See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/301723435

FEM model of the steel building roof includes stressed skin diaphragm action
effects

Chapter · May 2016

DOI: 10.1201/b21417-12

CITATIONS READS

7 189

2 authors:

Marcin Gryniewicz Jerzy Kazimierz Szlendak

Bialystok University of Technology State School of Higher Professional Education in Suwalki, Poland
11 PUBLICATIONS   9 CITATIONS    29 PUBLICATIONS   39 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Structural modelling of the steel buildings covered by the metal sheeting View project

Pitched roofs covered by metal sheeting - analysis includes stressed skin diaphragm action effects View project

All content following this page was uploaded by Jerzy Kazimierz Szlendak on 23 January 2021.

The user has requested enhancement of the downloaded file.

FEM model of the steel building roof includes stressed skin diaphragm
action effects
M. Gryniewicz & J.K. Szlendak
Faculty of Civil and Environmental Engineering, Technical University of Bialystok, Poland

ABSTRACT: The paper presents new numerical model for pitched roof of steel building covered by metal
sheeting. Stressed skin diaphragm action effect is considered as a main subject of the discussion. Structural
components of such buildings should be adequately taken into account in analytical models. Various bracings
locations or different main frames structure solutions lead to their different global stiffness caused by horizon-
tal forces. Examples of calculations are shown. Presented procedure consists of diaphragms assembly calcula-
tions made by using European Recommendations (ECCS) and other data from literature. Representing struc-
tural key components are under consideration. Fasteners are replaced by equivalent finite frame elements.
Trapezoidal sheeting is substituted by the equivalent orthotropic plate model. Interaction between the frame
and sheeting is also considered. Numerical results presented in the paper are verified by full scale tests made
by one of the authors and others.

1 INTRODUCTION by Dong Li (2006). Comparison of methods predict-

ing diaphragm behavior are described by Liedtke &
Light gauge steel cladding are widely used for Sherman (1982). Additional testing and numerical
buildings, however their including in structural cal- analyzes are given by Bakhdi et al. (2012).
culations as stressed skin diaphragms, up to now is
not common solution. Corrugation of sheets compli-
cate their in plane and out of plane behavior. More- 2 BACKGROUND OF THEORETICAL MODEL
over, sheets are connected each other and to sup-
porting structure in different configurations. Variety 2.1 Stressed skin design principles
of connector types can be used. Additionally, open- In ER (1995) formulas and procedures required to
ings for windows, doors and others can be applied. calculate bearing capacity and stiffness of shear
Analytical studies have to involve many various panels are given for separate cases. There are two
failure modes e.g. local buckling, global stability general types: sheeting spanning perpendicular to
loss, connections collapse etc. the span of the diaphragm or parallel to it. Depend-
Over the past decades design manuals and practi- ing on using walls bracings, “diaphragm beam” or
cal instructions have been prepared. In Europe, re- “cantilever diaphragm” can be recognized. Next step
search carried out by Baehre with accompany of is to describe rest of the structure elements configu-
Bryan & Davies, leads to recommendations provided ration and types (connectors, edge beams, purlins
by European Convention for Constructional Steel- etc.), where design procedures can be recognized in-
work (1995). General review and the recently state to next sub-groups.
of the art in the stressed skin design of steel struc- As written by Davies (2006) the strength of a dia-
tures is given by Davies (2006). The current version phragm is defined by the weakest potential failure
of Eurocode 3 for the design of cold-formed mem- mode. In case of serviceability conditions, flexibility
bers and sheeting is referred to the ECCS Recom- of the diaphragm is determined by the sum of a se-
mendations (ER) (1995). Useful comments and addi- ries of component flexibilities:
tional practical examples can be found in handbooks  Sheeting profile distortion: c1.1
e.g. Höglund (2002).  Shear strain in the faces of the profile: c1.2
The result of tests and researches in America (e.g.  Sheet to support fasteners deformation: c2.1
Nilson 1960 and many others), is AISC design man-  Seam fasteners deformation: c2.2
ual (Lutrell 2006). Explanations and summary of  Gable connections deformation: c2.3
major changes made in recent editions are presented  Axial strain in the edge members: c3
 For each failure mode and each flexibility com- V
 xy  (5)
ponent simple design expressions are available with bp t
the addition of comprehensive examples.
where u = diaphragm displacement; V = shear force;
bp = depth of the shear panel in the direction parallel
2.2 FEM analysis of shear diaphragms to the corrugations, t = plate thickness.
Authors works on numerical approaches and solu-
tions in the context of stressed skin diaphragm ac-
tion have been inspired by Nilson (1960) and further V
Atrek & Nilson (1976) investigations. Their methods

p
make greater use of the orthotropic plate conception

u
to describe behavior of trapezoidal sheets. The
stress-strain relationship for orthotropic plate in

bp
plane stress continuum mechanics problem, can be
written as:
 E   ap
   yx Exx 0
 x   xx   x 
 y  
1  xy E yy E yy 0    y  (1) Figure 1. Calculation model for diaphragm shear flexibility.
  1   xy yx   
 xy   0

0 1- xy yx  Gxy   xy 
 Diaphragm displacement u can be determined
Local coordinate system axis y is perpendicular to with values described in paragraph 2.1 for sheeting
cross section plane of folded plate and x is parallel to plate flexibility component:
it. Equivalent elastic modulus expressions are not u  V (c1.1  c1.2 ) (6)
the same, determined by geometry of corrugations
and their physical behavior. It can be estimated, as Applying formulas (4), (5) and (6), shear modulus
described by Atrek & Nilson (1976), using two ex- for corrugated panel is obtained:
pressions below: ap
Gxy  (7)
I x0 b p t  c1.1  c1.2 
E xx  Es (2)
Ix
As a last unknown variable used in orthotropic
lp
plate stiffness matrix showed in formula (1), Pois-
E yy  Es (3) son’s ratio xy is equal to 0,3 according to EC3. At
ap yx direction this value must be multiplied by the
where Es = elasticity modulus of base steel material; quotient of Exx and Eyy.
Ix0 = equivalent plate moment of inertia about its
own plane; Ix = corrugated plate section moment of a) b)
inertia about its own plane; lp = the developed width
Y
of the corrugation; ap = width of the shear panel in ky j F
the direction perpendicular to the corrugations (their
pitch in this case).
kx
0

For the Eyy more sophisticated and accurate ex- ky

pressions are exist in recent scientific studies. It is
assumed that for described practical method formula i X
kx
0
(2) is sufficient. The equivalent modulus Eyy is simp-
ly depend on fact that both sheets (real one and its u
model) must have the same longitudinal deformation Figure 2. Connection idealization a) theoretical model, b) typi-
under applied load. cal relationship force-displacement for connectors (real behav-
Shear modulus can be determined experimentally ior and linear approximation).
or numerically. For described model purposes ana-
lytical approach will be used. As seen in Figure 1, The connectors model are two perpendicular
expression for Gxy can be found using shear strain springs with stiffness kx and ky. Those springs con-
(4) and stress (5) relationship: nect two nodes having the same coordinates (dis-
u  xy tance between them equals zero, Figure 2a). One
 xy   (4) node is a part of sheeting plate while another one is
a p Gxy
connected with structural element or adjacent sheet.
Figure 2b describe typical force-displacement rela-
tionship. Generally, linear approximation can be
used for practical calculations (Bródka et al. 1999).

2.3 Diaphragm-frame interaction

Trapezoidal sheeting as a shear panels is a perfect
solution for building structure stabilization. The
stiffness of the sheeting is usually greater than the
stiffness of the wind trusses which are now common
used to stabilize building structures (Höglund 2002).
Stressed skin action effect leads to reduction in
the associated forces and bending moments in sup-
porting frames. This depends of the in-plane flexibil-
ity of shear diaphragm in relation to the frame flexi- Figure 4. Isometric view of tested steel hall (transverse axes
bility. numbers starts from left at “1”).
ER (1995) give some procedures how to solve
this problem, however with some limitations and as- Columns bases are designed as hinges. Gable
sumptions. In one of the ER sections, recommenda- frames are different than main ones and have I-beam
tions are given only to single-bay flat roof frames member instead of truss.
and to symmetrical single-bay pitched roof frames. Gable frame in axis “4” (see Figure 6 for axes lo-
General assumptions are as follows: all frames in a calization) is stiffened in its plane by diagonal brac-
building has uniform construction, all shear panels ings. Roof and longitudinal walls are also braced be-
are the same and all foundations and other condi- tween axes “4” and “5” . Between “1” and “2” only
tions are similar. walls are stiffened. Columns, at the middle of their
height, are prevented from out of plane buckling by
wall girts.

39 R3 266
40

R3
R3

108

54°
Figure 5. Corrugated plate geometry.

Figure 3. Example reference model for computational calcula-

tions method given in ER for flat roof buildings.
Roof is covered by trapezoidal sheets T 40 placed
in negative position (Figure 5) on purlins made with
For other cases computer analysis methods are cold-formed sections Z 200 x 68 x 60 x 1.5 mm.
allowed. ER gives procedure where sheeting panels Sheet profile has height 40.0 mm and thickness
are replaced by system of springs (see example in 0.7 mm. Pitch of corrugation is equal to 266.0 mm.
Figure 3). Similar approach with using equivalent
diagonal members is proposed e.g. Garncarek & 3.2 Tests procedure
Dziatkowski (2000) and Szlendak (2002).
Tests were performed on two stages. First one,
where building steel structure was tested without any
3 NATURAL SCALE TESTS coverings. In second approach roof was fully cov-
ered with TR sheets. At both approaches three
3.1 Prototype steel building frames were under examination: “3”, “4” and
The steel structure which is the subject of presented “5” axis ( Figure 6).
analysis is the building with dimensions measured in Experiments were performed by applying horizon-
the axes of the main profiles: 12.0 m (width), 20.0 m tal force at the top joint of the truss (or gable beam)
(length) and 6.1 m (height of the main frame). It is a where it is connected to column. Registration of the
single-bay hall composed of five frames spaced at displacements was carried out at the same node with
equal distances of 5.0 m (Figure 4). The main struc- dial gauge. Range of testing force has been estab-
ture of the roof is truss (chords made with RHS pro- lished from 1 kN to 15 kN.
files and diagonals with angles), which are support-
ed by wide flange HEA 160 columns.
 5000 5000 5000 5000
Test
Table 1. Main frames theoretical stiffnesses.
______________________________________________
Force Frame type Axis Symbol Stiffness

6000
[-] [-] [kN/m]
12000
______________________________________________
Gable (unbraced) 1 kf.1 116
Middle (main) 2-4 kf.2-4 150
1 2 3 4 5 Gable (braced) 5 kf.5 937
Purlin fastener ______________________________________________
Seam fastener
Main test force
Connectors type: For comparison purposes, main frame behavior is
self drilled screws GT5 5,5x25
right to choose since is located in a middle area
where only purlins and wall girts are mounted with
theoretical pinned joints (relatively minimal influ-
ence). Stiffness obtained by test in complete struc-

1500
ture was 186 kN/m and by only-frame calculations
150 kN/m. The higher value from test can be caused
by interaction of purlins (continuous beams) which
involved other frames to act together. Additional di-
al gauges were applied to every column to check this
1500
SPANS OF PURLINS
relationship. Finally, building 3D structure calcula-
tion model confirmed this conclusion.
Value 12.5 mm of maximum deflection has been
obtained after 10 kN force applying for building
with roof sheeting. Deflection of this point in struc-
1500

ture without covering (in the first test) has been

equal to 53,7 mm (Figure 7). Thus, real displace-
ments reduction caused by application of diaphragm
is equals to 78%.
1500

5000
DISTANCE BETWEEN FRAMES

Figure 6. Tested roof diaphragm geometry and fasteners ar-

rangement.

Horizontal displacements values were recorded at

every step with 1 kN increment. There were no con- Figure 7. Test results with and without sheeting
struction permanent deformations noted. Geomet-
rical imperfection were reduced to minimum by ap-
plying initial force. 4 CONCEPT OF ANALYTICAL MODEL

Diaphragm-frame interaction procedure could be

3.3 Results and static analysis carried out, e.g. with usage of European Recom-
As it was expected, quasi-linear behavior of frames mendations. As long as every standard requirement
under horizontal point load was registered. At first given in ER is fulfilled, there should not be any dif-
stage tests were performed to check structure stiff- ficulties. Otherwise, the designer have to use non
ness used for analytical analysis. Second stage was common and more complicated solutions. Building
made to define real deformation with stressed skin described in Chapter 3 is an example of such a case,
diaphragm action effect. mainly due to its irregular bracings configuration
Specific frame in-plane stiffness determination by and different frames construction.
testing it in complete structure may be difficult. Authors propose theoretical model to estimate
Bracings, wall girts or purlins have also some influ- stressed skin diaphragm action with using ER key
ence on frame flexibility. In that case, static calcula- principles, previous research and application of Fi-
tions were made to estimate own theoretical stiff- nite Element Method. Every element of diaphragm
nesses for each type of frames (Table 1). structure is used in this model to check deflection of
the building and compare it with test results.
4.1 Model components description
Table 2. Equivalent characteristics of diaphragm
4.1.1 Corrugated plate connections.
__________________________________________________
Formulas given in ER are sufficient and can be used Connection type Element Theoretical Moment
in this approach. First feature to define is sheeting length slip of inertia
profile distortion: [mm] [mm/kN] [10-3mm4]
__________________________________________________
Main fastener s 1.0 0.35 4.54
a p p 2,5
f 1 4 K
c1.1  (8) Seam fasteners 2.0 0.25 50.79
Et 2,5b 2p Purlin-girder (a)* 10.0 1.40 1133.79
Purlin-girder (b)** 10.0 0.005 317460.32
__________________________________________________
Second one is shear strain in the faces of the profile: * out-of-plane direction
** in-plane direction
  2h  
2a p 2 1    1   
  p f  
c1.2  (9) Direction of span sheeting
Etb p
a) b)
In formulas (8) and (9) variables are (except defined Purlin
before): pf = pitch of the corrugations; α1-α4 = fac-
tors to allow for the number of purlins and sheet
connectors; K = sheeting constant which is a func-
tion of the shape of the cross-section; E = modulus Purlin
Seam fastener
of elasticity of the sheet material;  = Poisson’s ratio c)
of the sheet material; h = height of the sheeting
profile.
For analyzed diaphragm c1.1 is equal to
Purlin
0.149 mm/kN and c1.2 has value 0.013 mm/kN. Us-
Main fastener
ing expressions shown in Chapter 2.2, final form of
the stiffness matrix for trapezoidal panel as an Shear panels
equivalent orthotropic plate is given as bellow: Figure 8. General concept of connectors modeling
 21.5 6.5 0  a) Diaphragm fragment, b) Seam fastener model, c) Main fas-
 kN tener (sheeting-purlin connection) in isometric view.
 Eeq   6.5 167253.5
 0  (10)
m
 0 0 5118.5 Equivalent members for fasteners are described in
model as 3D frame finite elements with the same
section properties at both directions. In the Figure 8
4.1.2 Connections general approach for connectors modeling is shown,
General assumption in case of connections model- where seam fasteners connects adjacent sheets in
ling has been explained before (using two springs their plane. Main connectors are joints between
connected two nodes with zero distance). Authors sheets and purlins.
proposed another solution. Every connector in sys-
tem will be replaced by equivalent cantilever frame 4.1.3 Building structure members
finite element (Gryniewicz & Szlendak, in press). Gable walls beams, purlins, bracings etc. are applied
Needed properties of the equivalent members sec- as 3D frame finite elements. Bars sections are the
tions can be determined using ER or other sources same as in the real structure.
and obtained by test or numerical simulations. For In this diaphragm analysis full model of trusses is
connections the most important feature is slip of fas- not necessary, only upper chords are needed to con-
teners and other connectors in diaphragm structure. nect purlins (maintain local support flexibility) and
Theoretical connections flexibilities are shown in create required roof geometry (Figure 9). Each steel
Table 2. Slips of fasteners are adopted from ER. Pur- frame behavior is carried out by using spring sup-
lin-girder connection is not typical and is tested with ports with stiffnesses calculated in Chapter 3.3.
numerical simulation as a cantilever plate with dif- Purlins-rafters connections are described using
ferent characteristics at both directions. the same method as main sheeting fasteners but with
Representative moment of inertia in specific di- different properties in both directions (see Table 2).
rection can be calculated with following expression: To avoid non-linear calculations only diagonal
3
leq braces under tension are included into the model as
I eq  (11) round bars with 12 mm diameter.
3sE
where: leq = chosen length of equivalent finite ele-
ment member; s = connection slip
4.2 Diaphragm numerical calculations will be interesting, ER provide solutions to estimate
this effect.
The studied building have pitched roof frames. Ac-
cording to guidelines given in ER the in-plane shear
flexibility of a diaphragm could be calculated using a)
expressions given for simple roofs, taking the depth
Applied Applied
of shear panel as the length of one pitched roof load load
slope. The horizontal shear flexibility is given by:
ch  c sec 2  (12)
b) Applied
load
where c = flexibility of shear panel calculated as
simple roof diaphragm; θ = angle of the rafter to the
horizontal line.
Slope of analyzed roof is equal to 5.7°. Thus:
sec 2 5.7  1.01 (13) Figure 10. Pitched roof portal frame deformations a) Sway un-
der horizontal load, b) Spread under vertical load.
According to fact that roof slope is nearly flat,
multiplication factor calculated in (13) is relatively
small. It is assumed that obtained in presented model 5 CALCULATIONS RESULTS
shear flexibility of diaphragm can be calculated as
horizontal flexibility. Calculations were successfully finished without any
warnings about instabilities and convergence alerts.
6000 Diaphragm deformation shape is shown in Fig-
ure 11. Maximum deflection at node where 15 kN
5 k f.5
force was applied is 18,80 mm. Relative value ob-
Bracing rod tained in test was 19,70 mm. Accuracy of model
5000

prediction is about 95%. Comparison between test

and theoretical analysis is presented in Figure 12.
4 k f.4
Figure 11 is an original graphic taken from static
calculation software without any modification but in
enlarged scale relative to construction dimensions.
5000

Truss upper chord Orientation of drawing is identical to Figure 9, thus

both can be qualitatively compared.
Test It can be easily seen that deflection of frame lo-
3 k f.3
Force cated in axis “5” (braced gable wall and roof) is sub-
stantially less than in axis “1” (opposite, unbraced
5000

Purlin gable wall and roof). This kind of cantilever-like be-

havior is due to fact, that braced construction region
2 k f.2 have higher stiffness than unbraced one. It is im-
portant to emphasize that frame “5” is not fully rig-
id, like it is necessary in standard procedures provid-
5000

Shear panels
ed by European Recommendation (1995). Its
flexibility is taken into account in this analysis.
1 k f.1 Shear panels acting as a separate sheets is also
Sliding support to allow recognized in the diaphragm deformation form. Ad-
in-plane movements jacent sheets are connected only with flexible con-
nectors which is the reason of this effect.
Figure 9. Geometry of diaphragm assembly - calculation model
Value of roof real deformation is known. Thus
bending moments in columns can be calculated us-
Figure 9 shows geometry of diaphragm assem-
ing multiplication factors (different for each frame
bly. Every member and connecting methods are de-
under consideration), which is obtained from ex-
scribed in Chapter 4.1. In natural scale tests, force is
pression:
applied to frame in axis “3”. Range of testing force
is from 1 kN to 15 kN. (with 1 kN step). Calculation s 
uc
(14)
are made in linear-elastic range. ub
The behavior of the pitched roof frame can be an-
alyzed as sway and spread structure (Figure 10). where uc = clad frame deflection obtained by numer-
Spread is not under consideration in this study. If it ical analysis; ub = bare frame deflection.
 Finally, bending moment in column in clad frame
can be calculated as follows:
M c  M ns   s  M s (15)
where Mns = no-sway bending moment in bare
frame; Ms = sway bending moment in bare frame.
Conception of this approach comes from ER
principles, where deflection and bending moment of
specific frame with sheeting can be computed with
using the reduction factors η obtained by reading it
from tables. It is mainly determined by two varia-
bles: relative flexibility (between diaphragm assem-
bly and frame stiffness) and frame position in the
system. In presented method it is no necessary to de-
fine it in that way.

6 SUMMARY AND CONCLUSIONS

As the reference to calculations results, accuracy of

model prediction is near 95%, what is very accurate.
This model could find practical use in steel buildings
static analysis for structural design or scientific stud-
ies.
Carried out experiments, numerical calculations
and its result, following conclusions could be listed
as follows:
1 Stressed skin diaphragm action effects has signif-
icant influence in structure deformation and in-
ternal forces. Reduction of sway tested and calcu-
lated equals 78% in presented industrial hall with
only roof being under consideration.
2 For non-common structures shapes some methods
and approaches can be find in literature. Authors
propose one solution with emphasis on practical
Figure 11. Diaphragm deformation shape under horizontal using.
force – model presented in enlarged scale. 3 Described approach is consistent with European
Recommendations (1995), which allow to use Fi-
nite Element Method.
4 Linear force-displacements characteristics for
connections are used. It is sufficient to analyze
structure behavior in elastic range. For more
complex solutions non-linear connector model
would be used e.g. to include contact defor-
mations with sheeting plate.
5 Analyzed roof has uniform shape (rectangular in
plane) and have no openings. Proposed method
could be applied in any shapes with various loca-
tions of openings.

ACKNOWLEDGEMENTS

Paper was created as part of the statutory work

S/WBiIŚ/2/2012.

Figure 12. Results of test and theoretical analysis.

REFERENCES

Atrek E. & Nilson A.H. 1976. Non-linear finite element analy-

sis of light gage steel shear diaphragms. New York: School
of Civil and Environmental Engineering at Cornell Univer-
sity.
Bakhdi F., Tremblay R. & Rogers C.A. 2012. Revisiting the
SDI and ECCS methods for in-plane shear flexibility of
metal roof deck diaphragms using 3D non-linear finite ele-
ment analysis. 15th World Conference on Earthquake En-
gineering Paper No. 2482. Lisbon.
Bródka J., Garncarek R. & Miłaczewski K. 1999. Blachy fał-
dowe w budownictwie stalowym. Warsaw: Arkady.
Davies J.M. 2006. Developments in stressed skin design. Thin-
Walled Structures 44 (2006) 1250–1260. Elsevier.
ECCS-TWG 7.5. 1995. European Recommendations for the
application of metal sheeting acting as a diaphragm. Euro-
pean Convention for Constructional Steelwork No. 88.
Dong Li 2006, The new SDI diaphragm design manual. Inter-
national Specialty Conference on Cold-Formed Steel Struc-
tures Paper 10: 437-452.
Garncarek R., Dziatkowski T. 2000. Obliczanie tarcz dacho-
wych z otworami. Konstrukcji stalowe 6(43): 45-47.
Gryniewicz M. & Szlendak J.K. In press. Modelowanie nume-
ryczne konstrukcji dachu hali stalowej z uwzględnieniem
współpracy poszycia. 61. Konferencja Naukowa KILiW
PAN oraz KN PZIiTB Krynica-Bydgoszcz.
Höglund T. 2002. Stabilisation by stressed skin diaphragm ac-
tion Publication 174. Stockholm: SBI.
Liedtke, P.E. & Sherman D.R. 1982. Comparison of methods
predicting diaphragm behavior. International Specialty
Conference on Cold-Formed Steel Structures Paper 1: 369-
389. St. Louis: Department of Civil Engineering University
of Missouri – Rolla
Lutrell, L.D. 2006. Steel deck institute diaphragm design man-
ual: third edition. AISI-Specifications for the Design of
Cold-Formed Steel Structural Members Paper 82.
Nilson, A. 1960. Shear Diaphragms of Light Gage Steel. Jour-
nal of Structural Division. ASCE 86:11: 111-139.
PN-EN 1993-1-3:2006 Eurokod 3 - Projektowanie konstrukcji
stalowych - Część 1-3: Reguły ogólne. Reguły uzupełniają-
ce dla konstrukcji z kształtowników i blach profilowanych
na zimno.
Szlendak, J.K. 2002. Design concept of pitched portal frame
covered by metal sheeting. Proceedings of the International
Colloquium on Stability and Ductility of Steel Structures,
Budapest, 293-300.

View publication stats