Вы находитесь на странице: 1из 17

Composites: Part B 31 (2000) 593±609

www.elsevier.com/locate/compositesb

A systematic analysis and design approach for single-span FRP


deck/stringer bridges
Pizhong Qiao*, Julio F. Davalos, Brian Brown
Department of Civil and Environmental Engineering, West Virginia University, Morgantown, WV 26506-6103, USA

Abstract
There is a concern with worldwide deterioration of highway bridges, particularly reinforced concrete. The advantages of ®ber reinforced
plastic (FRP) composites over conventional materials motivate their use in highway bridges for rehabilitation and replacement of structures.
In this paper, a systematic approach for analysis and design of all FRP deck/stringer bridges is presented. The analyses of structural
components cover: (1) constituent materials and ply properties, (2) laminated panel engineering properties, (3) stringer stiffness properties,
and (4) apparent stiffnesses for composite cellular decks and their equivalent orthotropic material properties. To verify the accuracy of
orthotropic material properties, an actual deck is experimentally tested and analyzed by a ®nite element model. For design analysis of FRP
deck/stringer bridge systems, an approximate series solution for orthotropic plates, including ®rst-order shear deformation, is applied to
develop simpli®ed design equations, which account for load distribution factors under various loading cases. An FRP deck fabricated by
bonding side-by-side box beams is transversely attached to FRP wide-¯ange beams and tested as a deck/stringer bridge system. The bridge
systems are tested under static loads for various load conditions, and the experimental results are correlated with those by an approximate
series solution and a ®nite element model. The present simpli®ed design analysis procedures can be used to develop new ef®cient FRP
sections and to design FRP highway bridge decks and deck/stringer systems, as shown by an illustrative design example. q 2000 Elsevier
Science Ltd. All rights reserved.
Keywords: Approximate series solution

1. Introduction changes in the geometry of FRP shapes can be easily related


to changes in stiffness, changes in the material constituents
Fiber-reinforced plastic (FRP) composites are increas- do not lead to such obvious results. In addition, shear
ingly used in civil infrastructure. In recent years, attention deformations in pultruded FRP composite materials are
has been focused on FRP shapes as alternative bridge deck usually signi®cant and, therefore, the modeling of FRP
materials because of their high speci®c stiffness and structural components should account for shear effects.
strength, corrosion resistance, lightweight, and potential For applications to pedestrian and vehicular FRP bridges,
modular fabrication and installation that can lead to there is a need to develop simpli®ed design equations and
decreased ®eld assembly time and traf®c routing costs. procedures which should provide relatively accurate predic-
Despite the overall bene®ts of using FRP sections, they tions of bridge behavior and be easily implemented by
are as yet not widely used. Besides concerns with material practicing engineers.
costs, in comparison with conventional materials, the A number of theoretical and experimental investigations
complexity of material properties and structural behaviors have been conducted to study stiffness, strength, and
has deterred the rapid development of design codes for stability characteristics of FRP composite bridge decks.
practicing engineers. Henry [1] and Ahmad and Plecnik [2] examined the
A critical obstacle to widespread use and applications of performance of several glass reinforced polymer bridge
FRP structures in construction is the lack of simpli®ed and deck con®gurations using ®nite element analysis. In their
practical design guidelines. Unlike standard materials, FRP study, an FRP deck was modeled as a truss system in the
composites are typically orthotropic or anisotropic, and their direction perpendicular to the traf®c and as a beam system in
analyses are much more dif®cult. For example, while the direction parallel to the traf®c; their results indicated that
the design was always controlled by de¯ection limit state
rather than strength limit states. Later, FRP decks were
* Corresponding author. Present address: Department of Civil Engineer- fabricated using a combination of ®lament winding and
ing, The University of Akron, Akron, OH 44325-3905, USA. hand lay-up processes, and the static and fatigue behaviors
1359-8368/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved.
PII: S 1359-836 8(99)00044-X
594 P. Qiao et al. / Composites: Part B 31 (2000) 593±609

highway bridge materials. An overview of the current status


on research and applications of ®ber-reinforced polymeric
bridge decks has been presented by Zureick [11].
Although several experimental and numerical efforts
have been conducted, there is no simpli®ed design protocol
available for FRP composite bridges. The design procedures
developed for bridges composed of isotropic materials [12]
cannot be directly applied; they need to be modi®ed to
incorporate the anisotropy of composite materials. Thus a
simple but accurate solution for the analysis and design of
FRP composite bridges is needed. This solution should
account for the geometry and material properties of FRP
decks and stringers; and it should provide reasonably
accurate predictions for performance and load distribution
of the system.
In this paper, a systematic approach for analysis and
design of all FRP deck/stringer bridges is presented. This
approach (Fig. 1) is based on analyses at micro-level
(material), macro-level (structural component) and system
level (structure) to design all FRP deck/stringer bridge
systems. First, based on manufacturer's information
and material lay-up, ply properties are predicted by
Fig. 1. Systematic analysis protocol for FRP bridge systems.
micromechanics. Once the ply stiffnesses are obtained,
macromechanics is applied to compute the panel mech-
of decks were determined experimentally [3]; from the anical properties. Beam or stringer stiffness properties are
experimental observation, it was evident that damage then evaluated from mechanics of thin-walled laminated
under fatigue loading consisted primarily of interfacial beams (MLB). Using elastic equivalence, apparent stiff-
delamination initiation and local buckling of thin delam- nesses for composite cellular decks are formulated in
inated layers under compressive service loads. Bakeri and terms of panel and single-cell beam stiffness properties,
Sunder [4] used balanced symmetrical laminate to study the and their equivalent orthotropic material properties are
feasibility of two FRP bridge deck systems: a deck solely further obtained. For design analysis of FRP deck/stringer
made of a glass-reinforced polymer and a deck with hybrid bridge systems, an approximate series solution for ®rst-
materials of a glass-reinforced polymer, a carbon-reinforced order shear deformation orthotropic plate theory is applied
polymer and a light-weight concrete; they concluded that to develop simpli®ed design equations, which account for
the hybrid system concept achieved less de¯ection and had a load distribution factors for various load cases. As illu-
promising future in infrastructure applications. With the strated in Fig. 1, the present systemic approach, which
objective of developing an FRP bridge deck system, accounts for the microstructure of composite materials and
Mongi [5] tested three full-scale ¯oor systems which differ geometric orthotropy of a deck system, can be used to
in their size, joint type, and loading conditions. GangaRao design and optimize ef®cient FRP deck and deck/stringer
and Sotiropoulos [6] assembled and tested two FRP bridge systems.
superstructure systems consisting of bridge decks and To introduce the systematic approach shown in Fig. 1,
stringers; a simpli®ed ®nite-element model, in which this paper is organized in the following three main sections:
equivalent plates were used as a substitute for the stringers (1) panel and beam analyses by micro/macromechanics and
and the cellular deck, was used to correlate results with mechanics of thin-walled laminated beams, (2) FRP cellular
experimental data. Burside et al. [7] presented a design decks by elastic equivalence analysis, and (3) analysis of
optimization of an all-composite bridge deck; cellular-box deck/stringer system by an approximate series solution
and stiffened-box geometries were optimized with consid- technique. To verify the accuracy of the equivalent ortho-
eration of de¯ection and buckling. Most recently, two high- tropic material properties, a multi-box-beam deck fabricated
way bridges were constructed with a modular FRP by bonding side-by-side box FRP beams is experimentally
composite deck in West Virginia [8]: one was built as an tested and analyzed by a ®nite element model. To validate
all-composite short-span deck/stringer system and another the approximate series solution, the multi-box-beam deck is
one was constructed with a modular FRP deck supported by attached to FRP wide-¯ange (WF) beams and tested and
steel stingers. Fatigue and failure characteristics of a modu- analyzed as a deck/stringer bridge system. The box beams
lar deck were investigated by Lopez-Anido et al. [9] and a for decks and WF beams for stringers were both produced
satisfactory performance was observed. Several projects by pultrusion. Both deck and bridge systems are tested
[10] also demonstrated the potential of FRP composites as under static loads for various load conditions. To
P. Qiao et al. / Composites: Part B 31 (2000) 593±609 595

Fig. 2. Microstructure and dimensions of a FRP box beam section.

illustrate the systematic design procedures developed in material architectures that can be simulated as laminated
this paper, an example of an FRP deck/stringer bridge is con®gurations. A typical pultruded section mainly includes
presented. the following three types of layer [13] (see Figs. 2 and 3):
(1) continuous strand mats (CSM); (2) stitched fabrics (SF);
and (3) rovings or unidirectional ®ber bundles. Each layer is
2. Panel and beam analyses
modeled as an homogeneous, linearly elastic, and generally
orthotropic material. Based on information provided by the
Extensive research has been conducted in the area of
manufacturer, the ®ber volume fraction (Vf) can be
analysis and design of composite materials at micro- and
evaluated and used to compute the ply stiffnesses from
macro-levels. The analysis of FRP beams from micro/
micromechanics models [14]. For the box section of Fig. 2
macromechanics to beam response has been presented in
and the wide-¯ange section of Fig. 3, the predicted ply
Ref. [13]. In this section, the analyses of micro/macro-
properties are given in Tables 1 and 2, respectively. Once
structure and beam component are brie¯y reviewed and
the ply stiffnesses for each laminate (panel) of a FRP beam
include: (1) constituent materials and prediction of ply
are computed, the stiffnesses of a laminated panel can be
properties, (2) laminated panel engineering properties, and
computed from macromechanics [15]. For example, for
(3) beam or stringer stiffness properties.
the box beam shown in Fig. 2, the panel properties
2.1. Panel analysis by micro/macromechanics (Ex, Ey, nxy and Gxy) predicted by the micro/macro-
mechanics models [14] correlate well with experimental
Although pultruded FRP shapes are not laminated results for coupon samples (Table 3) tested in tension
structures in a rigorous sense, they are pultruded with and torsion.

Fig. 3. Panel ®ber architecture of a wide ¯ange beam.


596 P. Qiao et al. / Composites: Part B 31 (2000) 593±609

Table 1
Ply material properties of a box section (Fig. 2) computed by a micromechanics model [14]

Lamina E1 ( £ 10 6 psi) E2 ( £ 10 6 psi) v12 G12 ( £ 10 6 psi)

1/2 oz CSM 2.093 2.093 0.407 0.744


1 oz CSM 1.710 1.710 0.402 0.610
15.5 oz 908 SF 4.118 1.183 0.389 0.457
12 oz 1/2458 SF 3.505 1.056 0.396 0.405
61 yield roving 8.469 3.374 0.343 1.429

2.2. Beam analysis by mechanics of laminated beams bridge systems, as presented in Section 4, and they can
also serve to simplify modeling procedures either in numer-
The response of FRP shapes in bending is evaluated using ical or explicit formulations. The design equations neces-
the mechanics of thin-walled laminated beams (MLB) [16]. sary for such a model are presented in this section, along
In MLB, the stiffness coef®cients (axial, A; bending, D; with numerical and experimental veri®cation of the results.
axial-bending coupling, B; and shear, F) of a beam are In this section, the development of equivalent stiffness for
computed by adding the contributions of the stiffnesses of cellular decks consisting of multiple FRP box beams is
the component panels, which in turn are obtained from the presented. Multicell box sections are commonly used in
effective beam moduli. Based on MLB, engineering design deck construction because of their light-weight, ef®cient
equations for FRP beams under bending have been geometry, and inherent stiffness in ¯exure and torsion.
formulated [17], and they can be easily adopted by Also, this type of deck has the advantage of being relatively
practicing engineers and composite manufacturers for the easy to build. It can be either assembled from individual
analysis, design, and optimization of structural FRP beams box-beams or manufactured as a complete section by pultru-
or bridge stringers. MLB is suitable for straight FRP beams sion or a vacuum-assisted resin transfer molding process.
or columns with at least one axis of geometric symmetry The elastic equivalence approach [20] used in this paper
and can be used to evaluate the stiffness properties and accounts for out-of-plane shear effects, and the results for
response of bridge stringers. As an example, the bending a multicell box section are veri®ed experimentally and by
(D) and shear (F) stiffnesses of a box beam (Fig. 2) and a ®nite element analyses.
wide-¯ange beam (Fig. 3) by MLB are listed in Table 4, and
experimental results for de¯ections and strains compared 3.1. Equivalent stiffness for cellular FRP decks
favorably with MLB predictions [13,18,19].
The panel and beam properties obtained above by micro/ As an illustrative example, we derive the bending, shear
macromechanics and MLB can be ef®ciently implemented and torsional equivalent stiffnesses for a deck composed of
in deck and deck/stringer system designs, as described in multiple box sections (Fig. 4).
Sections 3 and 4.
3.1.1. Longitudinal stiffnesses of cellular FRP deck
The bending stiffness of the deck in the longitudinal
3. FRP cellular decks: elastic equivalence
direction, or x-axis in Fig. 4, is expressed as the sum of
A multicellular FRP composite bridge deck can be the bending stiffness of individual box beams (Db, see
modeled as an orthotropic plate, with equivalent stiffnesses Table 4):
that account for the size, shape and constituent materials of Dx ˆ nc Db …1†
the cellular deck. Thus, the complexity of material aniso-
tropy of the panels and structural orthotropy of the deck where nc ˆ number of cells. For the section shown in Fig. 4,
system can be reduced to an equivalent orthotropic plate b ˆ width of a cell, h ˆ height of a cell, tf ˆ thickness of the
with global elastic properties in two orthogonal directions: ¯ange, and tw ˆ thickness of the web. If all panels have
parallel and transverse to the longitudinal axis of the deck identical material lay-up and tf ˆ tw ˆ t, Eq. (1) becomes
cell. These equivalent orthotropic plate properties can be   ht
directly used in the design and analysis of deck/stringer Dx ˆ nc Ex h2 1 3bh …2†
6
Table 2
Ply material properties of a wide-¯ange section (Fig. 3) computed by a micromechanics model [14]

Lamina E1 ( £ 10 6 psi) E2 ( £ 10 6 psi) v12 G12 ( £ 10 6 psi)

3/4 oz CSM 1.710 1.710 0.402 0.610


17.7 oz 1/2458 SF 4.157 1.191 0.294 0.460
62 yield roving 6.732 2.077 0.278 0.826
P. Qiao et al. / Composites: Part B 31 (2000) 593±609 597

Table 3
Panel properties of a box section (4 00 £ 8 00 £ 1/4 00 )

Ex Ey vxy Gxy
6 a 6 a a
Experimental 3.293 £ 10 psi 2.491 £ 10 psi 0.269 8.599 £ 10 5 psi b
Micro/macromechanics 3.377 £ 10 6 psi 2.620 £ 10 6 psi 0.285 8.760 £ 10 5 psi
%Difference 12.6% 15.2% 15.9% 11.9%

a
From tension tests.
b
From torsion tests.

where Ex ˆ modulus of elasticity of a panel in the x-direc- where the moments of inertia I are de®ned as:
tion computed by micro/macromechanics or obtained ÿ 
wtf3 w 2tw 3
experimentally (Table 3). If ˆ ; Iw ˆ …7†
The out-of-plane shear stiffness of the deck in the long- 12 12
itudinal direction, Fx, is expressed as a function of the For tf ˆ tw ˆ t, Eq. (6) can be simpli®ed as
stiffness for the individual beams (Fb):
2E wt3
F x ˆ nc F b …3† Fy ˆ  y  …8†
h
b b1
where Fb is given in Table 4, and nc ˆ number of cells. This 4
expression can be further approximated in terms of the in- where Ey is the modulus of elasticity of a panel in the
plane shear modulus of the panel, Gxy (see Table 3) and y-direction (Table 3).
cross-sectional area of the beam webs:
Fx ˆ nc Gxy …2t†h: …4† 3.1.3. Torsional stiffness of cellular FRP deck
The torsional rigidity of a multi-cell section, GJ, is
evaluated by considering the shear ¯ow around the cross-
3.1.2. Transverse stiffnesses of cellular FRP deck section of a multi-cell deck. For a structure where the webs
An approximate value for the deck bending stiffness in and ¯anges are small compared with the overall dimensions
the transverse direction, Dy, may be obtained by neglecting of the section, Cusens and Pama [21] have shown that the
the effect of the transverse diaphragms and the second torsional rigidity may be written as
moment of area of the ¯anges about their own centroids. 4A2 Gxy X t3
For a deck as shown in Fig. 4 with tf ˆ t: GJ ˆ 1 Gxy …ds† …9†
P ds 3
1 t
Dy ˆ E …w†…t†h2 …5†
2 y where A ˆ area of the deck section including the void area
P
where w is the length of the deck in the longitudinal and is de®ned as A ˆ nc bh, and ds=t represents the
direction and Ey is the modulus of elasticity of the panel summation of the length-to-thickness ratio taken around
in the y-direction (Table 3). the median line of the outside contour of the deck cross-
For multiple box sections, the simplest way to obtain the section. For a constant panel thickness t, the torsional
deck's out-of-plane transverse shear stiffness is to treat the rigidity can be simpli®ed as
structure as a Vierendeel frame in the transverse direction ÿ 
2 nc bh 2 Gxy t 2ÿ 
[21]. For the Vierendeel frame (Fig. 5), the in¯ection points GJ ˆ ÿ  1 nc b 1 h Gxy t3 …10†
are assumed at the midway of top and bottom ¯anges nc b 1 h 3
between the webs. The shear stiffness in the transverse The above approximate equation is justi®ed by the fact that
direction, Fy, for the cross-section shown in Fig. 4 may be for a multi-cell deck, the net shear ¯ows through interior
written as webs are negligible and only the shear ¯ows around the
V 12Ey outer webs and top and bottom ¯anges are signi®cant. The
Fy ˆ ˆ   …6† second term in Eq. (10) is relatively small compared with
u h b
b 1 the ®rst term and can be ignored.
Iw 2If
If the deck is treated as an equivalent orthotropic plate, its
Table 4 torsional rigidities depend upon the twist in two orthogonal
Strong-axis beam bending and shear stiffness coef®cients by MLB directions. Thus, torsional stiffness Dxy may be taken as one-
half of the total torsional rigidity given by Eq. (10) divided
Beam stiffness Db (lb/in 2 2 in 4) Fb (lb/in 2 2 in 2)
by the total width of the deck:
Box 4 00 £ 8 00 £ 1=4 00 1:795 £ 108 3:474 £ 106
GJ
WF 12 00 £ 12 00 £ 1=2 00 1:706 £ 109 5:026 £ 106 Dxy ˆ …11†
2nc b
598 P. Qiao et al. / Composites: Part B 31 (2000) 593±609

Fig. 4. Geometric parameters of a multi-cell box deck.

Substituting Eq. (10) into Eq. (11) and neglecting the second 3.2. Veri®cation of deck stiffness equations by ®nite element
term in Eq. (10), we get: analysis
nc Gxy bh2 t The formulae for bending and torsional stiffnesses
Dxy ˆ ÿ  …12†
nc b 1 h obtained in Section 3.1 are based on the assumption that
the deck system behaves as a beam and does not account
where D xy is the torsional stiffness per unit width for the Poisson effects of the deck. To verify the accuracy of
(lb 2 in 3 /in 2 ). the above deck stiffness equations, a ®nite element analysis
of the deck system is performed. The model is shown in Fig.
4 and consists of box beams (Fig. 2) bonded side-by-side to
form an integral deck. The computer program NISA [22] is
used, and the panels are modeled with 8-node isoparametric
layered shell elements. The cellular decks subjected to line-
loading for longitudinally and transversely supported
conditions are shown in Figs. 6 and 7, and the model for
torsional loading is given in Fig. 8.

3.2.1. Veri®cation of bending and shear stiffnesses


The deck bending and shear stiffnesses in the longitudinal
and transverse directions are used to evaluate midspan
de¯ections from the following

PL3 PL
d3 ˆ 1 …3-point bending† …13†
48Di 4 k Fi

23PL3 PL
d4 ˆ 1 …4-point bending† …14†
1296Di 6kFi
where P ˆ total applied load, L ˆ span length, k ˆ shear
correction factor (k ù 1:0 is assumed in the analysis), and
Di and Fi ˆ bending and shear stiffness (i ˆ x for longi-
tudinal or y for transverse directions). The de¯ections by
Eqs. (13) and (14) in terms of stiffness properties are
compared with results from the ®nite element model for
actual cellular systems under line loading (Figs. 6 and 7).
For the longitudinal stiffness veri®cation, the length of the
decks is kept constant (L ˆ 108 in), and the de¯ection in
terms of bending and shear stiffnesses is a function of the
Fig. 5. Vierendeel distortion in a multi-cell box-beam [21]. number of cells. Each deck is simply supported and
P. Qiao et al. / Composites: Part B 31 (2000) 593±609 599

Fig. 7. FE model for veri®cation of transverse bending and shear stiffnesses


equations.

displacements and rotations in all three principal directions


and all three rotations, and the other end subjected to a
uniform torque. The longitudinal torsional rigidity of a
Fig. 6. FE model for veri®cation of longitudinal bending and shear deck is expressed in terms of the angle of twist f and the
stiffnesses equations. torque applied at the end of the section as
TL
subjected to either 3-point or 4-point bending due to GJ ˆ …15†
f
uniformly distributed line loads. The comparisons between
the predictions of Eqs. (13) and (14) based on simpli®ed where T ˆ 2qnc bh (as shown in Fig. 8) is the applied torque.
stiffness formulae and the ®nite element results for actual The specimen length L is held constant (L ˆ 108 in), and
decks are presented graphically in Fig. 9. the number of cells is used as the design variable. The ®nite
Similarly, the midspan de¯ections in the transverse element results are compared with the theoretical predic-
direction are found by modeling several multicellular tions of Eq. (10), and the results are presented in Fig. 11.
decks comprised of 4 00 £ 8 00 £ 1/4 00 box sections (Fig. 7).
For these models, the width (w) is kept constant 3.2.3. Comparisons and remarks
(w ˆ 12 in), and the de¯ection is a function of the number As shown in Figs. 9±11, a good correlation is obtained
of cells. The model is simply supported and subjected to between the theoretical predictions based on the simpli®ed
either 3-point or 4-point bending due to uniformly distri- stiffness formulas and the ®nite element analyses of an
buted line loads. The results of the ®nite element models and actual deck. For the de¯ection in terms of longitudinal stiff-
theoretical predictions are shown in Fig. 10. nesses (Dx and Fx), the maximum percent difference is 4%,
and for the de¯ection in terms of transverse stiffnesses (Dy
3.2.2. Veri®cation of torsional stiffness of the deck and Fy), the maximum difference is about 10%. For the
The simpli®ed formula for the torsional rigidity, GJ, of longitudinal torsional stiffness, the discrepancy of results
the deck system was also veri®ed using ®nite element increases steadily from 6% for one cell to 22% for 15
analyses, which indirectly serve to verify the torsional stiff- cells. Some limited experimental data available for one
ness of the deck (Dxy). The model shown in Fig. 8 consisted and two cells [23] match closely the analytical results.
of a multicellular deck with one end ®xed, by constraining The favorable de¯ection comparisons between beam
600 P. Qiao et al. / Composites: Part B 31 (2000) 593±609

Fig. 10. Transverse central de¯ection of a multi-cell deck (Fig. 7).

Fig. 8. FE Model for veri®cation of the torsional rigidity equation.


deck can further simplify the design analysis of deck and
deck/stringer bridge systems.
equations and ®nite element results indirectly verify the To calculate the moduli of elasticity …Ex †p and …Ey †p for
accuracy of the deck bending stiffness equations. Similarly, the equivalent orthotropic plate, the relationship D ˆ EI is
the torsion results indicate that the simpli®ed torsional stiff- used, leading to
ness equations are acceptable for practical applications.
Therefore, the proposed relatively simple stiffness equations ÿ  D  
Ex p ˆ 12 3 x 1 2 nxy nyx …16†
account for both shape and material anisotropy of the deck t p bp
and can be used with relative con®dence in design analysis
of cellular bridge deck systems.
  Dy  
Ey ˆ 12 3 1 2 nxy nyx …17†
p tp lp
3.3. Equivalent orthotropic material properties
where the subscript ªpº indicates property related to
Once the stiffness properties of an actual deck are the equivalent orthotropic plate; tp ˆ thickness of the
obtained, it is a simple matter to calculate effective material plate ( ˆ h for the actual deck, Fig. 4), bp ˆ width of
properties for an equivalent orthotropic plate. To obtain the the plate ( ˆ ncb for the actual deck), and lp ˆ length
equivalent orthotropic plate material properties for an actual of the plate ( ˆ w for the actual deck). The Poisson's

Fig. 9. Longitudinal central de¯ection of a multi-cell deck (Fig. 6). Fig. 11. Torsional rigidity vs number of cells (Fig. 8).
P. Qiao et al. / Composites: Part B 31 (2000) 593±609 601

Fig. 12. Experimental setup of a multi-cell box deck (5 0 £ 10 0 £ 8 00 ).

ratios n ij are de®ned as Finally, to calculate the in-plane shear modulus …Gxy †p , we
2 1j use
nij ˆ …18†  
1i Dxy
Gxy ˆ 6 3 …22†
where 1 is the strain in the i or j direction. For orthotropic
p tp
materials, the Poisson's ratio must obey the following With these equivalent material properties, it is now easy to
relationship: use explicit plate solutions (see Section 4) for analysis and
nij nji nxy nyx design of cellular decks.
ˆ or ˆ …19†
Ei Ej Dx Dy
3.4. Experimental and numerical veri®cation of equivalent
In this study, we use the approximation nxy ˆ 0:3, which is orthotropic material properties
typically used for pultruded composites.
To calculate the out-of-plane shear moduli …Gxz †p and To indirectly verify the accuracy of the equivalent ortho-
…Gyz †p , the relationship F ˆ GA is used, leading to tropic material properties given in Eqs. (16)±(22) (see
Section 3.3), a multi-box-beam deck of 5 0 £ 9 0 £ 8 00 (Fig.
ÿ  F
Gxz p ˆ x …20† 12a) subjected to a patch load is tested and analyzed for
t p bp three load conditions: (1) at the center of the deck, (2)
at 16 00 to one side from the center along the line AA 0 , and
  Fy
Gyz ˆ …21† (3) at 16 00 to the other side from the center along the line
p tp lp AA 0 .
602 P. Qiao et al. / Composites: Part B 31 (2000) 593±609

Fig. 13. FE simulation and de¯ection contour of a multi-cell box deck under symmetric loading.

The ®nite element program NISA [22] is used to conduct 3.4.1. Experimental details
two distinct analyses: (1) the actual deck (Fig. 13) is The test sample was fabricated by bonding FRP box
modeled using 8-node isoparametric layered shell elements beams side-by-side with epoxy [24]. For each load condi-
and the material properties of Table 1; (2) an equivalent tion, the displacements are recorded at several locations
solid orthotropic plate of the same global dimensions as with LVDTs (see Fig. 12b), and the strains in the longi-
the actual deck is modeled using the same elements tudinal and transverse directions are obtained at three
and the equivalent material properties computed from locations by bonding 350 ohm strain gages at the bottom
Eqs. (16)±(22) and given in Table 5. The experimental of the deck (Fig. 12c). Note that from Fig. 12, for the
results and correlations with ®nite element analyses are asymmetric load cases 2 and 3, the following displacement
presented next. values should be approximately equal: d 1 and d 5, and d 2 and

Table 5
Equivalent deck stiffness properties and orthotropic material properties for cellular deck 5 0 £ 9 0 £ 8 00

Dx (lb 2 in 4/in 2) Dy (lb 2 in 4/in 2) n xy Dxy (lb 2 in 4/in 2) Fx (lb 2 in 2/in 2) Fy (lb 2 in 2/in 2)

2:689 £ 109 2:250 £ 109 0.3 1:153 £ 107 4:896 £ 107 3:662 £ 105
Ex (psi) Ey (psi) n yx Gxy (psi) Gxz (psi) Gyz (psi)
5 5 5 5
9:713 £ 10 4:515 £ 10 0.25 1:351 £ 10 1:020 £ 10 4:238 £ 102
P. Qiao et al. / Composites: Part B 31 (2000) 593±609 603

Table 6 functions, wheel load distribution factors are derived, which


Experimental and ®nite element comparison for a multi-cell box deck under are used later to provide design guidelines for deck/stringer
load case 1 (centric)
bridge systems. Finally, the approximate series solution is
Parameter Experiment FE (actual deck) FE equivalent plate veri®ed by testing a 10 0 £ 10 0 £ 8 00 multi-box-beam deck
supported by WF 12 00 £ 12 00 £ 1=2 00 FRP beams; this system
d 1 (in/kips) 0.00721 0.00627 0.00602
is also analyzed by ®nite element model [22].
d 2 (in/kips) 0.00971 0.00861 0.00947
d 3 (in/kips) 0.02036 0.01644 0.01900
d 4 (in/kips) 0.00964 0.00861 0.00947 4.1. First-order shear deformation theory for FRP
d 5 (in/kips) 0.00710 0.00625 0.00603 composite deck
d 6 (in/kips) 0.01544 0.01033 0.01242
d 7 (in/kips) 0.01421 0.01033 0.01242
A ®rst-order shear deformation theory [25] is applied to
1 1 (m1 /kips) 28.111 28.463 31.752
1 2 (m1 /kips) 214.701 25.832 27.813 analyze the behavior of a geometrically orthotropic FRP
1 3 (m1 /kips) 68.584 61.043 73.539 composite deck. Instead of direct modeling of the actual
1 4 (m1 /kips) 215.266 226.522 223.643 deck geometry, an equivalent orthotropic plate, as discussed
1 5 (m1 /kips) 28.361 28.463 31.752 in Section 3, is used to simplify the analysis. The formulae
1 6 (m1 /kips) 212.966 25.832 27.813
for equivalent orthotropic material properties accounting for
deck geometry and panel laminated material properties are
d 4. Similarly, the following strains should correspond to given in Section 3. The equilibrium equations accounting
each other: 1 1 and 1 5, and 1 2 and 1 6. for ®rst-order shear deformation of an orthotropic plate are:
   
2 2w 0 2 2w 0
3.4.2. Experimental results and correlation A55 cx 1 1 A44 cy 1 1 q…x; y† ˆ 0
2x 2x 2y 2y
Comparisons of experimental results and ®nite element
analyses for FRP deck under centric loading (load case 1)    
and asymmetric loading (load cases 2 and 3) are shown in 2 2c 2cy 2 2cx 2cy
D11 x 1 D12 1 D66 1
Tables 6 and 7, respectively. Tables 6 and 7 indicate that the 2x 2x 2y 2y 2y 2x
measured displacements and strains compare relatively well  
2w0
with FE models of actual deck and equivalent plate for both 2 A55 cx 1 ˆ0 …23†
2x
the symmetric loading (case 1) and asymmetric loading
(cases 2 and 3). For the symmetric load case (case 1), the    
difference of de¯ection (d 3) under load point between the 2 2cx 2cy 2 2c 2cy
D66 1 1 D12 x 1 D22
experiment and the FE equivalent plate model is about 2x 2y 2x 2y 2x 2y
6.5%; whereas there are 6.7% difference for longitudinal  
2w0
strain (1 3) at the center of the deck (Fig. 12). As noted in 2 A44 cy 1 ˆ0
Table 7, values for the asymmetric loading (cases 2 and 3) 2y
also compare favorably between the average experimental where Aij (i; j ˆ 4; 5) are the intralaminar shear stiffnesses,
data and FE equivalent plate model when the measurements and Dij (i; j ˆ 1; 2; 6) are the bending stiffnesses for an
were close to the applied load; the differences are about orthotropic material.
0.9% for de¯ection and 5.0% for longitudinal strain under The deck/stringer bridge system can be ®rst analyzed as
applied load. The good correlation between experimental an orthotropic plate stiffened by edge stringers (or beams)
data and FE models validates the orthotropic plate material [26]. Then, the contributions of interior stringers are
properties obtained by elastic equivalence analysis, which accounted for in the formulation by considering the inter-
can be then be used in the analysis of FRP deck/stringer action forces and the comparability conditions along rib
systems. lines between the deck and stringers. The analysis is general
with respect to: (1) size and stiffness of the deck, and (2)
type of loading (uniform and/or concentrated). The formul-
4. Analysis of FRP deck/stringer bridge system ation is concerned ®rst with symmetric and antisymmetric
loading conditions.
The equivalent properties for cellular decks and
stiffnesses for FRP beams can be ef®ciently used to analyze
and design deck/stringer systems. In this section, we present 4.1.1. System under symmetric loading case
an overview of a series solution for stiffened orthotropic A Fourier polynomial series is employed to obtain the
plates based on ®rst-order shear deformation theory and solutions for the equilibrium equations [Eq. (23)]. The
transverse interaction of forces between the deck and the solution for a symmetric loading is
stringers. The solutions for symmetric and antisymmetric X
1
load cases are used to obtain the solution for asymmetric w0 …x; y† ˆ Wij sin ax…sin by 1 W0 †
loading. Based on deck-stringer transverse interaction force i;jˆ1
604 P. Qiao et al. / Composites: Part B 31 (2000) 593±609

X
1

FE equivalent plate
cx …x; y† ˆ Xij cos ax…sin by 1 X0 † …24†
i;jˆ1

20.00117
0.00252
0.00901
0.02255
0.02104
0.00632
0.00632

30.010

84.803
8.327

226.792
20.692

26.142
X
1
cy …x; y† ˆ Yij sin ax cos by
i;jˆ1

where a ˆ ip=a and b ˆ jp=b, and Wij, Xij, and Yij are the
coef®cients to be determined to complete the solution.
Note that these series approximations satisfy the essential
FE actual deck

boundary conditions. The generalized loading can be


0.00094
0.00352
0.00860
0.01940
0.01729
0.00610
0.00566
written as the following in®nite double series
22.176

25.086
12.826

28.528

70.076
229.444
X
1
q…x; y† ˆ Qij sin ax sin by: …25†
i;jˆ1

Qij are the Fourier coef®cients in the representation of the


load q(x,y). By substituting the general solution Eqs. (24)
Average of load cases 2 and 3

and (25) into Eq. (23) and reducing by orthogonality condi-


tions [12,23], we obtain the following system of equations
for any number of terms (i,j):
2 38 9 8 9
K11 K12 K13 > > Wij >
> >
> Qij >
>
7< = < =
20.02801
0.00356
0.00993
0.02235
0.01883
0.00399
0.00815

6
6 K21 K22 K23 7 Xij ˆ
24.201

28.579

81.666
219.956

226.379

230.770

4 5> 0 …26†
> >
> > >
: ; > : >
;
K13 K23 K33 Yij 0
where Kij are the deck stiffness coef®cients (for a symmetric
Experimental and ®nite element comparison for a multi-cell box deck under load cases 2 and 3 (asymmetric)

loading) [24]. For a one-term approximation, the constants


W0 and X0 are obtained by satisfying the boundary
Experiment Load Case 3

conditions of the edge-stiffened orthotropic plate [Fig.


14(b)]:
 
Y
20.05410
0.00412
0.01091
0.02330
0.01901
0.00996
0.00872

W0 ˆ A44 c 11 1 b
23.750
223.297
29.289
231.206
83.287
234.467

W11
 
A44 Y11 1 bW11
X0 ˆ 2 …27†
a3 D X11
where
 
Experiment load case 2

1 1 1
cˆ 2 1 2 ;
a kF aD
k ˆ the stringer shear correction factor, and F and D are,
20.00191
0.00299
0.00895
0.02139
0.01865
0.00798
0.00757
24.652
216.614
27.868
221.552
80.044
227.072

respectively, the shear and bending stiffnesses of the


stringer and are obtained based on mechanics of laminated
beams (MLB) (Table 4) [16]. For any interior stringer at any
location r (r ˆ 0; 1; ¼; n) [see Fig. 14(c)], the generalized
de¯ection function for any symmetric loading is [24]
   
1 1 1 px pr
Parameter load cases 2 and 3

wR …x; r† ˆ R11 2 1 2 sin sin 1 W0


a kF aD a n
…28†
1 1 and 1 5 (m1 /kips)
1 2 and 1 6 (m1 /kips)
1 3 and 1 3 (m1 /kips)
1 4 and 1 4 (m1 /kips)
1 5 and 1 1 (m1 /kips)
1 6 and 1 2 (m1 /kips)
d 1 and d 5 (in/kips)
d 2 and d 4 (in/kips)
d 3 and d 3 (in/kips)
d 4 and d 2 (in/kips)
d 5 and d 1 (in/kips)
d 6 and d 6 (in/kips)
d 7 and d 7 (in/kips)

where
Q
R11 ˆ   11  :
Table 7

1 1 1 Q11 n 4W0
1 2 1 11
a2 kF a D W11 b p
P. Qiao et al. / Composites: Part B 31 (2000) 593±609 605

Fig. 14. Deck/stringer bridge system.

4.1.2. System under antisymmetric loading case 4.1.3. System under asymmetric loading case
Analogous to the symmetric case, Eqs. (24) and (25) are The asymmetric case is obtained by superposition of the
modi®ed for a ®rst-term approximation of an antisymmetric symmetric and antisymmetric load conditions. By simply
loading as adding the symmetric and antisymmetric responses, the
   generalized de¯ection function for an interior stringer
2y
w0 …x; y† ˆ W12 sin ax sin 2by 1 W1 1 2 under an asymmetric load is written as
b
wR …x; r† ˆ
  
2y
cx …x; y† ˆ X12 cos ax sin 2by 1 X1 1 2 …29† " ! !#
b
pr 2pr pr
R11 sin 1 W0 1 R12 sin 1 W1 cos
cy …x; y† ˆ Y12 sin ax cos 2by n n n

!
q…x; y† ˆ Q12 sin ax sin 2by 1 1 1 px
 2 1 2 sin (32)
By substituting Eq. (29) into Eq. (23), we obtain the stiffness a kF aD a
matrix for an orthotropic deck under antisymmetric loading
[24]. The constants W1 and X1 are determined as
0 1 4.2. Wheel load distribution factors
 
Y B 1 C The above solution is used to de®ne wheel-load
W1 ˆ A44 c 12 1 2b B @
C
A
W12 2 distribution factors for any of the stringers. The load
1 1 A44 c
b distribution factor for any interior stringer ith is de®ned as
  the ratio of the interaction forces R(x,r) for the ith stringer to
A44 Y12 1 2bW12 the sum of interaction forces for all stringers. The general
X1 ˆ 2 …30†
a3 D X12 expressions of load distribution factors in terms of the
number of stringers m (where, m ˆ n 1 1) for symmetric
where c is the same as for Eq. (27). The generalized and asymmetric loads [24] are, respectively
de¯ection function for antisymmetric loading is
  r21
1 1 1 sin p 1 W0
wR …x; r† ˆ R12 2 1 2 WfSym …r† ˆ m21
a kF aD 2
…m 2 1† 1 mW0
p
 
px 2p r pr
 sin sin 1 W1 cos …31† WfAsym …r† ˆ
a n n
     
where r21 r21 r21
R11 sinp 1 W0 1 R12 sin2p 1 W1 1 2 2
m21 m21 m21
m      
Q X r21 r21 r21
R12 ˆ   12  : R11 sinp 1 W0 1 R12 sin2p 1 W1 1 2 2
1 1 1 Q12 4n 2W1 rˆ1
m21 m21 m21
1 2 1 11
a2 kF a D W12 b 3p …33†
606 P. Qiao et al. / Composites: Part B 31 (2000) 593±609

Fig. 15. FE simulation and de¯ection contour of a deck/stringer system under symmetric load.

4.3. Design guidelines load case as


ÿ 
Pd ˆ Pe NL Wf max …35†
Based on the wheel distribution factors obtained above,
the number of stringers necessary for a given bridge deck where NL is the number of lanes and (Wf)max is found from
can be determined. The dimensions of the deck are used Eq. (33) as a function of number of stringers m. Two design
to evaluate the maximum allowable moment per lane criteria based on the performance of stringers and deck can
(Mmax) according to AASHTO [27]. Then an equivalent be used to design the system.
concentrated load (Pe) is calculated as [23,24]
4.3.1. Design criterion based on performance of stringer
4Mmax
Pe ˆ …34† The midspan de¯ection d LL of a stringer is evaluated as
L !
L3 L
where L is the length of a stringer (span of the bridge). dLL ˆ Pd 1 …1 1 DLA† …36†
48D 4k F
The equivalent deck properties (Section 3) and the bend-
ing and shear stiffnesses (D and F) for a given type of where DLA is the dynamic load allowance factor, and for
stringer (Section 2) are then used to calculate the edge short-span bridges DLA ù 0:2 [26]. Eq. (36) is then set
de¯ection coef®cient W0 and/or W1. Next, a design load equal to the maximum allowable de¯ection (from AASHTO
(Pd) is de®ned either for the symmetric or asymmetric [27]) to determine the number of stringers required for
P. Qiao et al. / Composites: Part B 31 (2000) 593±609 607

Fig. 16. Comparison for a deck/stringer bridge system.

the bridge deck. Once a suitable system is chosen, the 4.3.2. Design criterion based on performance of deck
maximum moment due to live load (MLL) is calculated from Excessive local deck deformation and punching-shear
failure may be observed in FRP bridge applications. Thus
Pd L it is necessary in the design process to check the local deck
MLL ˆ …1 1 DLA† …37†
4 de¯ection and bending and shear stresses in a deck section
between two adjacent stringers [28,29]. Further research is
Finally, the approximate maximum extreme ®ber normal needed to address these issues.
stress (s c) in the stringer can be found from
4.4. Experimental testing and numerical analysis of FRP
MLL y 0
sc ˆ …38† deck/stringer systems
I
To validate the approximate series solution presented
where y 0 is the distance from the neutral axis of the stringer above, an FRP deck (10 0 £ 10 0 £ 8 00 ) is fabricated by bond-
to the top surface of the stringer and I is the moment of ing side-by-side box beams of 4 00 £ 8 00 £ 1=4 00 (Fig. 2); the
inertia of the stringer. This stress can then be compared deck is attached to FRP I-beams 12 00 £ 12 00 £ 1=2 00 (Fig. 3)
with the material compressive or tensile strength to con®rm and tested and analyzed as a deck/stringer bridge system.
that the system will be effective. Also, as an approximation, The deck with either three or four stringers is subjected to
shear stress in the stringer can be estimated as: various static load conditions [24]. The ®nite element model
with NISA [22] is shown in Fig. 15 for a 3-stringer system
Pd under a concentrated centric loading. The comparisons
tˆ …1 1 DLA† …39†
2Aw among the FE, series solution and experiments for both
3-stringer and 4-stringer systems under centric loading are
where Aw is the area of the web panels. The shear stress shown in Fig. 16, and relatively consistent trends are
in Eq. (39) should be less than the shear strength of the observed. The maximum differences of stringer de¯ec-
stringer. tions between experiments and series approximation are
608 P. Qiao et al. / Composites: Part B 31 (2000) 593±609

Fig. 17. Dimensions and panel ®ber architectures of an optimized winged-box beam [30].

about 16% for a 3-stringer system and 23% for a 4- 5. check the stress level on the stringers;
stringer system. The detailed study on the experimental 6. check the local stresses, de¯ections, and other details.
program and comparisons for various cases can be found
in Ref. [24]. 5.2. Design example

As an illustration, a single-lane short-span bridge of


15 ft width and 25 ft span is designed using side-by-side
5. Design analysis procedures and illustrative example
4 00 £ 8 00 £ 1=4 00 bonded FRP box sections for the cellular
General guidelines for the applications of the above series deck assembly and optimized FRP winged-box beams
approximation solution for design analysis of FRP 12 00 £ 24 00 [30] for the stringers (Fig. 17). The material prop-
composite deck-and-stringer bridge systems and several erties for the stringers are: D ˆ 1:248 £ 1010 lb-in4 =in2
illustrative design examples are given in Ref. [24]. and F ˆ 1:940 £ 107 lb-in2 =in2 , which are computed by
micro/macromechanics and MLB. The de¯ection limit of
L/500 and the loading of AASHTO HS-20 [27] are consid-
5.1. General design procedures ered, and the number of stringers (m) is used as a design
variable. The edge de¯ection coef®cient for symmetric
The following step-by-step design procedures are loading is evaluated from Eq. (27) as W0 ˆ 1:691, and
recommended: the de¯ection limit is written as a function of the
1. de®ne bridge dimensions and allowable loads; number of stringers m:
2. obtain deck panel and bridge stringer properties by micro/ 0 1
macromechanics [13±15] and mechanics of laminated  
L 4 B 1 1 W0 C
beams (MLB) [16]; dLL ˆ ˆ Mmax NL B
@
C
500 L 2 A
3. determine deck equivalent material properties by the mW0 1 …m 2 1†
p
equations derived in Section 3;
4. perform the series approximation analysis and determine !
the number of stringers m based on the required L3 48 L
 1 …1 1 DLA† …40†
de¯ection limit; D 4k F
P. Qiao et al. / Composites: Part B 31 (2000) 593±609 609

where NL ˆ 1:0 and k ˆ 1:0. In this example, the dynamic Fatigue and failure evaluation of modular FRP composite bridge
load allowance DLA ˆ 0:20 and AASHTO lane-moment deck. Proceedings of Int Composites Expo 0 98, Composites Institute,
1998. p. 4-B (1±6).
M max ˆ 207:4 kip-ft are used. Solving for the number of [10] Richards D, Dumlao C, Henderson M, Foster D. Method of installa-
winged box beams (stringers) (m) required for this single- tion and the structural analysis of two short span composite highway
span bridge, we get m ˆ 6:15, and therefore, m ˆ 7 is bridges. Proceedings of Int Composites Expo 0 98, Composites
used, which corresponds to 30 in center-to-center spacing Institute, 1998. p. 4-E (1±6).
of seven longitudinal stringers. The maximum stress in [11] Zureick A. Fiber-reinforced polymeric bridge decks. Seminar Note,
National Seminar on Advanced Composite Material Bridges, FHWY,
the stringer becomes sc ˆ 1:48 ksi, which is below the Washington DC, 1997.
allowable stress of 21.2 ksi [30]. [12] Salim HA, Davalos JF, GangaRao HVS, Raju P. An approximate
series solution for design of deck-and-stringer bridges. Int J of
Engineering Analysis 1995;2:15±31.
6. Conclusions [13] Davalos JF, Salim HA, Qiao P, Lopez-Anido R, Barbero EJ. Analysis
and design of pultruded FRP shapes under bending. Composites, Part
As described in this paper, a systematic approach for B: Engineering J 1996;27(3-4):295±305.
design analysis of FRP deck/stringer bridge systems is [14] Luciano R, Barbero EJ. Formulas for the stiffness of composites
with periodic microstructure. Int J of Solids and Structures
proposed, and the constitutive material properties and
1994;31(21):2933±44.
micro/macrostructure of a composite system are accounted [15] Barbero EJ. Introduction to composite materials design. Philadelphia,
for in the design. This design approach (Fig. 1) includes the PA: Taylor & Francis, 1999.
analyses of ply (micromechanics), panel (macromechanics), [16] Barbero EJ, Lopez-Anido R, Davalos JF. On the mechanics of thin-
beam or stringer (mechanics of laminated beam), deck walled laminated composite beams. J Composite Materials
1993;27(8):806±29.
(elastic equivalence model), and ®nally combined deck/
[17] Davalos JF, Qiao P, Barbero EJ, Troutman D, Galagedera L. Design
stringer system (series approximation technique). This of FRP beams in engineering practice. Proceedings of Int Composites
relatively simple and systematic concept accounts for the Expo 0 98, Composites Institute, 1998. p. 12-E (1±6).
complexity of composite materials and geometry of the [18] Salim HA, Davalos JF, Qiao P, Barbero EJ. Experimental and
bridge system. The approximate series solution, which analytical evaluation of laminated composite box beams. Proc of
40th Int SAMPE Symposium 1995;40(1):532±9.
is used to obtain wheel load distribution factors for
[19] Davalos JF, Qiao P, Barbero EJ. Multiobjective material architecture
symmetric and asymmetric loading, is an ef®cient way optimization of pultruded FRP I-beams. Composite Structures
to analyze and design single-span FRP deck/stringer 1996;35:271±81.
systems. The present design analysis approach can be [20] Troitsky MS. Orthotropic bridges, theory and design. Cleveland,
ef®ciently used to design bridge systems and also Ohio: The James F. Lincoln ARC Welding Foundation, 1987.
[21] Cusen AR, Pama RP. Bridge deck analysis. John Wiley & Sons, 1975.
develop new design concepts for single-span FRP
[22] Numerically Integrated Elements for System Analysis (NISA), Users
deck/stringer bridges. Manual, Version 94.0, Troy, MI: Engineering Mechanics Research
Corp, 1994.
[23] Salim HA. Modeling and application of thin-walled composite beams
References in bending and torsion. PhD Dissertation, West Virginia University,
Morgantown, WV, 1997.
[1] Henry JA. Deck girders system for highway bridges using ®ber rein- [24] Brown B. Experimental and analytical study of FRP deck-and-
forced plastics. MS Thesis, North Carolina State University, 1985. stringer short-span bridges. MS Thesis, West Virginia University.
[2] Ahmad SH, Plecnik JM. Transfer of composite technology to design Morgantown, WV, 1998.
and construction of bridges. US DOT Report, September 1989. [25] Reddy JN. Energy and variational methods in applied mechanics.
[3] Plecnik JM, Azar WA. Structural components, highway bridge deck New York: John Wiley, 1984.
applications. In: Lee I, Stuart M, editors. International encyclopedia [26] Salim HA, Davalos JF, Qiao P, Kiger SA. Analysis and design of ®ber
of composites, vol. 6, 1991. p. 430±45. reinforced plastic composite deck-and-stringer bridges. Composite
[4] Bakeri B, Sunder SS. Concepts for hybrid FRP bridge deck system. Structures 1997;38:295±307.
Proceedings of 1st Materials Engineering Congress, ASCE, vol. 2. [27] Standard Speci®cations for Highway Bridges, The American Associa-
Denver, CO, 1990. p. 1006±14. tion of State Highway and Transportation Of®cials (AASHTO).
[5] Mongi ANK. Theoretical and experimental behavior of FRP ¯oor Washington DC, 1989.
system. MS Thesis, West Virginia University, Morgantown, WV, 1991. [28] Davalos JF, Salim HA. Effective ¯ange-width of stress-laminated
[6] GangaRao HVS, Sotiropoulos SN. Development of FRP bridge T-system timber bridges. J Structural Engineering, ASCE
superstructural systems. US DOT Report, June 1991. 1993;119(3):938±53.
[7] Burnside P, Barbero EJ, Davalos JF, GangaRao HVS. Design [29] Davalos JF, Salim HA. Local deck effects in stress-laminated T-
optimization of an all-FRP bridge. Proceedings of 38th Int SAMPE system timber bridges. Int J of Structural Engineering Review
Symposium, 1993. 1995;5(1):1143±53.
[8] Lopez-Anido R, Troutman DL, Busel JP. Fabrication and installation [30] Qiao P, Davalos JF, Barbero EJ. Design optimization of ®ber-
of modular FRP composite bridge deck. Proceedings of Int reinforced plastic composite shapes. J of Composite Materials
Composites Expo 0 98, Composites Institute, 1998. p. 4-A (1±6). 1998;32(2):177±96.
[9] Lopez-Anido R, Howdyshell PA, Stephenson LD, GangaRao HVS.

Вам также может понравиться