i

Thin-Walled Structures Vol. 27, No. I, pp. 43-53, 1997

Copyright ,~) 1996 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0263-8231/97 $15.00
ELSEVIER PI I:S0263-823 I (96)000 ! 9-5

Design Procedures for Profiled Metal

Sheeting and Decking

J. M. Davies & C. Jiang

Manchester School of Engineering, University of Manchester, Manchester, U.K.

ABSTRACT

When designing profiled metal sheeting and decking at the ultimate limit
state, the conditions at the internal supports, where the bending moment
interacts with the support reaction force, are critical. Design code provi-
sions./br estimating the moment of resistance over an internal support are
known to be inadequate and the problem is aggravated ([account is taken
q[" the redistribution o[" bending moment accompanied by post-yieM buck-
ling. This paper shows that the per[ormance at the internal support can be
pretficted b)'.[inite element analysis and that continuing the analysis into the
drooping post-yield range allows a relatively simple pseudo-plastic design
method to be developed. Copyright ~ 1996 Elsevier Science Ltd.

1 INTRODUCTION

Cold-formed profiled metal sheeting and decking are well-established

construction products and manufacturers make strenuous efforts to keep
these products competitive in terms of the carrying capacity for a given
weight of material. In the past, this has generally involved extensive test-
ing. Design by calculation, using for instance the 'European Recommen-
dations for the design of profiled sheeting '~ is possible but results in
significantly less economical designs.
In another application, cold-formed steel decking is often used as
permanent form work for the slab in reinforced concrete floor and roof
construction. Here, the profiled steel decking may be designed to behave
compositely with the in situ concrete by introducing a mechanical

43
44 J. M. Davies, C. Jiang

connection in the form of dimples or other indentations in either the

flanges or webs of the deck profile. An important design criterion for this
type of construction is that the decking should be able to support the wet
concrete and other construction loads without the need for temporary
propping between the beams.
The European Recommendations for the design of profiled sheeting j
have been shown to give an excellent prediction of the bending strength
for a wide range of sheeting and decking profiles including those with web
and/or flange stiffeners. 2 However, for decking which is continuous over
two or more spans, the design is decided by the interaction of hogging
bending moment and reaction force at an internal support. For this case,
the European Recommendations (and also other related standards)
provide an interaction equation which attempts to predict the yieldin&
buckling capacity under the combined forces. The accuracy of this equa-
tion can be investigated by comparing the predicted failure conditions
with those observed in simulated central support tests in which a line load
is applied to the centre of a simply supported deck whose span is chosen to
give the same ratio of bending moment and reaction as at the internal
support in a two-span test. However, the results of this comparison show a
huge scatter with very poor correlation between the test results and the
formula. 2
Most sheeting and decking profiles can sustain considerable redis-
tribution of bending moment so that the attainment of the calculated
moment of resistance (reduced if necessary for the interaction with the
support force) at an internal support is not immediately followed by
failure. After yielding and/or buckling at the support, 'plastic hinge'
action occurs, possibly accompanied by a reduction in the moment of
resistance, until failure takes place when the full moment of resistance is
attained within the span. Because the mid-span moment of resistance is
often greater than the (reduced) moment of resistance at a support, the
increase in the load carried as a result of this moment redistribution can
be considerable. 3
The present state of the art therefore requires either testing or quasi-
elastic design based on the calculated moment of resistance at the internal
support. However, it is shown in this paper that this situation can be
improved by the use of a second-order, elastic-plastic finite element
analysis to predict the moment-rotation relationship at an internal
support. This can then be combined with the prediction of the span
moment of resistance given by the European Recommendations to
produce a suitable mathematical model for the pseudo-plastic design of
continuous decking. It is shown that the results of this procedure compare
well with those obtained from double span tests.
 Design of metal sheeting and decking 45

The influence of ribs and dimples used to promote composite action is

also considered. It is shown that their influence on load-carrying capacity
depends on the configuration and placement of the dimples.

2 P S E U D O - P L A S T I C DESIGN

There are considerable benefits to be gained by taking advantage of plas-

ticity at the internal support. To date, this has generally required the test-
ing of two complete spans. A comprehensive and expensive testing
programme is therefore required if a family of sheeting or decking profiles
is to be designed on this basis.
A series of tests on some of the profiles commonly used in composite
deck construction was carried out at the University of Salford 4 and, as a
consequence, a simple 'quasi-plastic' design of continuous decking was
developed by permitting partially plastic rotation at the internal supports.
This allowed longer spans and greater economy while maintaining an
adequate factor of safety against failure during concreting. Thus, the ulti-
mate load w,lt per unit length is given by
8
wull = ~ [ M 2 + 0.46KM1] (1)

where M2 is the moment capacity in the span and the term K (less than
unity) represents the proportion of MI, the bending m o m e n t capacity over
the support, which can be mobilised at plastic collapse of the deck. It was
shown that safe designs were achieved for decks current at the time with
values of 0.46 K between 0.1 and 0.4.

3 F I N I T E E L E M E N T ANALYSIS

It is now apparent that the behaviour at an internal support can be predicted

by finite element analysis as well as by testing. As analysis is considerably
less expensive than testing, and as it opens up the possibility of parametric
studies, this development represents a considerable step forward.
Some profiled steel decks which have previously been tested 4 are shown
in Fig. 1. These have been analysed by the finite element method using
second-order elastic-plastic nine-node Heterosis elements. 5 In this analy-
sis, an idealised perfect elastic-plastic stress-strain relationship
(E = 180 k N / m m 2, v = 0.3) was adoped and, in the thickness direction, a
four- to six-layer model was employed to detect yield. Because the decks
were of periodic construction, it was sufficient to analyse only one quarter
46 J. M. Davies, C. Jiang

Holorib

07j
r- r
16o 4

L 61.2 J
Metecno r 1

rL 61.2 -r'L 86.8 ,~

Holodek t 300
L 112 a
r - '1

Fig. 1. Dimensions of tested sections.

of a complete rib in order to reduce the amount of computation. The

boundary conditions were the same as in the simulated central support
test in that, at the free end, the decking was simply supported at all nodes
and, at the internal support, all nodes were sliding clamped. A single point
load was applied at the web-flange intersection 75 mm from the centreline
simulating four such loads 150 mm apart on the complete flute.
In order to obtain the drooping section of the moment-rotation curve,
the displacement at the internal support was controlled during the
computation. By means of convergence studies, it was found that the
element mesh divisions of the cross-section listed in Table 1 were sufficient
to give good accuracy. At the internal support, a close mesh was required
in the length direction in order to model the deformations so that the
region near the support was divided into 10 or 11 divisions. A further 15
divisions were then required in the length direction in order to model the
remainder of the section.

TABLE 1
Number of Elements used to Describe the Cross-Section

Section Bottom flange Web Top .]~aFIge

Holorib 4 5 2
Metecno 3 6 2
Holodek 3 5 3
 Design o['metal sheeting and decking 47

14 m

12
- Mn

K " "
10

E 8

6
E

O/+F Test O +
2 Finite elements - -

0E I I L I I I ]
0 10 20 30 40 50 60 70
Deflection (mm)
Fig. 2. M o m e n t ~ l e f l e c t i o n r e l a t i o n s h i p s for t w o M e t e c n o sections.

Figure 2 shows a typical comparison between the results of simulated

central support tests and the corresponding analysis of two Metecno
decks. The total deflection is shown including that due to local crushing at
the end supports. The plastic rotation may be calculated from the plastic
deflection component of the theoretical curves. It is seen that the experi-
mental and analytical results agree well and a similar level of agreement
was obtained for all of the profiles investigated.
When the ultimate value of the bending moment was reached, the decking
section was still elastic-plastic but the slope of the load~teflection curve
became negative due to the local buckling of the compression flange and
web. For the profiled decks with a wider compression flange, e.g. Holorib
1.2/50 and Holorib 0.9/50, the slope of the curve drooped more gently. This
is typical of unsymmetrical cross-sections where yield begins on the tension
side. A typical buckling configuration, plotted from the analytical results for
Holodek 1.2/76, is shown in Fig. 3. It can be seen that the decking buckled
in local modes at the internal support due to the combined action of
compression and bending and that there was some crippling of the web.

4 I N T E R A C T I O N BETWEEN B E N D I N G M O M E N T A N D
REACTION

Because it was difficult to use an analytical method to investigate the rela-

tionship between bending moment and reaction, previous research has
concentrated on using experimental methods to evaluate empirical formu-
48 J. M. Davies, C. Jiang

Fig. 3. Buckled shape of Holodeck 1.2/76.

lae. In all the existing formulae, it has been assumed that the strength at an
internal support could be expressed in terms of two parameters, M/Md and
R/Rd, where Md and Rd are the ultimate resistance moment in pure bending
and the ultimate web crippling load, respectively.
In order to investigate in more detail the relationship between the
bending m o m e n t and reaction in a simulated central support test, a para-
metric study in which the M to R ratio was systematically varied has been
carried out on two typical second generation trapezoidal decks, namely
5202 and 6096 using the finite element method. The computational results
are shown in Fig. 4 and reveal that M/M,~ can be significantly greater
than 1 when R/Rd is less than 0.4. This unexpected result explains why it
has been so difficult to find a satisfactory empirical design expression for
Rd. It also explains a large part of the scatter observed by previous
researchers.

1.5 F Finite
/ ~ e l . . . . ts

1.0

M___
Md
0.5

ol I I
0 0.5 1.0
R
Rd

Fig. 4. Theoretical M - R curves.

 Design of metal sheeting and decking 49

A possible explanation for this increase of strength is that the interac-

tion of bending moment and reaction causes a complicated stress combi-
nation in the deck. This includes a compressive stress at and a shear stress
caused by the transverse loads, as well as the longitudinal stress o- which
dominates the resistance bending moment. If the material follows the von
Mises yield criterion, the following equation can be obtained:

V/a 2 - oat + a~ + 3z 2 <~Fy (2)

where Fy is the yield strength. It can be seen that the allowable value of a
increases with the value of a t until at is about 0.5Fy. This means that the
stress interaction can improve the stress distribution in a member when
the values of at and z are relatively small. Conversely, if the value of at is
higher than 0.5Fy, the moment of resistance will decrease rapidly due to
either web crippling or shear buckling.

5 AN IMPROVED THEORY FOR PSEUDO-PLASTIC DESIGN

The plastic collapse mechanism for a two-span deck is shown in Fig. 5. If

M~ and M2 are defined as the plastic bending moments at the support and
in the span, respectively, at collapse of the deck, and wc is the ultimate
load per unit length, equating the internal and external virtual work for a
small rotation of the collapse mechanism gives

2E
W~=Lx M I + M 2 ~ _ x L1 (3)

The distance x from the support to the plastic hinge in the span can be
obtained by differentiating with respect to x and equating to zero which
yields

x = M, + M2 - [(M1 + M2)M2] '/2 (4)

L M~

we per unit langth

l I l | l l I l l I I l | l l l I l l l l

M2 MI M2

Fig. 5. Collapse mechanism for a two-span deck.

50 J. M. Davies, C. Jiang

The necessary compatibility relationship in terms of the plastic rotation 0p

at the support and the residual support moment MI is

3EI Ml ] (5)

where I is the average inertia of the section. Equations (3) and (4) may be
solved to find wc and x knowing the relationship between M~ and 0p from
the simulated support test.
Based on the above discussion, a design procedure for profiled decking
is suggested:
(1) Determine the moment-rotation characteristic at the internal
support by either the simulated central support test or by finite
element computation.
(2) Determine the maximum value of positive bending moment M2 in
the span by either a simply supported decking test or by calculation
according to the European Recommendations or by finite element
analysis.
(3) Calculate the collapse load Wc using eqns (3) and (4). The value of
M1 can be determined from the moment~teflection characteristics
and eqn (5) using iteration.
A comparison of the results from the design approach and some two
span tests 4 is given in Table 2 and shows good agreement.
The above iterative method is suited to hand calculation. In order to
simplify the calculation, it is safe to draw a straight line as shown in Fig. 6
in which the bending moment M~ is equal to the ultimate bending moment
Myl at 0p ----0 and is asymptotic to the drooping curve for higher values of
0p. This line is given the equation:
M1 ~- Myl - kOp (6)

TABLE 2
Comparison of the Results from Two-Span Tests and the Design Method

Section Span L Thickness t Fy Test load Design wc wet

(m) (m) (MPa) wet (kN) (kN) w~

Holorib 0.9/50 3.0 0.890 313.7 25.77 23.83 1.08

Holorib 1.2/50 3,5 1.195 302.5 34.81 31.31 I. I 1
Metecno 0.8/55 3.0 0.760 316.0 15.69 17.15 0.91
Metecno 1.25/55 3.5 1.200 317.0 32.60 31.21 1.04
Holodek 0.9/76 3,5 0.880 324.0 25.97 24.67 1.05
Holodek 1.2/76 4,2 1.185 300.5 31.25 28.13 1.11
 Design of metal sheeting and decking 51

Central bending moment M ~ M

My1 . . . . My1

[ Plastic rotation ep

Fig. 6. Interpretation of the simulated central support test.

Together with eqn (5), this yields

wc L 2 E1
Mt - 8 3 ~ ( M y l - MI) (7)

If M2 is obtained as before, the values of x, 0~ and wc can be obtained by

solving the nonlinear eqns (3), (4) and (7) using any suitable computer
method such as N e w t o n - R a p h s o n or the standard procedure embraced in
MathCAD.

6 B E N D I N G C A P A C I T Y OF DECKS WITH D I M P L E S

Because of the complications caused by the location, dimensions and

shape of dimples designed to promote composite action with in situ
concrete, until now only very limited experimental research has been
carried out in order to investigate the effect of these dimples on the
strength of the deck. It has been found 7 that the presence of dimples in
webs reduced the web crippling strength, although the reduction occa-
sioned by longitudinal dimples is usually less than 10%. More recently,
Luure & Crisinel s concluded that the influence of web dimples was to
decrease the bending strength by less than 10%. The effect of flange
dimples on the bending strength does not appear to have been studied.
In order to investigate the effect of dimples on the bending strength, two
decking profiles have been analysed using the finite element method
described previously. There are square dimples in the flanges of the
Holorib profile and rectangular dimples in the webs of the Metecno
profile, the dimensions of which are shown in Fig. 7. The results obtained
for the sagging bending strength of the decks with and without dimples
are given in Table 3 where FEM - refers to analysis ignoring the dimples
52 J. M. Davies, C. Jiang

j
1• 150 . I r 88, r /

Fig. 7. Decks with dimples to promote composite action.

TABLE 3
Comparison of the Bending Strength with (+) and without ( - ) Dimples

Ultimate moment ( k N m / m ) (M'u - M~)

Dimple M'o
Section location F E M - M~u F E M + M'u T E S T + Mut (%)

Metecno 0.8/55 web 4.53 4.38 4.38 3.3 I

Metecno 1.25/55 web 9.95 9.65 9.65 3.02
Holorib 0.9/50 top flange 6.33 5.81 5.66 8.21
Holorib 1.2/50 top flange 9.94 8.85 10.58 10.96

and F E M + refers to an analysis which takes them into account. The test
results are for single span vacuum tests. 4
It can be seen that the dimples decrease the bending strength. In the
case of the dimples in the compression flange, the effect is more significant
than in the case of the dimples in the webs. This decrease is approximately
inversely proportional to 2p -- b/t(x/~y/E).
If elastic buckling of flanges
dominates the strength of a steel deck, i.e. ~.p has a relatively large value,
dimples in the compression flanges can increase the bending strength. The
relationship between compression flange dimples, 2p and bending strength
remains to be investigated in further parametric studies.

7 CONCLUSIONS

A design procedure has been described which is based on the formation of

a pseudo-plastic collapse mechanism which utilises the redistribution of
 Design of metal sheeting and decking 53

bending m o m e n t following initial yielding or buckling at the internal

support.
The presence of the support reaction force can actually raise the
m o m e n t of resistance over a significant part of practical range.
Dimples in the compression flange may affect the bending strength of
composite decking and should be considered in the design for the wet
concrete stage,

REFERENCES

1. European Recommendations for the design of profiled sheeting. European

Convention for Constructional Steelwork, ECCS-TC7-1983.
2. Baehre, R. & Fick, K-F., Berechnung und Bemessung von Trapezprofilen -
mit elauterungen zur DIN 18807, Berichte der Versuchsanstalt fur Stahl,
Holz und Stein, Universitat Fridericiana Karlsruhe, 1982.
3. Davies, J.M., Sheeting and decking. In Design o[Cold.[brmed Steel Members,
Chap. 10, ed. J. Rhodes, Elsevier Applied Science, Amsterdam, 1991.
4. Bryan, E.R. & Leach, P., Design of profiled sheeting as permanent form-
work. CIRIA Technical Note 116, London, 1984.
5. Figueiras, J.A. & Owen, D.R.J., Analysis of elasto-plastic and geometrically
non-linear anisotropic plates and shells. In Finite Element Software/or Plates
and Shells(Edited by E. Hinton and D,R.J. Owen). Pineridge Press, Swansea,
UK, 1984.
6. Hoglund, T., Design of trapezoidal sheeting provided with stiffeners in the
flanges and webs. Swedish Council for Building Research, Document D28,
Liber Tryck Stockholm, 1980.
7. Yu, Wei-Wen. Web crippling and combined web crippling and bending of
steel decks. Dept. of Civil Eng., Univ. of Missouri-Rolla, April 1981.
8. Luure, P. & Crisinel, M., Comparison tests for cold-formed sheeting with
and without embossments used in composite floors. ICOM 289 E, Lausanne,
Nov. 1993.
9. Jiang, C., Stability analysis of light gauge steel members using the finite
element method and the generalized beam theory. Ph.D thesis, Univ. of
Salford, 1994.