Highway IDEA Program

Manufacture and Testing of a Filament

Wound Composite Bridge Superstructure

Final Report for Highway IDEA Project 63

Prepared by:
I.D. Parsons, S. White, D. Therriault, J. Bignell, University of Illinois

February 2002
INNOVATIONS DESERVING EXPLORATORY ANALYSIS (IDEA)
PROGRAMS
MANAGED BY THE TRANSPORTATION RESEARCH BOARD (TRB)

This NCHRP-IDEA investigation was completed as part of the National Cooperative

Highway Research Program (NCHRP). The NCHRP-IDEA program is one of the four
IDEA programs managed by the Transportation Research Board (TRB) to foster
innovations in highway and intermodal surface transportation systems. The other three
IDEA program areas are Transit-IDEA, which focuses on products and results for transit
practice, in support of the Transit Cooperative Research Program (TCRP), Safety-IDEA,
which focuses on motor carrier safety practice, in support of the Federal Motor Carrier
Safety Administration and Federal Railroad Administration, and High Speed Rail-IDEA
(HSR), which focuses on products and results for high speed rail practice, in support of
the Federal Railroad Administration. The four IDEA program areas are integrated to
promote the development and testing of nontraditional and innovative concepts, methods,
and technologies for surface transportation systems.

For information on the IDEA Program contact IDEA Program, Transportation Research
Board, 500 5th Street, N.W., Washington, D.C. 20001 (phone: 202/334-1461, fax:
202/334-3471, http://www.nationalacademies.org/trb/idea)

The project that is the subject of this contractor-authored report was a part of the Innovations Deserving
Exploratory Analysis (IDEA) Programs, which are managed by the Transportation Research Board (TRB) with the
approval of the Governing Board of the National Research Council. The members of the oversight committee that
monitored the project and reviewed the report were chosen for their special competencies and with regard for
appropriate balance. The views expressed in this report are those of the contractor who conducted the investigation
documented in this report and do not necessarily reflect those of the Transportation Research Board, the National
Research Council, or the sponsors of the IDEA Programs. This document has not been edited by TRB.

The Transportation Research Board of the National Academies, the National Research Council, and the
organizations that sponsor the IDEA Programs do not endorse products or manufacturers. Trade or manufacturers'
names appear herein solely because they are considered essential to the object of the investigation.
 TIBRARY
TR,q.NspcRT,{TroN
liËöe^ncr} ¡-+ç.4 p ¡,
Table of Gontents L,I
1. INTRODUCTION AND OBJECTIVES .......... I

2. FRp BRtDcE DES|GNS ............... 3

2.1 Overview of Previous Research ........3

2.2 The Proposed Filament Wound Bridse Superstructure

3. INITIAL DESIGN OF THE PROTOTYPE BRIDGE 10

4. MANUFACTURING TOOLING DESIGN ....14

4.1 Inner Cell Mandrel ...14

4.2 Support for Outer Shell \ilindins l5

5. FILAMENT WINDING MANUFACTURTNG PROCESS 27

5.1 Manufacturinq Equioment ,n

5.2 FRP Comoosite Material 28

5.3 Manufacture of the Inner Cell 28

5.3.1 Winding Pattern Creation ......................29
5.3.2 Mandrel Coatins ...............29
5.3.3 Inner Cell'Winding ...........29
5.3.4 Vacuum Baeging and Curing ................ 30
5.3.5 Final Inner Cell Part Preparation ........... 30

5.4 Manufacture of the Outer Shell

5.4.1 Location
Inner Cell ........... 3l
5.4.2 Expanding Foam Application ................32
 5.4.3 Winding of the Outer Shell ....................32
5.4.4 Vacuum Bagging and Curing ................32
5.4.5 Final Outer Shell Preparation ................33

6. EXPERTMENTAL EQUIPMENT .................. 61

Test Confìsuration

6.2 Test Equipment .............61

6:2.1 Two Cell Bridge Test ....... 6l
6.2.2 Six Cell Bridge Test .........62

7. RESULTS OF THE EXPERIMENTAL TESTS ............ 69

7.1 Two Cell Bridees .................69

.l.l Specimen Description
7 .......69
.1.2 Experimental Procedure
7 ........................ 70
7.1.3 Experimental Results ........ 70
7.1.4 Discussion of Results ........ 70

7.2 Six Cell Bridees.

7.2.1 $pecimen Description .......71
7.2.2 Experimental Procedure ........................ 7l
7.2.3 Exoerimental Results for the Glass Bridses ...............72
7.2.4 Exps:rimental Results for the Glass-Carbon Bridees ............................................ 73
7.2.5 Experimental Results for the Glass Bridee Fatigue Test .................74
7.2.6 Summary of the Experimental Results .......................75

8. FINITE ELEMENT ANALYSIS OF THE PROTOTYPE BRIDGES 113

8.1 Two Cell Bridees

8.2 Six Cell Bridees 115

9. DISCUSSION OF RESULTS .... 131

Manufacture of the Proposed Bridee Sunerstructure .................... 131

Testine of the Fabricated Soecimens .............. 132

9.3 Finite Element Analvsis of the Manufactured Specimens ....133

9.4 Conclusions...

10. REFERENCES ...... 139

 Draft - 9/24/Ol Composite Bridge Superstructure

1. lntroduction and Objectives

This report documents the research performed under contract number NCHRP-63,
"Manufacture and Testing of a Filament Wound Composite Bridge Superstructure," by I.
D. Parsons (Associate Professor of Civil Engineering), S. White (Associate Professor of
Aeornautical and Astomautical Engineering), D. Therriault (Graduate Research
Assistant, Department of Aeronautical and Astronautical Engineering) and J. Bignell
(Graduate Research Assistant, Department of Civil and Environmental Engineering) at

the University of Illinois at Urbana-Champaign from October 1999 to August 2001.

The primary aim of this work was to determine the manufactuability and
experimental performance of a conceptual design for a filament wound bridge
superstructure described in [3,4]. Finite element simulations [4] demonstrated that the
proposed design can withstand the loads of heavy traffic, with a weight saving of at least

50% over a conventional steel-concrete bridge. This weight saving can be translated into
lower construction costs, and increased seismic resistance (since the supporting columns
have to carry less load). Furthermore, the durability of the fiber glass reinforced material
will substantially reduce the life cycle cost of the bridge (i.e., the material eliminates
corrosion problems associated with reinforced concrete decks).

However, before the construction industry can be expected to invest in the

development of this structural system necessary to produce fully functioning structures,
the manufacturing process and theoretical predictions must be verified by building and
testing some small scale prototypes. The primary objectives of this research project were
therefore:

l. Build a small number of scaled prototypes to demonstrate the feasibility of the

manufacturing procedure;

2. Test the prototypes measure their stiffrress, strength and fatigue properties;

3. Build finite element models of the fabricated specimens to determine if their

structural properties can be predicted.
Draft -9/24/01 Composite Bridge Superstructure

The report is organized in the following manner. Section 2 summarizes relevant work
by other researchers and discusses the features of the conceptual design. Section 3
described the procedures that were adopted to produce the specifications for the small-
scale protot¡pes that were manufactured and tested. The design of the fabrication
equipment is documented in Section 4. The manufacturing process is described in Section
5. The experimental equipment used to test the structural performance of the prototype
bridges is described in Section 6. Section 7 contains the experimental data obtained using

this equipment. Section 8 presents the results of finite element simulations of the
manufactured prototlpes. The results of this research project are discussed in Section 9.
 Draft - 9/24/01 Composite Bridge Superstructure

2. FRP Bridge Designs

This section presents a brief overview of fiber reinforced plastic bridge research
performed by other workers, and outlines the structural features and manufacturing
processes of the bridge design considered in this report.

2.1 Overuiew of Previous Researcå

Bridges are vital components of the nation's infrastructure, many of which are
deteriorated. The total estimated cost of bringing deficient bridge superstructures to an
acceptable level is $110 billion [6i. Fiber-reinforced plastics present several advantages
over conventional bridge materials. One of the primary sources of bridge deterioration is
steel corrosion. The de-icing chemicals used in winter deteriorate the structural sections

and the reinforcing re-bars in conventional designs. Fiber reinforced material is a viable
solution to this problem. h earthquake prone areas, lighnveight modular bridge
components can be used as permanent replacements for damaged bridges to reduce the
intemrption of traffic that normally accompanies conventional construction. The
availability of such designs will aid in restoring bridges faster than other methods of
construction. Moreover, the low weight of composite materials, resulting in lower inertia
forces, can be a signifrcant factor in designing bridges subjected to seismic activity.

These observations have motivated several researchers to consider fiber reinforced

plastics as an alternative bridge construction material, primarily because of the corrosion
resistance of this class of materials. These efforts can be divided into three categories:
retrofitting schemes to repair and upgrade bridge components (e.g., l7l,[ll]); design of
replacement components made from fiber reinforced plastics (e.g.,
[11],[13],[15]); ana
the design of new structures fabricated from composites (e.g.,
[5],[10],[16]).
A number of fiber-reinforced polymer (FRP) bridges have been designed or
constructed in the United States and other countries. In England, the first FRp bridge to
carry trucks was recently completed (t10]). The bridge, with a 26.9 ft. rp* and a width
of 14.25 ft., crosses the Stroudwater Canal and is considered an important example of the

use of FRP in a civil engineering application. Lockheed Martin Corporation researchers

designed and tested a 30 ft. span bridge ([5]) in which pultruded panels form the bridge
Draft -9/24/01 Composite Bridge Superstructure

deck and are attached to three U-shaped girders with mechanical fasteners. This structure
carried a load of up to 100 kips without failure. Measurements of strain in the deck and
the beam components of this bridge revealed a significant margin of safety when
comparing these values with the material's ultimate strain.

The design and construction of a 450 ft. cable-stayed bridge that will cross U.S. 5 in
California is being studied by government, industry, and the University of California at
San Diego (UCSD). UCSD researchers also have developed modular short-span bridge
systems ([6]), employing a fiberglass bridge deck supported by a number of carbon
shells filled with lightweight concrete. More recently, a 33 ft. span FRP bridge was
designed and constructed in Ohio through a collaborative effort between the State of
Ohio, private industry, and Wright Patterson Air Force Base engineers ([8]). A bridge
'West
deck system developed by researchers at Virginia University and constructed of
pultruded sections was used to replace the deck of Laurel Lick Bridge in V/est Virginia
(t12]). ln New York State, two FRP bridge systems are planned for construction. A
bridge superstructure built by Hardcore Composites of New Castle, Delaware, has been
completed, and another FRP bridge deck is still in the planning process ([l]). A
comprehensive discussion of the current state-of-the-art is given in [1a].

2.2 The Proposed FilamentWound Bridge Superstructure

The proposed structural system studied in this report evolved to its final stage by
considering manufacturing techniques, material capabilities and structural behavior. The
resulting design represents a new integral bridge superstructure that does not mimic
current bridge design. Deviation from curent practice is essential if FRP is to be used
successfully in bridges, since conventional design and construction methodologies are
tailored specifically to the inherent advantages of steel and concrete, and using the same
designs cannot be expected to produce an efficient fiber reinforced bridge structure.

2.2.1 Structural Sysfem

The bridge superstructure is shown in Figure 2.1 and consists of two components: a
series of inner cells, lying parallel to the direction of traffic, and an outer shell. Special

consideration was given to three factors when this design was developed. First, filament
 Draft - 9/24/01 Composite Bridge Superstructure

winding was chosen as the basic manufacturing procedure, since it allows for automated
manufacturing, with faster fabrication cycles and reduced manufacturing costs. Second,
attention was given to the transfer of shear between the different components; the oval
inner cells provide sufficient contact area to reduce the shear stress to acceptable values.
Third, additional stiffrress and strength was developed by incorporating shell behavior
into the structure, rather than relying solely on plate bending cornmon in standard bridge
deck designs. Shell action is obtained by providing contact between the inner cells.

The proposed design has several important features. The designer can readily choose
the material, winding angles and thickness of plies to comply with appropriate strength
and stiffrress criteria. The circular shapes used in the inner cells and the outer shell avoid

stress concentrations that are present at comers of an I- or box-section. Effective transfer

of shear through the bridge is achieved byproviding large contact areas between the inner
cells and the outer shell. The contact area can be chosen based on the shear strength of
the bonding material. The proposed design develops shell action in a structure primarily
acting in bending by providing contact between the inner cells at the mid-surface.

Previous research t4l used finite element analysis to investigate the potential
performance of this structural system. For example, a two lane highway bridge with a
span of 60 ft., a width of 30 ft. and a depth of 3 ft. was designed. These dimensions are
tlpical of many highway bridges found in the U.S.A., and produce a length-to-depth
aspect ratio of 20:1, which is within the ranges found in bridges constructed using
conventional materials. The numerical model of the bridge was subjected to the traffic
loads specified by AASHTO [2]. The analysis demonstrated that the L/800 deflection
criterion controls the design with a weight saving of at least 50Yo over a conventional
steel-concrete bridge.

2.2.2 Manufacturing Process

The bridge is manufactured using the filament winding process. Filament winding is a
manufacturing technique that has been used extensively in the aerospace industry since
the early 1960's. The principles of filament winding are presented in Figure 2.2. Filament
is wound onto a mandrel surface through a payout head that traverses along the length of
a rotating mandrel. The angle at which fibers are placed in relation to the mandrel axis
 Draft - 9/24/01 Composite Bridge Superstructure

can be controlled by the ratio of the traverse speed of the payout

head to the rotational
speed of the mandrel. This delivers high-speed and accurate lay-up
of fiber reinforcement
a¡ound the mandrel. The fiber winding angles are machine controlled
to form the required
stacking sequence specified by the designer. Several types of reinforcement
can be used,
including fibers, prepreg tape and rovings. Subsequent to filament winding of the
structure, the entire assembly is then cured either in an autoclave
or oven, with a vacuum
bag being used around the structure to draw off voids during
the cure cycle.

The filament wound bridge considered in this report is manufactured

in a two stage
process. First, the inner cells of the bridge deck are filament
wound and cured separately
using a mandrel that can be extracted after cure. After manufacturing
the inner cells are
trimmed and prepared for integration with the outer shell.

The outer shell structure is constructed by winding filament directly

onto the inner
cells as shown in Figure 2.3. A series of inner cells are coupled together
with a mandrel
framing system. The frame is assembled to end pieces that contain a chuck
coupling that
fits into the filament winder chucks. Once the inner cell mandrel system is placed
into the
filament winder, the outer shell filaments are wound onto the inner cell
surfaces. The
entire assembly is then cured in an autoclave under vacuum pressure to produce
the
finished part.
Draft - 9/24/01 Composite Bridge Superstructure

outer shel

inner cells

Figure 2.1: Filament wound bridge superstructure conceptual design.

Draft - 9/24/01 Composite Bridge Superstructure

payout head
chuck
mandrel

traverse ;

Figure 2.2:The filament winding process.

Draft - 9/24/Ol Composite Bridge Superstructure 9

chuck coupl¡ng
i.r
filaments

inner cells

Figure 2.3: Filament winding of the outer shell onto the inner cells.
Draft -9/24/01 Composite Bridge Superstructure l0

3. lnitial Design of the Prototype Bridge

This section describes the initial design of the small-scale prototlpe filament wound
bridges that were manufactured and tested. Small-scale versions of the bridge design
described in Section 2 were necessitated by the dimensions of the filament winder and
autoclave available at the University of lllinois. The filament winder had a maximum part
diameter of 32 in. and a maximum length of 8 ft., and the autoclave used to curing of the
wound parts had an inner diameter of 32 in. and a total length of 80 in. Thus, in order to
preserve the aspect ratios encountered in full-size highway bridges, each model was
approximately 60 in. long, 24 in. wide and 3 in. deep. Six inner cells were used in each
specimen.

In order to make more detailed selections for the various dimensions of the model
bridges, an Excel worksheet was created to optimizetheprototlpe dimensions, see Figure
3.1. The spreadsheet computes the prototype aspect ratios with those of a hlpothetical
full-size structure. The input pararneters are indicated with yellow cells and are:

o Cell and shell thickness;

o Inside radius of the cell;

o Width the flat surface of the cell;

o Number of cells;

o Total length of the bridge.

The computed ouþuts are:

o Bridge length to width aspect ratio;

o Bridge length to depth aspect ratio;

o Bridge width to depth aspect ratio;

o Shell thickness to cell thickness ratio;

Total coverage area to area of the cross-section ratio;

Weight of the bridge;

 Draft - 9124/01 Composite Bridge Superstructure 1l

. Approximate price of the bridge.

The price reflects the cost of materials only, and is computed by multiplying the weight
of the bridge with the price per pound of the material.

After a number of iterations, the initial bridge geometry shown in the spreadsheet
column denoted as "standard" was selected. This geometry has aspect ratios that are very
similar to those that would be expected in the full-size structure. Note that the input
parameters of the inner cell (inner radius and flat surface) for this geometry are rounded

to preferred sizes in order to construct the mandrel with standard tubing.

After the geometric dimensions of the model bridge had been selected, appropriate
winding angles of the inner cell and the outer shell had to be chosen. The limitations of
the manufacturing equipment were the controlling factor. Ideally, the winding angle and
thickness of each layer in the inner cells and outer shell would be specified used the
optimization procedure described in [3]. However, given that the length of the test
specimens were close to the maximum mandrel length of the filament winding machine,
only a limited range of winding angles could be considered. This is as result of the need
to provide sufficient excess material when winding the outer shell to allow the payout
head to change direction. Small winding angles (where 0' and 90' represent winding
angles parallel and perpendicular to the axis of the rotating mandrel, respectively) would

require more excess material, and would consequently timit the final length of the
finished part. Large winding angles would avoid this problem, but would tend to produce
specimens with low stiffness and strength. Therefore, two winding angle configurations
were chosen that represent a compromise between these two competing factor:

o lnner cell 45' , outer shell 30';

o kmer cell 45", outer shell 45'.

Table 3.lsummarizes the selected nominal dimensions of the specimens. It should be

noted that these dimensions were limited by the available manufacturing equipment, and
were not chosen to optimize the structural perfonnance of the filament wound bridge
superstructure.
Draft - 9/24101 Composite Bridge Superstructure t2

Shell Cell

Number 1 6

Length 60 in. 60 in.

width 23.875 in. 3.938 in.

Depth 2.9375 in. 2.688 in.

Thickness 0.125 in. 0.09375 in.

Table 3.1: Summary of the prototype bridge dimensions.

Draft -9/24/01 Composite Bridge Superstructure 13

r. l

'Pert:: ç4, : .,:..Pa ra mgtgr:,r:

."ì..tÌ; -,i'Unlt.' ':Scaledl'
' i1,r:ì¡ tD' Full Slze Scaled2 Sta nda rr
r ntcKness t rn 0.75 0. 1 25 0.0625 0.09375
Outer radius ro tn 16.772 I .313 1.406 1.344
lnner radius n tn 16.O22 1.188 1.344 1.250
Flat distance d tn 8.986 1.313 1 .156 1.250
CELL width wcell tn 42.531 3.938 3.969 3.936
Height hôell tn 33.54s 2.625 2.813 2.688
Volume ræeil in^3 65340.3 78.6 41 .1 59.9
Weigth Wcell tb 4247.1 5.1 2.7 3.9
I ntcKness T tn 1.141 0.1 875 U.U93 /¡¡ 0.125
Outer rad¡us Ro tn 17.913 1.500
lnner radius Ri ln
^-' 16:.772
--
1.313
' 1.500
1.406
1.469
1.344
SHELL Width (total W¡dth) wshell tn 300 24 24 23.875
Height (total Depth) hshell tn 35.8267 3 3 2.9375
Volume whell in^3 523567.0 571.9 287.6 380.3
Weisht Wshell tb 34031.9 97.2 18.7 24.7
Number of cells N 7 o 6 t
Total Length Lcell ln 72C 60 60 6(
Ratio:Length/Width Rlw in/in
BRIDGE Ratio: Length / Depth Rtd in/in
Ratio:Width/Depth Rwd in/in
Ratio:Tshell /Tcell RTt inl ¡n
Ratio : Area cov. / Area mat. Rcor¡mat in^2tin^2
Total We¡ght of cells Wcellt tb 29729.8 30.7 16.0 23.4
Total Weioht of bridoe Wtot tb 63761 .7 67.8 34.7 48.1
Weight for 3 briclges W3tot tb 191 28s zv3 104 144
3 Approx. Price for fiber Pf $ 2r 51 96 229 117 162
3RIDGES Approx. Price for resin Pr $ 717919 763 391 541
Total Price Ptot $ 932514 992 508 703

Figure 3.1: Excel worksheet used for selection of the bridge geometry.
 Draft - 9/24/Ol Composite Bridge Superstructure l4

4. Manufacturing Tooling Design

This section described the tooling that was designed and fabricated to facilitate the
manufacture of the bridge superstructure using the equipment available at the University
of lllinois. Two sets of tooling were required: a mandrel for winding the inner cells, and
supports to hold the inner cells in place during the winding of the outer shell.

4.1 Inner Cell Mandrel

An aluminum mandrel was designed and manufactured to serve as the mandrel used
to wind the inner cells. Aluminum was chosen because of its high coefficient of thermal
expansion and its ease of machining. In order to reduce cost, this mandrel was assembled
from standard box and circular tube stock to create the desired cross-section shown in
Figure 4.1. Figure 4.2 shows engineering drawings of the complete mandrel. A 1.25 in.
by 2.5 in. rectangular tube was connected to two halves of a 2.5 in. diameter tube using
steel screws spaced at 6 in. intervals along the 84 in. length of the mandrel. Two inner
cells mandrels were built to facilitate rapid manufacture of the required cells.

Each end of the mandrel required an extension so that the mandrel could be installed

on the filament winder. Figure 4.3 shows drawings of the aluminum block extensions that
were fabricated for this purpose. Figures 4.4 and, 4.5 show photographs of these
extensions. The tailstock end of the mandrel incorporates an aluminum block fixed inside
the mandrel with two lateral screws. The headstock end of the mandrel, which locates
into the chuck of the filament winder), requires a circular cross-section extension block,
which is also fixed into the mandrel with lateral screws.

Each mandrel was tapered using several spacers located along the mandrel length.
The spacers were inserted between the cut circular tube edges and the lateral side of the
rectangular tube as shown in Figure 4.6. The spacer size was 0.040 in. at the tailstock end
of the mandrel and decreased to zero at the headstock end in increments of 0.010 in. This
taper greatly eased the removal of the cured inner cells from the mandrel.
 Draft - 9/24/Ol Composite Bridge Superstructure 15

4.2 Support for Outer Shell WindÍng

The bridge design considered in this report use the inner cells as the mandrel during
the winding of the outer shell. Tooling was therefore required that would hold the
selected number of inner cells (that was either two or six for the protoþpe bridges
considered in this project) in place curing this phase of the manufacturing procedure. This
tooling had to be adjustable so that different cell of different thickness could be
accommodated and so that gaps of various dimensions could be introduced between the
cells. Furthennore, this tooling had to be designed to fit inside the available autoclave
that was used to cure the outer shell after the filaments had been wound in place. Figure
4.7 shows engineering drawings of the tooling that was manufactured to fulfill these
objectives;this figure also indicates the dimensions of the autoclave employed.

The outer shell support tooling consists of two aluminum bars to which up to six
inner cell extensions can be attached. An aluminum bar is shown in Figure 4.8, and
Figure 4.9 shows six inner cell extensions attached to one of these bars. Figures 4.10 and
4.11 show engineering drawings of these components. The inner cell extensions are

simply 12 in. long versions of the inner cell mandrel, with rectangular blocks attached to
the inside the rectangular tube in a manner similar to the tailstock extension described
above. The long aluminum ba¡s are 32 in.long and contain two grooves and slots. The
slot provides the required adjustment between the positions of the iru:er cells by allowing
the inner cells to be moved laterally across the a,xis of the bridge. The groove is 0.25 in.

deep and constrains the rotation of the inner cell extension about its longitudinal axis.
Draft - 9/24/01 Composite Bridge Superstructure r6

2.50 in.

3.75 in.
Figure 4.1: Inner cell mandrel cross section.
Draft - 9/24/01 Composite Bridge Superstructure I7

RECIAI{GI'LA.R, ¡¡5 X2JO CTRSIJL¡R II'BE DIÀ. zJO

o o

DANIEL ΡERRTÂULT AÂE .068I

Figure 4.2:Engineering drawing of the inner cell mandrel.

Draft - 9/24/01 Composite Bridge Superstructure 18

I ¡-'-+-J

LEFT T]P
H¿F
RECTANGULAR RIGHT TIP

RUSBER VRAP
I.,/S THICK

CHUCK CtrUPLTNG MANDREL CRESS SECTITSI CHUCK COUPLING

NEAR TERMINAL RIGHT SIDE

Figure 4.3: Enginnering drawing of the inner cell mandrel and extension blocks.
Draft - 9/24/01 Composite Bridge Superstructure 19

r.l

Figure 4.4: Tailstock end of the inner cell mandrel.

Draft - 9/24/01 Composite Bridge Superstructure 20

Figure 4.5: Headstock end of the inner cell mandrel.

Draft - 9/24/01 Composite Bridge Superstructure 2t

i.;

¡'Ì

Figure 4.6: Location of the inner cell mandrel spacers.

Draft -9/24101 Composite Bridge Superstructure 22

+
J-
+
t
uE!¡uE
-¡.

dþilÞd!{ÞdÞd}
EIEü[E
Ë[HHTE
.:,.-.
I

Figure 4.7: Enginnering drawing of the outer shell support tooling.

Draft -9/24/01 Composite Bridge Superstructure 23

Figure 4.8: Aluminum ba¡ used in the outer shell support tooling.
Draft - 9/24/01 Composite Bridge Superstructure 24

Figure 4.9: Aluminum bar with inner cell support blocks used in the outer shell support

tooling.
Draft -9/24/01 Composite Bridge Superstructure 25

r SMAII-HOLEFOR
/ TArr.cHud(PrN

DA¡trEL TI¡ERRIAT'LÎ AÂE .Oú8I

Figure 4.10: Engineering drawing of the aluminum bar used in the outer shell support

tooling
Draft -9/24/01 Composite Bridge Superstructure 26

DAÌIEL ÎERRIÀI'LÎ AAT {68{

Figure 4.1 1: Engineering drawing of the inner cell support blocks used in the outer shell
support tooling.
Draft -9/24/01 Composite Bridge Superstructure 27

5. Filament Winding Manufacturing Process

This section describes the filament winding manufacturing process that was used to
manufacture the test specimens examined in this study. This process delivers high-speed
and accurate lay-up of fiber reinforcement around a mandrel. Since large volumes of
materials are involved in the construction of bridge superstructures, the choice of
filament winding is an attractive solution for automating the manufacturing process. It
also allows the designer to optimize the winding angle and ply thickness for mCIrimum
structural performance.

The two components of the bridge superstructure (the inner cells and the outer shell)
were made using the filament winding process. In this section, the equipment nad
material used to manufacture the bridges are discussed, followed by a step-by-step
description of the manufacture of the inner cells and the outer shell.

5.1 Manufacturing Equipment

5.1.1 The Filament Winder

The filament winder employed in this study an En-Tec four a:ris machine shown in
Figure 5.1. The four winding axes shown in this figure are:

o Motion of the payout head along the longitudinal æcis of the mandrel;

o Motion of the payout head along the vertical a,xis perpendicula¡ to the mandrel;

o Rotation of the eye on the payout head;

o Rotation of the mandrel.

A sketch of the winder is shown in Figure 5.2. The principal components of the winder
are: a payout head to wind the fiber that translates along a><es 1 and 2, a rotating eye to
wind a helical pattern that rotates along axis 3, a mandrel that rotates to wind the fiber, a

tailstock chuck that holds the tailstock end of the mandrel, a headstock chuck that holds
the headstock end of the mandrel, a control terminal, a material rack that contains the
spools of material and extensioners which control the fiber tension during the winding,
and a personal computer that creates the winding pattern file.
 Draft -9/24101 Composite Bridge Superstructure 28

The maximum part dimensions that can be manufactured on this machine is a

diameter of 32 in. and a length of 8 ft.. The winding pattern is created on a PC with the
software program FGX; the filament winder is operated using the program WIMAX 2.

5.1.2 Autoclave
A filament wound composite part is cured using the autoclave shown in Figure 5.3.
The autoclave heats the part using the selected temperature curing cycle while applying a
vacuum pressure to the part. Thermocouples are placed on the part an in the interior of
the autoclave to monitor the temperature time history.

5.2 FRP Composife Materials

The spools of FRP composite material must be installed in the material rack shown in
Figure 5.4. Two types of FRP material were used in this project: an E-glass/epoxy
prepreg tow and a carbon/epoxy prepreg tow, both manufactured by TCR Composites. In

both cases, the resin used was designated UF3325 by the manufacturers. The E-glass
fibers were Roving l58B AB 450 Tack-Pak Wrap Bulk Unit Stretch Wrap, and the
carbon fibers were Toray T700 -12000 -50C.

When winding a component, it is best to use two spools of material at the same time
to reduce the winding time. The utilization of a prepreg material requires the use of the
prepreg fiber disposal of the payout head, see Figure 5.5. After the tows were correctly
in order to control the tension during the
installed, the extensioners were calibrated
winding. The fiber tension was controlled mechanically by the springs inside the
extensioners.

The prepreg FRP material was stored in a freezer to extend its storage life. The
suggested temperature curing cycles are listed in Table 5.1. An example of a

recommended cure profile is illustrated in Figure 5.6.

5.3 Manufacture of the lnner Cell

Many cells were manufactured during the course of this project. The preliminary
design calculations discussed in Section 3 of this report provided the cell dimensions.

Each cell has a total wall thickness of 0.85 in., a winding angle of +45', and a length of
 Draft - 9/24/01 Composite Bridge Superstructure 29

69 in.. Two layers of two superposed tows of prepreg were required to obtain a part
approximately 0.085 in. thick.

5.3.1 Winding Pattern Creation

The helical winding pattem was created using the program FGX. The part diameter
was defined as the average between the width and the height of the mandrel since the
inner cell cross-section is not circula¡. The winding pattern was generated so that the
individual tows were placed one beside each other during the winding. This reduced the
number of crossovers to give a more compact and stiffer part.

5.3.2 Mandrel Coating

One manufacturing problem that was encountered during this study was the sticking
of the cured inner cell to the mandrel. This prevented the cured inner cell from being
removed from the mandrel. It was found that the combination of a tapered mandrel and a
mandrel wrapping of Uß in. Teflon film solved this problem.
The film was cut to the desired length (75 in.) and width (10.625 in) and wrapped
around the inner cell mandrel. The film was heated with a hot blower to facilitate its
application on the mandrel surface. High temperature tape was used to attach the film
edges together and to seal the joint as shown in Figure 5.7.

5.3.3 lnner Cell Winding

The leading end of the tow must be taped on the mandrel surface before starting the
inner cell winding to ensure that the tows will be pulled during the mandrel rotation. The
cell winding was performed at about 50% of the machine speed, see Figure 5.8. During
the winding operation, it was ensured that there was enough material on the spools, there
was no fiber slipping on the rotating eye rollers, and the fiber was not sliding on the
Teflon film. Two layers of 20 minutes duration were required to obtain the desired cell
thickness.
Draft -9/24/01 Composite Bridge Superstructure 30

5.3.4 Vacuum Bagging and Curing

A vacuum bag was used to apply pressure on the inner cell during its curing at high
temperature inside the autoclave. This pressure was required to ensure a good resin flow
between the fiber tows. The components of the vacuum bag used for the inner cells were:

. TeflonÆiberglass porous release film;

o Two layers of breather/bleeder cloth;

. Top and Bottom parts of vacuum port;

. Baggrng film;

o Silicone tape;

r High temperature tap.

First, the layer of release film was wrapped around the wound part. Then two layers
of bleeder cloth were applied to absorb the extra resin during the cure cycle. The bottom
part of the vacuum port was then installed on top of the bleeder cloth. All of the
components were taped with high temperature tape. A large piece of bagging film was
used to wrap the part and the mandrel completely using the silicone tape. A small hole
was made in the bagging film to assemble the top and bottom parts of the vacuum port.
The inner cell was cured inside the autoclave under vacuum pressure at the temperature
cycle prescribed by the material manufacturer. Figure 5.9 shows two inner cells inside the
autoclave ready to cure.

5.3.5 Final lnner Cell Pa¡t Preparation

The vacuum bag was cut open after completion of the curing cycle. An opened bag is
shown in Figure 5.10. The cured inner cell is then slide off the aluminum mandrel. This
was facilitated by the use of a tapered mandrel and the Teflon file. Figure 5.11 shows a
finished inner cell next to the filament winder with inner cell mandrel.

5.4 Manufacture of the Outer Shell

The construction of the outer shell of the bridge presents two basic problems. First,
the cross-section is further removed from the ideal circular shape than the inn cells. The
 Draft -9/24/01 Composite Bridge Superstructure 3r

relatively high width to depth ratio creates a problem with the creation of the winding
pattern using the FGX computer progr¿rm, since this software assumes that the part has a

circular cross-section. Thus there will be a difference between the predicted and the
actual winding angles. The winding process is further complicated by the tendency of the
fiber to slip at each end of the outer shell because the winding angle is low and the shell
is thick. Second, the gap between the outer shell and the inner cells shown in Figure 5.12
complicates the curing process. It is essentially to be able to apply pressure during the
cure cycle to the unsupported portion of the outer shell that is located over this gap. This
is necessary to ensure adequate resin flow during cwe so that voids are eliminated form
the frnal part and to ensure that the outer shell is flat.

These two issues were solved using the techniques described in the remainder of this
section so that several two and six cell bridges could be manufactured.

5.4.1 lnner Cell Location

The first task during the manufacture of the outer shell manufacture is to install the
required number of outer cells on the support tooling described in the previous section.
Figure 5.13 shows the support tooling at one end of the bridge prior to the location of the
inner cells; note the application of the Teflon film around the inner cell support blocks.
This film was required to ensure a tight fit of the inner cells over these support blocks.
Figure 5.14 shows six inner cells located on the outer shell support tooling prior to the
winding of the outer shell.

Durng the course of this research project, various gaps were introduced between the
inner cells on some of the two cell confrgurations. This is documented in subsequent
sections. These gaps were accommodated by using the adjustment provided through the
slots in the outer cell support tooling. \ilhen a gap of zero size was required (i.e., when
the inner cells were positioned to contact their neighbor - the most common approach
adopted in this study), the sides of each cell were sanded to remove any surface
imperfections. The edges of the inner cells were then bonded together using a five minute
epoxy glue.
 Draft - 9/24/01 Composite Bridge Superstructure 32

5.4.2 Expanding Foam Application

After the inner cells had been positioned in the support tooling, an expanding foam
material was applied in the gap between the outer shell and the cells to provide support
for the outer shell fibers during cure. This expanding foam offers a cheap, light and easy
to apply material to fill the gap. In addition, the foam resists relatively high temperatures
(up to a maximum of 120'C).

Figure 5.15 shows this foam being sprayed over the whole length of the inner cells.
After the foam had been applied, an aluminum plate coated with a Teflon film was used
to apply pressure to the foam as shown in Figure 5.15. This pressure aids in producing a
flatter foam surface. The foam was then trimmed and sanded; Figure 5.16 shows a six
cell bridge after the application of all of the expanding foam.

5.4.3 Wínding of the Outer Shell

The outer shell can be wound after the expanding foam has been applied to fill the
gap between the inner cells and the outer shell. The outer shell requires three layers of
two-tow thick material in order to obtain the desired shell thickness of 0.lZS in. Figure
5.17 and 5.18 show the winding of a two cell bridge with an outer shell angle of t30'
and a six cell bridge with an outer shell angle of t45", respectively. The winding time is
approximately 45 minutes and2 hours per layer for the two cell and six cell bridges for a
machine speed of 50 o/o,
respectively.

5.4.4 Vacuum Bagging and Curing

After the outer shell has been wound, the entire bridge superstructure was placed in a
vacuum bag and cured in the autoclave. The composition of the vacuum bag was similar
to that used during the manufacture of the inner cell.

Two aluminum plates were placed between the bleeder and breather layers as shown
in Figures 5.19 and 5.20. These plates were used to apply a pressure during cure over the
top and bottom surfaces of the bridge so that these surfaces remained flat. Two and four
vacuum ports were used in the bags of the two and six cell bridges, respectively. After the
vacuum bag had been installed, the bridge was placed inside the autoclave for curing (see

Figures 5.21 and 5.22 for the two and six cell bridges, respectively.
Draft - 9/24/01 Composite Bridge Superstructure 33

The outer shell was cured in a similar manner to the inner cells: inside the autoclave
under a specified temperature cycle and vacuum pressure. A different curing cycle was
designed because the recommended curing temperature exceeded the mæcimum foam
temperature. The objective of this modification was to gel the resin at a lower
temperature (110"C) before the complete curing at 155 oC as shown in Figure 5.23. This

modified curing cycle required a total time of five hours.

5.4.5 Final Outer Shell Preparation

After the curing cycle was complete, the vacuum bag was cut and removed from the
bridge to produce the parts shown in Figures 5.24 md 5.25. The final step involved
trimming the ends of the cured part. A completed six cell bridge is shown in Figures 5.26.
Draft - 9/24/01 Composite Bridge Superstructure 34

Temperature Time
@egree F) / (degree C) (Hour)
270 / t35 4
290 / tAs 2

310 / 155 I

Table 5.1: Manufacturer's recorlmended cure temperature and hold time for E-
glass/epoxy prepreg.
Draft - 9/24/01 Composite Bridge Supersfructure 3s

\-,

\
t
I

Figure 5.1: The En-Tec four axis filament winder employed in this study.
Draft - 9124/01 Composite Bridge Superstructure 36

Payout head

Winder Computer
Figure 5.2: Schematic representation of the filament winder.
Draft - 9/24/01 Composite Bridge Superstructure 37

Figure 5.3: The autoclave employed in this study.

Draft -9/24/01 Composite Bridge Superstructure 38

Extentioner

.¡t
Ij;""--

Spool of E-glassÆpoxy prepreg

tow (around 12 lbs.)

Figure 5.4: Installation of the FRP composite material spools on the rack.
Draft - 9/24/01 Composite Bridge Superstructure 39

Figure 5.5: Prepreg winding configuration for the payout head.

Draft - 9/24/01 Composite Bridge Superstructure 40

Temperature
@egree F)

310

150

70 Ramps a¡e at * or - 5 F/min

1.0 hr Time

Figure 5.6: Manufacturer's recommended cure profile for E-glass/epoxy prepreg.

Draft - 9/24/Ol Composite Bridge Superstructure 4l

Figure 5.7: Inner cell mandrel with Teflon film installed on the filament winder.
Draft -9/24/0t Composite Bridge Superstructure 42

Figure 5.8: Manufacture of an inner celr at a winding angle of +45".

Draft - 9/24/Ol Composite Bridge Superstructure 43

Figure 5.9: Two inner cells inside vacuum bags in the autoclave prior to curing.
Draft - 9/24/01 Composite Bridge Superstructure 44

Figure 5.10: Inner cell on mandrel and opended vacuum bag after curing in the autoclave.
Draft - 9/24101 Composite Bridge Superstructure 45

Figure 5.11: Cured inner cell after removal from the mandrel.
Draft -9/24101 Composite Bridge Superstructure 46

Figure 5.12: Gap between the outer shell and the inner cells.
Draft - 9l24l0I Composite Bridge Superstructure 47

Figure 5.13: Support tooling used for the outer cell winding of a six cell bridge.
Draft - 9/24/01 Composite Bridge Superstructure 48

Figure 5.14: Inner cell location for the outer shell winding of a six cell bridge.
Draft - 9/24/01 Composite Bridge Superstructure 49

Figure 5.15: Application of the expanding foam.

Draft -9/24/01 Composite Bridge Superstructure 50

Figure 5.16: Expanding foam after trimming.

Draft - 9/24/01 Composite Bridge Superstructure 51

Figure 5.17: rü/inding of the outer shell for a two cell bridge at an angle of +30'.
Draft -9/24/01 Composite Bridge Superstructure 52

Figure 5.18: V/inding of the outer shell for a six cell bridge at an angle of +45'.
Draft - 9/24/01 Composite Bridge Superstructure 53

Figure 5.19: Vacuum bag assembly before the insertion of the aluminum plates.
Draft - 9124/01 Composite Bridge Superstructure 54

Figure 5.20: Vacuum bag assembly after the insertion of the aluminum plates.
Draft - 9/24101 Composite Bridge Superstructure 55

Figure 5.21: Two cell bridge inside the autoclave after complete assembly of the vacuum
bag.
Draft -9/24/01 Composite Bridge Superstructure 56

Figure 5.22: Six cell bridge inside the autoclave after complete assembly of the vacuum
Draft - 9124/01 Composite Bridge Superstructure 57

Temperature
(" c)
155 0c Maximum fbam
temperature (120'C)
ll0"c
Time
(min)

Figure 5.23: Modifred curing cycle employed during the manufacture of the outer shell.
Draft -9/24/01 Composite Bridge Superstructure 58

Figure 5.24: Two cell bridge after outer shell curing.

Draft -9/24/01 Composite Bridge Superstructure 59

Figure 5.25: Six cell bridge after outer shell curing.

Draft - 9/24/01 Composite Bridge Superstructure 60

Figure 5.26: Final six cell bridge.

Draft -9/24/Ol Composite Bridge Superstructure 61

6. Experimental Equipment
This section describes the experimental equipment that was used to test the prototlpe
filament wound bridges. Structural testing of the superstructure prototypes was
performed to determine their strength and stiffrress as well as their fatigue strength. The
bridge strength is charactenzedby the first ply failure (FPF) of the composite. The same
basic testing configuration was used for both the two cell and the six cell bridges.

6.1 Test Configuration

The three point bending configuration was chosen to represent tlpical bridge
superstructure loadings. No attempt was made to scale the standard design loadings with
the physical dimensions of the small scale prototypes; instead a uniform load was applied
over a steel plate located at the center of the span of the bridges. This provided sufficient
information to both assess the stn¡ctural performance of the manufactured protot¡pes and
investigate the validity of finite element models of the bridges.

6.2 Test Equipment

Figures 6.1 and 6.2 show diagrams of the test equipment employed. The main
components of the setup are similar for the two and six cell bridge tests. A 20 kips
capacity tensile machine applied the load through a square steel plate (4 in. x 4 in. x 0.5
in.). The ends of the bridge were placed on fwo 4 in. diameter steel circular rollers to
produce the required simply-supported end conditions. Two l0 ft. long steel l-beams
were placed on the base plate of the testing machine; the circular rollers were fixed to
these I-beams to prevent any motion of the roller supports. A computer was used to
manage the data acquisition system.

6.2.1 Two Cell Bridge Test

Figure 6.3 shows a photograph of the specific arangement used to test the two cell
bridges. During these tests, only the applied load and vertical displacement of the

acfuator were measured.

Draft - 9/24/01 Composite Bridge Superstructure 62

6.2.2 Six Cell Bridge Test

Figures 6.4 and 6.5 show the testing configuration employed for the six cell bridges.
In addition to the applied load and vertical deflection of the actuator, 4 LVDT's and 14
strain gauges were used to record the deformation of the specimen. Figure 6.6 shows the
location of these transducers on the top and bottom surfaces of the bridge.
Draft - 9/24/01 Composite Bridge Superstructure 63

MTS Machine
Composite Bridge
Load Application Fixture (24x3x60)

Roller

Dia.4 in.

Figure 6.1: Side view of the experimental apparatus.

Draft - 9/24/01 Composite Bridge Superstructure 64

Load Application
Fixture

Composite Roller
Bridge

I-Beams

Base of
MTS

Figure 6.2: Front view of the experimental apparatus.

Draft - 9l24l0l Composite Bridge Superstructure 65

Figure 6.3: Test setup for a two cell bridge.

Draft -9/24/01 Composite Bridge Superstructure 66

14 strain gages (0 and 90o Orientation)

Rollers at each end

Figure 6.4: Test setup for a six cell bridge.

Draft - 9/24/01 Composite Bridge Superstructure 67

Èr-r.-...-,J{
æ-::':r-lã--

rvlTS
@

Figure 6.5: Front view of the test setup for a six cell bridge.
Draft - 9/24/01 Composite Bridge Superstructure 68

!
0"

l*
!
I

1i
!
I

!
I

5r !

f
i I

FI
!

2i G

TOP \¡IEW BOTTOM VIEW

Leeend:

l: Location of LVDT CNorth) A to G: Location of Strain Gages

2: Location of LVDT (South)
3: Location of LVDT (V/est) L: Length of Bridge
4: Location of LVDT (East) W: rwidrh of Bridge
5: Location of LVDT (Actuator)

Figure 6.6: Location of LVDT's and strain gauges on the six cell bridge.
 Draft -9/24/01 Composite Bridge Superstructure 69

7. Results of the Experimental Tests

This section presents the data obtained by testing the various prototype bridges that
\ilere manufactured using the procedures documented in Section 5. The testing equipment

and instrumentation were discussed in Section 6. Two categories of bridges were tested:
two cell and six cell bridges. Various configurations were examined within these two
categories to explore the structural properties of the general design. The objectives of the
tests were to measure the stiffiress and strength of the different specimens, and, in the
case of a six cell bridge, to investigate the fatigue life of the structure. The measured data
were used to assess the behavior of the structural system and to investigate the validity of
finite element models of the prototlpes.

7.1 Two Cell Bridges

7.1.1 Specimen Description

A total of five two cell bridges, each 51 in. long, were manufactured and tested.
These specimens were used to check the integrity of the manufacturing process and the

experimental equipment. Table 7.1 lists the characteristics of these bridges. The first four
bridges (bridges A, B, C and D) were constructed entirely of glass fiber reinforced
polymer wound at !45'. The inner cells of bridges A and D were placed so that there was
no gap between them. Bridges B and C were constructed with gaps of %u in. and /ru in.,
respectively, between the inner cells. This was done to investigate the additional stiffrress
provided by inner cell contact. Bridge E was manufactured with inner cells composed of
carbon fiber reinforced polymer wound at +45" and an outer shell composed of glass

fiber reinforced polymer wound at +45'. This specimen served to examine the feasibility
of varying the materials used to change the stiffness, strength and cost of the bridge.
Bridge A was tested with aluminum inserts placed inside the ends of the inner cells, see
Figure 7.1. This was done to determine whether it was necessary to provide support to the
ends of the inner cells in order to develop the full stiffüess of the structural system. All of
the other bridges were tested without these inserts, as shown in Figure 7.2 for bridge D.
 Draft -9/24/01 Composite Bridge Superstructure 70

7.1.2 Experimental procedure

Only the actuator displacement and the applied load were measured during the tests
of the two cell bridges. The bridges were simply supported by rollers located
/, in. from
the ends of the specimen. Thus the distance between the end supports was 50 in. Each
specimen was positioned so that the load was applied in the center of the span. In order to
remove any looseness from the testing equipment and to verify the functioning of the
instrumentation, two loading cycles were conducted up to a load of 1,500 lbs.
(corresponding to an actuator deflection of about /, in.).These initial loading cycles
were well within the elastic range of the specimens. The load was then increased until the

bridge failed. Initial failure of the specimen was indentified by either the first big crack or
the first load drop. When the bridge had been completely broken, the load was removed
and the data acquisition stopped.

7.1.3 Experimental Results

Figure 7.3 shows plots of the measured applied load versus actuator displacement for
the five two cell bridges. Table 7.2 lists the observed failure loads and stiffrress of each
specimen. The stiffrresses were obtained by measuring the slopes of the actuator load-
deflection curves within their linear ranges between applied load levels of 1,000 lbs. and
2,000 lbs. The failure mode of all of the two cell bridges was the development of cracks
parallel to the fibers in the compressive side of the outer shell near the load application
region as shown in Figure 7.4 for bridge D. These cracks were followed by localized
buckling of the outer layers of the outer shell. Figure 7.5 shows the final state of bridge
A. These large deformations were representative of the final state of the other two cell
bridges. When the load was removed, the bridges all retumed to almost perfectly straight
structures.

7.1.4 Discussion of Resu/fs

Comparison of the measured data for bridges A and D indicate that placing rigid
supports at the ends of the inner cell has little effect. The stiffrress differed by So/o,and the

measured load-deflection behavior was similar. These observations are within a

reasonable estimate of the repeatability of the manufacturing process and cannot be
 Draft - 9/24/01 Composite Bridge Superstnrcture 7l

attributed to the different end conditions. No deformation was observed at the ends for
either of these specimens, further indicating that the rigid supports had no measurable
effect on the behavior of the bridges. The data obtained for the four glass bridges (i.e.,
bridges A, B, C and D demonstrate that providing contact between the inner cells by
setting the gap size to 0 produces an increase in stiffrress. The use of carbon fiber inner
cells in bridge E results in a stiffer structure, as was expected, but is accompanied by a
reduction in strength, which was unexpected.

V.2 Six CelI Bridges

7.2.1 Specimen Description
A total of four six cell bridges were manufactured and tested using the procedures
documented in this report. Table 7.3 lists the characteristics of these bridges. The first
three bridges (i.e., bridges #1,#2 and #3, see Table 7.3) were manufactured entirely out
of glass fiber reinforced plastic. Bridges #4 and#S were manufactured to investigate the
performance of hybrid material systems: bridge #4 was composed of glass fiber inner
cells and a hybrid carbon and glass fiber outer shell (using three layers of equal thickness,
with a carbon layer between the two glass layers), whereas bridge #5 was composed of
ca¡bon fiber inner cells and a glass fiber outer shell. In all cases, the inner cells were
placed so that there was contact between adjacent cells.

The objectives of the tests were to measure the stiffrress and strength of the different
specimens. Bridge #3 was constructed to investigate its structural performance under
fatigue loading.

7.2.2 Experimental Procedure

The apparatus used for the tests of the six cell bridges is shown in Figure 7.6 and is
very similar to the apparatus used to test the two cell bridges. However, more
measurements were taken during the tests as described in Section 6, namely:

l. The deflections measured by the four LVDTs;

2. The displacement of the actuator;
3. The load applied by the actuator;
4. The strain measured by 14 strain gages.
 Draft - 9/24/01 Composite Bridge Superstructure 72

The bridges were simply supported by the long steel rollers located 0.5 in. from each end
of the bridge.

The testing protocol employed for the six cell bridges was the same as that used for
the two cell bridges. Three load cycles are used to measure the stiffrress of the bridge.
The first two cycles deformed the bridge in its elastic range by applying a load that
increased monotonically from 0 to 1,500 lbs. These two cycles were applied to eliminate
any looseness in the testing equipment. On the third load cycle, the specimen was loaded
until failure. A different set of loads were applied to bridge #3, since the objective of the
test on this bridge was to investigate its fatigue properties. This loading protocol is
described in the section below that presents the results of the fatigue test.

7.2.3 Experimental Results for fhe G/ass Bridges

Two all-glass bridges were tested: bridges #1 and #2. This section presents the
observations and measured data obtained from these tests.

7.2.3.1 Bridge #1

Figure 7.7 shows plots of load versus displacements measured at the actuator and the
four LVDT's for bridge #1. Early in the loading, there was some light cracking sounds
and some debonding at the shell-cell interface ( see Figure 7.8). A loud cracking sound
was heard at an applied load of about 2 kips. This was accompanied by the appearance of
cracks near the load application region as shown in Figure 7.9. Debonding between the

inner cells and the outer shell became visible at a load of about 3 kips, see Figure 7.10.
The bridge fails at a load of about 9 kips; this failure is due to the development of large
deflections and cracks in the compression side of the bridge near the load application
regions as shown in Figure 7.1 l.

7.2.3.2 Bridge #2

Figure 7.12 shows plots of load versus displacements measured at the actuator and the
four LVDT's for bridge #2. No cracking sounds or debonding at the cell-shell interface
were observed during the first 2 loading cycles (Figure 7.13).
 Draft -9/24/Ol Composite Bridge Superstructure 73

The first load cracking sound (accompanied by a small drop in the applied load)
occurred at a load of approximately 2.6 kips. The first visible cracks appeared near the
load application zone on the top surface of the bridge is shown in Figure 7.14. Failure of
the specimen occulred at a load of 7.8 kips; as with bridge #1, this was a compression
failure of the outer shell near the load application region. Figre 7.15 shows the state of the
bridge at the collapse load.

7.2.4 Experimenfal Resu/fs forthe G/ass-Carb on Bridges

Two hybrid glass-carbon bridges were tested: bridges #4 and #5. This section
presents the observations and measured data obtained from these tests.

7.2.4.1 Bridge tt4

Figure 7.16 shows plots of load versus displacements measured at the actuator and the
four LVDT's for bridge #4. Manufacturing defects were visible in this specimen before
the test was conducted. Figure 7.17 shows a crack that was observed inside one of the
inner cells. During the first load cycle, some cracking sounds were heard at a load of
about 800 lbs. A louder cracking sound was heard at a load of 1.4 kips during the final
load cycle. The first visible cracks are shown in Figure 7.18 and are located farther from
the load application zone compared to the cracks observed in bridges #1 and #2. At
higher loads, the inner cells experience significant deformation and cracking, see Figure
of the bridge experiences significant deformation as shown in
7.19. The top surface
Figure 7.20. Complete failure of the bridge occurs at a load if 5.2 kips and is
accompanied by cracks in the top surface, see Figure 7.21.

7.2.4.2 Bridge #5

Figure 7.22 shows the load-displacement plots obtained during the test of bridge #5.
Figure 7.23 shows the testing configuration for this specimen. As with bridge #4,
manufacturing defects were present prior to the test; Figure 7.24 shows longitundinal
cracks that were observed inside the ca¡bon-epoxy inner cells. No cracking noises were
heard during the first two loads cycles. During the final load cycle, minor cracking
sounds were heard at a load of 2,200Ibs. A major cracking noise was heard at 2,600 lbs.
Crack were observed on the top surface of the outer shell at a load of 2,900 lbs. Figure
 Draft - 9/24/01 Composite Bridge Superstructure 74

7.25 shows the deformation experienced by the inner cells near the load application
region during the application of the load. Figure 7.26 shows the crack that developed on
the surface of the outer shell. Complete collapse of the bridge occu¡red at a load of about
5 kips.

7.2.5 Experimenfal Resu/fs forthe G/ass Bridge Fatigue Test

Bridge #3 was built to the same specifications as bridge #2 (see Table 7.3), and was
used to examine the performance of the proposed bridge design under fatigue loads. Once

again, the bridge was simply-supported at end using the long steel rollers as seen in
Figure 7.27.However, the rollers were located at 1.0 in. for each end of the bridge to
reduce the chance that the specimen would slide off its supports during the long duration
fatigue test. Thus the unsupported length of bridge #3 was 58 in., compared with 59 in.
for the otherwise identical bndge#2.

The fatigue test was similar to a test performed on another fiberglass bridge [9]. The
loading was applied under displacement control conditions as a sine \¡/ave with a
frequency of 5 Hz. The minimum value of the applied displacement was the displacement
that corresponded to apreload of 50lbs. The maximum value of the applied displacement
was 0.090 in., which is equal to the maximum allowable displacement ofZl800 plus 20%
(i.e., I .2x L 1800 =1.2x0.075 in.). This applied load was similar to that used in [9] and is
a reasonable simulation of the actual loads that a full-size bridge may experience.

The performance of the bridge under fatigue loading was examined in two ways: by
measuring the stiffiress of the specimen before, during and after the application of the
three million loading cycles; and by measuring the failure load of the bridge after the
application of the three million loading cycles. These me¿rsurements were obtained in the
following manner.

The stiffrress of the specimen was measured at the start of the test and after one, two
and three million loading cycles in the following way. The load actuator was moved a
distance of 0.3 in. after application of the 50lbs. preload (this corresponded to an applied
of load of approximately 1.5 kips). This produced a linear elastic deformation in the
specimen, and enabled the stiffrress of the bridge to be measured using the various
LVDT's employed in the test apparatus. The fatigue test was intemrpted every million
Draft - 9/24/01 Composite Bridge Superstructure 75

cycles and the elastic stifftess measured in this manner. After completion of three million
cycles, the bridge stiffrress was measured and the specimen loaded to failure.

Figure 7.28 shows bridge #3 during the fatigue test; the displacement of the bridge is
clearly visible. No cracking sounds or debonding of the bridge components (see Figure
7.29) were perceived during the application of the cyclic loads and during the various
stiffness measurements. After the fatigue test was complete, the bridge was loaded until
failure. The failure load of the bridge after the application of three million loading cycles
was 7.6 kips. As with bridge #2, the collapse of the specimen was initiated by the
development of cracks in the compression side of the outer shell near the load application
regions as shown in Figure 7.30.

Figure 7.31 shows the measured deflections of bridge #3 after the application of three
million loading cycles. Figure 7.32 shows a comparison between the measured stifftesses
of the bridge before, during and after the fatigue loading. These data are discussed further
in Chapter 9.

7.2.6 Summary of the Experimenfal Resu/fs

Tables 7.4 and 7.5 summarizethe measured stiffüess and failure loads of the six cell
bridges. The stiffrresses were calculated by measuring the initially constant slopes of the
load-deflection plots obtained during the final loading of the specimens (i.e., the loading
that caused the specimen to fail). These values are expected to produce a reasonable
approximation to the linear stiffrress of the bridge specimens. The failure loads
correspond to the observed load carrying capacities of the various bridges, and are

expected to be significantly higher than the first ply failure loads.

In Chapter 9, the data presented in Tables 7.4 ad,7.5 are discussed and compared with
the predicted specimen stiffrresses and strengfhs computed using the finite element
models described in Chapter 8.
Draft - 9/24/01 Composite Bridge Superstructure 76

lnner lnner Cell Outer Shell

Name Cell Gap Material Material

Bridge A 0 in. glass glass

Bridge B 3/16 in. glass glass

Bridge C 1/16 in. glass glass

Bridge D 0 in. glass glass

Bridge E 0 in. carbon glass

Table 7.1: Summary of the two cell bridge specimens.

Draft - 9/24101 Composite Bridge Superstructure 77

First Ply
Name Stiffness Failure

Bridge A 3,610 tb/in

Bridge B 3,080 lb/in

Bridge C 3,260!b/in

Bridge D 3,430 lb/in 2,400 lbs

Bridge E 3,990 lb/in 1,100 lbs

Table 7.2: Measured stiffrresses and failure loads for the two cell bridges.
Draft - 9124/01 Composite Bridge Superstructure 78

lnner Cell Outer Shell

Name Length
Material I Angle Material I Angle
I I

Bridge #1 41in. glass t45' glass +30"

Bridge#2 60 in. glass +45" glass !45"

Bridge #3 60 in. glass !45" glass 145'

Bridge #4 60 in. glass +45" carbon/glass !45"

Bridge #5 60 in. carbon !45" glass +50'

Table 7.3: Summary of the six cell bridge specimens.

Draft - 9/24101 Composite Bridge Superstructure 79

Bridge #1 Bridge #2 Bridge #3 Bridge #4 Bridge #5

Actuator 5,823|b/in 4,335|b/in 4,792tblin 2,601 lb/in 4,740lblin

LVDT 1 8,805 tb/in 7,242lblin 8,550 tb/in 5,334 tb/in 7,5351b/in

LVDT 2 8,533|b/in 7,283 tblin 8,163 tb/in 5,846|b/in 7,Q47 tblin

LVDT 3 15,728 tblin 5,813|b/in _ 7,266lblin 4,768 tbtin 6,445 tblin

LVDT 4 17,604 tblin 6,052 lb/in 6,854 lb/in 6,407 tb/in

Table 7.4: Measured stiffrresses for the six cell bridges.

Draft - 9/24/01 Composite Bridge Superstructure 80

First Ply
Name Failure

Bridge #1 2,000|bs

Bridge #2 2,600 tbs

Bridge fÉ4 3,000 lbs

Bridge #5 2,200|bs

Table 7.5: Measured failure loads for the six cell bridges.
Draft -9/24/01 Composite Bridge Superstructure 81

Figure 7.1: Testing configuration for bridge A.

Draft - 9/24/01 Composite Bridge Superstructure 82

¡'l¡
',t-'í -i

:li:
:iit
i
.i t::
','f!
;Ì+,

Figure 7.2: Testing configuration for bridge D.

Draft -9/24/01 Composite Bridge Superstructure 83

i. r

Figure 7.3: Measured actuator load-deflection data for the two cell bridges.
Draft - 9/24/01 Composite Bridge Superstructure 84

Figure 7.4: Failure of bridge D.

Draft - 9/24/01 Composite Bridge Superstructure 85

Figure 7.5: Ultimate collapse of bridge A.

Draft -9/24/01 Composite Bridge Superstructure 86

Figure 7.6: Experimental apparatus used of the six cell bridges (bridge #1 shown).
Draft - 9124101 Composite Bridge Superstructure 87

Figure 7.7: Measured load-deflection data for bridge #1.

Draft -9/24/01 Composite Bridge Superstructure 88

Figure 7.8: Debonding between the inner cells and the outer shell of bridge #1.
Draft - 9124/01 Composite Bridge Superstructure 89

Ç*:---. ,:
--#
fu .tt
Figure 7.9: Initial cracking of the outer shell of bridge #1.
Draft -9/24/01 Composite Bridge Superstructure 90

Figure 7.10: Debonding at inner cell-outer shell interface of bridge #1.

Draft -9/24/01 Composite Bridge Superstructure 9l

:-ù.îr--,

Figure 7.11: Final failure of bridge #1.

Draft - 9/24/01 Composite Bridge Superstructure 92

Figure 7.12:Measured load-deflection data for bridge #2.

Draft - 9/24/01 Composite Bridge Superstructure 93

)'ffi| ffircffiffi f':

:ä'.il:

Figure 7.13: Maintenance of the bond between the inner cell and outer shell of bridge#2.
Draft -9/24/Ol Composite Bridge Superstructure 94

... ' ' -:..

.l'

Figure 7.14: Initiation of cracks in the outer shell of bridge #2.

Draft - 9/24/01 Composite Bridge Superstnrcture 95

t\:

1r¿
'Þ:l

r . - -.;.,-j

:ä:=:

Figure 7.15: Final failure of bridge #2.

Draft - 9/24/01 Composite Bridge Superstructure 96

Figure 7.16: Measured load-deflection data for bridge #4.

Draft - 9/24/01 Composite Bridge Superstructure 97

\ùìï,HryE
\l:þq
'\¡+¿,
Draft -9/24/01 Composite Bridge Superstructure 98

q
ï

-*;t
Figure 7.18: Initial cracking in the outer shell of bridge #4.
Draft -9/24101 Composite Bridge Superstructure 99

,ridffiiˡilfr4r I
;{ìi.l$¿i 'r¡ia!

rlJ€,E¡:.i:
"f-
'ri;rT . :i.:
'.
'::

Figure 7.19: Inner cell buckling and cracking in bridge #4.

Draft - 9124/01 Composite Bridge Superstructure 100

J{,

Figure 7.20 Deformation of the top surface of bridge *H.

 Draft - 9/24/01 Composite Bridge Superstructure 101

ë;= t.

":-
:ïÈ
:- l,\*
!!'iÞ
.f>:

: lliil
*,. ¡

----,- \l.t

Figure 7.21:Final failure of bridge *å4.

Draft - 9/24/01 Composite Bridge Superstructure 102

8000

7000

6000

5000
an
EI
Þ 4000
(u
o
3000

2000

I 000

0
0.0 0.5 1.0 1.5 2.0 25 3.0
Defieclim (h.)

Figure 7.22:Measured load-deflection data for bridge #5.

Draft - 9/24101 Composite Bridge Superstructure 103

Figure 7.23:Testing configuration for bridge #5.

Draft - 9/24/01 Composite Bridge Superstructure 104

Figure 7.24:Initial longitudinal cracks in inner cells of bridge #5.

Draft - 9/24/01 Composite Bridge Superstructure 105

_r
;::Ë:'r'' "_r
,i..,t F'
j -.".ì: ..
- ,itl

Figure 7.25: Localized deformation of bridge #5.

Draft -9/24101 Composite Bridge Superstructure 106

-'e

Figure 7.26: Outer shell cracks in bridge #5.

Draft - 9/24/Ol Composite Bridge Superstructure 107

Figure 7.27:Experimental apparatus employed for the fatigue test of brid ge #3.
Draft -9/24/01 Composite Bridge Superstructure 108

Figure 7.28: Bridge #3 during the fatigue test.

Draft -9/24/01 Composite Bridge Superstructure

the fatigue test ofbridge#3.

Draft -9/24/01 Composite Bridge Superstructure 110

¡\t ¡ ¡ U t-
I
Eø
I
@

Figure 7.30: Final failure of bridge #3.

Draft - 9124/01 Composite Bridge Superstructure 111

8000

7000

6000

5000
at
€
4000
ão
- Actuator
3000 - LVDTI (North)
- LVDT2 (South)
2000 - LVDT3 (East)
- LVDT4 (West)
1000

0
00 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Deffedion (in.)

Figure 7.31: Measured load-deflection data for bridge #3 after three million loading

cycles.
Draft - 9/24/01 Composite Bridge Superstructure tt2

10000

9000

8000

7000
1
L
6000
=
€
ø
u,
5000
(¡)
c
E
.E 4000
U)

3000

2000

1000

Numberof Cycles lxtO€)

Figure 7 .32: Measured stifûresses of brid ge #3 before, during and after the fatigue test.
 Draft - 9/24/01 Composite Bridge Superstructure 113

8. Finite Element Analysis of the Prototype Bridges

This chapter presents the methodology and the results of a series of finite element
analyses that were conducted in an attempt to predict the experimentally observed
specimen stiffrresses and strengths documented in Chapter 7. In all cases, the
commercially available analysis code Abaqus version 5.7 was used.

8.1 Two Cell Bridges

Finite element models were constructed of bridges D and E. These were the only two
cell bridges modeled since their construction were closest to those of the six cell bridges
(i.e., a REPEAT parameter equal to one and no gap between the inner cells).

Figure 8.1 shows the finite element mesh that was constructed to model bridges D
and E. It consisted of 6,283 nodes and 6,528 S4R5 four-node shell elements. Note that
nodes were placed at the locations of the four LVDT's employed during the specimen
tests. Also, the mesh was designed so that a contiguous patch of elements was placed
over the load application region. The glass and carbon fi,ber were assigned the linear
elastic material properties listed in Tables 8.1; these data were assembled form
information provided by the manufacturer and by independent tests. Table 8.2 lists the
lamina thicknesses and winding angles that were used in the inner cells and the outer
shells. The total thickness of each component was measured at several locations using a
caliper. The average values were then used I n the finite element model. The winding
angles were also measured directly form the manufactured specimens. The models of the

bridges are simply supported at the lower edge of each end, consistent with the roller
supports used in the tests. A uniform pressure is applied over a square surface located at
the center of the bridge and having the same dimensions as the square plate used to load
the specimens. The pressure was chosen to produce a total load of I lb. Note that this is
not an exact representation of the manner in which the loads were applied during the
tests. In the tests, the actuator was subjected to displacement control; thus the pressure

applied to a specimen through the loading plate would be expected to be non-uniform.

The model replaces the load application under displacement control with application
 Draft - 9/24/01 Composite Bridge Superstructure tt4

under load control. This would be expected to result in localized differences between the
physical and simulated deformation.

Figures 8.2 and 8.3 show the vertical displacements computed using the finite
element models of bridges D and E, respectively. Noting that these deflections were
computed using an applied load of unit magnitude enables the specimen stiffrress and ñrst
ply failure load to be calculated in the following way.

The actutator stiffrress was measured during the tests of the two cell bridges, and was
therefore computed using the results of the finite element analyses of the specimens. The
actuator stiffness w¿ts computed by dividing the applied load by an average of the vertical

displacements computed for the load application region. This vertical displacement was
calculated as the average vertical displacement of the four nodes located at the corners of
the load application region and the node located at the center of the load application
region.

First ply failure was estimated using the Tsai-Wu failure criterion, which predicts first
ply failure when the failure index

.=[ä)'[;)[;).(;)'.(#)' (l)

is equal to one. In Equation (l), ø, is the normal stress in the longitudinal (i.e., fiber)
direction, ø, is the normal stress in the transverse direction and o* is the in-plane shear
stress. The subscript [/denotes the ultimate values of these stresses.

Since the structure will deform linearly up to first ply failure, the results of the linear
analyses described above can be scaled according to the value of the failure index
computed at the unit applied load: the applied load is scaled so that it produces a
maximum failure index equal to one.

Table 8.3 shows the model stiffüess and first ply failure loads computed using the
results of the finite element analysis described above.
 Draft -9/24/01 Composite Bridge Superstructure ll5

8.2 Six Cell Bridges

Finite element models were constructed of bridges #1, #2, #4 and #5, i.e., all of the
six cell bridges except the specimen used in the fatigue test. This bridge (bridge #3) was
identical to bridge #2 with the exception that the supported length of bridge #3 was 2 in.
shorter that bridge #2.

Figures 8.4 and 8.5 show the finite element mesh constructed to model these four
bridges. The mesh used for the shorter bridge #1 consisted of 16,019 nodes and 17,056
S4R5 four-node shell elements. The meshes used for bridges #2,#4 and #5 were identical
and consisted of 23,353 nodes and 24,960 S4R5 four-node shell elements. As was done
with the meses for the two cell bridges, nodes were placed at the locations of the four
LVDT's employed during the specimen tests and the meshes were designed so that a
contiguous patch of elements was placed over the load application region. Table 8.4 lists
the lamina thicknesses and winding angles that were used in the inner cells and the outer
shells for each of the six cell bridges. The boundary conditions and applied loads were
similar to those used for the two cell bridges.

Figures 8.6, 8.7, 8.8 and 8.9 show the vertical displacements computed using the
finite element models of the four six cell bridges. Since these deflections were computed
using an applied load of unit magnitude, the specimen stiffüess and first ply failure load
were calculated using the same procedure employed for the two cell bridges. An
additional set of specimen stiffrress measurements were made during the experimental
testing of the six cell bridges, namely the LVDT stiffrress. The LVDT stiffrress was
calculated by dividing the unit applied load by the vertical displacements computed at the
nodes positioned at the locations of the LVDT's.

Tables 8.5 and 8.6 show the model stiffrress and first ply failure loads computed using
the results of the finite element analyses of the four six cell bridges.
Draft *9/24/01 Composite Bridge Superstructure 116

Material Property Symbol E-glass Epoxy Carbon Epoxy

Longitundinal Young's modulus EL

5.7 Msi
20 Msi

Transverse Young's modulus Eî

0.7 Msi 1.5 Msi

Poisson's ratio Drr

0.31 o.21

Shear modulus G,, 0.7 Msi 0.95 Msi

c'.t4
Longitudinal tensile strength 160 ksi 210 ksi

Longitudinal compressive strength d.t<

90 ksi 210 ksi

Transverse tensile strength d-,1 14 ksi 6.5 ksi

Transverse compressive stren gth ô,, 120 ksi 34 ksi

Shear strength ôE 12 ksi 9 ksi

Table 8.1: Material properties used in the finite element analysis.

Draft - 9/24/0I Composite Bridge Superstructure tt7

Bridge D Bridge E

Length (in.) 51 51

Unsupported 50 51
Length (in.)

lnner Cell
Winding Angle +145 +145
(degrees)

Outer Shell
Winding Angle +l-29 +l-29
(degrees)

lnner CellThickness
- Four Layers (in.) 0.0704 0.0680

Outer ShellThickness
- Six Layers (in.) 0.1110 0.1110

Table 8.2: Dimensions of the two cell bridges.

Draft -9/24/01 Composite Bridge Superstructure 118

Name Stiffness First Ply Failure

Bridge D 3,513!b/in 1,177 lbs

Bridge E 4,130|b/in 826 lbs

Table 8.3: Computed specimen stiffiresses and strengths for the two cell bridges.
Draft - 9/24/01 Composite Bridge Superstructure 119

Bridge #1 Bridge #2 Bridge #4 Bridge #5

Length (in.) 41 60 60 60

Unsupported 40 59 59 59
Length (in.)

lnner Cell
Winding Angle +145 +145 +145 +145
(degrees)

Outer Shell
Winding Angle +/-33 +/-50 +/-50 +/-50
(degrees)

lnner CellThickness
- Four Layers (in.) 0.0704 0.0732 0.0704 0.0680

Outer ShellThickness
- Six Layers (in.) 0.1122 0.1146 0.1158 0.1110

Table 8.4: Dimensions of the six cell bridges.

Draft - 9/24/01 Composite Bridge Superstructure t20

Bridge #1 Bridge #2 Bridge #4 Bridge #5

Actuator 7,019 lb/in 4,116 tb/in 4,574lb|in 5,620!b/in

LVDT 1 12,051 lb/in 7,039 lb/in 7,8591b/in 9,3601b/in

LVDT 2 12,051 lb/in 7,039lb/in 7,8591b/in 9,360|b/in

LVDT 3 23,596 !b/in 6,531 lb/in 7,423lblin 8,151 lb/in

LVDT 4 23,5961b/in 6,531 lb/in 7,423 tblin 8,151 tb/in

Table 8.5: Computed specimen stiffrresses for the six cell bridges.
Draft - 9124/01 Composite Bridge Superstructure t2l

First Ply Failure

Bridge #1 1,3l8lb

Bridge #2 1,358 tb

Bridge #4 1,433 lb

Bridge #5 989 rb

Table 8.6: Computed specimen strengths for the six cell bridges.
Draft -9/24/01 Composite Bridge Superstructure 122

Figure 8.1: Finite element mesh employed for the two cell bridges.
Draft - 9/24/01 Composite Bridge Superstructure t23

-:jt?€,.**
Vertical
Displacement
(¡n.)

@æu
Figure 8.2: Vertical displacements computed for bridge D.
Draft - 9/24/01 Composite Bridge Superstructure t24

. -;:ü|r',:--
'$i;-$,.$,
Vertical
Displacement
(in.)

,'ijâ.-
¿a;íl:,i-!::î.-.-
l!J;;;jl¡iiri,';q:;:r'-
"a|!i¡ii.7:;;i;i:¡i;,:,
'-!t-..i;!lolt:1''
_--¿r?r..

Figure 8.3: Vertical displacements computed for bridge E.

Draft - 9/24101 Composite Bridge Superstructure r25

Figure 8.4: Finite element mesh employed for the bridge #1.
Draft - 9/24/01 Composite Bridge Superstructure 126

Figure 8.5: Finite element mesh employed for bridges #2, #4 and #5.
Draft - 9/24/01 Composite Bridge Superstructure 1,27

Vertical
Displacement
(in.)

irit?:ï.i:
f.':i::'

Figure 8.6: Vertical displacements computed for bridge #1.

Draft - 9/24/01 Composite Bridge Superstructure 128

Vertical
Displacement
(in.)

Figure 8.7: Vertical displacements computed for bridge #2.

Draft - 9/24/Ol Composite Bridge Superstructure t29

Vertical
Displacement
(in.)

Figure 8.8: Vertical displacements computed for bridge tf4.

Draft - 9/24/01 Composite Bridge Superstructure r30

Vertical
Displacement
(in.)

Figure 8.9: Vertical displacements computed for bridge #5.

 Draft - 9/24/01 Composite Bridge Superstructure 131

9. Discussion of Result;*
The objectives of the research described in this report were to: (l) demonstrate that
the proposed bridge superstructure could be manufactured; (2) perform experiments to
measure the stiffness, strength and fatigue properties of small-scale fabricated specimens;

and (3) build finite element models to determine whether these models could be used to
predict the performance of the manufactured specimens. This chapter discusses the
results presented in the preceding chapters in the context of the project objectives.

9.1 Manufacture of the Proposed Bridge Superstructure

Chapters 4 and 5 presented the methodology adopted to manufacture small-scale

versions of the proposed bridge design. The physical size of the specimens was limited
by the available manufacturing equipment (i.e., the filament winder and the autoclave).
However, sufficient experience was gained to develop a repeatable manufacturing
process that was used to construct a number of specimens having different characteristics,

including length, number of cells, material type and winding angles. The basic process of
winding and curing individual inner cells followed by the winding of the outer shell over
the chosen number of inner cells is potential scalable to full-size structures. Special care
is required to ensure that the inner cells can be removed from the mandrel after curing
and that support is provided for the portion of the outer shell that is not directly attached

to the inner cells.

Two significant problems arose during the manufacture of the specimens that were
not fully solved in this project. First, the relatively small outer shell winding angle
employed for bridge #1 produced significant variability in the quality of the final part.
Specifically, the length of the filament winding machine coupled with the desire to
produce a bridge with at least a 2:1 length:width aspect ratio resulted in fiber placement
at the ends of the outer shell that was not consistent with the desired winding angle. This
problem would be solved by using a longer winder or by reducing the width of the
strucfure.

Second, the manufacture of hybrid bridges (i.e., bridges composed of glass and
carbon reinforced fiber) proved to be problematic. Tow approaches were tried: placing
 Draft - 9/24/01 Composite Bridge Superstructure t32

carbon fibers in the outer shell (bridge #4) and making the inner cells out of carbon
(bridges E and #5). The two hybrid six cell bridges (bridges *t4 atd#5) developed cracks

in the inner cells during the curing of the outer shell. The cause of this cracking was not
clea¡ and could have been due to thermal effects. However, the hybrid two cell bridge
(bridge E) did not appear to have any such defects. Further work would be required to
investigate this problem further before attempting to construct larger scale hybrid bridges.

9.2 Testing of the Fabricated Specr'mens

Chapters 6 and 7 describe the test procedures employed in this project and the results
that were obtained from the experiments. Five two-cell bridges were fabricated and tested
to refine the test procedures and investigate the effect of gaps between the inner cells.
The tests of the two-cell bridges demonstrated that it was not necessary to place end
supports in the inner cells to develop the full capacity of the bridges. This was apparent
by noting that the failure of the unsupported bridges did not involve local collapse of the
inner cells. Introduction of a gap between the inner cells resulted in a reduction of
specimen stiffüess; this reduction increased with the gap size. This effect was predicted in

[a] and may be attributed to the membrane stiffness of the inner cells. The hybrid two-
cell bridge was observed to have a higher stiffness than the glass fiber two-cell bridges;
however, the strength of the hybrid bridge was lower than the glass fiber bridges. All of
the two-cell bridges exhibited essentially linear behavior until their load capacity was
reached. The failure of the specimens was sudden and brittle in nature. Various cracking
noises were evident during the tests, and were sometimes accompanied by small
reductions in the applied loads.

Some of the data obtained from the testing of the two-cell bridges was used for
comparison with frnite element models of these specimens. This comparison is discussed
in Section 9.3.

Five six-cell bridges were manufactured and tested. The overall behavior of these
specimens was similar to that observed during the tests of the six-cell bridges. The
specimens behave linearly until failing in a sudden and brittle manner. Cracking noises
were perceived at various intermediate loads, together with reductions in the applied load.

Specimen stiffnesses were measured using the actuator instrumentation and four LVDT's
 Draft - 9/24/01 Composite Bridge Superstructure 133

placed on the top surface of the outer shell. Several strain gauges were also used;
however, the data obtained from these strain gauges did not produce any useful insights
into the behavior of the specimens and is therefore not presented in this report.

The six-cell bridges can be arranged into three groups: (1) glass fiber bridges used to
measure stiffrress and strength þridges #1 and #2); (2) a glass bridge used to investigate

the behavior of the design under cyclic loads (brid ge #3); and (3) hybrid bridges used to
measure stiffüess and strength (bridges lÍ4 and #5). The stiffrress and strength of bridges
#1,#2,#4 and #5 (i.e., the frist and last of these groups) are presented in Chapter 7 and
were used to investigate the validity of finite element models of these specimen (see
Section 9.3).

The fatigue test of bridge #3 demonstrated that the stiffrress and strength of the
proposed design was not effected by the application of 3 million cycles of an applied load

chosen to produce a 20Yo greater deflection than the AASHTO standard of Z/800.
Measurements of the specimen's stiffüess \Ã/ere taken at the start of the cyclic load test
and after every I million cycles; no reduction in stiffüess w¿ts evident. The strength of
bridge #3 at the end of the cyclic load test was observed to be similar to the strength of
bridge #2, which had identical nominal geometric and material properties.

9.3 Finite Element Analysis of the Manufactured Specimens

Chapter 8 describes the finite element analyses that were conducted to predict the
behavior of some of the specimens, namely bridges D, E, #1, #2, ll4 and #5. Tables 9.1,

9.2 and 9.3 present the errors between the computed and measured specimen stiffrresses
and first ply failures. In these tables, the error in a given stiffrress is defined as

êrt¡tr=Ffu
' (e.l)
Krn

where krru and ku, are the stiffrress computed using the finite element model and the
stiffness measured from the test, respectively. The error in a given first ply failure is
defined as

D
^
gED?
_rFEM
_ (e.2)
rDExP
 Draft -9/24/01 Composite Bridge Superstructure 134

where Prru and Pro are the first ply failure loads computed using the finite element
model and observed during the test, respectively.

The data presented in Tables g.l, g.2 and 9.3 show that the finite element models
were generally unsuccessful in predicting the performance of the various specimens.
However, there were some successes. The actuator stiffrress of the two-cell bridges
(bridges D and E) was successfully predicted by the finite element
models, as was all of
the stiffrress for bridge #2. The lack of agreement between the measured and predicted
stiffrress for the remaining six-cell bridges could be attributed to the
manufacturing
defects present in these three specimens (see Section 9.1). Local defects in
the winding of
the outer shell (bridge #1) and the development of crack in the inner tubes
during curing
(bridges ll4 and #5) could have resulted in a loss of stifûress.

The predictions of first ply failure were substantially lower than those observed
during the tests. It should be noted that the observed values were made based
on the
perception of audible cracking sounds, and did not always correlate with
any noticeable
change in the measured load-deflection response. Moreover, it is conceivable
that first
ply failure did initiate at loads lower than those reported and was simply not detected
due
to the lack of appropriate instrumentation. The ultimate load carrying capacity of all of
the specimens \il¿ts substantially greater than both the computed and observed first ply
failures.

9.4 Conclusions
The research documented in this report has demonstrated that small-scale prototypes
of the proposed bridge superstructure can be manufactured using basic filament winding
equipment. Care was required to ensure that the specimens were free of manufacturing
defects. Some specimens contained defects, but these were attributed to the limitations
of
the equipment available and unsolved problems that arise when different materials are
integrated into the same specimen. The structural tests of the manufactured specimens
demonstrated that the structural properties of the prototypes did not deteriorate when the
bridge was subjected to cyclic loads. The test also produced measurements of the
stiffrress and strength of the various specimens. These data were used in attempt to
Draft - 9/24101 Composite Bridge Superstructure 135

validate finite element models of the prototypes. The models were generally unsuccessful
in reproducing the measured structural properties. However, the finite element model did
accurately predict the specimen stiffüess when the specimen was free of manufacturing
defects. First ply failure was consistently under-estimated by the models; this may have
been due to the difficulties inherent in observing fist ply failure in complex structures.
The ultimate loads of the specimens were substantially greater than the first ply failure
loads predicted by the models or observed during the tests.

This research indicates that full-scale prototypes of the proposed design could be
manufactured using appropriate equipment. Finite element models could be expected to

accurately predict the stiffrress of defect-free specimens, but would tend to underestimate

the specimen failure loads. However, previous work [3,4] indicates that the design of this
type of structure would be govemed by stifûress considerations. Therefore, the finite
element models would be valuable in arriving at appropriate designs of the full-scale
prototypes. More sophisticated models that account for the material nonlinearity may
give further insight into the failure mechanisms at work.
Draft - 9/24/01 Composite Bridge Superstructure 136

First Ply
Stiffness Failure

Bridge D 2.3o/o 49.1o/o

Bridge E 3.4o/o 75j%

Table 9.1: Finite element analysis errors for the two-cell bridges.
Draft - 9/24/01 Composite Bridge Superstructure 137

Bridge #1 Bridge#2 Bridge #4 Bridge #5

Actuator 20.5% 5.0% 75.8o/o 18.6%

LVDT 1 36.9% 2.8% 47.3% 24.2o/o

LVDT 2 41.2% 3.4% 34.4% 32.8%

LVDT 3 50.0% 12.4% 55.7o/o 26.5o/o

LVDT 4 34.0% 7.9% 27.2o/o

Table 9.2: Finite element analysis errors for the stiffrresses of the six-cell bridges.
Draft - 9/24/01 Composite Bridge Superstructure 138

First Ply
Name
Failure

Bridge #1 65.9%

Bridge #2 52.2%

Bridge #4 47.6%

Bridge #5 45.íyo

Table 9.3: Finite element analysis errors for the first ply failure
of the six-cell bridges.
Draft - 9/24/01 Composite Bridge Superstructure 139

10. References

tll S. Alampalli,A. Yannotti and J. O'Connor. Extending life of bridges using FRP
composites. International Conference and Exhibition on Structural Faults and
Repair, London, July 1999.

t2l American Association of State Highway and Transportation Officials, AASHTO

LRFD Bridge Design Specifications. 1994.

t3] A. Aref and I. D. Parsons. Design optimization procedures for a fiber reinforced
plastic bridge. Journal of Engineering Mechanics, 125, ro40-l047,lggg.

t4l A. J. Aref and I. D. Parsons. Design and performance of a modula¡ fiber

reinforced plastic bridge. composites - Part B Engineering, 31, 619-629,2000.

t5l C. Dumlao et al. Demonstration low-cost modular composite highway bridge.

Proceedings of the First International Conference on Composites in
Infrastructure, saadatmanesh, H. and Ehsarri, M.R., editors, l14l-1155, 1996.

t6l Federal Highway Administration. National Bridge Inventory lgg1.

t7l W. Finch et al. Bridge rehabilitation using composite materials. lnfrastructure:

New Materials and Methods of Repaiç K. D. Basham, editor, l l35-l l40,lgg4.

t8l D. C. Foster, D. Richards and B. R. Bogner. Design and installation of fiber-

reinforced polymer composite bridge. Joumal of Composites for Construction 4,
33-37,2000.

tel Hayes, M. D. et al. Perfonnance of tube and plate fiberglass composite bridge
deck. Journal of Composites For Construction 4, 48-55, 2000.

ll0l P. R. Head. Advanced composites in civil engineering - a critical overview at

this high interest, low use stage of development. Advanced Composite Materials
Bridges and Structures, M. M. Elbadry, editor, 3-13,1996.
Draft -9/24/01 Composite Bridge Superstructure 140

[11] E. M. Lenoe. Attempts to introduce new materials into bridging. Infrastructure:

New Materials and Methods of Repair, K. D. Basham, editor, 64-71,1994.

[12] R. lopez-Anido et al. Development and demonstration of a modular FRP deck

for bridge construction and replacement. International Composites Expo 97
conference, Composites Institute 1997, l6-D.

[13] F. C. McCormick. Field study of a pedestrian bridge of reinforced plastic.

Transportation Research Record 1118, 1987.

[14] Plecnik, J. M. and Henriquez, O. Applications of composites in highway

bridges. Chapter 51 in Bridge Engineering Handbook, Chen, W.-F. and Duan,
L., editors. CRC Press, 1999.

[5] S. N. Sotiropoulos, H. V. S. GangaRao and A. N. Mongi. Theoretical and

experimental evaluation of FRP components and systems. Joumal of Structural

Engineerin g, I20, 1994.

[16] F. Seible. Advanced composites materials for bridges in the 2lst century.
Advanced Composite Materials Bridges and Structures, M.M. Elbadry, editor,
t7-40, t996.