You are on page 1of 9

Struct Multidisc Optim (2010) 41:979–987

DOI 10.1007/s00158-010-0481-2

INDUSTRIAL APPLICATION

Optimization of piers for retaining walls


Luisa María Gil-Martín · Enrique Hernández-Montes ·
Mark Aschheim

Received: 5 March 2009 / Revised: 15 December 2009 / Accepted: 1 January 2010 / Published online: 23 January 2010

c Springer-Verlag 2010

Abstract The ultimate strength design of reinforced con- 1 Introduction


crete members under flexure and axial loads is well known,
with solutions dating back to the 1950s. There is little The present work focuses on circular cross sections, where
motivation for engineers to change solution approaches the potential advantages of optimal reinforcement have
in the absence of a clear benefit, particularly given that more readily realized benefits. Traditionally, longitudinal
current solution approaches are well-known and do not reinforcement for circular cross sections has consisted of
raise controversy. Recently, a different solution approach a number of bars of uniform diameter spaced uniformly
has been presented in the form of Reinforcement Sizing around the circumference of the section, inset from the face
Diagrams and Optimal Domains, but applications to rect- of the member by the required cover distance (see Fig. 1).
angular sections were limited and thus the approach has Circular sections have many applications in civil engi-
been mainly of academic or theoretical interest. This paper neering. Design solutions for circular sections obtained with
presents an application to circular sections, which allow the available computer programs presume that the longitudinal
advantages to be realized in the commonly encountered sit- reinforcement consists of a uniform bar diameter (∅) dis-
uation of retaining walls supported by circular section piers. tributed uniformly around the perimeter of the section, at
Using the approach presented herein, longitudinal rein- a spacing, s (see Fig. 1). One such program is IECA 3.0
forcement can be reduced up to 50% compared with tradi- (2001), which was written to satisfy the requirements of
tional designs, allowing significant financial savings while Eurocode 2 (EC2). Davalath and Madugula (1988) provide
also reducing the greenhouse gas emissions associated a numerical procedure for the analysis/design of reinforced
with the reinforcement component of reinforced concrete concrete circular cross sections subjected to axial loads
construction. (compression or tension) and bending moments, based on
ACI 318 assumptions. Their procedure determines uni-
Keywords Retaining walls · Reinforced concrete · formly distributed reinforcement for any given size of circu-
Optimal reinforcement lar cross section or an optimal combination of reinforcement
and section diameter to minimize cost, based on uniformly
distributed reinforcement having a uniform bar diameter.
As the number of bars in the circular section increased,
the N –M interaction curve approaches one that has
polar symmetry, and thus the conventional reinforcement
configuration is ideally suited to members that experience
N –M demands that lay within an N –M interaction sur-
L. M. Gil-Martín · E. Hernández-Montes (B) face that is approximately symmetric in three-dimensions
Structural Mechanics, University of Granada, Granada, Spain (Mθ = constant, where θ is the angle from the Mx axis in
e-mail: emontes@ugr.es the Mx − M y plane). In some applications, such as retain-
M. Aschheim ing walls supported by circular piers, the N –M demands
Civil Engineering, Santa Clara University, Santa Clara, CA, USA lack this symmetry and hence substantial advantages may
980 L.M. Gil-Martín et al.

Minimum
Constant separation reinforcement
d1

Longitudinal
reinforcement

α
Transversal
reinforcement
Main reinforcement

Fig. 1 Traditional reinforcement of a circular section Fig. 3 Cross-section as defined by Weber and Ernst

be realized by using non-uniformly reinforced circular piers Kanagasundaram and Karihaloo (1990) formulated the
(see Fig. 2). minimization of cost designs of reinforced concrete struc-
Weber and Ernst (1989) developed interaction dia- tures problem as a non-linear mathematical programming
grams for reinforced concrete circular cross-sections having problem. Later, Adamu and Karihaloo (1994) and Adamu
asymmetrically distributed reinforcement. The interaction et al. (1994) proposed a numerical procedure for the eco-
diagrams allow engineers to design a reinforced circular nomic design of reinforced concrete element on the basis of
cross-section with asymmetrical reinforcement. Reinforce- Discretized Continuum-type Optimality Criteria (DCOC).
ment was assumed to be smeared into continuous plates, The costs to be minimized involved, among other fac-
with a thicker plate representing the main reinforcement tors, those of concrete and reinforcing steel and limits on
over an included angle of α, and a thinner plate representing bending and shear strengths were included as design con-
minimum reinforcement over the remainder of the section straints. This procedure was applied to rectangular and sym-
(Fig. 3), with engineers having to determine an equivalent metrically reinforced cross-section columns of RC frames
number of discrete reinforcing bars to be used for a given (Adamu and Karihaloo 1995).
set of values of α. The distance from the edge of the section In recent years the authors have developed a new
to the centroid of the reinforcement plate is given by d1 , as approach to the optimal proportioning of reinforcement in
shown in Fig. 3. the design of reinforced concrete cross sections. Classical
approaches, including the division between small and large
eccentricities described by Whitney (Nawy 2003), have
Upper view been subsumed by a more informed view in the form of
Piers wall
Reinforcement Sizing Diagrams (Hernández-Montes et al.
2004, 2005), Optimal Domains (Aschheim et al. 2006), the
Theorem of Optimal Section Reinforcement (Hernández-
Montes et al. 2008), and the treatment of multiple load
combinations (Lee et al. 2009). Although these contribu-
A A
tions are theoretically and pedagogically significant, further
clarifying the understanding of the design of sections for
axial load and moment, applications to rectangular sections
are limited in practice, thus reducing the utility of the
approach over traditional methods.
Section A-A The preceding work recognized, perhaps for the first
time, that an infinite number of reinforcement solutions can
Piers wall provide a section with adequate strength, and thus engi-
neers have the opportunity to select a combination of top
B B and bottom reinforcement that meets other objectives, such
as minimization of reinforcement, cost, or environmental
consequences (e.g. embodied greenhouse gas emissions),
or enhancement of curvature ductility, while providing
adequate strength.
The existence of an infinite number of reinforcement
Fig. 2 Example of pier wall retaining structure solutions for a cross section subjected to a combination
Optimization of piers for retaining walls 981

of axial load and moment can be understood by recogniz- σs (MPa)


ing that reinforcement can be sized to provide the tension
400
and compression forces required for equilibrium for many
different neutral axis depths. In this paper, the minimum
200
area of reinforcement is located among the infinite number
of possible solutions for particular configurations of lon-
gitudinal reinforcement that are defined parametrically. In −0.01 −0.008 −0.006 −0.004 −0.002 0.002 εs
the present work, an optimization algorithm is presented.
To facilitate simplicity in construction and drafting, bar −200
positions are considered explicitly using a simple param-
eterization scheme. As much as 50% of the longitudinal −400
steel weight can be saved by using this approach, relative to
that obtained with traditional procedures. Several examples
illustrate the approach and the reduction in reinforcement Fig. 5 Bilinear steel model, for steel of characteristic resistance of
500 MPa
that may be achieved.

2 Flexural strength design assumptions


cern for the locations where steel is present, and the stress
The strength design problem for combined flexure and axial
assigned to any steel within this zone is reduced by ηfcd .
load involves the simultaneous consideration of equilib-
Compression is considered positive.
rium, compatibility, and the constitutive relations of the
All cross sections of a concrete member are required to
steel and concrete materials at the section level. Some
resist combined moment and axial load without failure, as
assumptions required for ultimate strength design are pre-
determined according to conventional design calculations.
scribed by codes of practice, including the Bernouilli
Failure is considered to occur at critical sections by crush-
assumption that plane sections remain plane and the
ing of the concrete in compression. The possible distribution
assumption that strains in the reinforcement are equal to the
of strain over the cross section associated with the bend-
strains in the adjacent concrete at any level. Without limit-
ing ultimate state is illustrated in Fig. 6 according to the
ing the generality of the method, the provisions of Eurocode
design assumptions given in Eurocode 2. The strain εc3 is
2 (2002) are used herein for illustration.
the concrete strain corresponding to the maximum concrete
The concrete stress block prescribed in Eurocode 2 is
compressive strength according to EC2 and εcu3 is the con-
illustrated in Fig. 4, where the factor λ defines the effective
crete ultimate strain. Both strain limits, εcu3 and εc3 , which
height of the compression zone and factor η defines the
define points B and C, respectively, are represented in Fig. 6.
strength of the rectangular stress block. A bilinear, elastic-
The variable yna represents the depth of the neutral axis.
plastic model for steel is used (no strain hardening), as
The area (Ac ) and center of gravity of the compression zone
shown in Fig. 5, for a steel having a characteristic strength
(ycgc ) are function of yna :
of 500 MPa. In this case, the concrete compressive stress
was reduced to ηfcd . To facilitate computation, the concrete
Ac = Ac (yna )
stress is applied over the compression zone without con- (1)
ycgc = ycgc (yna )

Nc is the axial compressive force carried by the concrete


and Mc is the moment of force Nc with respect to the cen-
εc η fcd
ter of the circular section. Nc and Mc can be expressed as
function of the variable yna :
λyna
yna
Nc (yna ) = η f cd Ac (yna )
Neutral axis
(2)
Mc (yna ) = η f cd Ac (yna ) · ycgc (yna )

Each value of yna defines a unique ultimate strain


distribution—for example, plane A–B of Fig. 6. The strain
Strain distribution Concrete stress block
distribution can be expressed as a function of two variables:
Fig. 4 Concrete stress block and strain distribution the position of the neutral axis (yna ) and the position of
982 L.M. Gil-Martín et al.

Fig. 6 Strain distributions


B
considered at the ultimate
limit state λyna yna (1-εc3/εcu3)/(2R)

y C
s2

A
s1
0 εc3 εcu3

A An arbitrary point at the bottom fiber


B Concrete compression strain limit
C Concrete pure compression strain limit

the fiber (y), ε = ε(y,yna ). Strain in the longitudinal bars and n 2 is the number of bars separated by s2 . Areas A1 and
follows the same function ε = ε(yi , yna ), where yi is the A2 are the cross sectional areas of individual bars having
y-coordinate of the bar i, and yna is the depth of the neutral diameters ∅1 and ∅2 respectively (see Fig. 7).
axis.
The equations of equilibrium for Nd and Md , applied at
the center of gravity of the gross section are:


n1
Zone 1
Nd = Nc (yna ) + A1 σs (ε (yi , yna ))
i=1
ξ

n2
  
+ A2 σs ε y j , yna
j=1


n1
Md = Mc (yna ) + A1 σs (ε (yi , yna )) · yi
i=1 Zone 2


n2
   Maximum flexural moment
+ A2 σs ε y j , yna · y j (3) Md (kN.m)
800
j=1

600
The number of longitudinal bars is partitioned into n 1
and n 2 ; n 1 is the number of bars separated by a spacing s1 ,
400
ξ (º)
90 110 130 150

Nd (kN)
y N-M Interaction diagram
εcu3 η fcd 8000
∅1
yna λ yna 6000
ξ
90º
4000 100º
110º
120º
130º
2000 140º
150º
Md (kN.m)
∅2 200 400 600 800 1000

Strain distribution Concrete stress block -2000

Fig. 7 Definition sketch, showing notation and variables y and yna Fig. 8 Results for Example 1
Optimization of piers for retaining walls 983

Eurocode 2 requirements for longitudinal reinforcement Non-symmetrical circular piers for retaining walls are
of columns (also applicable to piers) are: protected by international patents property of the University
of Granada.

– Bars should have a diameter of no less than 8 mm.


– The minimum amount of total longitudinal reinforce-
ment must not be less than 0.10 N E D / f yd or 0.002 Ac , 3 Flexural strength solutions
where N E D is the design (factored) axial compression
force and Ac is the gross sectional area. For a given cross section and reinforcement configuration,
– Maximum area of reinforcement (away from splices) the N –M interaction diagram can be obtained by consider-
must not exceed 0.04 Ac . ing equilibrium for different values of yna , ranging from 0 to
– A minimum of four longitudinal bars are required in a ∞. The following simple example shows the advantage of
circular column. the application of non-uniformly distributed reinforcement
in a circular section.

The minimum clear separation between bars (according


to EC2) is the maximum of: the bar diameter, 20 mm, and
dg + 5 mm, where dg is the maximum size aggregate. 4 Example 1
Though not required by Eurocode 2, a maximum sepa-
ration among longitudinal bars should also be established. Figure 6 represents a cross section of a pier wall, i.e. a wall
In the examples presented, the maximum clear spacing was consisting of circular piers. The section analyzed is part of a
taken equal to 300 mm. wall constructed in the southern part of Spain, as a retaining

Fig. 9 Flow chart of


Optimization Process 1 D, fck, fyk, ∅, Md, Nd

Calculate s1 and s2

n1 bars @ s1

Add 1 bars @ s2
(*)

Yes
No
Finish Resist Md and Nd?

(*) Bars added at spacing s2 have to be added at the bottom, at the position that generates the
highest lever arm relative to the center of gravity of cross-section. When adding bars at s2, there
may be interference with the bars having spacing s1. Any bar provided in the first step that is
located within the region containing bars spaced at s2 or within a distance less than s2 from this
region must be removed.

Step 1 Step 2 Step 3


984 L.M. Gil-Martín et al.

wall for a high speed railway line. The pier has a diameter This example makes clear that rearranging the position
of 800 mm and is to be made from C-25 concrete (charac- of the bars causes the ultimate flexural strength to increase
teristic compressive strength of 25 MPa) and steel having a from 530 to nearly 740 kN m (see Fig. 8). The parameter ξ is
characteristic strength of 400 MPa. As originally designed, used here as a way to clarify the potential effect of reinforce-
its longitudinal reinforcement consists of 16∅20 (16 bars, ment configuration. In the next section, two optimization
each 20 mm in diameter) distributed uniformly as indicated procedures are presented.
in the figure, for a total reinforcement area of 5027 mm2 , or The evolution of the N –M interaction diagram with the
1.0% of the gross area. Concrete cover is 60 mm. For this variation of parameter ξ is also shown in Fig. 8, for the
example, two different separations are considered. Let ξ be example analyzed. N –M.
the angle that defines Zone 1, and therefore distinguishes
Zone 1 from Zone 2 (see Fig. 8). In this example, each zone
contains the same number of bars, and the total number of 5 Optimization approach
bars is maintained (i.e. 16∅20). So, ξ = 90◦ corresponds
to equal spacing (s1 = s2 ) of the bars in both zones, and The optimization process must address several variables.
increasing ξ means an increment in the clear spacing of Some of these can be established based on practical con-
bars in Zone 1 and a decrease in the clear spacing between siderations. It is clear, for example, that the clear spac-
bars in Zone 2. Under the constraint that each zone con- ing (or separation) between longitudinal bars in Zone 1
tains the same number of bars, the evolution of flexural should be the maximum allowed while separation in Zone 2
strength with ξ , assuming zero axial load, is plotted in should be the minimum allowed. Thus, a few approaches to
Fig. 8. optimization may be considered.

Fig. 10 Flow chart of


Optimization Process 2 D, fck, fyk, ∅1, ∅2, Md, Nd

Calculate s1 and s2

n1 ∅1 bars @ s1

Add 1 bars ∅2 @ s2
(*)

Yes
Finish Resist Md and Nd?

(*) Bars ∅2 added at spacing s2 have to be added at the bottom, at the position that generates
highest lever arm relative to the center of gravity of cross-section. When adding bars ∅2 at s2,
there may be interference with bars located at a spacing s1. Any bar (∅1) provided in the first
step that is located within the region containing bars spaced at s2 or within a distance less than
s2 from this region must be removed.

∅1 bars

∅2 bars
Iterarion 1 Iteration 2 Iteration 3
Optimization of piers for retaining walls 985

Table 1 Iteration to obtain a solution according to Optimization obtained, the bar diameter can be reduced, but this will
Procedure 1, for Example 2
require a larger number of bars. Similarly, a smaller num-
Iteration Bar composition Mu (kN m) ber of larger diameter bars may be found acceptable. The
discrete nature of the problem, associated with the use of
1 10∅20 572 an integer number of bars in each zone and the availabil-
3 3∅20 @ 45 mm + 9 ∅ 20 @ 270 mm 792 ity of bars in particular bar diameters, rather than diameters
5 5∅20 @ 45 mm + 9 ∅ 20 @ 270 mm 910 that would be optimal for a given problem, suggests that one
7 7∅20 @ 45 mm + 9 ∅ 20 @ 270 mm 1,212 cannot predict in advance which bar diameter solution will
9 9∅20 @ 45 mm + 9 ∅ 20 @ 270 mm 1,409 use the least weight of reinforcement. A similar reasoning
10 10∅20 @ 45 mm + 9 ∅ 20 @ 270 mm 1,502 can be applied to the section diameter, which for prag-
matic reasons, must be limited to the diameters for which
formwork and/or soil augers are readily available.

5.1 Procedure 1: optimization for a common


bar diameter (∅1 = ∅2 ) 5.2 Procedure 2: optimization for two bar diameters
(∅1 = ∅2 )
The steps of the optimization procedure are shown in Fig. 9.
For a given bar diameter, s1 and s2 (Fig. 6) are chosen to The steps of the optimization procedure are shown in
be the maximum and the minimum allowed, respectively. Fig. 10. If two different bar diameters may be selected,
Bars are located at a spacing s1 , and bars are added in the then minimization of total bar area (or weight) will lead to
tension zone (Zone 2) at s2 up to the point that the ultimate A1 and A2 being equal to the minimum and maximum bar
flexural strength is adequate (M Rd ≥ M E D ). The forego- sizes allowed (or considered acceptable by the engineer),
ing presumes that a solution can be obtained for the given respectively. The clear spacing between the bars (s1 ) will be
bar diameters and section dimensions. If a solution can be the largest allowed, for the smaller bar size, while s2 will

Fig. 11 Solution and Nd (kN)


interaction diagrams for 15000
Example 2, Optimization N-M Interaction diagram for ∅20 bars
Procedure 1, with ∅20 bars Iterations 1 to 10

10000
Iterations
1 Iteration 10
3
5
7
5000 9
10

Md (kN.m)
500 1000 1500 2000

Iteration 1 Iteration 3 Iteration 10


10∅20 @ 45mm + 9 ∅ 20 @ 270mm

10∅ 20 @ 45mm
986 L.M. Gil-Martín et al.

Table 2 Iteration to obtain a solution according to Optimization In a particular area, the retaining wall is composed of piers
Procedure 2, for Example 2
that are 1-m in diameter. Longitudinal reinforcement con-
Iteration Bar composition Mu (kN m) sists of 20∅25, and circular hoops consist of ∅8@300 mm.
The design flexural moment was 1,592 kN m and axial force
1 10∅10 158 is approximately zero. Steel reinforcement is B-500-S, hav-
2 1∅32 @ 64 mm + 9 ∅ 10 @ 270 mm 430 ing a characteristic strength of 500 MPa. Concrete, C-30,
4 3∅32 @ 64 mm + 9 ∅ 10 @ 270 mm 1,000 has a compressive strength of 30 MPa. The total area of lon-
6 5∅32 @ 64 mm + 9 ∅ 10 @ 270 mm 1,521 gitudinal reinforcement (20∅25) is 9,817 mm2 , or 1.25% of
the gross area.
be the minimum allowed for the maximum bar size. Thus,
the larger bar size will be located within Zone 2, and the 6.1 Optimization for a common bar diameter
number of bars, n 2 , will be calculated to provide adequate
strength for a given design moment (Md ) and design axial In this case we determine the optimum distribution for sev-
force (Nd ), following the optimization procedure of Fig. 10. eral bar diameters (20, 25 and 32 mm). The maximum size
As n 2 increases, reductions in n 1 have to be identified, based of coarse aggregate is 20 mm, and the cover considered is
on the space occupied by the larger bars, to avoid physical 60 mm. Because the allowable distance between bars (cen-
interference with the smaller diameter bars, and to maintain ter to center distance) is imposed to be less than 300 mm,
adequate spacing limits (separation) with the smaller bars. the minimum number of bars is 10.
The design moment can be resisted using a single bar
size (A1 = A2 ) at uniform spacing (s1 = s2 ) using 12∅32
6 Example 2 placed symmetrically (total area of 9,651 mm2 ). With a
reduction to bars of 25-mm diameter, 14∅25 are found ade-
The need for 500 m of retaining wall has been projected quate, using nine bars in Zone 1 and five bars in Zone 2.
along the national freeway A2 from Madrid to Barcelona. This solution has a total area of 6,872 mm2 , representing a

Fig. 12 Solution and 15000 Nd (kN)


interaction diagrams for
Example 2, Procedure 2, ∅10 N-M Interaction diagram for ∅10 and ∅32 bars,
and ∅32 bars Iterations 1 to 6

10000
Iterations
1 Iteration 6
2
4
6
5000

Md (kN.m)
500 1000 1500 2000

Iteration 1 Iteration 3 Iteration 6


5∅32 @ 64mm + 9 ∅ 10 @ 270mm

5∅32 @ 64mm
Optimization of piers for retaining walls 987

savings of 29% in reinforcement area. Using 20-mm diame- References


ter bars, 19∅20 are found adequate, with nine bars in Zone
1 and 10 bars in Zone 2). This solution has a total area Adamu A, Karihaloo BL (1994) Minimum cost design of RC beams
of 5,969 mm2 , representing a savings of 40% in reinforce- using DCOC. Part II: beams with uniform cross-section. Struct
Optim 7:252–259
ment area. Since for this solution, the total number of bars Adamu A, Karihaloo BL (1995) Minimum cost design of RC frames
is the same as in the original design, labor costs should be using the DCOC method. Part I: columns under uniaxial bending
very similar to those originally envisioned. Table 1 shows actions. Struct Optim 10:16–32
the six iterations needed to converge on the final solution. Adamu A, Karihaloo BL, Rozvany GIN (1994) Minimum cost design
of reinforced concrete beams using continuum-type optimality
Figure 11 presents the N–M interaction diagrams for the criteria. Struct Optim 7:91–102
solutions obtained for each of the iterations described in Aschheim M, Hernández-Montes E, Gil-Martín LM (2006) Optimal
Table 1. domains for strength design of rectangular sections for axial load
and moment according to Eurocode 2. Engineering Structures
29(8):1752–1760
6.2 Optimization for two bar diameters Davalath GSR, Madugula MKS (1988) Analysis/design of rein-
forced concrete circular cross sections. ACI Struct J 85(6):
Further savings can be realized if different bar diameters are 617–623
allowed. In this case, if ∅10 are used in Zone 1 and ∅32 are Eurocode 2 (2002) Design of concrete structures—part 1: general
rules and rules for buildings prEN 1992-1-1 (July). European
used in Zone 2, a solution is obtained for 9∅10 in Zone 1 Committee for Standardization. Brussels, Belgium
together with 5∅32 in Zone 2. This solution uses a total area Hernández-Montes E, Aschheim M, Gil-Martin LM (2004) The impact
of 4,728 mm2 , representing a saving of 51% of the orig- of optimal longitudinal reinforcement on the curvature ductility
inal reinforcement area. Table 2 shows the four iterations capacity of reinforced concrete column sections. Mag Concr Res
56(9):499–512
needed to reach the final solution. Figure 12 presents the Hernández-Montes E, Gil-Martín LM, Aschheim M (2005) The design
N–M interaction diagrams obtained for each of the itera- of concrete members subjected to uniaxial bending and com-
tions, corresponding to those identified in Table 2. pression using reinforcement sizing diagrams. ACI Struct J
102(1):150–158
Hernández-Montes E, Gil-Martín LM, Pasadas-Fernández M,
Aschheim M (2008) Theorem of optimal reinforcement for
7 Conclusions reinforced concrete cross sections. Struct Multidisc Optim
36(5):509–521
Engineers have better tools and greater responsibility to IECA 3.0 (2001) Prontuario Informático del Hormigón Estruc-
tural. (www.ieca.es) by Corres Peiretti, Martínez Martínez JL,
rethink design practice with the advent of widely avail- Pérez Caldentey A and López Agüí JC. Accessed 16 February
able computational power in design offices and the press- 2008
ing needs to achieve economy and reduce anthropogenic Kanagasundaram S, Karihaloo BL (1990) Minimum cost design of
emissions of greenhouse gases. Substantial reductions in reinforced concrete structures. Struct Optim 2:173–184
Lee HJ, Aschheim M, Hernández-Montes E, Gil-Martin LM (2009)
the amount of steel reinforcement required in design can Optimum RC column reinforcement considering multiple load
be achieved, by coupling a novel solution approach with combinations. Struct Multidisc Optim 39(2):153–170
widely accepted assumptions for ultimate strength analy- Nawy EG (2003) Reinforced concrete. A fundamental approach,
sis. Examples illustrate potential reductions of up to 51% 5th edn. Prentice-Hall, New Jersey
Weber K, Ernst M (1989) Entwicklung von Interaktionsdiagrammen
in the longitudinal reinforcement required, relative to con- für asymmetrisch bewehrte Stahlbeton-Kreisquerschnitte. Beton -
ventional, uniformly distributed reinforcement solutions. und Stahlbetonbau V84(7):176–180

You might also like