You are on page 1of 21

ACI Structural Journal/November-December 1999 1033

ACI Structural Journal, V. 96, No. 6, November-December 1999.

Received August 7, 1998, and reviewed under Institute publication policies. Copyright
1999, American Concrete Institute. All rights reserved, including the making of copies
unless permission is obtained from the copyright proprietors. Pertinent discussion
including authors closure, if any, will be published in the September-October 2000 ACI
Structural Journal if the discussion is received by May 1, 2000.
This study presents experimental data of the reinforced concrete
corbels that, after cracking, were strengthened by external
prestressing or passive (nonprestressing) steel bars. The purpose
of the experimental tests was to recognize the influence of external
bars on cracking and on the load-carrying capacity of the tested
specimens. Nine corbels with various shear span-effective depth
ratio (a/d) were tested. The degree of prestress determined by the
value of the prestressing force was assumed constant for all three
tested prestressed specimens. To recognize the effect of
prestressing, the test results of the prestressed corbels were
compared with test results of the ultimate load and cracking of
corbels without strengthening, and also with corbels strengthened
by external barsthe same as the prestressing bars but without
external prestressing. Such passive bars are often used to
strengthen overloaded and cracked reinforced concrete (RC)
corbels. It turned out that the effectiveness of external prestressing
bars was significant, particularly for corbels with greater value
ratio (a/d 1.0). External prestressing is a useful solution for
strengthening corbels and reducing their crack widths. Until now,
there has been no recognition of crack morphology and the load-
carrying capacity of cracked corbels strengthened by external
prestressing bars. The results of the experimental tests also show
that the truss analogy, or shear-friction theory used for designing,
in practice cannot be applied (without modification) to the proper
determination of the ultimate load of the RC corbels strengthened
by external bars.
Keywords: corbels; external prestressing; strengthening; ultimate load.
The experimental data indicate that crack pattern and the
load-carrying capacity of RC corbels depend on the shear
span-to-effective depth ratio (a/d), on the amount of the main
reinforcement, and on the shape and amount of the stirrups.
The first vertical flexural crack (r
in Fig. 1) appears at a very
early stage. As shown in Fig. 1, in the corbel subjected to the
vertical load, the first crack appears approximately at the
load V 0.25V
(where V is the applied load and V
is the
collapse load). This first crack is situated at the face of the
column. Inclined cracks (r
) are developed later when the
vertical load varies between 0.4V
to 0.6V
. The practice
proves that the strengthening of RC corbels should be done
when inclined cracks start to develop, thus at load V in the
range of 0.5V
to 0.6V
According to Chakrabarti et al.,
and Tan et al.,
prestressing can delay flexural and shear cracking, so the
same effect can be expected at external prestressing of RC
corbels. External prestressing is more easily executed in
cases where a need of strengthening corbels exists. This
method protects corbels from corrosion, improves their
serviceability, and enhances their load-carrying capacity. In
practice, sometimes strengthening of even noncracked
corbels is needed. The modernization of construction when
the external load will increase is a good example. External
strengthening is relatively simple to carry out, is cost effective,
and is often used in practical cases.
The purpose of the experimental tests described in this
paper is to experimentally evaluate the real effectiveness of
the external prestressing bars used in strengthening the
Nine RC corbels (three sets of three corbels) with various
a/d (a/d = 1.0, 0.6, and 0.3) were tested. From each set one
of three corbels did not have external bars indicating that the
corbel was not strengthened. These three corbels were called
the basic corbels and were described as No. 4. The next three
corbels (No. 2) were prestressed by two external bars (the
diameter of each was d
= 25 mm and the yield strength f
= 396 MPa). The last three corbels (No. 3) had the same
external two bars as corbel No. 2 had, but these bars were
passive (they were nonprestressed during the test).
All the specimens were made from the same concrete
mixture. The average concrete cylinder compressive
Title no. 96-S115
Behavior of Corbels with External Prestressing Bars
Experimental Study
by Krystyna Nagrodzka-Godycka
Fig. 1Typical crack pattern of double corbel.
ACI Structural Journal/November-December 1999 1034
strength was f
= 25 MPa (f

30 MPa) at the time of
the specimen test. Ordinary portland cement 35, natural
sand, and an aggregate of 35 mm maximum size were used
for the concrete mixture (water-cement ratio [w/c] = 0.67).
The wooden forms were placed horizontally when the
concrete was laid.
The yield strength of ribbed steel bars of the diameter d
10 mm( No. 3) as the main reinforcement was f
= 493
MPa, and for smooth steel bars (as the stirrups) d
= 6 mm
was f
= 291.4 MPa. For the corner ribbed bars d
= 8 mm
and f
was 483 MPa. The yield strength of the longitu-
dinal column bars was 390 MPa (d
= 32 mm).
Details of specimens
The shapes of the tested corbels for each Set WI (a/d =
1.0), WII (a/d = 0.6), and WIII (a/d = 0.3), as well as the
arrangement and amount of reinforcement in specimens, are
shown in Fig. 2.
The reinforcement ratio for the main internal reinforce-
ment was equal to = 0.0104.
The steel strains were measured with electrical resistance
gages for each load increment to the failure. The gage length
was 20 mm. The strain gages were placed on the main rein-
forcement at the face of the column, and in the middle of
distance a (where a is the distance between vertical load and
the column face). As for the stirrups, the strain gages were
mounted at the supporting section (at the column face) and in
the half-length of each stirrup.
The concrete strains were measured with a mechanical
extensometer in the direction of the compressive principal
stresses and along the slope edge of the corbel. The readings
were taken for each load increment until the failure occurred.
The crack pattern was also recorded. Crack widths were
measured with a crack detection microscope with a magnifi-
cation of 40.
Test procedure
First, basic corbels No. 4 from each set were tested. All the
specimens were tested in an inverted position. The corbels
were subjected to vertical load V applied symmetrically at
the upper edge. The load was increased from V = 0 to V =
(where V
is the ultimate failure load for basic corbel).
When the ultimate load for the basic corbels was determined,
the next two corbels from each set (described as No. 2 or 3)
were tested until their failure. The load increment was
. When the load reached 60% of the ultimate load of
the comparable basic corbel, the tested corbel was unloaded.
Corbels No. 2 (one from each set) were prestressed by two
external bars (d
= 25 mm) placed at the upper tension edge
of the corbel on both sides. Prestressing was conducted by
means of special screws, as shown in Fig. 3. At the begin-
ning, the prestressing force in each external bar was 80 kN.
Therefore, the prestressing force was equal to P
= 0.53P
(where P
= A
The No. 3 corbels were strengthened by the same two bars
= 25 mm); however, they were passive. The tests were
run in the way previously described.
Crack morphology
The width of the first crack appearing at the tension junc-
tion was in a range between 0.04 and 0.1 mm. It depended on
Krystyna Nagrodzka-Godycka is a lecturer in the faculty of civil engineering at the
Technical University of Gdansk, Poland. She received her PhD in 1988. Her research
interests include reinforced concrete and prestressed structures, and behavior of
repaired and strengthened concrete structures.
Fig. 2Details of tested corbels (all dimensions in cm).
ACI Structural Journal/November-December 1999 1035
a/d. Inclined cracks developed in all the tested corbels under
the load V 0.6V
, and their maximum crack widths were 0.1
mm for Corbels WI and approximately 0.2 mm for Corbels
External prestressing (applied for the cracked Corbels WI-
2, WII-2, and WIII-2) caused the cracks to close. Later, when
the load acting on prestressed corbels was increased, the
cracks opened again. The crack widths were smaller,
however, than the corresponding crack widths of the corbels
strengthened by passive bars (WI-3, WII-3, and WIII-3). The
crack patterns of Corbel WI-2(WPS) with a/d = 1.0 (WPS =
corbel with prestressing bars) is shown in Fig. 4.
Figures 5 and 6 present the crack patterns of Corbel WII-
2(WPS) with a/d = 0.6 and for Corbel WIII-2(WPS) with a/d
= 0.3, respectively.
The maximum crack widths for Corbels WI-2, WI-3;
Corbels WII-2, WII-3; and Corbels WIII-2, WIII-3 at the
two levels of the load were V = 0.6
and V = 0.9V
is the ultimate failure load for the corbel with external
bars), are shown in Fig. 7.
Fig. 4Scheme of crack propagation of WI-2(WPS): (a)
at V =0.6V
; and (b) at V
Fig. 5Scheme of crack propagation of WII-2(WPS): (a) at
; and (b) at V
Fig. 3Anchorage of external bars for prestressing.
1036 ACI Structural Journal/November-December 1999
Stress-strain and load-carrying capacity
The reason for the failure of Corbel WI and WII sets (a/d
= 1.0 and 0.6) was diagonal splitting, independent from the
type of external bars (prestressing or passive). The mode of
the failure for Corbels WIII (a/d = 0.3) can be described as
the mode between diagonal splitting and shear failure.
The stresses of the main reinforcement of corbels under the
load V = 0.6V
were averaging between 0.6f
and 0.7f
The stresses at the same load in stirrups were 0.8f
compressive strains of concrete (
) were from 0.6 to 1.0%.
Figures 8 through 10 present the stresses of the main
internal reinforcement of the corbels and the stresses in
external prestressing bars for prestressed corbels (WPS).
At the load V = 0.9V
u ,
the stresses of the internal main
reinforcement close to the tension zone almost reached the
Fig. 6Scheme of crack propagation of WIII-2(WPS):
(a) at V = 0.06V
; and (b) at V
Fig. 7Crack widths of tested corbels: (a) at
V = 0.6V
; (b) at V = 0.9V
Fig. 8Stresses of main internal reinforcement and external
prestressing bars for corbel WI-2(WPS).
ACI Structural Journal/November-December 1999 1037
Fig. 9Stresses of main internal reinforcement and external
prestressing bars for Corbel WII-2(WPS).
Fig. 10Stresses of main internal reinforcement and
external prestressing bars for Corbel WIII-2(WPS).
Fig. 11Stresses of stirrups for Corbel WII-2(WPS).
Fig. 12Stresses of stirrups for Corbel WIII-2(WPS).
Fig. 13Stresses of main internal reinforcement and
external passive bars for Corbel WI-3(WPP).
yield stress while the stresses in the external prestressing
bars reached approximately 280 MPa (70% f
Figures 11 and 12 present the stresses of the stirrups for
the prestressed corbels: WII-2(WPS) and WIII-2(WPS). The
stresses in stirrups at the ultimate failure load reached the
yield stress.
The stresses of the main internal reinforcement and the
external strengthening passive bars of the corbels (WPP) are
shown in Fig. 13 through 15. The test results indicated big
differences between the stresses for prestressing and passive
bars. (70 MPa for the passive bars and 280 MPa for the
prestressing bars).
1038 ACI Structural Journal/November-December 1999
Figure 16 presents the effectiveness of the corbels with the
external bars, expressed as the ratio of the load-carrying
capacity of corbels with external bars to the load-carrying
capacity of basic corbels. Table 1 presents the ultimate
failure load for the tested corbels.
Comparing the authors test results with the calculated
ultimate load obtained from a few well-known design
methods based on truss model or shear-friction theory, it was
found that the ultimate load could not be properly estimated.
The analysis results of the load-carrying capacity of the
prestressed corbels with external bars based on a few
selected methods
(the methods are summarized in the
) are given in Table 2. They prove the methods
overestimate or underestimate the value of the load-carrying
capacity (Table 2).
The test results indicate that both the strain and the stress
increase were greater in the external prestressing bars in
comparison with the external passive bars (Fig. 8 through 10
and Fig. 13 through 15).
The stresses in external prestressing bars reached V

. These stresses were almost half as much as the
stresses in the internal main reinforcement of the prestressed
corbels. At the same load, the stresses in the external passive
bars reached only 0.17f
, therefore, they were from six to
seven times less than the stresses of the internal main rein-
forcement of the corbels with external passive bars.
The Appendix is available in xerographic or similar form from ACI headquarters,
where it will be kept permanently on file, at a charge equal to the cost of reproduction
plus handling at time of request.
Fig. 15Stresses of main internal reinforcement and
external passive bars for Corbel WIII-3(WPP).
Fig. 16Effectiveness of external bars for tested corbels.
Table 1Experimental ultimate load for tested
Corbel a/d
Load at failure V
, kN
WI-4 1.0 250
WI-2(WPS) 1.0 350
WI-3(WPP) 1.0 275
WII-4 0.6 475
WII-2(WPS) 0.6 525
WII-3(WPP) 0.6 500
WIII-4 0.3 650
WIII-2(WPS) 0.3 700
WIII-3(WPP) 0.3 650
Fig. 14Stresses of main internal reinforcement and
external passive bars for Corbel WII-3(WPP).
ACI Structural Journal/November-December 1999 1039
To obtain a reasonable agreement between calculated and
experimental load-carrying capacity, the differences
between the stresses of the internal main reinforcement and
external strengthening reinforcement should be taken into
account. This agreement could be reached assuming a coef-
ficient k = 0.5 reduces stresses in the external prestressing
bars on the load-carrying capacity (Table 3).
Taking this assumption into account, it is possible to estimate
the ultimate load of corbels with a/d = 1.0 according to Kriz
and Raths method
and for the shortest corbel (a/d = 0.3)
using Walravens method.
In both cases, for the satisfactory
agreement between the calculated and experimental ultimate
load of the corbels, the coefficient k = 0.5, should be
The authors procedure,
based on Mohrs failure criterion,
could be very useful in the case of strong main reinforce-
ment, when the concrete strength determines the load-carrying
capacity of the corbel. To obtain the properly estimated
calculated load-carrying capacity, the compression zone (c)
should be limited. For example: for the ratio a/d = 1.0
= 0.4d; for a/d = 0.6 c
= 0.6d and for a/d = 0.3
= 0.9d. For practical purposes, the load-carrying
capacity of the prestressed corbels can be calculated prop-
erly by taking into account limited c
and assuming k =
0.5. The results of the calculation are given in Table 3.
The experimental test results indicate that the effectiveness
of the external prestressing depends on a/d. For a/d = 1.0, the
ultimate load due to prestressing increased by 40%. This
effectiveness decreased with the decrease of a/d. For a/d = 0.6
or 0.3 (shorter corbels), the ultimate load increased by a
maximum of only 12%.
The crack widths due to external prestressing also depended
on a/d. Crack widths decreased with the increase of a/d.
At the failure of the corbels, the yield stresses in each case
were reached in part of the internal main reinforcement, and
horizontal stirrups were placed inside the corbel. The
maximum stresses in external prestressing bars were approxi-
mately 30% smaller than the yield stress (
= 0.7f
If, assuming for calculation purposes, that stresses in the
internal main reinforcement and external strengthening bars
are equal to the yield stresses, the load-carrying capacity is
overestimated. These improper overestimates could be
particularly large when strengthening the corbels using the
passive bars. Thus, to calculate the load-carrying capacity
of the corbels with external prestressing bars, the stress in
the prestressing bars might be approximately assumed
= 0.5f
This research was supported by Grant No. 7-7233-92-03 from the Polish
State Research Committee (KBN), which is gratefully acknowledged.
1. Chakrabarti, P. R.; Farahani, D. J.; and Kashou, S. I., Reinforced and
Precompressed Concrete CorbelsAn Experimental Study, ACI Structural
Journal, V. 86, No. 4, July-Aug. 1989, pp. 405-412.
2. Tan, K. H., and Mansur, M. A., Partial Prestressing in Concrete Corbels
and Deep Beams, ACI Structural Journal, V. 89, No. 3, May-June 1992,
pp. 251-262.
3. Kriz, L. B., and Raths, C. H., Connections in Precast Concrete
StructuresStrength of Corbels, Journal of the Prestressed Concrete Institute,
V. 10, No. 1, Feb. 1965, pp. 16-61.
4. Franz, G., Column Corbels, Beton und Stahlbetonbau, V. 71, No. 4, Apr.
1976, pp. 93-102. (in German)
5. Hagberg, T., Design of Concrete Brackets: On the Application of the Truss
Analogy, ACI JOURNAL, Proceedings V. 80, No. 1, Jan.-Feb. 1983, pp. 3-12.
6. Walraven, J.; Frenay, J.; and Pruijssers, A., Influence of Concrete Strength
and Load History on the Shear Friction Capacity of Concrete Members, Journal
of the Prestressed Concrete Institute, V. 32, No. 1, Jan.-Feb. 1987, pp. 66-84.
7. Nagrodzka-Godycka, K., Contribution to Design of Reinforced Concrete
Corbels under Short-Term Load on Upper Edge, Archives of Civil Engineering
(Warsaw), V. 37, No. 2, 1991, pp. 221-248. (in Polish)
Table 2Comparison of test results with the calculated ultimate load kN
Method corbel a/d
Kriz and
u, exp
u, steel
u, concr
u, steel
u, concr
WI-4 1.0 205 308 287 210 288 288 250
WI-2 (WPS) 1.0 262 552 494 210 486 486 350
WII-4 0.6 287 471 444 326 471 432 432 475
WII-2 (WPS) 0.6 366 776 704 326 654 699 699 525
WIII-4 0.3 379 727 664 428 475 639 639 650
WIII-2 (WPS) 0.3 485 1246 1015 428 659 1246 712 700

Table 3Ultimate load V
with k = 0.5 assumption
Method corbel Kriz and Raths
u, calc
u, exp Walravens
u, calc
u, exp Authors
u, calc
/ V
u, exp
u, exp
a/d = 1.0
237 0.68
c = 0.4d
= 298
0.85 350
a/d = 0.6
331 0.63 571 1.09
c = 0.6d
= 472
0.9 525
a/d = 0.3
438 0.63 576 0.82
c = 0.9d
0.87 700