Professional Documents
Culture Documents
Feasibility Study of
Standard Sections for
Segmental Prestressed
Concrete Box Girder Bridges
Felix Kulka
Consulting Engineer
(At time study was conducted,
Mr. Kulka was President of
T. Y. Lin International,
San Francisco, California)
S. J. Thoman
Structural Engineer
T. Y Lin International
San Francisco, California
54
Synopsis
Presents the highlights of a study which investigated
the feasibility of developing standard sections for
segmental prestressed concrete box girder bridges.
The report is based on an extensive survey of
segmental box girder bridges in the United States and
Canada. Recommendations are given for specific items
that could be standardized, while also discussing areas
which might not be appropriate to standardization.
pand their economical use by instilling specific areas which should be standar-
confidence among bridge engineers dized are listed and discussed in the
and by producing a cost effectiveness report, as are those which are not cur-
through uniformity in design, thus rently subject to standardization and
permitting precasters and contractors to those which are questionable.
invest in forms and equipment on a
broader basis than is done today. Scope of Study
This report deals with the feasibility
of standardizing segmental prestressed Standardization of highway construc-
concrete box girder bridges in the tion elements is a long-standing prac-
United States. The study relied heavily tice in the American highway industry.
on a survey of bridge engineers in the Development of the AASHTO-PCI I-
United States and Canada, which pro- girders is one example; precast con-
duced valuable information on all crete culverts, traffic barriers, and piles
bridges of this type. Statistical studies are other examples. It is fairly well
were conducted to determine correla- agreed that standardization has merits
tions and uniformity of significant pa- in cost savings, reduction of construc-
rameters, particularly with respect to tion time, and improved product qual-
geometry. ity.
Analytical design studies, mainly to It was felt that for standardization of
determine the economical use of mate- box girder sections to succeed, a uni-
rials, were made to augment the statis- form approach should be used in order
tical analyses. The results were to permit bridge engineers to design
evaluated both qualitatively and quan- such sections with a sufficient degree of
titatively, and an advisory technical re- uniformity and to allow precasters and
view committee was formed to review contractors to bid and build them as
the content of the study and its recom- they would any other advanced type of
mendations. structure.
The report takes the position that The object of this study, then, was to
standardization of segmental pre- consider all the advantages and disad-
stressed concrete box girder bridges is vantages of standardization and make
possible and should be initiated. The appropriate recommendations for future
56
i d
0 ^ qç/
C
/
Hawaiian < a_
Islands.'----- -
rte ] - _.
'I
f I ^
Koror Babelthaup
Puerto Rico
construction was not uniform, owing The conditions surrounding the pres-
mostly to a wide variety of site condi- ent state of segmental box girder con-
tions, design practices, specifications struction raise the obvious question of
and bidding requirements. The 1980 standardizing at least some aspects of
dollar cost per square foot of bridge its design and detailing. Ideally, stan-
deck of some 37 segmentally con- dardization could bring about cost ben-
structed box girder bridges appears to efits by permitting contractors to invest
vary widely from $30 to $150 (8323 to in forms, installations and equipment
$1615/m 2 ). Nevertheless, sufficient which could be reused more often, thus
cases of successful and economical con- reducing the cost of mobilization. De-
struction exist to make the segmental tails and joinery could he simplified in
box girder a very viable choice in the the process of standardization, and
concrete bridge market. overall safety and integrity could he
15 15
cn
L
m
0 10
L
0
E
7
z
O V] G Lo 0 0
a' (0 P- ti CO CO
r r r r r
Bid Year
58
in progress or completed
35 1- has not been bid or was not selected
30 _Q
Q. 3 V
o° 25 s ° m
U 7
C 61 :s Cl
20 a ca C G
w a ro
U l a U V
>
15 c m I -n to
m m m W
E U U
C
z 10 o
Construction Method
Fig. 3. Frequency of various construction methods.
d
0)
3- 0.
a)
m r c m I
O
c2 a ro as ao
02
0)— } Q U
fb _
O c m
d N
E 0 U U
C G
IIIII t0 ^C
ay 1 v a m M
a N . p
J
C.
0)
Construction Method
Fig. 4. Total square footage of bridge deck for various construction methods.
placing represent only one bridge each, equivalent 40- ft (12 m) width, and the
hence, cannot necessarily be consid- resulting total bridge length was then
ered representative. divided by the number of particular
The bridge deck area for the various bridges, thus obtaining an average
methods of construction is shown in length of bridge for each construction
Fig, 4. It may be seen that span-by-span type. The numbers show that the aver-
construction encompasses about 30 age length for span-by-span construc-
percent of the total bridge deck area, tion is about 40 percent larger than for
but as shown in Fig. 3, represents only balanced cantilever. In other words, it
about 10 percent of the total number of
bridges. That indicates that this method Table 1. Average bridge lengths for
obviously was applied to very large various construction methods.
projects. In balanced cantilever con-
struction the total bridge deck area is Construction Average length for a
about equally divided between cast- method 40-ft (12 m) roadway
in-place and precast structures as de-
signed, but the proportion of bridges Incremental
actually constructed to the total designed launching 1087 ft (331 m)
is higher for the cast-in-place structure. Progressive
placing 1165 if (355 in)
The project size greatly influences
the method of construction to he used. Span-by-span 5347 ft (1630 in)
Table 1 shows the average length of Balanced
segmental bridges surveyed. Here the cantilever
total deck area of bridges .for each type (precast) 3133 ft (955 m)
(cast in place) 2818 ft (850 rn)
of construction was normalized to an
60
in
emental launching
progressive placing
v A
0
4A 0.30m=1.Oft
AAA
s
pan-byspan
0
•
o M• 0
.• •i • •
C • • •0"• •
U
balanced cantilever, PC
■
• ■ ••■■ ■■
••. ■■ ■ ■■ ME ■ ■ ■ ■
-4^ balanced cantilever, cip !^
100 200 300 400 500 600 700 800
(ft)
Span Length
Fig. 5. Span ranges of box girder bridges for various construction methods.
200
o a
C 150 •• a °
• C
UI
0— 100 •• •• U r
o o
a Q • •• •^■ e^ ,,
tO
50 • m• • c 0.09m2-1.Osf
ar ♦ A '' mE 0.30m=1.Oft
a - A
0 a C
V
Construction Method
Fig. 6. Cost of segmental prestressed concrete box girder bridges for different
construction methods.
0 20
a
E
Z 10
Cross-Sectional Configuration
Fig. 7. Frequency of various cross-sectional configurations of box girder bridges.
2.5
0.30m3/m2=1.Ocy/sf
0.30m =1 .Oft
wing section
2.0
1.5
m
nC
0
U .. rder
w
m 7• 1.0
E V
v
7
O
J
0.5
50 100 150
62
210
1-cell
2-cells ---
190
0.30m2/m =1.0sf/ft
0.30m= 1.0ft
d 160 /
o //
0
- 130 £(
a / A
100 /
•• Ar_` W= 70
U
70 W - 50•
c tEEEE W - 30^
40
ii
1ii
10 20 30 40
(ft)
Girder Depth
Fig. 9. Internal surface forming area for various deck widths and girder depths.
takes a larger project with many short Fig. 6 shows the costs of these
spans, as for example a causeway, for bridges. It may be seen that there is no
this method to be economical as com- obvious uniformity to be discerned
pared to other structural sections. from these cost figures, Partially, the
The distribution of construction reason for this is the fact that accurate
methods for segmental bridges with re- costs are very difficult to establish. First
spect to span length is shown in Fig. 5. of all, cost figures are not readily avail-
It can be noted that the balanced can- able; secondly, when they are available
tilever method was used primarily for it is not totally clear what the costs
spans greater than 200 ft (61 m). The cover. However, costs do vary widely
span-by-span method was used for principally as the result of lack in uni-
spans between 80 and 180 ft (24 and 55 formity of design and construction
m) as were the incremental launching practices. In any event, the figures
and progressive placing methods. might demonstrate qualitatively the fact
For spans longer than 450 ft (137 m), that costs of the bridges varied consid-
only cast-in-place segments were erably within the construction method
employed, most likely because of the itself, in addition to differences be-
increased weight of precast segments tween the various construction
needed for long spans. methods.
W=70'
10
m /
U
/
0
// W 5O8
E
5 ^y W=30'
o
>
0.30 m= 1.Oft
0.09m3/m = 1.Ocy /ft
10 20 30 40
(t t)
Girder Depth
Fig. 10. Volume of concrete for various deck widths and girder depths.
100
1 -cell
2-cells -----
$0 4.45N =1.01b
0.30m -1.0ft
c /
m 60
of
a "
40 J(
ò /
20
30 40 (ft) 50 60 70
Width of Deck
Fig. 11. Weight of transverse prestressing steel for various deck widths.
64
correlation = 0.85
A cast-in-place = 0.87
• precast = 0.70
0.30m = i.Oft
3.0
W
T 1D
A
WT A
2.0
I-
3:
I-
m..
mv
r ^' r-
•
aL ..A •
a
AA
0.003L+0.026 (A)
L
•
Q
1.0
•
... 0.002L + 0.024 (•)
3:
.-:;-:r_A
....:r ST ♦
3.0
l^ v
SW /
(A) 0.005L-0.654
AL
2.0
t
(D
E A'
I- AA
&
• •
• ♦
0
0 1.0 • • ♦0.003L +0.234 (•)
A
• •
• ♦
N f
The weight of mild reinforcing steel It can therefore be deduced that the
is less for the single cell section than single cell section is more economical
for the double cell section for all sec- than the multiple section in all aspects,
tion widths. This, again, is reasonable, except for the transverse prestressing
since much of the mild steel is nominal steel. This is true up to a width of ap-
reinforcing and the loads are carried proximately 70 ft (21 m), at which point
largely by the post-tensioning tendons. twin single cells should be considered.
The internal surface forming area is
considerably less in the single cell sec-
tion, which translates into great econ- Statistical Studies of
omy for formwork. The elimination of
interior webs also produces a more con- Dimension Parameters
structable section. The required In order to determine the degree of
amount of transverse post-tensioning is, uniformity in dimensions, parameters
of course, higher for the single-cell in the transverse and longitudinal di-
section than for the double-cell section. rection were studied statistically for the
66
3•1- CL
correlation = 0.60
0.30m =1 .Oft
m
m 7
> 2.0 • •
• • • 0.064CL + 0.81 4
1.0
0
U
O
5 10 15 20
(ft)
Length of Cantilever
Fig. 14. Variation of cantilever deck thickness with cantilever length.
bridges surveyed. Linear regression span length is shown in Fig. 12. The
curves were fitted through the data correlation coefficient was 0.85 when
points using a least square criterion. combined and 0.87 and 0.70 when
Correlation coefficients were calcu- studied independently for cast-in-place
lated to determine the uniformity be- and precast bridges, respectively.
tween the parameters. The parameters These values indicate uniformity,
with correlation coefficients greater which suggests the feasibility of stan-
than approximately 0.80 were consid- dardization.
ered to be related, indicating uni- It is interesting to note that the func-
formity. Such uniformity would suggest tion for these precast bridges was
that the parameters lend themselves to below and somewhat parallel to cast-
standardization. Note that precast and in-place bridges. This indicates that for
cast-in-place bridges were considered the same span length the precast seg-
together and also independently. ments incorporate thinner webs than
To study the web dimensions for a their cast-in-place equivalents, which
particular span length, the web area for may be related to weight reduction
those bridges surveyed was normalized strived for in plant production.
by the bridge width. This accounted for The study of the soffit parameter,
the varied number of traffic lanes and shown in Fig. 13, was defined by di-
loading conditions. The web parameter viding the soffit cross-sectional area (lo-
was defined as the total area of the web cated near the pier) by the bridge
divided by the bridge width. The re- width, which normalized the different
lationship between web parameter and bridges surveyed. Quantitatively, when
7
• precast
0.30m=1.Oft
D
11
15
m
0
a
m
Z.
0
0.042L+0.45
i0 10
• slope 1:22 to 1:23
• • ^
•
•
the structural system is continuous over The low correlation may be attri-
a support, the bottom soffit near the buted to the varying amount of trans-
support must develop a compressive verse prestressing in the deck, which
force to resist the induced moment. was not included in the study. Also, the
Since this induced moment is related to deck thickness of the cantilever at its
the span length, the bottom soffit area support may he controlled by dimen-
must also increase with increasing span sioning requirements to accommodate
length. The correlation coefficient con- the longitudinal tendon anchorages, in-
sidering both precast and cast-in-place stead of providing the amount of resis-
segments was 0.83, indicating good cor- tance to induced forces.
relation. A high correlation was found be-
In Fig. 14, the deck thickness at the tween span length and girder depth for
cantilever base is plotted against the balanced cantilevers with constant
length of cantilever. The figure repre- depth sections, as shown in Fig. 15. A
sents the results of a study of the deck correlation coefficient of 0.95 was cal-
thickness at the transverse cantilever culated for the bridges considered. Re-
support as a function of the cantilever sults show that the average span-to-
length. AIthough a low correlation depth ratio was between 22 and 23 for
coefficient of 0.60 was calculated, the span ranges between 130 and 450 ft (40
deck thickness could intuitively be and 137 m). Also, the majority of con-
standardized For a particular bridge stant depth structures are precast as op-
width. posed to cast in plaee.
correlation
50 r A cast-in-place =0.84
• precast =0.60
0.30m =1.Oft
MIOP L
0
0 4.0 AL
s
A
a
W
L
^Q
a 3.0 -0.004L+O.9 (A)
c ..
a i
f • ^
2
0
2.0
0
a-
A 0.003E+1.13 (•)
•
The longitudinal haunch ratios, de- segments. Also, the infrequent use of
fined as the pier-to-midspan-depth haunched precast concrete segments
ratios, were studied for those bridges resulted in insufficient data for statisti-
employing balanced cantilever con- caI analysis.
struction. The results are shown in Fig.
16. The cast-in-place haunch ratios
varied from 1.7 for the shorter spans to Preliminary Designs for
4.3 for the longer spans. A correlation
Various Construction
coefficient of 0.84 was calculated for
cast-in-place bridges, indicating high Methods
uniformity. The low correlation coeffi- Preliminary designs were made to
cient for precast construction may determine the cost effectiveness of the
suggest difficulties or reluctance as- various construction methods. Quan-
sociated with using precast haunched tities of materials rather than cost fig-
0.30m31m2 =1.Ocf/sf
0.30m = 1.0ft
4.0
3.0
E v/c
-precast
0
balanced & progressive
cantilever non-prismatic
ressive placing
span-by-span
ures were used, since the latter are too The weight of prestressing steel ver-
variable. sus span lengths is shown in Fig. 18.
Fig. 17 shows the volume of concrete The relationship is similar to that of
plotted against span lengths for the volume of concrete for the various
various construction methods. Span- methods ofconstnrction.
by-span construction is more efficient From these curves and from other
in the lower span ranges, with balanced data presented it may he concluded that
cantilever being more efficient in the balanced cantilever is the most preva-
higher ranges. Incremental launching is lent method of construction for spans
cost-effective up to about 200-ft (61 m) over 150 It (46 m). Up to 300 ft (91 m),
spans, but becomes inefficient beyond precast construction is advantageous
that point, apparently because of the because such spans permit a constant
need to employ concentric prestressing. depth of section. Once a parabolic
Progressive placing shows economy of haunch is necessary to accommodate
concrete volume up to about 200 ft (61 the span, the cast-in-place section be-
m). comes more appropriate. It has been
70
20
incremental launching
0.30m =1 Oft
47.9NJm2=1.Olb/sf
15
/
/
N ^^ /
5,
1 fl non–prismatic
progressive placing-''
o ^! prismatic
L 5 ^'
a,
balanced & progressive
cantilever
ti- span-by-span
used for spans up to about 800 ft (244 tion of use, which was based on an ad-
m). Span-by-span construction is 're- ditional questionnaire sent to bridge
stricted to the shorter spans, perhaps up engineers in the United States and
to 150 ft (46 in). Canada, It shows that the projected use,
Items suitable for standardization are in their opinion, will feature to a great
summarized in Table 2. extent spans between 80 and 120 ft (24
Fig. 19 shows an interesting projec- and 37 m). This indicates that segmen-
80
60
m r,
m ^
^ c
0
=
o E40
20
o 0 0 0
0 o o MD o
0 Ui 4 W u]
N
o o 0 0
0 0 0 0 0 0'
CD 0 IL) o MD a
a m r c*) V W m
Fig. 19. Comparison of projected total square footage of bridge deck within span
ranges for steel, reinforced and prestressed concrete structures.
72
12
47.9Nlm2=1.Olb/sf
0.30m =1.Oft
10
wing section
8
1-girder
^ a 6
o ^. box girder
z
as
4 -T-girder
50 100 150
(ft)
Span Length
tal box girder construction will have to Fig. 21 plots the volume of concrete
compete with other types of construc- versus span length, showing that the
tion which have shown economies in volume is lowest for the T-section,
these particular span ranges. For that There is a cross-over point at about 90 ft
purpose a comparative study was made (27 m) between the box section and the
between the box section and other sec- I-section. Comparing the weight of lon-
tional forms. gitudinal post-tensioning steel, as
shown in Fig. 22, the I-section is low-
est, but there is a cross-over point at
Comparative Studies of
about 110 ft (33 m) with the box section.
Bridge Sections Comparing the mild steel required, the
An analytical study produced the T-section is again lowest (see Fig. 23).
comparison of quantities for the box The I-section and the box section have
section, I-girder, T-section, and wing a cross-over point at about 85 ft (26 m).
section, as shown in Fig. 20. Assessing the cost, as shown in Fig, 24,
4 w
N N
- m 3
N ^ rder
Q)
5-
0
.c 2
a)
50 100 150
(ft)
Span Length
Fig. 22. Weight of longitudinal prestressing steel for various girder types.
74
1300 1-cell
2 -cells — — —
0.30m=t.Oft /A
1100 14.6N/m =1.01b/ft
A/
A/
900 Aj A&
m
u1 W = 70`
m 700 l/
o^ /
s A / W = 50'
500 f/
300
10 20 30 40
(ft)
Girder Depth
Fig. 23. Weight of mild reinforcing steel for various deck widths and girder depths.
30 r wing section
w N
o
U ^
20
box girde I—girder
T—girder
10
50 100 150
(ft)
Span Length
and 91 m). For spans greater than 300 ft considered, especially in conjunction
(91 m), standardization does not yet with the transverse prestressing tendon
seem practical, but guidelines for de- layout in the deck.
sign and construction could be pro- 7. Transverse prestressing design and
vided. Similarly, recommendations for tendon layout should be standardized,
the accommodation of curved bridges since they can seriously affect top slab
should be included. dimensions.
5. Construction methods themselves 8. Vertical prestressing design and
should not he standardized, but the tendon layout for the webs should not
standardization of sections should con- be standardized, since they are gener-
sider the balanced cantilever method ally not needed fbr spans under 250 ft
and the span-by-span method, both cast (76 in). The design procedure and de-
in place and precast. The progressive tailing may he recommended.
placing and incremental launching 9. The possible use of external ten-
methods are not as yet sufficiently in dons for shorter spans should be
use to be included in standardization of treated, and recommendations fbr dif-
sections. ferential localities and environments
6. Longitudinal prestressing design should be made.
and tendon layout should not be stan- 10. Joints, both match-cast and wet,
dardized, but the magnitude of pre- should be standardized. Single or mul-
stressing force and eccentricity re- tiple shear key designs using epoxy
quired for the final condition of the between abutting precast segments
structure should be indicated. Their could be standardized. The possibility
effect on section dimensions should be of eliminating the epoxy between the
76
precast segments should be studied zation could be carried out is not as
further. clear in the case of box girder bridges,
11. Typical designs of anchorages and given their complexity, difference in
blisters for continuity and cap tendons construction methods, span ranges, and
may be suggested, but not standar- other variables. In any event, stan-
dized. dardization should be done to enhance
12. The use of bonded mild steel re- the use of segmental bridges by design-
inforcement for partial prestressing, ers and contractors, but it should be
temperature and shrinkage control, conceived and applied so as not to im-
stress concentration, prevention of de- pede new developments which might
lamination, and other local problems bring about greater economy, higher
may be recommended. safety and better performance.
13. Design of sidewalks, bicycle
paths, barriers, and railings should not
be standardized. ACKNOWLEDGMENT
14. Deflection control, both during
construction and after completion, This investigation was sponsored by
should he taken into account in the di- the Office of Research and Develop-
mensioning of standard sections. ment, Federal Highway Administration,
15. Location of expansion joints, both U.S. Department of Transportation,
temporary and permanent, may affect Washington, D.C., under the direction
the design of standard sections, and of Thomas Krylowski and Craig A. Bal-
guidelines should be established. linger.
16. Uniformity in design and specifi- The authors wish to thank the
cations should be addressed. FHWA's technical review committee
for help and guidance throughout the
It would be very desirable, indeed, if study and preparation of the report.
standardization of segmental pre- Members included Thomas Alberdi,
stressed concrete box girder bridges John Breen, Clifford Freyermuth,
could be accomplished to the same de- Wayne Henneberger, Jerry Jacques and
gree as AASHTO I-girders. The ap- Gordon Ray. The authors also wish to
proach could certainly be similar to that express their appreciation to the bridge
of the I-girders, inasmuch as dimen- engineers in the United States and
sional standards and construction prac- Canada for the information they pro-
tices could he made uniform, vided in the bridge survey question-
The extent to which such standardi- naire.