O presente trabalho compreende a execução de 16 experimentos, realizados em duplicata, visando avaliar a influência de fatores, tais como iluminância, concentração de NaH2PO4.H2O, concentração de NaNO3, fotoperíodo e comprimento de onda, além da interação entre a iluminância e os demais fatores, na produção de biomassa e lipídios pela microalga marinha Chlorella minutíssima. Os experimentos estiveram dispostos em 2 arranjos ortogonais L8 de Taguchi, cada um dos quais recebeu dados obtidos em experimentos com diferentes comprimentos de onda (470 nm, além da luz branca), nos quais eram variados, em dois níveis, a iluminância (26 klx e 13 klx), as concentrações de NaH2PO4.H2O (5 mg/L e 15 mg/L) e NaNO3 (45 mg/L e 75 mg/L) e o fotoperíodo (12: 12 horas claro: escuro e 8 ciclos diários de 2: 1 horas claro: escuro). Os experimentos se processaram em reatores de vidro com formato de prisma de base quadrada com volume útil de 1,1 L, isolados da influência da luminosidade externa. Para a luz branca, o melhor resultado para a produção de biomassa foi de 1300 mg/L, ao passo que o melhor resultado para o teor lipídico foi de 22,75% em relação à biomassa seca. Para a luz de comprimento de onda equivalente a 470 nm, o melhor resultado para a produção de biomassa foi de 1270 mg/L, ao passo que o melhor resultado para o teor lipídico foi de 19,03% em relação à biomassa seca. / This work comprises the execution of 16 experiments, performed in duplicate, in order to evaluate the influence of factors such as illuminance, NaH2PO4.H2O concentration, NaNO3 concentration, photoperiod and wavelenght, in addition to the interaction between the illuminance and the others factors, in the production of biomass and lipids by the m arine microalgae Chlorella minutíssima. The experiments were arranged in 2 Taguchi\'s L8 orthogonal arrays, each of which received data obtained in experiments with different wavelengths (470 nm, in addition to the white light), in which the illuminance (26 klx and 13 klx), the NaH2PO4.H2O (5 mg/L and 15 mg/L) and NaNO3 (45 mg/L and 75 mg/L) concentrations and the photoperiod (12: 12 hours light: dark and 8 daily cycles of 2: 1 hours light: dark) were varied, in two levels. The experiments were performed in quadrangular prism glass reactors with a useful volume of 1.1 L, isolated from the influence of external luminosity. For white light, the best result for biomass production was 1300 mg/L, while the best resulto f lipid contente was 22,75% in relation on the dry biomass. For light with a wavelength of 470 nm, the best result for biomass production was 1270 mg/L, while the best result of lipid contente was 19,03% in relation to the dry biomass

Abstract: Radio Frequency Identification (RFID) and related technologies have been touted to allow exponential improvements in supply chain logistics and management. However, many industrial users have indicated that these technologies have not provided the anticipated benefits. The two complimentary strategies required to address the RFID reliability are: to improve the reliability of RFID technology and to design the supply chain infrastructure that enables RFID. The focus of this paper is on designing the supply chain infrastructure to enable RFID by developing guidelines for “RFID Ready Facilities”. These guidelines were developed based on a set of experiments conducted in the RFID supply chain laboratory. These guidelines were developed by using Design of Experiments (DOE) to determine the operational and facility factors that impact RFID reliability. The three different packaging strategies were tested on packages, boxes and their various combinations. The main factors considered in the experiments were the following among many others: Package Orientation (PO), Tag Placement (TP), Package Placement (PP), Reader Location (RL), Box Orientation (BO), Tag Placement on Box (TPB) and Tag Placement on Package (TPP). Based on the DOE results, general guidelines were developed for RFID packaging.

Motivated by an electric power portfolio selection problem, a sampling method is developed for simulation-based robust design that builds on existing multi-objective statistical improvement methods. It uses a Bayesian surrogate model regressed on both design and noise variables, and makes use of methods for estimating epistemic model uncertainty in environmental uncertainty metrics. Regions of the design space are sequentially sampled in a manner that balances exploration of unknown designs and exploitation of designs thought to be Pareto optimal, while regions of the noise space are sampled to improve knowledge of the environmental uncertainty. A scalable test problem is used to compare the method with design of experiments (DoE) and crossed array methods, and the method is found to be more efficient for restrictive sample budgets. Experiments with the same test problem are used to study the sensitivity of the methods to numbers of design and noise variables. Lastly, the method is demonstrated on an electric power portfolio simulation code.

An analysis of the effect of passing vehicles’ characteristics and their impact on other drivers’ velocities was investigated. Three experimental studies were proposed and likely outcomes were discussed. Experiment 1 focused on the effect of passing vehicle type (SUV, sedan or truck) on driver speed. Drivers were hypothesized as going faster when the same vehicle type as they were driving passed them versus when no vehicle or a different vehicle passed them. Experiment 2 focused on the effect of passing SUV age on driver’s speed. Evidence suggests passing older SUVs will increase the driver’s speed more than new SUVs. Experiment 3 focused on the effect of passing SUV color on speed. Drivers were hypothesized to go faster when brighter colors (red and yellow) rather than cooler colors (grey and black) were painted on the vehicle.

In this paper we discuss an algorithm for the construction of D-optimal experimental designs for the parameters in a regression model when the errors have a correlation structure. We show that design points can collapse under the presence of some covariance structures and a so called nugget can be employed in a natural way. We also show that the information of equidistant design on covariance parameter is increasing with the number of design points under exponential variogram, however these designs are not D-optimal. Also in higher dimensions the exponential structure without nugget leads to collapsing of the D-optimal design when also parameters of covariance structure are of interest. However, if only trend parameters are of interest, the designs covering uniformly the whole design space are very efficient. For illustration some numerical examples are also included. (author's abstract) / Series: Research Report Series / Department of Statistics and Mathematics

Mathematical modeling is a key component of various disciplines in science and engineering. A mathematical model which represents important behavior of a real system can be used as a substitute for the real process for many analysis and synthesis tasks. The performance of model based techniques, e.g. system analysis, computer simulation, controller design, sensor development, state filtering, product monitoring, and process optimization, is highly dependent on the quality of the model used. Therefore, it is very important to be able to develop an accurate model from available experimental data. Parameter estimation is usually formulated as an optimization problem where the parameter estimate is computed by minimizing the discrepancy between the model prediction and the experimental data. If a simple model and a large amount of data are available then the estimation problem is frequently well-posed and a small error in data fitting automatically results in an accurate model. However, this is not always the case. If the model is complex and only sparse and noisy data are available, then the estimation problem is often ill-conditioned and good data fitting does not ensure accurate model predictions. Many challenges that can often be neglected for estimation involving simple models need to be carefully considered for estimation problems involving complex models. To obtain a reliable and accurate estimate from sparse and noisy data, a set of techniques is developed by addressing the challenges encountered in estimation of complex models, including (1) model analysis and simplification which identifies the important sources of uncertainty and reduces the model complexity; (2) experimental design for collecting information-rich data by setting optimal experimental conditions; (3) regularization of estimation problem which solves the ill-conditioned large-scale optimization problem by reducing the number of parameters; (4) nonlinear estimation and filtering which fits the data by various estimation and filtering algorithms; (5) model verification by applying statistical hypothesis test to the prediction error. The developed methods are applied to different types of models ranging from models found in the process industries to biochemical networks, some of which are described by ordinary differential equations with dozens of state variables and more than a hundred parameters.

Centrifugal compressors are widely used in propulsion technology. As an important part of turbo-engines, centrifugal compressors increase the pressure of the air and let the pressurized air flow into the combustion chamber. The developed pressure and the flow characteristics mainly affect the thrust generated by the engine. The design of centrifugal compressors is a challenging and time consuming process including several tests, computational fluid dynamics (CFD) analyses and optimization studies. In this study, a methodology on the geometry optimization and CFD analyses of the centrifugal compressor of an existing small turbojet engine are introduced as increased pressure ratio being the objective. The purpose is to optimize the impeller geometry of a centrifugal compressor such that the pressure ratio at the maximum speed of the engine is maximized. The methodology introduced provides a guidance on the geometry optimization of centrifugal impellers supported with CFD analysis outputs. The original geometry of the centrifugal compressor is obtained by means of optical scanning. Then, the parametric model of the 3-D geometry is created by using a CAD software. A design of experiments (DOE) procedure is applied through geometrical parameters in order to decrease the computation effort and guide through the optimization process. All the designs gathered through DOE study are modelled in the CAD software and meshed for CFD analyses. CFD analyses are carried out to investigate the resulting pressure ratio and flow characteristics. The results of the CFD studies are used within the Artificial Neural Network methodology to create a fit between geometric parameters (inputs) and the pressure ratio (output). Then, the resulting fit is used in the optimization study and a centrifugal compressor with higher pressure ratio is obtained by following a single objective optimization process supported by design of experiments methodology.

This thesis involves the development of digital algorithms for the microstructural analysis of metallic deposits produced through the use of Scanning Laser Epitaxy (SLE). SLE is a new direct digital manufacturing (DDM) technique which allows for the creation of three dimensional nickel-based superalloy components using an incremental layering system. Using a bed of powder placed on an underlying substrate and a laser propagating a melt-pool across the sample, a layer of material can be added and through the careful control of SLE settings various microstructures can be created or extended from the substrate. To create parts that are within specified microstructure tolerances the ideal SLE settings must be located through experimental runs, with each material needing different operating parameters. This thesis focuses on improving the microstructural analysis by use of a program that tracks various features found in samples produced through the SLE technique and a data analysis program that provides greater insights into how the SLE settings influence the microstructure. Using this program the isolation of optimal SLE settings is faster while also providing greater insights into the process than is currently possible. The microstructure recognition program features three key aspects. The first evaluates major characteristics that typically arise during the SLE process; such as sample deformation, the aspects of a single crystal deposit, and the total deposit height. The second saves the data and all relevant test settings in a format that will allow for future analysis and comparison to other samples. Finally, it features a robust yet rapid execution so it may be used for entire runs of SLE samples, which can number up to 25, within a week. The program is designed for the types of microstructure found in CMSX-4 and Rene-80, specifically single crystal and equiaxed regions. The data fitting program uses optimally piecewise-fitted equations to find relationships between the SLE settings and the microstructure traits. The data is optimally piecewise fitted as the SLE process is a two-stage procedure, establishing then propagating the melt-pool across a sample, which creates distinct microstructure transitions. Using the information gathered, graphs provide a visual aid to better allow the experimenter to understand the process and a DOE is performed using sequential analysis; allowing the previously run samples to influence the future trials, reducing the amount of materials used while still providing great insight into the parameter field. Having access to the microstructure data across the entire sample and an advanced data fitting program that can accurately relate them to the SLE settings allows the program to track and optimize features that were never before possible.

To model shrink-fitting in metal components, an analytical model for two long compound cylinders with temperature dependent material properties and interference between them is developed for calculating transient temperatures and stresses. A finite element model is developed for the same geometry which incorporated the temperature dependent material properties. A convergence study is performed on the finite element and analytical model. The finite element model is validated by comparing the approximations of finite element model with the analytical solution. In an assembly procedure of fulcrums for bascule bridges, called AP1, the trunnion is shrink-fitted into a hub, followed by shrink fitting the trunnion-hub assembly into the girder of the bridge. In another assembly procedure called AP2, the hub is shrink-fitted into the girder, followed by shrink-fitting the trunnion in the hub-girder assembly. A formal design of experiments (DOE) study is conducted on both AP1 and AP2 using the finite element model to find the influence of geometrical parameters such as radial thickness of the hub, radial interference, and various shrink-fitting methods on the design parameter of overall minimum critical crack length (OMCCL) - a measure of likelihood of failure by cracking. Using the results of DOE study conducted on both the assembly procedures, AP1 and AP2 are quantitatively compared for the likelihood of fracture during assembly. For single-staged shrink-fitting methods, for high and low hub radial thickness to hub inner diameter ratio, assembly procedure AP1 and AP2 are recommended, respectively. For fulcra with low hub radial thickness to hub inner diameter ratio and where staged shrink-fitting methods are used, for AP2, cooling the trunnion in dry-ice/alcohol and heating the girder, and for AP1, cooling the trunnion-hub assembly in dry-ice/alcohol followed by immersion in liquid nitrogen is recommended. For fulcra with high hub radial thickness to hub inner diameter ratio and where staged shrink-fitting methods are used, cooling the components in dry-ice/alcohol and heating the girder is recommended for both AP1 and AP2. Due to the limitations of AP2, assembly procedures by heating the girder with heating coils instead of dipping an already stressed trunnion-hub assembly in liquid nitrogen are studied for decreasing the likelihood of failure by cracking and yielding. In an assembly procedure called AP3-A, only the girder is heated to shrink-fit the trunnion-hub assembly in the girder. This assembly procedure AP3-A is found to be infeasible because the girder fails by yielding if heating is expected to be completed in a reasonable amount of time. An alternative assembly procedure called AP3-B is suggested for shrink-fitting where the heating of the girder is combined with cooling the trunnion-hub assembly in dry-ice/alcohol mixture. This assembly procedure AP3-B is found to be feasible. A complete DOE study is conducted on AP3-B to find the influence of parameters like hub radial thickness and radial interference at trunnion-hub interface on the design parameter of overall minimum critical crack length. The design parameter, OMCCL values during the assembly procedure AP3-B are quantitatively compared with the widely used assembly procedures (AP1 single-stage shrink-fitting and AP1 multi-staged shrink fitting). The results of this work suggest that increasing the hub radial thickness decreases the likelihood of fracture significantly. For hubs with large radial thickness, heating the girder combined with cooling the trunnion-hub in dry-ice/alcohol mixture (AP3-B) is recommended but for hubs with low radial thickness, multistage cooling of the trunnion-hub assembly in dry-ice/alcohol mixture followed by dipping in liquid nitrogen (AP1- multistage cooling) is recommended.

Nagrinėjama besiūlio įrišimo (klijavimo) kokybė. Aprašomas besiūlio įrišimo specifiškumas, klijavimo mašinos parametrai, darantys įtaką suklijavimo kokybei, ypač pirmam ir paskutiniam puslapiui. Pateikiami rezultatai klijuojant skirtingais suklijavimo parametrais. Rezultatai apdorojami eksperimentų planavimo metodu. Parodytas principas ieškant rezultatų eksperimento planavimo metodu. Darbą sudaro 8 dalys: įvadas, literatūros apžvalga, tyrimų metodika, rezultatai ir jų aptarimas, išvados ir rekomendacijos, santrumpos, literatūros sąrašas, priedai. Darbo apimtis – 57 p. teksto be priedų, 9 lentelės, 25 paveikslai, 10 bibliografiniai šaltiniai. / It is presented the research of glue binding machine. It is discribed the specificity of glue binding, the parameters, which are the most important for glue binding quality, especially for binding first and last pages. There are given results of binding, when different parameters are chosen. It is also shown the principle of counting by method of planning experiment. Thesis consist of: 57 p. text without appendixes, 9 tables,25 pictures, 10 bibliographical entries.