You are on page 1of 12

Structures Congress 2015

435

A Simplified Method for Modeling Soil-Structure Interaction


for Rigid Frame Structures
Hari Aamidala, PE, M.ASCE1; and John Kim, Ph.D., PE2
1

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

Project Manager and Lead Structural Engineer, Parsons Brinckerhoff, Inc., 465
Springpark Place, Herndon, VA 20170. E-mail: aamidala@pbworld.com
2
Senior Professional Associate and Supervising Bridge Engineer, Parsons
Brinckerhoff, Inc., 530 East Main Street, Suite 701, Richmond, VA 23219. E-mail:
KimJo@pbworld.com
Abstract
A structure can be modeled using a multitude of methods, ranging from simple
mathematical models with assumptions based on engineering judgment, to complex
3D models that attempt to capture every single variable that affects the behavior of
the system. This paper, utilizing analysis of Rigid Frame Structures, explores
different methods of modeling structures in an effort to understand the benefits, or
lack thereof, of introducing more variables and complexity into a model. The focus is
on a simplified method for the modeling of soil-structure interaction and analyzing a
rigid frame structure. Through this effort it was found that, rather than use an
expensive 3D modeling software, commonly available structural analysis software
like LARSA can be used to create a simple soil-structure interaction model. This
simple method allowed us to quickly create numerous models to efficiently analyze
different alternatives.
INTRODUCTION
In developing models, an engineers primary goal is to understand the behavior of a
simple or complex structure, as best as possible. A model is designed to represent the
behavior of a system by using a set of variables and a set of mathematical equations
that establish relationships between those variables. Ultimately, models are used to
help us study the effects of different variables and make better decisions. This study
was primarily focused on evaluating different methods to model soil-structure
interaction for rigid frame structures.
Rigid Frame type construction has been utilized for cut-and-cover tunnels and bridges
for more than 100 years (PCA 1936). Some of these structures have utilized arch
concept while others are simply designed as frames. This concept was originally
developed as a cost effective and attractive alternative to conventional arch bridges;
however, it is extensively used today in top down construction.

ASCE
Structures Congress 2015

Structures Congress 2015

436

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

Prior to computer software, numerous manual methods like column analogy (Cross
1930) and moment distribution methods (Timoshenko et al 1965) were utilized to
model rigid frame structures, which involved complex time-consuming analyses. In
recent years this process has been replaced with computer analysis that involves 3D
modeling, using software specifically intended for soil-structure interaction. Though
3D modeling using these complex software for soil-structure interaction could yield
good results, it can be clumsy to work with and could inadvertently introduce errors
into the analysis due to the use of large number of variables with inherent
assumptions, and it could also be cost prohibitive.
In an effort to simplify the modeling and to improve the performance of the analysis,
a simplified structural model was created, which would only use the basic structural
elements, and grounded linear and non-linear spring elements to model soil-structure
interaction. To evaluate this simple method and to compare with other modeling
methods, a couple of case studies are being utilized Load Rating of Rigid Frame
Bridges and Design of a Rigid Frame Rail Structure using top-down construction.
MODELS
A model is a physical or mathematical representation of a real world system. A
mathematical model is designed to represent the behavior of a system by using a set
of variables and a set of mathematical equations that establish relationships between
those variables. A physical model can especially be useful in validating a
mathematical model when Structural Engineers are trying to understand the behavior
of a real world structure (Gomez-Rivas et al 2012). However, models cannot be
expected to represent the behavior of a real world system with 100% accuracy. A
good model is the one that provides enough accuracy to make sound decisions. A
model cannot ever be totally complete; therefore, there is no need to refine the model
unless the quality of the decisions made is affected by the inaccuracy of the model.
When credible observations conflict with the predictions made by the model, either
the data/assumptions are incorrect or the model itself is incapable of providing
enough accuracy. Understanding what needs refinement is important to improve the
accuracy of the results.
As an example, Figure 1 shows a mathematical model that is represented by an
equation, which helps predict compressive strength of a particular concrete at a given
point of time, based on past concrete compressive strength test data. If the model
represented by the equation in Figure 1 cannot accurately predict the compressive
strength of this particular concrete, it could either be because the concrete test data
was incorrect or the mathematical equation used to best fit the data was incorrect. So
in this particular situation rather than only focus on the equation, the modeler should
also look at the data and the strength tests used to collect that data, to improve the
accuracy of the model.

ASCE
Structures Congress 2015

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

Structures Congress 2015

437

Figu
ure 1. Comp
pressive Streength of a particular
p
Concrete Mixx
ODEL COM
MPLEXITY
Y
MO
mportant to
As mentioned earlier, no model is ttotally compplete; howevver, it is im
mplify the model
m
to onlly include whats
w
needeed to make good prediictions at a
sim
dessired level off accuracy.
In an
a effort to understand
u
w
what
model complexity means, we will
w look at Columbus
voyyage to Indiia (Stern 2007). When Columbus proposed
p
to reach Indiaa by sailing
wesst from Spaiin, he knew
w that Earth was round. Experts turnned him dow
wn because
bassed on Erato
osthenes esttimate, from
m more than 2000 yearss ago, distannce to India
wass far too greeat to accom
mplish. Colum
mbus estim
mate of the earths
e
radiuus, based on
incoorrect units of distance, showed thatt India was lot closer thaan the distancce based on
Eraatosthenes estimate.
e
Coolumbus connvinced Queeen Isabella to overrule the experts
andd Columbus sets sail to India.
I
If the American coontinent hass not existed, Columbus
witth his tiny ships could never
n
have crossed an ocean as wiide as the Atlantic
A
and
Paccific combin
ned. His spheerical modell of the eartth was correect but the data
d
he used
wass based on incorrect unnits. Refininng the modeel would nott yield accuurate results
wheen the data used
u
in the model
m
was incorrect.
i
It wouldnt haave helped iff the model
com
mplexity waas increased by refining it with the effect of roough sea andd mountain
ranges on the earth surface,, when the innaccuracy sttems from thhe data used.
In order
o
to maake good decisions baseed on structuural models, it is imporrtant for an
enggineer to un
nderstand alll the variabbles used annd assumptioons made with
w
all the
varriables in thee model. Thiss is only posssible by redducing the coomplexity off the model.

ASCE
Structures Congress 2015

Structures Congress 2015

438

RIGID FRAME STRUCTURES

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

A rigid frame concrete bridge is a bridge structure where the deck is integrally or
rigidly connected to the abutments and piers of the structure. It is typically simpler
and more economical to build a rigid frame structure for top-down construction. One
of the direct benefits from the continuity or rigidity is that the moments are small in
the sections near the center of the deck of the rigid frame structure compared with
corresponding moments in a simply supported deck of the same length (PCA 1936).
The other significant benefit is reduced maintenance cost by eliminating bearings and
joints at the abutments compared to a conventional beam and deck type bridge.
Therefore, reducing the life cycle costs significantly. Also Rigid Frame Concrete
Bridges may be widened with little alterations to the existing bridge, without
interfering with normal traffic.
Figure 2 shows a 2-span continuous Concrete Rigid Frame Bridge for I-395 SB over
George Washington Memorial Parkway. This type of construction was typical for
bridges over George Washington Memorial Parkway, which was designed for
recreational driving built in 1950s thru 1960s, in the Virginia suburb of DC. All Rigid
Frame Bridges on GW Pkwy utilize the arch concept resulting in economical and
aesthetically pleasing bridge structures.

Figure 2. 2-Span continuous Rigid Frame Bridge, I-395 SB over GW Pkwy

ASCE
Structures Congress 2015

Structures Congress 2015

439

MANUAL ANALYSIS METHODS


Before the advent of computers, all structural analysis was performed manually by
solving equations developed using calculus or using graphical methods developed to
simplify these analyses. Some of the methods used to manually analyze are discussed
below.

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

Column Analogy Method


This was a method proposed by Hardy Cross where bending moments in simple
frames may be computed by a procedure analogous to the computation of fiber
stresses in short columns subject to bending, and that slopes and deflections in these
structures may be computed as shears and bending moments, respectively, on
longitudinal sections through such columns (Cross 1930). Professor Cross analyzed
24 frames by this method and presented a chart, shown in Figure 3, which could be
used for estimates and preliminary design of typical rigid frame structures that were
built during that time (PCA 1936).

Figure 3. Chart for Column Analogy Method (PCA 1936)

Table 1. Crown Moments created by uniformly distributed load (PCA 1936)

ASCE
Structures Congress 2015

Structures Congress 2015

440

Moment-Distribution Method
This method first introduced by Hardy Cross was widely used for the analysis of
statically indeterminate frame structures (Patil 2013). Typically analysis of frames
required solving a system of linear algebraic equations, which can be readily solved.
However, in many cases, it required solving more complicated system of equations,
and their solution was cumbersome. Moment-Distribution Method was one of the
methods where successive approximations were used to analyze these structures.
Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

In this iterative method, all joints are first assumed to be restrained against rotation so
that fixed-end moments are calculated. The joints are then successively released. The
unbalanced moment, which is the algebraic sum of all the fixed-end moments at a
joint is calculated, this represents the total turning effort exerted on the joint
(Timoshenko 1965). Releasing a joint is equivalent to distributing the unbalanced
moment among the members meeting at the joint based on the relative stiffness of the
members, using a factor called distribution factor. The final end moments at the joint
are then obtained by superimposing the distributed moments on the previously
calculated fixed-end moments. One half of the distributed moment is carried over to
the far end of the member. The released joint is again restrained before moving to the
next joint. The same sets of operations are carried out at each joint till all the joints
are completed. This completes one cycle of operations. The process is repeated for a
number of cycles till the values obtained are within the desired accuracy (Patil 2013).
CASE STUDY 1: LOAD RATING OF RIGID FRAME BRIDGES
As part of a Maintenance and Rehabilitation effort, I-395 Bridges over George
Washington Memorial Parkway were evaluated for structural capacity. These bridges
showed cracks, delamination and significant efflorescence on the underside of the
deck, which possibly stemmed from drainage issues on the top of the deck.
I-395 NB Bridge over George Washington Memorial Parkway was concrete rigid
frame structure that was composed of a two single span rigid frames with a
continuous deck, which was built in 1950. It utilized an arch concept with a thin deck
at the crown. I-395 SB Bridge over George Washington Memorial Parkway was a
two span continuous concrete rigid frame structure, which was built in 1960. It had a
similar thin deck at the crown utilizing an arch concept. See Figure 2 for the picture
of the I-395 SB Bridge.
To load rate these structures, we utilized LARSA for modeling the structure. The
tapered leg elements of the frame were modeled as single members while the deck of
the frame was divided into several tapered members to model the parabolic underside
of the deck, thereby allowing us to model the arch effect of the parabolic underside.
Also as part of this effort, several different manual analysis methods were looked at,
which were possibly used during the original construction of these structures.

ASCE
Structures Congress 2015

Structures Congress 2015

441

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

In an effort to compare the LARSA structural model with Column Analogy Chart, a
simple 1 kip/ft. uniformly distributed load was applied on a single span rigid frame
LARSA model, shown in Figure 4, and the results were compared with the results
obtained from the Chart, shown in Figure 3. Moments at the crown and the knees,
comparing Column Analogy Chart vs. LARSA model, are shown in Table 2.

Figure 4. Model for comparison of Column Analogy Chart vs. LARSA model

Mcrown

Chart
3.81 k-ft

LARSA
3.68 k-ft

Mcorner

-10.87 k-ft

-10.65 k-ft

Table 2. Results of Comparison of Column Analogy Chart vs. LARSA model


It is important to note that in our load rating model, the effect of horizontal earth
pressure was ignored for these structures. In comparison with the effect of the other
loads, earth pressure moments are generally of little importance in stress
determination (PCA 1936). At the crown, both vertical dead and live loads create
tension at the bottom of the deck, while the earth pressure creates tension at the top of
the deck; it is therefore conservative to ignore the earth pressure at the crown. The
corner moment due to earth pressure, for this type of concrete rigid frame structure,
would be negligible compared to the moment due to dead plus live loads therefore
can be ignored. However, it is important to note that soil stiffness definitely affects
the fixity at the foundation. See Figure 5 for the illustration of foundation fixity.

Figure 5. Foundation Fixity, which is to be assumed based on soil stiffness


If the soil behind the abutments is too stiff then regardless of the foundation type,
supports will need to be modeled as fixed. On the other hand if the soil is too soft
then supports will need to be modeled as hinged. Actually, the supports are rarely
ASCE
Structures Congress 2015

Structures Congress 2015

442

hinged or fixed, but the foundation conditions lie somewhere in the range between
these two extremes (PCA 1936). Therefore, it is important to model it correctly.
CASE STUDY 2: DESIGN OF A RAIL STRUCTURE RIGID FRAME

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

As part of Dulles Corridor Metrorail Project Phase 2, WMATA Metrorail System is


being extended through a busy Dulles Toll Road (DTR) Corridor. Based on an early
2012 study by Fairfax County Department of Transportation, extending Town Center
Parkway across DTR Corridor was identified as a crucial connection between Reston
downtown on the north of DTR, and businesses on the south. See Figure 6 for the
aerial view of the project location. After evaluating several overpass and underpass
alternatives, an underpass alternative was identified as the most feasible solution.
However, this connection was only in the Countys long-term plan. WMATA was
concerned that once the Metrorail was built through the DTR corridor, construction of
a future underpass would lead to severe service disruptions.
To minimize future service disruptions, through a multi-agency and consultant effort,
it was decided that a rail structure should be constructed as part of the Dulles
Corridor Metrorail Phase 2 Project that could allow for a future underpass with
minimal service disruptions to the Metrorail. This was only possible with a top-down
construction concept that could be achieved by a 2-Span Rigid Frame Concrete
Structure that is composed of Secant Pile Wall Abutments and Piers, and Reinforced
Concrete Slab, cast integral with the walls. This would have an added advantage of
eliminating bearings, and abutment and pier joints, which would significantly reduce
the future life cycle costs. Also it was important to build a structure that would not
require bridge inspections in the interim condition before the underpass is
constructed.
Top-Down Construction
To utilize top-down construction concept for this project, Secant Pile Abutments and
Pier, from the existing ground, will be constructed first. Therefore, the support of
excavation walls will serve as permanent substructure units, in the future. Next the
roof slab is constructed as a slab-on-grade and is rigidly connected to the Secant Pile
walls. Secant Pile Walls were preferred over Tangent Pile Walls because the superior
performance of the Secant Pile Walls was expected in this shallow ground water
condition. Once the Roof Slab is constructed, the Metrorail tracks will be placed on
the top. Until the future underpass needs to be constructed, this structure will be
hidden from plain sight. See Figure 7 for the illustration of Rail Structure elevation
prior to construction of the underpass. When we are ready to construct the underpass,
in the future, bridge structures will need to be constructed for Dulles Toll Road and
Airport Access Road, and the underpass can be extended by excavating under the roof
slab of the Rail Structure. Apart from any architectural treatment to the wall face,
Secant Pile Walls, without any additional work, should be capable of serving as
Bridge Substructures. See Figure 8 for the illustration of Rail Structure elevation with
the underpass.

ASCE
Structures Congress 2015

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

Structures Congress 2015

443

Figurre 6. Aerial view


v
of the location of the Proposeed Underpaass

Figure 7. Rail Struccture Elevattion Prior to Underpaass Constru


uction

Figure 8. Rail
R Structurre Elevation
n with Underpass
ASCE
Structures Congress 2015

Structures Congress 2015

444

Soil-Structure Interaction Modeling


One of the key concerns with the analysis of this structure was that we needed to
model the Structure and the Soil-Structure Interaction using a method that would
allow us to quickly modify the span lengths, under-clearance, different soil
parameters, etc.

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

After evaluating different modeling software and different methods of modeling soilstructure interaction, it was decided that a complex finite element model, with
Structure and Interface elements, and hardening soil model, would be too clumsy to
allow us to evaluate different scenarios for this project. See Figure 9 for an
illustration of a possible 3D finite element soil model for a similar structure. It is
important to note that though finite element software offer flexibility to create
complex soil models, they would create a model with too many elements based on too
many assumptions, creating an unwieldy model.

Figure 9. Illustration of a 3D Finite Element Soil Model


Creating a simple model that can provide the desired level of accuracy is really
important to any Structural Engineer. With that in mind a LARSA model was created
using simple structural elements, and linear and non-linear grounded spring elements
to model soil stiffness (Kim 2009). This significantly reduced the number of variables
in the model, thereby providing the Structural Engineer with good control of the
analysis without sacrificing the quality of the results. The model was simple and
allowed us to pre and post process data using excel spreadsheets. See Figure 10 for
the LARSA model created for this project.

Figure 10. Illustration of the LARSA model

ASCE
Structures Congress 2015

Structures Congress 2015

445

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

Soil Stiffness for the grounded spring elements can be calculated using modulus of
subgrade reaction for the different soil material present at different elevations.
Modulus of subgrade reaction can be calculated using triaxial compression tests on
compacted soil samples, plate load tests, or estimated from Standard Penetration Test
blow counts, based on the soil type. For a structure width of 7 feet, if the grounded
spring elements are placed at every 1 foot vertically, for a modulus of subgrade
reaction of 75pci, the Spring Stiffness (K) can be calculated as shown in the equation
below.

Grounded spring element is defined using a single joint, in LARSA, with no length
associated with the element. The element represents a spring connecting a joint in the
structure to the ground at its undeformed location. Linear Grounded Spring Elements,
with same stiffness in tension and compression, are used to model the foundation.
Non-Linear Grounded Spring Elements, when soil has stiffness in compression but
not in tension or vice versa, are used to model the soil fill behind the abutments.

Figure 11. Illustration of Spring Elements in the LARSA model


CONCLUSIONS
The key takeaway from this effort is that, no model is totally complete, it is important
to simplify the model to only include whats needed to make sound decisions. It is
important for an engineer to understand all the variables used and assumptions made
with all the variables in the model. Also based on the level of accuracy that can
reasonably be achieved, Structural Engineers should consider increasing the safety
factors, beyond code requirements, to account for the model uncertainty (Alvi, 2013).
The use of linear and non-linear grounded spring elements is an efficient method to
model soil stiffness, even when there is a significant variation in the soil profile,
compared to creating an unwieldy model built using hardening soil model and
interface elements. This simplified approach can allow engineers to make quick
changes to the model and evaluate different alternatives with relative ease.

ASCE
Structures Congress 2015

Structures Congress 2015

446

REFERENCES

Downloaded from ascelibrary.org by David Bonilla on 01/20/16. Copyright ASCE. For personal use only; all rights reserved.

Alvi, I. (2013). Engineers Need to Get Real, But Cant: The Role of Models.
Proceedings of 2013 ASCE Structures Congress, Pittsburg, PA.
Cross, H. (1930). The Column Analogy, University of Illinois, Urbana, IL.
Gomez-Rivas, A., Pincus, G., and Tito, J.A. (2012) Models, Computers and
Structural Analysis Proceedings of 10th Latin American and Caribbean
Conference for Engineering and Technology, Panama City, Panama.
Kim, J.S., Garro, R., and Doty, D.A. (2009). A simplified Load Rating Method for
Masonry and Reinforced Concrete Arch Bridges. Proceedings of AREMA
Annual Conference, Chicago, IL, September, 2009.
PCA (1936). Analysis of Rigid Frame Concrete Bridges without Higher Mathematics,
4th Ed., Portland Cement Association, Chicago, IL.
Patil, P.R., Pidurkar, M.D., and Mohankar, R.H. (2013) Comparative Study of End
Moments Regarding Application of Rotation Contribution Method (Kanis
Method) & Moment Distribution Method for the Analysis of Portal Frame.
IOSR Journal of Mechanical and Civil Engineering, 7(1), 20-25.
Stern, D.P. (2007). From Stargazers to Starships. NASA, Greenbelt, Maryland
http://www-istp.gsfc.nasa.gov/stargaze/Scolumb.htm (Jan. 10, 2015)
Timoshenko, S.P., and Young, D.H. (1965). Theory of Structures, 2nd Ed., McGrawHill, New York.

ASCE
Structures Congress 2015

You might also like