March-April 2004 1

Load Testing of Prestressed

Concrete Double Tees
Without Web Reinforcement
Alex Aswad, Ph.D., P.E., FPCI This paper describes three static load tests that were
Professor of Engineering
The Pennsylvania State University conducted on full-scale pretopped 12DT30 double
at Harrisburg tees. All specimens were typical production double
Middletown, Pennsylvania tees spanning 62 ft (18.9 m) and representing
typical units used in parking structures. Design of
the units used heavy live loads and the ACI 318-02
Building Code for load and resistance factors. The
bottom tensile stress under full live load was in the
range of 10.3 fc′ to 11.9 fc′ , which places these
units in the Class T category of ACI 318-02, Section
George Burnley, P.E.
Senior Engineer 18.3.3. Although the specimens’ stem dimensions
High Concrete Structures, Inc. varied, the double tees shared many common
Denver, Pennsylvania features, including total omission of shear
reinforcement in the web. The paper concludes
with appropriate recommendations related to the
omission of web reinforcement in prestressed
Ned M. Cleland, concrete double tees of similar geometry.
Ph.D., P.E., FPCI
President
Blue Ridge Design, Inc.
Winchester, Virginia he pretopped double tee is a very common precast,

T prestressed concrete product in North America. The

recent ACI 318-02 Building Code requirements for
load and φ-factors,1 the impact of bottom tensile stress, es-
pecially for the Class T category under Section 18.3.3
David Orndorff, P.E. (7.5 fc′ < |ft | ≤ 12 fc′ ), and the interest in omitting web
Engineering Manager reinforcement for shear by Section [Link] have attracted
The Shockey Precast Group the interest of designers and precast concrete manufacturers
Winchester, Virginia
alike. Furthermore, the design live load of 80 psf (3.83 kPa)
for roof tees in some mid-Atlantic areas is quite high and is
often used in conjunction with a unit length of 60 to 61 ft
(18.3 to 18.6 m).

BACKGROUND
Under the current ACI 318-02 Code, the load combina-
Charles Wynings, P.E. tion for ultimate strength is 1.2D + 1.6L, and the φ-factors
Engineering Manager
for flexure and shear are 0.90 and 0.75, respectively. These
Tindall Corporation are significant changes from previous editions. Since 1971,
Petersburg, Virginia a number of load tests have been conducted on full-size un-

66 PCI JOURNAL
 12DT30 PRECAST SECTION DIMENSIONS

CALCULATED GROSS SECTION PROPERTIES

Fig. 1. Cross section

and section
properties of the
three test
specimens.

topped and topped double tees by at tude is also provided in ACI 318-02, thick flange. Prestressing strands were
least eight companies or plants.2-6 The Section [Link]. straight. Geometry and material pa-
objective of these tests was to show In 1986, a 10 ft 8 in. (3.25 m) wide rameters may be obtained from High
that web reinforcement can be elimi- DT24 section was load tested using a Concrete Structures, Denver, Pennsyl-
nated without impairing the flexural span of 56.5 ft (17.2 m) center to cen- vania.
and shear capacities of prestressed ter of bearings. A 3 in. (76 mm) com- Test specimens in References 3 to 6
concrete double tees. This objective posite topping was made monolithic had no web reinforcement except for
was met, and producers invoked the with the 2 in. (51 mm) thick flange. some welded wire fabric near the
provisions of Chapter 11 of ACI 318 Strands were draped, with no debond- ends.
to waive the minimum area of shear ing at the ends. The test setup pro-
reinforcement. vided a horizontal force equal to 20
In Reference 3, three 1975 load tests percent of the vertical reaction. Other TEST OBJECTIVES
on double tees are discussed in detail. details of the test are included in Ref- For this test program, a series of
These tests involved 8DT12, 8DT18, erence 4. three load tests was needed to supple-
and 8DT24 sections with spans of 30, Reference 5 discusses two full-size ment the available data, including data
48, and 50 ft (9.14, 14.6, and 15.2 m), load tests on 10DT34 and 10DT26 on flanges wider than 10 ft (3.05 m),
respectively. All sections had a 2 in. pretopped specimens, which were con- straight strands with partial debond-
(51 mm) flange thickness, and strands ducted in 1987 and 1988, respectively. ing, and higher bottom tensile stresses.
were draped for all specimens. The design span for both specimens The objective was to observe the be-
In 1983, the ACI 318-83 Code was 62 ft (18.9 m). The flange thick- havior when the web reinforcement
added a new requirement for tests ness was 4 in. (102 mm). Design live was totally omitted. The design of
used as a basis for the omission of loads were 70 psf (3.35 kPa) for the such test units also needed to comply
minimum web reinforcement. For the 10DT34 and 54 psf (2.59 kPa) for the with the new ACI 318-02 Code, espe-
last two decades, Section [Link] 10DT26. Strands were draped with cially in regard to the load and capac-
stated that such tests shall simulate the one depression point at midspan and ity reduction factors.
effects of differential settlements, partially debonded near the ends. A The test locations were the produc-
creep, shrinkage, and temperature horizontal force equal to 20 percent of tion plants of High Concrete Struc-
change. It is common practice among the reaction force was applied simulta- tures, Inc., Denver, Pennsylvania, Tin-
many prudent designers to simulate neously with the gravity loads. dall Corporation, Petersburg, Virginia,
such volume change effects on the Reference 6 briefly describes the and The Shockey Precast Group,
bearings by applying a horizontal ten- 1996 testing of a 15DT34 specimen to Winchester, Virginia. Details, test re-
sile force, N, equal to 20 percent of the failure. The double tee had a 57 ft sults, and typical design parameters
vertical reaction, V. This force magni- (17.4 m) span and a 4 in. (102 mm) are summarized in the following sec-

March-April 2004 67
Fig. 2. Test setup for the three test specimens.

1
tions along with the conclusions and new ASCE 7-02 standard is adopted / 2 in. (12.7 mm) diameter strands
recommendations. The full report7 is by most jurisdictions, the design live stressed to 69 percent of the UTS.
available from the project sponsors.* load in these tests was kept at a uni- Shielding involved four strands
form 80 psf (3.83 kPa). This simulates debonded for 10 ft (3.05 m) from each
the maximum combination of snow end. Average stem width was 6.45 in.
DESCRIPTION AND plus vehicle loads for parking struc- (164 mm).
PARAMETERS tures in many geographic areas. • Specimen 2 (by Tindall Corpora-
Test specimens in this program con- All specimens used straight Grade tion) used fourteen 9/16 in. (14.3 mm)
sisted of 12.0 ft wide (3.66 m), 30 in. 270 ksi (1862 MPa) low-relaxation diameter strands stressed to 75 percent
(760 mm) deep precast, pretopped strands conforming to ASTM A 416. of the UTS. Shielding involved four
double tees cast at various times be- Areas of special 1/2 in. (12.7 mm) and strands debonded for 5 to 8 ft (1.52 to
9
tween June and October 2002. The / 16 in. (14.3 mm) diameter strands 2.44 m) from each end. One No. 6
span length (center to center of bear- were equal to 0.167 and 0.192 sq in. (19.1 mm) reinforcing bar, 20 ft (6.1
ings) was a uniform 62 ft (18.9 m) and (108 and 124 mm2), respectively. Due m) long, was used in each stem. Aver-
selected in such a way as to envelope to different section geometries and age stem width was 5.38 in. (137
all practical future parking structure production practices, the prestressing mm).
decks. The flange thickness was 4 in. reinforcement, amount of end shield- • Specimen 3 (by Shockey Precast
(102 mm). The typical cross section is ing, and jacking ratio (percentage of Group) used twelve 9/16 in. (14.3 mm)
shown in Fig. 1, along with a tabula- the ultimate tensile strength, or UTS) diameter strands stressed to 75 percent
tion of the section properties for the varied among the specimens. Further- of the UTS. Sleeving involved two
three specimens. For all test speci- more, two producers elected to use strands debonded for 6 ft (1.83 m)
mens, normal-weight concrete was mild reinforcing steel in the middle from each end. One No. 7 (22.2 mm)
used, and web reinforcement was en- third of the span to supplement the reinforcing bar, 24 ft (7.32 m) long,
tirely omitted. flexural capacity furnished by the pre- was used in each stem. Average stem
Although design live loading for stressing strands. Web reinforcement width was 7 in. (178 mm).
parking structures will decrease if the was totally omitted from all speci-
mens. A summary of the reinforce-
ment is listed as follows: TEST SETUP
* High Concrete Structures, Inc. (717-336-9300),
Tindall Corporation (804-861-8447), and The Shockey • Specimen 1 (by High Concrete The test setup was the same for all
Precast Group (540-667-7700). Structures, Inc.) used twenty special cases. Structural steel supports were

68 PCI JOURNAL
 Fig. 3. Specimen 1,
Stage 2: Full live
load (20 blocks).

Fig. 4. Specimen 1,
Stage 5: Required
failure load with
φ = 1.0.

used to provide a pinned bearing con- tilt further and create a horizontal LOADING STAGES
dition and the desired horizontal ten- force ratio that is greater than 0.20. In this test series, 10, 18, or 20
sile force N (equal to at least 20 per- Note that it is common practice stacks of large concrete blocks were
cent of the end reaction V), as shown among many design engineers to used to simulate the quasi-uniform,
in Fig. 2. These test supports were ele- apply the 0.20 force ratio only to the static loading condition. The place-
vated to hold the slab at 3 to 4 ft (0.91 factored dead loads. Therefore, the ment of the blocks was structured such
to 1.22 m) above the ground so that tests described in this paper used a that loading would proceed in stages.
the double tee could deflect freely. significantly higher horizontal force. This process made it possible to com-
Pivots on one side had a link that was As a safety measure, large timbers pare the actual flexural behavior with
tilted 11.3 degrees from vertical to or concrete blocks were stacked near that predicted by calculations. Each of
generate the desired horizontal force. the supports. However, several inches the five benchmark loading stages pro-
The 20 percent force ratio is valid for of clearance were still available under duced a midspan moment (or bottom
small cambers or deflections under the test unit. Changes in the slab stress ft in Stage 2) nearly equal to the
typical service loads. Under heavier midspan camber were measured from designated target for that stage. The
loads, however, the bottom fiber of the the original (unloaded) and cambered nominal concrete compressive
double tees would elongate somewhat position – designated as the zero load strength fc′ was assumed to equal 6000
and cause the left pivot (see Fig. 2) to position. psi (41.4 MPa) for all specimens. The

March-April 2004 69
Table 1. Input and output data for specimens subject to uniform live loads. strength, assuming φ = 1.0. In a well-
Specimen 1 Specimen 2 Specimen 3 balanced design situation, this
Section properties and loads amounts to a total load w3 = (1.2D +
Gross area, A 911.4 sq in. 855.5 sq in. 940.0 sq in. 1.6L)/0.9.
Moment of inertia, I 61,708 in.4 54,827 in.4 68,322 in.4 As an example, Figs. 3 and 4 show
Section modulus, Sb 2706.3 in. 3
2350.1 in.3 3054.0 in.3 the loading patterns for Specimen 1 in
Unit weight 0.949 kips/ft 0.891 kips/ft 0.979 kips/ft Stages 2 and 5, respectively.
Live load 0.960 kips/ft 0.960 kips/ft 0.960 kips/ft Schematic diagrams showing the
Concrete properties and reinforcement information placement of various weights, bending
Design fci′ 4400 psi 4500 psi 3800 psi moment values, and total camber
Eci 4021 ksi 4067 ksi 3737 ksi changes at the end of each stage for all
Design fc′ 6000 psi 6000 psi 6000 psi specimens are discussed in detail in
Ec 4696 ksi 4696 ksi 4696 ksi Reference 7. Placement of blocks was
Number of strands carried out in a continuous (mono-
20 14 12 tonic) process from start to finish over
below the flange
Area of each strand 0.167 sq in. 0.192 sq in. 0.192 sq in. three to four hours, with occasional
C.G. of strands, Ycg 9.63 in. 8.00 in. 5.82 in. minor interruptions to permit visual
Reinforcing bars per stem 1 No. 6 at 5 in. 1 No. 7 at 7 in. crack inspection, deflection measure-
None ment, or a lunch break. After Stage 2,
(20 ft long) above bottom above bottom
Strands’ pull ratio 69 percent 75 percent 75 percent crack observations were made from a
Total prestress loss 27.08 ksi 28.2 ksi 27.8 ksi safe distance. Two items in regard to
Load and resistance multipliers by ACI 318-02 the testing program are worthy of dis-
For dead loads 1.2 1.2 1.2 cussion:
For live loads 1.6 1.6 1.6 1. In the calculation of midspan mo-
φ-factor, flexure 0.90 0.90 0.90 ments and bottom tensile stress, the ef-
φ-factor, shear 0.75 0.75 0.75 fect of the horizontal force, N, and its
Selected output data from the analysis program eccentricity, yb, would add to the total
Self-weight moment 456.1 kip-ft 428.2 kip-ft 470.6 kip-ft moment, M T , and bottom tensile
Live load moment 461.2 kip-ft 461.2 kip-ft 461.4 kip-ft stress, ft. However, this effect was dis-
Maximum stress, ft 894 psi 809 psi 826 psi regarded because design software is
Required Mu at midspan 1285 kip-ft 1252 kip-ft 1303 kip-ft normally not equipped to consider N.
Available φMn (φ = 0.9) This practice is considered to be ac-
1331 kip-ft 1266 kip-ft 1314 kip-ft ceptable.
at midspan
Mu /0.9 1428 kip-ft 1391 kip-ft 1448 kip-ft 2. Another consideration is the ef-
Required Vu 79.6 kips 77.5 kips 82.2 kips fect of curing temperatures on increas-
(at distance x from left support) (at x = 1.25 ft) (at x = 1.25 ft) (at x = 0.67 ft) ing the initial prestress loss.8 Past tests
Required Av,ci (at quarter point) 0.06 sq in./ft 0.073 sq in./ft 0.03 sq in./ft have shown that curing temperatures
Required min. Av by ACI 318-99 0.13 sq in./ft 0.108 sq in./ft 0.09 sq in./ft will increase initial prestress loss by 4
Predicted cambers and camber change to 8 percent for several weeks after re-
Camber at release 0.72 in. ↑ 0.98 in. ↑ 0.81 in. ↑ lease of prestress. On the other hand,
Camber at erection 1.18 in. ↑ 1.58 in. ↑ 1.39 in. ↑ the time lag between release time and
Final camber change under testing date was not long enough to
2.01 in. ↓ 1.34 in. ↓ 1.64 in. ↓ allow creep and shrinkage losses to
full uniform live load
take their full effect on strand losses.
Note: 1 in. = 25.4 mm; 1 sq in. = 645.2 mm2; 1 in.3 = 16387 mm3; 1 in.4 = 416231 mm4; 1 ft = 0.3048 m;
1 kip = 4.448 kN; 1 kip/ft = 14.59 kN/m; 1 kip-ft = 1.356 kN-m; 1 psi = 0.0069 MPa; 1 ksi = 6.90 MPa. Because these two effects oppose each
other in their influence on ft, they were
ignored in the bottom stress calcula-
five loading stages were as follows: 318-02, Section 20.3.2). In strength tions under service loads (Stages 1 and
• Stage 1, first expected flexural evaluation of some products, this load 2).
crack — This stage corresponds to a is the required test load. It is not used The following sections provide con-
calculated nominal bottom tensile in the design of new products. cise descriptions and observations for
stress of ft = fr = 7.5 fc′ = 581 psi • Stage 4, factored load w2 = wu = each one of the three load tests.
(4.0 MPa). 1.2D + 1.6L — This load intensity is
• Stage 2, full live load — This mandated by ACI 318-02. It is used in
TEST RESULTS FOR
load is equivalent to the full service conjunction with φ = 0.9.
live load of 80 psf (3.83 kPa). • Stage 5, expected failure load w3 SPECIMEN 1
• Stage 3, intermediate phase — = (1.2D + 1.6L)/0.9 — The load inten- Specimen 1 consisted of a 12.0 ft
This stage corresponds to a load inten- sity w3 is such that the midspan mo- (3.66 m) wide, 30 in. (760 mm) deep
sity of w1 = 0.85(1.4D + 1.7L) (ACI ment is equal to the nominal flexural double tee cast on June 17, 2002, with

70 PCI JOURNAL
a total length of 62.67 ft (19.1 m). The
specimen contained 20 special 1/2 in.
(12.7 mm) diameter straight strands,
which were tensioned to 69 percent of
the UTS. Strands in Rows 4 and 6
from the bottom were debonded for 10
ft (3.05 m) from each end.
The nominal concrete strengths
were fci′ = 4400 psi (30.3 MPa) and fc′
= 6000 psi (41.4 MPa). The actual av-
erage strength at release was fci′ = 4492
psi (31 MPa). On the test date (June
20), the actual average strength was fc′
= 6268 psi (43.2 MPa). The welded
wire fabric in the deck was supported
by six slightly tensioned 1/4 in. (6.4
mm) diameter wires. Each end bearing
area was reinforced with a steel plate
welded to two 3 ft (0.91 m) long No. 5
(15.9 mm) reinforcing bars in the bot-
tom.
The design live load was 80 psf
(3.83 kPa). A listing of properties re- Fig. 5. Specimen 1: Superimposed load moment versus camber change.
sulting from this loading is provided
in Table 1. The computer analysis
shows a predicted release camber of At the end of Stage 2 (Fig. 3), with blocks, weighing 132 kips (587 kN),
0.72 in. (18 mm) and a camber at erec- a bottom tensile stress equal to 924 psi were placed over the double tee, re-
tion of 1.18 in. (30 mm). On the test (6.4 MPa), or 11.9 6000 , under a su- sulting in an applied moment M T =
date, the measured camber (using a perimposed load of 20 blocks, still no 1476 kip-ft (2001 kN-m) at midspan
string line) was approximately 1.6 in. flexural cracking was observed. Total versus the required moment Mu/0.9 =
(41 mm). camber change was 1.06 in. (27 mm). 1428 kip-ft (1937 kN-m). The calcu-
After adding two blocks near each lated available capacity was M R =
end, the first two small hairline cracks 1479 kip-ft (2006 kN-m). Deflections
Loading stages and observations appeared at distances of 10 to 15 in. increased noticeably. The last mea-
To simulate each of the five loading (254 to 381 mm) away from the cen- sured camber change was approxi-
stages, a series of 3 kip (13.3 kN) con- terline. A very tiny (and nearly hori- mately 39 in. (1 m). Failure occurred
crete blocks were placed in a symmet- zontal) 30 in. (0.76 m) long crack ap- when an extra 3 kip (13.3 kN) block
rical layout at specific locations to peared near the bearings after six and was added near midspan (greater than
reach the target midspan total mo- four blocks were added near each end, Stage 5 loading).
ment, M T . The exact positions for respectively. This hairline crack, how- For Stage 5, the actual shear force at
Stages 2 and 5 are shown in Figs. 3 ever, did not widen under further load- h/2 =1.25 ft (381 mm) from the bear-
and 4, respectively. Note the tilted pin ing. At the end of Stage 3, many other ings was equal to 94.2 kips (419 kN),
support on the left in Fig. 2. The com- flexural cracks were observed in the while the required design value as pro-
prehensive curve showing the midpsan middle half of the span. No shear vided by ACI was Vu = 79.6 kips (354
superimposed load moment versus the cracking appeared at the end of this kN). Based on the uniform live load,
camber change for all stages is shown stage. Total camber change was 6 in. the maximum required shear rein-
in Fig. 5. (152 mm). forcement, Av,ci, was equal to 0.06 sq
The load test started at 9:00 a.m. At the end of Stage 4, flexural in./ft (127 mm2/m) (see Table 1).
Total duration of loading to failure cracks progressed toward the supports
was approximately 3.5 hours. The as expected – up to approximately 14
weather was cloudy, with tempera- ft (4.27 m) from each end. Some of Discussion of testing
tures between 71° and 79°F (22° and the cracks started showing deviation The curve representing the superim-
26°C). from the vertical line. Under the posed moment versus the measured
At the end of Stage 1, with a calcu- weight of 38 blocks, the total camber total camber change is plotted in Fig.
lated bottom tensile stress equal to 572 change was 11.8 in. (0.3 m). 5. This curve shows that the loss of
psi (3.9 MPa), or 7.4 6000 , under a At the end of Stage 5 (Fig. 4), crack stiffness was smooth and gradual.
superimposed load of 18 blocks, no widths expanded significantly. A few Throughout the loading test, flexural
flexural cracking was observed. Cam- cracks were approximately 1/2 to 5/8 in. cracks did not propagate to cause early
ber change was 0.88 in. (22 mm). (13 to 16 mm) wide. A total of 44 flange separation, nor was there any

March-April 2004 71
 sulting from this loading is provided in
Table 1. The computer analysis shows
a predicted release camber of 0.98 in.
(24.9 mm) and a camber at erection of
1.58 in. (40.1 mm). At test time, the
measured camber (using a surveying
instrument) was 1.62 in. (41 mm).

Loading stages and observations

To simulate each of the five loading
stages, a combination of 5.70 and 2.75
kip (25.4 and 12.2 kN) concrete
blocks were placed in a symmetrical
layout at specific locations to reach
the target midspan total moment, MT.
The exact positions for Stage 5 are
Fig. 6. Specimen 1: Loading at end of Stage 2 (full live load). shown in Fig. 8. The comprehensive
curve showing the midpsan live load
moment versus the camber change for
the various stages is shown in Fig. 9.
The load test started at 9:20 a.m.
Total duration of loading to failure
was approximately three hours. The
test was conducted indoors. Camber
was measured on both sides of the tee
using a surveying instrument. The av-
erage results are plotted in Fig. 9.
At the end of Stage 1, the calculated
bottom tensile stress was equal to 700
psi (4.8 MPa), or 9 fc′ , under a su-
perimposed load of eight 5.7 kip (25.3
kN) blocks. Only two very small flex-
ural cracks were observed near
midspan. Their width was on the order
of 0.004 in. (0.1 mm), and their length
did not exceed 3 in. (76 mm). Actual
Fig. 7. Specimen 1: Stage 5 loading (near failure). camber change was 1.47 in. (37 mm).
At the end of Stage 2, the bottom
tensile stress was equal to 798 psi (5.5
sign of impending shear failure at the prestressing reinforcement, one 20 ft MPa), or 10.3 fc′ , under a superim-
ends or near the quarter points. Figs. 6 (6.1 m) long, No. 6 (19.1 mm) Grade posed load of ten 5.7 kip (25.3 kN)
and 7 show the end of Stages 2 and 5, 60 ksi (414 MPa) mild steel reinforc- blocks (one full layer). The two previ-
respectively. ing bar was placed at 5 in. (127 mm) ous hairline flexural cracks remained
from the bottom in each stem and cen- the same without an increase in height.
tered within the span. Total camber change was 1.63 in. (41
TEST RESULTS FOR The nominal concrete strengths mm).
SPECIMEN 2 were fci′ = 4500 psi (31 MPa) and fc′ = At the end of Stage 3, many flexural
Specimen 2 consisted of a 12 ft 6000 psi (41.4 MPa). The actual aver- cracks were present in the middle half
(3.66 m) wide, 30 in. (760 mm) deep age strength at release was fci′ = 4529 of the span. Crack width was limited
double tee cast on August 5, 2002, psi (31.2 MPa). On the test date (Au- to 0.013 in. (0.3 mm). Most crack
with a total length of 62.5 ft (19.05 gust 8) the actual average strength was spacings ranged from 8 to 12 in. (203
m). The specimen contained fourteen fc′ = 5892 psi (40.6 MPa). Each end to 305 mm), although occasionally the
9
/ 16 in. (14.3 mm) straight strands, bearing area was reinforced with a spacing reached 15 in. (381 mm). No
which were tensioned to 75 percent of steel plate welded to two 3.17 ft (0.97 shear cracks appeared at the end of
the UTS. Strands in Rows 3 and 5 m) long No. 4 (12.7 mm) reinforcing this stage. Total camber change was
from the bottom were debonded for 8 bars in the bottom. 6.19 in. (157 mm).
and 5 ft (2.44 and 1.52 m) from each The design live load was 80 psf At the end of Stage 4, flexural
end, respectively. In addition to the (3.83 kPa). A listing of properties re- cracks propagated vertically and to-

72 PCI JOURNAL
 Fig. 8. Specimen 2,
Stage 5: Required
failure load with
φ = 1.0.

Fig. 9. Specimen 2:
Superimposed load
moment versus
camber change.

ward the supports as expected – up to moment for this stage was MT = 1384 (384 kN), whereas the required design
14 ft (4.27 m) from each end. At a dis- kip-ft (1877 kN-m) versus a required value by ACI was Vu = 77.5 kips (345
tance of about 21 ft (6.40 m) from the moment Mu/0.9 = 1391 kip-ft (1886 kN). Based on the uniform live load,
ends, minor hairline cracks started de- kN-m) and a calculated available ca- the maximum required shear rein-
viating from the vertical line and curv- pacity MR = 1406 kip-ft (1907 kN-m). forcement, Av,ci, was equal to 0.073 sq
ing diagonally. Total camber change The last measured camber change with in./ft (155 mm2/m) (see Table 1).
was 8.13 in. (207 mm). full load was approximately 47 in.
At the end of Stage 5 (Fig. 8), exist- (1.19 m), and the double tee was close
ing crack widths expanded signifi- to touching the ground. Discussion of testing
cantly and new flexural cracks ap- Failure occurred when an additional As seen from the curve in Fig. 9, the
peared at a distance of 12 ft (3.66 m) 5.7 kip (25.4 kN) block was added loss of stiffness was gradual. The su-
from the supports. Near the end of this near the third point (greater than Stage perimposed moment versus the cam-
stage, with one large block missing, 5 loading). This caused a major diago- ber change curve has a bilinear shape.
two diagonal cracks appeared at dis- nal crack to form suddenly at the bear- Throughout the loading test, flexural
tances of 15 and 21 ft (4.6 and 6.4 m) ings on one end. At midspan, the dou- cracks did not propagate to cause early
from the bearings, respectively. The ble tee had already touched the flange separation, nor was there any
one at 15 ft (4.6 m) was hairline in ground. For this stage, the actual shear sign of premature shear failure at the
size, whereas the one at 21 ft (6.4 m) force at h/2 =1.25 ft (381 mm) from ends or near the quarter points. Flex-
was wider. The total applied midspan the bearings was equal to 86.3 kips ure and shear strengths appeared to be

March-April 2004 73
 3
/8 in. (9.5 mm) diameter strands. Each
end bearing area was reinforced with a
special steel plate welded to one No. 6
(19.1 mm), 4.0 ft (1.22 m) long bot-
tom reinforcing bar.
The design live load was 80 psf
(3.83 kPa). A listing of properties re-
sulting from this loading is provided
in Table 1. The computer analysis
shows a predicted release camber of
0.81 in. (21 mm) and a camber at erec-
tion of 1.39 in. (35.3 mm). On the test
date, the measured camber (using a
string line) was approximately 1.7 in.
(43 mm).

Loading stages and observations

Fig. 10. Specimen 2: Stage 5 loading (near failure).
To simulate each of the five loading
stages, a series of concrete blocks with
different weights were placed in a
reached nearly simultaneously. Fig. 10 ft (1.83 m) at each end. In addition to symmetrical layout at specific loca-
shows the specimen near the end of the prestressing reinforcement, one 20 tions in order to reach the target
Stage 5. ft (6.1 m) long No. 7 (22.2 mm) Grade midspan total moment, MT. The exact
60 ksi (414 MPa) mild steel reinforc- positions and equivalent point load
ing bar was placed at 7 in. (127 mm) magnitudes for Stage 5 are shown in
TEST RESULTS FOR from the bottom in each stem and cen- Fig. 11. The comprehensive curve
SPECIMEN 3 tered within the span. showing the midspan live load mo-
Specimen 3 consisted of a 12 ft The nominal concrete strengths ment versus the camber change for all
(3.66 m) wide, 30 in. (760 mm) deep were fci′ = 3800 psi (26.2 MPa) and fc′ stages is shown in Fig. 12.
double tee cast on October 10, 2002, = 6000 psi (41.4 MPa). The actual av- The load test started at 8:15 a.m.
with a total length of 62.67 ft (19.1 erage strength at release was fci′ = 3938 Total effective duration of loading to
m). The specimen contained twelve psi (27.2 MPa). On the test date failure was approximately four hours.
9
/ 16 in. (14.3 mm) diameter straight (November 1), the actual average The weather was cloudy, with temper-
strands, which were tensioned to 75 strength was fc′ = 7130 psi (49.2 MPa). atures between 40° and 48°F (4° and
percent of the UTS. Strands in Row 3 Welded wire fabric in the deck was 9°C).
from the bottom were debonded for 6 supported by four slightly tensioned At the end of Stage 1, with a calcu-

Fig. 11. Specimen 3,

Stage 5: Required
failure load with
φ = 1.0.

74 PCI JOURNAL
lated bottom tensile stress equal to 592
psi (4.1 MPa), or 7.6 6000 , under a
superimposed bending moment of
400.3 kip-ft (543 kN-m), no flexural
cracking was observed. Camber
change was 0.88 in. (22 mm).
At the end of Stage 2, with a bottom
tensile stress equal to 834 psi (5.8
MPa) or 10.8 6000 , under a superim-
posed bending moment of 463.1 kip-ft
(628 kN-m), the first vertical hairline
flexural crack appeared at a distance
of 2.67 ft (0.81 m) from the double tee
centerline. Somewhat earlier, very tiny
(and nearly horizontal) cracks ap-
peared at approximately 8 in. (203
mm) above the bearing location in
three stems. These cracks did not
widen under further loading. Total
camber change was 1.0 in. (25 mm).
At the end of Stage 3, with a super- Fig. 12. Specimen 3: Superimposed load moment versus camber change.
imposed bending moment of 755.6
kip-ft (1025 kN-m), several additional
flexural cracks appeared in the middle
third of the span. The maximum crack
width was approximately 0.009 in.
(0.23 mm). Crack length varied, with
the maximum value occurring at 1.83
ft (558 mm) above the bottom of the
stem. No shear cracks appeared at the
end of this stage. Total camber change
was 4.5 in. (114 mm).
At the end of Stage 4, flexural
cracks progressed toward the supports,
up to approximately 14 ft (4.27 m)
from each end. Upon addition of the
eleventh of sixteen railroad ties (in
Stage 4), an inclined hairline shear-
flexure crack appeared at approxi- Fig. 13. Specimen 3: Loading at Stage 5 near failure (two diagonal cracks close to
mately 12 in. (305 mm) from the in- L/4 at right).
side of the quarter point (toward the
supports) and declined toward the
bearing. This crack did not expand Deflections increased in a notice- (381 kN), whereas the required design
significantly until midway through able (but slow) fashion. The last mea- value by ACI was Vu = 82.2 kips (366
Stage 5 loading. Total camber change sured camber change was approxi- kN). Based on uniform loading, the
was 6.5 in. (165 mm). mately 16 in. (406 mm) immediately maximum required shear reinforce-
Throughout Stage 5 (Fig. 11), the prior to the addition of the final block. ment, Av,ci, was equal to 0.03 sq in./ft
flexure crack widths did not expand At a superimposed load moment of (64 mm2/m) (see Table 1).
significantly. Upon the addition of 976.5 kip-ft (1324 kN-m), the double
four 1.1 kip (4.9 kN) blocks at one end tee finally failed and hit the ground.
and two similar blocks at the opposite The total applied midspan moment for Discussion of testing
end, the initial shear crack mentioned this stage was MT = 1447 kip-ft (1962 Despite a significant bending curva-
in Stage 4 abruptly opened to an ap- kN-m) versus a required moment ture in the specimen and a camber
proximate width of 1 in. (25 mm). M u/0.9 = 1448 kip-ft (1963 kN-m). change exceeding 16 in. (0.41 m) (see
Once the deflection stabilized, the re- The calculated capacity was M R = Fig. 12), the loss of stiffness occurred
maining six blocks were added, with 1460 kip-ft (1980 kN-m). as an abrupt quarter-point shear failure
large shear cracks opening near the For this stage, the actual shear force at approximately 110 percent of the
quarter point at the other end of the at a distance of 0.67 ft (204 mm) from factored load, wu. Although an immi-
double tee. the bearings was equal to 85.6 kips nent shear failure appeared at slightly

March-April 2004 75
less than the superimposed load mo- 4. To prevent the serious effects that Strength Reduction Factor
ment M = 976.5 kip-ft (1324 kN-m), it loss of bond has near the ends of dou- Due to the small steel ratio, pre-
is worth noting that the horizontal ble tees without shear reinforcement, topped double tee units generally qual-
force ratio near failure was somewhat strand bond should be verified or cer- ify as tension-controlled sections. The
larger than 20 percent because of the tified to conform to the development applicable φ -factor under current
test setup. In fact, elongation of the length as predicted by Eq. (12-2) of codes is 0.90. However, this factor ap-
bottom fibers exaggerated the tilt of ACI 318-02. pears to be conservative when applied
the supporting steel pivot at one end. 5. Based on the test specimen data to double tees reinforced with only
Therefore, the actual horizontal force, supplied by the sponsors (see Table straight strands. The chances for mis-
N, and the applied moments were 1), the values of Av,ci should be lim- placement of the strands are greatly
higher than those based on the gravity ited. The required shear reinforcement reduced. As for the concrete strength,
loads plus the effect of a horizontal based on Eq. (11-10) of ACI 318-02 fc′, its impact on the section’s design
force amounting to 20 percent of the should not exceed 0.080 sq in./ft (169 strength does not exceed 1 percent
reaction. Fig. 13 shows the load test mm2/m). even with a drop in the strength on the
near the end of Stage 5. It was observed that Specimen 3 ex- order of 1000 psi (6.9 MPa).
hibited less ductility than Specimens 1 In the AASHTO Standard or LRFD
and 2, and that some significant shear specifications,10,11 the φ-factor is set at
CONCLUSIONS AND
cracks appeared at the quarter points 1.0 because the quality control process
RECOMMENDATIONS during Stage 5. After evaluation of precludes significant understrength in
various parameters, it appears that the the materials or strand misplacement.
Special Vertical Reinforcement
major variable that distinguished Consequently, it is our considered
at Bearings
Specimen 3 from the other two speci- opinion that, for pretopped double tees
In Specimens 1 and 3, a nearly hori- mens was the relatively low average with straight strands, and appropriate
zontal crack appeared at the end of Pi /A, where Pi is the prestress force at Pi/A values, the φ-factor should be re-
Stage 2 or at the beginning of Stage 3. transfer. For Specimen 3, Pi /A = 479 viewed and could probably be raised
Although the crack was extremely thin psi (3.30 MPa) versus 657 and 601 psi to 0.95.
and did not widen under increased (4.53 and 4.14 MPa) for Specimens 1
loading, it is recommended that suffi- and 2, respectively. Incidentally, this
cient vertical reinforcement (Ash) be low Pi /A ratio was due to the more ef- ACI 318-02 for Strength Evaluation
provided as recommended in the PCI ficient use of strands.
Design Handbook,9 Section 4.6.2, for Section 20.3.2 of ACI 318-02 re-
Therefore, it is recommended that
controlling the so-called potential hor- quires the same load intensity, w1 =
whenever the steel ratio for 12DT30
izontal crack. 0.85(1.4D + 1.7L), for testing certain
double tees is such that Pi/A is less
products as previous code editions did.
than approximately 550 psi (3.8 MPa),
For many pretopped double tees, this
Omission of the Web the φ-factor for strength reduction in
intensity is approximately 94 percent
Reinforcement flexure be lowered to 0.85 for spans
of the required factored load, wu, com-
close to 62 ft (18.9 m) and carrying
In consideration of the positive re- pared to 85 percent in past code edi-
heavy live loads such as 80 psf (3.83
sults from these three load tests (with tions. Although this load intensity
kPa). For shorter spans, lighter live
at least 20 percent horizontal forces at does not affect the design of new
loads, or both, the 550 psi (3.8 MPa)
the supports) and the successful use of products, it has a significant effect on
limit should be reduced based on engi-
double tees with little or no web rein- the acceptance of prestressed concrete
neering judgment. For wider sections,
forcement on hundreds of projects products that show minor localized
or when section geometry deviates
since 1971,2-6 it is concluded that shear cracks or that do not appear to satisfy
significantly from the three speci-
reinforcement in pretopped double the code after release.
mens, a test is suggested to determine
tees having similar geometries can be The authors believe that the
the cut-off limit for Pi/A. An example
omitted, provided that the following 0.85(1.4D + 1.7L) requirement is too
of one such section is the 15DT34,
conditions are applied: conservative and should be revised in
which was successfully tested and ex-
1. Double tees are simply supported view of the lower load factors for
hibited ductile behavior.6 For such a
and designed for a Class U or Class T bending in current codes. It is doubtful
section, φ = 0.90 is appropriate.
type bottom tensile stress in accor- that future prestressed concrete prod-
dance with ACI 318-02, Section ucts can pass the residual deflection
18.3.3, with appropriate end bearing test, ACI 318, Section 20.5.2, since
plates and reinforcement. FURTHER RESEARCH this intensity is too close to the fac-
2. Loads are substantially uniform. The following paragraphs offer sug- tored load. It is worthy of note that, in
3. For shear, the designer considers gestions for further research on the the tests presented here, none of the
the stress reduction at the double tee subject of double tees with straight test specimens showed any premature
ends due to partial strand debonding. strand patterns. shear cracks at the end of Stage 3.

76 PCI JOURNAL
Minimum Shear Reinforcement REFERENCES on Prestressed Double Tees by ACI
318-02 with Web Reinforcement
Section [Link] of ACI 318-02 al- 1. ACI Committee 318, “Building Code Omitted,” A joint report published by
lows a minimum area of shear rein- Requirements for Structural Concrete High Concrete Structures, Inc., The
forcement, Av, to be set equal to the (ACI 318-02) and Commentary (ACI Shockey Precast Group, and Tindall
smaller of the two values obtained 318R-02),” American Concrete Insti- Corporation, December 2002.
from Eqs. (11-13) and (11-14). This tute, Farmington Hills, MI, 2002.
8. Frost, R. J., “Rational Development of
provision is based on research at the 2. Ghosh, S. K., and Fintel, M., “Excep- Release Strength Criteria,” Master of
University of Illinois12 dating back to tions of Precast, Prestressed Members Engineering Thesis, The Pennsylvania
to Minimum Requirements,” Research State University, Graduate School,
1967. As an example, for Specimen 3
Project No. 2, Chapter 3, Precast/Pre-
and using fc′ = 6000 psi (41.4 MPa), stressed Concrete Institute, Chicago,
Harrisburg, PA, 1997.
Eq. (11-13) yields Av = 0.16 sq in./ft 9. PCI Design Handbook: Precast and
IL, 1986.
(339 mm 2/m) while Eq. (11-14) re- Prestressed Concrete, Fifth Edition,
3. Mack, P., Aswad, A., et al., “Shear Precast/Prestressed Concrete Institute,
quires Av = 0.09 sq in./ft (191 mm2/m). Test Report: Elimination of Shear Re- Chicago, IL, 1999.
It would be desirable to investigate inforcement in Prestressed Double
10. AASHTO, Standard Specifications for
these two equations further and deter- Tees,” A report by Stanley Structures,
Highway Bridges, 16th Edition, Amer-
mine whether using the smaller value is Inc., Denver, CO, 1977, 36 pp.
ican Association of State Highway and
sufficient to develop a ductile behavior, 4. Aswad, A., Black, G., et al., “ Elimi- Transportation Officials, Washington,
as implied in the ACI Commentary. nation of Shear Reinforcement in Pre- DC, 1996, with 1997 to 2000 Interims.
stressed Double Tees,” Test Report
11. AASHTO, AASHTO LRFD Bridge
No. 2, Stanley Structures, Inc., Corpo-
Design Specifications, Second Edition,
ACKNOWLEDGMENT rate Engineering, Denver, CO, 1986,
American Association of State High-
This investigation was funded by 40 pp.
way and Transportation Officials,
three precast, prestressed concrete pro- 5. Aswad, A., and Burnley, G., “Omis- Washington, DC, 1998, with 1999,
sion of Shear Reinforcement in Pre- 2000, and 2001 Interims.
ducers in the mid-Atlantic region. The
stressed Double Tees,” PCI JOUR-
principal author and contributors 12. Olesen, S. E., Sozen, M. A., and Siess,
NAL, V. 34, No. 2, March-April 1989,
would like to gratefully acknowledge C. P., “Investigation of Prestressed Re-
pp. 48-65.
the financial and staff support of High inforced Concrete for Highway
6. “High Concrete Structures Introduces Bridges, Part IV: Strength in Shear of
Concrete Structures, Inc., Tindall Cor- Super Tee with Open House and Test Beams with Web Reinforcement,”
poration, and The Shockey Precast Loading,” PCI JOURNAL, V. 42, No. Bulletin 493, University of Illinois,
Group. The authors also want to thank 1, January-February 1997, p. 129. Engineering Experiment Station, Ur-
the PCI JOURNAL reviewers for their 7. Aswad, A., Burnley, G., Cleland, N., bana, IL, 1967.
thoughtful and constructive comments. Orndorff, D., and Wynings, C., “Tests

APPENDIX — NOTATION
A = area of the concrete section Mu = factored moment at a section
Ash = vertical reinforcement across potential horizontal N = tensile force normal to cross section occurring
cracks at bearings with V and to include the effects of creep and
Av = area of shear reinforcement within a unit distance shrinkage
Av,ci = area of shear reinforcement within a unit distance Pi = initial prestressing force at transfer
as required by Eq. (11-10), ACI 318-02 V = vertical reaction at support
D = dead loads, or related moments and shears Vu = factored shear at section
fc′ = specified compressive strength of concrete w1 = 0.85(1.4D + 1.7L)
fci′ = compressive strength of concrete at time of initial w2 = factored load per unit length = wu = 1.2D + 1.6L
prestress w3 = (1.2D + 1.6L)/0.9
fr = modulus of rupture of concrete x = distance of section from the left support
ft = maximum bottom tensile stress φ = strength reduction factor
L = live loads, or related moments and shears = 0.9 for flexure
MR = available moment capacity at a section = 0.75 for shear
MT = total moment at a section, sum of the double tee self-
weight moment and the superimposed loads moment

March-April 2004 77
78 PCI JOURNAL