H. Pina

Multi-objective optimization of structures topology by genetic algorithms

This work develops a computational model for topology optimization of linear elastic structures for situations where more than one objective function is required, each one of them with a different optimal solution.The method is thus developed for multi-objective optimization problems and is based on Genetic Algorithms. Its purpose is to evolve an evenly distributed group of solutions (population) to obtain the optimum Pareto set for the given problem.To reduce computational effort, optimal solutions of each of the single-objective problems are introduced in the initial population.Two numerical examples are presented and discussed to assess the method.

The influence of radiation absorption on solar pond stability

This paper presents a study of the gravitational stability of a salty layer of a fluid subject to an adverse temperature gradient as a result of heat absorption. This is intended to model solar ponds where an artificial gradient of salt concentration in water is used to prevent convective motions induced by the absorption of solar radiation. The stability of the Boussinesq approximation of the Navier-Stokes equations is analysed for perturbations of a certain kind imposed on the stationary solution. The marginal states for the onset of convection are obtained using a Galerkin method based on a weak formulation of the governing equations. The analysis considers solar energy absorption in the layer and assumes prescribed heat flux values as boundary conditions for the temperature equation. Results are compared with those obtained earlier by different authors for a layer of fluid, heated from below, with linear profiles of both salt concentration and temperature.

The influence of non-constant diffusivities on solar ponds stability

A solar pond is a basin of water where solar energy is trapped due to an artificially created gradient of salinity that prevents convective motions. The present study intends to clarify the contribution of non-constant diffusion coefficients for the stability of the gradient layer together with the influence of solar radiation absorption, the thermal and molecular diffusivities being assumed to be linear functions of the vertical co-ordinate z. The analysis shows that the consideration of these two effects decreases the margin of stability in comparison with previous studies based on a layer of fluid heated from below with constant diffusivities coefficients and linear profiles for both temperature and salt.

Iterative techniques for 3-D boundary element method systems of equations

A key issue in the boundary element method (BEM) is the solution of the associated system of algebraic equations whose matrices are dense, nonsymmetric and sometimes ill conditioned. For large scale tridimensional problems, direct methods like Gauss elimination become too expensive and iterative methods may be preferable. This paper presents a comparison of the performances of some iterative techniques based on conjugate gradient solvers as conjugate gradient squared (CGS) and bi-conjugate gradient (Bi-CG) that seem to have the potential to be efficient and competitive for BEM algebraic systems of equations, specially when used with an appropriate preconditioner. A comparison with the direct application of the conjugate gradient method to the normalized systems of equations (CGNE and CGNR) is also presented.

Iterative solvers for BEM algebraic systems of equations

A key issue in the boundary element method (BEM) is the solution of the associated system of algebraic equations with matrices that are dense and sometimes ill-conditioned. For tridimensional problems, with large scale systems (several thousands of equations) direct methods like Gauss elimination become too expensive and iterative methods may be preferable. For these problems there are already many algorithms, namely for general non-symmetric systems. Most of them can be viewed as Lanczos or conjugate gradient-like methods. Here we present some iterative techniques based on conjugate gradient solvers as descent methods (DM), bi-conjugate gradient (Bi-CG), conjugate gradient squared (CGS) and bi-conjugate gradient stab (Bi-CGstab), that seem to have the potential to be competitive for BEM algebraic systems of equations, especially when used with an appropriate preconditioner.

Nonlinear analysis of the orbital motions of immersed rotors using a spectral/Galerkin approach

Abstract We recently developed a symbolic-numerical formulation for the nonlinear planar motion of rotors under fluid confinement, based on a spectral/Galerkin approach, for gap geometries of about δ = H / R ≈0.1––where H is the average annular gap and R is the rotor radius. Results showed a quite good agreement between the class of approximate models generated, the corresponding analytical exact planar model and experiments. This methodology can be almost entirely automated on a symbolic computing environment. In the present paper this symbolic-numerical spectral/Galerkin procedure is extended in order to deal with nonlinear orbital motions–– X ( t ) and Y ( t ) taking place in orthogonal directions. Numerical simulations performed over a centered rotor configuration maintained by nonisotropic supports ( K st Y / K st X =0.4, where K st X and K st Y stand for the structural stiffnesses), which exhibit interesting dynamics, show a quite good agreement between this type of approximate models and the corresponding analytical exact (but quite involved) model, developed in the past by the authors. With the proposed symbolic-numerical approach one can obtain accurate nonlinear dynamical formulations enabling the study, understanding and prediction of nonlinear orbital rotor dynamics.

An improved linear model for rotors subject to dissipative annular flows

Abstract In a previous paper, Antunes, Axisa, and co-workers, developed a linearized model for the dynamic of rotors under moderate fluid confinement, based on classical perturbation analysis, covering two different cases: (i) dissipative motions of a centered rotor; (ii) motions of an eccentric rotor for frictionless flow. Following the same procedures and assumptions, we derive here an improved model for the more general case of dissipative linearized motions of an eccentric rotor . Besides the rotor motion variables, a new variable—which can be interpreted as the fluctuating term of the average tangential velocity—is introduced, yielding an additional eigenvalue in the linear analysis. The new variable introduced, which is coupled with the rotor motions, is very convenient when frictional effects are not neglected. Under dissipative flows, a richer modal behavior is highlighted, which can be related to delay effects of the flow responses to the rotor motions. Our approach can be applied as well to other flow-excited systems, for example, those subjected to axial or leakage flows. Because rotor-dynamics are strongly dependent on the mean rotor eccentricity, the adequacy of this (or any other) model rely on using the actual value for such parameter.

Nonlinear Analysis of the Orbital Motions of Immersed Rotors Using a Spectral/Galerkin Approach

We recently developed a symbolic-numerical formulation for the nonlinear planar motion of rotors under fluid confinement, based on a spectral/Galerkin approach, for gap geometries of about δ = H/R ≈ 0.1 — where H is the average annular gap and R is the rotor radius. Results showed a quite good agreement between the class of approximate models generated, the corresponding analytical exact planar model and experiments. This methodology can be almost entirely automated on a symbolic computing environment. In the present paper this symbolic-numerical spectral/Galerkin procedure is extended in order to deal with nonlinear orbital motions — X (t) and Y (t) taking place in orthogonal directions. Numerical simulations performed over a centered rotor configuration maintained by non-isotropic supports (Ky st /Kx st = 0.4, where kx st and ky st stand for the structural stiffnesses), which exhibit an interesting dynamics, show a quite good agreement between this type of approximate models and the corresponding analytical exact (but quite involved) model, developed in the past by the authors. With the proposed symbolic-numerical approach one can obtain accurate nonlinear dynamical formulations enabling the study, understanding and prediction of nonlinear orbital rotor-dynamics.Copyright

Editorial to special issue of Computers & Structures

This special issue includes twenty-three extended, reviewed and revised papers originally presented at the Eleventh International Conference on Computational Structures Technology (CST2012) and the Eighth International Conference on Engineering Computational Technology (ECT2012) held concurrently in Dubrovnik, Croatia, 4–7 September 2012. The CST2012 Young Researcher Best Paper Prize was presented to Dr J.-W. Simon, of the RWTH Aachen University, Germany, for his paper “Limit States of Structures in n-dimensional Loading Spaces with Limited Kinematical Hardening”. The ECT2012 Young Researcher Best Paper Prize was presented to Dr L. Barbie, of CEA-Cadarache, France, for the paper “An Automatic Multilevel Refinement Technique based on Nested Local Meshes for Nonlinear Mechanics” by L. Barbie, I. Ramiere and F. Lebon. The CST2012 Prize Paper is the first paper included in this special issue; while the ECT2012 Prize paper is the second paper included in this issue. The standard of the papers considered for these prizes was very high and the conference editors wish to thank the members of the conference editorial board for their work in considering all the applicants

Optimization of baffle configurations to prevent aeroacoustic instabilities in heat exchangers: Preliminary experiments

Gas heat exchangers are prone to aeroacoustic instabilities, which often lead to severe noise levels, structural vibrations, and fatigue. Actually, this problem is solved by placing rigid baffles inside the container, which modify the acoustic modal fields and eventually inhibit the instability. For realistic industrial components using a restricted number of acoustical baffles, their optimal location is a challenging problem, as trial and error experimentation is often a costly and frustrating procedure. Recently, some strategies were proposed for the optimal location of a single baffle in a typical re‐heater from a power station boiler, based on simulated annealing as well as genetic algorithm approaches. In this paper and using the above‐mentioned optimization strategies, a more complex case of the problem—the optimal location of a given number (two or more) of baffles in a typical re‐heater, was addressed. Some preliminary experiments were performed and compared with the simulation results. From the d...