Report No. SAC 95-05
| Link Here
for the Reference Library and Downloadable Documents
Parametric Analytical Investigations of Ground Motion and Structural
Response, Northridge Earthquake of January 17, 1994, by S. Campbell,
M. Englehardt, J. Hall, G. Hart, L. Ho, S. Huang, A. Husain, W. Iwan, H.
Kim, K. Kim, R. Lobo, T. Sabol, M. Skokan, and J. Uzarski, December, 1995.
This volume includes the technical reports documenting a number of parametric
analytical investigations related to ground motions and the structural
performance of welded steel moment frame construction in the January 17,
1994 Northridge Earthquake. These studies comprise Tasks 3.4 and 3.5 of
Phase 1 of the SAC Steel Project. The goal of subtask 3.4 was to develop
a joint fracture modeling element for incorporation in an available nonlinear
analysis program, that could be used in individual building analyses performed
in subtask 3.1 (See SAC 95-04
). The contents of this volume and SAC
95-04 comprise the majority of the analytical investigations performed
in Phase I of the SAC Steel Project.
A solicitation for proposals to perform parametric investigations of
steel moment frame building performance (subtask 3.5) was circulated during
December, 1994. Over 20 proposals were received for this task. A total
of five Subcontractors were selected. These investigations included the
The work done in subtask 3.4 by Campbell rose out of the need to analytically
investigate the effect of damaged connections on the behavior of steel
moment frame structures. None of the nonlinear analytical packages available
for use by the investigators undertaking the detailed building analyses
in subtask 3.1 included an element which simulated the fracture and post-fracture
behavior of these connections. Modified beam-column elements were produced
to model these effects within the DRAIN series of programs. Versions were
produced for both two and three dimensional analyses. Example problems
using the new elements were also developed and made available.
A large number of inelastic building analyses on two model buildings based
on extensive inelastic analytical models for both beam-column and column
splice connections, varying assumptions for connection capacities, and
a suite of ground motion time histories that included large, thrust fault
The development of drift demand spectra for important strong ground motion
records of the Northridge earthquake and other ground motions, and the
implications of these spectra on the design and evaluations of existing
The development of simplified nonlinear models for use in the development
of nonlinear response spectra. This investigation was done in conjunction
with one of the building analyses in task 3.I. (Note that the report for
this study is included in SAC
95-04, in the paper by Krawinkler et. al.).
Analyses to study the effects of vertical ground accelerations on the response
of steel moment resisting frame buildings. This investigation was done
in conjunction with one of the building analyses in task 3.1.
Parametric studies on the sensitivity of expected building and connection
performance due to variations in modeling assumptions and procedures.
A parametric study by Hall focused on the effects of near fault motions
from moderate size (Magnitude 7) earthquakes on moment resisting steel
frame buildings. Analytical models of two buildings, one six stories and
the other twenty stories, were subjected to a series of ground motions,
including the Olive View Hospital free-field record in the Northridge earthquake,
and three earthquake simulations that included blind thrust earthquakes
in the Magnitude 7 range. The building models varied parameters such as
material yielding, weld fracture, composite floor slabs, accumulation of
damage from a second earthquake, and vertical ground motion. The basic
results of the study were as follows:
lwan processed ground motion data for selected near field sites where high
peak velocities were observed during the Northridge earthquake. Drift demand
spectra were derived from this data, and also from synthetic Northridge
ground motions. The results of these analyses are then compared with those
on actual and prototype buildings to evaluate the efficacy of using the
drift demand approach as a simple evaluation tool. The drift demand spectra
developed indicate demands in the Northridge Earthquake in the range of
two per cent, which appears to be consistent with more detailed analyses,
and appears to be consistent with the level of damaged experienced by a
number of these buildings.
Large building deformations were predicted for many of the ground motions,
especially for the six story building, even when all welds were assumed
to be "perfect".
The near fault ground motions appear to be well above the building code
spectrum levels. Such effects should be considered in design.
Including weld fracture into the model increased the building displacements,
and the potential for excessive damage and collapse. In several of these
analyses, the buildings collapsed.
Including the effects of the floor slabs increased the amount of column
yielding, but was not felt to improve the system behavior significantly.
A second large earthquake can lead to increased levels of damage, especially
if many fractured welds occurred during the first event.
Vertical ground motions appeared to be of minor importance.
Inelastic building response prediction is sensitive to the many modeling
assumptions that are made. Careful analysis and interpretation of the results
is therefore required.
Hart, et. al. performed a series of analyses to investigate the response
of welded steel moment resisting frames subjected to both horizontal and
vertical ground motion. A portion of a building that sustained damage during
the Northridge Earthquake was modeled for use in these analyses. The sensitivity
of this model to varying lateral and vertical mass was investigated using
a suite of nine synthetic time histories in a response spectrum approach.
This study indicated that the effects of vertical ground motion are likely
more significant for interior columns than exterior columns. The effect
on beam demands appeared to be insignificant for the models studied. The
effects of varying horizontal ground motions indicated a very direct impact
on element internal forces for all members, except perhaps interior columns.
The last report addresses a brief study examining the effects of various
modeling assumptions on the predicted response of a steel moment resisting
frame. A variety of modeling assumptions were used in a series of analyses
of a six story single bay steel moment frame. A "baseline" model with bilinear
representation of inelastic response and bare steel frame behavior and
a "refined" model were developed. The "refined" model included hardening
rules that better represent experimentally observed inelastic behavior,
and included composite floor slab effects. These models were subjected
to both a static pushover and a series of three strong ground motion records.
The results indicated the following:
As a quality assurance measure, all SAC Steel Project investigations are
overseen by a Technical Advisory Panel (TAP). The panel for the Task 3
investigations were experienced in the fields of elastic and inelastic
structural analysis, modeling techniques and structural steel system behavior.
The predicted response of these frames can be quite sensitive to modeling
assumptions. Predicted levels of beam plastic rotations varied up to 100%
between models in this study.
The presence of a composite floor slab can significantly affect the predicted
response of a steel moment frame, Neglecting these effects can substantially
underestimate both the frame strength and stiffness, and can alter the
expected distribution of plastic rotation demands.
The assumed level of yield stress in steel members, and especially the
relative value between the beams and columns can have a significant affect
on the predicted response.
The predicted location of yielding within a beam-column connection (joint
panel zone, beam flanges, column flanges, etc.), can be shifted by varying
the modeling assumptions.
Local response prediction indices (e.g., beam plastic rotation) are more
sensitive to variations in modeling assumptions than global indices such
as interstory drift. This confirms that large uncertainties will exist
in the prediction of plastic rotation demands, and the need to recognize
these uncertainties in the course of steel moment frame building designs