You are on page 1of 7

Behaviour of Roofing Sheet Systems under Static Uplift Loads

Dr S Narayanan, Associate Member


Dr M S Mathews, Non-member

Roofing sheets of industrial sheds have been observed to fail in uplift under extreme wind conditions as in a cyclone.
This paper presents the research carried out to study the behaviour under static uplift loads of asbestos cement and
galvanised iron corrugated roofing sheet systems used in India using gravity load method. About 250 full-scale tests
were carried out on roofing sheet systems with different connections for a span/purlin spacing of 1.4 m. The behaviour
of the systems as well as the ultimate loads is reported and analysed. Tests were also conducted to study the effect of
reduced number of centre support bolts and the purlin type on the load-carrying capacity of the roofing sheet system.
Small-scale models were also used to model the uplift behaviour of the roofing sheet systems at the connections and study
the effect of dimensions of washers and types of bolts on the behaviour of roofing sheet system.

Keywords: Windspeed; Roof; Cladding; Damage; Connection

INTRODUCTION standards of various countries are given in Figure 2. The


The most common roof claddings used in industrial sheds are standard distance over which the roofing sheet is supported
asbestos cement (AC) and corrugated galvanised iron (GI) (the purlin to purlin distance) is 1.4 m. The pitch of the roof is
roofing sheets. The post-disaster cyclone damage assessment of normally 18o . The roofing sheets are normally connected
the Chennai cyclone (November, 1994) highlighted the using J-bolts, conical metal washers and nuts, made watertight
damages suffered by the industrial sheds like (1) breaking of by the use of bituminous felt washers. While the standard
roofing sheets (limited to a few locations like ridges and eaves); thickness of AC sheets is 6 mm, the GI sheets are available in
(2) deformation of bolts that hold down the sheets; and (3) bolt
different gauges, the commonly used thicknesses being 24 gauge,
shearing through roofing sheets1. The roofing sheet systems
26 gauge and 28 gauge (0.60 mm, 0.50 mm and 0.40 mm,
have performed well under normal conditions of load where
dead load and live load are the deciding factors. However, respectively).
their performance under cyclonic wind loads has been
Metallic and Bitumen Washers
disastrous. The predominant failure mode under such
conditions is uplift. No study has been reported on the Different sizes of washers are available (Figure 3(a) and Table 1).
behaviour of roofing sheet systems under uplift loads. While MW-1 to MW-5 are conical washers, MW-6 is a flat
washer. Among the washers, MW-1 and MW-2 had thickness
LITERATURE REVIEW less than the minimum specified by IS 8869-19786, which is 0.9 mm
The different types of roofing materials used in industrial while MW-3, MW-4 and MW-5, had thickness greater than the
sheds and their connections are discussed here. specified minimum. Washer type MW-3 has been used
throughout the test programme. The flat washer had an
Roof Cladding Materials and Roof Cladding Connections average thickness of 1.5 mm while IS 8869-1978 specifies a
thickness of 1.6 mm.
Figure 1 shows a portion of the roof giving details of the con-
nections of AC roofing sheets to the purlins as recommended The average thickness of the available bitumen washer was
by IS 3007-19642. At each end of the roofing sheet, two bolts 2.6 mm and diameter 23.4 mm (Figure 3(b)). The inner
have been used to connect the sheet to the purlin and this is
Table 1 Dimensions of metal washers
called a ‘two bolts per purlin connection’ (2A). The other
connections recommended by various agencies are Type of Notation Outer Inner Height, Thickness,
washer used diameter, diameter, mm mm
summarised in Figure 23-5. There is no provision in the Indian
mm mm
Standards for connection of GI roofing sheets. The
MW-1 22.90 8.50 3.60 0.40
recommendations of the various organisations and the
MW-2 24.94 8.32 7.64 0.80
Dr S Narayanan is with Department of Civil Engineering, TKM College of Conical MW-3 30.40 8.10 8.60 1.00
Engineering, Kollam, 691 005; while Dr M S Mathews is with Department washer MW-4 34.60 8.14 9.20 1.20
of Civil Engineering, IIT Madras, Chennai 600 036. MW-5 38.00 8.24 9.50 0.66
This paper (revised) was received on November 26, 2002. Written discussion
on the paper will be entertained till February 29, 2004. Flat washer MW-6 26.00 8.38 — 1.50

216 IE (I) Journal—CV


Figure 1 Details of connecting the roofing sheets to purlins (asbestos cement roofing sheets)

(c) The arrangement, positions and number of bolts per


sheet that may depend on the location of the sheet on
the roof.
(d) Effectiveness of the washers.

STATIC UPLIFT TESTS ON ROOFING SHEET


SYSTEMS
The provisions of IS: 5913-1970 are inadequate for testing roofing
sheets in uplift10. BS 5950 (Part 6)-199511 and ENV 1993-1-3:199612
contain provisions for uplift test on roofing sheets. Gravity
Figure 2 Connections used for fixing AC and GI roofing sheets loading may be used to simulate uplift load by inverting the
sheet. In this method, the roofing assembly is set upside down
diameter is provided to suit the diameter of the bolts. Bitumen on a test bed specially prepared for this work, with a provision
washers generally satisfy the specifications of IS 8869-19786. of supporting three purlins at a spacing of 1.4 m (Figure 4). The
purlins are rigidly fixed to the supports. The troughs of the
Bolts corrugated sheets are first filled uniformly with sand of known
weight, and then loaded with layers of brick. This simulates
Though different types of bolts ( J or hook-bolts, L and U-bolts)
uniform uplift loading which can be evaluated without any
are recommended for use by IS:730-19787, the U-bolt is not
approximation. The deflections at the centre of the span have
generally used (Figure 3(c)). They have also not been used in
been measured with a specially fabricated deflectometer
this research programme. IS:1367 (Part 3) 1979 specifies that
having an accuracy of 0.1 mm. The tests have been carried out
the bolts shall be manufactured from low or medium carbon
on AC and GI sheets for the connections given in Figure 2 and
steels8. The chemical composition of the material of the bolt
for different thicknesses of sheets.
was determined using vacuum emission spectrometry. The
bolts satisfy the specifications. The failure criteria is taken as any one of the following:
IS:3007-1964 specifies that the fixing bolts for AC sheets shall (i) Local failure and pull-through of the bolt.
be 8 mm or more in diameter2. The British Steel Producers (ii) Failure by bending or dimpling of crests.
Conference (1966) also recommends the use of 8 mm bolts for
GI sheet roofing9. However, only 6.8 mm diameter bolts are (iii) Tearing of the sheet at the edges.
available and these have been used in the tests. (iv) Failure of bolts and washers.
The aspects studied during the course of the investigation
were: Load Deflection Behaviour
Two types of the deflection measurements were made, namely
(a) The bending capacity of the roofing sheet system.
central deflections and the transverse cross sectional
(b) Proper size and type of bolts. deflections of the sheets at midspan and at supports (Figure 5).

Vol 84, November 2003 217


Figure 3 Accessories for connections

The observed deflections of the roofing sheets in wind uplift Analysis of Failure Patterns
were larger than the theoretical deflections assuming that the The transverse flexural strength of the AC sheet is very low.
roofing sheets are simply supported at both ends and acted This is clearly reflected in the patterns of failure observed. AC
upon by dead load and live load and using the equation sheets with connections 2A, 2B and 3A fail in transverse
(deflection = 5 wl4/(384EI)). For roofing sheet with theoretical bending. Cracks form along one of the centre corrugations or
fixed support, the central deflection is calculated using the along the corrugation containing the bolt connection and
formula (deflection = wl4/(384EI)). The roofing sheet in wind failure occurs by extension of these cracks. Similar failures are
uplift loses contact with purlin and is supported by a few observed in two-span roofing sheets with connections 2A, 2B
numbers of bolts and hence, they are less rigidly connected and 3A. This type of failure is not seen in roofing sheet with
than the simply supported case under dead load and live load connections 3B and 4. They fail by cracking perpendicular to
where it is fully supported by the purlin. In case of GI roofing the corrugations. The failures are of brittle nature and sudden.
sheet, the distortion in geometry and the large transverse
bending has also contributed to the larger deflection. The asbestos fibres are generally oriented along the
longitudinal direction. The corrugations too impart
For the parts of the eastern coast of India prone to cyclones, longitudinal bending strength. The absence of both these
IS 875 (Part 3)-1987 recommends a wind speed of 50 m/s. For factors results in a low bending strength in the transverse
these regions the service wind load can be calculated as 1400 N. direction. Thus, in order to ensure longitudinal bending
Referring to Figure 5(a), the deflection for AC sheet with action, provision of more bolts may be necessary.
connection 2A under this load is 8 mm and still lesser for
roofing sheet with other connections. The limiting deflections GI sheets exhibit high tensile strength in both directions.
are 9.3 mm (British Steel Producers Conference, 1960)9 or 15.6 mm However, they undergo large deflections and fail due to
(AS 1562)3 (Table 2). Thus, the deflection of AC sheets at buckling. The sheets also undergo local deformations at the
service loads is very small compared to the limiting deflection bolt points due to load concentration. This is substantially less
at service loads (Table 2). For GI roofing sheets, the deflection when the numbers of bolts are increased.
ranges from 2.5 to 8.5 times the values observed in AC sheets.
Ultimate Loads
This is due to the lower flexural rigidity (EI) of the GI roofing
sheets (Table 3). The deflections of the 0.4 mm and 0.5 mm The uplift capacities determined from tests on AC and GI
thick GI sheets are very large at service loads (approximately roofing sheets of different thickness and types of connections
18 mm for a load of 1400 N as seen in Figures 5(c) and 5(d)) and are given in Table 4. Each load value in the table is an average
hence, are to be avoided in cyclone zones (to prevent water of three test results. No closed form equation is available to
intrusion). predict the deflection for roofing sheet systems with different

218 IE (I) Journal—CV


Table 2 Limiting deflection for wind uplift Effect of Reduced Number of Centre Support Bolts
There is always the chance of leaving out one or more centre
Authority Limiting deflection Limiting deflection at service
support bolts when the sheets are laid over three supports
(formula) loads for 1.4 m span, mm
(purlins) on the roof. Tests were done to evaluate the
AS1562; CP 143 reduction in strength of the roofing sheet when one or more of
(Part 10) 1973; BS span 15.6 the centre support bolts are omitted.
5950 Part 6 1995 90 Table 5 shows the uplift capacities for the different cases. In
each case, the end supports of the two span roofing sheets had
British Steel Producers span 9.3
the full number of bolts while the middle had lesser number of
Conference (1960) 150 bolts. Reducing the number of centre support bolts in the two
span roofing sheet systems drastically reduces the load
Table 3 Flexural rigidities of roofing sheets carrying capacities of both AC and GI roofing sheets. In the
case of AC sheets, the reduction is capacity due to omission of
Roofing sheet Thickness, mm Flexural rigidity (EI)
one bolt is 25% (in case of four bolts per purlin case) and 42%
(1010 Nmm2)
in case of three bolts per purlin case. In the case of GI roofing
AC 6.0 7.0020 sheets, reduction of centre support bolts results in large
GI 0.4 0.3275
deflections and lesser load capacities. The increase in
deflection is very marked in connections with three bolts per
GI 0.5 0.4090 purlin on the end purlin and two bolts at the middle support.
GI 0.6 0.5540 For double span AC and GI sheets tested in uplift, the
omission of centre support bolts results in large unsupported
span in longitudinal and transverse directions (since, in uplift
the purlins do not provide support to the roofing sheet). This
explains the large reduction in strength of AC and GI roofing
sheet systems when the centre support bolt is omitted.

Effect of Using Different Types of Purlins


These tests were conducted to determine whether the type of
purlin affects the load carrying capacity of the bolts. Angle,
channel and pipe purlins were used (Figure 3(c)). The tests
were done on double span AC roofing sheets with 3B
connection (Figure 2). The failure load for each case is shown
in Table 6 (column 2).
The corresponding load on the critical central support bolt of
the centre purlin is determined (column 3). This is compared
with the bolt opening out load obtained by testing bolts
separately (column 5).
The load on the critical middle bolt of the central support
(column 3, Table 6) calculated from the failure loads in full-
scale test on roofing sheet (column 2) supported on channel or
pipe purlin is higher than the bolt opening load obtained by
testing bolt alone (column 5 and Figure 6(b)). This explains the
failure of the centre support bolt by opening out in the full-
scale tests. Where angle purlins were used, the strength of the
bolt being larger, the AC sheet fails before the bolt opens out.
It is also clear that the use of channel or pipe purlin reduces the
load carrying capacity of roofing sheet since, the bolt opens
out at lower loads.
USING SMALL-SCALE MODELS

Figure 4 Static uplift load tests on roofing sheets


Modelling Bolt to Roofing Sheet Region
Small-scale model testing has been carried out to study the
connections. However, the deflections for the different localised failure of GI roofing sheet around the bolt holes of
configurations can be determined by a finite element analysis. roofing sheets. The small-scale model includes a small area
This has not been attempted in this work. around the critical central support bolt. These models can be

Vol 84, November 2003 219


(a) Single span AC corrugated roofing sheet; (b) Double span AC corrugated roofing sheet; (c) Single span GI roofing sheet t = 0.4 mm; (d) Single span GI roofing sheet t
= 0.5 mm; (e) Single span GI roofing sheet t = 0.6 mm; (f) Double span GI roofing sheet t = 0.4 mm; (g) Double span GI roofing sheet t = 0.5 mm; (h) Double span GI
roofing sheet t = 0.6 mm; (i) Transverse deflection between the bolts at support of single span GI roofing sheet (thickness 0.40 mm) with 3 bolts per purlin connection;
(j) Transverse deflection at centre of single span GI roofing sheet (thickness 0.40 mm) with 3 bolts per purlin connection; (k) Transverse deflection between the bolts at
support of single span GI roofing sheet (thickness 0.40 mm) with 6 bolts per purlin connection; and (l) Transverse deflection at centre of single span GI roofing sheet (thickness
0.40 mm) with 6 bolts per purlin connection
Figure 5 Experimental load deflection behaviour of roofing sheets systems

220 IE (I) Journal—CV


Table 4 Failure loads (N/m2) for a purlin spacing of 1.4 m

Roof Thickness Type Different bolt connections for roofing


sheet of sheet, of 2A 2B 3A 3B 4 5 6
mm span

AC 6 Single 1816.0 2425.0 3462.5 5341.8 4838.0 * *


sheet Double 1494.7 2110.0 3492.0 6878.0 5859.7 * *

28 gauge Single * * 3662.0 * 3931.0 4055.2 4745.0


(0.4 mm) Double * * 4684.5 * 6119.0 5937.0 5565.0
GI 26 gauge Single * * 5480.5 * 5504.0 5588.7 5559.0
sheet (0.5 mm) Double * * 5633.1 * 5294.8 4509.3 4821.2
24 gauge Single * * 6716.5 * 7235.0 9210.0 ND
(0.63 mm) Double * * 7165.7 * 8820.5 7696.4 7799.9

Note : * denotes that the test configuration is not applicable; and ND denotes test not done

Table 5 Ultimate capacity (N/m2) of AC roofing sheet system with replaced by a channel for testing L-bolts or by a pipe for testing
reduced number of centre support bolts the capacity of pipe bolts.
Bolts per Number of bolts at the centre purlin Effect of the Size of Metal Washer
purlin at the 0 1 2 3 4
end supports The effect of washer size or absence of washer in the
connection was studied by conducting tests on small-scale
2 691.4 1758.9 1494.7 — — models of GI roofing sheet (Figure 6(a)). The pull-through
3 1286.5 2194.5 2016.5 3492.0 — loads on small-scale models using different types of washers
for different thicknesses of the roofing sheets are given in
4 1501.2 2378.5 2016.5 4391.1 5859.7
Table 7.

Table 6 Failure loads for two-span AC sheets fixed to different types of The pull-through loads decrease with a decrease in the
purlins thickness of the roofing sheet, the outer diameter and
thickness of the metal washer. The thickness of the washer is a
Type of Failure load Load on Type of Opening very important factor in its performance. The 1.0 mm thick
purlin on the the critical failure in out load washer (MW-3) underwent very little deformation at much
sheet, middle bolt full-scale for bolt higher loads (column 2 of Table 7). The 0.4 mm thick washer
N/m2 of the centre test from (MW-1) easily flattened out and pulled through the sheet
support at small resulting in a much lower capacity. The least pull-through
failure from scale capacity is in the case where the connection is done without a
full-scale test, N metal washer (column 5 of Table 7). The dimensions of the
test, N conical metallic washer recommended by the Indian
Angle 65 × 65 8373.7 4414.8 AC sheet failure 6240 Standards are satisfactory since, MW-3 has performed
relatively well. In AC roofing sheets, such tests could not be
Channel 100 × 50 7049.3 3716.5 L-bolt opens 3170 carried out due to difficulties in making the small-scale model.
Pipe 76.2 mm
diameter 2850.0 1502.5 Pipe bolt opens 925 GENERAL GUIDELINES

Based on the observations made during the tests, some general


used to evaluate the pull-through load. The size of the model
guidelines to be followed while using AC and GI roofing sheet
chosen for the current test is 15 cm × 15 cm, which
system are given below:
corresponds to the case of bolts on alternate corrugations (or at
15 cm centres). The roofing model sheet is fastened to a small 1. Avoid over tightening of the bolts while fixing the
rectangular frame, which in turn is fastened to the lower AC sheet since, it causes cracks in the AC sheets in the
platten of a tensile testing machine (Figure 6(a)). The bolt fixed vicinity of the bolts leading to failure at low loads.
on the roofing sheet is free to move vertically. By pulling the
2. Use good quality metallic and bitumen washers as per
bolt upwards the pull-through load of the roofing sheet can be
relevant codes of practice. Lower quality washers
assessed.
reduce the strength of the roofing sheet system.
Modelling Purlin to Bolt Region 3. Great care is required in drilling the boltholes. The
This set-up shown in Figure 6(b) is used to evaluate the holes for fixing AC sheets must be drilled and not
capacity of the bolt. In the frame shown, the angle can be punched.

Vol 84, November 2003 221


Table 7 Pull-through loads (N) of GI sheet with different washers

Thickness of Size of washers (in order of decreasing thickness)


the sheet, mm MW3 MW2 MW1 No washer

0.6 6050 5750 4275 4150

0.5 6080 5025 3567 2800

0.4 6000 4600 3200 2775

CONCLUSIONS
Uplift deflections observed in AC roofing sheets under an
uplift force is very small at working loads. The failure at
ultimate load is brittle and sudden in nature. The GI roofing
sheet systems exhibit very large deflections under uplift force.
Use of thicker sheets (0.6 mm) is recommended. The omission
of central support bolts results in a large reduction in the load
capacities. The use of channel and pipe purlins reduces the
capacities of the roofing sheet systems.
ACKNOWLEDGEMENTS
The financial support provided by the All India Council of
Technical Education and the material support provided by the
Ramco Industries (India) is acknowledged gratefully. The
authors are also grateful to IIT Madras for providing all the
facilities for carrying out the experimental work.
REFERENCES
1. S Narayanan and M S Mathews. ‘Behaviour of Roofs of Industrial Structures
to Cyclonic Wind Loads.’ Proceedings of International Conference on Disasters
and Mitigation, January 19-22, 1996, pp B1.68-B1.73.
2. IS 3007-1964. ‘Code of Practice for Laying of AC Sheets.’ Bureau of Indian
Standards, New Delhi.
3. AS 1562-1992. ‘Design and Installation of Sheet Roof and Wall Claddings,
Part 1: Metal.’ SAA Standards Australia, Sydney, Australia.
4. CP 143: Part: 10 :1973. ‘Code of Practice for Sheet Roofing and Wall
Coverings: Galvanised Steel.’ British Standards Institution.
5. S Narayanan. ‘Improving Cyclone Resistant Characteristics of Roof
Cladding of Industrial Sheds.’ PhD Thesis, Department of Civil Engineering, IIT
Madras, 1999.
6. IS 8869-1978. ‘Specifications for Washers for Corrugated Sheet Roofing.’
Bureau of Indian Standards, New Delhi.
7. IS 730-1978. ‘Specifications for Hook Bolts for Corrugated Sheet Roofing.’
Bureau of Indian Standards, New Delhi.
8. IS 1367:Part 3:1979. ‘Mechanical Properties and Test Methods for Bolts,
Screws and Studs with Full Weldability.’ Bureau of Indian Standards, New
Delhi.
9. British Steel Producers Conference. ‘Steel Designers’ Manual.’ ELBS and
Crosby Lockwood and Sons Ltd, London, 1966, pp 887-901.
10. IS 5913-1970. ‘Methods of Test for AC Products.’ Bureau of Indian
Standards, New Delhi.
11. BS 5950 Part 6 1995. ‘Structural use of Steelwork in Buildings: COP for
Design of Light Gauge Steel Sheeting, British Standards Institution.

(a) Modelling bolt to roofing sheet region; and (b) Modelling purlin to 12. ENV 1993-1-3:1996 Eurocode 3: Design of Steel Structures — Part 1:3
bolt region General Rules — Supplementary Rules for Cold-formed Thin Gauge Members
and Sheeting.
Figure 6 Modelling of the connection

222 IE (I) Journal—CV

You might also like