Elisabeth Longatte

Application of Arbitrary Lagrange Euler Formulations to Flow-Induced Vibration Problems

Most classical fluid force identification methods rely on mechanical structure response measurements associated with convenient data processes providing turbulent and fluid-elastic forces responsible for possible vibrations and damage. These techniques provide good results; however, they often involve high costs as they rely on specific modelings fitted with experimental data. Owing to recent improvements in computational fluid dynamics, numerical simulation of flow-induced structure vibration problems is now practicable for industrial purposes. As far as flow structure interactions are concerned, the main difficulty consists in estimating numerically fluid-elastic forces acting on mechanical components submitted to turbulent flows. The point is to take into account both fluid effects on structure motion and conversely dynamic motion effects on local flow patterns. This requires a code coupling to solve fluid and structure problems in the same time. This ability is out of limit of most classical fluid dynamics codes. That is the reason why recently an improved numerical approach has been developed and applied to the fully numerical prediction of a flexible tube dynamic response belonging to a fixed tube bundle submitted to cross flows. The methodology consists in simulating at the same time thermo-hydraulics and mechanics problems by using an Arbitrary Lagrange Euler (ALE) formulation for the fluid computation. Numerical results turn out to be consistent with available experimental data and calculations tend to show that it is now possible to simulate numerically tube bundle vibrations in presence of cross flows. Thus a new possible application for ALE methods is the prediction of flow-induced vibration problems. The full computational process is described in the first section. Classical and improved ALE formulations are presented in the second part. Main numerical results are compared to available experimental data in section 3. Code performances are pointed out in terms of mesh generation process and code coupling method

Low Mass-Damping Vortex-Induced Vibrations of a Single Cylinder at Moderate Reynolds Number.

The feasibility and accuracy of large eddy simulation is investigated for the case of three-dimensional unsteady flows past an elastically mounted cylinder at moderate Reynolds number. Although these flow problems are unconfined, complex wake flow patterns may be observed depending on the elastic properties of the structure. An iterative procedure is used to solve the structural dynamic equation to be coupled with the Navier-Stokes system formulated in a pseudo-Eulerian way. A moving mesh method is involved to deform the computational domain according to the motion of the fluid structure interface. Numerical simulations of vortex-induced vibrations are performed for a freely vibrating cylinder at Reynolds number 3900 in the subcritical regime under two low mass-damping conditions. A detailed physical analysis is provided for a wide range of reduced velocities, and the typical three-branch response of the amplitude behavior usually reported in the experiments is exhibited and reproduced by numerical simulation.

Numerical simulation of an elementary Vortex-Induced-Vibration problem by using fully-coupled fluid solid system computation

Numerical simulation of Vortex-Induced-Vibrations (VIV) of a rigid circular elastically-mounted cylinder submitted to a fluid cross-flow has been extensively studied over the past decades, both experimentally and numerically, because of its theoretical and practical interest for understanding Flow-Induced-Vibrations (FIV) problems. In this context, the present article aims to expose a numerical study based on fully-coupled fluid-solid computations compared to previously published work [34], [36]. The computational procedure relies on a partitioned method ensuring the coupling between fluid and structure solvers. The fluid solver involves a moving mesh formulation for simulation of the fluid structure interface motion. Energy exchanges between fluid and solid models are ensured through convenient numerical schemes. The present study is devoted to a low Reynolds number configuration. Cylinder motion magnitude, hydrodynamic forces, oscillation frequency and fluid vortex shedding modes are investigated and the “lock-in” phenomenon is reproduced numerically. These numerical results are proposed for code validation purposes before investigating larger industrial applications such as configurations involving tube arrays under cross-flows [4].

Benchmark of Numerical Codes for Coupled CSD/CFD Computations on an Elementary Vortex Induced Vibration Problem

Numerical simulation of vortex-induced-vibrations (VIV) of an elastically supported rigid circular cylinder in a fluid cross-flow has been thoroughly studied over the past years, both from the experimental and numerical points of view, because of its theoretical and practical interest in the understanding of flow-induced vibrations problems. In this context, the present paper aims at exposing a numerical study based on a coupled fluid-structure simulation, compared with previously published studies [34], [36]. The computational procedure relies on a partitioned method ensuring the coupling between fluid and structure solvers. The fluid solver involves a moving mesh formulation for simulation of the interface motion. Energy exchanges between both systems are ensured through convenient coupling schemes. The present study is devoted to a low Reynolds number configuration ( Re = 100). Cylinder motion magnitude, hydrodynamic forces, oscillation frequency and fluid vortex shedding modes are investigated with the intention to observe the “lock-in” phenomenon. These numerical simulations are proposed for code validation purposes prior to industrial applications to tube bundle configurations [4].Copyright

Cross Flow Induced Vibration in a Single Tube of Square Array Using LES

Large eddy simulations (LES) of a single phase water flow through a square normal tube bundle at Reynolds numbers from 2000 to 6000 is performed to investigate the fluid-elastic instability. A single cylinder of the array is allowed to oscillate in one degree of freedom (1-DOF) in the flow normal direction, similar as in the corresponding experiments. The fluid-structure coupling is simulated using the Arbitrary Lagrangian-Eulerian (ALE) approach. The subgrid scale turbulence is modeled using the standard Smagorinsky’s eddy-viscosity model. The LES results show a good agreement with the experimental results, in terms of the response frequency and damping ratio of the cylinder vibration. The dynamic case simulations are compared with static cases over the range of Reynolds numbers by means of the pressure profiles on the cylinder surface and the probe velocity spectra.

Large Eddy Simulation of Fluid-Elastic Instability in Square Normal Cylinder Array

Large Eddy Simulations (LES) are performed at low Reynolds number (2000 upto 6000) to investigate the dynamic fluid-elastic instability in square normal cylinder array for a single-phase fluid cross flow. The fluid-elastic instability is dominant in flow normal direction, at least for all water-flow experiments (Price et al. [18]). The instability appears even in the case of single moving cylinder in an otherwise fixed-cylinder arrangement resulting in the same critical velocity (Khalifa et al. [1]). Therefore, in the present work only a central cylinder out of 20 cylinders is allowed to vibrate in flow normal direction. The square normal (90°) array has 5 rows and 3 columns of cylinders with 2 additional side columns of half wall-mounted cylinders. The numerical configuration is a replica of the experimental setup except for the length of cylinders, which is 4 diameters (4D) in numerical setup against about 8D in the experiment facility. The single-phase fluid is water. The standard Smagorinsky turbulence model is used for the sub-grid scale eddy viscosity modeling. The numerical results are analysed and compared with the experimental results, for a range of flow velocities in the vicinity of the instability. Moreover, instantaneous pressure and fluid-force profiles on the cylinder surface are extracted from the LES calculations in order to better understand the dynamic fluid-elastic instability.Copyright

Phase Lag Model for Fluidelastic Instability in Square Cylinder Arrangement

In this paper, a phase lag model is proposed in order to predict the fluid velocity threshold for fluidelastic dynamic instability of a square cylinder arrangement under cross flow. A theoretical formulation of a total damping, including the added damping in still fluid, the damping due to fluid flow and the damping derived from the phase shift between the fluid force and tube displacement, is given. A function of fluid and structure parameters, such as reduced velocity, pitch ratio and Scruton number, is thus obtained. It is shown that this function, taken as function of the reduced velocity variable, vanishes at the critical reduced velocity from which the fluidelastic dynamic instability of the tube occurs. Obviously, the value of the critical velocity is depending on other fluid-structure parameters. The obtained results are compared to experimental ones and those obtained from other theoretical models.Copyright

Study of a Fluid-Structure Interaction Instability Mechanism in a Tube Bundle With Multiphase-POD Approach

Multiphase-Proper Orthogonal Decomposition Reduced-Order Method has been proven to be efficient for the low-cost study of fluid-structure interaction mechanisms. Applications to a single tube under cross-flow, then to a tube bundle system revealed good behaviours of this method, which was shown to be able to accurately reproduce the velocity flow field as well as the solid displacement, even in the case of large magnitudes. The goal here is to go further by studying an instability mechanism with the Multiphase-POD technique, involving a tube array configuration because of its high interest in the nuclear domain. We first want to know if this method can reproduce critical to unstable cases and finally, we are interested in the possibility of leading a parametric study coupled with the Multiphase-POD Method in order to evaluate the instability threshold. Indeed, parametric studies coupled with a reduced-order method could lead to a CPU time additional gain, since only one basis calculation could cover several configurations with low computational cost.Copyright

Numerical Study of Fluid-Structure Interactions in Tube Bundles With Multiphase-POD Reduced-Order Approach

Fluid-Structure Interactions are present in a large number of systems of nuclear power plants and nuclear on-board stoke-holds. Particularly in steam generators, where tube bundles are submitted to cross-flow which can lead to structure vibrations. We know that numerical studies of such a complex mechanism is very costly, that is why we propose the use of reduced-order methods in order to reduce calculation times and to make easier parametric studies for such problems.We use the multiphase-POD approach, initially proposed by Liberge (E. Liberge; POD-Galerkin Reduction Models for Fluid-Structure Interaction Problems, PhD Thesis, Universite de La Rochelle, 2008). This method is an adaptation of the classical POD approach to the case of a moving structure in a flow, considering the whole system (fluid and structure) as a multiphase domain. We are interested in the case of large displacements of a structure moving in a fluid, in order to observe the ability of the multiphase-POD technique to give a satisfying solution reconstruction. We obtain very interesting results for the case of a single circular cylinder in cross-flow (lock-in phenomenon). Then we present the application of the method to a case of confined cylinders in large displacements too. Here again, results are encouraging.Finally, we propose to go further presenting a first step in parametric studies with POD-Galerkin approach. We only consider a flowing-fluid around a fixed structure and the Burgers’ equation. A future work will consist in applications to fluid-structure interactions.Copyright

Tackling FSI Simulation for FIV Problems in Tube Bundle Systems With POD Approach

Tube bundles in steam boilers of nuclear power plants and nuclear on-board stokehold are known to be exposed to high levels of vibrations under flowing fluid. This coupled fluid-structure problem is still a challenge for engineers, first because of the difficulty to fully understand it, second because of the complexity for setting it up numerically. Although numerical techniques could help the understanding of such a mechanism, a complete simulation of a fluid past a whole elastically mounted tube bundle is currently out of reach for engineering purposes. To get round this problem, the use of a reduced-order model has been proposed with the introduction of the widely used Proper Orthogonal Decomposition (POD) method for a flow past a fixed structure [M. Pomarede, E. Liberge, A. Hamdouni, E.Longatte, & J.F. Sigrist - Simulation of a fluid flow using a reduced-order modelling by POD approach applied to academic cases; PVP2010, July 18–22, Seattle]. Interesting results have been obtained for the reconstruction of the flow. Here a first step is to propose to consider the case of a flow past a fixed tube bundle configuration in order to check the good reconstruction of the flow. Then, an original approach proposed by Liberge (E. Liberge; POD-Galerking Reduction Models for Fluid-Structure Interaction Problems, PhD Thesis, Universite de La Rochelle, 2008) is applied to take into account the fluid-structure interaction characteristic; the so-called “multiphase” approach. This technique allows applying the POD method to a configuration of a flow past an elastically mounted structure. First results on a single circular cylinder and on a tube bundle configuration are encouraging and let us hope that parametric studies or prediction calculations could be set up with such an approach in a future work.© 2011 ASME