You are on page 1of 22

Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 1 of 22

4 Design
Design procedures are, even more than scientific theories and mathematical techniques of
structural analysis, the essential tools of the engineering designer’s trade [Addis 1990, p. 174]. As
theories of structures and materials advance it becomes difficult to apply them in a traditional
design process. This creates challenging issues to expand the philosophy of structural design.

In the previous chapters the stringer-panel model was presented and its reliability was
investigated. In this chapter the model is used in the design process of reinforced concrete walls. 1

4.1 Introduction
The flow of forces in a structure is essential in structural design. Two extreme approaches can be
distinguished in designing the force distribution with a nonlinear model.

1. Optimise the structure allowing the maximum amount of yielding.


2. Carry a force where it appears and do not accept any yielding.

Related to these approaches are the simplified material models shown in Figure 33. At the left-
hand side of the figure, the classic linear-elastic material behaviour is shown for stress analyses.
The middle graph is the plastic behaviour which is used in limit analysis (approach 1). At the
right-hand side a new behaviour is shown which we will refer to as strengthening behaviour
(approach 2). At first sight this seems to be a very strange material but its advantage in design can
be easily explained.

Figure 33: Different material models can be used in design.

To explain strengthening behaviour we choose a hypothetical material that is represented with


curve a in Figure 33. We increase the load at the structure and at some point the material starts to
yield. If we want the material to carry this stress without yielding, we have to strengthen it locally,
for example with steel reinforcement. Subsequently, the material behaves according curve b and
the load is increased again. When the ultimate elastic strain is reached again and the load is to be
increased, once more we need to strengthen the material and curve c is obtained. The subsequent
strengthening of the material can be replaced by the vertical thick line in the figure. Thus,
strengthening behaviour appears.

Strengthening behaviour is the opposite of plastic behaviour. It is very stable in load controlled
computations and in only one design cycle the final distribution of forces can be obtained.
However, no solution exists when large described displacements or strains are applied to the
structure. For example foundation settlements and temperature loading cannot be computed when
a strengthening material is used.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 2 of 22

As explained in Section 2.1, also elastic-plastic behaviour has limitations in the design process. It
shows that the structure fails but does not tell how we can improve it. Between the extremes of
plastic behaviour and strengthening behaviour we need to use a material model that leads to a
good design in only a few design cycles.

4.2 Design Process


We started with simulating the behaviour of walls designed with a linear-elastic stringer-panel
model. One of the observations was that during cracking, forces were redistributing away from the
linear distribution but during yielding the redistribution was towards the linear distribution. The
latter should not be a surprise because the yield strength of the reinforcement was based on the
linear distribution. In theory when the load is increased and much of the reinforcement is yielding
we should arrive at the elastic distribution again. However, these redistributions occur with large
strains.

Another observation was that in walls the concentrated reinforcement (in the stringers) is only
about 20% of the total reinforcement. So, the distributed reinforcement (in the panels) is about
80% of the total reinforcement. The distributed reinforcement can often not be reduced because it
is required for crack control (see Section 4.3).

Finally, it was observed that the amount of concentrated reinforcement can be selected much more
accurately than the distributed reinforcement. For example, 578 mm2 steel can be provided by 9
combinations of bars while only 3 standard meshes are available to provide 578 mm2/m. 2

Subsequently, we temporarily adopted a cyclic design process with a full nonlinear stringer-panel
model as advised in literature [for example Gerstle 1991].

1. Choose shape and dimensions using experience and rules of thumb.


2. Establish all load cases and load combinations.
3. Perform a linear analysis for all load combinations.
4. Select reinforcement and improve the concrete dimensions.
5. Perform a simulation of each load combination.
6. Improve the designed reinforcement.
7. repeat 5 and 6 until no further improvement is needed.
8. Detail the reinforcement.

Compared to a traditional design process, the steps 5 and 6 are added. The stresses in a simulation
are always smaller than or equal to the strength of the material, so, they cannot be used to
determine dimensions. 3 Instead we have to look for damage as already pointed out in Section 2.1.
We soon found out that it requires many design cycles to arrive at reasonably optimised
reinforcement. Moreover, the economies in reinforcement were very disappointing. For example if
we can reduce the concentrated reinforcement by 30% using heavy redistributions then only 6% of
the total wall reinforcement is saved. 4 Often crack widths in serviceability conditions and
ductility restricted the optimisation.

Addressing the problem of many design cycles, we needed something that would tell us how much
reinforcement to use. The answer was found in leaving large parts of the model linear. If we
expected that yielding in a particular part of the structure would create a more favourable force
distribution, then only the stringers and panels in that part were made nonlinear and all other
components remained linear. In this way it could be observed how forces in other parts of the
structure would increase as a result of the yielding. Reinforcement was selected accordingly.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 3 of 22

Considering the limited advantages of redistributions we focused on a fast design procedure


excluding the design cycles. It was found that it is successful when the reinforcement in the
stringers continues to behave linearly after the yield limit is reached. So, yielding of steel is
postponed but cracking of concrete is fully considered. In this way the amount of reinforcement
can be calculated from the stringer forces as is common in a traditional design process.
Additionally in the stringers, the concrete in compression is approximated linearly. The latter
makes the model less vulnerable to mistakes in the effective stringer area (see page 18). In Figure
34 the stringer constitutive models are illustrated.

In this process the panel behaviour is full nonlinear. When a panel yields the forces in the adjacent
stringers will increase and more stringer reinforcement will be required. The advantage compared
to a partly linear panel is that we often can just choose the minimum reinforcement mesh in the
panels. Any extra reinforcement can be put in the stringers which can be selected more accurately
than a panel mesh as already mentioned above.

As already presented in Chapter 2, the final design procedure we developed for structural concrete
walls has the following steps: 5

1. Choose shape and dimensions using experience and rules of thumb.


2. Establish all load cases and load combinations.
3. Perform a linear analysis for all load combinations.
4. Select the reinforcement and improve the concrete dimensions of the stringers.
5. Perform a nonlinear analysis of each load combination with non-yielding stringer
reinforcement.
6. Improve the reinforcement.
7. Perform a simulation with accurate material behaviour of the dominant load combinations
up to failure in order to check the design. 6
8. Detail the reinforcement.

Compared to a traditional design process, the steps 5, 6 and 7 are added. Despite the extra steps
the proposed procedure can be faster than what is currently being used because it is embedded in a
computer aided design environment.

Figure 34: Several constitutive relations can be selected for the stringer behaviour. The linear
behaviour is useful for initial design, the non-yielding behaviour is useful for improving the design
and the full nonlinear behaviour can be used when the design gets a final check.

Experience shows that a good initial design in step 3 and 4 can save a lot of time in the successive
steps of the design process. We adopted a linear-elastic stress analysis for the initial design with a
specially developed linear stringer-panel model which shows to perform well (see Section 2.2).

An alternative for the initial design could be plastic optimisation. Computer based optimisation
procedures have been developed for stringer-panel models to find the stress distribution with the
least reinforcement requirements [Damkilde 1993]. Two reservations made us decide not to
implement this optimisation. First, reinforced concrete is not always as ductile as necessary for a

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 4 of 22

plastic distribution of forces. Second, cracks in a reinforced concrete wall are completely ignored
in a plastic optimisation while experience indicates that durability requirements of a structure can
be more demanding than strength. Nevertheless, this subject is not exhausted and it would be
interesting to investigate plastic optimisation as a tool for initial design.

In step 5 and 6 the initial design is improved by taking into account the change of stiffness that
occurs when concrete cracks. A nonlinear computation is done for all load combinations
individually. The serviceability combinations can be used to assess deformations and crack
widths. A special feature of the nonlinear computation is that the reinforcement of the stringers
does not yield. Instead it continues to behave linearly beyond its yield strength (see Figure 34).
The envelopes of stringer forces for all ultimate load combinations are used to select new
reinforcement if necessary. 7 Of course in this, appropriate design values of the material have to
be used. The material safety factor can depend on the load combination that governs in a particular
part of the structure (see Section 2.7).

In the simulation of step 7, the design gets a final check using full nonlinear behaviour of the
material. The ductility of the design can now be assessed which for example is important for
earthquake loads.

Finally in step 8, the reinforcement is provided with anchorages and stresses at the support platens
are checked.

4.3 Durability Condition


Conditions can be derived that a good design has to fulfil. They provide an important assistance
for the designer. In this and the next section, design conditions are derived for respectively
durability and strength of stringers and panels.

The durability of concrete is commonly related to crack widths. When the concrete cracks there is
a danger that this crack creates a weak spot in the structure. Any increase of the deformation will
localise at the weak spot as one wide crack. In order to get many small cracks instead of one large
crack the reinforcement alone needs to be stronger than the reinforced concrete just prior to
cracking. This amount of reinforcement is commonly referred to as minimum reinforcement. 8
Besides prevention of localisation, the minimum reinforcement prevents abrupt failure and
increases the ductility of the structure. 9

Stringer
In case of a stringer, the force in the reinforcement at the first crack should be larger than the force
to create a second crack (see Figure 35).

In this fy is the yield stress of the steel, ss is the steel stress in the potential second crack and fcr the
cracking strength of the concrete. As and Ac are the cross-section area of the reinforcement and the
concrete, respectively. Note that Ac only includes the concrete, so, the total stringer area is A = As
+ A c.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 5 of 22

Figure 35: The force in the reinforcement at a crack should be large enough to trigger a second
crack in the stringer.

Before the second crack appears the reinforcement strain es equals the concrete strain ec.

es = ec

So, at the onset of cracking

In this Es and Ec are the Young’s moduli of reinforcement and concrete, respectively. With
substitution of this relation in the inequality the minimum reinforcement ratio is simply derived.
10

Often this is approximated as which is about 3 % too small.

Panel
In case of a panel, the force in the reinforcement at the first crack should be larger than the force to
create a second crack (see Figure 36). We assume a reinforcement ratio r in both the x and the y
direction and consider equilibrium perpendicular to the crack. 11

In this is the reinforcement area in the x direction, is the


reinforcement area in the y direction and is the concrete area of the crack. The length
of the crack is l and t is the wall thickness.

Figure 36: The forces in the reinforcement at a panel crack should be large enough to trigger a
second crack.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 6 of 22

Substitution in the equilibrium relation gives

Just before cracking the steel strain equals the concrete strain.

Because the largest principal strain e 1 is larger than both e xx and e yy the relation can be safely
replaced by

Just prior to cracking the strain is

With this the inequality can be simplified to

This result is consistent with that found for the stringer. If cracks need to be controlled, we have to
apply minimum reinforcement where the concrete is substantially tensioned. When a part of
concrete in a direction is compressed for all load combinations, obviously no tension
reinforcement is required in this direction. For example due to prestressing or dead load. Also
when the concrete is slightly tensioned it does not necessarily need to be reinforced because even
if one localised crack occurs it can have an acceptable crack width. Nevertheless, for these
situations codes of practice prescribe some reinforcement. For example a minimum reinforcement
of 8 mm bars with a spacing equal to the wall thickness but less than 300 mm. The minimum
reinforcement often is the largest contribution to the total amount of reinforcement in a structural
concrete wall.

4.4 Strength Condition


A distribution of forces can be computed with a stringer-panel model. If possible an envelope can
be constructed of the forces of all load combinations of the ultimate limit state. This can
conveniently be used to select reinforcement in a wall.

Stringer

The software computes stringer forces N1 and N2 at both ends of each stringer. It is generally
impossible to choose or improve the reinforcement in such a way that no further redistributions
occur. The best we can do is to simply select reinforcement that can carry the computed forces.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 7 of 22

In this N is the largest of N1 and N2. As is the cross-section area of the reinforcement and fy is the
yield strength of the reinforcing steel. If the stiffnesses of the newly designed stringers is close to
the old stiffnesses not much redistribution can be expected and no further analysis is required.

Panel

We restrict ourselves to a panel with orthogonal reinforcement in the x and y direction in order to
obtain a practical reinforcement layout. The software will compute the forces nxx, nyy, nxy, and the
principal stresses n1, n2 in each panel. The principal stresses are positive if tensile and n1 is larger
than n2 (in an algebraic sense). The panel forces n are per unit length. For example kN/m.

As with the stringers it is impossible to avoid redistributions. Applying local plasticity we can
choose the reinforcement in both directions to carry the largest principal force

In this nsx is the force in the x reinforcement and nsy is the force in the y reinforcement. No crack
rotation is necessary and the concrete stress increases from nc = n2 before cracking to nc = n2 - n1
after cracking (see Figure 37 middle picture) [Nielsen 1984, pp. 90]. However, this is not the most
economic reinforcement especially when the bars are placed in the principal directions.

Figure 37: Forces in reinforced concrete. At the left-hand side are shown the forces that have to
be reinforced. The plastic forces in the middle picture do not require a crack rotation while the
forces in the right-hand picture need a crack direction of p/4 rad with the orthogonal
reinforcement directions.

It is also possible to optimise the amount of reinforcement allowing crack rotations. The result is
(see also Figure 37 right-hand picture)

These reinforcement forces are valid for tension in both panel directions. For other forces less
reinforcement can be applied as shown in Table 6 [Nielsen 1984, pp. 75-80] [CEB-Bulletin 1993,
p. 189]. The left-hand two columns of the table give the load situation of the panel, while the
right-hand three columns give the accompanying reinforcement and concrete forces.

nxx nyy nsx nsy nc

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 8 of 22

<0 0 0 n2

Table 6: Optimal reinforcement forces according to local plasticity

In case of the linear stringer-panel model as used in the initial design, only shear forces occur in a
panel. So,

and

This simple result makes it easy to select mesh reinforcement for initial design of the structure.
The reinforcement ratio in each direction x and y becomes

We can decide to use less reinforcement than this ratio. As a consequence the panel reinforcement
will yield and if possible its force will be redistributed to adjacent stringers and panels.

Figure 38: The deep beam is loaded with one load case consisting of concentrated and distributed
loads. Shown are design loads for the ultimate limit state. The beam thickness is 250 mm

4.5 Example
The deep beam of Figure 38 was first designed by Despot at the ETH in Zürich as an example of
plastic optimisation [Despot 1995, pp. 97-103]. In this section we show the design of this beam

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 9 of 22

using a stringer-panel model. The 8 steps of the design procedure are outlined on page 23 and
page 53.

1 Choose shape and dimensions using experience and rules of thumb.

Often, the geometry of the wall at hand is already available as an AutoCAD drawing. The stringer-
panel model is drawn on top of this drawing with commands similar to the standard AutoCAD
commands. Stringers are drawn around the hole, at the edges of the wall, at supports and at
concentrated loads (see Figure 39). The panels are simply drawn in-between the stringers. 12 The
thickness of each panel is 250 mm and each stringer has a cross-section of 300 x 250 mm. The
stringers and panels have no reinforcement yet.

2 Establish all load cases and load combinations.

The load on the beam is shown in Figure 38 too. It consists of two horizontally and two vertically
distributed line loads, in total about 5000 kN. The top edge of the beam is loaded with two
concentrated forces of 1000 kN and 150 kN. The two vertically distributed forces at the right-hand
side of the hole and at the right-hand edge of the beam are due to walls perpendicular to the face
of the beam. The drawn forces belong to one load case and consequently only one load
combination is present for the ultimate limit state and one for the serviceability limit state.
Distributed forces are lumped to stringer ends. Self weight of the material - about 400 kN - is
neglected.

Figure 39: The stringer-panel model is drawn on top of the deep beam.

Figure 40: Linear stringer forces and panel forces are computed and displayed in only a second.
Tensile forces are black and compression is grey. The panel shear forces are shown in kN/m for a
convenient selection of the distributed reinforcement.

3 Perform a linear analysis for all load combinations.

The program computes the model deformation and the forces in the elements in only a second (see
Figure 40). Since the panels carry only shear forces and no normal forces, all normal forces are

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 10 of 22

carried by the stringers. The stringers are positioned at the edges of the model resulting in a large
lever arm for the moment and an efficient reinforcement layout.

4 Select the reinforcement and improve the concrete dimensions of the stringers.

The forces are used to select reinforcement in the stringers and panels. Of course, a minimum
reinforcement has to be present for crack control in panels which are tensioned. The minimum
reinforcement ratio is calculated with a characteristic value of the concrete tensile strength fcr and
a characteristic value of the reinforcement yield strength fy (see page 56 and Table 7). 13

A standard mesh at both surfaces of Ø12-150 = 1508 mm2/m can fulfil this requirement. The panel
with the largest shear force needs 659000 / 460 = 1430 mm2/m of reinforcement for strength in
both directions (see Figure 40 and Table 6). So, the minimum reinforcement is enough for this
panel and more than enough for the other panels.

We choose not to reinforce the horizontal tensioned (black) stringers in the middle of the wall
above the right-hand support because it is expected that the distributed reinforcement can carry
this force too. We just remove these stringers and see what will happen in the subsequent analysis.
Redistributions will occur and perhaps reinforcement elsewhere has to be increased.

Figure 41 shows the selected reinforcing bars in the stringers and standard meshes at both surfaces
of the beam. The concrete dimensions of the stringers are 250 mm thick and 140 to 900 mm wide
depending on the available wall material and the effective tension area. The panels are each 250
mm thick. The properties of the wall materials are summarised in Table 7.

Concrete Steel Reinforcement


Compressive Strength - 19.5 MPa Yield Strength 460 MPa
Young’s Modulus 30000 MPa Young’s Modulus 200000 MPa
Tensile Strength 2.2 MPa Hardening Modulus 0 MPa
Ultimate Strain - 0.0035 Ultimate Strain 0.05

Table 7: Design properties of the wall material for the ultimate and serviceability limit state. The
compressive and yield strength are factored characteristic values and the remaining properties
are average values.

Figure 41: The selected reinforcement consists for a large part of meshes and a number of bars.
Each standard welded mesh is drawn with a diagonal line and a label. The label A refers to two

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 11 of 22

meshes with bars Ø12-150 in both directions. The meshes overlap to provide development length.
The bars in the beam hole are to be cut at the construction site.

5 Perform a nonlinear analysis of each load combination with non-yielding stringer


reinforcement. 14

In a nonlinear analysis of the model the load factor l is incremented from 0 until 1 at which the
total load combination is carried by the model. A linear constitutive relation was selected for the
stringers without reinforcement and a non-yielding relation for the reinforced stringers (see page
52). The computation takes about one minute during which the load-displacement curve of the
structure is plotted on the computer screen. The resulting stringer and panel forces are presented in
Figure 42. The numbers plotted in the squares are the shear stresses in the middle of the panels.
Also normal forces occur in the panels, but these are not shown in the figure. Figure 43 shows the
principal stresses in the middle of the panels which gives a good impression of the flow of forces.

Figure 42: The stringer forces and panel forces at ultimate design load (l = 1.0) show the
redistributions. At the right-hand side the panel reinforcement carries a substantial part of the
forces which were carried by the stringers before (compare with Figure 40). At the left-hand side,
forces are redistributed from below to above the hole of the deep beam

Figure 43: The principal stresses in the panels give an impression of the flow of forces through
the structure.

In this example we define the design load in serviceability conditions as 70 % of the design load
for ultimate conditions. 15 The largest crack occurs in the vertical stringers at the left-hand edge of
the beam with an average width of 0.53 mm.

6 Improve the reinforcement.

The analysis shows that at the right-hand side of the model much of the previous stringer forces
are now carried by the panels (see Figure 42). Some redistribution occurs from below the hole to
above the hole because the bottom stringers crack over their full length and elongate considerably.
The panel above the hole starts to fail and in doing so it dilates substantially. Consequently, the
adjacent stringers have to elongate and since they are not allowed to yield the stringer forces

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 12 of 22

increase, showing exactly how much reinforcement is needed to prevent failure. The stringer
reinforcement was improved as shown in Figure 44.

Figure 44: The final reinforcement layout consists of 1220 kg steel in the panels and 285 kg steel
in the stringers.

Figure 45: The simulation of the beam behaviour up to the ultimate limit state. At the vertical axis
the load factor l is displayed and at the horizontal axis the vertical displacement of the top of the
beam at the concentrated load just right of the hole. The model can carry 20 % more than the load
combination requires.

7 Perform a simulation with accurate material behaviour of the dominant load combinations up to
failure in order to check the design.

The simulation was done using full nonlinear behaviour of stringers and panels. The ultimate load
occurs at a load factor l = 1.2 and the deflection at collapse is 28 mm (see Figure 45). The cracks
in the stringers and panels at service load (l = 0.7) are shown in Figure 46. The largest cracks
have an average width of 0.29 mm which is acceptable in normal environmental conditions.

The total mass of steel reinforcement is 1500 kg, which consists of 1220 kg minimum
reinforcement in the panels and only 290 kg in the stringers. This is practically the same as first
design based on the linear stringer-panel model which had totally 1530 kg of reinforcement. The
plastic optimisation - as referred to in the beginning of this section - gives a reinforcement mass of
1070 kg. The difference is mainly caused by a larger minimum reinforcement used in the stringer-
panel model which was considered to be necessary for crack control. The reinforcement quantities
mentioned do not include detailing reinforcement or losses due to cutting.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 13 of 22

Figure 46: Only a few elements are cracked at serviceability conditions (l = 0.7). The cracks are
displayed in the panel midst and stringer ends. The largest crack width is 0.29 mm.

This example clearly shows that the distributed reinforcement is much more than the concentrated
reinforcement in the stringers. The amount of minimum distributed reinforcement depends
strongly on the thickness of the wall. Consequently, the most important design parameter of a
concrete wall is its thickness. A thin wall does not only need less concrete material but less steel
reinforcement as well.

8 Detail the reinforcement.

Proper detailing of the reinforcement is essential in order to prevent premature failure. For
example at the right-hand top corner and the left-hand bottom corner of the hole a localised crack
will occur despite the presence of a minimum reinforcement. It is good practice to include a few
inclined bars in this corner to disperse the cracks as much as possible. Also sufficient development
length or hooks have to be applied because the stringer-panel model assumes that anchorage is
present at all bars. This subject is separately addressed in Section 4.7.

Figure 47: Deformed model and crack distribution in the ultimate limit state.

4.6 Ductility
In the previous section a deep beam was designed with a nonlinear model and it was shown with a
simulation that it fulfils its requirements. In this section the same beam is considered, first
designed with an elastic model and subsequently designed with a plastic model.

In a traditional design based on elasticity theory we jump from step 4 in the design procedure to
step 8 in which the bars are detailed (see page 53). This design, using a linear-elastic stringer-
panel model, was shown in Figure 41 on page 64. 16 A simulation of its behaviour showed that
after some redistributions the beam fails at a load factor l = 1.2. So, the strength of this design is
more than sufficient. Its largest average crack width at service load is 0.5 mm as found with the
nonlinear analysis before. This is clearly too large because the factored average crack width
should be smaller than about 0.4 mm for normal environmental conditions.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 14 of 22

Figure 48 shows the reinforcement that was obtained by plastic optimisation. A stringer-panel
model was made with this reinforcement and its behaviour was simulated. Already before the
service load is reached, excessive cracking occurs at the left-hand top of the model and above the
hole at its right-hand corner. The stringer-panel model fails at l = 0.7 when the reinforcing bars
above the hole break. The ductility of the material above the hole is not enough to allow the
plastic distribution of forces to develop. So, an important conclusion is that - according to the
stringer-panel model - for this wall a design based on the plastic force distribution is not safe.

Figure 48: Reinforcement according to plastic optimisation [Despot 1995, p. 102]. The total mass
of the reinforcement in this design is only 1070 kg.

However, the influence of hardening of the reinforcing bars on the ductility of the stringers is not
included yet in the stringer-panel model. Hardening can trigger yielding in more than one crack,
which improves the concrete ductility substantially. In addition, the peak in the force distribution
of the horizontal stringers at the point of failure might be less pronounced in reality than predicted
by the stringer-panel model. This could also contribute to the local ductility. Nevertheless, the
stringer-panel model warns us to be extra careful. It would be very interesting to investigate this
deep beam experimentally, with both hardening and non-hardening reinforcement.

4.7 Detailing
The stringer-panel model assumes perfect bond between bars and concrete both during nonlinear
analysis and simulation. In Figure 49 a column overhang is shown with the force distribution
according to a stringer-panel model. The gradient of the stringer force shows where large bond
stresses are present between bars and concrete. At these stringers, the bars might be pulled out of
the concrete. 17 This has to be prevented with sufficient anchorage of the bar ends, like
development length, hooks or headed bars. Clearly, the top horizontal bars need to be anchored at
both sides and the vertical bars at the left-hand side need to be anchored at the top.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 15 of 22

Figure 49: A column overhang is modelled with a stringer-panel model. At the left-hand side the
stringer-panel model is shown and at the right-hand side its distribution of forces.

When an anchorage is present the local distribution of forces changes considerably. The shear
force at the bars will decrease and a part of the force will go directly to the anchor. In the extreme
situation when the anchorages carry all the force we obtain the distribution as shown by the strut-
and-tie model in the left-hand part of Figure 50.

Figure 50: Force distribution in the column overhang with anchored reinforcement. The left-hand
figure shows a strut-and-tie model while the right-hand figure shows the corrected forces of the
stringer-panel model.

In reality the distribution of forces will be in-between the distribution of the stringer-panel model
and that of the strut-and-tie model. The strut-and-tie distribution is safe since it clearly shows the
need for anchorages where the stringer-panel model does not. The remedy for detailing with a
stringer-panel model is simple: Continue the stringer force towards the edge of the structure
constantly, select reinforcement accordingly and provide proper anchorage.

Even when the average bond stress can be carried, it is recommended to provide anchorage
because a stringer-panel model is too course to display a local peak in the distribution of bond
stresses. This results in some more reinforcement than necessary but is a safe solution.

Obviously, a strut-and-tie model can be used best for the design of details while the stringer-panel
model is used best for the global distribution of forces in a wall. Many examples of proper
detailing with strut-and-tie models are provided in references [Schlaich 1987] and [Bergmeister
1993].

4.8 Applications
Several potential applications of stringer-panel models are shown in this section. Figure 51 shows
a continuous deep beam with a distributed top load supported by columns. The stringer-panel
model of this beam is small because symmetry of the structure and the load has been used. The

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 16 of 22

figure shows two models, the left-hand model is the stringer and panel layout and in the right-hand
model the force distribution is shown. The force distribution is computed with linear material
behaviour.

Figure 51: A continuous deep beam with a distributed top load

Diaphragms in box-girder bridges transfer support reactions into the box-girder. Figure 52 shows a
cross-section of a box-girder with diaphragm and maintenance hole. In this example the bridge
carries a full traffic load, however obviously, also other load cases need to be considered in design
like eccentric load and wind load. Compared to strut-and-tie models the stringer-panel model has
the advantage that only one model needs to be made for all load combinations.

Figure 52a: A diaphragm in a box-girder bridge. The figure shows a cross-section of the box-
girder with a top load. Note that the single support at this diaphragm does not carry a possible
torsional moment in the bridge.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 17 of 22

Figure 52b: A diaphragm in a box-girder bridge. The top figure shows the stringer-panel model.
Most of the top load is transferred through the webs of the box-girder to the bottom of the model.
The bottom figure shows the linear distribution of forces in the diaphragm.

Figure 53 shows a deep beam with a hole. It illustrates that, already for a small wall, a strut-and-
tie model can become quite complicated.

Figure 53: A deep beam with a hole. The left-hand figure shows a stringer-panel model with its
linear force distribution. The right-hand figure shows a strut-and-tie model of the beam [Schlaich
1987, pp. 119-123].

In his dissertation, Hottman gives a design example of a 16 m high and 8 m wide wall with 2
rectangular holes of 3 m edge length each [Hottmann 1995, pp. 137-140]. Figure 54 shows this
wall with a stringer-panel model and its linear distribution of forces.

Figure 54: A stringer-panel model of a high wall with holes.

A stringer-panel model can also be used for three-dimensional assemblies of walls. Figure 55
shows the deformation of a box-girder bridge with a curved bottom flange. 18 Figure 56 shows a
stringer-panel model of a caisson that was used in the foundation of the Storebælt bridge in

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 18 of 22

Denmark. Of course, the walls in these structures do not only have membrane forces but bending
moments as well. However, the contribution of bending to the overall strength is often small.
Especially when concrete cracks, the moments occur mainly because bending has to follow the
global deformation (compatibility moments). So, a stringer-panel model can be used to obtain the
global force distribution and deformation while subsequently local behaviour can be analysed with
finite element models or strut-and-tie models.

Figure 55: A stringer-panel model of a hollow box-girder. The top picture shows the model and
the bottom picture the deformations due to a point load. Only panels in half the box-girder are
displayed so that distortion of the cross-section can be clearly observed.

Figure 56: Three dimensional stringer-panel model of a caisson

Footnotes
1.

It can be expected that some of the conclusions on design of structural concrete walls with
advanced models also apply to other structural elements and structures like concrete floors
and steel frames.

2.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 19 of 22

If we accept that the reinforcement can be 3% too small or 7% too large then the following
bars provide 578 mm2 of reinforcement: 2Ø16+6Ø6, Ø20+5Ø8, 4Ø12+4Ø6, 5Ø12,
5Ø10+Ø16, 4Ø12+3Ø8, 3Ø16, 2Ø16+4Ø8, 5Ø12+Ø8.

If we accept the same tolerance for distributed reinforcement we find the following standard
meshes to provide 578 mm2/m: 2Ø8.5-200, 2Ø7.5-150, 2Ø6-100.

3.

The Eurocode suggests that reinforcement in walls or slabs can be based on the forces or
moments computed with nonlinear models [ENV 1992, A2.6 (3), A2.7 (3)]. Obviously this
is not always the case. For example, consider a reinforced concrete beam that fails
prematurely. The stresses in the bars can be computed at the location of the failure with a
nonlinear model. However, if we use these stresses to design new reinforcement we get the
bars that are already there.

4.

It has to be mentioned that only one load combination was used in the referred designs.
More than one combination creates more room to optimise reinforcement because the
strength for one combination can be used in another combination too. However, it makes
optimising even more complicated and the restrictions imposed by crack widths remain.

5.

SPanCAD does not restrict the order of the design process: Many procedures are possible
depending on the preferences and the experience of the designer.

6.

It is common to use the word analysis or stress analysis for the phase of the design in which
linear-elastic stresses are computed. If a nonlinear model is used it is commonly referred to
as a nonlinear analysis. However, since the nonlinear computation in step 7 has very little to
do with stresses, the author prefers to use the word simulation instead.

7.

It is not always possible to draw an envelope of physical quantities. For example a largest
stress tensor is not defined. Sometimes it is possible to define an envelope but when plotted,
it cannot be interpreted properly. For example, if we plot the largest displacements of each
point of a three dimensional model for all load combinations, we get a strange shape with
holes and overlaps which by no means improves our understanding of its structural
behaviour.

The linear stringer-panel model has only shear forces in the panels for which an envelope
can be computed. In the nonlinear model, however, an envelope of the panel forces does not
exist because it includes not only shear forces but normal forces as well. In this case we can
determine the reinforcement in a panel for each load combination and use the largest
requirement in each direction. In theory it is even possible to optimise the local
reinforcement requirement including all load combinations. As pointed out in footnote 3 on
page 51, this is only useful for panels that did not yield in the nonlinear analysis.
Fortunately, often it is sufficient just to choose a convenient minimum reinforcement in the
panels.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 20 of 22

Minimum reinforcement is not enough to prevent localisation due to geometrical stress


concentrations at for example re-entrant corners of a hole in a wall.

9.

While large cracks are clearly a serviceability limit state, abrupt failure or ductility is
important for ultimate conditions. This introduces confusion as to what design values to use
in the calculation of the minimum reinforcement.

The ductility of the failure not only depends on the distributed reinforcement but on the
main reinforcement in adjacent stringers as well. So in general, a ductility requirement
cannot be used to derive a minimum reinforcement ratio for panels.

As a consequence the minimum reinforcement is only for crack control and should be
calculated with serviceability design values. If a design is not sufficiently ductile a
simulation will show this and appropriate improvements can be made.

It is noted that for slender beams it is possible to derive a minimum reinforcement ratio
from the requirement of ductile failure. Of course, this ratio should be calculated with
ultimate design values.

10.

Note that the definition of the reinforcement ratio r = As / A, in this text differs from that
what is often used in literature r = As / Ac. The reason is that many formulations become
simpler with the definition of this text. However, practically the difference is negligible.

11.

Different reinforcement ratios in the x and y direction result in the same amount of
minimum reinforcement.

12.

SPanCAD automatically draws the components of a stringer-panel model on special layers


of the AutoCAD drawing. This makes it possible to view types of components and the
forces of each load case separately.

13.

Despot chooses a minimum reinforcement ratio of 0.00268 in both directions which is very
small even in relation to the small concrete compressive strength [Despot 1995, p. 99].

14.

In the computations of this text a constant shear field is used in the panel formulation (see
Appendix 3). In an early publication [Blaauwendraad 1997] we used instead a linear varying
shear field in the panel, which, however, results in a less close resemblance with
experiments. In the mentioned publication a more developed redistribution was obtained for
the deep beam of Despot because the early panel failed somewhat prematurely.

15.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 21 of 22

This is equivalent to a partial safety factor on the load case of g = 1.43 in the ultimate limit
state.

16.

Note that a design with a linear finite element model is not the same as a design with a
linear stringer-panel model. The latter lumps the normal stresses to the edges of the model
which is in fact a local plastic redistribution that optimises the reinforcement layout.

17.

We considered to model slipping of the reinforcement bars in the concrete of stringers. This
would certainly contribute to the accuracy of the stringer-panel model. In that case a
simulation would show if a model fails prematurely due to insufficient anchorage. This is a
challenging task, however, not relevant to design because the danger of pull out can be
easily recognised in a distribution of forces and subsequently prevented with anchorage.

18.

With SPanCAD three-dimensional models can be drawn and the results of an analysis or
simulation (displacements, forces, cracks) can be displayed. However, drafting a three-
dimensional model requires much more skill in using AutoCAD than a plane model.
Moreover, the displayed results are more difficult to interpret.

Literature
• Addis, W., "Structural Engineering, The Nature of Theory and Design", Ellis Horwood
Limited, 1990, ISBN 0-13-850611-6
• Bergmeister, K., J.E. Breen, J.O. Jirsa, M.E. Kreger, "Detailing for Structural Concrete",
Center for Transportation Research, University of Texas at Austin, CTR 0-1127-3F, May
1993.
• Blaauwendraad, J. and P.C.J. Hoogenboom, "Discrete Elements in Structural Concrete
Design", Heron, Vol. 42, No. 3, 1997, pp. 159-168.
• CEB-Bulletin 213/214: CEB-FIP Model Code 1990, Design Code, Thomas Telford London,
1993, ISBN 0 7277 1696 4.
• Damkilde, L., J. F. Olsen and P. N. Poulsen, "A Program for Limit State Analysis of Plane,
Reinforced Concrete Plates by the Stringer Method", Technical University of Denmark,
Department of Structural Engineering, 1993.
• Despot, Z., "Methode der finiten Elemente und Plastizitätstheorie zur Bemessung von
Stahlbetonscheiben", (Finite Element Method and Plasticity Theory for Dimensioning of
Reinforced Concrete Disks), Institut für Baustatik und Konstruktion, ETH Zürich, IBK
Bericht Nr. 215, Birkhäuser Verlag, Basel, 1995, (In German).
• ENV 1992, "Eurocode 2, Design of Concrete Structures, Part 1, General Rules and Rules
for Buildings". European Prestandard, Brussel, CEN, 1991.
• Gerstle, K.H., A.K. Gupta, L.K. Nuss, C.G. Perchinelli, "Finite Element Analysis as a
Design Tool", International Workshop on Finite Element Analysis of Reinforced Concrete,
Columbia University, New York, June 3-6, 1991.
• Hottmann, H.U., "Bemessen von Stahlbetonbalken und -scheiben mit Öffnungen",
(Dimensioning of Reinforced Concrete Beams and Disks with Holes) Institut für
Tragwerksentwurf und -konstruktion Universität Stuttgart, 1995, (In German).
• Nielsen, M.P., "Limit Analysis and Concrete Plasticity", ISBN 0 13 536623 2, Prentice-
Hall, London, 1984.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019
Hoogenboom, Design of Structural Concrete Walls, Chapter 4: Design Page 22 of 22

• Schlaich, J., K. Schäfer, M. Jennewein, "Toward a Consistent Design of Structural


Concrete", PCI Journal, Special Report, Vol. 32, No. 3, 1987.

http://homepage.tudelft.nl/p3r3s/Dissertation/chap4.html 25-02-2019

You might also like