You are on page 1of 23

Consolazio, Lehr, McVay 1

DYNAMIC FINITE ELEMENT ANALYSIS OF


VESSEL-PIER-SOIL INTERACTION
DURING BARGE IMPACT EVENTS

By Gary R. Consolazio, Ph.D. 1, G. Benjamin Lehr 2, Michael C. McVay, Ph.D. 3

1 (Corresponding author) Assistant Professor, University of Florida, Department of Civil &


Coastal Engineering, P.O. 116580, Gainesville, FL, 32611 (Tel: 352-846-2220, Fax: 352-392-
3394, E-mail: grc@ce.ufl.edu)

2 Graduate Research Assistant, University of Florida, Department of Civil & Coastal


Engineering, P.O. 116580, Gainesville, FL, 32611 (Tel: 352-846-3431, Fax: 352-392-3394,
E-mail: blehr@ce.ufl.edu)

3 Professor, University of Florida, Department of Civil & Coastal Engineering, P.O. 116580,
Gainesville, FL, 32611 (Tel: 352- 392-8697, Fax: 352-392-3394, E-mail: mcm@ce.ufl.edu)

SUBMITTED: NOVEMBER 15, 2002

WORD COUNT : 5000 WORDS, 1 TABLE, 9 FIGURES

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 2

ABSTRACT

Current specifications for highway bridge design provide empirical relationships for computing lateral
impact loads generated during barge collisions, however, these relationships are based on limited
experimental data. In order to better understand and characterize such loads, dynamic finite element
analysis techniques have been employed to simulate vessel impact conditions never before tested
experimentally. Descriptions of the methods used to model a hopper barge, bridge piers, and soil-structure
interaction are given. Impact simulation results, including time histories of impact loads and barge
deformations, are presented and compared to data generated using current bridge design specifications.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 3

INTRODUCTION

Designing a bridge structure that spans a navigable waterway requires careful consideration be
given to the fact that vessels such as ships or barges may inadvertently collide with the piers that
support the bridge superstructure. Worldwide, vessel impacts occur frequently enough that, on
average, at least one serious collision occurs each year [1]. Several recent incidents involving
barges impacting bridge structures clearly demonstrate that the potential for catastrophic
structural failure and loss of life exist.
In September 1993, a multi-barge tow navigating at night and in a dense fog collided with
the Big Bayou Canot railroad bridge [2] near Mobile, Alabama (USA) resulting in a significant
lateral displacement of the structure. Moments later, a passenger train attempted to cross the
structure at 113 km/hr (70 mi/hr), resulting in catastrophic structural failure and forty-seven
fatalities. In September 2001 a barge tow navigating near South Padre, Texas (USA) veered off
course in strong currents and collided with piers supporting the Queen Isabella Causeway bridge.
As a result of this impact, three spans of the structure collapsed and several people lost their
lives. Most recently, in May 2002, an errant barge tow struck a bridge on interstate I-40 near
Webbers Falls, Oklahoma (USA). On an average day, the bridge carried approximately 20,000
vehicles across the Arkansas River. As a result of the barge impact, 177 m (580 ft.) of
superstructure collapsed, fourteen people died, and traffic had to be rerouted for approximately
two months.
Clearly, the lateral loads imparted to bridge structures during barge impact events have
the potential to cause significant loss of life as well as economic loss. Despite this fact, barge
impact loading is one of the less well quantified sources of load for which bridges must be
designed. The AASHTO vessel impact design guidelines and specifications [3, 4]hereafter
referred to simply as the AASHTO provisionsprovide bridge designers with a standard
methodology by which ship and barge impact loads may be determined for design purposes.
Using the mass and speed of an impacting barge, the vessels initial kinetic energy prior to
impact is computed. The kinetic energy is then used in an empirical relationship to predict the
depth of damage (crush depth) that would be sustained at the bow of the barge if the impact
condition in question actually occurred. In order to then compute the impact load imparted to a
bridge, additional empirical equations are employed that relate barge crush depth to impact load
magnitude.
Unfortunately, very little experimental data has ever been collected for the purpose of
quantifying the lateral loads and barge deformation that arise during impact events. AASHTOs
empirical relationships between energy, deformation, and load are based on a German study
conducted by Meir-Dornberg [5] that involved physical testing of partial sections of reduced-
scale standard European (Type IIa) barges. Using experimental data collected during the study,
relationships were developed between kinetic energy, barge crush deformation, and structural
load. AASHTOs barge impact expressions are essentially identical to those of Meir-Dornberg
except that minor modifications have been made to reflect differences in size between the
European barge and the jumbo hopper barge most often found operating in U.S. inland
waterways.
Despite the fact that barge impacts are dynamic events in which the magnitude of load
imparted to the bridge varies with time, the AASHTO provisions make use of static loads. These
static equivalents are intended to produce approximately the same structural response in the
bridge as would result from the application of true time-varying dynamic impact loads. In

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 4

general, however, due to the limited experimental data available the relationships between barge
kinetic energy, barge deformation, and lateral loads imparted to bridge structures have been
given only limited attention and are not well understood.
In this paper, nonlinear finite element impact simulation is used as an alternative to
experimental testing to study and characterize the loads generated on bridge piers during barge
impacts events. Finite element models have been developed for a typical hopper barge, for bridge
pier substructures, and for representation of soil-structure interaction. Results for a variety of
different impact scenarios, all simulated using the LS-DYNA [6] impact finite element code, are
presented along with discussions of impact load magnitudes, energy dissipation mechanisms, and
predominant modes of structural response (pseudo-static vs. dynamic).

CASE STUDY

In the United States, the jumbo class hopper barge makes up more than 50% of the entire barge
fleet [3] and is the type of barge most frequently found operating in the U.S. inland waterway
system. In addition, the jumbo hopper barge is the baseline vessel upon which the AASHTO
barge impact provisions are based. For these reasons, the finite element vessel model developed
in this study matched the mass, geometry, and structural configuration of a typical jumbo hopper
barge.
The St. George Island Causeway Bridge (Florida State Road 300) near Apalachicola,
Florida (USA) was chosen as the structure to be modeled in this study. It spans over an active
intracoastal waterway that regularly carries barge traffic, has a variety of different pier and span
types, and is typical of many bridges currently in-service in the U.S. Using structural plans and
soil boring logs for this bridge, finite element models were developed for two of the pier
substructures that support the bridge deck near the navigation channel. Descriptions of these
models are given later in this paper.
An added reason that the St. George Island Bridge was chosen for this study was that
full-scale experimental barge impact tests are scheduled to be conducted on the bridge in the near
future. A new bridge structurepresently under constructionwill be opened to traffic in 2003
that will replace the existing bridge. Prior to demolishing the existing bridge, full scale physical
barge impact tests will be conducted to directly measure the dynamic impact loads and structural
responses that arise during barge impact events. Much of the simulation data presented in this
paper was generated as part of the planning process [7] for the full-scale impact testing program.
While this physical test program will yield very valuable impact data, the time-frame in
which the testing must take place is relatively short and therefore only a limited number of tests
can be conducted. Thus, impact simulationas described hereinis an important tool because a
wide variety of impact conditions can be simulated with minimal additional cost. By combining
results obtained from both analytical simulation and physical testing, substantial improvements
in quantifying barge impact loads will be achieved.

DEVELOPMENT OF A HOPPER BARGE FINITE ELEMENT MODEL

Jumbo class hopper barges measure approximately 59.44 m (195 ft.) in length, 10.67 m
(35 ft.) in width, and 3.66 m (12 ft.) in depth, have a dry (empty) weight of approximately
1.8 MN (200 T) and a cargo capacity of 15.1 MN (1700 T). In order to obtain realistic results
from the impact simulations conducted in this study, the geometry, mass, and bow stiffness of
the barge needed to be accurately represented in the finite element model developed. To ensure

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 5

that a realistic model was constructed, structural plans for a typical jumbo class barge were
obtained from a leading U.S. barge manufacturer and were used in the development of the finite
element model.

Zones of the hopper barge model

All of the simulations conducted in this study involved head on impacts of barges striking bridge
piers in a direction perpendicular to the span of the bridge. Such conditions are typical of most
vessel collision events and can result in large scale deformations in the bow portion of the barge.
In order to accurately represent the stiffness, localized buckling, and extensive plastic
deformation that occur in this region, a high level of modeling detail is needed. However, rear
portions of the barge do not experience significant deformation during such impacts and thus
only need to be modeled such that the correct inertial properties are properly represented. For
computational efficiency, the barge model developed in this study was divided into three
separate zones (Figure 1), each modeled with a different level of detail.
Zone-1, the most detailed portion of the model, represents the first 2.67 m (8.75 ft.) of the
barge bow. Hot-rolled steel channel and angle members form internal stiffening trusses within
the bow, while steel plates of various thicknesses make up the outer hull shell. In order to capture
local deformations, buckling, and internal contact of structural components in this zone of the
barge, discrete shell element modeling was used. Hot rolled structural shapes as well as the outer
hull plates were modeled using shell elements (Figure 2). Steel material behavior was modeled
using nonlinear stress-strain relationships corresponding to steel having a yield stress of
248 MPa (36 kip/in2)marginally higher than the minimum allowable yield stress of 234 MPa
(34 kip/in2) specified by vessel construction rules [8] for inland barges. Contact definitions were
also established for all components of the barge bow to enable detection of internal contacts
during bow crushing.
Zone-2 of the barge model represented the 6.02 m (19.75 ft.) portion of the raked bow
that follows zone-1. Internal trusses extend throughout the entire raked bow. However, the truss
and hull components in zone-2 were not expected to sustain buckling, inelastic deformation, or
experience internal contact during typical impact events. As a result, numerically efficient
resultant beam elements were used to model the stiffness of the internal trusses in this zone. The
outer hull was modeled with shell elements, but with a lower resolution than that used in zone-1.
Zone-3 of the model represents the 50.75 m (166.5 ft.) hopper portion of the barge. No
significant deformations were expected in this zone. However, representing the inertial
properties of the overall barge required that an appropriate distribution of mass be modeled in
this zone, especially when partially or fully loaded payload conditions were being analyzed.
Zone-3 was modeled using a coarse mesh of brick elements whose mass density was chosen to
match the desired payload condition under consideration. Both gravitational effects and
buoyancy effects were also represented in modeling the barge (see Consolazio et al. [7] for
additional details).

DEVELOPMENT OF BRIDGE PIER MODELS

As was stated earlier, the bridge structure chosen for this study was the St. George Island
Causeway bridge near Apalachicola, Florida. For this bridge, all possible barge impact points are
located on the pier columns and direct contact between barges and the bridge superstructure
(deck) is not possible. Focus in this paper has been placed on the prediction and characterization

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 6

of barge impact loads on individual, isolated bridge piers. Thus, superstructure effects such as
load shedding to adjacent piers are not discussed herein.

Finite element models of pier-1 and pier-3

Models of both a channel pier and an approach span pierdenoted pier-1 and pier-3 respectively
based on their proximity to the navigation channelwere developed using structural plans for
the St. George Island bridge. Pier-1 and pier-3 were chosen for this study because they represent
a significant spread in both structural stiffness and impact resistance, and therefore yielded
impact force results representative of a wide variety of pier types.
Pier-1 is typical of what would likely have been considered to be an impact resistant
design at the time the bridge was constructed (1960s). Above the waterline, this reinforced
concrete structure consists of a bent cap, two large square pier columns, and a lateral stiffening
shear wall. The width of each pier column at the point where a barge would make contact during
an impact is 1.83 m (6 ft.) The mud-line footing consists of a single concrete pile cap (6.40 m
(21 ft.) wide, 11.89 m (39 ft.) long, 1.52 m (5 ft.) thick), a concrete tremie seal (7.32 m (24 ft.)
wide, 12.80 m (42 ft.) long, 1.83 m (6 ft.) thick) below the cap, and thirty-six steel HP 14x73
piles.
In contrast, pier-3 supports an approach span of the bridge and is more representative of a
non-impact resistant design. It consists of a reinforced concrete bent cap, two rectangular pier
columns, and a lateral stiffening beam. At the barge impact elevation, each pier column is 1.52 m
(5 ft.) in width. The foundation consists of two separate rectangular concrete pile caps (2.44 m
(8 ft.) wide, 3.05 m (10 ft.) long, 1.22 m (4 ft.) thick), each cast above the waterline and bearing
on four 508 mm (20 in.) by 508 mm (20 in.) square pre-cast, prestressed concrete piles.
Finite element models for both piers were constructed using high resolution meshes of
brick elements, resultant beam elements, and discrete spring elements (Figure 3). Because the
barge impact points on each pier were at relatively low elevations, catastrophic structural failures
initiating in the columns were not expected for the impact conditions being simulated. Thus,
representation of structural behavior at the ultimate strength limit state was not the focus
(although independent ultimate strength checks were later conducted using the FB-PIER [9]
analysis code). Instead, model development was primarily controlled by the need for accurate
representation of mass distribution and structural stiffness as these properties significantly affect
both the magnitude and time variation of impact loads developed during a barge collision.

Modeling soil-structure interaction

An important aspect of modeling bridge substructures subjected to lateral load involves


adequately representing the interaction that occurs between the piles and the surrounding soil.
For the St. George Island Causeway bridge, soil boring logs containing standard penetration test
(SPT) blow count data were used to generate models of soil-pile interaction. Soil-pile interaction
for pier-1 and pier-3 was modeled using hundreds of groups of nonlinear soil springs positioned
at nodal points along the axes of the piles (Figure 3). At each node, a unique spring group was
specified that consisted of a vertical spring for modeling axial (skin) resistance and four
horizontal springs for modeling lateral resistance (Figure 4). Each soil spring group had
potentially unique properties because the SPT data varied from location to location, soil
properties vary with depth, and because row-multipliers were used to represent pile group
effects. Soil springs were also placed at the base of each pile to represent end-bearing tip

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 7

resistance. Non-linear load-displacement relationships for the axial, tip, and lateral soil resistance
springs were generated from the SPT boring log data using traditional soil-pile interaction
mechanics equations and using the FB-PIER software [9].
Modeling lateral soil resistance requires the use of nonlinear inelastic relationships.
Traditional soil mechanics calculations for lateral stiffness (p-y curves) yield nonlinear load-
displacement relationships for the soil in the undisturbed condition. However, when soil is
displaced in an oscillatory fashionas can occur during a dynamic vessel impact eventit is
subjected to multiple cycles of load reversal and cannot be treated as undisturbed except at the
beginning of the analysis. Instead, a gap model should be used to represent the soil behavior
under such loading conditions. A soil gap model tracks permanent components of deformation
separately in each direction (positive and negative) as cyclic loading occurs (Figure 5). Plastic
deformation of the soil on one side of the pile is assumed to have no effect on the state of the soil
located on the opposite side of the pile. Even after permanent deformation has occurred in one
direction (Figure 5a), subsequent loading in the opposite direction (Figure 5b) follows an
undisturbed p-y loading curve. Records of maximum sustained permanent deformation are
maintained separately for each direction of loading so that proper reloading of previously
disturbed soil (Figures 5c and 5d) can occur.
In this study, non-linear discrete spring elements were used to model soil stiffness,
inelastic soil gap formation and the energy dissipation (damping) associated with plastic soil
deformation. Dynamic load rate effects were not included in the soil springs, however, a range of
different soil stiffnesses were considered.

RESULTS FOR SIMULATED IMPACT CONDITIONS

Using the barge, pier, and soil models described above, a variety of different barge impact
conditions were simulated so that dominant impact load characteristics could be identified.
Impact simulation models were constructed by merging the barge together with a pier model,
establishing the correct draft for the payload condition being simulated, initializing buoyancy
springs to support the barge appropriately, and then specifying the desired initial barge speed. In
each case simulated, the centerline of the barge bow struck the centerline of the pier in a head-
on, perpendicular manner. Additionally, the tug pushing the barge at the prescribed speed was
assumed to have disengaged and backed away prior to impact, i.e. the barge was not under power
during the impact event.
LS-DYNA contact algorithms were used to determine when and where portions of the
barge contacted and dynamically interacted with the bridge pier. Since very large normal forces
are generated at the contact interface, frictional effects between the barge and the pier can be
quite significant. Thus, static, dynamic, and decay coefficients of friction for the steel-to-
concrete contact between the barge and the pier columns were specified as 0.5, 0.3, and 0.3
respectively. Frictional parameters were also specified for other contact definitions in the model
(e.g. internal steel-to-steel contact within the bow of the barge).

Dynamic impact loads and barge deformation

In this section, two impact conditions were simulated using LS-DYNA that represented
reasonable bounds of expected response for single-barge impacts. In each case, the barge struck
pier-1 with an initial speed of 4 knots in a head-on manner. However, two different payload
conditions were simulated: empty (1.78 MN (200 T) barge) and fully loaded (1.78 MN (200 T)

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 8

barge plus 15.12 MN (1700 T) payload). Note that the mass of the barge and payload in the fully
loaded condition was approximately an order of magnitude larger than that of the empty
condition. By extracting contact forces acting between the barge bow and the surface of the pier
column throughout the simulation, a time history of impact load acting on the pier was
determined. In Figure 6, the time histories of impact load and the load-deformation relationships
for the empty and fully loaded conditions are presented. For all results presented herein, time
t = 0 sec. corresponds to the time of initial contact.
In both simulations, the magnitude of the impact load imparted to the pier rises very
rapidly reaching more than 4.00 MN (900 kips) in approximately 0.025 seconds. Both
simulations also follow nearly identical paths up to the initial peak load. However, after this
point, the characteristics of the load history differ greatly. In the empty barge impact event, the
load drops rapidly back to zero as the empty vessel essentially bounces off of the pier. In
contrast, the much greater momentum of the fully loaded barge generates a time history of load
that rises beyond its initial peak and then drops off relatively slowly. Overall, the empty barge
impact is a short term, transient event whereas the fully loaded impact is a sustained, nearly
pseudo-static application of lateral load.
Differences in impact severity between the two cases are also evident when one examines
the barge deformations generated in each case. Figure 7 illustrates these deformations
graphically. While the empty impact generated very little deformation, the fully loaded impact
caused substantial damage and crushing of the bow. By correlating time histories of impact load
and barge crush depth, a plot can be generated that relates dynamic impact load to barge crush
deformation. Figure 6b confirms quantitatively the results presented graphically in Figure 7. The
empty impact condition generates a maximum deformation of only about 76 mm (3 in.) whereas
the fully loaded barge sustains 686 mm (27 in.) of deformation before elastically rebounding
slightly.
Several key aspects of barge impact load characteristics are indicated in Figure 6b. After
approximately 51 mm (2 in.) of deformation, the bow of the barge sustains permanent, non-
recoverable, plastic deformation. As crushing continues beyond this point, the contact force
(impact load) continues growing until it reaches a peak at approximately 254 mm (10 in.) of
crush depth. After maximizing, the load decreases but then subsequently plateaus and remains at
roughly the same level for a considerable amount of additional deformation. Examining the
curve in Figure 6b for the fully loaded impact, it is evident that a large amount of non-
recoverable plastic deformation occurs in the barge bow. Calculations for this fully loaded case
and several other severe impact scenarios (involving substantial barge crushing) have revealed
that the dissipated energy associated with this plastic deformation can amount to more than 50%
of the initial kinetic energy of the impacting barge. Thus, one of the primary mechanisms of
energy consumption in severe but structurally non-catastrophic impact events has been identified
as being more a function of the vessel characteristics than of the structural characteristics.

Sustained versus oscillatory nature impact load

The load histories presented in the previous sections exhibited moderate dynamic variation
through time. In this section, additional simulation results are presented for pier-1 and pier-3 in
which extensive oscillation of impact load is present. Primary focus is placed on identifying
conditions under which sustained pseudo-static behavior can be expected and conditions under
which significant load oscillation is likely to occur.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 9

Shown in Figure 8 are time histories of impact force generated when pier-1 is struck by a
half loaded barge at speeds of 1 knot and 6 knots. Since the initial kinetic energy of the 6 knot,
half loaded condition is close to that of the 4 knot, fully loaded condition discussed in the
previous section, it is not surprising that the impact load histories for these two cases are similar.
Both exhibit sustained load over a relatively long period of time. In contrast, the impact load
history for the 1 knot, half-loaded impact condition exhibits two distinct oscillations of imparted
load.
Results for half loaded barge impacts on pier-3 at 0.5 knots and 1 knot are also given in
Figure 8. (Lower impact speeds had to be used for pier-3 because it was not an impact resistant
pier and was not structurally capable of sustaining barge impacts at the same speeds as those
imposed on pier-1.) Oscillation in the load history for pier-3 is very pronounced with several
cycles of loading and unloading occurring as the barge, pier, and soil interact dynamically.
The equivalent static load concept used in the AASHTO barge provisions clearly does
not apply to cases where the impact loading is strongly oscillatory. What sets these cases apart
from the sustained loading cases is dynamic interaction between the barge, bridge pier, and soil.
Based on the results shown in Figure 8, as well as additional data not shown in this paper,
conditions that lead to oscillatory loading have been identified as: small barge deformations and
flexible pier structures. When one or both of these conditions exist, the barge and pier
dynamically and repeatedly bounce against each other at the contact interface. Pier loading of
this nature cannot be modeled using equivalent static loads unless very conservative values are
used. Furthermore, such conditions must not be dismissed as minor or secondary simply
because they often correspond to low impact speed. The drifting barge impact condition,
specified in the AASHTO provisions for the design of piers distant from navigation channels but
still over water, falls directly into this category of load behavior.

Comparison impact simulation loads and AASHTO loads

The final set of results presented in this paper compare barge impact loads predicted by finite
element impact simulation with the corresponding equivalent static loads computed using the
AASHTO provisions. To accomplish such a comparison, barge impact load calculations were
performed for both pier-1 and pier-3 using the AASHTO provisions. Only vessel collision force
calculations were performedrather than carrying out a full AASHTO probabilistic analysis
because this investigation was concerned solely with characterizing the isolated, unfactored
impact loads associated with the actual barge impact event, not with determining the probability
that such an event would occur. In all cases discussed in this section, the barge considered was a
half loaded jumbo hopper barge. Collision speeds were chosen to produce impact loads close to
the lateral capacities of the piers. Table 1 compares results from AASHTO load calculations
(documented in detail in Ref [7]) with results extracted from corresponding impact finite element
simulations. For the purpose of comparing AASHTOs static loads to results from dynamic
simulations, instantaneous peak dynamic forces (from the simulations) were chosen for
comparison since they represent very conservative, worst case conditions.
Results shown in Table 1 indicate that the predicted barge crush deformations are in
reasonable agreement, especially for the more severe impact condition associated with pier-1. In
contrast, the predicted magnitudes of impact loads differ considerably. In the case of pier-1,
AASHTO predicts a load that is much larger than even the peak load extracted from the dynamic
impact simulation. For pier-3, however, AASHTOs prediction is much lower than that obtained
from impact simulation.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 10

To explore this comparison further, consider the dynamic impact simulation data and
AASHTO empirical relationship plotted in Figure 9. Each of the curves shown in this figure
expresses load as a function of barge crush deformation. In general, the same differences evident
in Table 1 are also evident in Figure 9. For severe impact conditions in which significant barge
deformation occurs, e.g. the 6-knot dynamic impact condition shown, dynamic simulation
predicts loads that are less than those predicted by AASHTO. However, for light to moderate
impact conditionsgenerating less than 64 mm (2.5 in.) of barge deformationsimulation
predicts loads that are larger than those predicted by AASHTO.
These differences suggest that further investigation in this area is needed to determine if
changes to AASHTOs provisions are warranted. While the loads predicted by simulation for
light to moderate severity impacts are larger than those predicted by AASHTO, it must also be
pointed out that such loads act on the bridge only for short durations of time and with significant
oscillation. The question that must be addressed in the future is whether the structural response
to the AASHTO equivalent static loads is on the same level of severity as the structural response
when subjected to the actual dynamic time-varying loads.
Under severe impact conditions, AASHTOs use of equivalent static loads appears
reasonable since the time-varying load histories produced from impact simulation are sustained
for relatively long durations of time. However, the load magnitudes predicted by AASHTO are
considerably larger than those predicted by impact simulation. Additional research is needed in
order to further investigate these differences.

CONCLUDING REMARKS

Finite element models of a jumbo hopper barge, two bridge pier substructures, and soil-structure
interaction have been developed for use in dynamic impact simulations involving barges striking
bridge piers. Key results presented included time histories of impact loads and barge crush
deformations. Through the use of impact simulation, it has been determined that severe impact
events can rationally be represented using equivalent static loads and that a large portion of the
impact energy in severe impact events is dissipated through permanent deformation of the barge
bow. Comparisons between impact loads and barge deformations computed using dynamic
impact simulation and the AASHTO barge impact provisions have been presented and discussed.
Results indicate reasonable agreement in predicted barge deformation, however, significant
differences have been observed in predicted design loads. These differences suggest that further
investigation, both through simulation as well as experimental testing, is warranted to determine
if changes to AASHTOs relationship between barge crush deformation and impact load is
needed. Full scale, experimental barge impact testing of the St. George Island bridgewhich is
planned for 2003will provide experimental data supplementing the simulation data presented
herein.

ACKNOWLEDGEMENTS

The authors wish to thank the Florida Department of Transportation (FDOT) for providing the
financial support that made this study possible. The material presented in this paper is based
upon research that was supported by the FDOT under Contracts BC-354 RPWO#23 and BC-354
RPWO#56.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 11

REFERENCES

1. Larsen O.D. Ship Collision with Bridges: The Interaction between Vessel Traffic and
Bridge Structures. IABSE Structural Engineering Document 4. IASBE-AIPC-IVBH, 1993.

2. Knott M., Prucz Z. Vessel Collision Design of Bridges: Bridge Engineering Handbook.
CRC Press LLC, 2000.

3. AASHTO. Guide Specification and Commentary for Vessel Collision Design of Highway
Bridges. American Association of State Highway and Transportation Officials, 1991.

4. AASHTO. LRFD Bridge Design Specifications and Commentary. American Association of


State Highway and Transportation Officials, 1994.

5. Meir-Dornberg K.E. Ship Collisions, Safety Zones, and Loading Assumptions for
Structures in Inland Waterways. VDI-Berichte No. 496, 1983, pp. 1-9.

6. Hallquist, J.O. LS-DYNA Theoretical Manual. Livermore Software Technology


Corporation, 1998.

7. Consolazio G.R., Cook R.A., Lehr G.B., and Bollmann H.T. Barge Impact Testing of the
St. George Island Causeway Bridge, Phase I : Feasibility Study. University of Florida
Engineering and Industrial Experiment Station, Structures Research Report No. 783, 2002.

8. American Bureau of Shipping. Rules for Building and Classing Steel Vessels for Service on
Rivers and Intracoastal Waterways, 1997.

9. Florida Bridge Software Institute (BSI). FB-PIER Users Manual. University of Florida,
Department of Civil & Coastal Engineering, 2000.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 12

LIST OF TABLES

TABLE 1 Comparison of predicted impact loads and barge deformations.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 13

LIST OF FIGURES

FIGURE 1 Schematic diagram of hopper barge model.

FIGURE 2 Detailed view of modeling used in zone-1 of the barge.

FIGURE 3 Finite element models of piers and soil springs.

FIGURE 4 Typical spring group representing lateral and vertical


soil stiffness at a node on the centerline of a pile.

FIGURE 5 Load vs. displacement diagrams (p-y curves) for gap model.

FIGURE 6 Results from simulated impacts on pier-1.

FIGURE 7 Maximum barge deformation sustained during impact with pier

FIGURE 8 Sustained and oscillatory time histories of impact force.

FIGURE 9 Comparison of force-deformation relationships based on


dynamic impact simulation and AASHTO provisions.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 14

TABLE 1 Comparison of predicted impact loads and barge deformations.

Pier-1 Pier-3
Impact
parameter AASHTO Impact AASHTO Impact
provisions simulation provision simulation
Half loaded Half loaded Half loaded Half loaded
Barge payload
9.34 MN 9.34 MN 9.34 MN 9.34 MN
condition
(1050 tons) (1050 tons) (1050 tons) (1050 tons)

Impact velocity 6.0 knots 6.0 knots 0.5 knots 0.5 knots

5.338 MN 1.246 MN
7.433 MN 0.427 MN
(1200 kips) (280 kips)
Impact force (1671 kips) (96 kips)
(peak (peak
(static equiv.) (static equiv.)
dynamic) dynamic)
Maximum barge 894 mm 838 mm 7.1 mm 4.6 mm
crush depth (35.2 in) (33.0 in) (0.28 in) (0.18 in)

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 15

Barge modeling Bridge Pier

Direction of barge Zone-1


motion
Barge modeling
Zone-2
Barge modeling
Zone-3

Waterline
Barge support
node

Base node

Buoyancy springs

FIGURE 1 Schematic diagram of hopper barge model.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 16

Plan view of barge bow


Internal trusses, discretely modeled
Outer plate on
with shell elements in zone-1
top of barge in
zone-1
Barge
head log
Head log

Outer plates on
side of barge

FIGURE 2 Detailed view of modeling used in zone-1 of the barge.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 17

Bent cap
Bent cap

Pier column
Pier column

Barge Pile cap


Barge impact
impact Shear wall
load
load
Pile cap
Tremie seal
Precast piles
H-piles

Soil springs Soil springs

a) Pier-1 b) Pier-3

FIGURE 3 Finite element models of piers and soil springs.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 18

z-direction
Nodal points along axis of
battered pile (typ.)
n
Resultant elements tio
di rec
representing pile y-

x-direction

Lateral (P-y and P-x)


Axial (T-z) soil springs
soil springs

FIGURE 4 Typical spring group representing lateral and vertical


soil stiffness at a node on the centerline of a pile.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 19

+p
1 2 +y

-y +y

+ Gap

a) Loading undisturbed soil in +y direction with permanent deformation


+p -y
2

4 3
-y +y

+ Gap

6 5
- Gap + Gap

b) Unloading from +y direction with +gap formation; loading


undisturbed soil in y direction with permanent deformation
+p 9 10 +y

7
-y 8 +y

- Gap + Gap

6 - Gap + Gap

c) Unloading from -y direction with gap formation; traversal of gap and +gap;
loading disturbed soil in +y direction with additional permanent deformation
+p -y
10

12 11
-y +y

- Gap + Gap

14 13 - Gap + Gap

d) Unloading from +y direction with expanded +gap; traversal of +gap and gap;
loading disturbed soil in -y direction with additional permanent deformation

FIGURE 5 Load vs. displacement diagrams (p-y curves) for gap model.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 20

1200
5 Pier-1, 4 knot, fully loaded
Pier-1, 4 knot, empty
1000
4
800
Impact Force (MN)

Impact Force (kip)


3
600

2
400

1 200

0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Time (sec)

a) Time history of impact force


Barge Crush Depth (in)
0 4 8 12 16 20 24 28
1200
5
Pier-1, 4 knot, fully loaded
Pier-1, 4 knot, empty 1000
4
800
Impact Force (MN)

3 Impact Force (kip)


600

2
400

1 200

0 0
0 100 200 300 400 500 600 700
Barge Crush Depth (mm)

b) Relationship between impact force and barge crush deformation

FIGURE 6 Results from simulated impacts on pier-1.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 21

Minimal bow
deformation
Width of
contact zone

a) Deformation caused by impact on pier-1, 4 knots, empty barge

Crush
depth

b) Deformation caused by impact on pier-1, 4 knots, fully loaded barge

FIGURE 7 Maximum barge deformation sustained during impact with pier.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 22

1200
5 Pier-1, 6 knots, half loaded
Pier-1, 1 knot, half loaded
1000

4
800
Impact Force (MN)

Impact Force (kip)


3
600

2
400

1 200

0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Time (sec)

a) Sustained and oscillatory impact force on pier-1


600
2.5 Pier-3, 0.5 knots, half loaded
Pier-3, 1 knot, half loaded
500
2
400
Impact Force (MN)

1.5 Impact Force (kip)


300

1
200

0.5 100

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (sec)

b) Oscillatory impact force on pier-3

FIGURE 8 Sustained and oscillatory time histories of impact force.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.
Consolazio, Lehr, McVay 23

Barge Crush Depth (in)


0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
7 1600

1400
6

1200
5
Impact Force (MN)

Impact Force (kip)


1000
4
800

3
600

2 Barge sustains only


minimal crushing 400
during 1-knot impact
on pier-3 Pier-1, 6 knot, 1/2 loaded, impact
1 Pier-3, 1 knot, 1/2 loaded, impact 200
AASHTO Relationship

0 0
0 100 200 300 400 500 600 700
Barge Crush Depth (mm)

FIGURE 9 Comparison of force-deformation relationships based on


dynamic impact simulation and AASHTO provisions.

TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.

You might also like