The fate and behaviour of hydrophobic organic compounds (e.g. pesticides) in soils are largely controlled by sorption processes. Recent findings suggest that the chemical properties of soil organic carbon (OC) significantly control the extent of sorption of such compounds in soil systems. However, currently there is no practical tool to integrate the effects of OC chemistry into sorption predictions. Therefore, the K [subscript]oc model, which relies on the soil OC content (foc), is used for predicting soil sorption coefficients (K[subscript]d) of pesticides. The K[subscript]oc model can be expressed as K[subscript]d = K[subscript]oc × foc, where K[subscript]oc is the OC-normalized sorption coefficient for the compound. Hence, there is a need for a prediction tool that can effectively capture the role of both the chemical structural variation of OC as well as foc in the prediction approach. Infrared (IR) spectroscopy offers a potential alternative to the K[subscript]oc approach because IR spectra contain information on the amount and nature of both organic and mineral soil components. The potential of mid-infrared (MIR) spectroscopy for predicting K[subscript]d values of a moderately hydrophobic pesticide, diuron, was investigated. A calibration set of 101 surface soils from South Australia was characterized for reference sorption data (K[subscript]d and K[subscript]oc) and foc as well as IR spectra. Partial least squares (PLS) regression was employed to harness the apparent complexity of IR spectra by reducing the dimensionality of the data. The MIR-PLS model was developed and validated by dividing the initial data set into corresponding calibration and validation sets. The developed model showed promising performance in predicting K[subscript]d values for diuron and proved to be a more efficacious than the K[subscript]oc model. The significant statistical superiority of the MIR-PLS model over the K[subscript]oc model was caused by some calcareous soils which were outliers for the K[subscript]oc model. Apart from these samples, the performance of the two compared models was essentially similar. The existence of carbonate peaks in the MIR-PLS loadings of the MIR based model suggested that carbonate minerals may interfere or affect the sorption. This requires further investigation. Some other concurrent studies suggested excellent quality of prediction of soil properties by NIR spectroscopy when applied to homogenous samples. Next, therefore, the performance of visible near-infrared (VNIR) and MIR spectroscopy was thoroughly compared for predicting both foc and diuron K[subscript]d values in soils. Some eleven calcareous soils were added to the initial calibration set for an attempt to further investigate the effect of carbonate minerals on sorption. MIR spectroscopy was clearly a more accurate predictor of foc and K[subscript]d in soils than VNIR spectroscopy. Close inspection of spectra showed that MIR spectra contain more relevant and straightforward information regarding the chemistry of OC and minerals than VNIR and thus useful in modelling sorption and OC content. Moreover, MIR spectroscopy provided a better (though still not great) estimation of sorption in calcareous soils than either VNIR spectroscopy or the K[subscript]oc model. Separate research is recommended to fully explore the unusual sorption behaviour of diuron in calcareous soils. In the last experiment, two dimensional (2D) nuclear magnetic resonance/infrared heterospectral correlation analyses revealed that MIR spectra contain specific and clear signals related to most of the major NMR-derived carbon types whereas NIR spectra contain only a few broad and overlapped peaks weakly associated with aliphatic carbons. 2D heterospectral correlation analysis facilitated accurate band assignments in the MIR and NIR spectra to the NMR-derived carbon types in isolated SOM. In conclusion, the greatest advantage of the MIR-PLS model is the direct estimation of Kd based on integrated properties of organic and mineral components. In addition, MIR spectroscopy is being used increasingly in predicting various soil properties including foc, and therefore, its simultaneous use for K[subscript]d estimation is a resource-effective and attractive practice. Moreover, it has the advantage of being fast and inexpensive with a high repeatability, and unlike the K[subscript]oc approach, MIR-PLS shows a better potential for extrapolating applications in data-poor regions. Where available, MIR spectroscopy is highly recommended over NIR spectroscopy. 2D correlation spectroscopy showed promising potential for providing rich insight and clarification into the thorough study of soil IR spectra. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1415416 / Thesis (Ph.D.) - University of Adelaide, School of Earth and Environmental Sciences, 2009

Made available in DSpace on 2016-12-12T20:27:38Z (GMT). No. of bitstreams: 1 Eduardo Henrique Couto.pdf: 1926472 bytes, checksum: 2b4464d74812fd9a1dc1cf69975b8f1f (MD5) Previous issue date: 2013-12-06 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / When the three phase induction motor is vector-controlled driven, is considered the mathematical model as a set of linear equations, so that the control is more effective only over an operation point. However, when changing that point, control can pass to not respond properly, causing unwanted errors of speed, position or even instability. Such behavior are due to mismatch of the real behavior of the motor, which is non-linear and time-varying system, and the mathematical model used in the project. Thus, parametric identification techniques allow the motor parameters to be determined online, tracking its time variation. The focus of the thesis is to present a general background of the subject and show the results obtained using techniques of parametric identification, more specifically exploring the least squares method to fit the parameters of the motor in closed loop. / Quando o motor de indução trifásico é acionado em malha de controle vetorial, é considerado o modelo matemático como um conjunto de equações lineares, por isso o controle é mais efetivo somente sobre este ponto de operação. Entretanto, quando se muda esse ponto, o controle pode passar a não responder mais adequadamente, causando erros indesejáveis de velocidade, posição e até mesmo a instabilidade. Tais fatos ocorrem devido à descasamentos entre o comportamento real do motor, que é não linear e variante no tempo, e o modelo matemático utilizado para o projeto. Dentro de tal contexto, as técnicas de identificação paramétrica permitem que os parâmetros do motor sejam determinados de forma online, monitorando sua variação no tempo. O foco principal deste trabalho é apresentar um cenário geral do assunto e mostrar os resultados obtidos usando as técnicas de identificação paramétrica, mais especificamente explorando o método dos mínimos quadrados para ajustar os parâmetros do motor em malha fechada

Resumo: O Sistema de Posicionamento Global oferece um poderoso e relativamente barato processo para se determinar órbitas de satélites artificiais da Terra. Este trabalho apresenta um método de determinação de órbita para satélites com um receptor GPS a bordo. Medidas de pseudo-distância são usadas para estimar o vetor de estado. O estimador considerado é o método dos mínimos quadrados recursivo, através de rotações ortogonais de Givens, com a finalidade de evitar problemas numéricos e de inversão de matrizes. É considerado a modelagem das forças devido ao geopotencial de alto grau e ordem. Resultados indicam que a precisão em posição melhor que 10 m foi obtido usando dados reais do satélite Topex (com um mínimo de duas horas de dados - aproximadamente um período orbital). O resíduo de pseudo-distância teve um desvio padrão cerca de 5 m. / Abstract: The Global Positioning System is a powerful and low cost process to compute orbits for some artificial Earth satellites. This work presents a method of orbit determination for satellites with an onboard GPS receiver. Pseudo-ranges are used in the measurements equations for the orbit estimator. The estimator considered is the recursive least squares method, numerically improved with orthogonal Givens rotations and thus avoiding problems concerning inversion of matrices. Up to high order geopotential perturbations are taken into account. Results indicate that precision better than 10 m is easily obtained using batches of one orbital period for the TOPEX satellite (two hours of orbital period). Standard deviation of about 5 m resulted for the residuals. / Orientador: Rodolpho Vilhena de Moraes / Coorientador: Hélio Koiti Kuga / Mestre

In order to meet tight product quality specifications for chemical batch processes, it is vital to monitor and control product quality throughout the batch duration. However, the frequent lack of in situ sensors for continuous monitoring of batch product quality complicates the control problem and calls for novel control approaches. This thesis focuses on the study and application of multivariate statistical methods to control product quality in chemical batch processes. These multivariate statistical methods can be used to identify data-driven prediction models that can be integrated within a model predictive control (MPC) framework. The ideal MPC control strategy achieves end-product quality specifications by performing trajectory tracking during the batch operating time. However, due to the lack of in-situ sensors, measurements of product quality are usually obtained by laboratory assays and are, therefore, inherently intermittent. This thesis proposes a new approach to realise trajectory tracking control of batch product quality in those situations where only intermittent measurements are available. The scope of this methodology consists of: 1) the identification of a partial least squares (PLS) model that works as an estimator of product quality, 2) the transformation of the PLS model into a recursive formulation utilising a moving window technique, and 3) the incorporation of the recursive PLS model as a predictor into a standard MPC framework for tracking the desired trajectory of batch product quality. The structure of the recursive PLS model allows a straightforward incorporation of process constraints in the optimisation process. Additionally, a method to incorporate a nonlinear inner relation within the proposed PLS recursive model is introduced. This nonlinear inner relation is a combination of feedforward artificial neural networks (ANNs) and linear regression. Nonlinear models based on this method can predict product quality of highly nonlinear batch processes and can, therefore, be used within an MPC framework to control such processes. The use of linear regression in addition to ANNs within the PLS model reduces the risk of overfitting and also reduces the computational e↵ort of the optimisation carried out by the controller. The benefits of the proposed modelling and control methods are demonstrated using a number of simulated batch processes.

This thesis is concerned with developing a control and condition monitoring system for a class of fault tolerant actuators with high levels of redundancy. The High Redundancy Actuator (HRA) is a concept inspired by biomimetics that aims to provide fault tolerance using relatively large numbers of actuation elements which are assembled in parallel and series configurations to form a single actuator. Each actuation element provides a small contribution to the overall force and displacement of the system. Since the capability of each actuation element is small, the effect of faults within the individual element of the overall system is also small. Hence, the HRA will gracefully degrade instead of going from fully functional to total failure in the presence of faults. Previous research on HRA using electromechanical technology has focused on a relatively low number of actuation elements (i.e. 4 elements), which were controlled with multiple loop control methods. The objective of this thesis is to expand upon this, by considering an HRA with a larger number of actuation elements (i.e. 12 elements). First, a mathematical model of a general n-by-m HRA is derived from first principles. This method can be used to represent any size of electromechanical HRA with actuation elements arranged in a matrix form. Then, a mathematical model of a 4-by-3 HRA is obtained from the general n-by-m model and verified experimentally using the HRA test rig. This actuator model is then used as a foundation for the controller design and condition monitoring development. For control design, two classical and control method-based controllers are compared with an H_infinity approach. The objective for the control design is to make the HRA track a position demand signal in both health and faulty conditions. For the classical PI controller design, the first approach uses twelve local controllers (1 per actuator) and the second uses only a single global controller. For the H_infinity control design, a mixed sensitivity functions is used to obtain good tracking performance and robustness to modelling uncertainties. Both of these methods demonstrate good tracking performance, with a slower response in the presence of faults. As expected, the H_infinity control method's robustness to modelling uncertainties, results in a smaller performance degradation in the presence of faults, compared with the classical designs. Unlike previous work, the thesis also makes a novel contribution to the condition monitoring of HRA. The proposed algorithm does not require the use of multiple sensors. The condition monitoring scheme is based on least-squares parameter estimation and fuzzy logic inference. The least-squares parameter estimation estimates the physical parameters of the electromechanical actuator based on input-output data collected from real-time experiments, while the fuzzy logic inference determines the health condition of the actuator based on the estimated physical parameters. Hence, overall, a new approach to both control and monitoring of an HRA is proposed and demonstrated on a twelve elements HRA test rig.

O estudo de escoamentos de fluidos não-Newtonianos através de expansões desperta um grande interesse em pesquisadores nas diversas áreas da engenharia, devido a sua ampla aplicação em indústrias e no meio acadêmico. O objetivo principal desta Dissertação é simular problemas de escoamentos envolvendo fluidos viscoplásticos através de expansões axissimétricas abruptas. O modelo mecânico empregado é baseado nas equações de conservação de massa e de momentum para escoamentos isocóricos acoplados com a equação constitutiva de um Fluido Newtoniano Generalizada (GNL), com a função de viscosidade de Herschel-Bulkley regularizada pela equação de Papanastasiou. O modelo mecânico é aproximado por um modelo estabilizado de elementos finitos, denominado método Galerkin Mínimos-Quadrados, ou Galerkin Least-squares (GLS). Esse método (GLS) é usado a fim superar as dificuldades numéricas do modelo de Galerkin clássico: a condição de Babuška-Brezzi e a instabilidade inerente em regiões advectivas do escoamento. O método é construído adicionando termos de malha-dependentes a fim aumentar a estabilidade da formulação de Galerkin clássica sem danificar sua consistência. A formulação GLS é aplicada para estudar a influência do índice power-law, da tensão limite de cisalhamento e razão de aspecto na dinâmica do escoamento de fluidos de Herschel-Bulkley através de expansões axissimétricas abruptas de razão de aspecto 1:2 e 1:4. Os problemas que envolvem números de Reynolds desprezíveis, para uma escala do número de Herschel-Bulkley entre 0 e 100 e índice de comportamento entre 0,2 e 1,0 são apresentados. Os resultados são fisicamente detalhados e estão de acordo com a literatura. / The study of non-Newtonian fluid flows in expansions is of great interest for researchers in the several branches of engineering, due to their wide application both in industry and academy. The objective of this Dissertation is to simulate flow problems involving a viscoplastic fluid through an axisymmetric abrupt expansion. The mechanical model employed is based on the mass and momentum conservative equations for isochoric flows coupled with the Generalized Newtonian Liquid (GNL) constitutive equation, with the Papanastasiou-regularized Herschel-Bulkley viscosity function. The mechanical model is approximated by a stabilized finite element scheme, namely the Galerkin Least-squares method. This method (GLS) is used in order to overcome the numerical difficulties of the classical Galerkin method: the Babuška- Brezzi condition and the inherent instability in advective flow regions. The method is built adding mesh-dependent terms in order to increase the stability of the classical Galerkin formulation without damaging its consistency. The GLS formulation is applied to study the influence of power-law index, yield stress and aspect reason in the flow dynamics of Herschel- Bulkley fluids through an axisymmetric abrupt expansions of aspect reason 1:2 and 1:4. Problems involving negligible Reynolds numbers, for a Herschel-Bulkley number range between 0 and 100 and e power-law index range between 0.2 and 1.0 are presented. The results are physically comprehensive and are in accordance with the literature.

Essa pesquisa avalia a estética de três praças em Pelotas, considerando as variáveis associadas aos diferentes elementos físico-espaciais que compõem as praças, bem como investiga a relação entre estética e uso das mesmas. Nessa investigação levam-se em consideração as percepções de diferentes faixas etárias de usuários: adolescentes, adultos e idosos. O objetivo central é fornecer subsídios teóricos para que possam ser produzidas praças mais satisfatórias para os usuários. As três praças selecionadas como objetos de estudo estão situadas na cidade de Pelotas (RS) e possuem características que as tornam pertinentes quanto a avaliação estética. Os métodos de coleta e análise de dados fazem parte daqueles utilizados na área de pesquisa Ambiente e Comportamento. Os dados foram obtidos a partir de levantamentos de arquivo e levantamento de campo, esse incluindo observações, mapas comportamentais, levantamento das variáveis associadas aos aspectos físico-espaciais e questionários. Os resultados dessa investigação demonstram que existe relação entre a estética e o uso das praças. Além disso, destaca que dentre as variáveis consideradas, as mais importantes para a definição da satisfação geral com as praças estão associadas às características das fachadas que delimitam as praças, principalmente com a composição arquitetônica e com características associadas ao espaço aberto das praças, salientando a presença de vegetação. Torna-se evidente para o desenvolvimento de bons projetos e para a existência de usuários satisfeitos a consideração de tais variáveis. Espera-se que os dados encontrados despertem o interesse pelo desenvolvimento de outros estudos sobre o tema da estética e sua relação com o uso das praças. / This research evaluates the aesthetics of three squares in Pelotas, considering the variables associated to the different physical and spatial aspects that compose the squares, as well as investigating the relation between aesthetic and use. In this inquiry the perceptions of different ages are taken into account: adolescents, adults and elder people. The main objective is to supply theoretical subsidies to produce more satisfactory squares for the users. The three selected squares used in this study are located in the city of Pelotas (RS – BRAZIL) and have characteristics that become them pertinent for the esthetic evaluation. The data collecting methods and data analysis are part of Environment and Behavior area. The data were obtained from file and field surveys, including observations, behavior maps, survey of the variables associated to the physical and spatial aspects and questionnaires. survey of the variables associated to the physical and spatial aspects and questionnaires. The results demonstrate that there is a relation between the aesthetic and the use of the squares. Moreover, it is features that the most important variable for the definition of the general satisfaction with the squares are associated with the characteristics of the facades that delimit the squares, mainly with the architectural composition and characteristics associated with the open space of the squares, pointing out the vegetation presence. The consideration of such variables becomes evident for the development of good projects and the existence of satisfied users. One expects that this data could awake the interest for the development of other studies on the subject of the aesthetic one and its relation with the use of the squares.

A grande maioria dos fluidos encontrados na natureza se comportam como fluidos não- Newtonianos o que torna o seu estudo muito importante para diversas áreas da engenharia. Este trabalho tem como objetivo simular o problema específico de escoamentos de fluidos viscoplásticos em expansões abruptas planares com razão de aspecto de 1:4. O problema em questão se mostra interessante pois em muitos sistemas industriais são apresentados a geometria proposta para estudo associada a fluidos viscoplásticos. Foi empregado o modelo mecânico multi-campos (mult-field) baseado nas equações de conservação de massa e balanço de momentum para escoamentos isocóricos acoplados com a equação constitutiva de um Fluido Newtoniano Generalizada (GNL), associada à função de viscosidade de Herschel-Bulkley regularizada através da equação de Papanastasiou. O modelo mecânico é aproximado por um modelo estabilizado de elementos finitos, denominado método Galerkin Mínimos-Quadrados (GLS). A fim de se pesquisar os fenômenos reológicos ali presentes é feito o estudo da influência do índice de power-law na topologia de um escoamento creeping flow (Re@0) para uma vasta faixa de números de Herschel-Bulkley variando entre 0.1 e 100. Os resultados mostraram-se satisfátorios, apresentando uma forte influencia do número de Herschel-Bulkley e do índice de power law na topologia e na dinâmica do escoamento. Obteve-se uma validação do trabalho comparando-se os resultados obtidos nesta dissertação com os obtidos em artigo de grande credibilidade e aceito por toda a academia científica. / The majority of fluids found in nature behave like non-Newtonian fluids what makes their study of great importance to various areas of engineering. The present work aims to simulate the specific problem of the viscoplastic fluids flows through in planar abrupt expansion with the aspect ratio of 1:4. The problem in question is interesting because in many industrial systems are the proposal to study geometry associated with viscoplastic fluid. It employed the multi-field mechanical model based on equations of conservation of mass and momentum balance for the isochoric flow coupled with the constitutive equation of the Generalized Newtonian Liquids (GNL), associated with the function of viscosity of Herschel-Bulkley regularized by the equation of Papanastasiou. The mechanical model is approximated by a stabilized finite element model, called Galerkin Least-Squares method (GLS). In order to research the rheological phenomena present is done studying the influence of power-law index in the topology of the creeping flow (Re@0) for a wide range of numbers of Herschel-Bulkley ranging between 0.1 and 100. The results have proved satisfactory, showing a strong influence on the number of Herschel-Bulkley and power law index of the topology and dynamics of the flow. Got to be a validation of the work comparing the results with those obtained in this thesis in an article of great credibility and accepted throughout the scientific academy.

Este estudo trata de uma abordagem histórico-cultural sobre as práticas corporais e esportivas realizadas nas praças e parques públicos da cidade de Porto Alegre nas décadas de 1920 a 1940. Desde o início deste período, as praças e parques públicos se afirmaram enquanto espaços de sociabilidade e lazer dos portoalegrenses. As formas de ocupação destes espaços sociais foram sendo reformuladas em decorrência das transformações urbanas e o advento das novas práticas deflagradas pela modernidade. Nesse contexto, começaram a ser promovidas competições esportivas entre os "clubes" das praças chegando-se até os anos 1940, quando os espaços foram potencializados enquanto palco de demonstrações de Educação Física em eventos com caráter cívico. O objetivo do presente estudo é compreender como ocorreu a emergência e difusão das práticas corporais e esportivas nas praças e parques públicos de Porto Alegre nas décadas de 1920 a 1940, identificando quais as primeiras práticas introduzidas nestes espaços públicos; como foram organizadas; e que representações culturais de uma identidade nacional foram construídas. A análise documental de fontes históricas impressas revelou que a utilização das praças e parques públicos, nos mandatos dos prefeitos Otávio Rocha (1924-1928) e Alberto Bins (1928-1937), foi privilegiada no que diz respeito à promoção de práticas corporais e esportivas. Ao traçarmos o panorama das praças e parques na cidade encontramos as denominadas Praças de Desportos, que se constituíram em espaço para as práticas organizadas e dirigidas pelo Serviço de Recreação Pública (SRP) instaurado em 1926. No ano seguinte a criação da SRP evidenciou-se a existência dos clubs da praça, que tinham diretoria estabelecida e registrada, equipes federadas e participação em campeonatos municipais. Com a instauração do Estado Novo, além do controle e organização do espaço público, uma das maneiras encontradas para efetivar o projeto de sociedade controlada com a participação popular e aceitação das idéias nacionalistas, foi acionar o civismo, a educação moral e as práticas corporais e esportivas para forjar uma nova ordem social. As demonstrações de Educação Física e competições esportivas realizadas nas festas cívicas sediadas nas praças e parques públicos foram mecanismos de construção de representações culturais de uma identidade nacional. / This study deals with a historical and cultural approach about the corporal and active sports practices accomplished at the public parks and squares from Porto Alegre City in the decades of 1920 to 1940. Since the beginning of this period, the public parks and squares have asserted as sociable and leisure spaces of the inhabitants from Porto Alegre. The means to occupy theses social spaces have been reformulated in consequence of urban transformations and the coming of new pratices aggravated by modernity. In this context, many sportive competitions have started to be promoted among the clubs of the squares arriving in the 1940s, when the spaces were made possible the creation of a stage for Physical Education demonstrations in civic feature events. The objective of this present study is to understand how it has happened the emergency and diffusion of the corporal and active sports practices in the parks and squares from Porto Alegre during the decades of 1920 and 1940, by identifying the first practices introduced in those public spaces; how they were organized; and what kind of cultural representations of a national identity were created. The documental analysis from printed historical sources have shown the utilization of public parks and squares during the mandates of the Mayors Otávio Rocha (1924-1928) and Alberto Bins (1928-1937), it was privileged concerning the promotion of corporal and sports active practices. By outlining the perspective of parks and squares of the city we find the so called "Praças de Desportos", which have established in spaces to the organized practices and they are ruled by the "Serviço de Recreação Pública" (SRP) set up in 1926. One year later the creation of SRP we could notice the existence of clubs in the squares, which had an established direction and they were registered, with federate teams and the participation in municipal championships. With the establishment of the Estado Novo (New State), besides the control and organization of public space, one of the ways found to put into effect the society project controlled by the popular participation and the acceptance of nationalist ideas it was to set in motion the civism, the moral education and the corporal and active sportive practices to forge a new social order. The Physical Education demonstrations and sportive competitions accomplished in civic festivities done in public squares and parks were mechanisms of construction of cultural representations of a national identity.

A literatura de Finanças oferece evidências de que, tanto no Brasil quantos nos E.U.A., as firmas de capital aberto estão retendo cada vez mais caixa ao longo das últimas décadas. Sobre a fonte de caixa retido, há algumas evidências do importante papel da emissão de ações para as firmas estadunidenses, porém não foram localizados trabalhos relacionados sobre o caso brasileiro. Usando uma base de dados em painel com dados contábeis e de mercado de firmas listadas na BMF&BOVESPA de 1995 a 2013 e o modelo proposto por McLean (2011), estimado através do método dos mínimos quadrados, este estudo objetiva obter evidências sobre as fontes de caixa retido pelas firmas no mercado brasileiro através de análise da amostra completa e de grupos de firmas classificadas como irrestritas e restritas financeiramente. Os resultados encontrados sugerem que a emissão de ações, o endividamento e o fluxo de caixa operacional têm correlação positiva e significante com a variação em caixa e que o fluxo de caixa operacional é a principal fonte de caixa retido das firmas brasileiras. Explorou-se o papel dos motivos precaucionários sobre a decisão de retenção e foram encontradas evidências de que o grupo de firmas restritas financeiramente retém mais a partir do fluxo de caixa operacional em tempos de crescimento nestes motivos. Por último, buscou-se verificar o papel de impactos macroeconômicos sobre a decisão de retenção, com os resultados mostrando que choques exógenos afetam de maneira diferente as firmas consideradas irrestritas e restritas financeiramente. / The Finance literature offers evidence that, in Brazil as in the U.S.A., publicly traded firms are saving more and more cash over the past decades. About the source of cash savings, there is some evidence of the important role of shares issuance for US firms, but no work was found related with the Brazilian case. Using a panel database formed by accounting and market data from listed firms in BM&FBOVESPA from 1995 to 2013 and the model proposed by McLean (2011), estimated by the least squares method, this study aims to gather evidence on the sources of cash retained by companies in the Brazilian market through analysis of the full sample and groups of firms classified as financially unconstrained and constrained. The results suggest that the shares issuance, debt and operating cashflow have positive and significant correlation with the change in cash savings and that operating cashflow is the main source of cash for Brazilian firms. The role of precautionary motives for the retention decision was explored and evidence was found that in times of growth in these motives, more is retained from operating cashflow. Finally, we sought to determine the role of macroeconomic impacts on the retention decision, with the results showing that exogenous shocks affect differently the firms considered financially unconstrained and constrained.