You are on page 1of 8

Articles Markus Drr Thomas Misiek* Helmut Saal

DOI: 10.1002/stco.201110033

The torsional restraint of sandwich panels to resist the lateral torsional buckling of beams
Experimental investigations and parametric finite element analyses show that the moment-rotation characteristic of sandwich panels providing support against the lateral torsional buckling of beams subjected to gravity loading can be represented as tri-linear. Formulae for calculating the parameters of this characteristic are given for different types of beam combined with sandwich panels for roofs and walls. According to available design codes, the torsional restraint coefficient for rotational stiffness c required for design against lateral torsional buckling of beams is governed by the rotational stiffness cA of the connection, which can be obtained from the characteristic mentioned as the secant stiffness for all common types of sandwich panel and different types of structural arrangement. These values allow the maximum spans of beams to be increased and help to improve the economy of lightweight structures. code 3-1-3 [7], they ignore the very essential beneficial contact effect with gravity loading. Significant influence factors, e.g. the mechanical properties of the foam, especially its Youngs modulus ES, are missing in [1]. Therefore, we see the need for some clarifications before introducing the new regulations of the German design code DIN 18800-2 [3] which are based on the results of the investigations documented in [4] and [5] and are presented here.

1 Introduction
Stabilizing beams against lateral torsional buckling contributes to the economy of loadbearing systems. This is especially true if this stabilization is achieved by building components that are part of the building anyway, e.g. roof or wall sheeting provided for the building envelope. Steel sandwich purlin-sheeting systems are such structures which, in addition, provide optimum insulation. This paper focuses on roof structures, but the approach applies to wall structures as well. In the latter case the gravity load has to be replaced by the load acting normal to the surface of the sandwich element in the direction of the supporting structure (beam), e.g. wind pressure, and the uplift load is the same type of load but in the opposite direction, e.g. wind suction. The way sandwich panels stabilize beams against lateral torsional buckling may be taken into account by considering the torsional restraint. This restraint acts as a rotational
Received 20 April 2011, revised 7. 6. 2011, accepted 12. 9. 2011 * Corresponding author: e-mail thomas.misiek@kit.edu

spring with stiffness c attached to the beam. The effective torsional restraint coefficient c is obtained from 1/c = 1/cM + 1/cP + 1/cA (1)

2 Some necessary clarifications 2.1 Lateral restraint


If the statement in [1] that the lateral restraint by the purlin is usually considered as rigid were true, then there would be no need to consider the rotational restraint in the many cases where the compression flange of the purlin is connected to the sandwich panel. However, for various reasons, it is very often not possible to consider this lateral support by the purlin as rigid. In principle, this justifies their investigation of the rotational restraint, which also refers to other studies on this subject, e.g. for trapezoidal sheeting.

where cM flexural stiffness of sandwich panel cP distortion of beam cA connection between beam and sandwich panel. If hot-rolled I-sections are used, the distortion of the beams can be neglected and due to the large values of cM for the flexural stiffness of sandwich panels, the stiffness cA of the connection governs the torsional restraint coefficient c. Katnam et al. [1] and Schueremans [2] deal with the subject of stabilization with regard to cold-formed steel purlins with Z, C or cross-sections. In contrast to the statement in [1], it is very often not possible to consider the lateral support provided by the purlin as rigid. It obviously escaped Katnam et al. [1] and Schueremans [2] that by relating clockwise rotation to gravity loading and anticlockwise rotation to uplift loading with the test setup described in Euro-

2.2 Rotation of the purlin due to loading


The rotation of purlins due to the loading is caused by the distance of the line of action from the shear centre. The shear centre is that point in the plane of the section through which a transverse load must act if bending deflection only is to be produced, with no twisting of the section [6]. The line of action under uplift loading is through the connection of the panel to the purlin, whereas for grav-

Ernst & Sohn Verlag fr Architektur und technische Wissenschaften GmbH & Co. KG, Berlin Steel Construction 4 (2011), No. 4

251

M. Drr/Th. Misiek/H. Saal The torsional restraint of sandwich panels to resist the lateral torsional buckling of beams

ity loading due to the stiffness of the web it is through the point on the top flange closest to the web. Thus, under gravity loading there is only little, if any, rotation when using cross-sections with double or point symmetry. Under uplift loading, the purlin will usually rotate such that the outer edge of its flange connected to the panel will come into contact with the panel; the rotation of the panel may be neglected when compared with the rotation of the purlin. A contact close to the web will only occur under uplift loading if the shear centre is on the side opposite to the web when compared with the line of action through the connection point. Relating clockwise rotation to gravity loading and anti-clockwise rotation to uplift loading as in [1] with regard to an arrangement according to Fig. 1 applies only to cross-sections where the shear centre is to the left side of the web (action line of gravity load) and to the left side of the screw (action line of uplift load) if the rotation is due to the imposed loading. This means that this rule does not apply to sections if the action line of the forces transferred from the panel does not pass through the shear centre, e.g. Hor T-sections. The effect of an inclination of the roof is not considered here or in the following.

washer sheeting) with a masterslave surface pair option in a physical and geometric non-linear analysis may be of academic interest; however, this will hardly result in an engineering model for design practice. Some of the influencing factors a) shape and thickness of sheeting, b) thickness of foam, c) cross-section of purlin, d) number of screws per unit length of purlin, e) types of screw and their positions, and f) types of washer listed by Katnam et al. [1] are negligible, whereas other significant factors are not mentioned there, e.g. the mechanical properties of the foam, especially its Youngs modulus ES.

tional restraint. In fact, for a well-designed system under uplift loading, the indentation of the fasteners will lead to a gap between the inner face of the panel and the flange of the purlin. Therefore, in contrast to the beneficial effect of gravity loading on the restoring moment, c 0 applies for uplift loading. Thus, by ignoring the effect of the vertical forces, the experiment according to [7] underestimates the restoring moments in the case of gravity loading and overestimates this effect in the case of uplift loading. In this context it has to be pointed out that the results for uplift loading are unsafe, whereas the results for gravity loading may be accepted as a safe approximation.

2.4 Direction of loading and test setup according to Eurocode 3-1-3


Both Katnam et al. [1] and Schueremans [2] use the test setup described in Eurocode 3-1-3 [7], Annex A (Fig. 1). Both papers claim to include both gravity and uplift loading merely by rotating the purlin in the positive and negative direction. This procedure does not cover the effect of the vertical load because due to the horizontal loading (d) in Fig. 1, the restoring vertical forces between purlin and sheeting are tension and compression of equal magnitude. In the real situation, however, these forces include the vertical loading. These tests thus underestimate the tensile forces transferred by the screws in the case of uplift loading and overestimate them in the case of gravity loading. This is important because the indentation of the heads of the screw fasteners into the outer face is a governing parameter for the rota-

3 Investigations 3.1 Scope of the present study


The objective of the investigation was to close the existing gap in knowledge concerning the values of cA in order to use the stabilizing effect of all kinds of sandwich panels to resist the lateral torsional buckling of beams. Extensive experimental testing was performed in the study in order to determine the torsional restraint cA of steel sandwich purlin-sheeting systems under gravity loading. A parametric study by numerical analyses with a finite element model confirmed by the tests was performed to extend the results and to develop general rules for the determination of cA. In contrast to the previous investigations mentioned above, the effect of a gravity loading of realistic magnitude was included. In this paper, the term gravity loading is used for a load that is transferred via the roof panel through contact with the compression flange of the beam. For uplift loading, c 0 has to be assumed as discussed above.

2.3 Influencing factors and their appearance in an engineering model


Taking into account strain hardening effects for the faces of the sandwich panel, nominal strains in the core material of up to 80 % and hyperelastic material behaviour of the neoprene washer as well as modelling the four contact surfaces (screw head circular neoprene washer rectangular washer rectangular neoprene

3.2 Experiments
The experimental setup (Figs. 2 and 3) consisted of a beam that was supported by self-aligning ball bearings on both sides and covered with two sandwich panels each fixed to the beam by two self-tapping screws. After loading the sandwich panels with a gravity load, a torque mt was applied to the beam by means of lever arms, which were connected to both

Fig. 1. Experimental setup according to [5]: (a) trapezoidal sheeting or sandwich panel, (b) fastener, (c) beam, (d) load, (e) restraint

252

Steel Construction 4 (2011), No. 4

M. Drr/Th. Misiek/H. Saal The torsional restraint of sandwich panels to resist the lateral torsional buckling of beams

Fig. 2. Experimental setup

ends of the beam. The displacements of the upper and lower flanges of the beam due to the rotation were measured and converted into corresponding rotations . The experiments were performed with sandwich panels for applications in roof structures and sandwich panels for applications in wall structures. The core material was polyurethane and in some cases mineral wool. Hot-rolled IPE 160 I-sections or cold-formed Z 140-65 Z-sections with nominal thickness tN = 3.0 mm were used for the beams. The screws for the hot-rolled I-sections were located alternating left and right of the web in the trough of the sandwich panel. The connection for the Zsections was located in the middle of the flange. The elements were connected in the longitudinal seam by self-drilling screws. The mechanical properties of the sandwich sheeting and the core material were determined for every element. Table 1 shows the range of parameters tested. With the

same gravity loading p1, the torque mt, increasing stepwise, was applied to the beam in the positive and negative directions until the maximum amplitude of rotation amounted to = 0.1 rad. This was then repeated with increased gravity loadings p2 and p3. The torque and the resulting rotation were recorded by means of an elec-

tronic measuring system and illustrated as a moment-rotation characteristic as depicted in Fig. 4. This figure shows a large stiffness in the range from 0.02 rad to +0.02 rad and a significantly smaller gradient for the values beyond this range. The mechanical model in Fig. 5 explains the large stiffness, which affects the rotational restraint decisively. The loading is uniformly distributed over the complete area of the upper flange of the beam until the torsional moment is applied. As the torsional moment and rotation increase, the resulting force of the gravity load shifts towards the outer edge of the upper chord. This results in a counteractive moment mK with a theoretical maximum value mK = p b/2 (2)

For mt > mK, the eccentricity of the gravity load is not sufficient to resist the applied torque. Therefore, tensile

Table 1. Range of parameters for experimental investigations


Sandwich panels for roof and wall applications with core material of Polyurethane (PUR) Density of core material Modulus of elasticity ES [kg/m3] 3843 3.44.6 0.400.75 4080 IPE 160 and Z 140-65 Connection to beam: self-tapping screw EJOT-JZ3-6.3 L with 16, 19 and 22 mm dia. washer Connection in longitudinal seam of panels: self-drilling screw EJOT-JT3-2H-5.5 25 Mineral wool (MW) 111117 6.49.0 0.500.75

[N/mm2]

Thickness of sheeting [mm] Panel thickness [mm] Sections

Types of screw

Fig. 3. Experimental setup mechanical principle

Steel Construction 4 (2011), No. 4

253

M. Drr/Th. Misiek/H. Saal The torsional restraint of sandwich panels to resist the lateral torsional buckling of beams

Fig. 4. Example of moment-rotation characteristic for varying gravity load p

screw fastening, were performed with the ANSYS program. Fig. 7 shows the FE model together with a deformation plot in the direction of the panel thickness. Bilinear material behaviour was implemented for modelling the steel sheeting and the beam as well as the washers. Uniaxial compression tests were performed with optical measurements for determining the sectional stress-strain characteristic of the sandwich core material. Preliminary investigations with the FE model showed that the non-linear distribution of the modulus of elasticity in the direction of the thickness of the panel can be approximated by a homogeneous, isotropic and elastic material behaviour. The load transfer from the sandwich panel to the rotated beam is taken into account by contact ele-

Fig. 5. Mechanical model

forces in the connection elements have to contribute to the transfer of mt. These tensile forces cause indentations in the panel and thus decrease the stiffness of the connection. This results in a small gradient for the moment-rotation characteristic. According to Fig. 6, both parts of the moment-rotation characteristics obtained from the tests are approximated by a straight line. The large stiffness for small rotations is denoted with c1 and the gradient of the curve for the range of larger rotations with c2. Local deformations in the core material result in a continuous transition from c1 to c2, starting at approximately 2/3 of mK and idealized as shown in Fig. 6.

Fig. 6. Idealization of moment-rotation characteristic

3.3 Numerical analyses


The finite element analyses for the model, which represents the beam and the sandwich panel as well as the

Fig. 7. Finite element model with deformation plot

254

Steel Construction 4 (2011), No. 4

M. Drr/Th. Misiek/H. Saal The torsional restraint of sandwich panels to resist the lateral torsional buckling of beams

ments: a contact surface between the face layer and the upper flange of the beam sections serves for transferring the contact between the sandwich element and the beam section. Between these pairs of contact elements, compression and shear forces can be transferred if the gap is closed. The contact elements have the same geometric properties as the elements connected to them. The load steps for the numerical analyses are defined according to the test procedure mentioned before. In the first step the gravity loading is applied to the sandwich panel, following a torquemt. The moment-rotation characteristics resulting from the numerical analyses are compared, with regard to the values c1 and c2, with those from the experiments. Table 2 compares the values c1 and c2 obtained from experiments with those obtained from finite element analyses for roof and wall panels each with a core thickness of 40 mm. The maximum deviation between the experimental- and numerical-based mean values is 16 % for c1 and 14 % for c2. In particular, the single values of c1 show close agreement between numerical and experimental results for high gravity loads. This demonstrates that the FE model is well

Fig. 8. Influence of gravity load on moment-rotation characteristic

suited to investigating the stabilizing effect of sandwich panels to beams. Using this validated model, parametric studies can be performed to determine the factors influencing the momentrotation characteristic. The influence of the gravity load on the moment-rotation characteristic for a type TL75 panel (roof panel with 40 mm continuous core depth and profiled external face with 35 mm deep profiling) is depicted in Fig. 8, which shows that the gradient is inde-

pendent of the gravity load for small values of . The transition in stiffness occurs at approximately 2/3 of mK. The curves are parallel for values > 0.03 rad. This shows that not only c1 but also c2 is independent of the gravity load. Depending on the production process and the foam formulation, the elastic modulus ES of polyurethane core material usually varies between 2.0 and 6.0 N/mm2. Fig. 9 shows that c1 and c2 increase remarkably with

Table 2. Comparison of experimental values c1 and c2 with values from FE analysis


Load step and direction 1_pos 1_neg 2_pos TL75 panel for roof applications 2_neg 3_pos 3_neg Mean value Deviation [%] 1_pos 1_neg 2_pos W40 panel for wall applications 2_neg 3_pos 3_neg Mean value Deviation [%] 1.5 1.5 2.5 2.5 3.2 3.2 5.88 6.12 6.93 7.37 7.20 7.69 6.87 2 6.70 Gravity load [kN/m] 2.0 2.0 4.0 4.0 6.5 6.5 c1 [kNm/m] experimental 5.90 6.30 8.50 9.60 9.50 10.90 8.50 16 0.84 1.06 0.99 1.02 0.96 0.95 0.97 14 1.11 9.80 numerical c2 [kNm/m] experimental 2.05 1.75 2.17 1.79 2.20 1.59 1.93 12 1.70 numerical

Steel Construction 4 (2011), No. 4

255

M. Drr/Th. Misiek/H. Saal The torsional restraint of sandwich panels to resist the lateral torsional buckling of beams

Fig. 9. Influence of modulus of elasticity ES on moment-rotation characteristic

Fig. 10. Influence of support width b on moment-rotation characteristic

ES for a W40 sandwich panel (wall panel with 40 mm deep core and lightly profiled face). This is due to the reduction in deformations of the core material under the head of the screw and in the compression zone between sandwich panel and beam. The evaluation of the results shows that c1 and c2 are proportional to ES. Technical approvals require a minimum width b of 60 mm for the supports to sandwich panels. The variation in the width of the upper flange with 60, 82 and 100 mm is shown in Fig. 10 for a W40 panel. Values c1 and c2 both increase with width b. This favourable effect of the width b is due to an increase in the lever arm for the counteractive moment and the screwing as well. For architectural reasons, screws for connecting panels to purlins are often located either to the left or the right of the web instead of alternating. The numerical results in Fig. 11 confirm the test results showing that for the three possible cases (favourable, alternating, unfavourable) the arrangement of the fasteners only influences the c2 value. For the case of the alternating arrangement, c2 can be approximated as the average of the favourable and unfavourable arrangements. When connecting sandwich panels to cold-formed Z-sections, the screws are normally arranged in the middle of the supporting flange. Fig. 12 shows how a Z-shaped section with a height of 160 mm and a width b of the supporting flange of 60 mm influences the moment-rotation characteristic. The bilinear curve for negative rotation agrees with the curve shown in Fig. 11 for a favourable arrangement of the fasteners, the only difference being that the transition from c1 to c2 occurs at mK = p bZ (3)

The small stiffness in Fig. 12 for a positive rotation is due entirely to the connection because there is no contact in this case.

3.4 Statistical evaluation and development of a design model


According to the results of experiments and numerical analyses, c1 is

Fig. 11. Influence of fastener arrangement on moment-rotation characteristic

256

Steel Construction 4 (2011), No. 4

M. Drr/Th. Misiek/H. Saal The torsional restraint of sandwich panels to resist the lateral torsional buckling of beams

bined with the factors c1 and c2 in Table 4.

4 Conclusions
Experimental investigations and finite element analyses of the torsional restraint of beams subjected to gravity loads against lateral torsional buckling show that in contrast to [1] and [2], the rotational stiffness cA of sandwich panels depends significantly on the type of beam, the type of sandwich panel and the magnitude of the vertical gravity load as well as the modulus of elasticity of the sandwich core material. Based on parametric studies with a validated finite element model and a statistical evaluation of experimental results, formulae are derived to calculate the parameters c1, c2 and mK for gravity loading. These parameters describe a tri-linear moment-rotation characteristic for all common supporting structures with torsional restraint provided by sandwich panels. The rotational stiffness cA is obtained from this characteristic as secant stiffness depending on the vertical load on the beam. With

Fig. 12. Moment-rotation characteristic for 160 60 mm Z-section

influenced by the width b of the beam and the modulus of elasticity ES of the core material. Due to the limited range of b in the experimental results and therefore in the range of applications, a linear relation between c1 and b was used for the arrangement of the design equations. This is confirmed by the exponent < 2 obtained from the numerical investigations. As a result of the conclusions from the numerical analyses, the experimental values c1 for panels with a polyurethane core material are statistically evaluated with regard to the respective modulus of elasticity and width b. The results from roof and wall elements are combined into one population each and statistically evaluated to give the 5 % fractile value at a confidence level of 75 % according to EN 1990, Annex D, using a Gaussian (normal) distribution [8]. The statistical evaluation was the same for the panels with mineral wool core material. The value c2 is additionally influenced by the arrangement and flexibility of the connection. The c2 values are statistically evaluated for each population and with regard to the respective modulus of elasticity ES, the width b and the core thickness of the steel sheeting under the head of the screw. The rotational stiffness cA of the connection is obtained from the moment-rotation characteristic depicted in Fig. 6 as secant stiffness by

cA =

mK (mK )

(4)

where (mK) 0.08 rad. The parameters required for the description of the moment-rotation characteristic are obtained from Eqs. (5) to (10) listed in Table 3, com-

Table 3. Values c1, c2 and mK for moment-rotation characteristic


I-sections with 60 mm b 100 mm c1 = c2 = mK = where 2.0 ES 6.0 0.42 tK 0.67 b pd Z-sections with 60 mm b 100 mm (5) c1 ES 0 pd bZ (8) (9) (10)

c1 ES

b 82 b 82

c 2 ES t K pd b 2

(6) (7)

c1, c2

modulus of elasticity of sandwich core material in N/mm2 core thickness of steel sheeting under screw head in mm width of upper flange of supporting beam vertical beam load factor accounting for arrangement of fasteners: =1 for alternating arrangement = 1.5 for favourable arrangement =0 for unfavourable arrangement factors according to Table 4

Table 4. Factors c1 and c2


Core material Polyurethane (PUR) Mineral wool (MW) Area of application Roof Wall Roof Wall Type of steel sheeting under head of screw profiled faces flat faces profiled faces flat faces c1 1.44 1.20 0.69 0.48 c2 0.22 0.38 0.18 0.16

Steel Construction 4 (2011), No. 4

257

M. Drr/Th. Misiek/H. Saal The torsional restraint of sandwich panels to resist the lateral torsional buckling of beams

the rotational stiffness cA of the connection, the overall rotational stiffness c, which is required for the design against lateral torsional buckling of beams, can be calculated easily. For uplift loading, c 0 applies in contrast to [1] and [2].

Acknowledgements
The financial support of IFBS for the experimental part of this investigation is greatly appreciated.
References
[1] Katnam, K. B., Van Impe, R., Lagae, G., De Strycker, M.: Modelling of coldformed steel sandwich purlin-sheeting systems to estimate the rotational restraint. Thin-Walled Structures 45 (2007), pp. 584590. [2] Schueremans, L.: Rotational restraint by profiled sheeting An extensive set

of Pekz tests validating existing mechanical models. In: Eurosteel 2008 5th European Conference on Steel and Composite Structures. Brussels: ECCS European Convention for Constructional Steelwork, 2008. pp. 8792. [3] DIN 18800-2:2008-11: Stahlbauten Stabilittsflle, Knicken von Stben und Stabwerken. Berlin: Deutsches Institut fr Normung, 2008. [4] Drr, M.: Die Stabilisierung biegedrillknickgefhrdeter Trger durch Sandwichelemente und Trapezbleche. Berichte der Versuchsanstalt fr Stahl, Holz und Steine der Universitt Fridericiana in Karlsruhe; series 5, No. 17. Karlsruhe, 2008. [5] Drr, M., Podleschny, R., Saal, H.: Untersuchungen zur Drehbettung von biegedrillknickgefhrdeten Trgern durch Sandwichelemente. Stahlbau 76 (2007), pp. 401407. [6] Roark, J. R., Young, W. C.: Formulas for Stress and Strain. McGraw-Hill Kogakusha Ltd, Tokyo, 1975.

[7] Eurocode 3: Design of steel structures Part 1-3: General rules Supplementary rules for cold-formed members and sheeting. Brussels: CEN, European Committee for Standardization, 2006. [8] Eurocode 0: Basis of structural design. Brussels: CEN, European Committee for Standardization, 2002. Keywords: lateral torsional buckling; sandwich panel; stabilization; torsional restraint; purlin

Authors:
Dr.-Ing. Markus Drr, Montana Bausysteme AG, Durisolstr. 11, 5612 Villmergen, Switzerland Dr.-Ing. Thomas Misiek, Univ.-Prof. Dr.-Ing. Helmut Saal, Versuchsanstalt fr Stahl, Holz und Steine, Karlsruhe Institute of Technology (KIT), Otto-Ammann-Platz 1, 76131 Karlsruhe

258

Steel Construction 4 (2011), No. 4

You might also like