The tuning of optimization software is of key interest to researchers solving mixed integer programming (MIP) problems. The efficiency of the optimization software can be greatly impacted by the solver’s parameter settings and the structure of the MIP. A designed experiment approach is used to fit a statistical model that would suggest settings of the parameters that provided the largest reduction in the primal integral metric. Tuning exemplars of six and 59 factors (parameters) of optimization software, experimentation takes place on three classes of MIPs: survivable fixed telecommunication network design, a formulation of the support vector machine with the ramp loss and L1-norm regularization, and node packing for coding theory graphs. This research presents and demonstrates a framework for tuning a portfolio of MIP instances to not only obtain good parameter settings used for future instances of the same class of MIPs, but to also gain insights into which parameters and interactions of parameters are significant for that class of MIPs. The framework is used for benchmarking of solvers with tuned parameters on a portfolio of instances. A group screening method provides a way to reduce the number of factors in a design and reduces the time it takes to perform the tuning process. Portfolio benchmarking provides performance information of optimization solvers on a class with instances of a similar structure.

Doctor of Philosophy / Department of Statistics / Dallas E. Johnson / Over the years sampling and experimental design have developed independently with little mutual compatibility. However, many studies do (or should) involve both a sampling design and an experimental design. For example, a polluted site may be exhaustively partitioned into area plots, a random sample of plots selected, and the selected plots randomly assigned to three clean-up regimens. In this research the relationship between sampling design and experimental design is discussed and a basic review of each is given. An estimator that combines sampling and experimental design is presented and it's development explained. Properties of this estimator will be derived and some applications of the estimator will be examined. Finally, a simulation study comparing this estimator with the traditional estimator will be presented.

Orientador: Rubens Maciel Filho / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Química / Made available in DSpace on 2018-08-18T17:38:14Z (GMT). No. of bitstreams: 1 Coelho_DayanadeGusmao_M.pdf: 1694005 bytes, checksum: d8d6330083def5f9d9b5e26276c66c26 (MD5) Previous issue date: 2011 / Resumo: Atualmente, a grande parcela de produção do ácido propanóico ocorre a partir da síntese química, tendo como matéria-prima o petróleo. O conhecimento das limitações das fontes não-renováveis, sobretudo o esgotamento das matérias derivadas do petróleo, constitui o grande desafio na busca de fontes alternativas, competitivas e sustentáveis, em face do intento na preservação ambiental. O presente trabalho versa acerca do estudo de processo fermentativo, visando a minimizar a exploração desenfreada dos recursos naturais, comandada atualmente pelas fontes petroquímicas, bem como propiciar caminhos paralelos e alternativos por meio de fontes renováveis. Desta forma, este projeto tem como objetivo investigar a produção do ácido propanóico, por meio de processo biotecnológico, utilizando como matéria-prima o glicerol e o microrganismo propionibacterium acidipropionici. Neste âmbito, para a análise do sistema em batelada, será realizada a modelagem matemática do processo, por intermédio de modelos não estruturados, e a simulação do referido processo pelo método de Runge Kutta. Os parâmetros operacionais e cinéticos foram otimizados por intermédio da aplicação da técnica de planejamentos de experimentos por Metodologia de Superfície de Resposta e pelo Algoritmo Genético. Deste modo, por meio de desenvolvimento dos modelos, otimização dos parâmetros e das condições de operação podemos, determinar as condições, limitações e viabilidade do processo constituindo ferramentas fundamentais na investigação do comportamento do processo a fim de maximizar a produção do ácido propanóico possibilitando uma rota alternativa para este processo / Abstract: Currently, a large proportion of propionic acid production occurs from chemical synthesis, with the raw oil. Knowing the limitations of non-renewable sources, especially the depletion of petroleum derived feedstock, is the great challenge in finding alternative sources, competitive and sustainable in the face of intent on environmental preservation. The present paper is about the study of the fermentation process, in order to minimize the uncontrolled exploitation of natural resources, currently led by petrochemical sources, as well as provide alternative and parallel paths through renewable sources. Thus, this project aims to investigate the production of propionic acid by biotechnology process using as raw glycerol and microorganism Propionibacterium acidipropionici. In this context, to analyze the batch system will be the mathematical model of the process through unstructured models, and simulation of this process by the Runge Kutta. The operational parameters and kinetics were optimized through the application of the technique of planning of experiments by Response Surface Methodology and the Genetic Algorithm. Thus, through the development of models, optimization of parameters and operating conditions can determine the conditions, limitations and feasibility of the process constitutes the fundamental tools for investigating the behavior of the process in order to maximize production of propionic acid providing an alternative route for this process / Mestrado / Desenvolvimento de Processos Químicos / Mestre em Engenharia Química

No description available.

The thermal stability, fire retardancy and basic mechanical properties, as a function of the mass fractions of the poly(vinyl chloride) (PVC) compound ingredients, can be modelled using 2nd order Scheffé polynomials. The empirical models for each response variable can be determined using statistical experimental design. The particular models for each response variable, which are selected for predictive ability using k-fold cross validation, can be interpreted using statistical analysis of the model terms. The statistical analysis of the model terms can reveal the synergistic or antagonistic interactions between ingredients, some of which have not been reported in literature. The interaction terms in the models also mean that the effect of a certain ingredient is dependent on the mass fractions of the other ingredients. Sensitivity analysis can be used to examine the overall effect of a change in a particular formulation on the response variables. The empirical models can be used to minimise the cost of the PVC compound by varying the formulation. The optimum formulation is a function of the costs of the various ingredients and the limits which are placed on the response variables. To analyse the system as a whole, parametric analysis can be used. The number of different parametric analyses which can be done is very large and depends on the specific questions which need to be answered. Parametric analysis can be used to gain insight into the complex behaviour of the system with changing requirements, as a decision making tool in a commercial environment or to determine the completeness of the different measuring techniques used to describe the thermal stability and fire retardancy of the PVC compound. Statistical experimental design allows for the above methods to be used which leads to significant time and labour savings over attempting to reach the same conclusions using the traditional one-factor-at-a-time experiments with changes in the phr of an ingredient. It is recommended that the data generated for this investigation is analysed in more detail using the methods outlined for this investigation. This can be facilitated by making the analysis of the data (and therefore the data itself) more accessible through a usable interface. The data set itself can also be expanded to include new ingredients requiring very few additional experiments. If a PVC compound that contains none of the ingredients that were used in this investigation is of interest a new separate data set needs to be generated. This can be done by following the same procedure used in this investigation. In fact the method that is used in this investigation can be generalised to optimise the proportions of the ingredients of any mixture. / Dissertation (MEng)--University of Pretoria, 2017. / Chemical Engineering / MEng / Unrestricted

Non-facial and non-verbal methods of affective expression are essential for naturalistic social interaction in robots that are designed to be functional and lack expressive faces (appearance-constrained)such as those used in search and rescue, law enforcement, and military applications. This research identifies five main methods of non-facial and non-verbal affective expression (body movement, posture, orientation, color, and sound). From the psychology, computer science, and robotics literature a set of prescriptive recommendations was distilled for the appropriate non-facial and non-verbal affective expression methods for each of three proximity zones of interest(intimate: contact - 0.46 m, personal: 0.46 - 1.22 m, and social: 1.22 - 3.66 m). These recommendations serve as design guidelines for adding retroactively affective expression through software with minimal or no physical modifications to a robot or designing a new robot. This benefits both the human-robot interaction (HRI) and robotics communities. A large-scale, complex human-robot study was conducted to verify these design guidelines using 128 participants, and four methods of evaluation (self-assessments, psychophysiological measures, behavioral observations, and structured interviews) for convergent validity. The study was conducted in a high-fidelity, confined-space simulated disaster site with all robot interactions performed in the dark. This research investigated whether the use of non-facial and non-verbal affective expression provided a mechanism for naturalistic social interaction between a functional, appearance-constrained robot and the human with which it interacted. As part of this research study, the valence and arousal dimensions of the Self-Assessment Manikin (SAM) were validated for use as an assessment tool for future HRI human-robot studies. Also presented is a set of practical recommendations for designing, planning, and executing a successful, large-scale complex human-robot study using appropriate sample sizes and multiple methods of evaluation for validity and reliability in HRI studies. As evidenced by the results, humans were calmer with robots that exhibited non-facial and non-verbal affective expressions for social human-robot interactions in urban search and rescue applications. The results also indicated that humans calibrated their responses to robots based on their first robot encounter.

In this paper we present some numerical results concerning the identification of material parameters in linear elasticity by dealing with small deformations. On the basis of a precise example different aspects of the parameter estimation problem are considered. We deal with practical questions such as the experimental design for obtaining sufficient data for recovering the unknown parameters as well as questions of treating the corresponding inverse problems numerically. Two algorithms for solving these problems can be introduced and extensive numerical case studies are presented and discussed.

Experimental design can be vital when experiments are resource-exhaustive and time-consuming. In this work, we carry out experimental design in the Bayesian framework. To measure the amount of information that can be extracted from the data in an experiment, we use the expected information gain as the utility function, which specifically is the expected logarithmic ratio between the posterior and prior distributions. Optimizing this utility function enables us to design experiments that yield the most informative data about the model parameters. One of the major difficulties in evaluating the expected information gain is that it naturally involves nested integration over a possibly high dimensional domain. We use the Multilevel Monte Carlo (MLMC) method to accelerate the computation of the nested high dimensional integral. The advantages are twofold. First, MLMC can significantly reduce the cost of the nested integral for a given tolerance, by using an optimal sample distribution among different sample averages of the inner integrals. Second, the MLMC method imposes fewer assumptions, such as the asymptotic concentration of posterior measures, required for instance by the Laplace approximation (LA). We test the MLMC method using two numerical examples. The first example is the design of sensor deployment for a Darcy flow problem governed by a one-dimensional Poisson equation. We place the sensors in the locations where the pressure is measured, and we model the conductivity field as a piecewise constant random vector with two parameters. The second one is chemical Enhanced Oil Recovery (EOR) core flooding experiment assuming homogeneous permeability. We measure the cumulative oil recovery, from a horizontal core flooded by water, surfactant and polymer, for different injection rates. The model parameters consist of the endpoint relative permeabilities, the residual saturations and the relative permeability exponents for the three phases: water, oil and microemulsions. We also compare the performance of the MLMC to the LA and the direct Double Loop Monte Carlo (DLMC). In fact, we show that, in the case of the aforementioned examples, MLMC combined with LA turns to be the best method in terms of computational cost.

<p>Fouling is a critical issue in commercial food manufacturing. Fouling can cause biofilm formation and pose a threat to the safety of food products. Early detection of fouling can lead to informed decision making about the product’s safety and quality, and effective system cleaning to avoid biofilm formation. In this study, a Non-Intrusive Continuous Sensor (NICS) was designed to estimate the thermal conductivity of the product as they flow through the system at high temperatures as an indicator of fouling. Thermal properties of food products are important for product and process design and to ensure food safety. Online monitoring of thermal properties during production and development stages at higher processing temperatures, ~140°C like current aseptic processes, is not possible due to limitations in sensing technology and safety concerns due to high temperature and pressure conditions. Such an in-line and noninvasive sensor can provide information about fouling layer formation, food safety issues, and quality degradation of the products. A computational fluid dynamics model was developed to simulate the flow within the sensor and provide predicted data output. Glycerol, water, 4% potato starch solution, reconstituted non-fat dry milk (NFDM), and heavy whipping cream (HWC) were selected as products with the latter two for fouling layer thickness studies. The product and fouling layer thermal conductivities were estimated at high temperatures (~140°C). Scaled sensitivity coefficients and optimal experimental design were taken into consideration to improve the accuracy of parameter estimates. Glycerol, water, 4% potato starch, NFDM, and HWC were estimated to have thermal conductivities of 0.292 ± 0.006, 0.638 ± 0.013, 0.487 ± 0.009, 0.598 ± 0.010, and 0.359 ± 0.008 W/(m·K), respectively. The thermal conductivity of the fouling layer decreased as the processing time increased. At the end of one hour process time, thermal conductivity achieved an average minimum of 0.365 ± 0.079 W/(m·K) and 0.097 ± 0.037 W/(m·K) for NFDM and HWC fouling, respectively. The sensor’s novelty lies in the short duration of the experiments, the non-intrusive aspect of its measurements, and its implementation for commercial manufacturing.</p>

The presentation of the search result information has become more important ever since the rapid development and use of smartphones. This thesis investigated what makes search engine result pages efficient and effective on the smartphone screen. For this thesis an experimental study of Karolinska Institutet’s search solution was executed. Three different versions of Karolinska Institutet’s search engine result pages were developed, by researching potential features that would improve the effectiveness and efficiency. The developed versions were compared to the original search engine result pages through experiments with students from Karolinska Institutet. The quantitative results showed no statistical significance of improvement in effectiveness or efficiency. The qualitative results indicated that categories and summary texts are preferable features in a search GUI in a smartphone. Thus the results were indecisive, the author reflects and provides ideas on how to achieve clearer results. / Hur man presenterar sökresultat har blivit allt viktigare i och med dagens av smarttelefoner. Det här arbetet undersökte vad som gör sökresultatssidor effektiva när man använder en smarttelefon. I arbetet utfördes en experimentell studie av Karolinska Institutets söklösning. Tre olika versioner av Karolinska Institutets smarttelefonsgränssnitt utvecklades genom att undersöka potentiella \emph{features} (funktioner, egenskaper) som förbättrar effektiviteten. De olika utvecklade versionerna jämfördes mot det original som Karolinska Institutet använder idag. Detta gjordes genom experiment med studenter från Karolinska Institutet. De kvantitativa resultaten kunde inte visa på någon statistisk signifikant effektivitetsförbättring. De kvalitativa resultaten pekade på att kategorisering och summeringstexter är föredragna features i ett sökgränssnitt i smarttelefonen.