With respect to the continuous growth of computing systems, the energy-efficiency requirement of their processes becomes even more important. Different configurations, implying different energy-efficiency of the system, could be used to perform the process. A configuration denotes the choice among different hard- and software settings (e.g., CPU frequency, number of threads, the concrete algorithm, etc.). The identification of the most energy-efficient configuration demands to benchmark all configurations. However, this benchmarking is time- and energy-consuming, too. This thesis explores (a) the effect of dynamic voltage and frequency scaling (DVFS) in combination with dynamic concurrency throttling (DCT) on the energy consumption of (de)compression, DBMS query executions, encryption/decryption and sorting; and (b) a generic approach to reduce the benchmarking efforts to determine the optimal configuration. Our findings show that the utilization of optimal configurations can save wavg. 15.14% of energy compared to the default configuration. Moreover, we propose a generic heuristic (fractional factorial design) that utilizes data mining (adaptive instance selection) together with machine learning techniques (multiple linear regression) to decrease benchmarking effort by building a regression model based on the smallest feasible subset of the benchmarked configurations. Our approach reduces the energy consumption required for benchmarking by 63.9% whilst impairing the energy-efficiency of performing the computational process by only 1.88 pp, due to not using the optimal but a near-optimal configuration.

Apresentam-se os resultados de uma investigação sobre os fatores que influem no comportamento de revestimentos asfálticos em relação à deformação permanente, avaliado através dos ensaios de fluência por compressão uniaxial estática e dinâmica, realizados em corpos-de-prova moldados segundo as normas do ensaio Marshall. Os fatores selecionados para estudo foram o tipo de asfalto, a distribuição granulométrica do agregado, o teor de asfalto utilizado na mistura e a tensão utilizada para realização dos ensaios de fluência. Foram testados três tipos de asfalto, CAP-20 convencional, CAP-20 modificado com 4,5% de polímero SBS e CAP-20 modificado com 20% de borracha reciclada de pneu, utilizados em quatro teores. A definição dos teores asfálticos utilizados foi realizada a partir de seis dosagens Marshall iniciais, correspondentes às combinações dos três tipos de asfalto e das duas distribuições granulométricas adotadas, referentes aos centros da faixas B e C do DNER-ES 313/97. Os ensaios de fluência foram realizados sob três diferentes tensões, 0,1 MPa, 0,4 MPa e 0,7 MPa. Para avaliação dos efeitos das variações dos fatores citados no comportamento das misturas foi empregada a técnica de planejamento e análise de experimentos fatoriais fracionários assimétricos. Modelos estatísticos, referentes ao comportamento das misturas asfálticas avaliadas com os ensaios de fluência estática e dinâmica, foram obtidos com razoáveis coeficientes de determinação. / This work presents research results about influential factors in the permanent deformation behavior of asphalt mixes, evaluated through static and dynamic uniaxial creep tests, accomplished in specimens prepared according to the Marshall mix design method. The selected factors used in the analysis are type of binder, aggregate gradation, binder content and stress level used in the creep tests. Three types of binder were tested, convencional CAP-20, CAP-20 modified with 4,5% of SBS polymer and CAP-20 modified with 20% of recycled tire rubber, used in four binder content. The definition of binder content was accomplished starting from six Marshall method, corresponding to the combinations of the three type of binder and of the two adopted gradations, regarding the centers of the gradations B and C of DNER-ES 313/97. The creep tests were accomplished under three different stress, 0,1 MPa, 0,4 MPa and 0,7 MPa. Design and analysis techniques of fractional factorial experiments were used to evaluate the effects of the variations of the above mentioned factors on the mixture behavior. Statistical models of behavior were obtained with reasonable determination coefficients from the static and dynamic creep test.

Apresentam-se os resultados de uma investigação sobre os fatores que influem no comportamento de revestimentos asfálticos em relação à deformação permanente, avaliado através dos ensaios de fluência por compressão uniaxial estática e dinâmica, realizados em corpos-de-prova moldados segundo as normas do ensaio Marshall. Os fatores selecionados para estudo foram o tipo de asfalto, a distribuição granulométrica do agregado, o teor de asfalto utilizado na mistura e a tensão utilizada para realização dos ensaios de fluência. Foram testados três tipos de asfalto, CAP-20 convencional, CAP-20 modificado com 4,5% de polímero SBS e CAP-20 modificado com 20% de borracha reciclada de pneu, utilizados em quatro teores. A definição dos teores asfálticos utilizados foi realizada a partir de seis dosagens Marshall iniciais, correspondentes às combinações dos três tipos de asfalto e das duas distribuições granulométricas adotadas, referentes aos centros da faixas B e C do DNER-ES 313/97. Os ensaios de fluência foram realizados sob três diferentes tensões, 0,1 MPa, 0,4 MPa e 0,7 MPa. Para avaliação dos efeitos das variações dos fatores citados no comportamento das misturas foi empregada a técnica de planejamento e análise de experimentos fatoriais fracionários assimétricos. Modelos estatísticos, referentes ao comportamento das misturas asfálticas avaliadas com os ensaios de fluência estática e dinâmica, foram obtidos com razoáveis coeficientes de determinação. / This work presents research results about influential factors in the permanent deformation behavior of asphalt mixes, evaluated through static and dynamic uniaxial creep tests, accomplished in specimens prepared according to the Marshall mix design method. The selected factors used in the analysis are type of binder, aggregate gradation, binder content and stress level used in the creep tests. Three types of binder were tested, convencional CAP-20, CAP-20 modified with 4,5% of SBS polymer and CAP-20 modified with 20% of recycled tire rubber, used in four binder content. The definition of binder content was accomplished starting from six Marshall method, corresponding to the combinations of the three type of binder and of the two adopted gradations, regarding the centers of the gradations B and C of DNER-ES 313/97. The creep tests were accomplished under three different stress, 0,1 MPa, 0,4 MPa and 0,7 MPa. Design and analysis techniques of fractional factorial experiments were used to evaluate the effects of the variations of the above mentioned factors on the mixture behavior. Statistical models of behavior were obtained with reasonable determination coefficients from the static and dynamic creep test.

This thesis consists of three chapters, covering topics in both the design and modeling aspects of computer experiments as well as their engineering applications. The first chapter systematically develops a new class of space-filling designs for computer experiments by splitting two-level factorial designs into multiple layers. The new design is easy to generate, and our numerical study shows that it can have better space-filling properties than the optimal Latin hypercube design. The second chapter proposes a novel modeling approach for approximating computationally expensive functions that are not second-order stationary. The new model is a composite of two Gaussian processes, where the first one captures the smooth global trend and the second one models local details. The new predictor also incorporates a flexible variance model, which makes it more capable of approximating surfaces with varying volatility. The third chapter is devoted to a two-stage sequential strategy which integrates analytical models with finite element simulations for a micromachining process.

Fractional factorial designs are a popular choice in designing experiments for studying the effects of multiple factors simultaneously. The first step in planning an experiment is the selection of an appropriate fractional factorial design. An appro- priate design is one that has the statistical properties of interest of the experimenter and has a small number of runs. This requires that a catalog of candidate designs be available (or be possible to generate) for searching for the "good" design. In the attempt to generate the catalog of candidate designs, the problem of design isomor- phism must be addressed. Two designs are isomorphic to each other if one can be obtained from the other by some relabeling of factor labels, level labels of each factor and reordering of runs. Clearly, two isomorphic designs are statistically equivalent. Design catalogs should therefore contain only designs unique up to isomorphism. There are two computational challenges in generating such catalogs. Firstly, testing two designs for isomorphism is computationally hard due to the large number of possible relabelings, and, secondly, the number of designs increases very rapidly with the number of factors and run-size, making it impractical to compare all designs for isomorphism. In this dissertation we present a new approach for tackling both these challenging problems. We propose graph models for representing designs and use this relationship to develop efficient algorithms. We provide a new efficient iso- morphism check by modeling the fractional factorial design isomorphism problem as graph isomorphism problem. For generating the design catalogs efficiently we extend a result in graph isomorphism literature to improve the existing sequential design catalog generation algorithm. The potential of the proposed methods is reflected in the results. For 2-level regular fractional factorial designs, we could generate complete design catalogs of run sizes up to 4096 runs, while the largest designs generated in literature are 512 run designs. Moreover, compared to the next best algorithms, the computation times for our algorithm are 98% lesser in most cases. Further, the generic nature of the algorithms makes them widely applicable to a large class of designs. We give details of graph models and prove the results for two classes of designs, namely, 2-level regular fractional factorial designs and 2-level regular fractional factorial split-plot designs, and provide discussions for extensions, with graph models, for more general classes of designs.

Made available in DSpace on 2014-12-17T14:07:11Z (GMT). No. of bitstreams: 1 PatriciaNM_DISSERT.pdf: 6134806 bytes, checksum: 48ee86ee041e8aea24a2f38fd4fbf43a (MD5) Previous issue date: 2013-03-01 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior / Synthetic inorganic pigments are the most widely used in ceramic applications because they have excellent chemical and thermal stability and also, in general, a lower toxicity to man and to the environment. In the present work, the ceramic black pigment CoFe2O4 was synthesized by the polymerization Complex method (MPC) in order to form a material with good chemical homogeneity. Aiming to optimize the process of getting the pigment through the MPC was used a fractional factorial design 2(5-2), with resolution III. The factors studied in mathematical models were: citric acid concentration, the pyrolysis time, temperature, time and rate of calcination. The response surfaces using the software statistica 7.0. The powders were characterized by thermal analysis (TG/DSC), x-ray diffraction (XRD), scanning electron microscopy (SEM) and spectroscopy in the UV-visible. Based on the results, there was the formation of phase cobalt ferrite (CoFe2O4) with spinel structure. The color of the pigments obtained showed dark shades, from black to gray. The model chosen was appropriate since proved to be adjusted and predictive. Planning also showed that all factors were significant, with a confidence level of 95% / Os pigmentos inorg?nicos sint?ticos s?o os mais utilizados em aplica??es cer?micas por apresentarem uma excelente estabilidade qu?mica e t?rmica e tamb?m, em geral, uma menor toxicidade para o homem e para o meio ambiente. No presente trabalho, o pigmento cer?mico preto CoFe2O4 foi sintetizado pelo M?todo de Polimeriza??o de Complexos (MPC) visando a forma??o de um material com boa homogeneidade qu?mica. Com o objetivo de otimizar o processo de obten??o do pigmento atrav?s do MPC foi utilizado um planejamento fatorial fracionado 2(5-2), com resolu??o III. Os fatores estudados na modelagem matem?tica foram: concentra??o de ?cido c?trico, tempo de pir?lise, temperatura, tempo e taxa de calcina??o. As superf?cies de resposta utilizando o programa statistica 7.0. Os p?s obtidos foram caracterizados atrav?s de an?lises t?rmicas (TG/DSC), difra??o de raios-x (DRX), microscopia eletr?nica de varredura (MEV) e espectroscopia na regi?o do UV-vis?vel. Com base nos resultados, foi verificada a forma??o da fase ferrita de cobalto (CoFe2O4) com estrutura espin?lio. A cor dos pigmentos obtidos apresentou tonalidades escuras, do preto ao cinza. O modelo escolhido foi adequado, visto que mostrou-se ajustado e preditivo. O planejamento tamb?m mostrou que todos os fatores foram significativos, com n?vel de confian?a em 95%

Solar energy is increasingly being considered a promising solution to reduce the emissions of CO2 and green house gas. The performance of solar collectors largely depends on the ability to absorb incoming solar radiation with minimal thermal radiation losses. To weigh the potential absorbed energy to thermal losses, the performance criterion (PC) can be used, calculated as PC =α−xε, where α is absorptance, ε is emittance and x is a scaling factor < 1. It has been shown by G. Vargas et al. that Co-Cr alloys excibit great potential (α = 0.98 and ε = 0.03) for use in solar concentrators. The main goal of this project is to quantify the impact of key factors (controlled input variables) on an electroplating process of Co-Cr alloys, using the design of experiment (DOE) methodology. It is part of an ongoing collaboration between Absolicon and the physics department at Umeå university. Six factors were investigated using a fractional factorial (FrF) design. Data was collected through a series of experiments where stainless steel substrates were electroplated with Co-Cr alloys. The resulting samples were analyzed in terms of α and ε as well as the quality of deposition (QD). Using the experimental results, three models were made in a DOE-software called MODDE. Models are used to correlate the factors with each response, i.e. α, ε and QD. Ideally the predictive power of the models (Q2) should be as high as possible, and at least > 0.5. The analysis of variance (ANOVA) test was used to determine the significance of the models. Based on the models, the ’Optimizer’ tool in MODDE was used to predict two set of optimum factor settings, producing two samples, S1 and S2. S1 and S2 were evaluated in terms of α, ε and QD as well as chemical composition and structural properties of the coatings. The predictive power of the models was 0.49 for α, 0.38 for ε and 0.53 for QD. The predictive power of the models were therefore limited. ANOVA-test showed that the models for α and QD were statistically significant. For all three responses the significant effects were mostly two factor interactions. All three models showed significant lack of fit (model error) as a result of high reproducibility. S1 had the best PCAbsolicon (performance criterion for Absolicons solar collectors) of all samples with 0.858. S2 was not as good, even though it was predicted to have a higher value of PCAbsolicon by MODDE. EDS, XPS and SEM measurements of samples S1 and S2 showed that the two samples were very similar in terms of chemical composition. The main difference was that the coating of S1 was more porous, and also thicker than S2, 0.81 μm compared to 0.26 μm. Even though the models showed some predictive capabilities, the impact of the factors could not be fully determined. That is due to the nature of the FrF-design, which cannot accurately determine two-factor interactions.

No description available.

In the constantly changing field of engineering and design, achieving solutions that are both resilient and optimized is crucial. This thesis introduces a robust design methodology (RDM) that integrates statistical concepts and Finite Element Analysis (FEA) to enhance the product reliability and durability. Altair Hyperstudy is used for design exploration, and it examines how changes in geometry and material properties influence the IKEA Wall-Mounting shelving unit. The objective is to efficiently parameterize geometric shapes, compare different Design of Experiments (DOE) methodologies, identify key input variables impacting outputs, and generate a response surface model based on training data. The parameterization of geometric shapes is achieved using the morphing tool in Altair Hypermesh, while the design exploration is conducted using Full factorial, Fractional factorial V, IV and III approaches with selected resolutions. A space-filling modified extensive lattice sequence design (MELS) is employed to examine the entire design space, providing input for the development of the response surface methodology (RSM). Finally, the RSM is developed for all design variables and the quality of fit is assessed. The methodology described is used to explore the design space and determine which parameters influence the system’s output response, ensuring these parameters are considered during the design phase.

Trata-se de uma investigação sobre a influência de alguns fatores no comportamento mecânico de misturas asfálticas densas quanto à estabilidade e à flexibilidade. Foram testados três tipos de ligantes, asfalto convencional, modificado com 4,5% de SBS e modificado com 20% de borracha reciclada de pneu; duas distribuições granulométricas do agregado, centros das faixas B e C do DNER (1997); quatro teores de ligante, escolhidos com base nos valores de volumes de vazios e espessuras de película almejados; três condições de envelhecimento a longo prazo, mistura não envelhecida, envelhecida em estufa ventilada a 85ºC por 5 dias e envelhecida por exposição ao tempo por 4 meses; e duas condições de envelhecimento a curto prazo, mistura não envelhecida e envelhecida em estufa ventilada a 135ºC por 4 horas. A técnica de planejamento e análise de experimentos fatoriais fracionários assimétricos foi utilizada para a consideração simultânea dos fatores citados. O comportamento mecânico das misturas foi avaliado através dos ensaios de resistência à tração, módulo de resiliência e fluência por compressão uniaxial estática e dinâmica. A análise de variância dos resultados permitiu a identificação dos fatores com influência significativa nas respostas dos ensaios. O modo como cada fator interferiu nas propriedades apresentadas pelas misturas foi estabelecido através da construção de modelos estatísticos de comportamento. Os resultados mostraram que a adição de modificadores ao asfalto melhora a resistência das misturas à fadiga e à deformação permanente. Os processos de envelhecimento aumentaram os módulos de resiliência das misturas mas diminuíram sua capacidade de recuperação elástica, o que significa queda de resistência à fadiga. Por outro lado, a resistência a deformações permanentes das misturas, inclusive daquelas compostas por asfaltos modificados, aumentou com o envelhecimento. / This work deals with the influence of some factors in the mechanical behavior of asphalt mixtures stability and flexibility. Three types of binders were tested, conventional one, modified with 4,5% of SBS and modified with 20% of recycled tire rubber. Two aggregate gradations were tested, center of B and C gradations of DNER (1997). Four binder contents were chosen based on the air voids and film thickness. Three types of long-term aging were tested, not aged, aged in a forced-draft oven for 5 days at 85ºC and aged through weather exposition for the period of 4 months. Two types of short-term aging were tested; not aged and aged in a forced-draft oven for 4 hours at 135ºC. The technique of design and analysis of asymmetric fractional factorial experiments was used for the simultaneous analysis of the factors. The mechanical behavior of the mixtures was evaluated based on indirect tensile strength test, resilient modulus test and static and dynamic creep tests. The results of the analysis of variance allowed the identification of factors with significant influence in the answers. The influence of the factors in the mixtures properties was established through statistical models of behavior. The results showed that modified binders improves the mixtures resistance in relation to fatigue and to permanent deformation. The aging processes increased the resilient modulus of the mixtures, but also decreased its capacity of elastic return, resulting in a lost of fatigue resistance. On the other hand, the permanent deformation resistance of the aging mixtures increased, including the mixtures with modified binders cases.