BOOK OF ABSTRACTS - CFDOIL BOOK OF ABSTRACTS 4th Latin American CFD Workshop Applied to the Oil and Gas Industry July 12-13, 2010 - Rio de Janeiro - Brazil Organization July 12th July

www.cfdoil.com.br

BOOK OF ABSTRACTS

4th Latin American CFD WorkshopApplied to the Oil and Gas Industry

July 12-13, 2010 - Rio de Janeiro - Brazil

Organization

BOOK OF ABSTRACTS

4th Latin American CFD Workshop Applied to the Oil and Gas Industry

July 12-13, 2010 - Rio de Janeiro - Brazil

www.cfdoil.com.br

July 12th July 13th

8:40 Plenary Session Plenary Session

10:20 Coffee Break & Poster Session

10:50 Plenary Session Plenary Session

12:20 Lunch

14:00 Special Topics Special Topics

14:40 Technical Session Technical Session

15:55 Coffee Break & Poster Session

16:40 Special Topics Special Topics

17:20 Technical Session Technical Session

PlenARy SeSSiOn

• Prof. Dr. Said Elghobashi, University of California, Irvine

• Prof. Dr. Marcus Herrmann, Arizona State University

• Dr. Herve P. Morvan, University of Nottingham

• Dr. Markus Braun, ANSYS Germany GmbH

• Dr. Sergio Butkewitsch Choze, Vale Soluções em Energia

SPeCiAl TOPiCS

The Leopoldo Américo Miguez de Mello Research & Development Center, CENPES – PETROBRAS, has been using Computational Fluid Dynamics solutions for troubleshooting and design of equipments and processes for over 10 years. The main objective of this session is to present a brief history of the usage of these tools in order to inspire new engineering applications and achieve improved products and processes.

• Daniel Fonseca de Carvalho e Silva CENPES/PDP/MC

• Guilherme da Silva, CENPES/EB-AB-G&E/CS

• Newton Reis de Moura, CENPES/PDEDS/GN

• Raphael Moura Lopes Coelho, CENPES/EB-E&P/EPEP

• Ricardo Serfaty, CENPES/EB-AB-G&E/EEQ

• Thiago Judson L. Oliveira, CENPES/PDP/TEP

• Waldir P. Martignoni, PETROBRAS - AB-RE/TR/OT

COnFeRenCe-AT-A-glAnCe

CFD OIL is well established as one of the most important discussion forums for end users and developers of Computational Fluid Dynamics software, especially the ones involved with heat and mass transfer, fluid mechanics and chemical reactions in both upstream and downstream activities. CFD OIL strongly incentives research and development both in industry and academia and also the exchange of experiences among all the involved experts.

www.cfdoil.com.br

Abstracts Index

PLENARy SESSION

MONDAy, JuLy 12TH

• HowdoInertialParticlesModifyTurbulentFlows? 1

• Multi-ScaleDetailedSimulationsofPhaseInterfaceDynamicsinTurbulentFlows 1

TuESDAy, JuLy 13TH

• IntegratedProductDevelopmentatVSE 1

• ModelingDispersedMultiphaseFlowswithLagrangianMethods 2

• ModellingFilmsandGas-LiquidSeparationinInternalFlowProblems 2

TECHNICAL SESSION

MONDAy, JuLy 12TH

• SolutionofaMultiphaseFlowinHorizontalWellsusingaDrift-FluxModel 3

• NumericalSimulationofaCycloneusedasanInletDeviceofaGravitationalSeparator 5

• CFD-DEMModelingoftheGravelPackingProcessduringPetroleumHorizontalWellCompletions 7

• MultiphaseModellingandFluidDynamicsSimulationofaBiomassGasifier 9

• EnhancedOilRecoveryinCapillaryPassagesbyInjectionofNon-NewtonianFluidwiththeCompleteWetting

oftheDisplacedFluid 11

• CFDEvaluationofaRefineryFurnaceBehaviorusingFlamelet-BasedModel 13

• AnalisysofSleeveRepairWeldingofIn-ServicePipelines 15

• ComputationalSimulationofNaturalConvectionofaFluidwithInternalHeatGeneration 17

• NumericalSimulationofaDummyWellPumpingModule 18

• ACFDStudyofanIndustrialGasTurbineCombustionChamberusingZimont’sModel 20

www.cfdoil.com.br

TuESDAy, JuLy 13TH

• SimulatingCokeDepositionintheSlopWaxColectorusingComputationalFluidDynamics 22

• 2DSimulationofLeakageandDamagedStabilityofOilCarrierbyMPSMethod 24

• SimulationofBubblingProcessesIncludingBubbleFormationEffects:IsothermalStripping 26

• CFDModeling of theCombustionGases inNon-PremixedBurners of aGroundFlare and theDispersion of

PollutantsGeneratedDuringtheProcess 28

• ModelingandSimulationofaThree-PhaseandThree-DimensionalFlowinaFCCIndustrialRiser 30

• SimulationofEnvironmentalLoadsonaSupplyBoat 32

• CFDusetoImproveDesignattheDetailedEngineeringStage:ColdBy-PassOptimization 34

• EdgeCFD-ALE:AFiniteElementSystemforComplexFluid-StructureInteractionsinOffshoreEngineering 36

• NumericalSimulationofOnshoreProcessSeparators-ResidenceTimeOptimization 38

• GasDetectorLocationUsingCFDSimulation 40

Abstracts Index

1

www.cfdoil.com.br

Monday, July 12th - 09:20

HOW DO INERTIAL PARTICLES MODIFyTuRBuLENT FLOWS?

Said Elghobashi

University of California, Irvine4200 Engineering Gateway Irvine, CA 92697-3975

e-mail: [email protected]

ABSTRACT

Particle-laden turbulent flows are ubiquitous in nature (e.g. aerosols in clouds, and dust storms on Earth and Mars) and in industrial applications (e.g. pneumatic transport, liquid fuel and pulverized coal sprays in combustion chambers). Experimental and numerical studies of these flows are quite challenging due to the wide spectra of length-and time-scales of the dispersed particles in addition to the spectra of scales intrinsic to the carrier fluid turbulence. The two-way and four-way nonlinear interactions between the dispersed particles and the turbulence result in complex multi-scale physical phenomena. The lecture focuses on the physical mechanisms of interactions between dispersed spherical particles and isotropic turbulence using Direct Numerical Simulation (DNS). Particles whose diameter is smaller than the Kolmogorov length scale of turbulence are simulated as point particles. Particles with diameter larger than the Kolmogorov length scale are fully resolved using the Immersed Boundary method.

Monday, July 12th - 11:20

MuLTI-SCALE DETAILED SIMuLATIONS OF PHASE INTERFACE DyNAMICS IN

TuRBuLENT FLOWS

Marcus Herrmann

Arizona State UniversityP.O. Box 876106, Tempe, AZ 85287-6106


ABSTRACT

The dynamics of phase interfaces play an important role in many natural phenomena and technical applications. For example, the atomization of liquid fuel

to generate a spray is typically the first in a sequence of steps that ultimately result in energy conversion in the form of combustion. As such, the dynamics of the liquid/gas interface can strongly impact combustion stability, efficiency, and pollutant production. Another example relates to wax deposition in deep water crude oil pipelines, where wax crystal growth, agglomeration, deposition on the pipeline walls and the resulting growth of the wax/liquid interface can lead to significant blockage of the flow. Common to both examples is the fact that they involve turbulent flows and are governed by processes occurring over vast ranges of both length and time scales thus posing a tremendous numerical challenge to detailed numerical simulation approaches.In this talk the requirements for numerical methods for detailed simulations of phase interface dynamics are discussed and a multi-scale modeling approach is presented to address the challenges in an efficient way. The resulting method is based on a hybrid Eulerian/Lagrangian, refined level set grid approach. It has been successfully applied to the detailed simulation of the primary atomization of Diesel jets and liquid jets in crossflow, giving new insights into the details of the atomization mechanisms. In addition to these atomization simulation results, multi-scale modeling strategies for Large Eddy Simulation sub-grid models of phase interface dynamics and the wax deposition problem will be discussed.

Tuesday, July 13th - 08:40

INTEGRATED PRODuCT DEVELOPMENT AT VSE

Sergio Butkewitsch Choze

VSE - Vale Soluções em Energia S.A Rodovia Presidente Dutra, km 138 Eugênio de Melo,São José dos Campos, SP, Brasil, 12247-004e-mail: [email protected]

ABSTRACT

This lecture will present a brief introduction to VSE and its major activities, followed by detailed description and examples on the computational framework aimed at supporting the modeling and simulation activities performed in the company. Specific explanation will be devoted to the CFD analyses and the use of open source tools to perform process and configuration management. At the concluding remarks, outlines of the next steps will be presented.

Plenary Session

www.cfdoil.com.br


MODELING DISPERSED MuLTIPHASE FLOWS WITH LAGRANGIAN METHODS

Markus Braun

Ansys Germany GmbHBirkenweg 14a, 64295 Darmstadt, Germany


ABSTRACT

An overview is given on modeling techniques for dispersed multiphase flows based on Lagrangian methods. This will cover the range from classical Euler/Lagrange models, over DEM methods, as well as hybrid methods, like the Dense Discrete Phase Model, recently presented by the authors [1], bridging the gap between Euler/Euler and Euler/Lagrange methods.

Particular emphasis is put on modeling of collisions as it becomes more important with increasing volume fraction of the dispersed phase. Another focus is put on ability of the models to consider size distributions in an efficient way as they appear in all technical multiphase flows applications.

Results were shown to discuss applicability of the various models. This includes simulation of basic hydrodynamic flow behavior as well as heat and mass exchange in dilute as well as dense multiphase flows.

REFERENCES

[1] Popoff, B.; Braun, M.; A Lagrangian Approach to Dense Particulate Flows, 6th International Conference on Multiphase Flows, Leipzig, Germany, July 2007.


MODELLING FILMS AND GAS-LIQuID SEPARATIO N IN I NTER AL FLOW PROBLEMS

Herve P. Morvan

University of NottinghamUniversity Park, Nottingham, NG7 2RD, UKe-mail: [email protected]

ABSTRACT

Modelling gas liquid flow is an ongoing challenge for a number of industries. Gas/liquid annular flow in pipes is indeed widely encountered in various industrial equipments such as nuclear reactors, chemical reactors, oil and gas production systems and gas turbines.

Some of the challenges associated with annular flow are the computation of the liquid film distribution, the mass transfer of fluid, the deposition and entrainment of droplets in these flows. A 3D CFD model has been developed by the Nottingham team to simulate annular and film flows pipes and in various chambers. In the approach presented here the liquid film is solved explicitly by means of a modified Volume of Fluid (VOF) method. The droplets are traced using a Lagrangian technique and the film to droplets (entrainment) and droplets to film (splashing, deposition) interactions are taken into account.

This work will constitute the main thrust of the presentation, though additional material including both other applications and modelling approaches will also be presented as the author endeavours to give a broader overview of the gas/liquid work he leads at the University of Nottingham and their associated challenges.

Plenary Session2

3

www.cfdoil.com.br

SOLuTION OF A MuLTIPHASE FLOW IN HORIZONTAL WELLS uSING A DRIFT-FLuX MODEL

Arthur Besen Soprano*, António Fábio Carvalho da Silva†, Clovis R. Maliska+

* UFSC - Federal University of Santa CatarinaFrancisco Goulart, nº 278, ap 14, Trindade - Florianópolis - SC/Brazil

e-mail: [email protected] †UFSC - Federal University of Santa Catarina

Florianópolis - SC/Brazile-mail: [email protected]

+ UFSC - Federal University of Santa CatarinaFlorianópolis - SC/Brazil


ABSTRACT

This works presents a procedure for solving two-phase (gas and liquid) flows along an oil well with lateral mass inflow coming from the reservoir. The flow is considered isothermal and one-dimensional. Equations are discretized using a Finite Volume Method with a C++ (OOP) code implementation. This algorithm is intended to be used with a reservoir simulator for solving the coupled flow between reservoir and well. A drift-flux model is used to model the two-phase flow and the solution is then validated with results available in the literature.From the drift flux model, the gas (disperse phase) velocity is calculated as ([3]): (1.1)where is the profile parameter, the drift velocity and is the total volumetric flux, defined as: (1.2)

The void fraction of a phase is given by: (1.3) The last equality results if we notice that, inside a control volume, all the properties are equal, so one can calculate the void fraction using the duct cross sectional area and the cross sectional area occupied by the phase ( ).The two-phase drift-flux model is based on three balance equations [2]

Mixture Continuity E.: (1.4)

Gas-phase Continuity Eq.: (1.5)

Mixture Momentum Eq.: (1.6)

where and are the total and gas mass influx (per unit volume) along the well lateral holes, respectively.

All the mixture properties (e.g. density, viscosity, etc.) are calculated by averaging values between the two

phases. The term is called modified drift velocity and can be calculated in terms of the mixture velocity

([4]):

(1.7)

Monday, July 12th - 14:40 - Room 1

Technical Session

4

www.cfdoil.com.br

The problem was solved using a Newton’s Method ([1]), generating a tridiagonal block matrix structure, since it’s a one-dimensional domain. Comparisons made with results available in [4], show that the results agreed well (Figure 1). The main advantage is that the approach presented in this work allows timesteps 100 times greater (or even more) than the maximum timesteps that could be used in the reference solution ([4] and [5]). Of course the greater timestep, the less accurate were the results of the transient solutions. Being able to use any timesteps, is an advantage not only when the steady state solution is the final goal, but also when the transient behavior is of interest. The great advantage of the method proposed herein is that one is not restricted to small time steps because of convergence. Recall that reservoir’s timestep has an order of days, while the well’s steady state is reached much earlier than that.

Figure 1 - Results obtained with 200 control volumes. Timestep equals 0.01s when not indicated. Pipe diameter: 0.1m, pipe lenght: 1000m.

REFERENCES

[1] DA CUNHA, A. R. uma Metodologia para Simulação Numérica Tridimensional de Reservatórios de Petróleo utilizando Modelo Black-Oil e Formulação em Frações Mássicas. Tese de Doutorado, Departamento de Engenharia Mecânica, Universidade Federal de Santa Catarina, Florianópolis, Brasil, 1996.[2] HIBIKI, T. ISHII, M. One-dimensional drift-flux model and constitutive equations for relative motion between phases in various two-phase flow regimes. International Journal of Heat and Mass Transfer, 2003.[3] ISHII, M. HIBIKI, T. Thermo-Fluid Dynamics of Two-Phase Flow. Springer, 2006.[4] PROVENZANO, C. E. C. Previsão numérica de escoamento bifásico em tubulações utilizando o modelo de deslizamento. Dissertação de Mestrado, PUC, Rio de Janeiro, 2007.[5] EVJE S., FJELDE K. K., 2003. On a rough AuSM scheme for a one-dimensional two-phase model. Comput. Fluids 32, 1497–1530, 2003.

Technical Session

5

www.cfdoil.com.br


NuMERICAL SIMuLATION OF A CyCLONE uSED AS AN INLET DEVICE OF A GRAVITATIONAL SEPARATOR

Lucilla Coelho de Almeida*, João Américo Aguirre Oliveira Jr.†, Carlos Alberto Capela Moraes+

*ESSSe-mail: [email protected]

†ESSSe-mail: [email protected]

+PETROBRASe-mail: [email protected]

ABSTRACT

In an oil field over its productive life, simultaneous production of gas, oil and water, along with contaminants such as sand, occurs frequently. This makes necessary the use of equipment to perform, under controlled conditions, the primary processing of the mixture, whose goal is the separation of oil, gas and water and the treatment of these currents in order to adapt them to the existing patterns, so that they can be transferred to the terminals and refineries.

The first stage of separation is located immediately downstream of the manifold of production and accomplishes the separation of the three phases: water, oil and gas. In general, gravity separators are used, where several internal devices can be present. Among them one can mention: cyclones and baffle plates, at the primary separation section; demisters at the agglutination section and structured packings at the secundary separation section.

This work is focused on the modeling and numerical simulation of the flow inside a gravitational separator in order to evaluate the influence of the presence of each of its internal devices in the separation efficiency. At this first stage, the flow inside the cyclone – which is located at the entrance of the vessel, the section known as primary separation – is investigated. The flow inside the cyclones presents the researcher rather complex and challenging characteristics, among which one mention: streamlines with high curvature, intense force field, interaction between primary and secondary flows and anisotropic turbulence. This paper presents a three-dimensional fluid dynamic study of a gas-liquid two-phase flow in a cyclone.

The base geometry studied was inserted into a CAE package and a mesh of hexahedral elements used in the solution of the flow by finite volume method was generated, as can be seen in Figure 1. The fluid fraction at the input current varies between 6 and 12% V/V. As a first approach, it was considered a two-phase flow; the gas phase being the primary phase, continuous, and a homogeneous liquid phase, dispersed with relatively large diameter, with properties obtained by the weighted average volume fraction of each fluid.

The two-phase flow was modeled using an Eulerian approach. The density and viscosity of the phases were considered constant, which corresponds to an isothermal approach. At the inlet of the cyclone, the velocity of the phases, as well as the volume fraction of liquid were prescribed. In the overflow the relative pressure of 0 [Pa] was prescribed, whereas in the underflow the hydrostatic pressure of the liquid column above it was given, since this cyclone works flooded. Finally, at the walls the non-slip condition for both phases was assumed. The SST model was used for modeling the turbulence. The domain was initialized with the normal level of fluid expected during normal operation. The pressure was also initialized this way, so that the hydrostatic pressure was taken into consideration.

The phase separation does not occur by centrifugal effect, since a rotating flow inside the device is not established, due to an ineffective inlet design. Since this centrifugal field is not generated, as expected for a cyclone, we can conclude that the device does not function as such. The gas phase can pass around the vortex finder and leaves the cyclone through the overflow. The liquid phase, on the other hand, due to its greater inertia, cannot do the same and ends up bumping into the walls of the device, as can be seen at Figure 2. It is possible to follow the formation of a liquid film as the liquid droplets collide with the walls and its movement toward the gas-liquid interface due to gravity, as shown at Figure 3 (isosurfaces of volume fraction equal 0.5). Due to this lack of rotation, the efficiency of the device is limited and it is expected that small droplets will be carried by the gas stream.

Technical Session

6

www.cfdoil.com.br

Thus, changes in the geometry of the entrance of the cyclone were proposed for the achievement of the cyclone effect in order to increase the separation efficiency.

Figure 1 - Hexahedral mesh used in the simulations. Figure 2 - Liquid phase streamlines.

Figure 3 - Isosurfaces of liquid volume fraction.

REFERENCES

[1] Drew, D. A., Mathematical modelling of two-phase flows. Annual Review of Fluid Mechanics 15 (4), pp. 261–291 (1983).[2] ANSYS CFX-Solver Modeling Guide - ANSYS CFX Release 11.0 (2009).

Technical Session

7

www.cfdoil.com.br


CFD-DEM MODELING OF THE GRAVEL PACKING PROCESS DuRING PETROLEuM HORIZONTAL WELL COMPLETIONS

Jairo Z. Souza*, João A. Aguirre Oliveira Jr.*, Carlos Eduardo Fontes*, João V. M. Magalhães†, André L. Martins†

*ESSS - Engineering Simulation and Scientific SoftwareRua Lauro Müller, 116 – Torre do Rio Sul – Sl. 1404 – Botafogo – Rio de Janeiro – RJ – Brasil – 22.290-160

E-mail: [email protected], [email protected], [email protected]†Petrobras/CENPES/PDP/TEP

Rua Horácio Macedo, 950 – Ilha do Fundão – Cidade Universitária – Rio de Janeiro – RJ – Brasil – 21941-915E-mail: [email protected], [email protected]

ABSTRACT

The increasing exploration challenges faced by the oil industry require the development of new technologies to achieve higher efficiency in the extraction of oil and gas, contemplating the lowest possible cost. Gravel packing is today the most frequently applied sand control technique by Petrobras in horizontal well completions. The gravel pack process is basically divided in three different stages: the injection, the alpha wave propagation and the beta wave propagation. Because of the critical conditions, such as operation in deep and ultra deep waters and limited flow rates (to avoid formation fracturing conditions), high precision methods of process design are required to assure gravel-packing success. There are several models available to predict the alpha wave height, but they are essentially empirical, resulting in imprecise predictions for extrapolated conditions (Martins et al, 2003). Mechanistic models are also available, such that developed by Martins et al (2003), but that model is not able to predict the bed height in all flows regimes. The modern CFD codes are able to simulate complex multiphase systems, but they are not so robust to simulate granular systems into the flow. Looking for other possibilities of multiphase simulation, Petrobras and ESSS has been working in development of a new methodology in order to reproduce granular flows numerically, specially for the gravel packing process. In this approach the Discrete Element Model (DEM) is used to perform particles simulation. The fluid field is solved by common CFD techniques and a coupling between CFD and DEM is established by a coupling module between FLUENT and EDEM softwares. The results show velocity and solid volume fraction profiles, which are in agreement with theoretical behavior of the flow. The Figure 1 shows the profiles obtained with the numerical simulations.

Figure 1 - Flow Profiles: (a) Solid Volume Fraction; (b) Fluid Velocity Profile.

Technical Session

8

www.cfdoil.com.br

The validation of this approach is done by the comparison of the alpha wave height obtained in the numerical simulations with experimental data from Petrobras. The comparison is showed in Table 1, where the cases present different operational conditions.

Petrobras DATA Simulation Results

Case Study 1 84 % 82,5 %

Case Study 2 85 % 84,74 %

Case Study 3 80 % 79,61 %

Case Study 4 79 % 81,82 %

Table 1 – Comparison - Alpha wave heights.

The results achieved in the present work turned out to be very promising, which means that it is possible to work numerically with the complex problem of high solids concentration. The main results of this study are the validation of the alpha wave height obtained in numerical simulation when compared with experimental data. The results proved the applicability of this approach, even in different operating conditions, and that the CFD-DEM coupling may, in the future, be used to aid in the design of Gravel Packing operation for horizontal wells.

REFERENCES

MARTINS, A. L. et al. A Mechanistic Model for Horizontal Gravel Pack Displacement. SPE European Formation Damage Conference. Netherlands. 2003.

Technical Session

9

www.cfdoil.com.br


MuLTIPHASE MODELLING AND FLuID DyNAMICS SIMuLATION OF A BIOMASS GASIFIER

Jairo Z. Souza*, Rodrigo P. M. Moreira*, Carlos Eduardo Fontes*, Carlos A. S. Paulo†,Marco A. H. Leite†, Álvaro Coppieters†, Marcelo R. Q. Medeiros†

* ESSSe-mail: [email protected]

† ESSSe-mail: [email protected]

+ PETROBRASe-mail: [email protected]

# PETROBRASe-mail: [email protected]

ABSTRACT

Due to the need of innovative ways to convert existing fuel reserves and renewable sources in useful energy, a renewed interest in thermochemical processes arose. Gasification of biomass in particular, brings the perspective of an efficient way to produce electric energy and biofuels. Nevertheless, biomass is a fuel that does not have homogeneity in physical and chemical properties, introducing challenges to the gasification process.

Cold mock-ups are used in experimental investigations to provide a better understanding of the multiphase flow inside of circulating fluidized beds. As part of the development of a cold mock-up, Petrobras and ESSS have made preliminary simulations using CFD. The goal was to evaluate operational conditions and to obtain an optimal configuration in order to refine the mock-up design.

Fine modeling of the riser, stand-pipe and loop-seal were performed providing detailed information about the gas-solid flow in those components. For the cyclone a simplified approach was adopted, whereas it caused a local pressure loss calculated by the correlation of Shepherd & Lapple.

The Figure 1 shows the geometry built to perform the numerical simulation. Riser, stand-pipe, and loop-seal are displayed in blue, red and light-blue, respectively.

Figure 1 - Gasifier Circuit Geometry in CFD model.

Technical Session

10

www.cfdoil.com.br

Figure 2 - Solid Volume Fraction profile.

The interaction between fluid-particle and particle-particle was reached through kinetic theory of granular flows. This model is widely used to reproduce granular flows in industrial equipments. The results showed the gas-solid flow profile, making possible identify recirculation zones, high velocities, high solid concentration regions, and the pressure profile. Loop-seal operating details have also been identified, allowing Petrobras engineers to evaluate the better physical and operational configuration for this specific component. Some results of the solid volume fraction profile are shown in Figure 2.

The wall shear analysis allowed to identify erosion zones in the gasifier circuit, pointing out regions that should be redesigned and/or receive special lining.

Technical Session

11

www.cfdoil.com.br


ENHANCED OIL RECOVERy IN CAPILLARy PASSAGES By INJECTION OF NON-NEWTONIAN FLuID WITH THE COMPLETE WETTING OF THE DISPLACED FLuID

Edson J. Soares† , Roney L. Thompson*

†Department of Mechanical Engineering, Universidade Federal do Espirito SantoAvenida Fernando Ferrari 514, Goiabeiras, Vitoria, ES. E-mail: [email protected]

*Department of Mechanical Engineering (PGMEC), Universidade Federal FluminenseRua Passo da Patria 156, Sao Domingos, Niteroi, RJ. E-mail: [email protected]

ABSTRACT

After primary and secondary oil recovery processes, in most of the cases, there still exists a significant amount of oil (in a mixture with water) inside the porous media of the reservoir. An interesting technique that can be employed at this stage is the injection of a polymer fluid in order to take advantages of its rheology to optimize the process. When the wetting conditions are such that displaced fluid wets completely the rock walls of the porous media, there remains a layer of this fluid attached to the wall. The idealized problem conceived to model the industrial process described can be simplified into a 2-D approach in two manners: the fluid-fluid displacement in a capillary tube or in a capillary plane channel. The former represents a useful simplification for the case where the cross area of the porous passage has an aspect ratio around the unity while the latter is more adequate to mimic the case of high values of aspect ratio. The objective of the present work is to study the displacement efficiency and flow patterns when a non-Newtonian fluid is injected.

MAIN FEATuRES

The flow is considered immiscible, incompressible, isothermal and with negligible inertia. The position of the interface is unknown a priori and, thus the domain of the problem is also part of the solution. We employed an elliptic mesh generation technique, coupled with the Galerkin Finite Element Method, solving the problem in a reference domain and mapping it on the physical one as depicted on the figure above (right), where the subdomains 1, 2, and 5 belong to the injected fluid while the sub-domains 3, 4, and 6 belong to the displaced one. We tested two non-Newtonian fluids: a power-law and a visco-plastic one.

RESuLTSWe can see in the figures below, the residual mass fraction as a function of the capillary number for two different viscosity ratio and different injected fluids, a power-law and a visco-plastic material. The respective rheological parameters are the power-law exponent and the yield number.

Technical Session

12

www.cfdoil.com.br

We can also see above, different flow patterns the problem exhibits depending on the rheological parameter of the injected fluid: bypass, transition, and full-recirculating regimes.

CONCLuSIONS

Low values of power-law exponent and high values the yield number lead to a more efficient displacement. These two parameters are also measurers of non-Newtonian behavior. In this sense, the more non-Newtonian, more percentage of oil recovery is obtained in the process.

REFERENCES

[1] E.J. Soares and R.L. Thompson, Flow regimes for the immiscible liquid–liquid displacement in capillary tubes with complete wetting of the displaced liquid, J. Fluid Mech. 641, pp. 63–84 (2009).[2] R.L. Thompson and E.J. Soares, Displacement efficiency and flow regimes when a power-law non-Newtonian fluid displaces a Newtonian liquid in capillary tubes, submitted to J. Fluid Mech. (2010).[3] J.F. Freitas, E.J. Soares and R.L. Thompson, Immiscible Newtonian displacement by a visco-plastic material in a capillary plane channel, submitted to Rheol. Acta (2010).

Technical Session

13

www.cfdoil.com.br

CFD EVALuATION OF A REFINERy FuRNACE BEHAVIOR uSING FLAMELET-BASED MODEL

Ricardo Serfaty*, Raphael D. A. Bacchi†

*PETROBRAS – Petróleo Brasileiro S.A. Centro de Pesquisas Leopoldo Américo Miguez de Mello - Av. Horácio Macedo, 950, Cidade Universitária. Rio de

Janeiro - RJe-mail: [email protected]

†ESSS – Engineering Simulation and Scientific SoftwareRua Lauro Muller 116, sala 1404, Botafogo. Rio de Janeiro - RJ


ABSTRACT

A three-dimensional computational fluid dynamics simulation using FLUENT of the radiant section of a double-fired furnace is performed in this work. Many aspects should be considered to increase the model accuracy with the real equipment. The furnace is modeled using real industrial size including burners and radiant section tubes. The influence of the furnace aspect ratio is also evaluated. The computational meshes vary from 1 to 3 million nodes to obtain the temperature profile near the tubes and species gradient in the flame zone. The fuel used in these simulations was pure methane.

One of the objectives of the present work is to evaluate the viability of flamelet-based model employment in reactive simulations. This model requires less computational efforts to achieve convergence than the Eddy-Dissipation and Finite Rate Chemistry models because the temperature and species composition are evaluated as a functions of two passive scalars fields, obtained from proper transport equations. Therefore, the non-linearity resulted from the difference between reactions timescales, which can be found, for example, in Arhenius-based model, is avoided. Another important feature of flamelet model is the possibility of evaluation of less stable or transitory species which can describe de flame topology instead of the temperature field.

All combustion models available for non-premixed flame in Fluent were used in order to make a comparison in the mean and maximum temperatures obtained among these methods. The radiation modeling with a spectral treatment is also an important issue in the simulations because the burning gases, in high temperatures, absorbs and reemits the radiation in different bands, depending on the concentration of the species involved and its temperatures.

One of the main objectives of these simulations is to evaluate the influence of variables such as aspect ratio for radiant section and different ratio for fuel gas staging in the overall heat transfer and local peak heat flux in the radiant section of a double fired furnace. Besides the reaction and radiation modeling challenges, the correct representation of the different length scales, which can vary over 1.000 times, is another important task for the correct flame topology representation.

The results show the viability of application of flamelet model to the problem and the difference in heat distribution between the tubes with the aspect ratio variation.


Technical Session

14

www.cfdoil.com.br

Picture 1 - Temperature plot in a cross section of the radiation chamber. The picture shows the same field for Finite Rate Chemistry (right) and Flamelet (left) models.

REFERENCES

[1] ANSYS CFX 12.1 Solver Documentation. ANSYS Inc., Cannonsburg, EUA (2009).[2] LI, L.; LIU, T.; PENG, X.F. Flow characteristics in an annular burner with fully film cooling. App. Thermal Eng. V.25, pp. 3013-3024 (2005).[3] MAGNUSSEN, B.; HJERTAGER, B. 16th Comb. (Int.) Symposium, pp. 719-729 (1976).[4] Hottel, H.C. and Sarofim, A.F.,“Radiative transfer”, McGraw-Hill, New York (1967).[5] Taylor, P.B. and Foster, P.J., “The total emissivities of luminous and non-luminous flames”, Int. J. Heat Mass Transfer, 17, pp. 1591-1605 (1974).[6] Hadvig, S.,“Gas emissivity and absorptivity”, J. Inst. Fuel, 43, pp. 129-135 (1970).

Technical Session

15

www.cfdoil.com.br

ANALISyS OF SLEEVE REPAIR WELDING OF IN-SERVICE PIPELINES

Esteban Gonzalez Garcia*, Mauricio Pacheco†, João Aguirre†, Giovani Dalpiaz+

*ESSS - Engineering Simulation and Scientific SoftwareBaltimore 645 - X5105AHG – Villa Allende

Córdoba - Argentinae-mail: [email protected]

†ESSS - Engineering Simulation and Scientific SoftwareRua Paulo Emílio Barbosa, 485 - Parque Tecnológico do Rio - Ilha do Fundão - CEP 21941-615

Rio de Janeiro - RJ - Brasile-mail: [email protected]

+CENPES – PETROBRAS Research and Development Center Cidade Universitaria Q.7 - Ilha do Fundao - Rio de Janeiro/RJ, Brazil. CEP: 21949-900


ABSTRACT

Welding is a manufacturing process widely used in the metalworking industry, due to their wide application. Among the various applications, we highlight the use of welding processes in operations relating to the petroleum industry, constantly rising in Brazil, not only for building structures, but also for the union and repair of pipelines.

The ability to weld onto in-service lines enables the installation of repair sleeves without interruption of flow. While there are economic benefits to this practice, certain issues must be addressed to ensure that public, environmental, and worker safety is maintained both during and after welding. Specifically, high cooling rates occur when welding onto an in-service pipeline because the flowing contents quickly remove heat from the pipe wall. These cooling rates promote the formation of hard heat-affected zone microstructures, making these welds susceptible to hydrogen cracking during or soon after welding, and to sulfide stress cracking in subsequent sour service.

The welding processes in metallic alloys have as a basic characteristic the presence of an intense localized heat source. This heat source may cause drastic changes in the structure`s material. This project involves the development of a CFD model describing the displacement of a thermal source of the welding process in a sleeve repair union of an in-service pipeline, considering the flow of different fluids inside the pipe. Through the use of a CFD solver it is possible to obtain more accurate temperature profiles in the welded parts and the determination of the cooling rate during the process.

The model was developed with ANSYS simulation tools, starting with the construction of a parametric geometry (pipeline, sleeve, weld bead and fluid) and mesh (Fig. 1). A small gap filled with air was also considered between the pipe and the sleeve. Material properties for the low-carbon steel and working fluid where inserted in CFX as expressions valid for the full range of expected temperatures.

The geometry of the heat source and the distribution of power within it are two very important factors for a correct representation of the process. In this case, a double-ellipsoid Goldak (Ref.1) time-dependent heat source was programmed in CFX (Fig. 2) and the material deposition effect is achieved by changing the weld bead material’s thermal conductivity along with the heat source movement.

The simulation of the fluid flow in the duct was performed independently of the simulation of heat transfer between bodies and fluid. This was done to reduce the computational cost of the cases to be executed, as “uncoupling” the solution of the flow from the thermal simulation, the field of fluid velocities can be resolved using a single steady state simulation.

Results compare well with the expected temperature profiles and exhaustive benchmarking is being made to calibrate the model with experimental data. The main conclusion is that this simulation approach gives, besides accurate results, much more insight of the process itself and unlimited flexibility when it comes to changing geometry, material properties or welding parameters.


Technical Session

16

www.cfdoil.com.br

Figure 1 - Geometry and mesh (left), Goldak heat source shape (right).

Figure 2 - Thermal conductivity and welding temperature (left), pipe inner wall temperature (right).

REFERENCES

[1] GONZALEZ, E.; AGUIRRE, J.; PACHECO, M; “Análise do Processo de Soldagem de Reparo em Dutos em Operaçao”, Relatório Técnico 2009-BRC-TMEC-CALHA-RT03, 2010.[2] GOLDAK, J., “Computational Welding Mechanics”, Springer, 2005.

Technical Session

17

www.cfdoil.com.br

COMPuTATIONAL SIMuLATION OF NATuRAL CONVECTION OF A FLuID WITH INTERNAL HEAT GENERATION

Camila B.Vieira and Jian Su

Universidade Federal do Rio de JaneiroNuclear Engineering Program

e-mail: [email protected] [email protected]

ABSTRACT

Natural convection in fluids with internal heat generation in two-dimensional square and semi-circular cavities was simulated numerically by using a tool of computational fluid dynamics (CFD), ANSYS CFX 12.0. The objective of the work was to study the influence of the geometry, boundary conditions, fluid properties and mainly the volumetric heat generation rate, as a function of Prandtl and Rayleigh numbers. Steady-laminar, transient-laminar and turbulent flow regime were identified. Spectral analysis of instantaneous velocities and temperature by using FFT was carried out to identify periodic or chaotic transient-laminar regimes. Critical Rayleigh numbers at which the onset of transient–laminar regime occurs were determined for Pr = 0.0321, 0.71 and 7.0, in two-dimensional square cavities with isothermal vertical walls and adiabatic horizontal walls. Turbulent natural convection at high Rayleigh numbers was simulated by using the Reynolds-averaged Navier-Stokes equations (RANS) with SST and SST-SAS turbulence models, and the large eddy simulation (LES-WALE). Some numerical results were in good agreement with available empirical correlations and others results were compared with computational simulation in the literature.

REFERENCES

[1] Emara, A. e Kulacki, F. (1980). A numerical investigation of thermal convection in a heat-generating fluid layer. Journal of Heat Transfer, 102:531–537.[2] Horvat, A., Kljenak, I., e Marn, J. (2001). Two-dimensional large-eddy simulation of turbulent natural convection due heat generation. Heat and Mass Transfer, 44:3985–3995.[3] Kulacki, F. A. e Emara, A. A. (1976). High rayleigh number convection in enclosed fluid layers with internal heat sources. Technical Report NUREG-75/065, U.S. Nuclear Regulatory Commission Report.[4] Mayinger, F., Jahn, M., Reineke, H., e Steinbrenner, V. (1976). Examination of thermohydraulic processes and heat transfer in a core melt. Technical report, Institut fur Verfanhrenstechnic der T.U.


Technical Session

18

www.cfdoil.com.br

NuMERICAL SIMuLATION OF A DuMMy WELL PuMPING MODuLE

Luiz Carlos Tosta da Silva*, Ricardo A. Medronho†

*PETROBRAS - Petróleo Brasileiro S.A. - CENPESAv. Horácio Macedo, 950 - Ilha do Fundão, Prédio 20, Sala 2055 CEP: 21941-915 - Rio de Janeiro - RJ / Brasil


† Dep. de Eng. Química, Universidade Federal do Rio de JaneiroEscola de Química/UFRJ - Av. Horácio Macedo, 2030 Edifício do Centro de Tecnologia, Bloco E, sala E-207

Cidade Universitária - Ilha do Fundão CEP: 21941-909 Rio de Janeiro - RJe-mail: [email protected]

ABSTRACT

Oil and gas production in deep and ultra deep waters require the development of new production techniques, as novel artificial elevation techniques. At the moment, PETROBRAS is developing a novel subsea pump module, equipped with an ESP – Electrical Submersible Pump, known as MOBO and applied on the flow of a multiphase mixture (oil, gas and/or water). In this work, the flow behavior inside of a MOBO system was simulated through computational fluid dynamics (CFD). Two geometries of MOBO were built – with and without side holes in a shroud pipe that encases the ESP intake - to simulate an internal two-phase (water and air) flow. For numerical results validation, a comparison with experimental results that had being previously acquired in the R&D Center (CENPES) of PETROBRAS, was performed and showed good agreement. It has been evidenced that the shroud pipe strategically equipped with side holes was able to attenuate in an expressive way the phase discontinuities in the ESP intake region, thus minimizing events of gas discharges and associated lack of operational continuity in the pump. Numerical simulations with oil and gas, exhibiting similar characteristics of those found in the Campos Basin, also showed absence of gas discharges within the tested conditions.

INTRODuCTION

A major challenge in the production of offshore oil reserves is the profitability assurance over the field concession exploitation period. In Brazil, approximately 86.3% of oil production and more than 69% of gas production come from offshore accumulations, while most of these is located in deep waters (300 to 1500 m deep) and ultra deep waters (deeper than 1500m). The discovery of new offshore fields of oil and gas, as the giant fields of Tupi and Jupiter, among other pre-salt accumulations, further raise the reserve mineral resource offshore areas in ultra deep water depths and ,consequently, also increase the need to use floating systems for exploration, production and maintenance of wells and subsea equipment.

MOTIVATION

The production of oil and gas in deep and ultra deepwater scenarios necessarily involve the development of various production techniques, including the artificial lift and flow techniques of the produced fluids to the operational processing facility. These artificial lift and flow techniques have a strong influence on the production planning and development of new fields. These techniques must ensure compensating flow rates, under an economical point of view, for the typically very high investment required in the oil and gas offshore production. Among these artificial lift techniques, the one based on an Electrical Submersible Pump (ESP), being deployed in the production and/or dummy well, is a very powerful one.


Technical Session

19

www.cfdoil.com.br

PETROBRAS has worked with some subsea equipment companies for designing and building a prototype called Pumping Module (MOBO) equipped with an ESP, to be installed in a dummy well close to an actual production well in the Brazilian coast in water depths of 1400 m. The prototype is designed to work with an ESP of about 30 m in length. The dummy well is located typically about 200 m from the production well.

METHODOLOGy

Making use of the computational tool CFX- Version 12, studies were initially performed by Tosta da Silva (2010) [1] in order to identify the best simulation mesh – residing such criteria on the incurred computational time and level of agreement between numerical computational results and experimental previously acquired data. The computational mesh was built using the same geometry as the one found in the experimental apparatus. This mesh allowed modeling the flow of two fluids (water and air) subject to operational conditions in order to simulate and analyze the dynamics of flow inside the MOBO. Such analysis also allowed a direct comparison between the numerical computational results and the experimental ones collected by Gaspari et al (2007) [2].

RESuLTS

This study found that the computational results showed good agreement with experimental results. It can also be concluded that holes in the shroud attenuate in an expressive way phase discontinuities in the region of pump suction, minimizing events of gas discharge in the well dummy / shroud annulus region, see Figure 1 and 2. Simulations with oil and gas, exhibiting similar characteristics of those produced in the Campos Basin, showed absence of gas discharge in the well dummy / shroud annulus, however these studies still need further clarification to investigate the real contribution of the fluid viscosity and some geometrical aspects of the MOBO.

Figure 1 - View with discharge of fluid in the dummy well / shroud annulus region, PG 7.0%, Shroud

without side holes and water flow rate of 9.02 l/min.

RefeRences

[1] Tosta da silva, l. C., Simulação Numérica de Poço Alojador de Bombeio, Dissertação M. Sc. Universidade Federal do Rio de Janeiro – UFRJ, Escola de Química – E.Q., Rio de Janeiro, 2010.

[2] Gaspari, E. F., Alves de Oliveira, G. P. H., Tosta da Silva, L. C. Estudo Experimental sobre a Dinâmica de Escoamento no Módulo de Bombeio de Jubarte, Comunicação Técnica TE 014/TE, Internal Document of the Research Center of PETROBRAS – CENPES, 2007.

Technical Session

Figure 2 - Visualization of the phase segregation in the MOBO, PG 8.0%, Shroud with side holes and

water flow rate of 9.02 l/min.

20

www.cfdoil.com.br

A CFD STuDy OF AN INDuSTRIAL GAS TuRBINE COMBuSTION CHAMBER uSING ZIMONT’S MODEL

Luís Fernando Figueira da Silva†, Thiago Koichi Anzai*, Carlos Eduardo Fontes*, Karolline Ropelato*

†Department of Mechanical Engineering, Pontifícia Universidade Católica do Rio de Janeiro, Rio de Janeiro, BRAZIL

e-mail: [email protected]*Engineering Simulation and Scientific Software (ESSS), Ltda. Rio de Janeiro, BRAZIL


ABSTRACT

The accurate prediction of pollutants emission from a gas turbine combustion chamber turns out to be one of the major concerns when such an equipment is subject to long periods of operation away from its design point. In such conditions, the flow field itself might also show big differences from the design point, leading, for instance, to the presence of undesirable hot spots or instabilities in the combustion process. Since a study of all possible operation conditions is economically unfeasible and as direct numerical simulation (DNS) of industrial combustors is also beyond reach of the foreseeable computational resources, models must be used for the analysis of such issue.

This works presents the results obtained using Computational Fluid Dynamics (CFD) of an industrial gas turbine combustion chamber reactive flow field. The model considers the geometrical details via a CAD model of such a combustion chamber and uses actual operating conditions, calibrated using an overall gas turbine thermodynamical simulation to provide the boundary conditions (Orbegoso et al., 2009). This model retains the basic information on combustion staging, which occurs in diffusion and lean premixed modes.

The turbulence has been modeled using the k-ω-SST model with a curvature correction. Combustion and turbulence interactions is accounted for by using the Zimont (2001) model. A high resolution scheme was used to model the advection terms of the momentum equation.

RESuLTS AND DISCuSSIONS

Figure 1 (a) presents the viscosity ratio plotted over stoichiometric surfaces issuing from the fuel injection, allowing to verify that slower fuel/air mixing occurs in stages A&B than at the pilot region. Figure 1 (b) shows methane mass fraction distribution, dark gray color indicates regions of high methane mass fraction and light gray color, low regions of methane concentration. The longitudinal velocity component in a plane which passes through the center of the injection masts of the different stages is represented on Figure 1 (c). Dark gray color indicates regions of high velocity and light gray color, low regions of velocity.

The swirling vanes of stages A&B impart a mild rotation movement to the flow, whereas the orientation of the pilot stage swirling vanes is such that a strong flow rotation results.

The combustion chamber thermal protection is partly guaranteed by four series of cooling air orifices, which are arranged in recesses of the combustion chamber wall and that lead to the formation of an adjacent fresh air film.


Technical Session

21

www.cfdoil.com.br

Figure 1 - (a) Distribution in a cut-off plane of methane mass fraction and (b) longitudinal velocity component.

A CFD study on an industrial gas turbine combustion chamber in order to determine NOx emissions was developed by the authors. The emission predictions were improved with a combustion model capable of describing combustion in partially premixed flow.

REFERENCES

[1] E.M.M., Orbegoso; C.D., Romeiro; S.B., Ferreira; L.F., Figueira da Silva. Emissions and thermodynamic performance simulation of an industrial gas turbine, 45th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit 2009.[2] D. C., Wilcox. Turbulence Modelling for CFD. DCW Industries, pp. 314, (2000).[3] V.L., Zimont; F. Biagioli. and K. Syed. Modelling turbulent premixed combustion in the intermediate steady propagation regime, Progress in Computational Fluid Dynamics, Vol. 1, pp. 14-28 (2001).

Technical Session

22

www.cfdoil.com.br

Tuesday, July 13th - 14:40 - Room 1

SIMuLATING COKE DEPOSITION IN THE SLOP WAX COLECTOR uSING COMPuTATIONAL FLuID DyNAMICS

Paulo L. C. Lage*, Alessandra B. Santos*†, Rodrigo B. Medeiros+, Júlio C. C. Venâncio+, Luiz F. L. R. Silva+

*Programa de Engenharia Química, COPPE, UFRJ C. P. 68502, Rio de Janeiro, RJ, 21941-972, Brazile-mail: [email protected]

+Escola de Química, Universidade Federal do Rio de Janeiro,21941-909 Rio de Janeiro, RJ, Brazil

e-mail: [email protected]†CENPES, PETROBRAS

Cidade Universitária, Q.7 Ilha do Fundão 21949-900, Rio de Janeiro, Brazile-mail: [email protected]

ABSTRACT

The aim of this study was to analyze coke formation and deposition in a slop wax collector of vacuum distillation column by computational fluid dynamics (CFD), using the commercial software ANSYS CFX 11.0 SP1 & 12.1.

In a previous work [1,2], a simplified kinetic model was used to estimate the coke formation from asphaltenes. The three-dimensional CFD simulations with heat transfer were carried out in the actual scale of the collector, but considering only the flow of the liquid phase. In the model, asphaltenes were set up as an additional variable and the coke formed as another additional variable that is simply accumulated in your point of formation. The liquid was considered incompressible, turbulence was considered by the SST model and the lower wall of the collector was assumed the temperature of the rising steam. The steady state was not obtained, but transient simulations showed that the flow has quasi-stationary characteristics. The results showed that coke is formed preferentially in hot spots of the collector.

The model used in this previous work can determine the mean volumetric rate of coke formation at each point in the liquid phase but that model did not consider the movement of the coke particles after their formation. Therefore, in the present work, a one-way Lagrangian-Eulerian pseudo-steady simulation of coke particles movement in the slop wax collector of vacuum distillation column was carried out using the results of the mean volumetric rate of coke formation and the quasi-steady liquid velocity fields from the previous simulations. A routine in user FORTRAN was implemented into ANSYS CFX 12.1 to inject a particle at each mesh node if the mass flow rate corresponding to the coke mass formation rate at the corresponding volume. A sensitivity analysis was made regarding the coke particle size and particle-wall restitution factors. The model was able to predict the wall regions where the particles tend to accumulate and, probably, where they will agglomerate and deposit, a phenomenon that cannot be predicted by the present model. A sample of the results is presented in Figures 1 and 2 for 50 micra coke particles and a restitution coefficient of 0.1 in a 270 thousand node mesh. Figure 1 shows the contour plot of the coke mass formation rate, normalized by its maximum value, at the lower heated surface. This show the regions that form more or less coke over this surface. Figure 2 shows the averaged volume fraction occupied by the coke particles at the same surface considering all the coke particles formed in the liquid phase. It is quite clear that the locations of coke formation and deposition are quite different.

Technical Session

23

www.cfdoil.com.br

Figure 1 - Normalizad coke mass formation rate at the bottom surface.

REFERENCES

[1] A. B. Santos, Estudo do Comportamento Térmico do Coletor de Líquido de Gasóleo Residual através da Fluidodinâmica Computacional, M.Sc. Dissertation, Programa de Engenharia Química, COPPE/UFRJ (2009).[2] A. B. Santos, P. L. C. Lage, Minimizing coke formation in a distillation column chimney tray by CFD análisis. 2009 ESSS South American ANSYS Users Conference, Florianópolis, November (2009).

Figure 2 - Averaged volume fraction of coke particles at the bottom surface.

Technical Session

24

www.cfdoil.com.br


2D SIMuLATION OF LEAKAGE AND DAMAGED STABILITy OF OIL CARRIER By MPS METHOD

Liang-yee Cheng1*, Diogo Vieira Gomes2, Kazuo Nishimoto3

1 Escola Politécnica, University of São PauloAv. Prof. Almeida Prado Trav. 2 No. 83 Ed. Eng. Civil Cidade Universitária, 05508-900 São Paulo, SP, Brazil

e-mail: [email protected] Escola Politécnica, University of São Paulo

Av. Prof. Mello de Moraes 2231, Cidade Universitária, 05508-900 São Paulo, SP, Brazile-mail: [email protected]

3 Escola Politécnica, University of São PauloAv. Prof. Mello de Moraes 2231, Cidade Universitária, 05508-900 São Paulo, SP, Brazil aulo


ABSTRACT

In case of damage of a crude oil carrier, both the stability of the vessel and oil leakage is of great concerns due to the safety and the environmental issues. Nowadays, there are several method and computational tools for the calculation of the damaged stability. However, due to the fluid-solid interaction with complex geometry and multiphase flow, the detailed investigation of the phenomena of the oil leakage caused by hull damage, including the coupled transient motions of the fluids and the vessels, still remains as a challenge.

The aim of the present research is to carry out a coupled transient analysis of the oil leakage process and the damaged stability. For this purpose, numerical method based on MPS (Moving particle Semi-Implicit) method [1] is adopted to model both the motion of the vessel and the oil-water multiphase flow with free surface. The effectiveness of the method for the analysis of the damaged stability was shown in a previous study [2] by considering water leakage and entrance in a floating tank. In the present paper, a multiphase flow modeling based on the interparticle potential force model proposed by Kondo et al [3] is applied to study the complex oil-water flow in case of oil leakage.

For sake of simplicity, a two dimensional (2D) rectangular shaped scaled model is considered. The beam, depth, mass, inertia, TCG and VCG are, respectively, 0.425 m, 0.325 m, 20.3 kg, 0.657 kgm2/m, 0.0 m and 0.1097 m. The scaled model has two internal tanks. The thickness of the walls is 0.02 m except in center, where the thickness is 0.025 m, and in the bottom, where the thickness of 0.055 m is used to model the double bottom. The opening for the oil leakage is 0.05 m. The total width of the towing tank used in the numerical simulations is 2.7 m with beaches of 0.8 m in both extremity of the towing tank.

Fig. 1 gives the snapshots of the animation obtained from the MPS simulation with 75% filling and damage at 0.10 m above the keel. The numerical simulations of the transient motion show that the sway motion induced by the leakage may occur in the beginning of the process when a relatively large volume of oil is released.

The validation of final equilibrium angle of list was carried out by using SSTAB. As SSTAB is a hydrostatic stability calculation program and unable to take into account the dynamic effects, the final list angle provided by SSTAB is determined by using the volume obtained by MPS simulation. Fig. 2 provides the final list angles in case of 75% filling the damage height of 0.14 m, and the good agreement between the results shows that the numerical approach is very effective.

Fig. 3 gives the volume of the oil leakage calculated by MPS, together with the leakage estimate by using SSTAB through quasi-static approach. The cause of the discrepancy seems to be the dynamic effects: for lower openings, the leakage is larger as well as the dynamic motion caused by the leakage. On the other hand, as the leakage reduces when opening becomes higher, the motion becomes near to the quasi-static assumption of the SSTAB.

Technical Session

25

www.cfdoil.com.br

Figure 1 - Snapshots of the simulation of the cases with 75% filling and damage height of 0.10 m.

Figure 2 - Time history of the list angle obtained by transient analysis of MPS and hydrostatic analysis of SSTAB for internal tank with 75% filling and damage

height of 0.14 m.

Figure 3 - Oil leakage obtained by MPS and by quasi-static calculation using SSTAB for 45% and 75% filling and damage height of 0.10 m,

0.14 m and 0.20 m.

Finally, as 2D modeling is a hypothetical situation in which the dimension of the opening is much larger than the actual cases. In this way, instead of extrapolating straightforwardly the 2D results to the actual situations, further complete 3D analysis should be done.

REFERENCES

[1] S. Koshizuka and Y. Oka, Moving-Particle Semi-Implicit Method for fragmentation of incompressible fluid. Nuclear Science and Engineering, vol. 123, pp. 421-434, 1996.[2] G. E. R. Silva, M. M. Tsukamoto, H. F. Medeiros, L. Y. Cheng, K. Nishimoto, A. Saad. Validation Study of MPS (Moving Particle Semi-implicit Method) for Sloshing & Damage Stability Analysis. In: The Proceedings of the 27th International Conference on Offshore Mechanics and Artic (OMAE’2008), Jun. 2008, Estoril, Portugal. Paper No. OMAE2008-57460.[3] M. Kondo, S. Koshizuka, M. Takimoto. Surface tension model using inter-particle potential force in Moving Particle Semi-implicit method. Transactions of JSCES, paper No. 20070021, (2007).

Technical Session

26

www.cfdoil.com.br


SIMuLATION OF BuBBLING PROCESSES INCLuDING BuBBLE FORMATION EFFECTS: ISOTHERMAL STRIPPING

Ricardo C. Rodrigues*, Paulo L. C. Lage*

*Programa de Engenharia Química, COPPE, UFRJ C. P. 68502, Rio de Janeiro, RJ, 21941-972, Brazil


ABSTRACT

Computational fluid dynamics simulations of bubbling processes using the Eulerian two-fluid model assume that the bubbles are “instantaneously” formed with a given size at the bottom of the column, ignoring completely the bubble formation process. This approximation eliminates some important characteristics related to the flow fluctuations promoted by non-simultaneous bubble detachments at neighbor orifices that affect fluid flow immediately above the gas distributor. Besides, in chemical engineering processes, mass and/or energy are transported between the bubbles and the liquid phase. Due to the liquid mixing in a bubble column, the mass and energy transport resistance inside the bubbles are usually the dominant ones. Previous works [1,2] have proven that these transport processes are very intense inside the bubble during its formation due to the large surface-to-volume ratio and the large convection promoted by the gas injection into the attached bubble through its formation orifice. The transport processes are so intense that more than 50% of the overall mass or heat available for exchange can be transferred during the bubble formation. Therefore, the current employed approximation of “instantaneous bubble formation” used in computational fluid dynamics simulations of bubbling processes using the Eulerian two-fluid model is wrong and may lead to large errors if the bubbling height is sufficiently small for the emerging bubbles not to be in equilibrium with the liquid phase.The present work develops a simplified method of incorporating the bubble formation mass transfer effect in the bubbling process of isothermal stripping of ethyl acetate from its water solution [3]. The method of coupling uses a previously developed mass and heat transfer model for a single bubble during its formation at a submerged orifice and during its ascension in a liquid column [1]. This model gives the bubble size, position and conditions at detachment and the mean mass transfer coefficient inside the bubble during its ascension [2]. This information is incorporated in the CFD simulation by changing the gas inlet conditions and by considering a bubble formation region at the bottom of the column in which a sink term is applied to the liquid phase that corresponds to the mass extracted from the bubbles during their formation.As examples of the results, Figure 1 shows the instantaneous ethyl acetate saturation and velocity fields of the gas phase after 20 seconds of simulation for the gas superficial velocity of 2 cm/s including or not the bubble formation effect. The difference in ethyl acetate saturation is quite clear near the column bottom. Figure 2 quantifies the error related to the approximation of “instantaneous bubble formation” by showing the area-average ethyl acetate saturation along these simulations at a horizontal plane distant 8 mm from the column bottom. It is quite clear that, for low bubbling heights, like in the stages of distillation towers, the error introduced by this approximation is too large.

REFERENCES

[1] C. P. Ribeiro Jr., C. P. Borges, P. L. C. Lage, Modelling of Direct Contact Evaporation Using a Simultaneous Heat and Multicomponent Mass Transfer Model for Superheated Bubbles, Chemical Engineering Science 60(6), pp. 1761–1772 (2005).[2] R. C. Rodrigues, C. P. Ribeiro Jr., P. L. C. Lage, On the estimation of the gas-side mass-transfer coefficient during the formation and ascension of bubbles, Chemical Engineering Journal 137(2), 282–293 (2008).[3] R. C. Rodrigues, Estudo do aumento de escala do processo de esgotamento de aromas em colunas de borbulhamento usando fluidodinâmica computacional, M.Sc. Dissertation, Programa de Engenharia Química, COPPE/UFRJ (2005).

Technical Session

27

www.cfdoil.com.br

Figure 1 - Instantaneous ethyl acetate saturation and velocity fields of the gas phase at t = 20 s for simulations (a) with and (b) without the bubble formation model for the gas superficial velocity of 2 cm/s.

Figure 2 - Ethyl acetate saturation in gas phase at a column section 8 mm above the gas distributor for the gas superficial velocity of 2 cm/s: effect of including the bubble formation model in the simulation.

Technical Session

28

www.cfdoil.com.br


CFD MODELING OF THE COMBuSTION GASES IN NON-PREMIXED BuRNERS OF A GROuND FLARE AND THE DISPERSION OF POLLuTANTS GENERATED DuRING THE PROCESS

Marcelo Kruger*, Daniel Ribeiro*, Carlos Eduardo Fontes*, Karolline Ropelato *, Paulo Roberto Pagot†

*ESSS - Engineering Simulation and Scientific SoftwareRua Lauro Müller, 116 – Torre do Rio Sul – Sl. 1404 – Botafogo – Rio de Janeiro – RJ – Brasil – 22.290-160

E-mail: [email protected], [email protected], [email protected], [email protected] †Petrobras/CENPES/PDP/MC

Rua Horácio Macedo, 950 – Ilha do Fundão – Cidade Universitária – Rio de Janeiro – RJ – Brasil – 21941-915E-mail: [email protected]

ABSTRACT

An Enclosed Ground Flare is provided for incinerating waste combustible gas in which the gases are conducted by horizontal supply pipes and delivered for burning by a great number of burners mounted at the top of vertical supply pipes, as showed in the Figure 1. Each burner is composed by a great number of gas nozzles. The gases are delivered to burn in the environment, where the mixture with the air is done and the combustion products are discharged at the flame end.The main advantages of these large installations are no flame visual impact on the neighborhood, smokeless flaring, reduced utility fluid usage, no thermal radiation toward the surrounding and wind sheltered flame combustion. Thus, the use of Ground Flares is increasing nowadays, especially in the proximity of densely populated and/or environmental sensitive areas.The height, number and arrangement of burners, the height and distance of fence, and many others parameters should be designed both to meet radiation and disperse pollutants down to the allowable limits.In this context, the present work is focused on the modeling and numerically simulating the non-premixed hydrocarbon-gaseous jet-flame burners of the Cabiúnas Ground Flare, Figure 1, and the consequent dispersion of the pollutants generated during the process.

Figure 1 - Cabiúnas Ground Flare.

Technical Session

29

www.cfdoil.com.br

The work was developed in two phases. In the first phase, it has been modeled a single first-stage burner flame. Some influences like the cross-wind on and the contact between the close flames were disregarded. In the second phase, the results achieved at the first phase were imposed to each one of thirty one-stage burners in the Ground Flare. Environmental characteristics, direction and wind velocity, were informed by Petrobras Technical Team.The Figure 2 shows the single-burner flame temperature central cross-section and the CO tri dimensional dispersion in the Ground Flare. This is a work-in-progress and it will proceed during these next coming twelve months. The main goal will be to calibrate and validate the CFD simulations with real field measurements.

REFERENCES

[1] Ferreira, C. J. Modelling of Combustion in Turbulent Reacting Flow, ANSYS Germany.[2] ANSYS CFX (2009) Theory Documentation version 12, abril de 2009. Cannonsburg, USA.[3] Famiani, F (2008) Ground flare and ground flare fence design. Control logic narrative for TA-5003 and TA-5004. Document I-MD-4150.50-5412-583-HSY-001.

Figure 2 - (a) Central cross-section map of Flame Temperature; (b) Iso-surface of CO Volume Fraction.

Technical Session

30

www.cfdoil.com.br


MODELING AND SIMuLATION OF A THREE-PHASE AND THREE-DIMENSIONAL FLOW IN A FCC INDuSTRIAL RISER

Gabriela Cantarelli Lopesa,*, Milton Moria, Waldir Pedro Martignonib

aUniversity of Campinas, School of Chemical EngineeringP.O. Box 6066, 13083-970, Campinas, SP, Brazil.


bPETROBRAS/AB-RE/TR/OT65 Republica do Chile Av, 20031-912, Rio de Janeiro, RJ, Brazil.


ABSTRACT

A three-dimensional and three-phase flow model to predict the dynamic behavior of a fluid catalytic cracking (FCC) industrial reactor was developed in this work. The study took into account heat transfer, feedstock vaporization and chemical reactions. The Eulerian-Lagrangian approach was used to simulate the dynamic flow inside the riser. A commercial CFD code (FLUENT) was used to obtain the numerical data, and appropriate user-defined functions were implemented inside the software to model the heterogeneous kinetics and the catalyst deactivation. Results show nonuniform tendencies inside the reactor, emphasizing the importance of using a more complete model in FCC process predictions.

INTRODuCTION

Momentum transfer, heat transfer, catalytic cracking reaction and droplet vaporization are some of many phenomena that occur simultaneously in commercial FCC riser reactors. These phenomena are interrelated and a disturbance in one, influences the others. Most studies in FCC process do not consider radial dispersions observed in experimental studies and which can affect the reactions yield. This work extends these contributions by considering a three-dimensional and three-phase model which takes into account these radial dispersions. It is an important consideration in the FCC reactor studies, since a shorter and more uniform catalyst distribution in the riser reactor could potentially produce a better reaction performance. Therefore it is fundamental to know the hydrodynamic behavior of this process.

SIMuLATION

The transient three-fluid model used in this work considers a three-dimensional flow, including heat transfer, phase change and chemical reactions. Therefore the velocity is increased by both the vaporization of the gas oil droplets and the molar expansion caused by cracking reactions. The three-phase flow model used an Eulerian description of the gas and solid phases, while the Lagrangian approach was used to describe the liquid droplets.

The commercial code ANSYS FLUENT 12.0 was used to obtain the numerical data for the model of this work. The Discrete Phase Model (DPM), present in the software, was applied to predict the gas oil vaporization from the discrete phase droplet to the continuous phase.

The properties of the solid granular flow were predicted using the kinetic theory and, the momentum transfer between the gas and solid phases was modeled using the Gidaspow drag model. The effects of turbulence in the gas phase were represented by the Reynolds Stress Model (RSM). The Ranz-Marshall model was used to predict the heat transfer between the phases. A four-lump model was applied to take in account the catalytic cracking reactions. The catalyst deactivation by the coke deposition in its surface was also considered in the model.

Technical Session

31

www.cfdoil.com.br

Figure 1 illustrates the geometry and mesh of the industrial riser considered. The mesh used in this work consists of approximately 1,200 thousand hexaedrical volumes. The gas oil droplets are injected by lateral smaller tubes, while the bigger side entrance is used to feed the hot catalyst. Steam is used as a fluidizing medium, being injected into the base of the reactor.The results presented bellow were obtained by simulating an unsteady state case reaching a time of 3 seconds, and do not represent the final results. More time of simulation is necessary for the variables have a cyclic behavior and the time-averaged data can be obtained.The model was verified by comparing the simulated gas oil conversion and gasoline and coke yields with the data found in the work Ali et al. (1997). Their data refers to an industrial riser with a height of 33m and a diameter of 0.8m, operating under similar conditions as those simulated in this work. The agreement between simulated and commercial data is reasonably good, as shown in Figure 2(a). Figure 2(b) presents the curl of the velocity field close the region where the gas oil droplets are fed. The gas oil vaporization occurs in this area causing intense turbulence. The nonuniformities of the flow extend along the length of the reactor, being observed at 20m from the entrance, as shown in Figure 2(c) where the radial dispersion of gasoline can be observed.

CONCLuSION

The results which are being obtained with this model highlight the importance of three-dimensional simulations to correctly predict geometric effects, especially at the feed injection area where the gas oil meets the hot catalyst and is vaporized, causing strong momentum and temperature gradients. The three-dimensional model provides more correct variables distribution, giving a better prediction of reactor performance. In conclusion, the model presented in this work is useful for the optimization and operation of FCC risers.

REFERENCES

[1] H. Ali, S. Rohani and J. P. Corriou, Modelling and control of a riser type fluid catalytic cracking (FCC) unit. Trans IChemE 75, Part A, pp. 401–412 (1997).

Figure 1 - Geometry and mesh of the simulated industrial riser.

Figure 2 - (a) Simulated results compared with plant data, (b) velocity vectors and (c) gasoline mass fraction.

Technical Session

32

www.cfdoil.com.br


SIMuLATION OF ENVIRONMENTAL LOADS ON A SuPPLy BOAT

Thiago S. F. C. Silva* ,Daniel F. C. e Silva†

PETROBRASAv. Horácio Macedo 950, Prédio 32

Cidade Universitária – Ilha do FundãoRio de Janeiro – RJ

CEP: 21941-915*e-mail: [email protected]

†e-mail: [email protected]

ABSTRACT

Offshore vessels must be designed according to the environmental loads they will be subjected during operation, such as current, wind and wave loads. Computational Fluid Dynamics has recently become one of the main engineering tools to assess those loads. Relevant studies with that purpose, such as current load analyzes for semisubmersibles [1, 2] and wind load analyzes for a FPSO[3] were already made by other research groups. The Scientific Methods group at Petrobras Research Center has also made current and wind loads simulations for different FPSO of the Petrobras fleet[4].

During offshore operations, it is usual to be necessary for tugboats or supply vessels to keep fixed position, when exposed to environmental loads. Since for many practical situations free surface effects can be neglected, wind and current loads problems can be uncoupled and independently evaluated. Figure 1 illustrates this procedure, showing the splitting process applied to a supply boat model. The present work do not take into account the wave loads, which uses to be considered by potential methods.

This work intends to validate the simulations conducted on ANSYS CFX® for the determination of wind loads on an offshore supply vessel comparing to experimental data collected on wind tunnel. The wind tunnel experiments took place on IPT (Instituto de Pesquisas Tecnológicas), and the simulations were conducted by the Scientific Methods group in Petrobras Research Center.

All wind flow generated forces can be evaluated using a steady-state simulation for the present case. Several vessel heading angles were evaluated, varying from 0º to 345º, due to the asymmetrical arrangement of deck equipments. The heading angles were automatically changed using the software modeFrontier®.

Figure 2 shows a comparison between the simulation and the experiments from wind tunnel measurements for force coeffiicients. The same kind of simulation was performed to determine the current forces on the hull of the vessel described before, but no comparison was made with experimental data from towing tank experiments yet.

Technical Session

REFERENCES

[1] H. Ottens, R. Dijk and G. Meskers, Benchmark Study between CFD and Model Tests on a Semi-submersible Crane Vessel, OMAE2009-79307, ASME 28th International Conference on Offshore Mechanics and Arctic Engineering, 2009.[2] G. Vaz et al., Current Affairs: Model tests, Semi-Empirical Predictions and CFD Computations for Current Coefficients of Semi-Submersibles, OMAE2009-80216, ASME 28th International Conference on Offshore Mechanics and Arctic Engineering, 2009.

33

www.cfdoil.com.br

Figure 1 - Geometry splitting for environmental loads calculation.

Figure 2 - Wind force coefficients (CFD Simulations x IPT Experiments).

Technical Session

[3] J. S. Kim, D. Y. Lee, C. B. Hong and S. M. Ahn, Prediction of Current Load Using Computational Fluid Dynamics, OMAE2009-79307, ASME 28th International Conference on Offshore Mechanics and Arctic Engineering, 2009.[4] D. F. C. Silva, Validação do Cálculo de Esforços de Correnteza e Vento em Embarcações via CFD, Comunicação Técnica PETROBRAS, CT MC001/2010).

34

www.cfdoil.com.br


CFD uSE TO IMPROVE DESIGN AT THE DETAILED ENGINEERING STAGE: COLD By-PASS OPTIMIZATION

Oscar Barroso*†, Juan Martín Catelén†, Leopoldo Millán†

† Techint Engineering & Construction (TEI&C)Hipólito Bouchard 557 - Piso 14 - Ciudad Autónoma de Buenos Aires - Argentina

(C1106ABG)e-mails: [email protected]; [email protected]; [email protected]

ABSTRACT

In the context of the Petrotrin project [1] within the acid regeneration unit, the acid (with lower concentration) coming from the alkylation unit reacts in a combustion chamber to obtain SO2. This gas is converted to SO3 at a reactor. After, the SO3 is absorbed in H2O to give H2SO4 with 98% and 99.2% purity as products. The reaction goes through four stages and temperature control is extremely important to guarantee a required conversion. Between stages is necessary to remove the heat released before. For this purpose there are two heat exchangers; each one counts with a cold bypass to control temperature (Figure 1). Process engineers suspected that temperature profiles problems may appear, since there was not enough space between exchangers 1 and 2; that could lead to deficient temperature control. The license owner was proposing to add two elbows to enlarge the connection, but this option was really expensive as the engineering was with a 90% of advance and the zone was crowded. So a different solution was required. The traditional tools were not useful at this level, so Computational Fluid Dynamic (CFD) software [2] was chosen: first, to analyze the original configuration, and then if problems arise, to try different modifications and see their effect.

On the first simulation, it was found that not only there were inhomogeneous temperature distribution within each entrance (second exchanger’s and second bypass’) but also there were significant differences between temperatures (averaged) at both entrances (352°C versus 273.6°C) when they should be theoretically equal. This temperature difference was not taken into account in the original design and could provoke problems with the temperature control of the second heat exchanger at operation.

Then, as said before, a group of simulations were developed in order to find a simple solution and improve the flow distribution entering from the first bypass to the connection to be able to generate better temperature profiles.

The modification consisted in extending the exit of the first bypass into the connection so the gas could be distributed in a more equilibrated way. The variables for each option were the extension of the pipe, the quantity and the dimension of the holes.

After three alternatives, two kinds of improvements were achieved: first, the difference between the two averaged temperatures at the entrances (second exchanger’s and second bypass’) was reduced from a difference of 78.4°C to a value of 10°C (335.5°C versus 325.5°C) on the final configuration; second, the temperature profile within each entrance and in general, through the entire model, was improved (more homogeneous) avoiding another cause of malfunction in the exchanger (Figure 2). Better profiles could be accomplished through more complex solutions, but this option was chosen for being a good compromise between good results and low construction complexity.

For every option the pressure drop was analyzed to inform the value and be sure that any significant restrictions to the flow would not appear.

As a conclusion it can be said that a CFD tool at the detailed engineering stage offered two possibilities that would have been impossible without it: to study if a possible problem could appear and quantify it; and to choose a simple solution that otherwise could not have been developed since it was out of the standard options.

Technical Session

35

www.cfdoil.com.br

Figure 1 - Process flow diagram of Acid Regeneration Unit.

Figure 2 - Comparison of temperature profiles between initial and final configurations.

REFERENCES

[1] Pointe-a-Pierre Refinery Gasoline Optimization Program; C3 / C4 Sulfuric Acid Alkylation / Sulfuric Acid Regeneration.[2] CFX 12.0, ANSYS, USA).

Technical Session

36

www.cfdoil.com.br


EDGE CFD-ALE: A FINITE ELEMENT SySTEM FOR COMPLEX FLuID-STRuCTuRE INTERACTIONS IN OFFSHORE ENGINEERING

Carlos E. Silva*, Nestor O. Guevara Jr*, José L. D. Alves*, Renato N. Elias†, Anderson L. Mendonça†, Paula A. Sesini†, Rafael March†, Alvaro L. G. A. Coutinho†, Gabriel Guerra+, Erb F. Lins+, Fernando A. Rochinha+,

Milton A. Gonçalves Jr¹, Paulo T. T. Esperança¹, Marcos A. D. Martins², Marcos. A. D. Ferreira²

*Laboratory for Computational Methods in Engineering, Civil Engineering Department, COPPE/UFRJ, Rio de Janeiro, RJ, Brazil

e-mail:{kadu, nestor, jalves}lamce.coppe.ufrj.br†High Performance Computing Center, COPPE/UFRJ, Rio de Janeiro, RJ Brazil

e-mail:{renato,Anderson,psesini,rafael,alvaro)@nacad.ufrj.br+Department of Mechanical Engineering, COPPE/UFRJ, Rio de Janeiro, Brazil

e-mail:{gguera,eflins, farochinha}@mecanica.coppe.ufrj.br¹Department of Ocean Engineering, COPPE/UFRJ, Rio de Janeiro, Brazil

e-mail:{milton,ptarso}@peno.coppe.ufrj.br²Petrobras Research Center, Rio de Janeiro, Brazil

e-mail:{marcos.martins, marcos.donato}@petrobras.com.br

ABSTRACT

This work presents EdgeCFD-ALE, a finite element system for complex fluid-structure interactions designed for offshore hydrodynamics. Green-water decks and wave impact on floating devices are a few examples of problems that can be solved with EdgeCFD-ALE. The fluid component of the software consists of an incompressible finite element edge-based flow solver able to treat free-surface flow problems by a VOF approach [1, 2]. EdgeCFD-ALE is based on the Streamline-Upwind Petrov-Galerkin/Pressure-Stabilized Petrov-Galerkin with the Least Squares Incompressibility Constraint (SUPG/PSPG/LSIC) finite element formulation. Turbulence in EdgeCFD-ALE has been treated by a Smagorinsky model. More recently the Residual-Based Variational Multiscale method to turbulence has been incorporated into EdgeCFD-ALE with success [3]. The fluid-structure problem is treated by the Arbitrary Lagrangian Eulerian (ALE) formulation in which parts of the computational mesh move attached to material particles in a Lagrangian description, while other parts of the mesh remain fixed in space, in an Eulerian description. In between these parts, there is a transition region, connecting them, where the mesh nodes move arbitrarily, so to say, irrespective of material particles motion. In EdgeCFD-ALE special attention is given to the description of the hydrodynamics of an immersed body. The mesh updating scheme adopted, the node repositioning in the neighborhood of a moving body in three-dimensions is accomplished by the solution of a scalar diffusion problem in each spatial coordinate. Problem statement for mesh updating process includes a variable and adaptive local diffusion coefficient, dependent on a relative geometric quality index, computed for each element throughout the mesh. Boundary conditions involve the motion of the immersed body´s surface, i.e., the fluid-structure interface, taken as the Lagrangian portion of the domain in the overall problem. Time integration in EdgeCFD-ALE is performed with a predictor-multicorrector algorithm with adaptive timestepping by a Proportional-Integral-Derivative (PID) controller [4]. Within the flow solution loop, the multi-correction steps correspond to the Inexact-Newton method. In this method the tolerance of the linear solver is adapted according to the history of the solution residua. As a linear solver, EdgeCFD employs the Preconditioned Generalized Minimal Residual Method (GMRES) for both flow and VOF equations. Mesh movement equations are solved by preconditioned conjugate gradients. Most of the computational effort spent in the solution phase is devoted to matrix-vector products. In order to compute such operations more efficiently, Edge-CFD uses an edge-based data structure. This data structure, when applied to problems as those described in this work, is able to reduce indirect memory access, memory requirements to hold the coefficients of the stiffness matrices and the number of floating point operations when compared to other traditional data structures such as element-by-element (EBE) or compressed sparse row (CSR). EdgeCFD-ALE computational kernel is a full parallel Fortran90 finite element code.

Technical Session

37

www.cfdoil.com.br

The computations are performed in parallel using a distributed memory paradigm through the message passing interface library (MPI). Collective and point-to-point communication between subdomains are currently supported. The parallel partitions are generated by Metis/Parmetis library, while the

Documents

BOOK OF ABSTRACTS - CFDOIL BOOK OF ABSTRACTS 4th Latin American CFD Workshop Applied to the Oil and Gas Industry July 12-13, 2010 - Rio de Janeiro - Brazil Organization July 12th July