Buckling Resistance Criteria of Prismatic Beams Under Biaxial Moment Gradient
 380 Downloads
Abstract
Laterally and torsionally unrestrained steel Isection beams are susceptible to torsional deformations between supports; therefore, according to Part 11 of Eurocode 3, they need to be designed to resist lateraltorsional buckling. Eurocode’s steelwork design criteria require safety checking based on two stability interaction formulae utilizing the socalled equivalent uniform moment factors and the crosssection resistance formula that, in the case of moment gradient, refer to the beam end section. Uncoupling the beam stability resistance criterion and the crosssection resistance criterion may result in a nonuniform safety assessment of Isection beams. Finite element simulations of the beam resistance for different moment gradient ratios are performed. Verification of the buckling resistance is conducted by varying the following parameters: the slenderness ratio, the location of maximum end moments about both axes and the section depthtowidth ratio (i.e., considering rolled I and Hsections). The variation in the accuracy of the current Eurocode resistance evaluation method is identified, and an approach for a better equalization of the safety predictions is suggested by considering different values of the most important factors influencing the stability performance of steel Isection beams.
Keywords
Steel Isection Biaxial bending Special monoaxial bending Imperfect unrestrained beam FEM nonlinear stability analysis Resistance verification Eurocodes1 Introduction
Part 11 of EN 1993 (2005) gives direct recommendations for the lateraltorsional buckling (LTB) of beams laterally and torsionally unrestrained between supports and under monoaxial bending about the section major axis y–y (M_{y}). This code does not, however, refer explicitly to the stability evaluation of beams subjected to biaxial bending (M_{y} and M_{z}, where M_{z} is bending about the section minor axis z–z). This case (M_{y} and M_{z}) is to be treated as a special case of the code’s resistance interaction formulae developed for compression and biaxial bending (N, M_{y} and M_{z}) when the axial force is zero.
Investigations on LTB of monoaxial major axis bending of Isections have been well researched with regard to many factors influencing the resistance of real beams, a survey of which may be found in Trahair (1993), Trahair et al. (2008), Weiss and Giżejowski (1991) and Rykaluk (2012). An evaluation of the design rules of the biaxial bending of beamcolumns is presented in Bradford (1995). The background information for the recommendations of the current Eurocode 3 (EN 199311 2005) with regard to combined compression and bending is presented in Boissonnade et al. (2004 and Greiner and Lindner (2006) and in the ECCS design manual (da Silva et al. 2010b). Further improvements for the revision of LT buckling assessment are the focus of Taras and Greiner (2010), and Greiner and Taras (2010). Recently, new methods for the design procedure of beamcolumns are presented in Boissonnade et al. (2017) and RodriguezGutierrez and AristizabalOchoa (2017) and can be applied to the case of biaxial bending. However, investigations of the stability behaviour in the case of biaxial bending of Isections alone are less reported in the literature; see analytical investigations reported in Baptista (2012a) on the evaluation of elastic and plastic resistances under axial force and biaxial bending of an Isection. There are more papers dealing with other section types. Research on the resistance of double symmetric rectangular hollow cross sections is presented in Baptista (2012b). In the case of angle sections, very similar surveys were provided in Vayas et al. (2009) and Aydın and Doğan (2007). This problem is a special case in which a normalized axial force is equal to zero. The examples presented in the literature are of a very clear form. An approach presented in the abovementioned papers is based on beam theory and some basic hypotheses; one hypothesis with the most farreaching consequences states that the plastic hinge will occur in the “most stressed” cross section. The “most stressed” cross section is the one in which the combined normal stresses resulting from bending moments and compressive forces integrated over the crosssectional area (as an absolute value) reach the maximum value. Clearly, the development of plastic hinges results from the interaction of all forces, the crosssection shape and some local phenomena, resulting for example, from boundary conditions and loading realization. Such influences are neglected in beam theory; note the need for FEM application in such cases. Therefore, the interaction of biaxial bending on the resistance curve in the case of thinwalled rectangular crosssection elements is analysed not only analytically but also numerically in papers (Kim and Wierzbicki 2000; Belingardi and Peroni 2005; Osterrieder and Kretzschmar 2006). In all stability problems, the finite element models were created from shell elements, the material was assumed to be elastic–plastic, and the obtained results were confronted with thinwalled beam theory, proving that some interesting conclusions that are contrary to the beam formulations can be drawn from the FEM analysis. Such conclusions should be verified by experimental investigations, especially in the case of aluminium extrusions, like those presented in Belingardi and Peroni (2007). For investigations related to rectangular thinwalled crosssection elements, a special stand for biaxial loading was designed to obtain a fundamental understanding of the influence of test loading conditions on element resistance. The obtained results demonstrated the element’s overall behaviour and the local phenomena accompanying the plastic hinge formation. Another experimental validation of numerical modelling was presented in the case of a steel Icross section element bent in one plane in Yang et al. (2017). The obtained and verified results demonstrate conservative predictions of three standards: Eurocode 3, American standards and Chinese standards.
The purpose of this paper is fivefold: (1) present the background to Eurocode’s LTB formulation (da Silva et al. 2010a; Giżejowski et al. 2016b), (2) develop a FE modelling technique for the verification of the resistance criteria of imperfect beams subjected to a moment gradient for biaxial bending (monoaxial bending as a special case), (3) conduct advanced finite element simulations using GMNIA with a shell modelling technique, (4) carry out the verification exercise of Eurocode’s resistance interaction criteria, and finally (5) suggest the improvements of codification to equalize the safety assessment for different section geometry, slenderness ratio and moment gradient cases.
2 Background of Eurocode’s Formulation for Resistance Interaction Criteria for Biaxial Bending

for the maximum applied moment over the left support:
$$M_{y,Ed} \left( x \right) = M_{ys,Ed} + \left( {1  \xi } \right)M_{ya,Ed} ,$$(3a)$$M_{z,Ed} \left( x \right) = M_{zs,Ed} + \left( {1  \xi } \right)M_{za,Ed} ,$$(3b) 
for the maximum applied moment over the right support:
$$M_{y,Ed} \left( x \right) = \left( {1  \xi } \right)M_{ya,Ed}  M_{ys,Ed} ,$$(4a)where \(\xi = 2x/L\) is the dimensionless beam axis coordinate and L—beam length.$$M_{z,Ed} \left( x \right) = \left( {1  \xi } \right)M_{za,Ed}  M_{zs,Ed} ,$$(4b)
Possible combinations of the moment gradient distributions
Loading case  Symbol  Discrete values of the ψfactors for the considered loading cases  

ψ_{y}= M_{y,d,min}/M_{y,d,max}  ψ_{z}= M_{z,d,min}/M_{z,d,max}  
1.  SMY–SMZ  1.0  1.0 
2.  TMY–SMZ  0.0  1.0 
3.  AMY–SMZ  − 1.0  1.0 
4.  SMY–TMZ  1.0  0.0 
5.  TMY*–TMZ*  0.0^{a}  0.0^{a} 
6.  TMY*–TMZ**  0.0^{a}  0.0^{b} 
7.  AMY–TMZ  − 1.0  0.0 
8.  SMY–AMZ  1.0  − 1.0 
9.  TMY–AMZ  0.0  − 1.0 
10.  AMY–AMZ  − 1.0  − 1.0 
In the following section, the authors of this paper present the finite element model for the verification of Eurocode’s approach to the buckling resistance assessment of Isection beams. The finite element method is used, and at first, its accuracy is evaluated for the LTB resistance due to monoaxial bending about the y–y axis. Refinements of the geometry discretization meshing with shell finite elements are discussed. Numerical simulations are carried out using GMNIA and a nonlinear solver option available in ABAQUS/Standard (ABAQUS Theory manual 2011; ABAQUS/Standard User’s manual 2011). For biaxial bending, the interaction resistance curves from the numerical simulations, expressed in the dimensionless coordinates m_{by} = M_{y,Ed,max}/M_{by,Rk} and m_{cz} = M_{z,Ed,max}/M_{cz,Rk} [where M_{by,Rk} and M_{cz,Rk} are the characteristic lateraltorsional buckling resistance and characteristic crosssection bending resistances according to EN 199311 (2005), respectively], are expected to be nonlinear and of varying degrees, dependent upon the gradient case. The obtained numerical results are used for the verification of Eurocode’s formulation for biaxial bending and for monoaxial bending. In the case of monoaxial LTB resistance, the discrete points from the finite element simulations are compared with the corresponding Eurocode buckling curves represented in the coordinate system defined by m_{by} and \(\overline{\lambda }_{LT,ref}\) (see explanation in Sect. 3.1). The case of outofplane buckling of unrestrained beams under biaxial bending may be regarded as similar to that used in Giżejowski et al. (2016c, 2017b) and for inplane bending and compression. The similarity yields from the same number of elementary resistance ratio components to be considered (two components) and the same nature of the components (one refers to the member buckling phenomenon, and the other refers one to the yielding criterion of the most stressed section).
3 Assumptions and Numerical Model Descriptions
3.1 Material Model and ISection Beams Considered
Typical grades of structural steel up to S 420 fulfil the minimum ductility requirements for the plastic design defined in Part 11 of Eurocode 3 (EN 199311 2005). The material constitutive model of all steel grades may then be approximated using an elastic–plastic model. The plastic strains are associated with the classical Huber–Mises yield condition, which is also used for plasticity detection. After yielding, the material undergoes plastically isotropic hardening, which means that in principal stress space, the yield condition is expanding to the same extent in every direction.
 1.
The elastic range, in which the material behaviour is described by Hooke’s relationship for isotropic materials, with Young’s modulus E equal to 210 GPa and Poisson’s ratio equal to 0.3.
 2.
The inelastic strain hardening range for strains in the interval \(\varepsilon \in \left( {\varepsilon_{y} ,\varepsilon_{u} } \right)\), where the isotropic strain hardening modulus is equal to \(\tilde{E} = {{\left( {f_{u}  f_{y} } \right)} \mathord{\left/ {\vphantom {{\left( {f_{u}  f_{y} } \right)} {\left( {\varepsilon_{u}  \varepsilon_{y} } \right)}}} \right. \kern0pt} {\left( {\varepsilon_{u}  \varepsilon_{y} } \right)}}\) and Poisson’s ratio is equal to 0.5 as a consequence of the plastic incompressibility assumption.
 3.
The ideally plastic behaviour range for strains \(\varepsilon > \varepsilon_{u} = 0.15\), in which the yield stress f_{u} = 1.1·f_{y}. In that range, the hardening modulus \(\tilde{E} = 0\), but the incompressibility assumption is still valid.
In the applied modelling technique, nonlinearity is taken into account in two ways. The deformation description is completed with large deformation theory (continuum solid mechanics), where the strain tensors are logarithms from the left Cauchy–Green stretch tensor. The multiplicative decomposition of the deformation gradient tensor onto elastic and plastic part is assumed. In the case of a logarithmic strain tensor, the multiplicative decomposition is equivalent to additive decomposition, which is the key feature of the applied finite element software ABAQUS/Standard. The decomposition of strains into elastic and plastic parts specifies the second nonlinearity consideration. The elastoplasticity constitutive model with isotropic strain hardening produces the nonlinear relationship between the stress and strain tensor. The applied model with the Mises yield condition, associated plastic flow law and local loading/unloading conditions allows for the stress redistribution phenomena, which are very important, mainly when some local buckling occurs with stress concentrations.
Buckling curve designation and corresponding imperfection factors
Isection  Buckling curve for LTB (general case)  Imperfection factor α_{LT} 

HEB 300  a  0.21 
IPE 500  b  0.34 
Comparison of slenderness ratio values for beams made of HEB 300
Description of the moment distribution case and corresponding LTB slenderness  Flexural slenderness ratio \(\overline{\lambda }_{z}\)  

0.5  1.0  1.5  2.0  2.5  3.0  4.0  5.0  6.0  
SMY  \(\overline{\lambda }_{LT,ref}\)  0.430  0.729  0.943  1.114  1.259  1.388  1.613  1.809  1.985 
TMY  \(\overline{\lambda }_{LT}\)  0.318  0.539  0.698  0.824  0.932  1.027  1.193  1.338  1.468 
AMY  \(\overline{\lambda }_{LT}\)  0.265  0.449  0.580  0.686  0.775  0.854  0.993  1.113  1.222 
Comparison of slenderness ratio values for beams made of IPE 500
Description of the moment distribution case and corresponding LTB slenderness  Flexural slenderness ratio \(\overline{\lambda }_{z}\)  

0.5  1.0  1.5  2.0  2.5  3.0  4.0  5.0  6.0  
SMY  \(\overline{\lambda }_{LT,ref}\)  0.432  0.815  1.136  1.404  1.632  1.831  2.171  2.460  2.715 
TMY  \(\overline{\lambda }_{LT}\)  0.319  0.603  0.841  1.039  1.208  1.355  1.606  1.820  2.008 
AMY  \(\overline{\lambda }_{LT}\)  0.266  0.502  0.699  0.864  1.005  1.127  1.336  1.514  1.671 
3.2 Finite Elements, Load Transfer and Boundary Conditions
A detailed sensitivity analysis in the case of buckling FEM modelling with 3D elements is presented in Kala and Valeš (2017) and Kala (2015). The results obtained from this analysis for the same support conditions showed that the type and value of the geometrical imperfection and the value of the yield stress are among the most important factors. Numerical simulations of LT buckling using 3D brick elements in comparison to shell and beam elements were also presented in Giżejowski et al. (2016c). The conclusion is that the application of such a modelling technique for the resistance evaluation of rolled Isections leads to timeconsuming and costly simulations, especially if a rational finite element aspect ratio is maintained for the slender elements, while the accuracy of the buckling strength predictions and the equilibrium path estimations remain practically the same when obtained with the use of the shell finite element modelling technique. Hence, shell finite element modelling of the behaviour of Isection beams is considered in this paper, using linear 4node thin shell elements S4R (hereafter called SM).
Special care has to be paid in modelling the boundary and load transfer conditions for the SM simulations. To compare the results from the BM and SM simulations, the boundary conditions and load application conditions should be the same in both modelling techniques. In the case of the FE BM (Fig. 3a), warping is allowed through the application of the B32 OS type of finite elements (open section), whereas in the case of the FE SM (Fig. 3b), the warping is possible due to the use of different modelling tools available in the ABAQUS/Standard software. Figure 3b shows the details of the SM modelling at the LHS support. The developed model of the three rigid subcontours for both flanges and the web allows for a convenient application of the action moments. The moment M_{y,Ed} is to be applied at the centroid of the web subcontour, while the moment M_{z,Ed} has to be divided into two components of M_{z,Ed,f} = M_{z,Ed}/2 to be applied at the centroids of the flange subcontours. Flexural rotations and warping are allowed for both supports.
3.3 Options Used for Modelling Imperfections

Approach 1 with individual modelling of both types of imperfections that have physical meaning.

Approach 2 with an equivalent bow imperfection to reproduce the stability behaviour modelled with inclusion of both types of imperfections.
It is noticeable that the amplitude of the equivalent geometric imperfection profile of the HEB 300 section is much greater than that of the IPE 500 section. This trend is opposite to that of the imperfection factor values α_{LT} for the buckling curves related to IPE 500 (greater for IPE 500 than for HEB 300, as given in Table 2). Furthermore, one can note that the outofplane equivalent imperfection amplitudes are much greater than the acceptable tolerance limit utilized in Approach 1, the values of which are presented in Fig. 5 as dotted lines.
3.4 Solution Options Used in the Numerical Analysis
The elastic–plastic tangent stiffness matrix and the residual load vector of the behaviour of the structural model are integrated numerically using the Riks–Crisfield implicit algorithm (Riks 1979; Crisfield 1981). The application of the Riks–Crisfield algorithm provides stability under nonzero stress boundary conditions, as in the analysed problem, and allows a critical point crossing on the global equilibrium path.
The results obtained from the FEM simulations calculated according to Eq. (5) are compared in Fig. 6 for ψ_{y}= 1, 0 and − 1 with the results from Eurocode’s General case (buckling curves χ_{LT} marked by Gc—red solid line) and Eurocode’s Special case (curves χ_{LT,mod} marked by Sc—blue solid line) as presented in Gizejowski and Stachura (2017). Furthermore, the results from the proposal presented in Taras and Greiner (2010) and Greiner and Taras (2010) for updating the Eurocode’s lateralbuckling curves (denoted by the dashed line and labelled “Greiner and Taras”) and from the authors’ proposal developed hereafter (denoted by the dotted line and labelled “Proposal”) are also included.
The results presented in Fig. 6 show that for class 1 and 2 sections, it is rational to use the equivalent imperfection profile calculated for e_{0,LT} based on the section plastic properties. The buckling resistance evaluated on the basis of the plastic section resistance is a lower value than that for the elastic section resistance. Therefore, this is hereafter taken into consideration.
The lateraltorsional buckling resistances χ_{LT,mod} according to the Sc approach of clause 6.3.2.3 of EN 199311 (2005), the author’s proposal and the Greiner and Taras (2010) approach allow for the evaluation of higher values in comparison to those of the Gc approach [according to clause 6.3.2.2 of EN 199311 (2005)]. They are closer to the results from the finite element simulations. For the considered HEB section in the case of SMY, the case of TMY for reference slenderness \(\overline{\lambda }_{LT,ref}\) greater than 1.3 and the case of AMY for \(\overline{\lambda }_{LT,ref}\) > 1.5, the results from the authors’ proposal are closer to the numerical results than the results of the Greiner and Taras (2010) approach. For the considered IPE section and in the cases of SMY and AMY, the results from the authors’ approach and Greiner and Taras (2010) approach are similar. Only in the case of TMY for the considered IPE section, the buckling curve from Greiner and Taras (2010) for \(\overline{\lambda }_{LT,ref}\) > 0.5 is slightly closer to numerical results than those of the authors’ proposal.
It is therefore rational to recommend the authors’ approach or Greiner and Taras (2010) proposal as an alternative approach to that of the current Eurocode 3 for rolled and equivalent welded I and Hsections (Sc) for the design of rolled Isection beams. Note that both proposals are more general than that of Eurocode 3 since they account for more parameters; therefore, these approaches are not restricted to the section dimensions of the rolled Isections. However, one must remember that the beam section must not be susceptible to either local or distortional buckling.
3.5 Imperfection Modelling Techniques and Loading History Adopted

Path A First application of support moments with the considered moment gradient ratio ψ_{z} to produce the required M_{z,Ed}(x) moment diagram; then, the application of the support moments with moment gradient ratio ψ_{y} to produce the required M_{y,Ed}(x) moment diagram. The Riks incrementation method of the ψ_{y} parameter is then applied until the buckling resistance under biaxial bending is attained.

Path B The load sequence application opposite to that of Path A by substituting subscripts “y” for “z”, and vice versa. In this case, care has to take with regard to modelling the flange moment responses of the Isections since this response is a combination of minor axis bending and warping torsion.

Path C The simultaneous application of support moments with the moment gradient ratios ψ_{y} and ψ_{z} for the loading cases identified in Table 1, with a predefined proportionality ratio ψ = ψ_{y}/ψ_{z} (or ψ = ψ_{z}/ψ_{y}) subjected to incrementation. The Riks incrementation method of the ψ parameter is then applied until the buckling resistance under biaxial bending is attained.
Figure 8a shows the effect of the path case on the biaxial bending resistance interaction curves. The results show that out of the three possibilities of defining the ultimate beam state under the combination of inplane and outofplane moments, there is a negligible effect of the loading path on the resistance interaction curves from the engineering application point of view. The differences are not greater than 1%, and it may be summarized that Path A gives the lower bound assessment of the buckling resistance, while Path B provides that of the upper bound. This result is easily explainable since the application of minor axis moments in the first stage of analysis may be treated as a superimposition of two initial states, one produced by imperfections and the second by minor axis bending.
Figure 8b presents the verification of the lower bound resistance estimation in the case of biaxial bending and the equivalent geometric imperfections approach as well as the resistance based on the explicit introduction of residual stresses and geometric imperfections at the manufacturing tolerance level in finite element simulations. By observing the results, one can come to the conclusion that both approaches of imperfection modelling provide results of a similar accuracy as that observed for monoaxial bending for lesser values of m_{cz} (cf. Fig. 7), while the differences continuously decrease with increase in m_{cz} values.
In conclusion, it is justifiable to use the loading history according to Path A and the imperfection modelling technique based on Approach 2 of the equivalent geometric imperfection profile. The profile was considered the lowest SMY buckling mode with a “plastic” amplitude according to Fig. 5.
3.6 Mesh Refinements
For the SM finite element simulations, refinements of the discretization scheme are checked to select the scheme that ensures that the accuracy is acceptable from an engineering point of view. The exercise is similar to that carried out for compression and monoaxial bending without lateraltorsional buckling, and the results are presented in Giżejowski et al. (2016d).
Mesh refinements from coarse to fine for the HEB 300 section
The obtained results confirm the conclusion made in Giżejowski et al. (2016d), that a sufficient accuracy is obtained with the use of the M3 mesh. This mesh is therefore used in all the FEM simulations presented hereafter.
4 Verification of Eurocode’s Resistance Strength Curves for Beams Under Biaxial Bending
In this section, in the Eurocode prediction of biaxial bending resistance, the values of χ_{LT,mod} according to the Sc approach for rolled Isections [according to clause 6.3.2.3 of EN 199311 (2005)] are used in the interaction equations of clause 6.3.3, in which N_{Ed} = 0 is assumed. Furthermore, Eurocode’s Method 1 is considered for the comparison of beam resistances under biaxial bending with those from the FEM simulations, and comments regarding the use of only Method 2 are included hereafter to provide more a brief discussion.
The finite element models were previously validated; for many cases, these validations were presented in Giżejowski et al. (2016c, d, 2017a, b) and Gizejowski and Stachura (2017). This paper is a continuation of longterm research and is part of the current research of Papp (2016). It must be stated that validation of the finite element results is performed only through a comparison with the results obtained by using the Eurocode 3 analytical formulation (EN 199311 2005) that was validated prior to its codification. These results, in most cases, are far more conservative than the FEM results, but all significant discrepancies between the FEM modelling and Eurocode 3 predictions are carefully analysed before formulating any conclusions. The conclusions presented in this paper are not an extrapolation of the results but are stated on the basis of solid and checked analysis. To present a wider view of the analysed topic in terms of fullscale experimental tests, the following classical papers may be addressed (Anslijn 1983; Kloppel and Winkelmann 1962; Chubkin 1959; Birnstiel et al. 1967).
4.1 Finite Element Simulations for Buckling Resistance for the Bending Moment Diagram of Case SMY
It is shown in Fig. 11a that the finite element interaction curves (represented by solid lines) for the medium slenderness ratio \(\overline{\lambda }_{z}\) of the HEB 300 section exhibit two distinctive parts, namely, a concave branch (for the low values of m_{cz}) and a convex branch (for higher values of m_{cz}). Eurocode’s recommendations of not considering the large twist rotations lead to interaction curves (represented by dashed lines), constituting a safe estimation of the beam biaxial bending ultimate state. The interaction curves for the IPE section that is less sensitive to the effect of prebuckling displacements on the critical state are presented in Fig. 11b. Since the buckling state under monoaxial bending about the y–y axis is reached in this section before the large inplane displacements could have occurred, the reduction of the buckling resistance in biaxial bending is faster than observed for the wide flange beams.
When the minor axis bending moment diagram becomes more asymmetric, the concave part of the FEM resistance curves gradually vanishes, and the curve becomes fully convex for the whole range of the slenderness ratio \(\overline{\lambda }_{z}\), regardless the webtoflange dimension proportion. The predictions made using Method 1 show an opposite trend. For uniform minor axis bending, the Eurocode 3 strength curves consist of two parts, the lower m_{cz} values of these curves become approximately linear as purposefully adopted for Method 2 in EN 199311 (2005). For a greater minor axis moment gradient ratio, the predictions made with the use of Eurocode 3 Method 1 create a more concave strength curve instead of a more convex curve, as predicted by the FEM simulations. It might be shown that the Eurocode 3 Method 2 linear branch (instead of the concave branch of Eurocode 3 Method 1) gives, in this case, less conservative predictions than those from the FEM simulations.
4.2 Finite Element Simulations for Buckling Resistance for the Bending Moment Diagram of Case TMY
Comparing the results, it has to be noted that Eurocode’s interaction curves in Fig. 12b and c are the same for the same section in the TMY*–TMZ* and TMY*–TMZ** cases, respectively. The finite element results are substantially different in both cases. One can conclude that the concept of equivalent uniform moment used in the code does not allow for a proper resistance assessment in the case TMY*–TMZ** since it underestimates the resistance by an unacceptable level, e.g., for the HEB section and the slenderness ratio \(\overline{\lambda }_{z}\) = 2.0, interaction between moments applied about both principal axes occurs for values of m_{cy} ≥ 0.8. Below this limit, the ultimate limit state is reached when either the minor axis moment attains the corresponding section resistance or the major axis moment attains its corresponding section resistance. Contrarily, Eurocode’s interaction curve requires the interaction in the range of approximately m_{cy} ≥ 0.35 for the same slenderness. Comparing the results presented in Fig. 12a, b, d, one can observe the similarity of the shape of the FEM interaction curves and its lesser sensitivity to the change towards a more convex shape when the minor axis moment gradient ratio increases. Accordingly, Eurocode’s Method 1 interaction curves and the FEM resistance curves are more similar in the TMY* cases than in the SMY cases; this is especially observed for TMY*–AMZ in comparison to SMY–AMZ.
4.3 Finite Element Simulations for Buckling Resistance for the Bending Moment Diagram of Case AMY
5 General Conclusions and Final Remarks
The buckling phenomenon of perfect Isection beams under compression and under twodirectional bending about the principal axes y–y and z–z, laterally and torsionally unrestrained between supports, is studied in this paper. Furthermore, sections of bending class 1 or 2 are taken into consideration, implying that local instability does not affect the stability behaviour of the considered beams. The overall buckling problem is therefore converted to a spatial secondorder bending and warping torsion problem under a general moment gradient loading, and the limit points on the inelastic equilibrium path of the imperfect elements are evaluated. The conventional simple boundary conditions with loading cases producing different moment gradient conditions along both principal axes are considered. The principal objective of the paper is to perform numerical simulations for all the considered loading cases and verify the Eurocode 3 interaction equation approach used for the buckling resistance evaluation in the case of biaxial bending when the stability load effect M_{y,Ed} produced by the applied moments M_{y,d} is associated with the nonstability load effect M_{z,Ed} produced by the applied moments M_{z,d}. Since the considered calculations refer to imperfect members, they allow for a better understanding of the behaviour of real class 1 or 2 section beams and a better prediction of the accuracy of the resistance criteria obtained from the simplification of Eurocode’s recommendations related to biaxial bending and compression.
Current stability criteria rely on uncoupling the beam stability resistance and the crosssection resistance, resulting in a nonuniform safety assessment of Isection beams, especially when subjected to a negative moment gradient ratio, i.e., when the moments in both planes are acting over the opposite beam supports. Since the safety assessment of beam buckling resistance evaluation in biaxial bending has not yet been widely investigated, this paper focuses on advanced analysis and GMNIAbased numerical simulations. The finite element simulations for different moment gradient situations are presented for the verification of Eurocode’s analytical approach to the buckling resistance of beams. Cases of monoaxial bending about the section major axis and biaxial bending, including the moment gradient effect for bending about both section principal axes, are evaluated.
The developed finite element model of the equivalent geometric imperfections, applied to biaxial bending, allows for the identification of the moment gradient parameters ψ_{y} and ψ_{z} for which the current design recommendations lead to quite conservative predictions of the buckling resistance. Specifically, the finite element results are substantially different in cases TMY*–TMZ* and TMY*–TMZ**, while Eurocode’s interaction curves for both cases are the same. One can conclude that such differences might be mainly attributed to the concept of equivalent uniform moment, which replaces the real moment gradient cases. An improved accuracy between the results of the finite element simulations is expected when a new formulation would be proposed that abandons the concept of equivalent uniform moment factors. This aspect will be a subject of the authors’ forthcoming research activities dealing with the resistance evaluation of imperfect beams for outofplane behaviour.
In Sect. 4, the comparison of the analytical results obtained from Method 1 of Eurocode 3 and the finite element method simulation results is presented. The analytical solutions were obtained for the considered rolled crosssections, and the moment distributions are denoted as SMY–SMZ, SMY–TMZ and SMY–AMZ and are plotted against the FEM solutions with shell modelling. The detailed conclusions on these comparisons are provided underneath the relevant graphs; however, it is worth noting that in all cases, Eurocode’s expectations are conservative. As it has already been mentioned, in some cases, the differences are of a magnitude of several percent. However, the differences are sometimes greater; for example, in the case of TMY–TMZ for both HEB 300 and IPE 500, the differences are substantial. Comparing the graphs in Fig. 12, it is possible to conclude that, for some initial moment distributions, the nonlinearity of the problem is clearly manifested. This nonlinearity cannot be taken into account with the beam theory (Bernoulli’s theory or the thinwalled theory) but requires full spatial modelling and proper constitutive modelling. The shell finite element models formulated by considering large deformation theory with elastoplasticity constitutive relationships like those presented here may be treated as the first reliable step in the movement towards proper stability problem modelling. The limitations resulting from the shell geometry description are not as conservative as those from beam theory, and the number of elements in the model are reasonable compared with, for example, 3D modelling (Giżejowski et al. 2016c).
References
 ABAQUS Theory manual, Version 6.11, Dassault Systèmes (2011).Google Scholar
 ABAQUS/Standard User’s manual, Version 6.11, Dassault Systèmes (2011).Google Scholar
 Anslijn, R. (1983). Tests on steel Ibeamcolumns subjected to thrust and biaxial bending. CRIF report MT157, Center De Recherches Scientifiques Et Techiques De L’industrie De, Fabrications Metalliques Brussels, Belgium.Google Scholar
 Aydın, R., & Doğan, M. (2007). Elastic, full plastic and lateral torsional buckling analysis of steel singleangle section beams subjected to biaxial bending. Journal of Constructional Steel Research, 63(1), 13–23.CrossRefGoogle Scholar
 Baptista, A. M. (2012a). Resistance of steel Isections under axial force and biaxial bending. Journal of Constructional Steel Research, 72, 1–11.CrossRefGoogle Scholar
 Baptista, A. M. (2012b). Analytical evaluation of the elastic and plastic resistances of double symmetric rectangular hollow sections under axial force and biaxial bending. International Journal of NonLinear Mechanics, 47(9), 1033–1044.CrossRefGoogle Scholar
 Belingardi, G., & Peroni, L. (2005). Numerical investigation on plastic collapse of thin walled beams subjected to biaxial bending. In Proceedings of VIII international conference on computational plasticity, COMPLAS VIII, Barcelona (pp. 1–4).Google Scholar
 Belingardi, G., & Peroni, L. (2007). Experimental investigation of plastic collapse of aluminium extrusions in biaxial bending. International Journal of Mechanical Sciences, 49(5), 554–566.CrossRefGoogle Scholar
 Birnstiel, C., Leu, K. C., Tesoro, J. A., & Tomasetti, R. L. (1967). Experiments on Hcolumns under biaxial bending. New York: New York University.Google Scholar
 Boissonnade, N., Hayeck, M., Saloumi, E., & Nseir, J. (2017). An overall interaction concept for an alternative approach to steel members design. Journal of Constructional Steel Research, 135, 199–212.CrossRefGoogle Scholar
 Boissonnade, A., Jaspart, J.P., Muzeau, J.P., & Villette, M. (2004). New interaction formulae for beamcolumns in Eurocode 3: The French–Belgian approach. Journal of Constructional Steel Research, 60, 421–431.CrossRefGoogle Scholar
 Boissonnade, N., & Somja, H. (2012). Influence of imperfections in FEM modeling of lateral torsional buckling. In Proceedings of annual stability conference: structural stability research council, Grapevine (pp. 1–15).Google Scholar
 Bonet, J., & Wood, R. D. (2008). Nonlinear continuum mechanics for finite element analysis (2nd ed.). Cambridge: Cambridge University Press.CrossRefzbMATHGoogle Scholar
 Bradford, M. A. (1995). Evaluation of design rules of the biaxial bending of beamcolumns. Civil Engineering Transactions, Institution of Engineers, Australia, 37(3), 241–245.Google Scholar
 Chubkin, G. M. (1959 ). Experimental research on the stability of thin plate steel members with biaxial eccentricity. Paper No. 6 in the book Analysis of spatial structures, Vol. 5, Moscow (G.I.L.S.).Google Scholar
 Crisfield, M. A. (1981). A fast incremental/iterative solution procedure that handles “snapthrough”. Computers & Structures, 13, 55–62.CrossRefzbMATHGoogle Scholar
 da Silva, L. S., Marques, L., & Rebelo, C. (2010a). Numerical validation of the general method in EC311 for prismatic members. Journal of Constructional Steel Research, 66, 575–590.CrossRefGoogle Scholar
 da Silva, L. S., Simoes, R., & Gervasio, H. (2010b). Design of steel structures, Eurocode 3: Design of steel structures, Part 11: General rules and rules for buildings, ECCS Eurocode design manual. Berlin: Ernst & Sohn.Google Scholar
 de Souza Neto, E. A., Peric, D., & Owen, D. R. J. (2008). Computational methods for plasticity: Theory and applications. New York: Wiley.CrossRefGoogle Scholar
 EN 10034 EN 10034. (1996). Structural steel I and H sectionstolerances on shape and dimensions. Brussels: CEN.Google Scholar
 EN 199311. (2005). Eurocode 3: Design of steel structures. Part 11. General rules and rules for buildings. Brussels: CEN.Google Scholar
 Gizejowski, M., & Stachura, Z. (2017). On LTB resistance assessment of prismatic Isection beams according to Eurocode 3. Journal of Civil Engineering, Environment and Architecture, 34(643/I/17), 427–436.Google Scholar
 Giżejowski, M. A., Stachura, Z., Szczerba, R. B., & Gajewski, M. D. (2016a). Numerical study of buckling resistance of steel Isection members under two directional bending. In Proceedings of 6th international conference on structural engineering, mechanics and computation (pp. 267–268) (ebook on CD: 758–763).Google Scholar
 Giżejowski, M. A., Stachura, Z., & Uziak, J. (2016b). Elastic flexuraltorsional buckling of beams and beamcolumns as a basis for stability design of members with discrete rigid restraints. In Proceedings of 6th international conference on structural engineering, mechanics and computation (pp. 261–262) (ebook on CD: 738–744).Google Scholar
 Giżejowski, M. A., Szczerba, R. B., & Gajewski, M. D. (2017a). LTB resistance of rolled Isection beams. FEM verification of Eurocode’s buckling curve formulation. In Proceedings of Eurosteel 2017 in: Online collection for conference papers in civil engineering (Vol. 1, No. 2–3, pp. 1325–1334). Berlin: Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG.Google Scholar
 Giżejowski, M. A., Szczerba, R., Gajewski, M. D., & Stachura, Z. (2016c). On the resistance evaluation of lateraltorsional buckling of bisymmetrical Isection beams using finite element simulations. Procedia Engineering, Open Access Journal, 153, 180–188.Google Scholar
 Giżejowski, M. A., Szczerba, R., Gajewski, M. D., & Stachura, Z. (2016d). Beamcolumn inplane resistance based on the concept of equivalent geometric imperfections. Archives of Civil Engineering, 62/II(4), 35–71.CrossRefGoogle Scholar
 Giżejowski, M. A., Szczerba, R., Gajewski, M. D., & Stachura, Z. (2017b). Buckling resistance assessment of steel Isection beamcolumns not susceptible to LTbuckling. Archives of Civil and Mechanical Engineering, 17(2), 205–221.CrossRefGoogle Scholar
 Greiner, R., & Lindner, J. (2006). Interaction formulae for members subjected to bending and axial compression in Eurocode 3: The method 2 approach. Journal of Constructional Steel Research, 62, 757–770.CrossRefGoogle Scholar
 Greiner, R., & Taras, A. (2010). New design curves for LT and TF buckling with consistent derivation and codeconform formulation. Steel Construction, 3(3), 176–186.CrossRefGoogle Scholar
 Kala, Z. (2015). Sensitivity and reliability analyses of lateraltorsional buckling resistance of steel beams. Archives of Civil and Mechanical Engineering, 15, 1098–1107.CrossRefGoogle Scholar
 Kala, Z., & Valeš, J. (2017). Global sensitivity analysis of lateraltorsional buckling resistance based on finite element simulations. Engineering Structures, 134, 37–47.CrossRefGoogle Scholar
 Khan, A. S., & Huang, S. (1995). Continuum theory of plasticity. New York: Wiley.zbMATHGoogle Scholar
 Kim, H.S., & Wierzbicki, T. (2000). Numerical and analytical study on deep biaxial bending collapse of thinwalled beams. International Journal of Mechanical Sciences, 42(10), 1947–1970.CrossRefzbMATHGoogle Scholar
 Kloppel, K., & Winkelmann, E. (1962). Experimental and theoretical investigations on the ultimate strength of biaxially compressed steel columns. Der Stahlbau, 31(2), 33.Google Scholar
 Osterrieder, P., & Kretzschmar, J. (2006). Firsthinge analysis for lateral buckling design of open thinwalled steel members. Journal of Constructional Steel Research, 62(1–2), 35–43.CrossRefGoogle Scholar
 Papp, F. (2016). Buckling assessment of steel members through overall imperfection method. Engineering Structures, 106, 124–136.CrossRefGoogle Scholar
 Riks, E. (1979). An incremental approach to the solution of snapping and buckling problems. International Journal of Solids and Structures, 15, 529–551.MathSciNetCrossRefzbMATHGoogle Scholar
 RodriguezGutierrez, J. A., & AristizabalOchoa, J. D. (2017). Bending momentaxial forcecurvature interactions for metal beamcolumn sections. Structures, 10, 139–146.CrossRefGoogle Scholar
 Rykaluk, K. (2012). Stability problems of metal structures. Wrocław: Dolnośląskie Wydawnictwo Edukacyjne. (in Polish).Google Scholar
 Taras, A., & Greiner, R. (2010). New design curves for lateraltorsional: Proposal based on a consistent derivation. Journal of Constructional Steel Research, 66(5), 648–663.CrossRefGoogle Scholar
 Trahair, N. S. (1993). Flexuraltorsional buckling of structures. London: CRC Press.Google Scholar
 Trahair, N. S., Bradford, M. A., Nethercot, D. A., & Gardner, L. (2008). The behaviour and design of steel structures to EC3 (4th ed.). London: Taylor & Francis.Google Scholar
 Vayas, I., Charalampakis, A., & Koumousis, V. (2009). Inelastic resistance of angle sections subjected to biaxial bending and normal forces. Steel Construction, 2, 138–146.CrossRefGoogle Scholar
 Weiss, S., & Giżejowski, M. A. (1991). Stability of metal structures. Rod structures. Warsaw: Arkady. (in Polish).Google Scholar
 Yang, B., Kang, S.B., Xiong, G., Nie, S., Hu, Y., Wang, S., et al. (2017). Experimental and numerical study on lateraltorsional buckling of singly symmetric Q460GJ steel Ishaped beams. ThinWalled Structures, 113, 205–216.CrossRefGoogle Scholar
Copyright information
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.