Scenario-focused three-dimensional computational modeling in geomechanics

Modélisation tri-dimensionelle de calcul en géomécanique base sur l’étude de scenarios

A. Elgamal & J. Lu
University of California, San Diego, La Jolla, California, USA
Z. Yang
California Department of Transportation, Oakland, California, USA
T. Shantz
California Department of Transportation, Sacramento, California, USA

ABSTRACT
Three-dimensional nonlinear finite element simulations are becoming increasingly feasible for geotechnical applications. This paper
presents a versatile framework that helps streamline the use of three-dimensional finite elements for analyses of soil and soil-structure
systems. In this regard, a Windows-based graphical-user-interface OpenSeesPL is developed for footing/pile-ground interaction
analyses. OpenSeesPL allows convenient studies of three-dimensional seismic (earthquake) and/or push-over pile analyses. Various
ground modification scenarios may be also addressed by appropriate specification of the material within the pile zone. The presented
analysis scenarios aim to highlight the analysis framework capabilities and range of potential applications.

RÉSUMÉ
Les simulations tri-dimensionelles nonlinéaires par éléments finis deviennent de plus en plus réalisables dans le domaine de la
géotechnique. Cet article présente un cadre de calcul polyvalent qui facilite l’utilisation d’éléments finis tri-dimensionelles pour les
analyses de sol et de systèmes sol-structure. A cet égard, un interface d’utilisateur graphique de calcul basé sur Windows,
OpenSeesPL, est dévelopé pour les analyses d’intéraction sol-fondation/pieux. OpenSeesPL permets d’étudier le comportement des
systèmes de fondation sur pieux soumis à des charges sismiques dynamiques et/ou des charges quasi statiques. Différents scénarios
de modification des sols peuvent être aussi examinés par spécification appropriée des matériaux dans la zone des fondations à pieux.
Les scénarios de modélisation présentés visent à mettre en évidence les capacités du cadre de calcul et le spectre d’applications
potentielles.

Keywords : Geotechnical, Finite Element, Pile Foundations, Earthquake, Seismic, Numerical modeling

1 INTRODUCTION 2 COMPUTATIONAL FRAMEWORK

Soil-Foundation-Structure Interaction (SFSI) is an important The open-source platform OpenSees (http://opensees.

aspect affecting the performance of structures such as buildings berkeley.edu, Mazzoni et al. 2006) is employed
and bridges. With the recent developments in numerical throughout. OpenSees is a software framework for
modeling techniques and high-speed efficient computers, linear developing applications to simulate the performance of
and nonlinear three-dimensional (3D) finite-element (FE) structural and geotechnical systems subjected to
methods are becoming an effective technique for understanding earthquakes. OpenSees can be used to study the
the involved SFSI mechanisms. Particularly suited to seismic performance of infrastructure facilities (bridges, buildings,
applications, the open-source computational platform OpenSees etc.) under static loads, and during earthquake events.
(Mazzoni et al. 2006) provides such 3D simulation capabilities. In the OpenSees platform, a wide range of linear and
However, in conducting numerical simulations, preparation of nonlinear soil and structural elements is available. The
the FE data file is a step that requires careful attention. A minor reported pre- and post-processing scenarios are generated
oversight might go undetected, leading to erroneous results. by the user interface OpenSeesPL (http://cyclic.ucsd.edu/
Numerous opportunities for such small errors abound, and a openseespl) which allows for: i) convenient generation of
user-friendly interface can significantly alleviate this problem, the mesh and associated boundary conditions and loading
and allow for high efficiency and much increased confidence. parameters (FE input file), ii) execution of the
On this basis, a user-interface “OpenSeesPL” is under computations using the OpenSees platform, and iii)
development (Figures 1 and 2), to allow for the execution of graphical display of the results for the footing/pile and the
push-over and seismic footing/pile-ground simulations (Lu et al. ground system.
2006, https://neesforge.nees.org/projects/openseespl/). Various
ground modification scenarios may be also studied by 3. MODELING CONFIGURATIONS
appropriate specification of the material within the pile zone.
In the following sections, an overview of OpenSeesPL The OpenSeesPL graphical interface (pre- and post-
capabilities is presented, followed by a range of potential processor) is focused on facilitating a wide class of 3D
simulation scenarios. As such, the aim is to highlight the analysis studies (with additional capabilities yet under
framework capabilities and range of potential applications. development). The basic default configuration is in the
form of a 3-dimensional soil island with the possibility of
including a footing/pile/pile-group model. Full-mesh, half-mesh,
or quarter mesh configurations may be analyzed, as dictated by
symmetry considerations (Figures 3-5).

Figure 5. Quarter 3D mesh-pile configuration.

In OpenSeesPL, the mesh configuration may be easily

modified to: i) change the pile diameter, depth of
embedment, height above ground surface, and number of
pile beam-column elements, and ii) refine the ground mesh
domain in the lateral and vertical directions (Figure 6).
Figure. 1. OpenSeesPL user interface with mesh showing a
circular pile in level ground (Lu et al. 2006).

Figure. 2. Push-over analysis and deformed mesh window in

OpenSeesPL (Lu et al. 2006).

Figure 6. Mesh refinement in OpenSeesPL.

Square or circular pile cross-sections may be specified.

As such, the pile model may be employed to study the
response of a tall building that can be modeled as a
Figure 3. Full 3D mesh-pile configuration. bending beam (Figure 7). Shallow foundations (rigid) in
square or circular configurations may be also conveniently
analyzed (Figure 8).

Independent control over the pile zone material may be

exercised, allowing for a wide range of ground
modifications studies (Figure 9). Of particular importance
and significance in these scenarios is the ability to simulate
the presence of a mild infinite-slope configuration,
allowing estimates of accumulated ground deformation,
efficacy of a deployed liquefaction countermeasure, pile-
pinning effects, and liquefaction-induced lateral pile loads
and resulting moments/stresses.
Figure 4. Half-3D mesh-pile configuration.
 4. LOAD APPLICATION

Static and dynamic loads may be applied. For static

loading, push-over type analyses may be conducted where
the loads/moments are directly applied to the pile top or
footing surface, in force or in displacement modes (Figure
12). Capabilities are provided for monotonic loading,
cyclic loading, and for user-defined load patterns to be
uploaded as text file. Push-over along the finite element
mesh boundary may be also specified, for instance to
explore loads on pile foundations due to lateral ground
Figure 7. Building modeled as a bending beam on a shallow displacement (Figure 13).
foundation embedded in the ground.

Figure 8. Circular shallow foundation model.

Material for the pile-soil interfacing zone may be also

specified by the user, permitting scenarios such as analysis of
cylindrical foundations, and/or control over pile-soil friction and
potential no-tension interaction during lateral deformation
(Figure 10). In addition to the footing and single pile
configurations, pile groups may be also represented in the free
head or fixed head configurations (Figure 11).

Figure 10. Control over specification of pile-soil interfacing zone.

Figure 11. Large pile group model (1/2 mesh configuration).

Dynamic and earthquake shaking may be also imparted

along the soil lower boundary (base). Shaking is allowed
in 3D with a small set of available motions, and a
capability to upload user specified base shaking excitation
Figure 9. Control over specification of soil inside the pile zone. (Figure 14).
 Figure 15. Specification of soil lateral boundary conditions.

6 SOIL PROPERTIES

Linear and nonlinear elasto-plastic cyclic soil modeling

capabilities are available. For nonlinear soil response,
Figure 12. Push-over load application in force or displacement pressure independent (Mises or J2) plasticity and pressure
modes. dependent (Drucker-Prager cone yield surface) models are
available (Elgamal et al. 2003, Yang et al. 2003). The
available solid-fluid coupled formulation allows for
conducting liquefaction-type analyses (Yang et al. 2003).
Selection may be made from a set of available soil model
properties, or by user-defined input modeling parameters
(Figure 16).

Figure 13. Push-over for lateral ground displacement studies.

Figure 14. User-uploaded earthquake base excitation.

5 SOIL MESH BOUNDARY CONDITIONS

For static loading on the pile or footing system, a fixed boundary

condition may be specified along the base and lateral boundaries
of the soil mesh. For dynamic/earthquake excitation, ground Figure 16. Menu of soil properties, and user-defined parameters.
motion is specified uniformly along the soil model base as
mentioned earlier. Along the lateral boundaries, users can choose 7 BEAM-COLUMN ELEMENTS
between fixed, shear beam, or periodic boundary conditions
(Figure 15). OpenSeesPL employs state-of-the-are beam-column
element formulations through the FE analysis engine
OpenSees (Mazzoni et. al 2006). In addition to static
analysis, these elements allow for dynamic/cyclic earthquake-
type simulations. Linear, bilinear hysteretic, and nonlinear fiber
element formulations are available (Mazzoni et al. 2006), based
on steel and concrete cyclic constitutive models. Using
OpenSeesPL, the beam column modeling properties may be
specified, and a display of the resulting moment-curvature
relationship can be generated as shown in Figure 17.

8. VISCOUS DAMPING

For dynamic computations, viscous damping at the level of the

entire model may be specified conveniently. A dedicated
interface allows users to define damping ratios at two different
frequencies, according to the Rayleigh mass-stiffness damping
logic. Conversely, the mass and stiffness matrix viscous damping
multipliers may be specified directly (Figure 18).

9. POST-PROCESSING

Upon specification of the model parameters, the interface

accesses the FE OpenSees platform to conduct the computations. Figure 18. Control over Rayleigh mass, stiffness viscous damping.
If needed, own weight is applied first (soil domain followed by
super-structure), nonlinear material properties are activated, and Upon completion of the computational phase, display
the specified loading scenario is finally executed (static or of the results is initiated by OpenSeesPL. The structure
dynamic/earthquake loading). response may be viewed as time histories and/or as
response at various levels of the applied static load
(Figures 19 and 20). The deformed mesh may be also
viewed (Figures 2, 11), with capabilities for animation and
display of conditions after application of own weight only,
and after execution of the static/dynamic load
computations. Contour quantities such as displacement,
strain, stress, pore pressure, and stress-ratio (stress-state
relative to failure condition) may be viewed (Figure 13).

Figure 19. Display of response time histories.

Figure 17. Fiber section and moment-curvature relationship. Figure 20. Pile displacement load-step display (monotonic load).
10. EXAMPLE SIMULATION SCENARIOS

I. Elgamal and Lu (2009a) conducted a pilot study of lateral

loading on a 3x3 pile group. A single-pile FE model was first
calibrated in the linear range based on the 3D analytical solution
of Abedzadeh and Pak (2004). Response of this linear pile in an
idealized nonlinear undrained-clay material was then computed
and compared to the linear solution. The corresponding 3x3 pile
group response was also addressed, as a function of pile-spacing
for the above linear and nonlinear soil cases (Figures 21 and 22).

Figure 23. Ground modification study for mitigation of liquefaction-

induced lateral deformation (above showing pattern of stone column
construction, and periodic boundary logic, and below showing plan
and side-views of FE mesh (1/2 mesh due to symmetry).

11. SUMMARY AND CONCLUSIONS

Figure 21. FE mesh of 3x3 pile group (1/2 mesh due to symmetry).
A robust and versatile framework for computational
analysis of pile-ground systems was presented. The open-
II. In a remediated area of large spatial extent (Figure 23), the
source platform OpenSees is employed throughout. For
periodic boundary technique offers an effective approach for
illustration, scenarios of lateral response of pile groups, as
conducting 3D analyses (i.e., symmetry allows the investigation
well as ground remediation against liquefaction-induced
of a representative remediated “cell”). On this basis, Elgamal et
lateral spreading were discussed. The conducted
al. (2009b) conducted a 3D FE ground modification parametric
investigations aim to highlight the analysis framework
study, to evaluate mitigation of liquefaction-induced lateral soil
capabilities and range of potential applications.
deformation by the stone column and the pile pinning
approaches. An effective-stress plasticity-based formulation was
ACKNOWLEDGMENT
employed. Using OpenSeesPL, a half-mesh was studied due to
symmetry (Figure 23). A 10 m depth mildly-inclined (4 degrees) The authors are grateful for the funding provided by the
saturated layer was analyzed, with the remediated zone diameter Pacific Earthquake Engineering Research (PEER) Center,
maintained at 0.6 m throughout. Liquefaction-induced lateral the US National Science foundation Grant No. 0529995,
deformation and remediation procedures for mildly sloping sand and the PEER Lifelines program.
and silt strata were investigated under the action of an applied
earthquake excitation. The extent of deployed remediation (area REFERENCES
replacement ratio) and effect of the installed stone column
permeability were analyzed. Effect of lateral spreading on the Abedzadeh, F., and Pak, Y.S. 2004. “Continuum Mechanics of
pile response was also investigated. Lateral Soil–Pile Interaction.” Journal of Engineering
Mechanics, Vol. 130 (11): 1309-1318.
Elgamal, A., Yang, Z., Parra, E., and Ragheb, A. 2003. Modeling
of cyclic mobility in saturated cohesionless soils.
International Journal of Plasticity, 19(6), 883-905.
Elgamal, A., and Lu, J. 2009a. A Framework for 3D finite element
analysis of lateral pile system response, Proceedings of the
2009 International Foundation Congress and Equipment
Expo, Contemporary Topics in In Situ Testing, Analysis, and
Reliability of Foundations, ASCE GSP 186, M. Iskander, D.
F. Laefer, and M. H. Hussein, Editors, Orlando, Florida,
March 15–19, pp. 616-623.
Elgamal, A., Jinchi Lu, J., and Forcellini, D. 2009b. Mitigation of
liquefaction-induced lateral deformation in a sloping stratum:
3D numerical simulation, Journal of geotechnical and
geoenvironmental engineering, ASCE, to appear.
Lu, J., Yang, Z., and Elgamal, A. 2006. Openseespl three-
dimensional lateral pile-ground interaction version 1.00 user's
manual, Report No. SSRP-06/03, Department of Structural
Engineering, University of California, San Diego, La Jolla,
CA.
Mazzoni, S., McKenna, F., and Fenves, G. L. 2006. Open system
for earthquake engineering simulation user manual, Pacific
Earthquake Engineering Research Center, University of
California, Berkeley (http://opensees.berkeley.edu/).
Yang, Z., Elgamal, A., and Parra, E. 2003. A computational model
Figure 22. Plan view of displacement around piles for 5 (above) and for cyclic mobility and associated shear deformation, Journal
7 (below) pile-diameter spacing (1/2 mesh configuration, with red of Geotechnical and Geoenvironmental Engineering, 129
color denoting the large displacement zones). (12), 1119-1127.