This research explored the use of CFD simulations to study single phase flows through structured packing. Flow rates were chosen to approximate those used in the vapor phase of industrial distillation columns. The results were evaluated against experimental results obtained with the same packing model and packed height. Several novel methods were employed to quickly obtain high validity results. A high-fidelity, digital copy of an actual packing element was created in seven hours through CT scanning. The meshing strategy employed adaptive, polyhedral meshing algorithms which resulted in high quality volume meshes with 80 percent less mesh elements than would be required with traditional tetrahedral meshing. Meshing and computation were performed on the TACC clusters. The permitted meshing with up to 57 million volume cells in less than 30 hours while simulations employing a realizable k-[epsilon] model converged in approximately two days using up to 544 processors. Nitrogen simulation predictions were found to be, on average, 7 percent below experimental measurements with water simulations showing considerably more error (~40%). The error is likely attributable a discrepancy between the simulation and experimental geometries. This discrepancy is due to an oversight in sample preparation and not a flaw in the CT scanning process of geometry creation. The volume of data generated in CFD simulation was found to be very valuable for understanding and benchmarking packing performance. Streamlines and contour plots were used to analyze the variation in performance both locally and throughout the packing stack. Significant variation was observed in flow pattern, velocity distribution, and pressure profiles throughout the column. However, the joint regions were found to be most adverse to column energy efficiency. / text

Submitted by Fabio Sobreira Campos da Costa (fabio.sobreira@ufpe.br) on 2017-07-14T12:28:59Z No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) Dissertacao Bautista Nilson.pdf: 11338318 bytes, checksum: ae22c70eb1719f066a5eeb3de436c953 (MD5) / Made available in DSpace on 2017-07-14T12:28:59Z (GMT). No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) Dissertacao Bautista Nilson.pdf: 11338318 bytes, checksum: ae22c70eb1719f066a5eeb3de436c953 (MD5) Previous issue date: 2016-01-26 / CNPQ / A modelagem e a simulação numérica do transporte de solutos, como por exemplo traçadores, em meios porosos heterogêneos e anisotrópicos, tais como aquíferos e reservatórios de petróleo constituem-se num grande desafio de natureza matemática e numérica. A modelagem de falhas selantes, canais, poços inclinados, pinchouts e outras características complexas demanda o uso de malhas não-estruturadas e não-ortogonais, capazes de se adaptar naturalmente ao domínio em estudo. Os pacotes computacionais utilizados comumente na indústria do petróleo, na sua grande maioria, se baseiam no Método das Diferenças Finitas com Aproximação de Fluxo por Dois Pontos (Two-Point Flux Approximation - TPFA) e no Método de Ponderação à Montante de Primeira Ordem (First Order Upwind Method - FOU), devido a sua facilidade de implementação e sua eficiência computacional. Infelizmente, os métodos TPFA são incapazes de produzir soluções convergentes em malhas não-ortogonais ou para tensores de dispersão ou permeabilidades completos e os métodos FOU produzem soluções com difusão numérica excessiva, exigindo malhas demasiadamente refinadas para obtermos soluções confiáveis. Uma alternativa ao TPFA, e que permite o uso de tensores completos e malhas não-ortogonais, é o Método dos Elementos Finitos de Galerkin (MEF), porém este método não produz soluções localmente conservativas, o que pode ser um problema sério para a modelagem de problemas envolvendo leis de conservação, como no escoamento em meios porosos. Outra alternativa são os Métodos de Volumes Finitos (MVF). Nas suas variantes mais robustas, estes métodos são capazes de lidar com malhas poligonais quaisquer e tensores de dispersão e permeabilidades completos e com razão de anisotropia arbitrária, além de produzir aproximações discretas de alta ordem e localmente conservativas. Neste contexto, no presente trabalho, apresentamos uma formulação MVF centrado na célula para a modelagem do transporte de um traçador não-reativo num escoamento monofásico em meios porosos heterogêneos e anisotrópicos. Para a discretização dos termos elípticos, tanto da equação de pressão quanto da equação de Advecção-Dispersão-Reação (ADRE), utilizou-se um MVF com aproximação de fluxo por múltiplos pontos que faz uso do estêncil diamante (MPFA-D) e para a discretização dos termos hiperbólicos, usamos o método FOU e um MVF do tipo MUSCL (Monotone Upstream Centered Scheme for Conservation Laws). A fim de testar nossa formulação, resolvemos alguns problemas benchmark encontrados na literatura. / Modeling and numerical simulation of solutes (e.g. Tracers) in heterogeneous and anisotropic porous media such as aquifers and oil reservoirs, constitute a bigger challenge of mathematics and numerical nature. Modeling sealants faults, channels, inclined wells, pinch outs and other complex features of these geological formations demand the use of unstructured and not orthogonal meshes, able to adapt naturally to the domain under study. The computational packages used commonly in the oil industry, mostly, are based on the Finite Difference Method with Two Point Flow Approximation (TPFA) and the Amount First Order Upwind method (FOU), due to its ease of implementation and its computational efficiency. Unfortunately, TPFA methods are unable to produce conver-gent solutions in non-orthogonal meshes or in permeability or dispersion full Tensor and FOU methods produce solutions with excessive numerical diffusion, requiring excessively refined mesh to obtain reliable solutions. An interesting alternative to TPFA, which allows the use of full tensor and not orthogonal meshes, is the Galerkin Finite Element Method (FEM), but this method does not produce solutions locally conservative, which can be a serious problem for modeling problems involving conservation laws as the flow in porous media. An interesting alternative is the Finite Volume Methods (MVF). In its most robust embodiments, these methods are able to cope with any polygonal mesh and full permeability or dispersion tensors and with an arbitrary anisotropy ratio, beyond producing discrete approximations of high order and locally conservative. In this context, the present study, we present one MVF formulation cell centered to modeling the transport of a non-reactive tracer in single-phase flow in heterogeneous and anisotropic porous media. For the elliptical discretization terms, both, the pressure equation as the equation advection-dispersion-reaction (ADRE), we used The FVMF multipoint flow approximation that uses the diamond stencil (MPPA-D) and for the discretization of hyperbolic terms, we use the FOU method and an MVF type MUSCL (Monotone Upstream Centered Scheme for Conservation Laws). In order to test our formulation, we solve some benchmark problems in the literature.