26.12.2013 Views

Report - Oregon State Library: State Employee Information Center ...

Report - Oregon State Library: State Employee Information Center ...

Report - Oregon State Library: State Employee Information Center ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

The model uses an explicit method, where the calculation timestep is very short compared with<br />

the time necessary for information (acceleration, velocity, displacement) to physically pass from<br />

one element to another. It also utilizes a Lagrangian formulation, where the incremental<br />

displacements are added to the coordinates at each timestep so that the grid moves and deforms<br />

with the material it represents.<br />

There are several advantages and disadvantages in using FLAC as compared to implicit finite<br />

element programs (Table 6.1). Because of the explicit Lagrangian methodology and the explicit<br />

use of the equations of motion, FLAC is advantageous in modeling non-linear, large strain,<br />

physically unstable situations (Itasca Consulting Group 1997).<br />

Table 6.1: Comparison of FLAC and Finite Element Numerical Programs<br />

FLAC (Explicit)<br />

Timestep must be smaller than a critical value for<br />

stability.<br />

Small amount of computation effort per timestep.<br />

No significant numerical damping introduced for<br />

dynamic solution.<br />

No iterations necessary to follow nonlinear<br />

constitutive law.<br />

Provided that the timestep criterion is always<br />

satisfied, nonlinear laws are always followed in a<br />

valid physical way.<br />

Matrices are never formed. Memory requirements<br />

are always at a minimum. No band-width<br />

limitations.<br />

Since matrices are never formed, large<br />

displacements and strains are accommodated<br />

without additional computing effort.<br />

Finite Element (Implicit)<br />

Timestep can be arbitrarily large, with<br />

unconditionally stable schemes.<br />

Large amount of computational effort per timestep.<br />

Numerical damping dependent on timestep present<br />

with unconditionally stable schemes.<br />

Iterative procedure necessary to follow nonlinear<br />

constitutive law.<br />

Always necessary to demonstrate that the above<br />

mentioned procedure is a) stable, and b) follows the<br />

physically correct path (for path-sensitive problems).<br />

Stiffness matrices must be stored. Ways must be<br />

found to overcome associated problems such as bandwidth.<br />

Memory requirements tend to be large.<br />

Additional computing effort needed to follow large<br />

displacements and strains.<br />

6.1 CONSTITUTIVE SOIL MODEL<br />

An effective stress Mohr-Coulomb constitutive model was used for this study. It is able to model<br />

plastic deformations utilizing a plastic flow rule. The elastic behavior of the soil is defined by the<br />

bulk and shear modulus, and the strength is defined by the angle of friction and cohesion. This<br />

significantly simplifies the dynamic soil behavior; however, it is not capable of accounting for<br />

the strain dependent dynamic properties such as damping and shear modulus. This model has<br />

been demonstrated to yield satisfactory displacement results for a variety of applications<br />

involving seismically-induced deformations of earth structures and retaining walls (Roth et al.<br />

1993; Roth and Inel 1993; Dickenson and Yang 1998; McCullough and Dickenson 1998). The<br />

elastic and strength properties of the soil were estimated from established correlations with<br />

normalized SPT values ((N 1 ) 60 ), unless otherwise noted.<br />

110

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!