mot varandra samt analyserades vad gäller respektive parametersignifikansen. Med optimering fanns sedan en optimal modell för att kunna jämföras med en nominell modell där en kryplag implementerades i lödfogen. Livslängden beräknades sedan för båda modeller.

[pt] A Cannabis é um insumo medicinal histórico e em atual expansão para as mais diversas aplicações medicinais, cosméticas, recreativas e têxteis. O mercado medicinal de Cannabis se encontra em destaque no cenário global, principalmente na forma de apresentação de extratos oleosos. Os extratos herbais de Cannabis (CHE) são um dos produtos de maior interesse e mais procurados para abordagens terapêuticas de uma diversidade de condições clínicas. Parâmetros de controle de qualidade regulados e métodos padronizados de avaliação de risco são atualmente demandados para produtos à base de Cannabis. O potencial medicinal da Cannabis é atribuído principalmente à biossíntese de uma classe especial de metabólitos: os fitocannabinoides. O canabidiol (CBD), o tetrahidrocanabinol (THC) e o canabinol (CBN) são destacados como os principais fitocanabinoides alvos de preocupação farmacêutica. Além disso, o monitoramento de impurezas e adjuvantes, como o teor de metais e metaloides, também é fundamental para garantir a segurança e a integridade destes produtos. Técnicas analíticas avançadas retratam alternativas poderosas para o monitoramento de produtos à base de Cannabis. No entanto, etapas analíticas adicionais precisam ser otimizadas criticamente para acompanhar o desempenho instrumental e o Design de Experimentos (DoE) fornece uma abordagem rápida, simples, confiável e eficaz para alcançar otimizações multivariadas bem-sucedidas. Neste trabalho apresentamos o desenvolvimento de dois métodos otimizados por DoE para análise de CHE: um método de quantificação de CBD, THC e CBN por UHPLC-HRMS/MS e três métodos de determinação multielementar por ICP-MS. Para a quantificação de fitocannabinoides, as condições instrumentais foram otimizadas frente a um planejamento do tipo Plackett-Burman para 7 variáveis, buscando-se otimizar a reprodutibilidade do fenômeno de ionização. Valores de desvio-padrão relativo de 2 por cento, 2 por cento e 5 por cento foram alcançados para CBD, THC e CBN, respectivamente. Além disso, planejamentos do tipo Fatorial Completo e Box-Behnken foram utilizadas para propor um protocolo otimizado de extração líquido-líquido assistida por ultrassom com 6,9 mL de metanol:hexano 9:1 v/v, 18 min de tempo de agitação e 25 min de tempo de sonicação. O modelo preditivo construído foi validado, apresentando valores de acurácia entre 86 e 120 por cento. O desempenho analítico foi validado por diretrizes farmacêuticas brasileiras de referência (ANVISA RDC 166/2017) frente a três diferentes abordagens de calibração: calibração externa, adição-padrão e Matrix Matching. Valores satisfatórios de exatidão, precisão, sensibilidade, linearidade e efeito de matriz foram alcançados com a utilização deste último, sendo representativo de uma alternativa eficiente, de maior custobenefício e de maior frequência analítica. A aplicação desta metodologia em um lote de 4 amostras reais revelou uma preocupação significativa em relação à avaliação de risco desses produtos, sendo observada uma discrepância significativa entre a descrição do rótulo e o conteúdo quantificado de CBD (mais de 10.000 vezes menor). THC e CBN não foram encontrados acima do Limite de Quantificação para nenhuma das amostras. Fenômeno que compromete não só seu potencial terapêutico, mas também revelando um ponto cego da segurança do consumidor. Para a determinação multielementar, parâmetros instrumentais atrelados ao plasma e à introdução de amostra foram otimizados por planejamentos do tipo Composto Central para maximização de sensibilidade e minimização de interferências, alcançando-se condições de compromisso com taxas de otimização globais superiores a 80 por cento. Quanto ao preparo de amostra, três métodos foram explorados: digestão ácida aberta em chapa de aquecimento, digestão ácida em vaso fechado e diluição direta com solvente orgânico, assim como três diferentes abordagens de calibração: calibração externa, Matrix Matching e adição-padrão. O desempenho de todos os métodos foi criticamente avaliado em relação à exatidão, precisão, sensibilidade, efeito de matriz e impacto ecológico. O método empregando decomposição em chapa de aquecimento com 6,9 mL de HNO3 diluído 10 por cento v/v com aquecimento por 60 min àa 100 graus C e uma abordagem de calibração por Matrix Matching forneceu o melhor desempenho geral e foi aplicado para analisar um lote de 6 amostras de Cannabis em comparação com óleo de gergelim, um óleo vegetal de consumo comum e muito utilizado como veículo farmacotécnico nos extratos de Cannabis. Em geral, foram determinados baixos teores de metais e metaloides e, comparando os extratos de Cannabis com o óleo vegetal, diferenças estatisticamente significativas foram identificadas apenas para Au, Cu, K, Li, Mg, Mn, Ni, Pb, Ti e Zn. O Pb foi encontrado em níveis mais altos em todas as amostras de Cannabis, variando de 11,7 a 12,4 (micro)g.g-1, em desacordo (teor quase 3 vezes maior) com as diretrizes da FDA para elementos potencialmente tóxicos. Por sua vez, Li, Mg, Mn, Ni, Ti e Zn foram encontrados em níveis discordantes entre as amostras, sugerindo uma heterogeneidade relevante na produção dos óleos e um controle de qualidade não padronizado para esses produtos. / [en] Cannabis herbal extracts (CHE) are one of the most interesting and sought products for therapeutic approaches of a diverse number of clinical conditions. Regulated quality control parameters and risk assessment methods are needed for Cannabis-based products. Advanced analytical techniques portray powerful alternatives to Cannabis-based products monitoring. However, further analytical steps need to be critically optimized to keep up with instrumental performance. In this work, analytical methods for phytocannabinoids quantification by UHPLCHRMS/MS and multielement determination by ICP-MS, both in CHE, were developed and optimized by Design of Experiments. For UHPLC-HRMS/MS, an ultrasound-assisted liquid-liquid extraction with methanol:hexane 9:1 v/v was proposed, analytical performance was successfully validated by gold-standard brazilian pharmaceutical guidelines, proving its efficiency in question. For multielement determination, three sample preparation methods were explored (acid decomposition in open-vessel, acid decomposition in closed-vessel, and organic solvent direct dilution). Their performances were critically evaluated regarding analytical metrics, ecological impact and user-friendliness. The open-vessel method with diluted HNO3 provided the overall best performance and was applied to analyze 6 CHEs and one sesame oil sample. The phytocannabinoids quantification suggested a major discrepancy between CHE label description and quantified content, as CBD was over 10,000-times lower and both THC and CBN could not be determined. Also, in general, low metal and metalloid contents were determined, but significant potentially toxic metals content was found. By comparing with sesame oil, statistically significant differences were identified only for Au, Cu, K, Li, Mg, Mn, Ni, Pb, Ti, and Zn. Lead were found at higher levels for all Cannabis samples in a range 11.7 to 12.4 (micro)g g-1 , in disagreement with FDA guidelines for potentially toxic elements (10 (micro)g g-1 ). Li, Mg, Mn, Ni, Ti and Zn were found at discordant levels between samples, suggesting a relevant heterogeneity and non- standardized quality control for these products.

Master of Science / Department of Animal Science and Industry / David M. Grieger / Controlling reproductive cycles during active cyclicity and seasonal anestrous in small ruminants is critical for profitability. The objective of this study was to evaluate the effect on estrous response and interval to estrus of two CIDR protocols in sheep and goats during breeding and non-breeding seasons. In experiment 1, 133 ewes were randomly assigned to 1 of 3 treatments during the breeding season. In the CIDR-7 group, ewes received a CIDR insert for 7 d. In the CIDR-7 + PGF treatment, ewes received a CIDR insert for 7 d and 20 mg of prostaglandin-F[subscript]2[subscript]α (PGF[subscript]2[subscript]α) upon CIDR removal. Ewes in the CIDR-14 treatment received a CIDR insert for 14 d. Following CIDR removal all ewes were exposed to a ram every 12 h until breeding. There was a shorter interval from CIDR removal to estrus in the CIDR-14 treatment compared to the CIDR-7 and CIDR-7 + PGF treatments (P<0.05). There was no difference in number of ewes per treatment displaying estrus. In experiment 2, 54 ewes were randomly assigned to one of two treatment groups during the anestrous season. Ewes in CIDR-7 and CIDR-14 treatments received a CIDR insert for 7 d and 14 d, respectively. Upon CIDR removal ewes were exposed to a ram every 12 h until breeding. There was a significantly shorter interval from CIDR removal to estrus in CIDR-14 ewes when compared with CIDR-7 ewes (P<0.05). For experiment 3, 37 Boer does were randomly assigned to one of two treatments. In the CIDR-10 treatment, does received a CIDR insert for 10 d and 20 mg of PGF[subscript]2[subscript]α at time of CIDR removal. In the CIDR-19 treatment, does received a CIDR insert for 19 d. Upon CIDR removal, does were exposed to a buck fitted with a marking harness and chalk marks were recorded every 12 h. The number of does displaying estrus was not different (CIDR-7, 85%; CIDR-14, 95%). There was no difference in interval from CIDR removal to estrus between treatments. Results from experiments 1 and 2 supported the hypothesis that long-term protocols yield a shorter interval to estrus when compared with short-term protocols.

This project aims to investigate the effect of varying different laser cutting parameters on the laser cut surface. The parameters that were varied were cutting speed, beam effect, gas pressure and focal point. The statistical method Design of Experiment was used to plan the experiments and two full factorial test plans were set up, one with argon as cooling and protective gas and one with nitrogen. Sheet metal consisting of the superalloy HAYNES ® 282 with a thickness of 2,54 mm was investigated. Samples from the material were cut with different laser cutting parameters according to the test plans. The samples were evaluated by measuring cracks, recast layer thickness, surface irregularity, burr height and distance to beam deflection point on the cut surface. The resulting data was inserted as responses into the statistical program Modde 10.1 for analysis. Modde 10.1 provided models of high significance for the responses of Recast layer, Tav; Recast layer, Tmax; Burrheight, max and Distance to deflection point for the argon series and Recast layer, Tav and Recast layer, Tmax for the nitrogen series. The results showed that both recast layer thickness and burr height decreases with increased values of cutting speed, gas pressure and focal point as well as decreased values of beam effect within the parameter window investigated. They also showed that distance to deflection point increased with increasing values of gas pressure and focal point and with decreased values of cutting speed and beam effect within the parameter window. The results also showed that for the samples cut with argon as cooling and assistant gas the parameters that affects the recast layer thickness the most are focal point and gas pressure, whereas for the nitrogen series it is gas pressure and beam effect. The parameter with greatest effect on burr height is the cutting speed and for the distance to deflection point it is focal point and gas pressure that have the greatest effect. When comparing the argon and the nitrogen series to each other it can be seen that the argon series generate less recast layer from the entry of the cut until between 60 – 80 % into the cut, after which the argon series bypasses the nitrogen series in amount of recast layer produced in the cutting process in the majority of the places measured. / Projektet syftar till att undersöka effekten på den laserskurna ytan då laserskärningsparametrar varieras. Parametrarna som varierades var skärhastighet, lasereffekt, gastryck och fokalpunkt. Den statistiska metoden Design of Experiments användes för att planera experimenten. Två stycken fullskaligt faktoriella försöksplaner skapades, en med argon som skärgas och en med kväve som skärgas. Undersökningen gjordes på plåt bestående av superlegeringen HAYNES ® 282 med en tjocklek på 2,54 mm. Provbitar från materialet skars ut utifrån försöksplanerna med olika laserinställningar under skärprocessen. Provbitarna utvärderades genom att mäta sprickor, tjockleken på det omsmälta materialet, ytojämnheten, gradhöjden och avståndet till laserstrålens avböjningspunkt på den laserskurna ytan. Resultaten användes för analys i det statistiska programmet Modde 10.1. Modde 10.1 bidrog med modeller av hög signifikans för responserna Recast layer, Tav; Recast layer, Tmax; Burrheight, max and Distance to deflection point för argonserien och Recast layer, Tav och Recast layer, Tmax för kväveserien. Resultaten visar att både tjockleken på omsmält material och gradhöjden avtar med ökande värden på skärhastighet, gastryck och fokalpunkt samt minskande värden på lasereffekt inom det undersökta parameterfönstret. De visade också att till laserstrålens avböjningspunkt på den laserskurna ytan ökade med ökande värden på gastryck och fokalpunkt och med minskande värden på skärhastighet och lasereffekt inom parameterfönstret. Resultaten visade också att de parametrar som i störst utsträckning påverkar tjockleken på det omsmälta materialet för provbitarna i argonserien är fokalpunkt och gastryck, medan det för kväveserien är gastryck och lasereffekt. Parametern som har högst inverkar på gradhöjden är skärhastigheten, medan fokalpunkt och gastryck har störst effekt på avståndet till laserstrålens avböjningspunkt på den laserskurna ytan. Om kväveserien och argonserien jämförs med varandra kan det observeras att argonserien generellt genererar mindre omsmält material än kväveserien från det att laserstrålen träder in i materialet till dess att den är mellan 60 och 80 % in i snittet, vartefter argonserien går förbi kväveserien i mängd omsmält material på majoriteten av mätställen.

Atualmente o Projeto de Experimentos (DOE) vem sendo largamente utilizado para determinar os fatores de projetos e processos mais significativos afetando uma variável resposta e para estabelecer modelos empíricos entre os fatores, entretanto este método ainda é pouco utilizado e processo de forjamento a quente. O presente trabalho analisa a influência individual e cumulativa de alguns parâmetros controláveis de um processo de forjamento a quente em matriz fechada na força de prensagem necessária para sua realização. Esta análise foi realizada utilizando a técnica de Projeto de Experimentos (DOE). Para a determinação da influência dos parâmetros de processo selecionados na variável resposta do Projeto de Experimentos (DOE), uma série de ensaios variando o lubrificante utilizado, o diâmetro da geratriz e a temperatura de forjamento foram realizados. A variável resposta para o experimento foi definida como a força de prensagem exigida do equipamento para a realização do forjamento. Para este trabalho foi utilizado o material ABNT 4140 fornecido em barras trefiladas de 28,6mm (1.1/8″) posteriormente forjado a quente em matriz fechada. Os resultados mostraram que o fator que mais influencia na força de prensagem para a situação ensaiada é o lubrificante. Os demais fatores, mesmo combinados, não apresentaram uma influência significativa na variável resposta. O presente estudo demonstra que é possível a utilização de técnicas de ajuste e definição de parâmetros de processo de forjamento a quente de maneira confiável e sem a necesside da aplicação dos complexos programas de simulação computacional e os métodos de tentativa e erro ainda presentes na indústria. Finalmente este trabalho reforça a versatilidade do Projeto de Experimentos (DOE) ainda pouco aplicado em processos de forjamento. / Nowadays the Design of Experiments (DOE) has been widely used to determine the most significant project and process factors affecting a response variable and to establish empirical models among the factors, although this method is still little used and the process of hot forging. This work analyzes the individual and cumulative influence of some controllable parameters in a closed die hot forging process in the pressing force required for its realization. This analysis was performed using a Design of Experiments (DOE) method. To determine the influence of variable factors selected in response Design of Experiments (DOE), a several tests varying the lubricant, the diameter of the billet and the forging temperature was performed. The response variable for the experiment was defined as pressing force. For this study, it was used the ABNT 4140 steel provided in drawn bars of 28,6mm (1 1/8 ″). The fator that had more influence over the pressing force was the lubricant. Other factors, whether or not combined, did not show a significant influence on the response variable. This study demonstrates that it is possible use techniques to set hot forging process parameters reliably and without necesside the application of complex computer simulation programs or the trial and error system. Finally, this work reinforces the versatility of Design of Experiments (DOE) wich is still little used in forging processes.

A retrospective mortality and cancer incidence study of former nuclear weapons assemblers from the Iowa Army Ammunitions Plant was conducted. This study examined whether or not workers at the plant exhibited higher rates of mortality or cancer as a result of their work-related activities. Potential exposures included radiation, beryllium, asbestos, and solvents. Cancer incidence was determined by calculating standardized incidence ratios (SIR) and using the Iowa population as reference. SIRs were calculated on 3,889 workers from1969-2005. Overall and cause-specific mortality was determined by calculating standardized mortality ratios (SMR) and using the U.S. and Iowa populations as reference. SMRs were calculated on 5,743 workers from 1947-2005. The SIR results showed that overall cancer incidence was lower than the Iowa population. Using the Iowa population as reference, the SMR analyses for men demonstrated excesses for all cancers (SMR 1.09, 95% CI 1.02-1.17), lung cancer (SMR 1.38, 95% CI 1.24-1.54), diseases of the respiratory system (SMR 1.15, 95% CI 1.03-1.46), mesothelioma (SMR 6.20, 95 % 1.28-18.1), asbestosis (SMR 9.28, 95% CI 1.12-33.5) and COPD (SMR 1.27, 95% CI 1.10-1.46). Significantly lower SMRs were observed stomach cancer and ischemic heart disease. For women excesses were observed for all cancers (SMR 1.41, 95% CI 1.17-1.69), lung cancer (SMR 2.47, 95% CI 1.72-3.44), ischemic heart disease (SMR 1.32, 95% CI 1.09-1.58), respiratory diseases (SMR 1.59, 95% CI 1.14-2.16), and COPD (SMR 2.47, 95% CI 1.60-3.65). Using the U.S. population, men experienced lower overall mortality while women had significantly higher overall mortality. In conclusion, the SIR portion of the study showed overall lower cancer incidence for both men and women. This may be due to the Healthy Worker Effect and the limited dates of study. There are no cancer registry data before 1969 thus missing cancers with short induction periods. Workers may have also moved out of the Iowa and had a cancer diagnosis in another state. Compared to Iowa population, there was an excess of respiratory disease deaths and deaths from lung cancer in both men and women. Considering the significant respiratory exposures workers may have experienced, further study with a nested case-control design is suggested.

<p>Luvata Sweden AB, Finspång, tillverkar tunna kopparband till fordons- och elektro-nikindustrin. Koppar har en förmåga att kemiskt reagera med den omgivande luften. Detta orsakar att kopparn missfärgas - det vill säga att metallen får en avvinkade färg. För att undvika att kopparn missfärgas under transport till kund, behandlas banden med ett missfärgningsskydd. Vid enstaka tillfällen fallerar dock detta skydd med missnöjda kunder och reklamationer som följd.</p><p>Syftet med detta examensarbete är att undersöka möjligheten att reducera problemet med missfärgning. Detta har gjorts genom att undersöka möjligheten att dels införa en nyutvecklad kontrollutrustning och dels genom att förbättra produktionsproces-sen.</p><p>Möjligheten att använda kontrollutrustningen för acceptanskontroll undersöktes grundligt. Parametrar som förekomsten av systematiska och slumpmässiga fel under-söktes. Dessa undersökningar visade att utrustningen inte är lämplig att användas som acceptanskontroll - dels för att utrustningen är svår att använda och ibland ger felaktiga resultat, men framförallt för att produktionsproverna generellt har så dåligt missfärgningsskydd att samtliga prover blir underkända.</p><p>Möjliga förbättringar av påverkbara parametrar i produktionsprocessen undersöktes. Detta utfördes med ett så kallat flerfaktorförsök (DoE). Undersökningarna visade att det finns parametrar som bör förbättras. Det är emellertid inte möjligt att införa före-slagna förbättringar utan ganska omfattande förändringar i processen.</p> / <p>Luvata Sweden AB, Finspång, manufactures thin copper strips for the auto- and elec-tronics industry. Copper has the ability to react chemically with the surrounding at-mosphere. This will cause the copper to become discolored. To avoid discoloring during transport to customers, a protecting agent is used. In some occasions this protector has failed with dissatisfied costumers and complaints as a result.</p><p>The aim for this thesis work is to investigate the possibilities to reduce the discolor-ing problem. This has been done by investigating the possibility to introduce a newly developed control equipment and by possible improvements to the production proc-ess.</p><p>The possibility to use the control equipment for acceptance control was investigated. Parameters such as presence of systematic and random errors were considered. The equipment turned out to be hard to use and occasionally returned faulty readings. The main concern is however that no production sample is good enough to pass the test. It is therefore concluded that this control equipment is not suitable for accep-tance control.</p><p>Possible improvements to controllable parameters in the production process were investigated. A Design of Experiments (DoE) was used to evaluate these tasks. The DoE indicated possible improvements to some parameters. To introduce the pro-posed improvements, some major alterations of the production process have to be done.</p>

Abstract SMT process is the most critical process to influence products quality in electronic assembly house process. Especially,the solder paste printing control is the key process to determine the yield rate of SMT quality. To own the capability to control quality of solder paste printing is the important thing what assembly house have to face it¡C For the purpose of gaining good SMT yield & solder paste printing stability ,the study use six sigma management technique & procedure (DMAIC--Define¡BMeasure¡BAnalyze¡BImprove & Control) make the main factors of solder paste printing¡Aincluding Printing speed¡BScraper pressure¡BStencil clean frequency¡Ato be the experiment factors of DOE(Design of Experiment)¡CThe DOE minimize the experiment number of times and provide a way of 2 levels factorial design combine RSM(Response Surface Methodology) experiment to get the optimization combination of factors¡¦ levels. Then evaluate how the optimization combination of factor levels to influence the quality of SMT¡C The study final found that there are a surprised & satisfied result on SMT yield improvement caused by optimization combination of factors¡¦ levels which 2 levels factorial design combine RSM experiment generated. It can provide the procedure & methodology for SMT assembly house reference to improve yield rate .