Luiz Landau

Simple zero thickness kinematically consistent interface elements

Abstract Interface finite elements have been used in many geotechnical and engineering applications. Essentially, these elements must allow relative displacements between two bodies in contact, or separated by a thin material layer. Frequently, interface elements behave as linear elastic bodies up to a limiting stress state. This linear behavior of interfaces is very important, because it will establish when the slip and/or separation occurs, causing stress redistribution over the mesh. In this paper, the mechanical behavior of interface elements is discussed. It is shown that the kinematic inconsistency pointed by Kaliakin and Li [Comp. Geotech. 17 (1995) 225] for the element proposed by Goodman et al. [ASCE J. Soil Mech. Fdns. Div. 99 (1968) 637] also occurs for other interface models, and new interface elements for 2D and 3D analyses without kinematic inconsistencies are proposed.

Virtual Control Desks for Nuclear Power Plant Simulation: Improving Operator Training

Current nuclear power plant simulators employ interfaces that use multiple computer displays to represent a simplified control desk. To make the simulation more realistic and effective, a proposed approach replaces the displays with a virtual control desk that precisely mimics the original.

The application of the Lanczos Mode Superposition Method in dynamic analysis of offshore structures

Abstract In this work the application of the Lanczos Mode Superposition Method in dynamic analysis of offshore structures is studied. It is shown, in two typical examples, that this approach reduces remarkably the computational effort spent on dynamic response computations without loss of accuracy. Furthermore, by an adequate choice of the Lanczos algorithm starting vector, it is also verified that the localized effects, which depend on the higher modes, are well represented.

A study of some triangular discrete Reissner–Mindlin plate and shell elements

Abstract A shell simultaneously supports membrane and plate bending actions, and both shear and membrane locking may occur. If the element is plane, bending and membrane effects are uncoupled, and the shell element is formed by the assembly of plate and membrane elements. This was proposed in the early sixties, but its success was hampered by the lack of good triangular plate elements. Shear locking has been the major problem, and it may be eliminated by using discrete Reissner–Mindlin models. In this paper, we used some discrete triangular plate elements to form flat shells. Membrane and shear locking are studied.

Comparison of lanczos and conjugate gradients for the element-by-element solution of finite element equations on the ibm 3090 vector computer

Abstract Experiences in the iterative solution of large systems of finite element equations on the IBM 3090 600 E/6VF vector computer are presented. The iterative drivers are the preconditioned conjugate gradient method and Lanczos algorithm. The algorithms are implemented utilizing the element-by-element scheme. The numerical experiments show the computer effectiveness of the present implementation, favouring the conjugate gradient algorithms.

RADARSAT-1 images in support of petroleum exploration: the offshore Amazon River mouth example

RADARSAT-1 images were used as an attempt to identify seepage slicks in the Foz do Amazonas Basin, one of the most promising oil and gas exploration frontiers in the Brazilian equatorial margin, a region that is cloud-covered for most of the year. An unsupervised semivariogram textural classifier algorithm was used to enhance areas of smooth texture and low radar backscatter, indicative of these seepage slick targets. Complementary information related to sea surface temperature, cloud top temperature, wind velocity, and modelling for the tidal regime (all obtained as close as possible to the RADARSAT-1 acquisition) was used to support image interpretation and to exclude false targets also characterized by low radar backscatter (e.g., local upwelling, heavy rain cells). Twenty-one target areas were interpreted as seepage slicks. Considering the influence of the oceanographic and environmental conditions on drifting of the interpreted seepage slicks, they were spatially compared with available geological information, including seismically derived structural and isopach maps, and bathymetric data. Most of the interpreted seepage slicks occur in an extensional structural domain of growth faults. These structures are related to Cenozoic gravity tectonics that result from tensional stress in the continental slope due to the huge load of fluvial sediments transported by the Amazon River. Together with previously discovered subcommercial shallow water oil accumulations, the remote detection of seepage slicks is additional evidence of present-day generation and migration phenomena. The results indicate that the use of a remote, swift method to identify offshore natural oil seepage related to active petroleum systems may constitute a well-accepted approach to support exploration in frontier areas like the Foz do Amazonas Basin.

Nonlinear dynamic analysis using the pseudo-force method and the Lanczos algorithm

Abstract This work presents a technique for nonlinear dynamic analysis based upon the combination of the Pseudo-Force method and the Lanczos algorithm. The equations of motion of the structural system, discretized by finite elements, are reduced through the Lanczos algorithm and the nonlinear time integration is performed by the purely iterative Pseudo-Force method. The accuracy and computer effectiveness of such methodology is shown in the analysis of a rectangular elastic plate resting entirely on nonlinear supports.

On the application of an element-by-element lanczos solver to large offshore structural engineering problems

Abstract Iterative methods for solving large sets of linear equations have been used as an alternative to direct methods of solution since the early beginning of numerical analysis. The conjugate gradient method (CCM), one of the most widely used, seeks a solution that minimizes the potential energy of the finite element assemblage. Recently, the use of Lanczos algorithm for the solution of large sets of linear equations has been re-examined. Lanczos biorthogonalization procedure is an oblique projection method that provides a solution approximation whose residual is orthogonal to a Krylov subspace. It has been shown that Lanczos and CGM share several properties but the former has the advantage of not being necessary to compute the approximated solution at each iteration. Jacobi preconditioning can also be employed in order to accelerate convergence. The Lanczos procedure was implemented using an element-by-element (EBE) scheme. The applications spread over typical offshore engineering problems encompassing regular and irregular meshes. These problems are normally ill-conditioned when compared with continuum problems. For all the analyses addressed the element-by-element Lanczos procedures presented outstanding computational efficiency.

A 4D-variational ocean data assimilation application for Santos Basin, Brazil

Aiming to achieve systematic ocean forecasting for the southeastern Brazilian coast, an incremental 4D-Var data assimilation system is applied to a regional ocean model focused mainly in the Santos Basin region. This implementation is performed within the scope of The Santos Basin Ocean Observing System (or Project Azul), a pilot project designed to collect oceanographic data with enough frequency and spatial coverage so to improve regional forecasts through data assimilation. The ocean modeling and data assimilation system of Project Azul is performed with the Regional Ocean Modeling System (ROMS). The observations used in the assimilation cycles include the following: 1-day gridded, 0.1° resolution SST from POES AVHRR; 1-day gridded, 0.3° composite of the MDT SSH from AVISO; and surface and subsurface hydrographic measurements of temperature and salinity collected with gliders and ARGO floats from Project Azul and from UK Met-Office EN3 project dataset. The assimilative model results are compared to forward model results and independent observations, both from remote sensing and in situ sources. The results clearly show that 4D-Var data assimilation leads to an improvement in the skill of ocean hindcast in the studied region.

Volume and heat transports in the world oceans from an ocean general circulation model

Monitoring the volume and heat transports around the world oceans is of fundamental importance in the study of the climate system, its variability, and possible changes. The application of an Oceanic General Circulation Model for climatic studies needs that its dynamic and thermodynamic fields are in equilibrium. The time spent by the model to reach this equilibrium is called spin-up time. This work presents some results obtained from the application of the Modular Ocean Model version 4.0 initialized with temperature, salinity, velocity and sea surface height data already in equilibrium from an ocean data assimilation experiment conducted by Geophysical Fluid Dynamics Laboratory (GFDL). The use of this dataset as initial condition aimed to diminish the spin-up time of the oceanic model. The model was integrated for seven years. Volume and heat transports in different sections around the world oceans showed to be in good agreement with the literature. The results showed a well defined seasonal cycle starting at the second integration year, also some important dynamic and thermodynamic aspects of the Global Ocean Circulation, as the great conveyor belt, are well reproduced. The obtained results could constitute an important data source to be used as initial and boundary conditions in regional ocean model experiments as well as for long integration runs in order to study oceanic climate variability.