Redesign of Submerged Structures by Large Admissible Perturbations

2001 ◽  
Vol 123 (3) ◽  
pp. 103-111 ◽  
Author(s):  
Vincent Y. Blouin ◽  
Michael M. Bernitsas
Keyword(s):  
Finite Element ◽  
Optimization Problem ◽  
Mass Matrix ◽  
Added Mass ◽  
Forced Response ◽  
Initial Structure ◽  
General Perturbation ◽  
Large Admissible Perturbations ◽  
Fluid Added Mass ◽  
Perturbation Equations

The method of large admissible perturbations (LEAP) is a general methodology, which solves redesign problems of complex structures without trial and error or repetitive finite element analyses. When forced vibration constraints are incorporated into the redesign problem, damping and added mass due to the presence of fluid must be included into the model. The corresponding terms introduce theoretical and numerical difficulties, which are treated in this paper. The LEAP method has been implemented into a Fortran computer code RESTRUCT, developed at the University of Michigan. The redesign process is mathematically formulated as an optimization problem with nonlinear constraints, called general perturbation equations. First, a finite element analysis of the initial structure is executed. Then, the results are postprocessed by code RESTRUCT using an incremental scheme to find the optimum solution for the problem defined by the designer. Accurate determination of nonstructural terms, such as fluid added mass, is generally detrimental as far as forced response analysis is concerned. In redesign problems, however, simple but realistic models can be used. A simple transformation of the structural mass matrix is used to compute the added mass matrix and its dependency on the redesign variables. The presence of non-structural terms in the general perturbation equations requires the development of a new LEAP algorithm for solution of the optimization problem. A simple cantilever beam with 100 degrees of freedom is used to validate the fluid added mass model. The developed method and algorithm are then applied to a partially submerged 4,248 degree of freedom complex structure modeled with beam elements.

Cognate Space Identification for Forced Response Structural Redesign

2003 ◽  
Vol 127 (3) ◽  
pp. 227-233
Author(s):  
Vincent Y. Blouin ◽  
Michael M. Bernitsas
Keyword(s):  
Direct Method ◽  
A Priori ◽  
Normal Modes ◽  
Forced Response ◽  
Response Amplitude ◽  
Computational Time ◽  
Practical Guidelines ◽  
Structural Redesign ◽  
Large Admissible Perturbations ◽  
Perturbation Equations

Large admissible perturbations (LEAP) is a general methodology, which solves redesign problems of complex structures with, among others, forced response amplitude constraints. In previous work, two LEAP algorithms, namely the incremental method (IM) and the direct method (DM), were developed. A powerful feature of LEAP is the general perturbation equations derived in terms of normal modes, the selection of which is a determinant factor for a successful redesign. The normal modes of a structure may be categorized as stretching, bending, torsional, and mixed modes and grouped into cognate spaces. In the context of redesign by LEAP, the physical interpretation of a mode-to-response cognate space lies in the fact that a mode from one space barely affects change in a mode from another space. Perturbation equations require computation of many perturbation terms corresponding to individual modes. Identifying modes with negligible contribution to the change in forced response amplitude eliminates a priori computation of numerous perturbation terms. Two methods of determining mode-to-response cognate spaces, one for IM and one for DM, are presented and compared. Trade-off between computational time and accuracy is assessed in order to provide practical guidelines to the designer. The developed LEAP redesign algorithms are applied to the redesign of a simple cantilever beam and a complex offshore tower.

Static Stress Redesign of Plates by Large Admissible Perturbations

2005 ◽  
Author(s):  
Bhineka M. Kristanto ◽  
Michael M. Bernitsas
Keyword(s):  
Plate Thickness ◽  
Forced Response ◽  
Static Stress ◽  
Perturbation Equation ◽  
Shell Elements ◽  
Modal Properties ◽  
General Perturbation ◽  
Structure Changes ◽  
Admissible Perturbation ◽  
Large Admissible Perturbations

The LargE Admissible Perturbation (LEAP) methodology is developed further to solve static stress redesign problems for shell elements. The static stress general perturbation equation, which expresses the unknown stresses of the objective structure in terms of the baseline structure stresses, is derived first. This equation depends on the redesign variables for each element or group of elements; namely the plate thickness. LEAP enables the designer to redesign a structure to achieve specifications on modal properties, static displacements, forced response amplitudes, and static stresses. LEAP is implemented in code RESTRUCT which post-processes the FEA results of the baseline structure. Changes on the order of 100% in the above performance particulars and in redesign variables can be achieved without repetitive FEA’s. Several numerical applications on a simple plate and an offshore tower are used to verify the LEAP algorithm for stress redesign.

Prediction of Low Frequency Vibrational Frequencies of Submerged Structures

1991 ◽  
Vol 113 (2) ◽  
pp. 187-191 ◽  
Author(s):  
G. C. Everstine
Keyword(s):  
Finite Element ◽  
Boundary Element ◽  
Mass Matrix ◽  
Low Frequency ◽  
Added Mass ◽  
Structural Acoustics ◽  
Mass Matrices ◽  
Element Approach ◽  
Fully Coupled ◽  
Vibrational Resonances

Practical numerical techniques are described for calculating the low frequency vibrational resonances of general submerged structures. Both finite element and boundary element approaches for calculating fully-coupled added mass matrices are presented and illustrated. The finite element approach is implemented using existing structural analysis capability in NASTRAN. The boundary element approach uses the NASHUA structural acoustics program in combination with NASTRAN to compute the added mass matrix. The two procedures are compared in application to a submerged cylindrical shell with flat end closures. Both procedures proved capable of computing accurate submerged resonances; the more elegant boundary element procedure is easier to use but may be more expensive computationally.

The Equilibrium Cell-Based Smooth Finite Element Method for Shakedown Analysis of Structures

2019 ◽  
Vol 16 (05) ◽  
pp. 1840013 ◽  
Author(s):  
P. L. H. Ho ◽  
C. V. Le ◽  
T. Q. Chu
Keyword(s):  
Finite Element Method ◽  
Finite Element ◽  
Optimization Problem ◽  
Conic Programming ◽  
Equilibrium Equations ◽  
Shakedown Analysis ◽  
Numerical Examples ◽  
Elastic Stresses ◽  
Gauss Points ◽  
Element Method

This paper presents a novel equilibrium formulation, that uses the cell-based smoothed method and conic programming, for limit and shakedown analysis of structures. The virtual strains are computed using straining cell-based smoothing technique based on elements of discretized mesh. Fictitious elastic stresses are also determined within the framework of finite element method (CS-FEM)-based Galerkin procedure, and equilibrium equations for residual stresses are satisfied in an average sense at every cell-based smoothing cell. All constrains are imposed at only one point in the smoothing domains, instead of Gauss points as in a standard FEM-based procedure. The resulting optimization problem is then handled using the highly efficient solvers. Various numerical examples are investigated, and obtained solutions are compared with available results in the literature.

A Simplified Finite Element for Added Mass and Inertial Coupling in Arrays of Cylinders

1985 ◽  
Vol 107 (2) ◽  
pp. 118-125 ◽  
Author(s):  
R. E. Harris ◽  
M. A. Dokainish ◽  
D. S. Weaver
Keyword(s):  
Finite Element ◽  
Time Integration ◽  
Element Model ◽  
Added Mass ◽  
Cylindrical Structures ◽  
Beam Elements ◽  
Inertial Coupling ◽  
Fundamental Frequencies ◽  
Tube Bundles ◽  
Primary Advantage

A simplified finite element has been developed for modeling the added mass and inertial coupling arising when clusters of cylinders vibrate in a quiescent fluid. The element, which is based on two-dimensional potential flow theory, directly couples two adjacent beam elements representing portions of the adjacent cylindrical structures. The primary advantage of this approach over existing methods is that it does not require the discretization of the surrounding fluid and, therefore, is computationally much more efficient. The fundamental frequencies of tube bundles of various pitch ratios have been predicted using this method and compared with experimental data. Generally, the agreement is good, especially for the bandwidth of fluid coupled natural frequencies. The transient response of tube bundles is also examined using time integration of the finite element model. The beating phenomenon and time decay characteristics exhibited by the experimental bundles under single-tube excitation are well predicted and valuable insights are gained into the measurement of damping in tube bundles.

Consistent Hydrodynamic Mass for Parallel Prismatic Beams in a Fluid-Filled Container

1984 ◽  
Vol 106 (3) ◽  
pp. 270-275
Author(s):  
J. F. Loeber
Keyword(s):  
Finite Element ◽  
Dynamic Response ◽  
Incompressible Fluid ◽  
Finite Element Methods ◽  
Mass Matrix ◽  
Three Dimensional ◽  
Beam Length ◽  
Step Process ◽  
Beam Bending

In this paper, representation of the effects of incompressible fluid on the dynamic response of parallel beams in fluid-filled containers is developed using the concept of hydrodynamic mass. Using a two-step process, first the hydrodynamic mass matrix per unit (beam) length is derived using finite element methods with a thermal analogy. Second, this mass matrix is distributed in a consistent mass fashion along the beam lengths in a manner that accommodates three-dimensional beam bending plus torsion. The technique is illustrated by application to analysis of an experiment involving vibration of an array of four tubes in a fluid-filled cylinder.

Measured Static and Rotordynamic Coefficient Results for a Rocker-Pivot, Tilting-Pad Bearing With 50 and 60% Offsets

2012 ◽  
Vol 134 (5) ◽  
Author(s):  
Chris D. Kulhanek ◽  
Dara W. Childs
Keyword(s):  
Dynamic Stiffness ◽  
Mass Matrix ◽  
Excitation Frequency ◽  
Quadratic Function ◽  
Added Mass ◽  
Damping Coefficients ◽  
Stiffness Coefficients ◽  
Tilting Pad ◽  
Frequency Independent ◽  
Direct Stiffness

Static and rotordynamic coefficients are measured for a rocker-pivot, tilting-pad journal bearing (TPJB) with 50 and 60% offset pads in a load-between-pad (LBP) configuration. The bearing uses leading-edge-groove direct lubrication and has the following characteristics: 5-pads, 101.6 mm (4.0 in) nominal diameter,0.0814 -0.0837 mm (0.0032–0.0033 in) radial bearing clearance, 0.25 to 0.27 preload, and 60.325 mm (2.375 in) axial pad length. Tests were performed on a floating bearing test rig with unit loads from 0 to 3101 kPa (450 psi) and speeds from 7 to 16 krpm. Dynamic tests were conducted over a range of frequencies (20 to 320 Hz) to obtain complex dynamic stiffness coefficients as functions of excitation frequency. For most test conditions, the real dynamic stiffness functions were well fitted with a quadratic function with respect to frequency. This curve fit allowed for the stiffness frequency dependency to be captured by including an added mass matrix [M] to a conventional [K][C] model, yielding a frequency independent [K][C][M] model. The imaginary dynamic stiffness coefficients increased linearly with frequency, producing frequency-independent direct damping coefficients. Direct stiffness coefficients were larger for the 60% offset bearing at light unit loads. At high loads, the 50% offset configuration had a larger stiffness in the loaded direction, while the unloaded direct stiffness was approximately the same for both pivot offsets. Cross-coupled stiffness coefficients were positive and significantly smaller than direct stiffness coefficients. Negative direct added-mass coefficients were obtained for both offsets, especially in the unloaded direction. Cross-coupled added-mass coefficients are generally positive and of the same sign. Direct damping coefficients were mostly independent of load and speed, showing no appreciable difference between pivot offsets. Cross-coupled damping coefficients had the same sign and were much smaller than direct coefficients. Measured static eccentricities suggested cross coupling stiffness exists for both pivot offsets, agreeing with dynamic measurements. Static stiffness measurements showed good agreement with the loaded, direct dynamic stiffness coefficients.

Sign in / Sign up

Export Citation Format

Share Document