You are on page 1of 8

Advances in Steel Structures, Vol.

I
Chan, Teng and Chung (Eds.)
© 2002 Elsevier Science Ltd. All rights reserved. 303

DETERMINATION OF THE FACTORS OF SAFETY OF


STANDARD SCAFFOLD STRUCTURES

B Milojkovic\ R G Beale^ and M H R Godley^

'Halcrow Group, 44, Brook Green, Hammersmith, London, W6 7BY, formerly Oxford Brookes
University, UK
^Department of Civil Engineering & Construction Management, Oxford Brookes University, Oxford,
0X3 OBP UK

ABSTRACT

A three-dimensional, non-linear, elasto-plastic, finite element model of a scaffold structure typical of


small domestic construction was developed with the objective of determining the factors of safety
against collapse. All connections between standards and ledgers and between standards and transoms
were considered to be semi-rigid. The influence of different types of fault in the structure on factors of
safety was determined. The solution obtained fi-om the elasto-plastic analysis was compared with an
alternative approach which combined a linear elastic analysis and eigenvalue buckling with the use of
member buckling interaction formulae given in the European Standard ENV 1993-1-1. This alternative
analysis was used to generate tables of factors of safety. In a standard finite element program elastic
buckling load factors are normally calculated by increasing all the applied loads simultaneously. For
the purposes of this research, however, it was necessary to determine buckling load factors for
imposed load only, dead and wind loads being kept constant. A regression curve was fitted to a series
of buckling load factors for different imposed loads. This curve was used to enable the factors of safety
against collapse of the structure to be calculated. The performance of scaffold structures is affected by
faults in both the design and site erection stage and in the operational stage. A table was constructed
which showed the effects of combinations of faults enabling an understanding to be achieved of those
faults which may make a small scaffold collapse under normal wind and imposed load combinations.

KEYWORDS

Steel, scaffold, factors of safety, semi-rigid, structures, non-linear analysis

INTRODUCTION

During the six years fi-om 1986-1993 the UK Health and Safety Executive (HSE) investigated 1091
safety related incidents (MARCODE HSE database). In addition, during the same period there were
471 reported collapses out of an estimated 7.5 million scaffold erections giving a failure rate of 6
collapses per 10000 erections. Scaffold collapses were primarily caused by missing ties (28%),
overload (25%), faulty components (13%) and missing bracing (9%). The objectives of this research
were to examine the effects of faults in design and construction on the safety factors of a scaffold
which represented typical domestic construction (see Figure 1). The model scaffold was designed
according to BS 5973 (BSI 1994).
304

Figure 1: Model scaffold


Traditionally engineers have used simple methods, such as linear analyses, to analyse scaffolds for
practical design. However, because linear analyses do not include geometrically non-linear effects the
results obtained are not accurate even in the range of working loads (Chan et al, 1995; Chu et al, 1996;
Huang et al, 2000). The authors have previously demonstrated the importance of the semi-rigid
connections between ledgers and standards, and between transoms and standards. (Godley and Beale,
1997 and 2001) on the behaviour of the structure. However the normal eccentricity of the connection
in traditional tube and fitting scaffolds has been shown both experimentally and theoretically to have
little effect (Milojkovic et al, 1996).

STRUCTURAL ANALYSIS

The scaffold shown in Figure 1 was analysed using the Lusas finite element program (Lusas 1998). A
full elasto-plastic analysis was initially conducted of the structure. Convergence difficulties caused by
the wide variation of stiffiiesses within the model and long processing times (each analysis was taking
several days) occurred with this model. In all the analyses the diagonal brace failed in buckling before
the rest of the structure had achieved its maximum capacity. As the structure was still capable of
carrying load a reduced model without the brace was used to validate an alternative procedure. The full
details of the analysis are given in Milojkovic (1999).

The failure of the structure was determined using the interaction formulae for member buckling
proposed in ENV 1933-1-1 (1993). In this context collapse was defined as the load factor at which the
structure first failed to satisfy the ultimate limit state requirement.

Buckling Analysis

This approach was applied member by member and every member was treated in the same way by
combining a linear elastic analysis with an elastic buckling analysis. The axial forces and the bending
moments in the members were taken from the results of linear elastic analyses, while the non-
dimensional slendemess ratio was calculated by combining results from the linear elastic and elastic
buckling analyses. Normally an interaction formula is used to check a member for which the end
moments and axial forces, as well as the elastic buckling load are known. If this expression is less or
equal to unity, the check succeeds. At failure, this expression is equal to unity. In this approach the
305

load factor was adjusted until the interaction formula equalled unity. This load factor was taken to be
the load factor at failure.

The finite element program Lusas calculated buckling load factors by uniformly increasing all loads
together at the same time; dead load, wind load and imposed load where appropriate. However, these
elastic buckling load factors were not appropriate in this case because the purpose of this project was
to determine the safety factors against collapse when the imposed load alone was increased. This could
not be done directly so the following method was adopted:

If the imposed load only is increased by a factor /, the force in any member F, is given by:

F= F,+F^+rF, (1)

where FD is the force in a member due to dead load, Fw is the force due to the wind load and Fj is the
force due to the imposed load.

If a buckling analysis is made for any value of y then buckling occurs when the value of the elastic
buckling load Fcr is

K =K{Fo+F^+F,) (2)

Naturally, A^^ will depend on the value of y chosen. We need to find the value of x called y^^ for which

F.r=F,+F^+F, (3)
and this was done by calculating values of A^^ for different values of^, and noting the value of / = y^^ at
which at>^^^ = 1 . The calculated y^^is the elastic buckling load factor when only the imposed load is
increased, while the dead load and the wind are kept constant. Values of A^^ were calculated for a range
of values of/and for a load combination of dead load and a working load of 1.5kN/m^ in every bay on
the top level are shown in Figure 2. From this curve, the function relating ;Kand the critical load factors
was given by, using regression analysis:

A^^ (y) = 8.823 - 2.649/ + 0.4185/' +0.00309/^ + 0.0008519/' (4)

From this polynomial the critical load was determined when 4^ =1. In this case /^^ =12.25. A
different regression curve was found for all structures analysed. Values of the elastic buckling load
factors /^^ are given for all combinations in Table 1.

Interaction formula

When an axially loaded member is also subjected to bending moments, the total effect cannot simply
be computed by the direct addition of the corresponding values of the component cases. The reason for
this is that the deflections caused by the bending moment increase or "amplify" the secondary effects
produced by the axial compression, leading to a non-linear interaction. To account for this effect an
interaction formula was used to estimate the overall factor of safety, X, against collapse for this
structure (members in compression) in accordance with ENV 1993-1-1, clause 5.5.4:

XAf, W,J^, W,J^


306

Graph of the variation of the critical load


against
the coefficient of the imposed load

o 5 \
^ \
"O \
Z 3 \

0 ; --- -• - - -- '.
0 5 10 15
Coefficient of imposed load ^

Figure 2: Graph of the variation of the critical load against the coefficient of imposed load

where NSD is the axial force, My^sD and MZ^SD are the bending moments about the y and z axes in a
scaffold element. z4fv^^ ^^^ characteristic resistance load of the element in compression,
^v ^(^pify) ^^^ ^z l^pify) ^^^ ^^^ characteristic resistance of the element in compression about the;;
and z axes. The forces were taken fi-om linear analyses of scaffolds and amplified by the buckling
loads determined above.

The non-linear analysis of the structure showed that local failure of the top ledger elements occurred
before overall collapse. In order to investigate overall collapse the moment of resistance against
bending of ledger elements was arbitrarily increased 1000 fimes. Failure of the structure then occurred
by buckling failure parallel to the facade. The interaction formula equation (5) was then applied to the
twelve central bottom elements on the fi'ont of the scaffold.

LOAD CASES AND COMBINATIONS

The dead load was considered with three alternative imposed load conditions to give the following
load cases:

Load Case 1: Dead load + Working load of 1.5 k N W in every bay at the top level.
Load Case 2: Dead load + Working load of 1.5 k N W in alternate bays at the top level.
Load Case 3: Dead load + Overload on the top platform comprising two 300 kg loads acting on an area
of 0.96m X 0.46m at the centre of each and every span.

In addition to this, wind loading was considered acting normal to and parallel with the fa9ade. Thus
each of the load cases was considered in the following combinations:

Combination 1: vertical load condition with no wind.


Combination 2: vertical load condition with wind parallel to the fa9ade, blowingfi-omleft side to right.
Combination 3: vertical load condition with wind perpendicular to the fa9ade - sucking, assuming that
the fa9ade itself offered no wind resistance.
307

There were therefore initially a total of nine load case combinations.

However, when Load Case 2 was analysed it was discovered that all standards had higher overall load
factors against collapse than occurred in Load Case 1. This is because at the top level the load applied
in Load Case 2 is only half (applied on alternate bays) of that in Load Case 1. The ledgers, however,
have lower overall load factors in Load Case 2, because bending moments are greater. This increase in
bending moment caused local failure of the structure but did not lead to an overall collapse. In
addition, the faults which are introduced to the structure in this research do not have any influence on
this local failure of the top ledgers. As the aims of this research were to investigate overall failure and
not local failure Load Case 2 was neglected. This reduced the number of combinations to be analysed
to six.

STRUCTURAL FAULTS

Inspection by the HSE has identified the most common faults which can exist in scaffold structures
(Maitra, 1997). The faults were:

1 A partial settlement caused by the base-plate of the scaffold not being level. This fault induces a
bending moment into the bottom leg of the scaffold. This was modelled by inclining the bottom
leg of the scaffold by 2%.

2 A gross settlement caused by a member not being supported by the ground. For maximum effect
this was modelled by removing the support at the most heavily loaded base.

3 Initially the standards were regularly placed at 2.4m centres. In this case the overall width of the
scaffold was unaltered but spacing of the central two standards was changed to 2.1m and 2.7m.
The effect of this fault was to increase the strength of the scaffold. This is due to the fact that for
regularly spaced scaffolds the flexural rigidity provided by the ledgers is a minimum.

4-7 The middle standards were assumed to be out-of-plumb by 1% (fault 4), 2% (fault 5) parallel to
the facade and by 1% (fault 6) and 2% (fault 7) normal to the facade.

8-10 Curvature was applied to the standards below the bottom lift of a maximum of 3mm (fault 8),
6mm (fauh 9) and 12mm (fault 10).

11 The height to the first lift was increased to 2.7m. In order to keep the whole scaffold to the same
height the top lift was reduced to 1.15m.

12 To gauge the effects of corrosion one standard was reduced from 4mm thickness to 3.25mm.

13 Connections between transoms and ledges adjacent to standards were made with right-angle
couplers. In order to model the common fault of using putlog connections, pinned joints were
inserted at all these positions.

14 Ledger bracing was initially placed as close to standards as possible. In practice this is not
possible and in faulty structures the ledger bracing was placed 300mm vertically away from the
correct position. In addition, swivel connectors instead of right-angled couplers were also used.

15 The perfect structure had ledger braces at every level. Commonly, however, the bottom diagonal
is omitted. This was taken to be fault 15.

16-18The most common cause of failure is inadequate tying. The perfect structure had two ties on the
middle standard. Fault 16 - top tie omitted; fault 17 - bottom tie omitted, fault 18 - no ties.
308

TABLE 1
LOAD FACTORS FOR THE WHOLE STRUCTURE

Overall Load Factors for Whole Structure (/I)


r^
Fault Description Load Load Load Case 1 Load Case 3 1 %age 1
Case Case Combination Combination reduc-
1 3 LCI LC2 LC3 LCI LC2 LC3 tion
in (A)
- Perfect Structure 11.79 6.45 10.91 9.88 9.88 6.02 5.39 5.25 -
1 Partial 11.82 6.49 10.34 10.32 9.19 5.66 5.98 4.78 8.5%
settlement
2 Gross settlement 11.54 6.37 7.00 7.00 6.63 3.36 3.10 3.24 41.0%
Irregular 11.78 6.46 10.81 9.87 9.83 6.03 5.45 5.27 -0.5%
1 ^ standards
4 1% out-of- 11.92 6.44 10.41 9.35 9.31 5.60 5.10 4.80 8.0%
plumb parallel
5 2% out-of- 11.72 6.43 9.76 8.90 9.23 5.39 4.93 5.05 6.0%
plumb parallel
6 1% out-of- 11.92 6.49 9.96 8.95 8.74 5.30 4.84 4.47 15.0%
plumb normal
7 2% out-of- 11.80 6.42 9.24 8.45 8.85 4.90 4.40 4.60 16.0%
plumb normal
8 3mm curvature 11.71 6.34 10.24 9.37 9.11 5.52 4.96 4.56 13.0%
9 6mm curvature 11.72 6.37 9.61 9.15 8.03 5.28 4.94 4.13 21.5%
10 12mm curvature 11.74 6.39 8.61 8.23 6.74 4.71 4.45 3.35 36.0%
11 2.7m bottom lift 10.26 5.52 9.61 8.69 8.60 5.32 4.79 4.48 14.5%
12 Corrosion 11.73 6.37 10.84 9.83 9.83 5.97 5.41 5.19 1.0%
13 Putlog couplers 9.68 5.01 9.23 8.19 7.98 4.87 4.35 4.00 24.0%
14 Eccentric 11.42 6.10 10.55 9.55 9.15 5.59 5.12 4.76 9.5%
bracing
15 No bottom 11.63 6.21 10.48 9.49 9.29 5.40 4.91 5.02 6.0%
brace
16 No top tie 11.21 5.89 10.45 9.37 9.11 5.52 4.96 4.56 13.0%
17 No bottom tie 9.69 5.16 9.07 7.66 8.03 4.92 4.16 4.14 21.0%
18 No ties 9.48 5.06 9.02 7.62 7.87 4.86 3.53 3.70 29.5%

RESULTS OF ANALYSES WITH SINGLE FAULTS

Table 1 is a summary of all the analyses. The lowest buckling load factors for Load Cases 1 and 3 are
given. These always occurred for Load Combination 3 (called LC3). The buckling load factors for
Load Combinations 1 (LCI) and 2 (LC2) were less than 5.6% higher for Load Case 1 and 9% higher
for Load Case 3. Full details are found in Miljokovic (1999). It can clearly be seen that the most
significant reductions in load factor occurred for the cases of Gross Settlement (Fault 2), excessive
curvature in the lowest standards (Fault 10), inadequate ties (Faults 17 and 18) and the incorrect use of
putlog couplers instead of right angled couplers (Fault 13).

COMBINED FAULTS

In practice when scaffolds are designed and erected single faults do not arise on their own. Analyses
were therefore firstly undertaken with combinations of faults. In most cases the reduction in capacity
of a set of faults could approximately be estimated by adding the reductions in load factor given by the
single faults in Table 1 above. For example, a combination of putlog couplers (instead of right angled
couplers) in combination with no ties (i.e. Faults 13+18) gave a reduction of 59%. The sum of the two
309

faults is 54.5%. The major exception occurred in the case of putlog couplers in combination with gross
settlement. In this case the analysis produced a reduction of 28% instead of the combined total of 65%.
The effect of putlog couplers is to change the structure from one containing semi-rigid connections to a
structure with pinned joints only. This therefore prevents moment transfer between adjacent elements
in a frame. When the interaction formulae are used to determine collapse the additional moments
transferred from the frame farthest from the fa9ade precipitate failure in the most heavily loaded
member. When putlog couplers are used these moments are not transferred. The most heavily loaded
members in this case are those adjacent to the 'settled' support in the same plane frame. They have
higher axial loads than members with right angle couplers but much lower moments. The use of the
interaction formulae produces a higher collapse load.

A typical scaffold is influenced by two important stages in its lifetime. The first of these stages is the
design and construction stage. In this stage, decisions are first made on paper about the specification
required to meet a defined set of purposes. Then the scaffold is erected in accordance (or not) with this
specification. The design and erection stage of the process may be carried out by experts, and would
then perform as intended, or it may be carried out less well, so that if the fiill demands specified for the
structure were actually asked of it, the scaffold would not perform to specification. The second stage in
the life of the scaffold occurs once it has been erected on site, when its performance is influenced by
the quality of the site control. Site control is good if the scaffold is correctly used, so that the imposed
loading never exceeds that specified for it, and if it is conscientiously inspected and serviced at regular
intervals so that no significant deterioration takes places during its life.

Both stages can be subdivided into three categories: good, average and poor according to the faults
which are present. Some faults may be generated in both stages of the life of the structure while others
may be generated in one or the other. For instance, ties may be missed by the designer or the erector,
or they may be removed while the scaffold is in use and, because of poor site control, not replaced.
This is a fault possible in both stages. Table 2 summarises the combinations of faults assigned to each
category. Note that overload. Load Case 3, should not occur in practice and has hence only been
assigned to faults in site control. Common geometric faults such as standards out-of-plumb by 1% or
with 6mm curvature and partial settlement have been assigned to average categories. Gross settlement,
or extreme errors of curvature (12mm), or large out-of-plumb (2%) have been assigned into the
category of poor site control. The use of putlog couplers was considered to be an element of poor
design or poor construction. Joint eccentricity usually arises due to poor design. The absence of ledger
braces was considered to belong to poor erection/site control as it should be noticed and corrected.

TABLE 2
FAULT COMBINATIONS

Site Control
Design/Erection Good Average Poor
Good - 17+LC3 15+18+LC3
Average 1+4/5+9+11+13+15 1+4/5+9+11+13+17+ 1+4/5+9+11+13+15+
LC3 18+LC3
Poor 2+3+6/7+10+11+12+ 2+3+6/7+10+11+12+ 2+3+6/7+10+11+12+
13+14+18 13+14+18+LC3 13+14+15+18+LC3

The resulting load factors are given in Table 3. The table indicates poor site control is potentially more
detrimental to the safety of scaffolds than poor design. Each combination has been analysed in the
same way as the single faults. It is interesting to note that the extreme case of poor design in
conjunction with poor site control led to a negative safety factor. In this case, if all the faults were
present, it would not be possible to erect the scaffold. In practice, not all the faults occur and so such
scaffolds are constructed but with very little margin of safety. Although the analyses undertaken in
310

this paper have dealt with only one scaffold it is thought that the principles and results obtained are
applicable to scaffolds in general; particularly, the absence of ties and use of putlog connectors.

TABLE 3
LOAD FACTORS FOR DIFFERENT SITE DESIGN/SITE CONTROL COMBINATIONS

Site Control
Design/Erection Good Average Poor
Good 9.88 4.14 0.98
Average 3.88 1.79 0.01
Poor 1.19 0.51 -0.18

CONCLUSIONS

The paper has described the analysis of a scaffold system representing typical domestic construction in
the UK. The derivation of the regression equation enabling the buckling load factor of a scaffold
against imposed loads only when static and wind loads are first applied is given and the use of the
interaction formula given in Eurocode 3 elaborated.

The effects of individual faults on the safety of the scaffold have been determined. A table of
combined faults has been drawn up to illustrate the relationship between design/erection and site
control has been evaluated. It can be seen that site control has a greater influence on safety than the
design of scaffolds.

REFERENCES

BS 5973: (1994). Code of Practice for Access and Working Scaffold and Special Scaffold Structures in
Steel, BSI, London
Chan W.F., Zhou, Z.H., Chen W.F., Peng, J.L. and Pan A.D. (1995), Stability Analysis of Semi-rigid
Steel Scaffolding, Engineering Structures, 17,568-574
Chu A.Y.T, Zhou Z.H., Koon, Z.H., Chan S.L., Peng, J.L. and Pan A.D. (1996), Design of Steel
Scaffolding using an Integrated Design and Analysis Approach, in Advances in Steel Structures, Hong
Kong 245-250
Godley M.H.R and Beale R.G. (1997), Sway Stability of scaffolding structures, The Struct. Eng
75:1:4-12
ENV 1993-1-1: 1992. (1993). Eurocode 3. Design of Steel Structures: Part 1.1: General Rules for
Buildings, BSI, London
Godley M.H.R and Beale R.G. (2001), Analysis of Large Proprietary Scaffold structures, Struct and
Buildings. 146:1:31-39
Huang, Y.L, Chen, H.J, Rosowsky, D.V and Kao, Y.G. (2000), Load-carrying capacities and failure
modes of scaffold shoring systems. Part 1: Modelling and experiments. Struct. Eng & Mech. 10:53-66
Lusas User Manual Version 12.2 (1998). FEA Ltd, UK
Maitra, A. (1997). Accidents Associated with Scaffolding in the UK, Technical seminar. Design and
use of Temporary Structures, Institution of Civil Engineers, London
Milojkovic, B. (1999), Factors of Safety for standard scaffold structures, PhD Thesis, Oxford Brookes
University
Milojkovic, B., Beale, R.G. and Godley, M.H.R. (1996), Modelling Scaffold Connections, in Proc 4*
ACME UK An Conf, Glasgow, 85-88

You might also like