Professional Documents
Culture Documents
alex_fce@uacg.bg
Abstract. Unlike older design codes, Eurocode 2 does not provide a procedure for calculation
of the capacity of axially loaded columns. For that purpose, a general algorithm is developed,
using the principles of structural mechanics and mathematics: It starts with constructing the P-
Δ relationship (buckling curve) for the element, accounting for initial imperfections, minimum
eccentricity, material nonlinearity and second order effects. Then, the respective bending
moments are determined from the displacements and the P-M relationship is obtained. Finally,
the axial load capacity in planar domain is determined by intersecting the P-M diagram with
the interaction curve for the RC section. In spacial domain, the P-Δ relationship is defined by
the respective “buckling surface”. Its intersection with the interaction surface represents a
spacial curve in the (P, Mx, My) coordinate system. The axial capacity is determined as the
minimum of P values along the curve. The solution is performed by numerical methods.
1. Introduction
Usually, design checks of RC columns for bending with axial force are performed by comparing the
external bending moment with the ultimate bending capacity (M Ed ≤ M Rd), providing that the internal
and external axial loads are in equilibrium (N Ed = N Rd). However, in most cases, columns are loaded
only with axial force and there are no external bending moments. Then, minimal eccentricities, initial
imperfections and second order effects are applied besides the axial load. In such cases, it is more
convenient to determine the ultimate axial force capacity N Rd,II and compare it to the imposed external
load N Ed . This approach provides more valuable information for the structural designer.
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
Strain diagram is varied from pure tension to pure compression through stages 1, 2 and 3 as shown on
Figure 3.
Figure 1. Stress-strain relationship for concrete. Figure 2. Stress-strain relationship for steel.
For a given strain diagram, the respective stresses in concrete and reinforcement bars are evaluated.
The resulting compression force in concrete is calculated via numerical integration using adaptive
Gauss-Lobatto quadrature [1]. The equilibrium of the internal and external axial forces is found by
solving the equation:
N Rd (ε )+ N Ed =0 (1)
,where ε is a general strain parameter that describes all possible positions of the strain diagram. The
equation is solved using the modified Anderson-Bjork root finding method [2]. The axial cross section
resistance is determined as follows:
N Rd = N c + ∑ N si , (2)
where:
x
{ [ ( )]
n
ε c ( z)
σ c ( z )= f cd 1− 1− ε for 0 ≤ ε c ( z ) ≤ ε c 2 - concrete stress; (4)
c2
f cd for ε c 2 ≤ ε c ( z) ≤ ε cu 2
2
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
4. Design criteria
As we mentioned above, most authors propose the equation M Ed ≤ M Rd for design check of RC
sections, where M Rd is obtained for the equilibrium of internal and external axial forces. However. this
approach may not be fully correct for unsymmetrical sections and/or reinforcement. For example, the
asymmetrically reinforced section, shown on Figure 6 will fail for the applied load even though M Ed ≤
M Rd. The correct approach for all cases is to check if the load is inside the interaction curve (Figure 4)
which can be expressed by the equation:
M Rd,min ≤ M Ed ≤ M Rd,max (8)
The same approach is applicable for biaxial bending, as shown on Figure 5. Here, M Rd,min, M Ed and
M rd,max are located on a horizontal line in the plane of the external moment. In this way, we provide
that the load point is inside the interaction surface. Once we have the surface constructed, we can use
it to check many load cases as displayed on Figure 5.
3
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
5. Capacity of columns
Besides the cross section and material properties, the capacity of the element depends on the minimum
eccentricity, initial imperfections and second order effects, that should be also taken into account.
Figure 8. Minimum eccentricity check with M–N Figure 9. Min eccentricity check
interaction diagram. with M x - M y interaction diagram.
4
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
For the spatial formulation, the eccentricity can be represented by inverted elliptic cone as
displayed on Figure 11. Since, it can act in any direction, the entire cone will be the relevant load
whenever the external moment falls inside the cone/pyramid, regardless of its sign and direction. Then
the check is reduced to a mere geometrical problem: We intersect the pyramid and the interaction
surface with a horizontal plane at N = NEd and compare both shapes. If the eccentricity polygon falls
entirely inside the M x –M y interaction diagram in the N = NEd plane, the check is satisfied (Figure 10).
The definition of the axial load capacity N Rd,0 is a bit more complicated. If we intersect the
inverted pyramid with the interaction surface, we will obtain a spatial closed curve. The capacity is the
minimum of the curve in respect to the N coordinate as displayed on Figure 11
5
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
There are three methods in Eurocode 2 to account for second order effects: general method,
nominal stiffness and nominal curvature. For the purpose of our research, we will use the nominal
stiffness method, according to EN 1992-1-1, section 5.8.7.2. However, any of the other methods may
be applied with equal success, without affecting the final conclusions.
The following procedure is used:
Bending moment with initial imperfections - M 0 , Ed = M Ed + N Ed · ei
6
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
a) 2D solution b) 3D solution
Figure 12. Buckling capacity of columns.
6. Proposed algorithm
Based on the above discussion, we propose the following algorithm for calculation of the axial load
capacity of arbitrary shaped RC columns, with account to minimum eccentricity, initial imperfections
and second order effects:
7. Additional considerations
7
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
7.3. Performance
At a first glance, the proposed 3D algorithm may look computationally intensive. However it can be
efficiently implemented on the contemporary hardware. All surfaces can be stored as 3D meshes
(polygons). Once they are constructed, the problem is reduced to pure geometric algorithms. The
proposed method is suitable for parallelization, so we can take advantage of the today’s multi-core
processors. The other possibility is to benefit from the computational power of 3D graphic cards, if
available. In all cases, we will be able to calculate hundreds or even thousands of columns in a second.
9. Worked example
To illustrate how the proposed method works, we developed a simple example for a rectangular
section with different boundary conditions in both directions. We calculated the buckling resistance
using both methods – planar and spatial and compared the results.
Cross Section – R50x30 + 4N28 Buckling Data
Concrete area - Ab =1500 cm2 Column length – L = 420cm
Bar count - nb = 4 Buckling length – Lox = 2*L
Bar diameter - db = 28mm Buckling length – Loy = 1*L
Reinforcement area - As = 24 cm2 Creep factor - φ(∞,t0) = 2.5
Reinforcement ratio - ρ = 1.6% Permanent load ratio - KG = 75%
Concrete grade C25/30
Characteristic compressive cylinder strength - f ck =25 MPa
Partial safety factor for concrete - γ c =1.5 , α cc =0.85
α cc · f ck 0.85 · 25
Design compressive cylinder strength - f cd = = =14.17 MPa
γc 1.5
Mean value of cylinder compressive strength - f cm =f ck +8=25+8=33 MPa
( )
0.3
( )
0.3
f 33
Secant modulus of elasticity - E cm =22· cm =22· =31.48 GPa
10 10
Ultimate compressive strain - ε cu 2=0.0035
Strain at the end of parabolic part of the diagram - ε c 2=0.002
Reinforcement grade B500
Characteristic yield strength - f yk =500 MPa
Partial safety factor for steel - γ s =1.15
f yk 500
Design yield strength - f yd = = =434.78 MPa
γ s 1.15
Modulus of elasticity - E s=2 00 GPa
8
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
Planar solution
Geometric imperfections and accidental eccentricity
Number of vertical members contributing to the total effect - m=1
2 2
Reduction factor for height - α h= = =0.976
√ L ·10 √ 4200 · 10−3
−3
√
Reduction factor for number of members - α m= 0.5 · 1+
1
( m1 )=√ 0.5 ·(1+ 11 )=1
Basic inclination value - θo= =0.005
200
Inclination - θi=θo · α h · α m =0.005 · 0.976 · 1=0.00488 [EN 1992-1-1 (5.1)]
θi · Lo 0.00488 · 8400
Eccentricity - ei= = =20.49 mm [EN 1992-1-1 (5.2)]
2 2
Minimum eccentricity [EN 1992-1-1, § 6.1(4)]
eox =max (h
30 )
; 20 =max (
500
30 )
; 20 =20 mm eoy =max
b
(
30 )
; 20 =max
300
30 (
; 20 =20 mm )
Second order effects based on the nominal stiffness method
N Ed 2271
Relative axial force - n= −3
= =1.07
A c · f cd ·10 150000 · 14.17 ·10−3
9
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
E cm 31.48
E cd = = =26.23 GPa
γ cE 1.2
Moment of inertia of the concrete cross section
b · h3 300 · 5003 h· b3 500 · 3003
I cx = = =3125000000 mm4 I cy = = =1125000000 mm4
12 12 12 12
Radius of gyration
√ √I
i x = cx =
Ac
Slenderness ratio
3125000000
150000
=144.34 mm i y=
√ √
I cy
Ac
=
1125000000
150000
=86.6 mm
( ) ( ) ( ) ( )
2 2 2 2
h h b b
I sx = A s 1 · −d 1 + A s 2 · −d 2 I sy = A s 1 · −d 1 + A s 2 · −d 2
2 2 2 2
( ) ( )
2 2
500 300
I sx =2 · 1200 · −50 =96000000 mm4 I s =2 ·1200 · −50 =24000000mm4
2 2
Nominal stiffness [EN 1992-1-1 (5.21)]
EI x =K c · E cd · I c +K s · E s · I s EI y =K c · E cd · I c + K s · E s · I s
EI x =0.0778· 26.23 ·3125 ·106 +1 · 200 · 96 · 106 EI y =0.0778 · 26.23 · 1125 · 106 +1· 200 · 24 · 106
EI x =25575184994 kN·mm² EI y =7095066598kN·mm²
Buckling load [EN 1992-1-1 (5.17)]
π 2 · EI x 3.142 · 25575184994 2
π · EI y 3.14 · 7095066598
2
N Bx = = =3577.34 kN N By = = =3969.7 kN
Lox 2 84002 Loy 2 42002
10
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
11
DCB-2022 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 1276 (2023) 012011 doi:10.1088/1757-899X/1276/1/012011
buckle only along one of the two orthogonal directions, the spatial method gives the same results as
the planar one.
10. Conclusions
Based on the above discussion, we may draw the following conclusions:
1. Design codes do not provide sufficient recommendations about the buckling capacity of arbitrary
shaped columns. Numerical methods and computer software/simulations are required.
2. Using one planar solution only may not be enough for certain shapes. On the other hand, it may
be too conservative to account for the minimum eccentricities and initial imperfections simultaneously
in both directions, as recommended in sections 5.8.9 of EN 1992-1-1 [3]. For example, for circular
sections, it will be simply equivalent to increase the eccentricities by approx. 1.4 times.
3. Even for simple cross sections, it may be difficult to predict the most unfavorable direction to
apply the minimal eccentricity and initial imperfections, if the boundary conditions of the element are
different in both directions.
4. The proposed algorithm gives a robust and solid mathematical approach to find the buckling
resistance of RC columns with arbitrary shapes. Additionally, it provides a clear and illustrative
graphical representation of the solution.
References
[1] Gander W and Gautschi W 2000 Adaptive Quadrature - Revisited BIT Numerical Mathematics
40 pp 84–101
[2] Anderson N and Björck Å 1973 A new high order method of regula falsi type for computing a
root of an equation BIT 13 pp 253–264
[3] EN 1992-1-1 (2004): Eurocode 2: Design of concrete structures. Part 1-1: General rules and
rules for buildings
12