Article publié par le Laboratoire de Construction en Béton de l'EPFL

 
Paper published by the Structural Concrete Laboratory of EPFL

Title: Discussion of paper 'Factors affecting strength of elements designed using
strut-and-tie models'
Authors: Muttoni A., Kostic N., Fernández Ruiz M.
Published in: ACI Structural Journal
Volume: 104-S36
Pages: pp. 233-235
Country: USA
Year of publication: 2008
Type of publication: Peer reviewed journal article
EPFL InfoScience link: https://infoscience.epfl.ch/record/116129

Please quote as: Muttoni A., Kostic N., Fernández Ruiz M., Discussion of paper 'Factors

affecting strength of elements designed using strut-and-tie models', ACI
Structural Journal, 104-S36, USA, 2008, pp. 233-235.

[Muttoni08] Downloaded from 190.26.37.11 on 20.08.2020 23:25
 DISCUSSION
Disc. 104-S26/From the May-June 2007 ACI Structural Journal, p. 267

Factors Affecting Strength of Elements Designed Using Strut-and-Tie Models. Paper by Sergio F. Breña and
Micah C. Morrison

Discussion by Emil de Souza Sánchez Filho, Júlio J. Holtz Silva Filho, and Maria Teresa Gomes Barbosa
ACI member, DSc, Professor, Fluminense Federal University, Rio de Janeiro, Brazil; DSc. Structural Engineer, Rio de Janeiro, Brazil; ACI member, DSc, Professor, Federal University of Juiz de
Fora, Minas Gerais, Brazil.

The authors have made an interesting contribution to the ments, similar to what is currently done for reinforced concrete
experimental study of strut-and-tie models. However, the deep beams with openings.
discussers would like to address some aspects in this study: 4. All four models shown in Fig. 2 of the paper are
1. The discussers have reviewed several publications composed by superimposed models, but this information is
listed in the References section of the paper regarding the not given by the authors. Strut-and-tie modeling is a rational
fundamental concepts of strut-and-tie models. An important and simple method for analysis and design, but for Models 2A
parameter of these modes is the concrete effectiveness factor and 2B, it is not true. These two models are very complex
that depends on the strut type, reinforcement’s arrangements, and inappropriate for an engineering design.
and so on, but the authors did not consider this parameter in the 5. All three possible different D regions are cataloged in
explanation of their modeling. The concrete effectiveness Jennewein and Schäfer (1992), where the expressions for
factor ν is an essential parameter that needs to be inserted forces, angles of the struts, and ties are given for each D
into the development of the plasticity theory. The best region, and it is a waste of time trying to compare with
agreement between theory and experimental data is obtained another special truss model for D regions near the openings,
by the appropriate choice of ν, but the authors did not which is the case of models of Specimens 2A and 2B.
provide the value used in their study. The authors statement, 6. The authors failed to explain the theoretical considerations
“The strength of the nodes, struts, and ties was calculated about the models. It would be interesting to know the values of
using procedures in Appendix A of the 2002 ACI Code the struts angles adopted in the analyses. Another consideration
(ACI Committee 318 2002)” is very unclear because this that requires a better explanation is the optimization of the
strength depends on the quite a lot of parameters and this models. Further precise information of the models conception is
code provides several expressions to calculate the necessary—for example, struts lengths and widths, node
concrete effectiveness factor. dimensions, types of the basic models that are superimposed,
and details about nonlinear analyses undertaken.
2. The authors adopted the strength reduction factor φ = 7. The discussers believe that it is impossible to check the
0.75 to find the ties armors, which is an inadequate and conser- theoretical values given in Tables 2 and 3 of the paper, and
vative approach for this type of research. The abundance of several topics of the paper are confusing. Therefore, the
armors in several regions of the beams is corroborated by the discussers would greatly appreciate if the authors could
low strain measured in the several ties. All specimens have provide some complementary information about the research.
unusual reinforcement arrangements. The secondary armors
on Specimens 1A and 1B certainly are responsible for the
REFERENCES
great discrepancies among theoretical and experimental Bergmeister, K.; Breen, J. E.; Jirsa, J. O.; and Kreger, M. E., 1993,
results. This fact is corroborated by the authors’ approximate “Detailing for Structural Concrete,” Research Report 1127-3F, University
procedure to estimate the contributions of these secondary of Texas at Austin, Austin, TX.
reinforcements, substantially reducing these differences. Brown, M. D., 2005, “Design for Shear in Reinforced Concrete Using
Strut-and-Tie and Sectional Models,” PhD thesis, University of Texas at
3. Specimens 2A and 2B were designed for an ultimate Austin, Austin, TX.
load with very complicated models, and this is not the basic Foster, S. J., and Mailk, A. R., 2001, “Evaluation of Compression Failures
idea of the strut-and-tie model approach. Instead of very in RC Non-Flexural Members,” UNICIV Report No. r-401, University of
New South Wales, Australia.
complex modeling for these specimens, it would be more Jennewein, M., and Schäfer, K., 1992, “Standardisierte Nachweise von
interesting to use the simplest steel reinforcement arrange- häufigen D-Bereichen,” DafStb, Heft 430.

Disc. 104-S26/From the May-June 2007 ACI Structural Journal, p. 267

Factors Affecting Strength of Elements Designed Using Strut-and-Tie Models. Paper by Sergio F. Breña and
Micah C. Morrison

Discussion by Rafael A. de Souza

ACI Member, PhD, Associate Professor, Universidade Estadual de Maringá, Brazil.

The experimental work conducted by the authors has conclusion as it can substantially affect the design based on
shown that shrinkage and temperature reinforcement strut-and-tie models (STMs). The discusser would like to
contributes significantly to the strength of reinforced offer the following comments to emphasize this specific
concrete deep beams with web openings. It is a very important conclusion made by authors:

232 ACI Structural Journal/March-April 2008

 Fig. D—Strut-and-tie model proposed for complex deep beam.
Fig. A—Complex deep-beam subjected to geometric
irregularities.
3. Take, for example, the complex deep beam presented in
Fig. A, subjected to a design load Fd = 100 kN (22 kips). This
beam has a width of 0.25 m (9.8 in.) and is supposed to be
molded with concrete with a compressive strength of 20 MPa
(2900 psi). Based on the elastic analysis shown in Fig. B and C,
the STM presented in Fig. D was developed. It is a very
complex STM and the time required to sketch the proposed
truss is a very demanding task. The time required for this
activity is not compatible with the time available for the
Fig. B—Principal tension stress for deep beam. structural engineers at their offices;
4. Based on the last paragraph, for many complex situations,
it is better to conduct a nonlinear analysis first rather than
expend a great deal of time trying to develop a truss model
based on the STM. Using some resources of nonlinear analysis
available in many package software programs, the mandatory
minimum reinforcement can be taken into account in the
design, and only some tie positions will need to be
strengthened. For the specific case presented in Fig. A, for
example, it is easy to prove by using nonlinear analysis that
minimum reinforcement control for shrinkage and temperature
Fig. C—Principal compression stress for deep beam. can carry approximately 60% of the design load. This design
procedure is attractive for structural engineers because it is
less time-consuming than STM;
1. STMs have usually been taken as a panacea for solving 5. Also, using nonlinear analysis resources for structural
any complex problem that arises when designing structural concrete, a quick estimate for both the ultimate and
concrete. Undoubtedly, STM is a very powerful tool, but serviceability limit state (usually a problem when using
designers should be aware that sometimes it can lead to an STMs) is available. Until now, the development of STMs has
exaggerated design, being that the collapse loads are much been focused on the ultimate limit state, and the serviceability
higher than the design loads; limit state is only implicitly considered through the selection
2. Taking into account the mandatory recommendations of of appropriate STMs; and
many normative codes about minimum reinforcement 6. Finally, taking into account the powerful computational
control for shrinkage and temperature (usually assumed as resources available today, STM could be used as a hand-made
10% of the concrete section area for each side of an structural verification proof to certify the answers provided by many
element), some idealized truss models may be substantially specific software programs. It seems to be a routine very
changed from the original truss sketch. For these cases, close to that one required by structural offices. Besides, the
minimum reinforcement may be higher than the reinforcement minimum reinforcement for crack control could always be
provided for the principal ties, and for this reason, an considered in design, providing a more economic answer for
exaggerated collapse load may be found; complex problems.

Disc. 104-S26/From the May-June 2007 ACI Structural Journal, p. 267

Factors Affecting Strength of Elements Designed Using Strut-and-Tie Models. Paper by Sergio F. Breña and
Micah C. Morrison

Discussion by A. Muttoni, N. Kostic, and M. Fernández Ruiz

École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland.

The authors of the paper investigated the suitability of of four 1/4-scale tests. The methodology followed by the
design of structural concrete members using strut-and-tie authors is much appreciated by the discussers. In the
models inspired by linear elastic (uncracked) stress fields. discussers’ opinion, papers providing experimental data that
The accuracy of this approach is checked against the results can be compared with strut-and-tie models (or stress fields)

ACI Structural Journal/March-April 2008 233

are still necessary to advance the state of knowledge on this conditions (Muttoni et al. 1997). According to this approach,
topic and to develop new strategies for development of suitable a licit mechanism first has to be selected for the member
strut-and-tie models and stress fields. (which yields an upper-bound solution of the theory of
In Table 2 of the paper, the theoretical strengths Qth plasticity). Second, a stress field is developed in the free-
expected by the authors according to strut-and-tie models bodies of the mechanism, respecting its kinematics and the
inspired by the uncracked stress field of the members are plasticity criterion (which yields a lower-bound solution of
compared with actual values measured in the tests Qtest. The
ratios between them (Qtest/Qth) vary between 1.72 and 3.19
(summarized in the first row of Table A), showing too
conservative estimates of the strength for the various
specimens. The differences are, according to the authors of
the paper, due to four phenomena:
1. The design method (strut-and-tie model) is a lower
bound solution of the theory of plasticity, thus leading to
conservative estimates of the actual failure load;
2. The contribution of the secondary (minimal) reinforcement
is neglected in the strength of the strut-and-tie models;
3. Significant stress redistributions developed during tests
(confirmed by test measurements); and
4. Also, according to the authors, concrete contribution to
tie strength could have played a non-negligible role in the
strength of the member.
Considering the influence of the minimal reinforcement
and performing a nonlinear analysis to account for stress
redistributions, the previous ratios (Qtest/Qth) are improved
by the authors and vary between 1.49 and 2.18 (refer to
Table A, second row).
The discussers are in complete agreement with the first three
phenomena mentioned by the authors. The differences between
the measured and the predicted strengths can be mostly
explained due to the fact that the selected strut-and-tie models,
although licit for design (because they give a lower-bound
solution), differ notably from the actual stress fields at failure.
In this sense, the minimal reinforcement of Specimens 1A and
1B shows a significant influence on the actual stress field and,
thus, on the strength of the members (refer to Table A). Also,
stress redistributions from the uncracked stress field to the
cracked stress field at failure (including yielding of the
reinforcement and changes in the angle and the location of
compression struts) has a clear influence on the strength
of the member.
On the contrary, in the discussers’ opinion, the fourth
phenomenon (concrete contribution to tie strength) can be
neglected in comparison with the other phenomena because
significant crack widths develop at failure in RC members.
Furthermore, this contribution is not reliable and should not
be considered in plastic analyses (Muttoni et al. 1997).
In addition to the previous phenomena, the discussers
think that the differences between the expected and the
measured strengths are also due to the fact that the proposed
strut-and-tie models do not account for a realistic kinematics Fig. E—Development of strut-and-tie model and stress
at failure. An approach overcoming most of the previous fields. Assumed kinematics at failure and discontinuous
problems (minimal reinforcement, stress redistributions, and stress fields in critical free-bodies of Specimens: (a) 1A;
suitable kinematics) can be easily developed, leading to (b) 2A; (c) 1B; and (d) 2B; and continuous stress fields
satisfactory results both at failure and under serviceability of Specimens: (e) 1A; (f) 2A; (g) 1B; and (h) 2B.

Table A—Ratio between measured and estimated failure loads for various specimens
Qtest /Qth 1A 1B 2A 2B Average Coefficient of variation
Breña and Morrison (strut-and-tie models inspired by linear-elastic
3.19 2.98 1.72 1.74 2.41 0.33
uncracked stress field)
Breña and Morrison (strut-and-tie models where minimal reinforcement is
1.72 2.18 1.49 1.49 1.72 0.19
considered and nonlinear analysis is performed)
Discontinuous stress field accounting for kinematics at failure 1.09 1.10 1.04 1.02 1.06 0.04
Continuous stress field 1.17 1.27 1.00 1.03 1.11 0.09

234 ACI Structural Journal/March-April 2008

the theory of plasticity). Consequently, this approach leads to strength of struts in strut-and-tie models. Past experimental
an exact solution according to the theory of plasticity. evidence suggests that the effective concrete strength is
For instance, Fig. E(a) to (d) show some possible failure lower than the uniaxial compression strength when concrete
mechanisms for the various specimens. According to these is subjected to transverse tension that induces cracking in
failure mechanisms, specimens with two openings fail on the conditions of biaxial stresses. Because the intent was to use
right side, whereas specimens with one opening fail on the available design recommendations to develop strut-and-tie
left side. Starting from such kinematics, and accounting for models in the design phase of the research, the authors chose
the reinforcement layout (including the minimal reinforcement) to use the effective strength factors included in Appendix A
and concrete strength, discontinuous stress fields can be of ACI 318-02. Effective strength factors in ACI 318-02 are
developed in the critical free-bodies (free-bodies governing calculated as 0.85βs, where βs ranges between 1.0 and 0.4,
the strength of the member), as shown in Fig. E(a) to (d). depending primarily on the level of transverse tension that
Such stress fields allow the location of the critical nodal the strut will experience. The authors used factors corre-
regions to be determined and, thus, realistic angles and sponding to either prismatic or bottle-shaped struts
locations of the critical struts to be estimated (indicated depending on the location in the specimens as indicated in
in dark gray in Fig. E). Table B.
When checking the strength of a member, as in this case, 2. The primary objective of the paper was to quantify
the suitable failure mechanism is the one having the lowest sources of potential overstrength when using strut-and-tie
strength. In doing so, the mechanisms shown in Fig. E(a) to models for design as stated in the Research Significance
(d) (which are found critical) lead to the failure loads section of the paper. To achieve this goal, the authors used a
detailed in Table A (third row) in excellent agreement with strength reduction factor φ = 0.75, in accordance with
the test results. This methodology can also be followed for ACI 318-02 for design of the laboratory specimens. This
design purposes, leading to stress fields with a satisfactory factor was later removed, or in other words, φ = 1.0 was used
behavior at SLS (crack control) and accounting for the when strength evaluation of the specimens was being
kinematics at failure (Muttoni et al. 1997). conducted after the tests. The procedure used to estimate
An alternative approach to account for the kinematics of the strength of the as-built specimens is discussed in
structural concrete members is the development of continuous detail in the paper.
stress fields (Fernández Ruiz and Muttoni 2007), where 3 and 4. The authors agree that the strut-and-tie models
compatibility conditions for concrete and for reinforcing that resulted in some cases were more complicated than
steel are introduced. Figures E(e) to (h) show the continuous would be desired. The intent was to examine whether
stress fields obtained for the four specimens. The ratios differences in models resulting in different reinforcement
between the measured and the estimated failure loads patterns would affect measured strength of the specimens.
according to continuous stress fields are also shown in Table A Even if nontraditional models are used for design, the
(fourth row). The results obtained are somewhat more reinforcement can be resolved in two orthogonal directions,
conservative than those obtained with discontinuous stress as is commonly done in practice. As can be observed from
fields accounting for kinematics at failure. This is due to the the reported test results, the specimens failed at approximately
fact that the strength of concrete is reduced in continuous the same load, so no apparent effect on reinforcement pattern
stress fields to account for transverse cracking. was identified for these specimens.
In any case, both approaches accounting for kinematics 5. The authors do not have a copy of the reference
provide a very good agreement with the actual failure loads mentioned in this paragraph, so we cannot comment on the
and show to be more accurate than those obtained with discussers’ statement. It is difficult, however, to think of a
elastically inspired strut-and-tie models. catalog with only three different D-regions that would
To conclude, the discussers would like to highlight that encompass all possible stress fields and boundary conditions
simple, suitable, and satisfactory strut-and-tie models and that might be encountered in practice.
stress fields can be developed if a realistic kinematics at failure 6 and 7. The discussers request additional information
is considered. This allows one to overcome most difficulties about the models used in the tests. This information was not
found when developing strut-and-tie models inspired by the provided in the original manuscript because of space
linear elastic (uncracked) stress field of a member. constraints. Complementary information is summarized in
Table B of this closure.
REFERENCES
Fernández Ruiz, M., and Muttoni, A., 2007, “On Development of Suitable
Stress Fields for Structural Concrete,” ACI Structural Journal, V. 104, Closure to discussion by de Souza
No. 4, July-Aug., pp. 495-502. In the authors’ experience, designs based on the strut-and-
tie method result in elements with higher strength than the
AUTHORS’ CLOSURE design load. The authors also believe, as the discusser
The authors appreciate the interest expressed by three suggests, that minimum reinforcement required to control
groups of discussers on our recently published paper. The thermal and shrinkage cracking will contribute to strength of
response to each group will be addressed in a separate some types of structural elements that are designed using
section as follows. strut-and-tie models (such as deep beams). Secondary
reinforcement is also used to prevent wide cracks from
Closure to discussion by de Souza Sánchez degrading the strength of struts after diagonal cracking
Filho et al. occurs. The effect that this type of reinforcement has on
Using the same order that the discussers’ used, the authors element strength needs to be studied in more detail.
would like to clarify the issues brought up by the discussers: The discusser suggests using nonlinear finite element
1. As the discussers point out, the concrete effectiveness analysis to calculate effects of crack control reinforcement in
factor is an important parameter when computing the structural concrete elements with complicated geometries.

ACI Structural Journal/March-April 2008 235

Table B—Summary of strut geometry and forces Table B (cont.)—Summary of strut geometry and
for strut-and-tie design models forces for strut-and-tie design models
Effective Design Effective Design
Strut force, width, strength, Strut force, width, strength,
Specimen Strut no. kN (kip) βs mm (in.) kN (kip) Specimen Strut no. kN (kip) βs mm (in.) kN (kips)
S1 –132 (–29.70) 1.000 168 (6.61) 330 (74.19) S14 –92 (–20.60) 1.000 152 (6.00) 312 (70.18)
S2 –53 (–11.87) 0.750 102 (4.00) 150 (33.66) S15 –80 (–18.05) 1.000 152 (6.00) 312 (70.18)
S3 –53 (–11.88) 0.750 102 (4.00) 150 (33.66) S16 –61 (–13.60) 1.000 102 (4.00) 208 (46.79)
S4 –49 (–11.12) 0.750 102 (4.00) 150 (33.66) S17 –24 (–5.38) 1.000 51 (2.00) 104 (23.39)
S5 –49 (–11.12) 0.750 102 (4.00) 150 (33.66) S18 –39 (–8.67) 1.000 102 (4.00) 208 (46.79)
S6 –53 (–11.88) 0.750 114 (4.50) 169 (37.87) S19 –74 (–16.60) 1.000 133 (5.25) 273 (61.41)
S7 –53 (–11.87) 0.750 114 (4.50) 169 (37.87) S20 –52 (–11.63) 1.000 102 (4.00) 208 (46.79)
S8 –85 (–19.09) 1.000 127 (5.00) 250 (56.10) S21 –34 (–7.60) 1.000 51 (2.00) 104 (23.39)
S9 –8 (–1.87) 1.000 15 (0.60) 30 (6.73) S22 –9 (–1.97) 1.000 0 (0.00) 0 (0.00)
S10 –11 (–2.56) 1.000 25 (1.00) 50 (11.22) S23 –43 (–9.57) 1.000 102 (4.00) 208 (46.79)
S11 –26 (–5.86) 1.000 38 (1.50) 75 (16.83) S24 –16 (–3.69) 1.000 38 (1.50) 78 (17.55)
1A
S12 –14 (–3.12) 1.000 25 (1.00) 50 (11.22) S25 –40 (–9.07) 1.000 76 (3.00) 156 (35.09)
S13 –20 (–4.43) 1.000 38 (1.50) 75 (16.83) S26 –60 (–13.43) 1.000 127 (5.00) 260 (58.48)
S14 –20 (–4.42) 1.000 51 (2.00) 100 (22.44) S27 –60 (–13.40) 1.000 127 (5.00) 260 (58.48)
2A
S15 –13 (–2.91) 1.000 25 (1.00) 50 (11.22) S28 –99 (–22.18) 1.000 127 (5.00) 260 (58.48)
S16 –18 (–4.12) 1.000 38 (1.50) 75 (16.83) S29 –42 (–9.36) 1.000 76 (3.00) 156 (35.09)
S17 –18 (–4.12) 1.000 32 (1.25) 62 (14.03) S30 –9 (–2.02) 1.000 13 (0.50) 26 (5.85)
S18 –32 (–7.19) 1.000 51 (2.00) 100 (22.44) S31 –11 (–2.56) 1.000 25 (1.00) 52 (11.70)
S19 –31 (–6.93) 1.000 51 (2.00) 100 (22.44) S32 –26 (–5.83) 1.000 38 (1.50) 78 (17.55)
S20 –34 (–7.69) 1.000 57 (2.25) 112 (25.25) S33 –13 (–2.97) 1.000 25 (1.00) 52 (11.70)
S21 –47 (–10.61) 1.000 76 (3.00) 150 (33.66) S34 –19 (–4.20) 1.000 25 (1.00) 52 (11.70)
S22 –47 (–10.61) 1.000 102 (4.00) 200 (44.88) S35 –19 (–4.20) 1.000 38 (1.50) 78 (17.55)
S1 –140 (–31.50) 1.000 168 (6.61) 330 (74.16) S36 –13 (–2.94) 1.000 25 (1.00) 52 (11.70)
S2 –68 (–15.36) 0.750 127 (5.00) 187 (42.08) S37 –21 (–4.65) 1.000 38 (1.50) 78 (17.55)
S3 –68 (–15.36) 0.750 127 (5.00) 187 (42.08) S38 –20 (–4.41) 1.000 38 (1.50) 78 (17.55)
S4 –64 (–14.38) 0.750 108 (4.25) 159 (35.76) S39 –42 (–9.35) 1.000 76 (3.00) 156 (35.09)
S5 –64 (–14.38) 0.750 108 (4.25) 159 (35.76) S40 –55 (–12.32) 1.000 102 (4.00) 208 (46.79)
S6 –68 (–15.36) 0.750 140 (5.50) 206 (46.28) S41 –55 (–12.32) 1.000 102 (4.00) 208 (46.79)
S7 –68 (–15.36) 0.750 140 (5.50) 206 (46.28) S1 –147 (–33.00) 1.000 164 (6.46) 323 (72.48)
S8 –90 (–20.25) 1.000 127 (5.00) 250 (56.10) S2 –102 (–23.02) 1.000 152 (6.00) 300 (67.32)
S9 –47 (–10.67) 1.000 76 (3.00) 150 (33.66) S3 –32 (–7.19) 1.000 76 (3.00) 150 (33.66)
S10 –32 (–7.26) 1.000 51 (2.00) 100 (22.44) S4 –26 (–5.91) 1.000 76 (3.00) 150 (33.66)
S11 –6 (–1.36) 1.000 13 (0.50) 25 (5.61) S5 –26 (–5.91) 1.000 76 (3.00) 150 (33.66)
1B
S12 –26 (–5.81) 1.000 38 (1.50) 75 (16.83) S6 –72 (–16.29) 1.000 170 (6.70) 335 (75.17)
S13 –37 (–8.26) 1.000 76 (3.00) 150 (33.66) S7 –168 (–37.70) 1.000 254 (10.00) 499 (112.20)
S14 –37 (–8.24) 1.000 76 (3.00) 150 (33.66) S8 –3 (–0.69) 1.000 102 (4.00) 200 (44.88)
S15 –20 (–4.48) 1.000 51 (2.00) 100 (22.44) S9 –52 (–11.72) 1.000 76 (3.00) 150 (33.66)
S16 –43 (–9.57) 1.000 76 (3.00) 150 (33.66) S10 –131 (–29.51) 1.000 330 (13.00) 649 (145.86)
S17 –28 (–6.34) 1.000 51 (2.00) 100 (22.44) S11 –83 (–18.65) 1.000 152 (6.00) 300 (67.32)
S18 –28 (–6.34) 1.000 76 (3.00) 150 (33.66) S12 –28 (–6.20) 1.000 127 (5.00) 250 (56.10)
S19 –43 (–9.57) 1.000 76 (3.00) 150 (33.66) S13 –94 (–21.21) 1.000 127 (5.00) 250 (56.10)
S20 –20 (–4.48) 1.000 38 (1.50) 75 (16.83) S14 –33 (–7.33) 1.000 51 (2.00) 100 (22.44)
2B
S21 –50 (–11.25) 1.000 76 (3.00) 150 (33.66) S15 –20 (–4.49) 1.000 51 (2.00) 100 (22.44)
S22 –50 (–11.25) 1.000 102 (4.00) 200 (44.88) S16 0 (0.00) 1.000 51 (2.00) 100 (22.44)
S1 –154 (–34.50) 1.000 164 (6.46) 336 (75.56) S17 –55 (–12.36) 1.000 51 (2.00) 100 (22.44)
S2 –107 (–24.07) 1.000 127 (5.00) 260 (58.48) S18 –36 (–7.99) 1.000 64 (2.50) 125 (28.05)
S3 –71 (–16.05) 1.000 127 (5.00) 260 (58.48) S19 –25 (–5.65) 1.000 44 (1.75) 87 (19.64)
S4 –36 (–8.02) 1.000 76 (3.00) 156 (35.09) S20 –36 (–7.99) 1.000 76 (3.01) 150 (33.77)
S5 –127 (–28.59) 1.000 159 (6.25) 325 (73.11) S21 –45 (–10.11) 1.000 114 (4.50) 225 (50.49)
S6 –41 (–9.19) 1.000 127 (5.00) 260 (58.48) S22 –16 (–3.62) 1.000 102 (4.00) 200 (44.88)
2A S7 –36 (–8.02) 1.000 76 (3.00) 156 (35.09) S23 –10 (–2.29) 1.000 127 (5.00) 250 (56.10)
S8 –36 (–8.02) 1.000 76 (3.00) 156 (35.09) S24 –15 (–3.43) 1.000 152 (6.00) 300 (67.32)
S9 –36 (–8.02) 1.000 76 (3.00) 156 (35.09) S25 –27 (–6.13) 1.000 51 (2.00) 100 (22.44)
S10 –8 (–1.78) 1.000 25 (1.00) 52 (11.70) S26 –57 (–12.71) 1.000 51 (2.00) 100 (22.44)
S11 –105 (–23.54) 1.000 127 (5.00) 260 (58.48) S27 –52 (–11.79) 1.000 70 (2.75) 137 (30.86)
S12 –3 (–0.69) 1.000 25 (1.00) 52 (11.70) S28 –52 (–11.79) 1.000 76 (3.00) 150 (33.66)
S13 –79 (–17.78) 1.000 102 (4.00) 208 (46.79)

236 ACI Structural Journal/March-April 2008

We believe that interpretation of nonlinear analysis results is Closure to discussion by Muttoni et al.
quite complex and requires significant experience with these As the discussers point out, the strut-and-tie models for
types of analyses. Adequate material models with appropriate this study were developed without consideration of kinematics
parameters (often calibrated through experimental testing) because these are commonly neglected for design. The
must be used to achieve adequate solutions. Inadequate use discussers have developed a method based on stress fields
of material properties may result in predicted loads that are that accounts for realistic kinematic conditions at failure.
higher than the actual force an element can carry. On the
The authors believe that these techniques are extremely
other hand, constructing and solving a relatively complicated
strut-and-tie model can be performed with ease and will promising and can provide much better estimates to actual
provide a lower bound to the true solution, which is desirable strength of elements than current models based on strut-and-
for design. Therefore, the authors still believe that strut-and- tie idealizations. The authors are extremely pleased that their
tie modeling techniques offers a viable option for safe design research is able to contribute to development of sophisticated
of members with discontinuities. tools for structural design of elements with complex geometries.

Disc. 104-S29/From the May-June 2007 ACI Structural Journal, p. 294

Lattice Shear Reinforcement for Slab-Column Connections. Paper by Hong-gun Park, Kyung-soo Ahn,
Kyoung-kyu Choi, and Lan Chung

Discussion by Ramez B. Gayed

ACI member, Research Associate, Civil Engineering Department, University of Calgary, Calgary, AB, Canada.

Several researchers15-19 have suggested configurations of REFERENCES

multi-leg prebent bars as shear reinforcement in flat plates. 15. Joint ACI-ASCE Committee 426, “Shear Strength of Reinforced
The ACI 318 Code20 permits the use of shear reinforcement Concrete Members—Slabs,” Journal of the Structural Division, ASCE,
in the form of closed or multi-leg stirrups. Essential criteria V. 100, No. ST8, Aug. 1974, pp. 1543-1559.
for shear reinforcement in flat plates are effective anchorage 16. Broms, C. E., “Shear Reinforcement for Deflection Ductility of Flat
Plates,” ACI Structural Journal, V. 87, No. 6, Nov.-Dec. 1990, pp. 696-705.
and ease of constructibility. The results of the presented tests
17. Melo, G. S.; Coelho, A. E. G.; and Oliveira, D. R. C., “Reinforced
show adequacy of the anchorage achieved by welding of the Concrete Flat Slabs with Inclined Stirrups as Shear Reinforcement,”
lattice shear bars to the flexural bars. The constructibility of Proceedings, International Workshop on Punching Shear Capacity of RC
the proposed system is questionable, however, particularly Slabs, Stockholm, Sweden, June 2000, pp. 155-162.
in prestressed slabs due to the reinforcement congestion. 18. Seible, F.; Ghali, A.; and Dilger, W. H., “Preassembled Shear
It is well established that the shear reinforcement is most Reinforcing Units for Flat Plates,” ACI JOURNAL, Proceedings V. 77, No. 1,
effective when it confines the maximum volume of concrete; Jan.-Feb. 1980, pp. 28-35.
thus, its overall height has to be as large as possible. 19. Yamada, T.; Nanni, A.; and Endo, K., “Punching Shear Resistance of
Flat Slabs: Influence of Reinforcement Type and Ratio,” ACI Structural
However, all reinforcement must be protected with the Journal, V. 88, No. 4, Sept.-Oct. 1992, pp. 555-563.
concrete cover specified in codes. In the tested slabs, the 20. ACI Committee 318, “Building Code Requirements for Structural
lattice shear reinforcement had no cover at the bottom. To Concrete (ACI 318-05) and Commentary (318R-05),” American Concrete
provide the required cover for the lattice shear reinforcement, Institute, Farmington Hills, MI, 2005, 430 pp.
its overall height has to be smaller than in the tests, resulting 21. ACI Committee 421, “Shear Reinforcement for Slabs (ACI 421.1R-99),”
in an adverse effect on the observed strength and ductility. American Concrete Institute, Farmington Hills, MI, 1999, 15 pp.
In the lattice shear reinforcing system, the shear is resisted
by the inclined legs running in almost two orthogonal directions. AUTHORS’ CLOSURE
A leg intercepting a crack at a right angle is most effective in The authors thank the discusser for his interest in this
controlling its width; a leg parallel to a series of cracks may paper. Each item of the questions and comments presented
not intersect any of them. If half the legs are perpendicular to by the discusser is discussed separately, as follows.
the cracks, the other half will be parallel to them. Thus, only
one half of the web bars can be fully effective in intercepting
and controlling the inclined shear cracks. Absence of web 1. Constructibility of proposed system
bars that intersect the shear crack at the critical section close The lattice system that the authors tested is an existing
to the column can induce failure at a low load level. For this commercial product that was originally developed as a part
reason, ACI 318-0520 and ACI 421.1R-9921 specify the of a form deck system for slab construction. For better
distance between the column face and the first peripheral constructibility, however, the configuration and shape of the
line of vertical shear reinforcement. lattice bars can be changed according to the engineer’s
The comparison between the results of tests with the desire. For example, the number of the lattice bars that are
lattice shear reinforcement and tests having headed studs is installed at the slab-column connection can be significantly
unpersuasive because: 1) the volumetric ratio of the inclined reduced by using large-diameter reinforcing bars for the
legs of the lattice shear reinforcement was 13 to 19% higher lattice. Further, though the lattice system that was used in
than that of the stems of the headed studs; and 2) the this study had a space truss configuration, it can be changed
compared tested slabs differed in fc′ (ρ fy) and the overall to a planar truss system for easy installation and avoiding
heights of the shear reinforcement assemblies. congestion of the reinforcing bars.

ACI Structural Journal/March-April 2008 237

2. Concrete cover for lattice shear reinforcement at 4. Comparison between results of tests with lattice
the bottom shear reinforcement and those with headed studs
In the lattice system, the legs of the inclined web bars do As shown in Tables 1, 2, and 4, the specimens using shear
not affect the structural capacity of the lattice, and play the stud rails have almost the same material properties as those
role of bar-chairs. Therefore, like ordinary bar-chairs, the of Specimens SL1, SL2, and SL4 tested by the authors. The
legs can be treated for corrosion protection. Otherwise, compressive strengths of the authors’ specimens are 25.9 to
the lattice can be manufactured without the legs, and 28.0 MPa (3.8 to 4.1 ksi) whereas those using shear studs rails
ordinary bar-chairs can be used to support the lattice bars are 29.0 to 49.0 MPa (4.2 to 7.1 ksi). The effective shear
during construction. strength (ρv fyv) of the authors’ specimens is 1.4 to 1.7 MPa
(0.20 to 0.25 ksi) whereas those using shear studs rails are
3. Distance between column face and first peripheral 1.3 to 5.0 MPa (0.19 to 0.73 ksi). As the discusser indicates,
line of shear reinforcement the amount of web shear reinforcement of the lattice is
As the discusser mentioned, for effectiveness of shear slightly greater than that of the stud rail. This is because a
reinforcement, the location of the first peripheral line of the part of the inclined web bars of the lattice is ineffective in the
shear reinforcement is important, and the specification of the shear contribution. However, the effective amount (the
ACI 318-05 Code20 and ACI 421.1R-9921 should be met. effective shear strength) of the web reinforcement of the
Unlike ordinary shear reinforcement including stirrups and lattice is similar to that of the stud rails. As such, because the
shear studs, however, the lattice shear reinforcement is a material properties of the specimens using shear stud rails
continuous truss system going through the column and are not significantly different from those of the lattice
extending beyond the region of the critical section. In the specimens, the authors believe that the comparison
authors’ opinion, the load-transfer mechanism of the lattice presented in this paper is meaningful enough.
reinforcement providing truss action and increased dowel Though the discusser indicated the difference in the
action is different from that of the ordinary shear reinforcing amount of web reinforcement of the lattice and the stud rails,
bars. In this paper, the shear strength of the lattice reinforcement from an economical standpoint, the cost for manufacturing
was calculated in the same manner as used for the stirrups. To the shear reinforcement is more important than the amount
clarify the load-transfer mechanism of the lattice, however, of the shear reinforcement. An application of the lattice
further intensive experimental and theoretical studies are reinforcement in Korea showed that the construction cost of
required in the future. the lattice was significantly less than that of the stud rails.

Disc. 104-S36/From the May-June 2007 ACI Structural Journal, p. 357

Distinction between Punching and Flexural Failure Modes of Flat Plates. Paper by Timm Stein, Amin Ghali,
and Walter Dilger

Discussion by Carl Erik Broms

Senior Consultant, WSP Sweden AB, Stockholm, Sweden.

The authors maintain that experimental research to study nominal strength level has to be chosen safely low. The level
the effectiveness of shear reinforcement in flat plates gives should allow the desirable structural behavior that all
conclusive results only if tests are designed so that the reinforcement will reach the yield stress without punching
predicted flexural capacity is at least 50% larger than the occurring if the nominal punching capacity of the slab
predicted punching capacity. As will be shown in the exceeds the nominal flexural capacity. This sound engineering
following, they thereby take advantage of a deficiency in the principle seems to have been ignored by the authors of the
ACI 318 Code, where the nominal punching shear strength paper and also by ACI 421.1R-99 on design of shear studs in
is considered to be independent of the provided amount of flat plates. In the latter, no indication is given that the
flexural reinforcement. It has been well known for more than proposed upper bound for the nominal punching strength
40 years, however, that the amount of flexural reinforcement with shear studs calls for more flexural reinforcement than
does indeed have a major impact on the punching strength, required for the bending moment.
which in fact is confirmed by the tests described in the paper. The discusser therefore believes that the experimental
More severe misinterpretations of test results than those principle described in the paper is of limited value if the tests
described by the authors may therefore be made if due are evaluated against the ACI 318 Code because the test
respect is not paid to important factors that influence the specimens would then not reflect normal design. In a real
punching strength. Such factors are, for instance, the flexural case, the amount of flexural reinforcement would not be
reinforcement ratio, the size effect, and the slenderness of chosen to resist a bending moment that is more than 50%
the test specimen—none of them covered by ACI 318. larger than the actual bending moment.
Kinnunen and Nylander (1960) showed that the punching A consequence of ignoring the influence of the flexural
shear strength of flat plates increases with increasing flexural reinforcement ratio on the punching capacity is demonstrated by
reinforcement ratio and decreases with increasing slenderness of the authors’ own tests. Specimen V2, with a reinforcement ratio
test specimens. Moe (1961) concluded that if the nominal of 0.0098, failed in shear outside the zone with shear studs at
punching shear strength is defined to be independent of the the load 438 kN (98.5 kip). Specimen V3 with a reinforcement
flexural reinforcement ratio (as by ACI 318-05), then the ratio of 0.0062 also failed in shear outside the studs at

238 ACI Structural Journal/March-April 2008

365 kN (82.05 kip). The authors conclude that this latter limit of the punching strength with shear reinforcement is, in
failure is due to a combination of flexural and shear failure. Europe, usually expressed as a function of the flexural
This seems to be an erroneous conclusion. The radial reinforcement ratio. The upper limit for the punching
bending moment is very low at the actual shear failure capacity with shear studs in ACI 421.1R-99 seems to be
position outside the shear reinforcement and should therefore based on these tests, without mentioning that the flexural
not influence the shear capacity of the slab. If the influence reinforcement thereby must be over-designed. Furthermore,
of the flexural reinforcement ratio on the shear strength is
slender test specimens simulating flat plates would display
taken into account by the principle of Eurocode 2, then
the theoretical relation between the shear capacities of lower punching capacity than the tested compact test specimens,
Specimens V2 and V3 becomes (0.0062/0.0098)1/3, which is which has been reflected by, for instance, the Swedish Code
close to the actual test relation 365/438 = 0.83. The lower for Concrete Structures since 1964.
shear capacity of Specimen V3 in relation to Specimen V2 is The nominal punching strength without shear reinforcement
thus fully explained by its lower flexural reinforcement ratio. according to ACI 318 is usually conservative when
This demonstrates the danger with the authors’ recommended compared with test results as noted by the authors. This can
testing principle in combination with evaluation according to be exemplified by the aforementioned Specimen AB1 with
ACI 318. If test results with high reinforcement ratios are the nominal punching capacity of 333 kN (75 kip) according
taken as an indication of the shear strength of the slab outside to ACI 318-05 to be compared with the test capacity of 408 kN
the studs, the consequence will be unsafe design for real (91.7 kip) (= 1.23 · 333). This conservatism, however, is only
structures with less flexural reinforcement.
experienced for specimens with a high reinforcement ratio
A similar evaluation mistake exists for Specimen V1. Its
and small effective depth and is not sufficient to give reasonable
punching capacity without shear reinforcement is assessed
by comparison with the previously tested Specimen AB1 safety for real structures with a low reinforcement ratio and
with the compression strength of 36 MPa (5221.4 psi). large effective depth (Gardner et al. 2000). At least the same
Specimen AB1 failed in punching for the load of 408 kN margin of approximately 1.25 for test results in relation to
(91.7 kip). The authors then assess the punching capacity the nominal shear strength according to ACI 318-05 should
Vc without shear reinforcement to 355 kN (80 kip) for be applied when evaluating test specimens with shear
Specimens V1 to V3 by proportioning to the square root of reinforcement failing in shear outside the shear-reinforced
the concrete strengths in accordance with ACI 318. They zone. This basic principle was not adhered to by, for
thereby disregard the considerable difference in reinforcement instance, Mokhtar et al. (1985), Megally (1998), ACI
ratios—0.013 for Specimen AB1 and 0.0045 for Specimen V1. Committee 421 (1999), and Gayed and Ghali (2006).
If the principle of Eurocode 2 is used for proportioning in In summary, the authors of the paper have hit some tender
relation to Specimen AB1, the probable punching capacity
spots of the ACI 318 Code. They have demonstrated that
without shear studs for Specimen V1 becomes
evaluation of punching experiments against ACI 318-05 or
ACI 421.1R-99 may lead to false conclusions because due
1
--- respect is not paid in those documents to important
0.0045 29.7 3
V c = 408 ⎛ ---------------- ⋅ ----------⎞ = 269 kN (60.6 kip) parameters that affect the punching strength. It is therefore
⎝ 0.013 36 ⎠
high time that the ACI 318 provisions for shear and punching
of slabs be modernized.
instead of the authors’ assessment of 355 kN (80 kip). It is
then evident from Fig. 9 that both the capacity and the ductile REFERENCES
behavior of Specimen V1 most probably should be attributed Andrä, H.-P., 1981, “Zum Tragverhalten von Flachdecken mit Dübel-
to the shear studs, contrary to the authors’ conclusion. In leisten—Bewehrung im Auflagerbereich,” Beton und Stahlbetonbau V. 76,
addition, the size effect should be considered when evalu- No. 3, pp. 53-57, and No. 4, pp. 100-104.
ating test results. The obtained shear strengths (expressed in DEHA 1996, “Bericht über Versuche an punktgestützten Platten bewehrt
stress units) of slab specimens with the small effective depth mit DEHA Doppelkopfbolzen,” Report No. AF 5 96/6-402/1, Institut für
of approximately 115 mm (4.5 in.) must be treated with Werkstoffe im Bauwesen, Universität-Stuttgart, Stuttgart, Germany, 81 pp.
caution when applied to real structures with larger effective Gardner, N. J.; Huh, J.; and Chung, L., 2000, “What Can We Learn from
depth. For instance, a real structure with an effective depth the Sampong Department Store Collapse,” Proceedings, International
of 200 mm (8 in.) and a reinforcement ratio of 60% of the test Workshop on Punching Shear Capacity of RC Slabs, Royal Institute of
Technology, Stockholm, Sweden, pp. 225-233.
specimen’s ratio would, according to Eurocode 2, have a
Gayed, R., and Ghali, A., 2006, “Seismic-Resistant Joints of Interior
shear strength or punching strength of only
Columns with Prestressed Slabs,” ACI Structural Journal, V. 103, No. 5,
Sept.-Oct., pp. 710-719.

1 + 200
Kinnunen, S., and Nylander, H., 1960, “Punching of Concrete Slabs
--------- 1
---
200 3 without Shear Reinforcement,” Transactions No. 158, Royal Institute of
---------------------- ⋅ ( 0.60 ) Technology, Stockholm, Sweden, 112 pp.
200
1 + --------- Megally, S. H., 1998, “Punching Shear Resistance of Concrete Slabs to
115 Gravity and Earthquake Forces,” PhD dissertation, University of Calgary,
Calgary, AB, Canada, 468 pp.
Moe, J., 1961, “Shearing Strength of Reinforced Concrete Slabs and
= 0.73 times the strength (in stress units) of the test specimen. Footings under Concentrated Load,” Bulletin D46, Portland Cement
Compact test specimens with a high flexural reinforcement Association, Research and Development Laboratories, Skokie, IL, 135 pp.
ratio have often been used in Europe for testing the punching Otto-Graf-Institut, 1996, “Durchstansversuche an Stahlbetonplatten mit
capacity of slabs with shear studs (Andrä 1981; DEHA 1996; Rippendübeln und vorgefertigten Groß-Flächentafeln, Report No. 21-21634,
Otto-Graf-Institut 1996), which is justified because the upper University of Stuttgart, Stuttgart, Germany.

ACI Structural Journal/March-April 2008 239

Disc. 104-S36/From the May-June 2007 ACI Structural Journal, p. 357

Distinction between Punching and Flexural Failure Modes of Flat Plates. Paper by Timm Stein, Amin Ghali,
and Walter Dilger

Discussion by Myoungsu Shin and Jacob Grossman

Project Engineer, Rosenwasser Grossman Consulting Engineers, New York, NY; President, Rosenwasser, Grossman Consulting Engineers.

The tests were performed with approximately 1/2-scale subjected to gravity and/or lateral loading. The middle
interior slab-column connection subassemblies subjected to strip reinforcement in the test specimen could have
both gravity and lateral loading. The discussers would like to participated in force transfer significantly more than in
deliberate several considerations on the test setup used in the actual prototype frame.
this study. There are several unrealistic features that could 4. In conjunction with Item 3, the support condition around
have directly affected the test results investigated, as the slab edges restrained relative rotational deformations
summarized in the following: (curvatures) along the slab edges parallel to the direction of
1. The continuous simple supports along the four peripheral lateral loading (x-direction) during testing, which would
lines of the slab in the test were set up at approximately 1/3 have occurred in the prototype structure.
of the span length apart from the column, assuming that the test Finally, the discussers would like to recommend that tests
specimen represented a roughly 1/2-scale model. There is no need to be conducted on three-dimensional frames with
explanation, however, whether the supports were devised to multiple spans in both principal directions that allow realistic
simulate inflection points under gravity loads only or under moment distribution across slab sections and moment redis-
combined gravity and lateral loads. The test specimen was tribution along spans in the nonlinear range, especially when
subjected to subsequent lateral loads, with gravity loads applied two-way shear and flexural behaviors and their interaction
first and sustained. In a real structure, inflection points for are investigated as in this study.
combined gravity and lateral loading would differ from those
under gravity loading only. In general, the inflection points AUTHORS’ CLOSURE
occurring in the prototype frame in an event of the design-level Closure to discussion by Broms
earthquake would typically be close to the midspan of the slab. The purpose of the presented tests was to show that test
In the test, however, the inflection points under the gravity specimens, aiming to study the effectiveness of shear
loading were identical to those during the lateral loading by reinforcement in resisting punching of flat plates, have to be
design (by applying compression on the column for simulating designed such that they fail by punching, not flexure. This
the gravity loads). Thus, vertical resultant forces due to the requirement is obvious in a shear strength test of a simple beam
lateral loading at the slab edges perpendicular to the loading subjected to gravity load. With a low flexural reinforcement
direction would have been larger in the test than those in the ratio, the test beam can fail at midspan, in a ductile form, by
prototype frame, which in turn would have produced larger yielding of the bottom flexural reinforcement combined with
direct punching shear forces around the column. Also, slab crushing of concrete at the top at large curvature; the flexural
moments generated in the slab-column interfaces due to the failure can occur before the shear strength is reached near the
gravity loading applied first would have been smaller in the test supports. The test results give information on the strength
than those in the prototype frame. In short, the magnitude of and the ductility in flexural failure; they can only indicate
direct shear or unbalanced moment occurring at the slab-column that the shear strength exceeds the maximum shear force that
interfaces depends on the locations of inflection points in the beam has been exposed to during the test, without giving
the test subassembly. information on the shear strength or the ductility of the shear
2. Along with Item 1, data analysis related to the story drift failure that has not occurred.
ratio may not be valid if the subassembly dimensions were The issue is the same in punching shear tests of flat plates,
not detailed in proportion to the prototype slab-column although it is somewhat obscured by the fact that punching
frame. The discussers assumed that the test specimens were or flexural failure occurs at the same location—in the
in roughly 1/2-scale in that the story height in a typical vicinity of the column. A test exhibiting flexural failure at a
building with a flat plate system ranges from 2.75 to 3.35 m low load level that does not demand the full shear strength
(9 to 11 ft). would not be indicative of the value of the shear strength, the
3. During testing, the continuous simple supports played a ductility, or the brittleness of the shear failure that has not
function similar to continuous wall supports, so that relative occurred. When searching for the strength and the ductility
vertical displacements along each of the four slab edges were in punching shear of flat plates, premature failure by flexure
restrained. When a slab span in a flat plate system is has to be excluded. This can only be achieved by the provision
supported by a long wall at one end, moments in the slab of a sufficiently high flexural reinforcement ratio ρ. This
section adjacent to the wall are much more evenly distributed logic does not appear acceptable to the discusser.
across the wall, compared with moment distribution in the The discusser dwells on what he calls deficiencies of the
column and middle strips at a slab span supported by two ACI 318 Code and the recommendations of ACI 421.1R-99.
columns at both ends; for an interior span, approximately 75 In particular, he criticizes that the shear strengths’ equations
and 25% of the total negative moment are resisted in column in these sources do not include ρ. The fact that the punching
and middle strips respectively, under gravity loads. In short, shear strength increases with the increase in ρ is well known
the moment distribution occurring in the test specimen and does not need the long explanation in the discussion. The
would have been different from that in the prototype frame calibration and the reasons behind the equations of the

240 ACI Structural Journal/March-April 2008

ACI 318 Code or the recommendations of ACI 421.1R-99 and the labor needed to test these ideal structures are prohibitive
are beyond the scope of the paper. Thus, a response to the for practically all researchers.
claims by the discusser of “false conclusions” in the paper The tests presented in the paper represented full-size
and the “high time” for ACI 318 to be “modernized” should isolated interior connections of a flat plate with spans of
not be in the authors’ closure of the paper. 4.8 m (15.75 ft) in two orthogonal directions. The supported
On the amount of ρ that should be provided in practice, the edges were at approximately 1/5 the span representing the
discusser states, “The level should allow the desirable structural inflection location due to gravity loads. Under gravity loads
behavior that all reinforcement will reach the yield stress combined with unbalanced moments, the zone of inflection
without punching occurring if the nominal punching in an actual structure would move away from the columns,
capacity of the slab exceeds the nominal flexural capacity.” but would practically not be at midspan.
Then he argues, “In practice, ρ would not be chosen to resist International codes for punching shear design of flat plates
a bending moment that is more than 50% larger than the are based on extensive experimental data. The majority of
actual bending moment.” The paper does not suggest or the tests are on isolated specimens, the type presented in the
imply that ρ in practice should exceed the demand by 50%. paper. These tests have the advantage that the magnitudes of
The design steps in practice are: select the thickness of the the shear force V and the unbalanced moment M are accurately
flat plate as a ratio of the span to avoid excessive deflection; measured at all loading stages. Thus, the zone of the slab in
determine ρ values at midspans and at the supports, to resist the vicinity of the column is transferring forces V and M of
the bending moments obtained by elastic analyses for the known magnitudes, and in the tests, the behavior of the zone
loading cases that produce maximum positive and maximum in the vicinity of the column is monitored with sufficient
negative values due to factored loads; check the punching accuracy. Scarce tests on three-dimensional structures give
shear strength to ensure that it exceeds the maximum demand information on the behavior of the entire structure (for
with the appropriate loading case; and provide shear reinforcing example, Sherif [1996], Sherif and Dilger [2001], and
means if necessary. The discusser’s recommendation that all Dechka [2001]), and at the same time confirm that the
flexural reinforcements (top and bottom) reach their yield behavior of the zone of the slab-column connections can be
stress requires a higher load intensity than the factored safely predicted by tests on isolated specimens, for the same
design load intensity used in the elastic analyses is neither V, or V combined with M.
required by codes nor complied with in practice.
REFERENCES
Dechka, D. C., 2001, “Response of Shear-Stud-Reinforced Continuous
Closure to discussion by Shin and Grossman Slab-Column Frames to Seismic Loads,” PhD thesis, University of Calgary,
The discussers propose an ideal test system that is relevant Calgary, AB, Canada, 2001, 472 pp.
to all research on punching shear. They recommend a three- Sherif, A., 1996, “Behavior of Reinforced Concrete Flat Slabs,” PhD
dimensional structure with multiple spans in two principal thesis, University of Calgary, Calgary, AB, Canada, 397 pp.
directions. Several structures of this type would be needed to Sherif, A., and Dilger, W. H., 2001, “Effect of Test Set-Up on the
Punching Strength of Slab-Column Connections,” Proceedings, CSCE
study the effect of varying one or two parameters. The cost Annual Conference, Victoria, BC, Canada, June.

ACI Structural Journal/March-April 2008 241