Postulates on shear in TECHNICAL PAPER

reinforced concrete JOURNAL OF THE SOUTH AFRICAN

INSTITUTION OF CIVIL ENGINEERING
M Gohnert Vol 48 No 4, 2006, Pages 9–13, Paper 609

Ambiguities associated with shear in reinforced concrete have led to a misunderstanding

of what shear is and how it affects structural members. This paper therefore explores the PROFESSOR M GOHNERT received
notion of shear, using mechanics and finite elements as a tool to describe this force. A his BSc in civil engineering in 1985 and
series of postulates are also presented in order to define the nature and characteristics of a master’s of engineering management
shear. The theory is then applied to structural elements such as beams and slabs. Various in 1987 from Brigham Young University,
USA. In 1995, he received a PhD in civil
shear concepts, such as shear enhancement, shear in slabs and punching shear are also engineering from the University of the
examined. Witwatersrand (Wits). He has worked
for various companies in the United
States and in South Africa as a structural engineer. In 1990, he
joined the School of Civil and Environmental Engineering at
INTRODUCTION what shear is, using mechanics as a tool to
Wits University, where he is currently head of the school.
An understanding of shear seems to elude describe this force. A set of postulates is
many practitioners and researchers alike. proposed for the purpose of formally defin- Contact details:
Some even question the existence of shear. ing the characteristics of shear and its effects School of Civil and Environmental Engineering
University of the Witwatersrand
For those who research shear, testing and on reinforced concrete. The theory is then Private Bag X3
attaining a shear failure is difficult to achieve. applied, with the help of finite elements, to PO Wits, 2050
Matching shear failure with theoretical mod- various structural elements such as beams T 011-717-7125
els is nearly an impossible task. Furthermore, and slabs. Concepts such as shear enhance- gohnert@civil.wits.ac.za
the complexity and uncertainty associated ment, shear in slabs and punching shear are
with shear has led to conservative code also considered and explained.
equations and is therefore an uneconomical
feature of structural design (Kong & Evans
1987). Perhaps our problem lies in the fact FUNDAMENTALS OF SHEAR
that our analysis methods are based on line- To understand shear, the internal forces
ar-elastic models; yet, our codes are based on within a structural member must be broken
limit state principals. This mismatch, for the down into fundamental postulates. The first
sake of making the design process practical, postulate states:
has led to confusion and a disparity between
theory and observed shear failures.
The purpose of this paper is to explore

Figure 1 Various combinations of principal tensile and compressive stresses

Journal of the South African Institution of Civil Engineering • Volume 48 Number 4 December 2006 9
 2 Shear stresses are components of tensile
and compressive stresses applied at an
angle to the x and y axes

Furthermore, what we call pure shear is the

case of stress applied at 45 degrees to the x
and y axes (figure 2). If stress is applied at
some other angle, the differential element
would be composed of normal stresses as
well as shears (Wang & Salmon 1985).
The second postulate could, perhaps,
be incorporated with the first postulate.
Figure 2 Shear stresses equivalent to components of tensile or compressive stresses applied at an angle However, it is separated to illustrate that
shear is not a unique force, but a tensile or
compressive stress applied at an angle to the
plane under consideration.
The third postulate states:

3 Two forces will always exist and

orientated orthogonally to one another

This postulate is based on the premise

that structural materials are compressible.
Compressive materials adhere to Poisson’s
effect and therefore two forces will always
exist at an element level, orientated orthogo-
nally (Ferguson et al 1988). These two forces
may be compressive or tensile.
Figure 3 Pin-ended member used to model Poisson’s effect of internal forces Poisson’s effect may be modelled by a
pin-ended mechanism attached to springs,
differential elements subjected to various as shown in figure 3. The model illustrates
combinations of principal tensile and com- how opposing forces (tensions and compres-
pressive stresses, orientated along the x and sions) are produced in compressive materi-
y axes. Figure 1 is the conventional method als – the model is not an indication of unit
of illustrating bi-axial principal stresses displacements.
along the x and y axes. Howeve, stress is Poisson’s effects are well known – a
not limited to the x and y directions, but stress applied in one direction will produce
may occur at any angle. Figure 2 illustrates an opposite stress at 90 degrees. Since all
a tensile force applied at the corners of an structural materials are compressible, the
element, orientated 45 degrees from the postulate is applicable and valid.
x-axis. The tensile stress may be broken The fourth postulate states:
Figure 4 Shear crack at end of beam down into components of stress, as shown
in the second element. The components 4 Only two modes of possible failure exist
1 Only two fundamental forces exist in of stress replace the tensile stress and are – tension and compression
structural members – tensions and graphically represented in the third dif-
compressions ferential element. The stress pattern of the If only two forces exist, then it is only logi-
third element is what we call shear. Thus, a cal that only two failure mechanisms exist
In structural members, there are three shear stress is a graphical representation of – tension and compression. If we consider
types of forces which include moments, a tension or compressive stress applied at structural members, rather than differential
shears and axial loads. These forces are in an angle to the differential element. elements, buckling is the third possible
fact resultants (integrated stresses) which If the tensile stress applied to the corner mode of failure in steel structures. In con-
may be broken down into fundamen- of the element is replaced with a compres- crete, however, buckling rarely occurs and
tal forces at a differential element level. sive stress, the resulting shear pattern is modes of failure are typically tension and
Although we have specific nomenclature to opposite to the former. Likewise, if stress is compression failures – tension failure being
describe these forces, all resultant forces applied to the other two corners of the ele- the most predominantly observed. The
comprise of tensions and/or compres- ment, the same two possible shear patterns question is, does the postulate apply to what
sions. A moment is merely a force couple, would result. we call a shear failure?
composed of tensions and compressions, As seen, shear stresses are merely com- In figure 4, a reinforced concrete beam
caused by a linear or non-linear variation ponents of tensile or compressive stresses was tested to failure. Cracking in the beam
of stress across a section. An axial stress is applied at an angle to the x and y axes. is visible, extending from the support
either compressive or tensile and therefore Figure 1 illustrates a principal stress condi- towards the first load point. The observed
complies with the postulate. The question tion, and by definition will have no shear failure line is often referred to as a shear
is whether or not shear is a tensile or com- stresses. However, if stress is applied at some crack. Literature has repeatedly described
pressive stress as postulated. The answer angle to the x and y axes, figure 1 would also the mechanisms by which shear forces
is neither. Shear stresses are, however, graphically include shear stresses. This leads affect structures and how they are resisted
components of tensile and compressive to the second postulate, which has already once cracking in the concrete has occurred
stresses. Figure 1 illustrates four possible been defined in the preceding discussion: – dowel action, aggregate interlock, inter-

10 Journal of the South African Institution of Civil Engineering • Volume 48 Number 4 December 2006
 a cube is a complex arrangement of triaxial
stresses and greatly influenced by friction
between the contact surfaces, the failure
lines are predominantly vertical rather than
horizontal (figure 7)(Ferguson et al 1988).
From postulate 3, a compressive stress
applied in one direction will cause a tension
stress orthogonally. The expansive tensile
stresses in the horizontal direction are due
to Poisson’s ratio (the vertically applied
compression will cause a horizontal ten-
sion). Since concrete is far weaker in tension
Figure 5 Vertical and horizontal modes of shear failure than compression, the tension strength is
reached prior to the compressive strength.
Therefore tension cracks form along planes
that are parallel to the compressive force.
Ironically, what we call a compressive test
is actually a measure of the tensile capacity
Potential shear cracks
of concrete.
Figure 6 Principal stress pattern in a simply supported There have been several attempts to
beam subjected to a UDL devise a test to determine the shear capacity
FRC beam of concrete. Figure 8 is one such apparatus
used to apply a pure shear condition to a
concrete beam reinforced with steel fibre
Figure 8 Shear test rig (JSCE No 3 1984). The theoretical failure
plane is as illustrated – a shear failure line
supported and subjected to a uniformly occurring between the faces of the applied
distributed load. The plot illustrates the load. The actual failure line, however, con-
relative magnitude of the principal stresses sistently forms at a slight angle between
Figure 7 Predominant vertical failure lines in cubes as well as the orientation. Arrows pointing the centre of the applied force and the
inward indicate a compressive stress and centre of the support. This mode of failure
face shear transfer, etc (MacGinley & Choo arrows pointing outward indicate a ten- is not due to shear. The applied load flows
1990; Kong & Evans 1987). However, what sion stress. The stress pattern exhibits directly into the support. The compressive
we call a shear failure is not congruous catenary arching of the compressive forces force is increased until the tensile capacity
with observed failure patterns. We are in (superimposed blue lines) as well as cat- is reached, at 90 degrees to the compres-
fact misled by the term. Furthermore, shear enary sagging of tensile forces (superim- sive thrust (postulate 3). Cracking of this
theory assumes linear elasticity, which is posed red lines). Although tension stresses nature is due to what is referred to as a split-
not appropriate at the onset of cracking. The exist in the top corners of the beam, ting force (Ferguson et al 1988). Although
influence of the reinforcement on the crack these stresses are negligible. The major- the test is intended to determine the shear
pattern is also not taken into account. ity of tensile cracking occurs within the strength, the specimens consistently fail in
Near the end of a simply supported compression arch. tension. This example illustrates the difficul-
beam, where shear is maximum and the The plot clearly illustrates the region ty in assessing the shear capacity of concrete
bending moment is minimum, the stresses where tension stresses reside – along the for the simple reason – the shear capacity of
resemble the pure shear case illustrated in bottom of the beam parallel with the span concrete is greater than the tensile capacity.
figure 2. The shear stresses are orientated and curving upward near the supports. Examples of failure lines are given in
vertically and horizontally which would Cracks will form perpendicular to the orien- figure 9. As discussed previously, a shear
precipitate failure patterns similar to those tation of these stresses matching the crack mode of failure would require slippage along
illustrated in figure 5 (Timoshenko 1970). pattern of the test beam in figure 4. Thus, the failure plane. No slippage is discern-
The diagonal failure line in figure 4 is not in the majority of cases, what are referred to ible. The failure consists of a perpendicular
vertical or horizontal, which would imply as shear cracks are actually diagonal tension separation of the concrete along the crack
that the crack is not the result of shear forc- cracks which form consistently with the ten- (indicative of tension) and some crushing of
es. Furthermore, a shear failure would cause sion stress regions of the beam. This leads to the concrete near the support.
material slippage along the failure plane the next postulate on shear: The shear strength is greater than the
(translational failure); this was not observed. tensile strength of concrete except in cases
Cracking commenced at the bottom of the 5 The shear and compressive strength where a weak plane exists. This is a typi-
beam, propagating at an angle and terminat- of concrete is greater than the tensile cal problem when dealing with composite
ing at a point near but falling short of the strength members. A potential weak plane will exist
first load. Parallel slippage along the failure between the two materials (ie, concrete on
plane was not observed; the direction of Shear forces exist in concrete, but concrete concrete or concrete on steel). Horizontal
crack displacement is perpendicular to the rarely fails in shear. The reason is that con- shear failure is common if the shear strength
failure plane. This would imply that the crete is an anisotropic material – the shear is reliant on chemical bonding. However, the
observed failure is a tension failure rather and compressive strength is greater than the horizontal shear strength is improved when
than a shear failure. tensile strength. Thus failures in concrete the two materials are mechanically bound
Finite elements are used to illustrate are primarily tensile. by reinforcement, shear studs or roughening
the stress pattern in a beam. Figure 6 is The anisotropic nature of concrete is the surface. Potential weak planes in rein-
an ABAQUS model of a beam composed of illustrated by compression tests on cube forced concrete also exist along the length of
plane stress elements. The beam is simply specimens. Although the stress field within the main steel. Thus some observed failure

Journal of the South African Institution of Civil Engineering • Volume 48 Number 4 December 2006 11
 It is therefore imperative that the main steel
extends beyond the edge of the support to
provide anchorage in simply supported or
continuous slabs. Leonhardt (1965) has also
a Point load applied at a distance of 2d observed that because slabs have a greater
av/d ratio (av/d > 6), slabs have a greater
propensity to fail in flexure rather than
shear (av is the distance from the applica-
tion of the load to the centre of the support).
b Point load applied at a distance of d Second, the flow of stress within a slab is
Figure 9 Two examples of failure lines in shear tests low compared to beams and columns. A
Figure 10 Stress distribution within a beam due to point roof drainage system is an appropriate anal-
lines run parallel with the main steel in the loads ogy. If the roof represents a slab, the gutter
regions of maximum shear. the beam and the down pipe a column, the
It is interesting to note that what we call to the centre of the tension steel. The blue flow of stress is similar to the flow of water.
shear steel (vertical links) will resist shears and the red lines indicate the magnitude The amount of water flowing off a sloped
running horizontally, but not vertically. and direction of the compression and ten- roof is far less than the flow in the gutter
From mechanics, the maximum shear stress sion stresses, respectively. A comparison of and down pipe. The same applies to slabs
on a cross-section is at the neutral axis. In a the two plots show that a point load placed (one-way or two-way) – the flow of force in a
beam, this region is typically unreinforced further away from the support will cause slab is far less than the flow of concentrated
horizontally and therefore susceptible to more bending and thus flexural tensions in force in beams and columns. Thus slabs are
vertical shear cracking. Yet, vertical shear the bottom of the beam. The crack pattern is less susceptible to shear failures (with the
cracks are rare, except in some cases where consistent with flexural failures. If the load exception of punching). Lastly, the closeness
the member is over-reinforced (this has been is placed near the support, the load tends to of the top and bottom reinforcement limits
observed in tests where horizontal shear fail- flow directly into the support and will fail the crack width, which increases aggregate
ure occurs along the main reinforcement and from splitting tension stresses orientated interlock; therefore the likelihood of shear
travels up vertically between the shear links at right angles to the compression thrust; failure is reduced.
to the first load point). Having said this, a diagonal failure line will form extending Although most slabs may be classed
whether or not a beam is reinforced horizon- from the support to the point load (similar as thin, many are not. The thickness of
tally near the neutral axis is not a concern to the failure lines of figures 4 and 9). silo slabs, for example, may exceed 2 m in
since the shear capacity of concrete is far From figures 10a and 10b it is observed depth and may be subjected to extremely
greater than the tensile capacity. The pur- that the flexural tensions are much greater high loads. In these types of structures,
pose of the shear steel is not to resist shears, than the splitting tensions for the same mag- the l/d ratio is so low that one would ques-
as the name would imply, but rather to resist nitude of load. Therefore, when the load is tion whether or not it should be considered
inclined tensile forces near beam supports. placed closer to the support, more load is a slab. Deep slabs react similarly to deep
required to produce a tensile stress sufficient beams in that loads are transferred to the
to cause cracking. Thus the beam appears to supports by arch action.
ENHANCED SHEAR STRENGTH increase in shear capacity, which is not the
NEAR SUPPORTS case. Again, a misnomer associated with shear. Distribution of shears in slabs
Beams that are subjected to point loads Yield-line analysis is seen as a useful tech-
near the supports exhibit what is referred nique to determine the collapse load of slabs
to as enhanced shear strength (SABS 0100- SHEAR STRESSES IN SLABS (Johansen 1963). However, the technique
1 2000; BS 8110: Part 1 1985; Rowe et extends beyond its traditional application.
al 1985). Tests seem to indicate that the Shear in slabs Yield-lines not only signify the location of
shear capacity increases as the point load is Another ambiguity with shear is why maximum principal moments (where yield-
placed closer to the end of the beam. shear steel is not required in slabs if the ing occurs), but also the location of zero
Enhanced shear capacity implies that the calculated shear capacity is greater than shears. In a beam, the maximum moment
shear strength of the beam increases as the the applied shear stress. This is contrary at mid-span is the location of a possible
load is placed closer to the support (ie, an to other elements (ie, beams and columns) hinge, the point of zero shear as well as the
increase in vc, which is a function of the con- where shear steel is necessary regardless of point in which the load sheds to each sup-
crete strength, quantity of the main tension the magnitude of the applied shear. There port. The same is true of slabs. The location
steel and depth of the member). However, the are several reasons: First, at working loads of a yield-line is also the location of maxi-
area and spacing of the shear steel, the con- (linear elastic) the shear stress distribu- mum principal moments, zero shear and
crete strength and the depth of the beam usu- tion along a vertical cross-section of a slab the dividing line in which loads are shed.
ally do not vary; therefore the shear strength is parabolic in shape and the maximum This explains why the slab demarcation
is constant from one end of the beam to the shear is located at the neutral axis. Slabs lines illustrated in figure 10 of SABS 0100-1
other. The term ‘enhanced shear strength’ is are relatively thin and the main steel is in (2000) resemble a yield-line pattern. The
therefore misleading, since the shear capacity close proximity to the neutral axis, which answer is simple: they are the same.
does not vary along the length of the beam. will resist the shear stresses (or, more cor- It is common knowledge that the maxi-
An ABAQUS finite element solution is rectly termed – inclined tension stresses). In mum shears in slabs exists along the edges.
used to observe the flow of stresses with beams, the neutral axis is further away from Knowing the pattern and location of zero
point loads placed at various distances the main steel and therefore this region is shears enables one to construct a three-
from the face of the support. In figure 10a, susceptible to failures associated with shear. dimensional distribution of shears in a slab.
a point load is placed at a distance 2d from Furthermore the main steel in slabs and In a beam subjected to UDL, the shear dis-
the end of the support and in figure 10b the beams has a dual function – the mid-span tribution is linear and triangular. In one and
point load is placed at a distance d, where steel resists the flexural stresses and the two-way slabs, the distribution is similarly
d is the distance from the compression face steel at the ends resist the inclined tensions. linear, but wedge shaped.

12 Journal of the South African Institution of Civil Engineering • Volume 48 Number 4 December 2006
 probable cause of cracking in a beam or slab
is a result of tension stresses.
The postulates present a series of laws
associated with stress. A combination of
these laws presents a formal interpretation
Figure 11 Principal stresses in a section of a flat slab of shear. Whatever the structural or load
configuration may be, the postulates are
Punching shear Figure 12 Earthquake devastation along Highway 1,
applicable and present a logical interpreta-
Point loads on slabs tend to punch through, California (Photograph by John O’Hara, The tion of crack patterns that we often incor-
which is referred to as punching shear Quake of ’89) rectly associate with shear.
(SABS 0100-1 2000; BS 8110: Part 1 1985).
A flat slab is also subjected to punching Vertical shear lines are visible, indicative of
shear at the location of columns. A section shear failure. The picture is a classic exam- REFERENCES
of a flat slab is illustrated in figure 11. The ple of a true punching shear failure – in this BS 8110: Part 1: 1985. Structural use concrete. London: British
stress pattern illustrates the flow of prin- case, the result of an earthquake which hit Standards Institute.
cipal tensile and compressive forces in the California in 1989. DePaiva, H A R & Siess, C P 1965 Strength and behaviour of
slab – the pattern is similar to water flowing deep beams in shear, ACSE Proceedings (ST-5), Part 1:19.
down a drain. Ferguson, P M, Breen, J E, & Jirsa, J O 1988. Reinforced con-
Also illustrated in figure 11 is a typi- CONCLUSIONS crete fundamental. 5th ed. New York: Wiley.
cal punching shear crack in a flat slab. The Mysteries associated with shear are gener- Johansen, K W 1963. Yield-line theory. Portland Cement
failure lines are parallel with the compres- ally of our own making. The profession has Institute.
sive trajectories. Tensile splitting forces exist somehow inherited incorrect nomenclature JSCE No 3 1984. Recommendations for the design and construc-
perpendicular to the compressive forces and to describe modes of tension failure. This tion of steel fiber reinforced concrete. Tokyo: Japan Society of
therefore crack lines are prone to develop in has led to an incorrect understanding of Civil Engineers.
these regions. A close inspection of figure what shear is and how it affects reinforced Kong, F K & Evans, R H 1987. Reinforced and prestressed con-
11 will also indicate that the compressions concrete. Our lack of understanding may crete. 3rd ed. Berkshire: Van Nostrand Reinhold.
increase in magnitude closer to the column. also be attributed to our persistence in Leonhardt, F 1965. Reducing shear reinforcement in concrete
Thus failure lines tend to form closer to the matching linear-elastic methods of analy- beams and slabs. Magazine of Concrete Research, 17(53):
column, which is consistent with conven- sis with ultimate limit state theory. Shear 187–198.
tional theory. However, the failure pattern is failure rarely occurs in concrete, with the MacGinley, T J & Choo, B S 1985. Reinforced concrete. 2nd ed.
actually inclined tensile cracking from com- exception of composite construction. Shear London: Spon.
pressive thrusts rather than punching shear. failure is a translation failure as opposed to Rowe, R E et al 1985. Handbook to British Standard
A true punching shear failure would other failure types such as bending, which is BS8110:1985 structural use of concrete. London: Palladian.
cause columns to displace vertically through rotational (or hinge). What we term a shear SABS 0100-1, Ed. 2.2 2000. The structural use of con-
the slab, similar to the failure shown in failure is actually an inclined tension failure crete, Part 1: Design. Pretoria: South African Bureau of
figure 12. As observed, several columns of or a splitting failure along a compression Standards.
a bridge deck punched through the slab. thrust. Whatever the case may be, the most

Journal of the South African Institution of Civil Engineering • Volume 48 Number 4 December 2006 13