Www.sefindia.org

Reliability based codification for the design of overhead
travelling crane support structures
by
Juliet Sheryl Dymond
Dissertation presented for the degree of
Doctor of Philosophy in Engineering

at the University of Stellenbosch
Department of Civil Engineering

University of Stellenbosch
Private Bag X1, 7602, Matieland, South Africa
Promoters:
Prof P.E. Dunaiski
Prof J.V. Retief
December 2005
Declaration
I, the undersigned, hereby declare that the work contained in this dissertation
is my own original work and that I have not previously in its entirety or in
part submitted it at any university for a degree.
Signature: . . . . . . . . . . . . . . . . . . . . . . . . . . .
J.S. Dymond
Date: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
iii
Synopsis

J.S. Dymond
Department of Civil Engineering
University of Stellenbosch
Private Bag X1, 7602, Matieland, South Africa
Dissertation: PhD (Engineering)

December 2005
Electric overhead travelling bridge cranes are an integral part of many indus-
trial processes, where they are used for moving loads around the industrial
area.
Codes of practice on loadings on buildings provide load models for the
calculation of the vertical and horizontal loads that cranes impose on their
support structures. The crane load models in the South African loading code,
SABS 0160:1989 [1], are over-simplistic, therefore it is currently under consid-
eration to adopt the crane load models from the Eurocode crane loading code,
prEN 1991-3 [2] into the updated South African loading code, SANS 10160 [3].
There is no reliability basis for the partial load factor applied to crane loads
in SABS 0160:1989.
This dissertation presents an investigation into electric overhead travel-
ling crane support structures, focussing on the crane load models from prEN
1991-3. The investigation takes the form of a code calibration in two parts:
calibration to current practice and reliability calibration.
The aims of the calibration to current practice were to investigate the
load models from prEN 1991-3 to determine their suitability for inclusion into
the proposed SANS 10160 and to assess the effect on the cost of the support
v
vi Synopsis
structure and the design effort required, of calculating crane loads using the
load models from prEN 1991-3 rather than SABS 0160:1989.
The aims of the reliability calibration were to investigate the current level
of reliability of crane support structures designed using crane loads calculated
from prEN 1991-3 and SABS 0160:1989 and, if necessary, to determine partial
load factors required to achieve a consistent, minimum level of reliability.
The calibration process was carried out on three representative cranes and
their support structures.
Statistical models for the hoistload lifted by cranes and the modelling un-
certainties in the calculation of the wheel loads were developed for use in the
reliability calibration.
It was found that the current level of reliability was inadequate and partial
load factors were determined, for ultimate limit state, accidental limit state
and fatigue, to achieve consistent, selected target levels of reliability.
Samevatting

J.S. Dymond
Departement Siviele Ingenieurswese
Universiteit van Stellenbosch
Privaatsak X1, 7602, Matieland, Suid-Afrika
Proefskrif: PhD (Ingenieurswese )

Desember 2005
Elektriese oorhoofse brugkrane vorm ’n ge-integreerde deel van baie nywer-

heidsprosesse, waar dit gebruik word om swaar laste in die nywerheidsaanleg
te verskuif.
Praktykkodes vir belastings op geboue gee lasmodelle vir die berekening
van vertikale en horisontale laste wat van die kraan na die ondersteunings-
struktuur oorgedra word. Die kraan-lasmodelle in die huidige Suid-Afrikaanse
laskode, SABS 0160:1989 [1], is oorvereenvoudig en daarom word dit oorweeg
om die kraan-lasmodelle van die Eurocode, prEN 1991-3 [2], in die nuwe Suid-
Afrikaanse laskode, SANS 10160 [3], op te neem. Daar bestaan ook geen
betroubaarheidsbasis vir die huidige parsiële lasfaktore vir kraanlaste in SABS
0160:1989 nie.
Hierdie proefskrif bespreek die navorsing oor ondersteuningsstrukture vir
elektriese oorhoofse krane, met die klem op die lasmodelle van prEN 1991-3.
Die navorsing neem die vorm aan van ’n tweeledige kodekalibrasie : kalibrasie
gerig op huidige praktyk en ’n betroubaarheidskalibrasie.
Die doel van die kalibrasie gerig op huidige praktyk is om die lasmodelle
van prEN 1991-3 te ondersoek, die geskiktheid daarvan vir die Suid-Afrikaanse
kode te bepaal en dit met die huidige lasmodelle in SABS 0160:1989 te vergelyk.
vii
viii Samevatting
Dit sluit ook die effek daarvan op die koste van ondersteuningsstrukture en die
omvang van die ontwerpwerk in.
Die doel van die betroubaarheids gebaseerde kalibrasie is om die huidige
vlak van betroubaarheid van ondersteuningsstrukture vir elektriese oorhoofse
krane wat volgens die kraanlaste van prEN 1991-3 en SABS 0160:1989 ont-
werp is, te bepaal, en indien nodig, parsiële lasfaktore te bepaal wat lei tot ’n
volhoubare minimum vlak van betroubaarheid.
Die kalibrasieproses is uitgevoer op drie verteenwoordigende krane en onder-
steuningsstrukture.
Statistiese modelle vir die laste wat deur die krane gehys word en die
modeleringsonsekerhede vir die bepaling van die kraanwiellaste is ontwikkel
vir die gebruik in die betroubaarheidskalibrasie.
Daar is gevind dat die huidige vlak van betroubaarheid ontoereikend is.
Parsiële lasfaktore wat ’n gekose vlak van betroubaarheid verseker is bepaal
vir die grenstoestand van swigting, die grenstoestand vir ongelukslaste en die
grenstoestand van vermoeidheid.
Acknowledgements
I would like to thank the following people for the various ways in which they
have assisted me during the course of my PhD:
• My supervisors, Prof PE Dunaiski and Prof JV Retief for their academic

guidance during the course of this study
• Prof M Holický from the Klokner institute at the Czech Technical Uni-
versity in Prague for his assistance and guidance with my study during
my two month stay in Prague
• The following crane manufacturers and operators for their time and as-
sistance in the collection of data: Coen Lubbe, Coen Jacobs, Trevor
Graham, Derek Lidston, Drikus Stander, Mark Walter and the late Tom
Reynolds.
• My parents for the support they have given me during the course of all
my studies
• My friend Michele van Rooyen, for all her encouragement
• My husband, Jacques, for always believing in me and supporting me and

for his assistance in the drawing of Figure 2.1
ix
Contents
Declaration iii
Synopsis v
Samevatting vii
Acknowledgements ix
Contents x
1 Introduction 1
1.1 Crane load models . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Reliability of crane support structures . . . . . . . . . . . . . . 2
1.3 Code calibration of crane load models . . . . . . . . . . . . . . 3
1.3.1 Calibration to current practice . . . . . . . . . . . . . . 4
1.3.2 Reliability calibration . . . . . . . . . . . . . . . . . . . 5
2 Design of crane support structures 11

2.1 Electric overhead travelling cranes . . . . . . . . . . . . . . . . 11
2.2 EOT Crane support structures . . . . . . . . . . . . . . . . . . 14
2.3 Design process for EOT crane support structures . . . . . . . . 15
2.3.1 Loads imposed by cranes on the support structure . . . 16
2.3.2 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.3.3 Crane - support structure interaction . . . . . . . . . . . 52
2.3.4 Support structure configurations and details . . . . . . . 53
2.3.5 Correct construction . . . . . . . . . . . . . . . . . . . . 56
2.4 Reliability of EOT crane support structures . . . . . . . . . . . 57
x
Contents xi
3 Scope of the code 59

3.1 Crane parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.1.1 Configuration of crane . . . . . . . . . . . . . . . . . . . 60
3.1.2 Nominal weights of crane and hoistload . . . . . . . . . 60
3.1.3 Crane geometry . . . . . . . . . . . . . . . . . . . . . . . 61
3.1.4 Travel and hoist speeds . . . . . . . . . . . . . . . . . . 62
3.1.5 Hoist type and characteristics . . . . . . . . . . . . . . . 62
3.1.6 Wheels and wheel drives . . . . . . . . . . . . . . . . . . 63
3.1.7 Guide means . . . . . . . . . . . . . . . . . . . . . . . . 65
3.1.8 Buffers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.1.9 Governing parameters . . . . . . . . . . . . . . . . . . . 66
3.2 Range of support structure configurations . . . . . . . . . . . . 69
3.2.1 Crane Girders . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2.2 Crane columns . . . . . . . . . . . . . . . . . . . . . . . 70
3.3 Representative cranes and their support structures . . . . . . . 70
3.3.1 5t crane . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.3.2 40t crane . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.3.3 260t crane . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4 Calibration to current practice 81

4.1 Crane code provisions . . . . . . . . . . . . . . . . . . . . . . . 81
4.1.1 Crane classification . . . . . . . . . . . . . . . . . . . . . 81
4.1.2 Load cases . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.1.3 Load combinations . . . . . . . . . . . . . . . . . . . . . 86
4.1.4 Fatigue loading . . . . . . . . . . . . . . . . . . . . . . . 90
4.2 Cost of support structure . . . . . . . . . . . . . . . . . . . . . 91
4.2.1 Structural elements . . . . . . . . . . . . . . . . . . . . . 93
4.2.2 Crane class . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.2.3 Comparison of load effects . . . . . . . . . . . . . . . . . 94
4.2.4 Design effects . . . . . . . . . . . . . . . . . . . . . . . . 98
4.3 Design effort . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.3.1 Information required for design . . . . . . . . . . . . . . 99
4.3.2 Work required for load calculations . . . . . . . . . . . . 101
4.3.3 Work required for support structure design . . . . . . . 101
4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
xii Contents
5 Development of limit states equations 103

5.1 Ultimate limit state . . . . . . . . . . . . . . . . . . . . . . . . 104
5.1.1 Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.1.2 Resistances . . . . . . . . . . . . . . . . . . . . . . . . . 108
5.1.3 Limit states equations . . . . . . . . . . . . . . . . . . . 112
5.2 Accidental limit state . . . . . . . . . . . . . . . . . . . . . . . . 113
5.2.1 Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.2.2 Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.2.3 Design equation . . . . . . . . . . . . . . . . . . . . . . 114
5.2.4 Limit state equation . . . . . . . . . . . . . . . . . . . . 114
5.3 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.3.1 Economic design method . . . . . . . . . . . . . . . . . 116
5.3.2 Economic design . . . . . . . . . . . . . . . . . . . . . . 118
5.3.3 Reliability analysis . . . . . . . . . . . . . . . . . . . . . 124
5.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6 Stochastic models 129

6.1 Basis for statistical modelling . . . . . . . . . . . . . . . . . . . 129
6.2 Sources of information . . . . . . . . . . . . . . . . . . . . . . . 130
6.3 Material properties . . . . . . . . . . . . . . . . . . . . . . . . . 130
6.3.1 Structural steel . . . . . . . . . . . . . . . . . . . . . . . 130
6.3.2 Bolts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
6.3.3 Welds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
6.3.4 Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
6.4 Geometric properties . . . . . . . . . . . . . . . . . . . . . . . . 136
6.4.1 Steel members . . . . . . . . . . . . . . . . . . . . . . . 136
6.4.2 Welds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6.4.3 Concrete column . . . . . . . . . . . . . . . . . . . . . . 138
6.5 Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
6.5.1 Permanent loads . . . . . . . . . . . . . . . . . . . . . . 139
6.5.2 Roof imposed loads . . . . . . . . . . . . . . . . . . . . . 140
6.5.3 Wind loads . . . . . . . . . . . . . . . . . . . . . . . . . 140
6.6 Modelling uncertainties . . . . . . . . . . . . . . . . . . . . . . 141
6.6.1 Basis for modelling uncertainties . . . . . . . . . . . . . 141
6.6.2 Distributions for modelling uncertainties . . . . . . . . . 142
6.7 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Contents xiii
6.7.1 Fatigue resistance . . . . . . . . . . . . . . . . . . . . . 150

6.7.2 Fatigue loading . . . . . . . . . . . . . . . . . . . . . . . 152
6.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
7 Stochastic modelling of crane hoistload 157

7.1 ‘One cycle’ distribution . . . . . . . . . . . . . . . . . . . . . . 157
7.1.1 Type of distribution . . . . . . . . . . . . . . . . . . . . 158
7.1.2 Upper limit of distribution . . . . . . . . . . . . . . . . 159
7.1.3 Shape of distribution . . . . . . . . . . . . . . . . . . . . 159
7.2 Extreme hoistload distributions . . . . . . . . . . . . . . . . . . 166
7.2.1 Basis for development . . . . . . . . . . . . . . . . . . . 167
7.2.2 Method of development . . . . . . . . . . . . . . . . . . 168
7.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
8 Code calibration method 175

8.1 Reliability analysis method . . . . . . . . . . . . . . . . . . . . 175
8.1.1 Economic design . . . . . . . . . . . . . . . . . . . . . . 175
8.1.2 Reliability assessment . . . . . . . . . . . . . . . . . . . 176
8.2 Definition of code objective . . . . . . . . . . . . . . . . . . . . 177
8.2.1 Ultimate limit state . . . . . . . . . . . . . . . . . . . . 177
8.2.2 Accidental limit state . . . . . . . . . . . . . . . . . . . 178
8.2.3 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
8.3 Definition of code format . . . . . . . . . . . . . . . . . . . . . 180
8.3.3 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
8.4 Calibration method . . . . . . . . . . . . . . . . . . . . . . . . . 183
8.4.3 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
8.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
9 Code calibration results 193

9.1 Ultimate limit state . . . . . . . . . . . . . . . . . . . . . . . . 193
9.1.1 Crane load only . . . . . . . . . . . . . . . . . . . . . . . 193
9.1.2 Combinations of time varying loads . . . . . . . . . . . . 212
9.2 Accidental limit state . . . . . . . . . . . . . . . . . . . . . . . . 220
xiv Contents
9.2.1 Results of parametric studies . . . . . . . . . . . . . . . 221

9.2.2 Calibration of partial load factors . . . . . . . . . . . . . 222
9.2.3 Verification of partial load factors . . . . . . . . . . . . . 222
9.3 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
9.3.1 Results of parametric studies . . . . . . . . . . . . . . . 225
9.3.2 Calibration of partial load factors . . . . . . . . . . . . . 232
9.3.3 Simulation of crane behaviour . . . . . . . . . . . . . . . 232
9.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
10 Discussion of results 239

10.1 Sensitivity factors . . . . . . . . . . . . . . . . . . . . . . . . . . 239
10.1.1 Sensitivity of reliability to modelling uncertainty para-
meters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
10.1.2 Sensitivity of partial load factors to modelling uncer-
tainty parameters . . . . . . . . . . . . . . . . . . . . . . 243
10.2 Code format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
10.2.3 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
10.3 Further work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
10.3.1 Multiple cranes . . . . . . . . . . . . . . . . . . . . . . . 250
10.3.2 Probabilistic optimisation . . . . . . . . . . . . . . . . . 252
11 Conclusions 253
11.1 Calibration to current practice . . . . . . . . . . . . . . . . . . 255
11.1.1 Crane load models from prEN 1991-3 . . . . . . . . . . . 255
11.1.2 Fatigue loading in prEN 1991-3 . . . . . . . . . . . . . . 257
11.1.3 Implications of adopting crane load models from prEN
1991-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
11.2 Reliability calibration . . . . . . . . . . . . . . . . . . . . . . . 260
11.2.1 Stochastic modelling . . . . . . . . . . . . . . . . . . . . 261
11.2.2 Code calibration . . . . . . . . . . . . . . . . . . . . . . 263
11.3 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . 269
11.4 Further work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
References 273
Contents xv
List of Figures 279
List of Tables 291
A Load calculations 295

A.1 Crane loads according to SABS 0160:1989 . . . . . . . . . . . . 295
A.1.1 Vertical loads . . . . . . . . . . . . . . . . . . . . . . . . 296
A.1.2 Horizontal transverse loads . . . . . . . . . . . . . . . . 296
A.1.3 Horizontal longitudinal load . . . . . . . . . . . . . . . . 297
A.1.4 End stop forces . . . . . . . . . . . . . . . . . . . . . . . 297
A.2 Crane loads according to prEN 1991-3 . . . . . . . . . . . . . . 298
A.2.1 Dynamic factors . . . . . . . . . . . . . . . . . . . . . . 298
A.2.2 Load combinations . . . . . . . . . . . . . . . . . . . . . 299
A.2.3 Vertical loads . . . . . . . . . . . . . . . . . . . . . . . . 300
A.2.4 Horizontal loads . . . . . . . . . . . . . . . . . . . . . . 302
A.3 Roof Imposed loads . . . . . . . . . . . . . . . . . . . . . . . . . 307
A.4 Wind loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
A.5 Permanent loads . . . . . . . . . . . . . . . . . . . . . . . . . . 311
B Load Effects 313

B.1 5t crane girder . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
C Graphs for stochastic modelling of hoistload 317

C.1 Class 1 cranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
C.2 Class 2 cranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
C.3 Class 3 cranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
C.4 Class 4 cranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
C.5 Extreme hoistload distributions . . . . . . . . . . . . . . . . . . 323
D Ratios of hoistload to total crane weight 325
E Code calibration results 329

E.1 Ultimate limit state - crane only . . . . . . . . . . . . . . . . . 329
E.1.1 5t crane girder . . . . . . . . . . . . . . . . . . . . . . . 329
E.1.2 5t crane column . . . . . . . . . . . . . . . . . . . . . . 334
E.1.3 40t crane girder . . . . . . . . . . . . . . . . . . . . . . . 336
E.1.4 40t crane column . . . . . . . . . . . . . . . . . . . . . . 342
E.1.5 260t crane girder . . . . . . . . . . . . . . . . . . . . . . 344
xvi Contents
E.1.6 260t crane auxiliary girder . . . . . . . . . . . . . . . . . 347

E.1.7 260t crane column . . . . . . . . . . . . . . . . . . . . . 349
E.2 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
E.2.1 5t crane corbel to column welded connection . . . . . . 351
E.2.2 40t crane girder bottom of intermediate stiffener . . . . 352
E.2.3 40t crane girder top flange to web weld . . . . . . . . . 354
Chapter 1
Introduction
Electric overhead travelling cranes are used in industrial applications for mov-
ing loads without causing disruption to activities on the ground. Overhead
cranes can be described as machines for lifting and moving loads, consisting of
a crane bridge which travels on wheels along overhead crane runway beams, a
crab which travels across the bridge and a hoist for lifting the loads.
Overhead cranes are often an integral part of the industrial process and
any time in which the crane is not able to be used can have severe financial
implications for the owner. Cranes can be classified as heavy machinery which
lift loads overhead and any mechanical or structural failure which causes the
crane or load to fall could become a serious safety hazard.
The overhead travelling crane runway beams and the structural elements
which support them are referred to as the crane support structure.
Overhead travelling cranes are supplied by the crane manufacturers whose
responsibility it is to ensure the working and safety of the crane itself. The
crane support structure is designed by a structural engineer who is responsible
for ensuring that the support structure is sufficiently strong and robust to
withstand the loads that are imposed by the crane.
Various problems have been encountered with crane support structures in
practice, many of these arise from the deflections that result from the crane
loads. Cranes, by their nature of moving loads by travelling along the crane
runway beams, imposed cyclic loading on the support structure which can
lead to fatigue. Fatigue failures are common problems that occur with crane
support structures.
These problems which have been observed with crane support structures
1
2 Chapter 1. Introduction
prompted this investigation into the design of crane support structures and
more specifically crane induced loads on the support structures. Two aspects
are considered here, the load models provided in codes of practice for the crane
induced loads and the structural reliability of the crane support structures.
1.1 Crane load models

Codes of practice on loadings on buildings give load models for the calculation
of vertical and horizontal wheel loads that overhead travelling cranes impose
on their support structures.
The crane load models in the South African loading code (SABS 0160:1989
[1]) are over-simplistic as will be shown by a comparison between the crane
load models in SABS 0160:1989 and those in the Eurocode 1 Part 3 (prEN
1991-3 [2]), the German code (DIN 15018 Part 1 1984 [4]), the International
standard (ISO 8686-1:1989 [5]), the Australian code (AS1418.1-1994 [6]) and
the American code (ASCE 7-98 [7]).
Because of the over-simplistic nature of the crane load models in SABS
0160:1989, the crane load models are currently under review by the South
African loading code committee, with the intention of adopting more sophisti-
cated crane load models into the updated South African loading code, SANS
10160 [3].
The prEN 1991-3 crane load models have been widely accepted, having
been developed from DIN 15018 and ISO 8686-1:1989 and forming the basis of
AS1418.1-1994. Because of their wide acceptance, the adoption of the prEN
1991-3 crane load models into proposed SANS 10160 has been proposed and
is currently under consideration.
An assessment of the prEN 1991-3 crane load models is required to de-
termine if they are suitable for inclusion into proposed SANS 10160 and to
investigate the implications of their adoption into proposed SANS 10160.
1.2 Reliability of crane support structures

There is no reliability basis for the crane partial load factor applied to crane
wheel loads in SABS 0160:1989. The imposed load partial factor of 1.6 is used
which has been calibrated for floor loads in office buildings. Crane induced
loads are caused by the mechanical operations of lifting loads and the crane
1.3. Code calibration of crane load models 3
movement along the runway beams whereas floor loads in office buildings are
due mainly to furniture and the occupants of the building. These loads do not
have the same origin, one is mechanical and the other is based on the weight
of objects and people, and there is no evidence to suggest that crane loads
have the same characteristics as floor loads in office buildings; so it is unclear
whether the same partial load factor is suitable for both.
In contrast, prEN 1991-3 prescribes a partial load factor of 1.35 for crane
loads which is the same as the permanent load factor in prEN 1991-3.
There is thus a difference in the approach taken to, and the value of, the
crane partial load factor between the two codes; SABS 0160:1989 applies a
factor equal to the imposed load factor of 1.6 and prEN 1991-3 applies a factor
equal to the permanent load factor of 1.35.
The lack of a reliability basis for the crane partial load factors in SABS
0160:1989 indicates a need for a reliability assessment of the crane load models
to assess the current level of reliability and if necessary to determine appro-
priate partial load factors for crane induced loads.
1.3 Code calibration of crane load models
The two aspects of the crane loading code which have been identified as critical
and requiring investigation are:
1. Crane load models
The crane load models in SABS 0160:1989 are over-simplistic, so it is

under consideration to adopt the prEN 1991-3 crane load models into
proposed SANS 10160.
An assessment of the prEN 1991-3 crane load models is required to in-

vestigate their suitability for inclusion into proposed SANS 10160 and
the implications of their inclusion into proposed SANS 10160.
2. Reliability of the crane load models.
Different partial load factors are applied to crane loads in SABS 0160:1989
and prEN 1991-3 and there is no reliability basis for these crane partial
load factors.
A reliability assessment of the crane load models is required to determine

the current level of reliability and, if necessary, to determine appropriate
partial load factors for the crane loads.
Both of these aspects of crane induced loads that require investigation

can be treated as code calibration problems, where the ‘code’ that is being
considered is the code of practice which specifies loads that cranes impose on
their support structures.
The code calibration consists of two parts, firstly calibration to current
practice and secondly reliability calibration.
1.3.1 Calibration to current practice
The calibration to current practice takes the form of a comparison between the
prEN 1991-3 crane load models and the crane load models in SABS 0160:1989.
The aims of the calibration to current practice are to assess the manner in
which the crane behaviour is modelled and to assess the effect on the cost of
the support structure and the difference in design effort, of using the crane
load models in prEN 1991-3 rather than those in SABS 0160:1989. The com-
parison is carried out with respect to these three aspects of the codes which
are discussed in more detail below:
1. Load situations
The crane load situations that are allowed for and the aspects of the
crane behaviour that are modelled are assessed. The load cases and load
combinations in each code are compared.
2. Cost of support structure
The effect on the cost of the crane support structure of calculating the
crane loads using the prEN 1991-3 crane load models is assessed.
3. Design effort
The amount of work required to calculate the factored load combinations

for each code and the work required for the subsequent design of the
support structure is compared.
1.3.2 Reliability calibration
Reliability based code calibration has not previously been applied to crane
support structures. In this investigation the methods of reliability code cal-
ibration are applied to crane loads to investigate suitable code formats and
partial load factors for crane loads as well as combination factors for combina-
tions of crane loads with other time varying loads, i.e. wind and roof imposed
loads.
Reliability code calibration has been defined by Faber & Sørensen [8] as
using reliability analysis methods to choose design equations, characteristic
values, combination schemes and partial load and resistance factors to maintain
a minimum and consistent target reliability over all choices of material, loading
conditions and structural configurations.
As mentioned above, the reliability code calibration is concerned with the
crane induced loads on the support structure.
In general code calibration for a body of codes, the calibration of the load
factors and resistance factors is carried out separately as described by Kemp
et al. [9]. This is because of the wide range of load and resistance parameters
which need to be taken into account and the need for partial load factors which
are independent of the resistance codes. The partial load factors are calibrated
first to obtain design loads which have a specified maximum probability of
occurrence. The second step is to calibrate the partial resistance factors that
result in a consistent probability of failure when combined with the calibrated
design loads.
Since a particular structural type, i.e. crane support structures, is consid-
ered here, it is practical to include the resistance modelling when calibrating
the partial load factors. The code calibration procedure followed here, there-
fore, will not follow the convention of separating resistances and loads, rather,
limit states equations including both loads and resistances are set up for the
reliability analysis.
The crane load models are the focus of the code calibration and all the
other loads, partial load factors, resistances and partial resistance factors are
taken as they are specified in the current South African loading and materials
codes (SABS 0160:1989 [1]; SABS 0100-1:1992 [10]; SANS 10162-1:2005 [11]).
The aims of the reliability calibration of the crane load models are given
below:
1. A reliability assessment of the two different sets of load models.

This reliability assessment is carried out to determine the level of relia-
bility of crane support structure for present practice of SABS 0160:1989
as well as for prEN 1991-3.
2. An investigation into different code formats.

The code format refers to the number and type of partial load factors
as well as the load combination schemes [8]. Different code formats are
considered here and recommendations for the best alternative are given.
3. Determination of partial load factors and combination factors.

The final code calibration objective is to find crane partial load factors
and combination factors which result in a consistent level of reliability
over a range of parameters.
The procedure for the code calibration of the crane load models was based
on that given by Faber & Sørensen [8] and is outlined below.
1. Definition of the scope of the code

The code calibration is focussed on the crane load models and ‘the code’
therefore refers to the crane loading code. The crane load models which
are specifically being considered are those in the Eurocode 1 Part 3,
prEN 1991-3 [2] because they are being assessed with a view to being
incorporated into proposed SANS 10160.
The definition of the scope of the proposed new code entails determining
the type of cranes whose wheel loads can be calculated by the crane
loading code and the range of these cranes in South Africa.
The code calibration is carried out on specific example structures (cranes
and support structures) as recommended by Faber & Sørensen [8] and
Hansen & Sørensen [12]. These structures are chosen to be representative
of the scope of the code.
2. Definition of the code objective

The code objective could either be maximisation of the expected utility
of the code or to achieve a target reliability [12]. The objective for
this calibration is the achievement of a consistent target reliability. An
assessment of recommended target reliabilities is carried out and suitable

values are chosen for the calibration of the crane load models for the
different limit states considered.
3. Definition of the code format

As mentioned above, the code format refers to the number and type of
partial load factors and load combination schemes.
Load combinations consist of combinations of crane loads with perma-
nent loads or other time varying loads such as wind or roof imposed loads
as well as combinations of actions from more than one crane. Actions
from more than one crane are not within the scope of this investigation;
so this code calibration focusses on single cranes only in combination
with permanent loads, wind loads and roof imposed loads.
The load combination schemes for crane loads with permanent loads
and wind or roof imposed loads are defined by the South African loading
code, SABS 0160:1989 [1], so the definition of the code format here will
focus on the number and type of crane partial load factors.
Different code formats are considered for this code calibration, to take
into account the different characteristics of various crane induced loads.
The different code formats considered are the use of one partial load
factor applied to the characteristic wheel loads as is the current practice
in both SABS 0160:1989 and prEN 1991-3, and an extension of this
format to the use of two partial load factors applied separately to the
crane self weight and hoistload before the calculation of the design wheel
loads and the option of including an additional partial load factor for the
horizontal crane wheel loads.
4. Identification of typical failure modes

The typical failure modes of the structural elements considered for the
code calibration are required to set up the limit states equations.
Because this code calibration considers the loading and resistance to-
gether, not separately as is the convention for the calibration of the
Eurocodes, the resistances of the structural elements considered are re-
quired for setting up the limit states equations. The elements chosen for
the limit states equations are those that are subject to crane loads i.e.
the crane girders, crane columns, roof columns, roof trusses and crane
runway bracing. The design of these elements is carried out according

to the South African materials codes [10; 11]. The critical failure modes
that govern the sizes of the elements are selected for the limit state equa-
tions.
Three limit states are considered: ultimate limit state, accidental limit
state and fatigue.
5. Development of stochastic models

The variables in the limit states equations that are considered as random
variables for the reliability analysis are identified. Most of the stochastic
models are taken from literature.
Statistical models for crane loads have been presented by Köppe [13] and
Pasternak et al. [14] but are not suitable for the reliability based code
calibration carried out for this investigation. Due to the lack of available
suitable models, stochastic models, which are suitable for a range of limit
states and load combinations, are developed for this investigation for
loads lifted by cranes and the modelling uncertainties in the calculation
of the crane wheel loads. The development of these stochastic models is
discussed
6. Determination of the optimal partial load factors

The method of determining the optimal partial load factors is to carry out
reliability analyses on the selected structural elements. The reliability
analysis is carried out as an iterative process in two steps.
a) Economic design
The code calibration exercise is specifically interested in the relia-
bility of the code, therefore the structural element is first designed
to exactly satisfy the code requirements. No practical rounding of
the element size is carried out. Practical rounding includes conser-
vatism in the resistance of the element so that it no longer represents
the code specifications and would have a higher reliability than that
implied by the code. Code calibration carried out on conservatively
designed elements would underestimate the size of the partial load
factor required to achieve a given target reliability.
The load models used for the crane loads in the economic design
are the models in prEN 1991-3, the resistances are calculated using
the South African steel design code SANS 10162-1:2005 [11] and
concrete design code SABS 0100-1:1992 [10].
The economic design is carried out using assumed values of the
crane partial load factors.
b) Reliability analysis
The size of the element obtained from the economic design is used
as the nominal size for the reliability analysis. A time integrated
approach using a first order reliability method (FORM) with the
Rackwitz-Fiessler procedure as given by Nowak & Collins [15] is
used for the reliability analysis.
These two steps are repeated, varying the value of the crane partial load
factors until the reliability of the element satisfies the code objective.
Finding the optimal partial load factors will entail assessing the different
code formats to determine which best meets the code calibration criterion
of obtaining a consistent level of reliability over a range of parameters.
7. Verification of partial load factors

The final step in the code calibration procedure is the verification of
the partial load factors. All the elements and limit states are assessed to
ensure that the code objective is met. The level of conservatism incurred
with the chosen partial load factors is evaluated.
The crane partial load factors will also be assessed on the basis of engi-
neering judgement and existing practice.
Chapter 2
Design of crane support

structures
Electric overhead travelling cranes (EOT cranes) are used in industrial appli-
cations for mechanically moving loads without interfering with activities on
the ground. An EOT crane can be defined as: ‘A machine for lifting and
moving loads that moves on wheels along overhead crane runway beams. It
incorporates one or more hoists mounted on crabs or underslung trolleys’ [2].
Cranes are an essential part of the industrial process and any ‘down time’ can
have severe financial consequences for the owner. For this reason it is essential
that the running of the crane is kept as problem free as possible, leading to
minimum disruption of service.
An overview is given below of electric overhead travelling cranes and their
support structures with a discussion of the common problems that are encoun-
tered.
2.1 Electric overhead travelling cranes

EOT cranes can either operate within an industrial building or outside a build-
ing. Industrial buildings housing EOT cranes generally consist of one or more
bays with one or more EOT cranes in each bay.
The magnitude of the loads lifted by EOT cranes varies from 2t up to 630t.
The rated weight of the load lifted by an EOT crane is referred to as the ‘safe
working load’ (SWL) and is the way in which EOT cranes are referred to (e.g.
a 40t crane).
11
12 Chapter 2. Design of crane support structures
EOT cranes can have various different configurations depending on the

application, the layout of the industrial building housing the crane and the
type of loads to be lifted. The three basic types of electric overhead travelling
cranes are bridge cranes as shown in Figure 2.1, portal cranes and semi-portal
cranes. Portal cranes are a portal frame structure with the bottom of the
frame legs running on rails and semi-portal cranes have one end of the crane
bridge running on an elevated rail and the other end supported on a column
(as with a portal frame) with the bottom of the column running on a lower
rail. EOT bridge cranes are the most commonly used cranes in industry and
are the type of cranes which fall under the scope of the crane loading codes.
These bridge cranes will be the focus of this investigation.
Figure 2.1: Main components of an EOT bridge crane
Within the category of EOT bridge cranes there are various different con-
figurations. The biggest distinction is between underslung overhead travelling
cranes and top mounted overhead travelling cranes. Underslung cranes are
supported on the bottom flanges of the runway beams and top mounted over-
head cranes are supported on rails on the top of the runway beams.
The main components of an EOT crane are the crane bridge which spans
the width of the bay between the runway girders and moves longitudinally
2.1. Electric overhead travelling cranes 13
down the length of the building, the crane crab which traverses the bridge
and houses the hoisting mechanism and the end carriages on either side of
the crane bridge which house the wheel blocks. The combined horizontal
longitudinal and transverse movements of the crane bridge and crab and the
vertical movement of the hoist allow the crane to reach any position in the
industrial building for the purpose of lifting or lowering a load. The elements
of an EOT crane are shown in Figure 2.1.
Top mounted overhead travelling cranes can have further different configu-
rations. EOT cranes lifting light loads typically have hot rolled I or H sections
for the end carriages and a single hot rolled I or H section for the crane bridge
with the crab hoist running along the bottom flange of the crane bridge. EOT
cranes lifting heavier loads normally have box girders for the end carriages
and crane bridge, the crane bridge consists of two parallel box girders with
the crab unit mounted on rails on top of the crane bridge girders. In cases of
cranes with extremely heavy applications, the crane bridge can consist of four
box girders.
There are a wide range of applications for which cranes are used which
require different load lifting mechanisms. Hooks are the most common type
of load lifting mechanism for general warehouse and industrial use. Ladles
are used in metal works for transporting molten metal from the furnace to the
casters. Another load lifting mechanism is a grab which is used for applications
such as lifting granular material or scrap metal. Cranes equipped with a
magnet lifting device are typically used for lifting steel plates. Specialised
load lifting mechanisms such as coil lifters are used for specific applications.
The wheels are a very important part of EOT bridge cranes because the
smooth running of the crane depends on the quality of the wheels and the
wheels transfer the loads from the crane to the support structure. Most cranes
have four wheels, two on each end carriage, but larger cranes lifting heavier
loads can have eight or sixteen wheels in total. The current practice of wheel
configurations is to have independent wheels which are not linked in any way
to the wheels on the opposite end carriages [16; 17; 18; 19]. The driven wheels
each have their own wheel drive.
Buffers are supplied on the end carriages to reduce the impact forces if the
crane runs into the end stops at the end of the runway. The different types of
buffers are rubber and cellular plastic which are used mainly for smaller cranes
as well as hydraulic buffers which are used mainly for larger cranes.
2.2 EOT Crane support structures
The movement of loads is an integral, essential part of many industrial appli-

cations and cranes therefore play a very important role in the smooth running
of the industrial process. A crane that is out of commission can have very
serious financial implications for the owner. For example, if a crane in a steel
works building which carries the molten metal from the furnace to the caster is
out of commission, the whole production line comes to a halt with a great loss
in production and hence a great financial loss. The problem of the crane be-
ing unable to fulfil its intended purpose is considered a serviceability problem
because the implications are financial and not safety related.
The crane is also an important aspect when considering the safety of the
industrial area. Cranes are a type of heavy machinery which lift large loads
and if something should cause the load or the crane to fall it would endanger
the lives of the people working in the industrial area.
Problems which could cause cranes to be out of commission or cause a

safety hazard could be related to either the crane itself or the support struc-
ture. Ensuring the functioning and safety of the crane itself when it is man-
ufactured is the responsibility of the crane manufacturer and ensuring the
continued functioning and safety of the crane by means of regular inspection
and maintenance is the responsibility of the owner. The functioning and safety
of the support structure initially is the responsibility of the structural engineer
and the responsibility for the ongoing functioning and safety of the support
structure lies partly with the structural engineer in designing a durable struc-
ture and partly with the owner in regular inspection and maintenance. The
aspects of the functioning and safety of the EOT crane in the industrial area
that will be focussed on here are those which are the responsibility of the struc-
tural engineer, viz. to design a support structure which is sufficiently reliable
over the lifetime of the crane.
Ensuring sufficient reliability of the support structure is a twofold process.

The first aspect is the accurate structural modelling of the support structure
in the design process, this includes both load modelling and structure response
modelling. The second aspect is ensuring sufficient statistical reliability of the
support structure by means of suitable partial safety factors.
2.3. Design process for EOT crane support structures 15
2.3 Design process for EOT crane support

structures
The crane support structure consists of the rails, rail fastening system, crane
runway girders, crane columns, crane column bracing, crane column founda-
tions, crane stops and conductor rail supports (Ricker [20]). The crane induced
loads are applied by the wheels to the rails which transmit the loads into the
girders which in turn transmit the loads to the columns and bracing and down
to the foundations. Figure 2.2 shows the main components of the crane support
structure.
(a) Crane bay
(b) Crane runway
Figure 2.2: Crane support structure

EOT cranes can either operate within an industrial building or outside a

building. In the case where EOT cranes operate inside a building, the building
withstands the environmental actions of wind or temperature and the crane
girders are subject to crane loads only. The crane support structure can be
integrated with the building structure (e.g. the crane girders are supported on
corbels connected to the building column) and in this case the roof members
are affected by the crane loads. Heavier cranes are typically supported on
columns that are separate from the building columns and in this case the roof
members are not subject to crane loads. In the case where EOT cranes operate
outside a building, the crane support structure carries the crane loads as well
as the environmental loads.
Ensuring an EOT crane support structure which does not cause down time
for the crane or cause a safety hazard is the responsibility of the structural
engineer. In order to properly design an EOT crane support structure the
designer must have a good understanding of all the steps in the design process:
1. Identifying the loads that the crane imposes on the support structure
2. Modelling the response of the structure
3. Choosing the correct structural system and details to ensure a durable

structure
2.3.1 Loads imposed by cranes on the support structure
In order to be able to design the EOT crane support structure, the designer
must understand and take into account the various loads that the support
structure will be subject to during its lifetime.
There is very little literature available on loads that EOT cranes impose
on their support structures. A series of articles has been published by Lobov
[21; 22; 23; 24] and Spitsyna & Anoskin [25] which investigate the dynamic
forces on the crane caused by acceleration of the crane bridge, skewing of the
crane bridge and contact of the wheel flange with the rail. The effects on the
crane have been investigated considering the equations of motion of the crane.
The results of these investigations are not in a form where they are applicable
for the calculation of loads that cranes impose on their support structures.
In the absence of literature on crane loads, recourse must be made to design

codes of practice which specify crane loads to be considered in the design of
support structures.
An important point is raised by Lobov [21], and that is that crane loads
are primarily dynamic loads. These dynamic loads are caused by the hoisting
of loads and movement of the crane and crab in order to transport the loads
around the industrial area. In the design codes these dynamic loads are treated
as static loads with amplification factors, usually represented by φ, to allow
for dynamic effects.
The crane loads are transferred to the support structure by the wheels. For
this reason the loads that cranes impose on their support structures are often
referred to as crane ‘wheel loads’.
Crane loads can be divided into three basic categories.
1. Loads arising from normal operation of the crane
2. Loads arising from accidental situations
3. Loads arising from improper construction, lack of maintenance or misuse

of the crane
Dunaiski et al. [26] give a review of crane load provisions from various
codes, considering only vertical and horizontal crane loads due to normal op-
eration of the crane. A more extensive discussion is presented here of the
crane provisions, with respect to the three types of loads listed above, in the
following design codes.
1. South African loading code SABS 0160:1989 [1]
2. Eurocode on crane loading prEN 1991-3 [2]
3. German standard on cranes DIN 15018 Part 1 1984 [4]
4. International standard on crane loading ISO 8686-1:1989 [5]
5. Australian code on crane loading AS1418.1-1994 [6]
6. American standard for loading on buildings ASCE 7-98 [7]
SABS 0160:1989, prEN 1991-3, DIN 15018, AS1418.1-1994 and ASCE 7-98
specify loads that cranes impose on the support structure. ISO 8686-1:1989
specifies the wheel loads that cranes is subject to, the support structure would
be subject to equal and opposite reaction forces from the crane wheels.
2.3.1.1 Classification of cranes
In many of the design codes on crane loading, the crane wheel loads depend
on the crane classification. The various classification systems in the codes are
discussed below.
SABS 0160:1989 classifies cranes into four classes based on a description of
the usage of the crane as given in Table 2.1.
Table 2.1: Classification of cranes in SABS 0160:1989

Class of crane Description of crane
Class 1 Hand cranes
Light duty
Class 2 Cranes for general use in factories and workshops
Medium duty Warehouse cranes - intermittent operation
Power station cranes
Machine shop cranes
Foundry cranes
Class 3 Warehouse cranes - continuous operation
Heavy duty Scrapyard cranes
Rolling mill cranes
Grab and magnet cranes - intermittent operation
Ladle cranes in steelworks
Class 4 Grab and magnet cranes - continuous operation
Extra Heavy duty Soaking pit cranes
Ingot stripping cranes
Furnace charging cranes
Forging cranes
Claw cranes
The classification of the crane influences the magnitude of the vertical and
horizontal forces as will be discussed in the sections below.
ASCE 7-98 does not have a classification system for cranes beyond dis-
tinguishing between hand operated cranes, powered cab or remotely operated
cranes and powered pendant operated cranes. This nominal classification af-
fects only the vertical impact forces.
prEN 1991-3, ISO 8686-1:1989, AS1418.1-1994 and DIN 15018 all classify
cranes into ‘Hoist Classes’. The hoist class of the crane affects only the dynamic
factor applied to the hoistload to model the dynamic effects of lifting the
hoistload off the ground.
The rationale behind the classification of the cranes into different hoist
classes is given by DIN 15018 as: ‘The softer the springing of the hoisting gear,
the larger the elasticity of the supporting structure, the smaller the actual
hoisting speed at the commencement of the hoisting of the useful load, the
smaller and steadier the acceleration and deceleration during changes in the
hoisting motion, the smaller the factor φ. Accordingly, the cranes are classified
into lifting classes . . . with different factors φ’ [4]. Where φ is the dynamic
factor applied to the hoistload to allow for the dynamic effects of lifting a load
off the ground.
prEN 1991-3 and DIN 15018 provide a table giving the hoist class of cranes
depending on a description of their usage, some types of cranes fall into more
than one class. The crane descriptions and classes are given in Table 2.2.
Table 2.2: Classification of cranes in DIN 15018 and prEN 1991-3

Hoist class of crane Description of crane
HC1 Hand cranes
Powerhouse cranes
HC1, HC2 Assembly cranes
HC2 Storage cranes - intermittent usage
HC2, HC3 Workshop cranes
Casting cranes
HC3, HC4 Storage cranes - continuous operation
Scrapyard cranes - continuous operation
Soaking pit cranes
HC4 Stripper cranes
Charging cranes
Forging cranes
ISO 8686-1:1989 recommends that the designer select the hoist class of the
crane on the basis of experience.
AS1418.1-1994 provides a table for the selection of the hoist class of the
crane which is related to the rationale behind the hoist classes as given by DIN
15018. The table relates the natural frequency of the crane structure in the
vertical plane to the hoisting acceleration and is shown in Table 2.3.
Whereas the classification method given in AS1418.1-1994 reflects more
the rationale behind the classification it would be more difficult to obtain the
information required for the classification.
Table 2.3: Classification of cranes into hoist classes in AS1418.1-1994

Fundamental natural Hoisting application group
frequency of Hoisting acceleration
structure m/s2
(vertical plane) Hz ≤ 0.2 0.2 - 0.4 0.4 - 0.6 >0.6
≤3.2 H1 H1 H2 H3
3.2 - 5.0 H1 H2 H2 H3 to H4
>5.0 H2 H2 H3 H4
2.3.1.2 Loads arising from normal operation of the crane
The loads that arise from normal operation of the crane can be divided into
vertical loads and horizontal loads. The vertical loads can be divided into two
parts, firstly the static part arising from gravitational effects on the crane and
hoistload and secondly the dynamic part which is caused by inertial effects
acting on the mass of the crane and hoistload. Horizontal transverse and
longitudinal loads arise due to the movement of the crane.
2.3.1.2.1 Vertical crane loads due to gravitational effects
All the codes agree that the nominal weights of the crane bridge, crab and pay
load as given by the crane manufacturer are to be used for the calculation of
characteristic vertical gravitational loads.
ISO 8686-1:1989, DIN 15018 and AS1418.1-1994 give no guidance on the
method of the calculation of the vertical gravitational loads. The calculation of
the vertical gravitational loads is generally carried out by considering equilib-
rium of the crane bridge supported by the wheels and as such does not require
specification.
SABS 0160:1989 recommends using the wheel loads supplied by the crane
manufacturer.
ASCE 7-98 states simply that the gravitational vertical wheel loads should
be calculated considering the hoistload and crab placed at the position where
the wheel load is a maximum.
prEN 1991-3 specifies the crab position and whether the crane is to be
loaded or unloaded for the calculation of the wheel loads. This results in four
values of wheel load:
• maximum - crab closest to wheel being considered, loaded crane

• accompanying maximum - crab furthest from wheel being considered,

loaded crane
• minimum - crab furthest from wheel being considered, unloaded crane
• accompanying minimum - crab closest to wheel being considered, un-

loaded crane
2.3.1.2.2 Vertical crane loads due to inertial effects
The codes differ more on the inertial effects to be considered for the vertical
crane loads. Four basic situations have been identified for which inertial effects
are taken into account.
(a) Generic dynamic effects

Both SABS 0160:1989 and ASCE 7-98 specify a dynamic factor to be applied
to the static vertical wheel loads to account for general dynamic and impact
effects. The value of the dynamic factor in SABS 0160:1989 depends on the
class of crane and in ASCE 7-98 on the type of crane. Tables 2.4 & 2.5 give
the dynamic factors for the vertical loads from SABS 0160:1989 and ASCE
7-98.
Table 2.4: Dynamic factors for vertical loads from SABS 0160:1989
Class of crane Dynamic factor φ
Class 1 1.10
Class 2 1.20
Class 3 1.25
Class 4 1.30
Table 2.5: Dynamic factors for vertical loads from ASCE 7-98
Powered monorail cranes 1.25
Powered cab-operated or remotely operated bridge cranes 1.25
Powered pendant-operated bridge cranes 1.10
Hand-operated bridge and monorail cranes 1.00
(b) Dynamic effects caused by lifting a hoistload off the ground
prEN 1991-3 and ISO 8686-1:1989 take into account the dynamic effects on
the crane self weight and hoistload caused by lifting a load off the ground. ISO
8686-1:1989 states that these dynamic effects are due to the drive coming up
to speed before the lifting attachment engages the loads and are the result of
a build up of kinetic energy and drive torque.
AS1418.1-1994 and DIN 15018 take into account the dynamic effect on the
hoistload only, caused by lifting a load off the ground.
prEN 1991-3 and ISO 8686-1:1989 specify a dynamic factor, φ1 , to be
applied to the crane self weight to account for dynamic effects of lifting a
hoistload off the ground.
prEN 1991-3 specifies the value of φ1 as:
0.9 ≤ φ1 ≤ 1.1 (2.3.1)
In a similar way, ISO 8686-1:1989 specifies the value of φ1 as:
φ1 = 1 ± a 0 ≤ a ≤ 0.1 (2.3.2)
The two values that are given represent the upper and lower values of the
vibrational pulse. For overhead travelling bridge cranes, typically only the
upper value is of interest. The lower value is relevant for tower or jib type
cranes which could have instability problems with lower hoistloads.
prEN 1991-3, ISO 8686-1:1989, AS1418.1-1994 and DIN 15018 specify a
dynamic factor, φ2 , to be applied to the hoistload to account for the dynamic
effects of lifting a load off the ground. The value of φ2 in all cases depends
on the Hoist Class of the crane and on the steady hoisting speed vh . The
equations are given below, in all cases the lifting speed is in m/s:
prEN 1991-3:
φ2 = φ2,min + β2 × vh (2.3.3)
ISO 8686-1:1989:
φ2 = φ2,min for vh ≤ 0.2 m/s

(2.3.4)
φ2 = φ2,min + β2 (vh − 0.2) for vh > 0.2 m/s
AS1418.1-1994:
φ2 = φ2,min for vh ≤ 0.2 m/s

φ2 = φ2,min + β2 (vh − 0.2) for 0.2 < vh ≤ 1.52 m/s (2.3.5)
φ2 = φ2,max for vh > 1.52 m/s
DIN 15018:
φ2 = φ2,min + β2 (vh − 0.2) for vh ≤ 1.5 m/s

(2.3.6)
φ2 = φ2,max for vh > 1.5 m/s
Table 2.6: Factors for the calculation of φ2

prEN 1991-3 ISO 8686 & AS1418 DIN 15018
Hoist β2 φ2,min β2 φ2,min φ2,max β2 φ2,min φ2,max
class
HC1 0.17 1.05 0.2 1.00 1.3 0.132 1.1 1.3
HC2 0.34 1.10 0.4 1.05 1.6 0.264 1.2 1.6
HC3 0.51 1.15 0.6 1.10 1.9 0.396 1.3 1.9
HC4 0.68 1.20 0.8 1.15 2.2 0.528 1.4 2.2
The equations for the dynamic factor φ2 as well as the values of φ2,min
and β2 are similar for all the codes. The effect of the differences is shown in
Figure 2.3 which shows the values of the dynamic factor φ2 for the different
codes over a range of lifting speeds ranging from 1 m/min to 120 m/min. The
largest difference in values between the codes is in the region of 1 - 20 m/min
which is the most common range of hoisting speeds. Enlarged graphs showing
the dynamic factors φ2 for a range of hoisting speeds from 0 - 30 m/min are
shown in Figure 2.4. DIN 15018 gives the largest values, AS1418.1-1994 and
ISO 8686-1:1989 have the lowest values and prEN 1991-3 gives intermediate
values.
This spread in the values of the dynamic factors indicate the uncertainty
about the true values of these dynamic factors.
Dynamic factors φ from the codes

2
2.6
Eurocode
Australian
HC4 International
2.4 German
2.2
HC3
2
φ2
1.8
HC2
1.6
1.4 HC1
1.2
1
0 20 40 60 80 100 120
Steady hoisting speed in m/min
Figure 2.3: Dynamic factor φ2 values for four design codes
Dynamic factors φ2 from the codes, for 0 ≤ vh ≤ 30 m/min, HC1 Dynamic factors φ2 from the codes, for 0 ≤ vh ≤ 30 m/min, HC2
Eurocode Eurocode
Australian Australian
1.6 International 1.6 International
German German
1.5 1.5
1.4 1.4
φ2
φ2
1.3 1.3
1.2 1.2
1.1 1.1
1 1
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Steady hoisting speed in m/min Steady hoisting speed in m/min
(a) Hoist Class 1 (b) Hoist Class 2

Dynamic factors φ2 from the codes, for 0 ≤ vh ≤ 30 m/min, HC3 Dynamic factors φ2 from the codes, for 0 ≤ vh ≤ 30 m/min, HC4
Eurocode Eurocode
Australian Australian
1.6 International 1.6 International
German German
1.5 1.5
1.4 1.4
φ2
φ2
1.3 1.3
1.2 1.2
1.1 1.1
1 1
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Steady hoisting speed in m/min Steady hoisting speed in m/min
(c) Hoist Class 3 (d) Hoist Class 4
Figure 2.4: Dynamic factor φ2 values for four design codes for 0 ≤ vh ≤ 30 m/min
(c) Dynamic effects caused by sudden release of part of the hoist-

load
prEN 1991-3, ISO 8686-1:1989, AS1418.1-1994 and DIN 15018 consider the
dynamic effects of the situation when a crane suddenly releases part of, or the
whole of, the hoistload. This occurs in cranes which use grabs or magnets to
lift loads and often release them suddenly as part of the normal operation of
the crane.
prEN 1991-3, ISO 8686-1:1989 and AS1418.1-1994 specify a dynamic factor
φ3 to be applied to the hoistload to allow for the dynamic effects of suddenly
releasing a load. The equation given in these three codes is:
∆m
φ3 = 1 − (1 + β3 ) (2.3.7)
m
Where:
∆m – released part of the load
m – total hoisted load
β3 = 0.5 for cranes with slow release devices like grabs
β3 = 1.0 for cranes with rapid release devices like magnets
The dynamic factor φ3 will always be less than one and, like the lower
values of φ1 , is not generally a critical consideration for EOT bridge cranes
but more for tower or jib cranes for instability considerations.
DIN 15018 specifically states that this load situation is only for jib cranes
and recommends a value of:
φ3 = −0.25φ2 (2.3.8)
(d) Dynamic effects caused by the crane travelling on the rails

prEN 1991-3 and ISO 8686-1:1989 adopt the same approach with regards to
the dynamic effects caused by the crane travelling at a constant speed along
the rails. Both codes recommend applying a dynamic factor φ4 to the crane
self weight and hoistload to allow for the dynamic effects of travelling along
uneven rails. The value of φ4 depends on the unevenness of the rails and both
codes state that if the tolerances of the rails meet specifications the value of
φ4 may be taken as 1.0.
Rails which have a vertical or horizontal step or gap at the rail splices can
induce large dynamic forces as the crane travels over the joint. These forces
may lead to a fatigue failure in the web of the crane girder at the weld to the
flange or stiffener, even at a relatively low number of cycles. A step or gap in
the rail may also lead to increased wear of the crane rails or wheels
In the event that the rail tolerances do not meet given standards prEN
1991-3 refers to an alternative model in EN 13001-2 [27], the standard for the
design of cranes. ISO 8686-1:1989 contains a model for the calculation of φ4
for the crane travelling over either a gap or a step in the rail. The model is
based on elasto-kinetic principles.
AS1418.1-1994 and DIN 15018 apply a dynamic factor φ1 to the crane self
weight to allow for the inertial forces caused by movement of the crane or
crane components. The value of φ1 depends on the type of wheel, type of
wheel suspension, type of runway, condition of runway and travel speed of the
crane. Tables 2.7 & 2.8 are given in the codes for the determination of φ1 .
DIN 15018 states that in the case of spring suspended wheels running on rails
φ1 may be taken as 1.1 regardless of the travelling speed or type of runway.
Table 2.7: Determination of φ1 in AS1418.1-1994
Dynamic multiplier φ1
Type of Condition Wheel Suspension Travel velocity, m/s
runway of runway type type 0 ≤1.0 >1.0 >1.5 >3.5
≤1.5 ≤3.5
Steel Smooth Unsprung 1.0 1.1 1.1 1.2 1.2
rails welded Steel
or continuous Sprung 1.0 1.1 1.1 1.1 1.1
beams Joints ≤ Steel Unsprung 1.0 1.1 1.2 1.2 1.2
4 mm wide Sprung 1.0 1.1 1.1 1.1 1.1
Concrete No joints Rubber Sprung 1.0 1.1 1.1 1.1 1.1
Jointed Rubber Sprung 1.0 1.2 1.2 1.25 1.25
Roadway Rubber Sprung 1.0 1.1 1.1 1.15 1.15
or flexible — Crawler Sprung 1.0 1.1 1.2 1.25 1.25
pavement tracks
The value of φ1 that would most commonly be used is φ1 = 1.1. This is the
same value given in prEN 1991-3 and ISO 8686-1:1989 for the factor applied to
the crane self weight to allow for the dynamic effects on the crane self weight
of lifting a load off the ground, although the rationale behind the factors is
different.
Table 2.8: Determination of φ1 in DIN 15018

Travelling speed in m/min
Runways Self
without weight
with rail joints rail joints factor
or irregularities or with welded φ1
(road) and machined
rail joints
Up to 60 Up to 90 1.1
Over 60 up to 200 Over 90 up to 300 1.2
Over 200 — ≥ 1.2
2.3.1.2.3 Horizontal crane loads due to inertial effects
As cranes accelerate or travel along the runway, horizontal transverse and

longitudinal wheel loads are developed due to eccentric masses or the direction
of long travel motion of the crane not coinciding with the longitudinal axis of
the rails. The different horizontal load situations which are taken into account
in the codes are discussed below.
(a) Generic horizontal loads

ASCE 7-98 specifies only general horizontal transverse and longitudinal loads
without giving any indication of the crane load situation which is being mod-
elled.
The transverse forces are taken to be 20% of the sum of the weights of the
hoistload and the crab. The forces act in either direction perpendicular to the
runway beam and are to be distributed taking into account the stiffness of the
runway beams and supporting structure.
The longitudinal forces are taken to be 10% of the maximum wheel loads
of the crane and act in either direction parallel to the beam.
Because of these specifications being all that is given in ASCE 7-98 with
regards to horizontal loads, this code will not be considered in the further
discussions.
(b) Acceleration and braking of the crane bridge

SABS 0160:1989 specifies horizontal longitudinal forces only for the case of
acceleration or braking of the crane bridge. The longitudinal force on each rail
is specified to be 10% of the sum of the maximum static wheel loads on that
rail.
ISO 8686-1:1989 recommends, in the main body of the code, the use of
a rigid body kinetic model to calculate the loads due to the acceleration and
braking of the crane bridge, taking into account the geometry and mass dis-
tribution of the drives and the crane. A dynamic factor φ5 is applied to allow
for the elastic effects of the drives. The values of φ5 range from 1.0 to 2.0
depending on the smoothness of change of the drive forces. A detailed model
for the calculation of these loads is given in an Appendix of ISO 8686-1:1989.
The model assumes wheel pairs that are fixed on one side and movable on
the other. However, this wheel combination is not common practice amongst
crane manufacturers today, the current practice is crane wheels that are fixed
laterally on both end carriages [16; 17; 18; 19]. The model is a complex fully
kinetic model of the drives and crane structure.
prEN 1991-3, AS1418.1-1994 and DIN 15018 specify transverse and longi-
tudinal loads caused by acceleration and braking of the crane. The longitudinal
loads are caused by the drive forces and the transverse loads are the result of
the crab being positioned eccentric to the centre of mass of the crane.
For independent wheels which are laterally fixed on both sides, the drive
force is calculated, assuming no slip, as the minimum possible wheel loads
multiplied by the friction coefficient between the wheel and rail:
K = µ (n × Vmin ) (2.3.9)
Where:
K – total drive force
µ – friction coefficient
n – number of driven wheels
Vmin – minimum possible wheel load (crab furthest from wheel being consid-
ered, unloaded crane)
The drive force acts at the geometric centre of the crane. The configuration
which results in the maximum transverse loads is the crab at the extreme of its
travel causing the maximum offset between the geometric centre of the crane
and the centre of mass. The drive force then causes a moment about the centre
of mass of the crane which is resisted by couples acting on the wheels of each
end carriage. Figure 2.5 shows the configuration of the crane and the resulting
forces.
Figure 2.5: Wheel loads due to acceleration of crane bridge
The values of the horizontal transverse loads are calculated using the equa-
tion below:
Kls
HT,1 = φ5 ξ2
a (2.3.10)
Kls
HT,2 = φ5 ξ1
a
Where:
HT,i – horizontal transverse force on wheels on rail i
φ5 – dynamic factor to take into account the dynamic effects of the change of
drive forces
ξi – distance from rail i to centre of mass of the crane
ls – distance between geometric centre and centre of mass of the crane
a – distance between the guide rollers or flanged wheels
The value of the dynamic factor φ5 in DIN 15018 and AS1418.1-1994 is

given as 1.5 on condition that the drives are free from backlash.
Different values of the dynamic factor φ5 are given in prEN 1991-3 depend-
ing on the type of drive and the behaviour of the drive. The values are shown
in Table 2.9.
Table 2.9: Values of the dynamic factor φ5 in prEN 1991-3

φ5 = 1.0 centrifugal forces
1.0 ≤ φ5 ≤ 1.5 systems in which drive forces change smoothly
1.5 ≤ φ5 ≤ 2.0 systems in which drive forces change suddenly
φ5 = 3.0 for drives with considerable backlash
(c) Acceleration and braking of the crab

SABS 0160:1989 specifies the total force due to the acceleration and braking of
the crab as equal to the weight of the hoistload and crab multiplied by a factor
depending on the class of the crane. The factors are given in Table 2.10. It is
recommended in the code that the load be divided among all the wheels of the
crane taking into account the transverse stiffness of the crane rail supports.
Table 2.10: Factors from SABS 0160:1989 for the loads due to acceleration and
braking of the crab
Class of crane Factor
Class 1 0.05
Class 2 0.10
Class 3 0.15
Class 4 0.20
prEN 1991-3 states that the transverse forces due to the acceleration or
braking of the crab are taken into account by the buffer forces resulting from
the accidental situation where the crab runs into the end stops on the end
of the crane bridge. This approach is problematic in that it is a conservative
estimate of the forces that would be caused due to the acceleration or braking
of the crab and its use in fatigue calculations, for example, could lead to over
conservative designs.
ISO 8686-1:1989 recommends the same method for calculating the forces
due to the acceleration and braking of the crab as those due to the accelera-
tion and braking of the crane, i.e. the rigid body kinetic model given in the
appendix of the code.
AS1418.1-1994 and DIN 15018 also take the same approach to the calcu-
lation of the loads due to acceleration or braking of the crab as to those due
to the acceleration or braking of the crane. The drive forces are calculated,
assuming no slip at the wheel-rail interface, by multiplying the friction factor
by the minimum possible crab wheel loads. The resulting transverse force is
distributed between the crane wheels taking into account whether they are
fixed/fixed wheel pairs (divide the total drive force by the number of wheels)
or fixed/movable wheel pairs (divide the total drive force by the number of
fixed wheels).
(d) Misalignment of crane wheels or gantry rails

SABS 0160:1989 takes into account the situation where there is misalignment
of either the crane wheels, in a ‘toe-in’ or ‘toe out’ manner, or gantry rails
which causes horizontal transverse forces on each side of the crane tending
to either pull the rails together or push them apart. These forces are due to
horizontal friction forces induced by the crane wheels rolling at an angle to
the longitudinal axis of the rail. Wheels or rails which are aligned according
to specified tolerances may still be at an angle sufficient to cause these mis-
alignment loads so this load case does fall under the ‘loads arising from normal
operation of the crane’. The value of the misalignment load for each wheel is
calculated according to the equation given below.
XM
HT,m = (2.3.11)
N
Where:
HT,m – horizontal transverse load on each wheel caused by misalignment of

wheels or rails
X – factor depending on the class of the crane, given in Table 2.11.
M – combined weight of the crane bridge, crab and hoistload
N – total number of crane wheels
prEN 1991-3, ISO 8686-1:1989, AS1418.1-1994 and DIN 15018 do not give
a load model for the calculation of loads due to misalignment of crane wheels
or gantry rails.
Table 2.11: Factors from SABS 0160:1989 for the loads due to misalignment of
crane wheels or gantry rails
Class of crane Factor
Class 1 0.05
Class 2 0.12
Class 3 0.15
Class 4 0.20
(e) Skewing of the crane in plan

Skewing of the crane in plan can be caused by a number of factors such as the
drive force on one side being larger than on the other side, one motor failing,
applying brakes too suddenly so one side takes before the other or different
diameters of wheels on either side of the crane. In short, any difference between
the long travel speed on one side of the crane from the other can cause the
crane to skew in plan thereby causing transverse forces to be applied to the
runway.
SABS 0160:1989 specifies loads caused by skewing of the crane in plan.
Different methods are given for cranes guided by guide rollers and those guided
by wheel flanges.
In the case of a crane guided by wheel flanges, the load applied to each
wheel is taken as 1.5 times the load due to misalignment of the wheels or rails.
The loads are applied as couples on the wheels of each end carriage.
In the case of a crane guided by guide rollers, the loads applied as a couple
to the guide rollers should result in a couple 1.3 times the couple that would
have been produced by the skewing forces on one end carriage if the crane
did not have guide rollers. The note in the code states that it is not known
precisely what the forces imposed on guide rollers are and the recommendation
given is empirically based and conservative.
prEN 1991-3, ISO 8686-1:1989, AS1418.1-1994 and DIN 15018 all have the
same basic load model for skewing. The model assumes a crane travelling at
a constant speed and rotating about an ‘instantaneous centre of rotation’ to
an angle to the runway. A guide force develops on the front guidance means
which comes into contact with the rail and tries to straighten the crane. This
guide force is then resisted by horizontal transverse forces on the wheels which
are due to the transverse component of friction as the wheels slide horizontally.
The model takes into account the angle of skew and the configuration of the
wheels.
The guide force and transverse forces due to skewing are calculated as given
below:
S = f λs,j ΣQr
(2.3.12)
HS,i,j,k = f λs,i,j,k ΣQr
Where:
S – guide force on the front guidance means
HS,i,j,k – transverse skewing forces on the wheels
f – friction coefficient in the transverse direction
λs,i,j,k – skewing force factors
ΣQr – total weight of the crane and hoistload
i – rail i
j – wheel pair j
k – direction of the force, T = transverse, L = longitudinal
Figure 2.6 shows the configuration of the skewing loads.
Figure 2.6: Configuration of skewing loads

prEN 1991-3 gives the most guidance on the skewing angle α, recommend-
ing that three aspects of skewing be taken into account: initial skewness from
slight rail and wheel misalignment, clearance between rail and guidance means
and wear of rail and guidance means.
The component of the friction coefficient in the transverse direction is
stated by ISO 8686-1:1989 to be empirically based and is given by all four
codes as:
f = 0.3 (1 − exp (−250α)) ≤ 0.3 (2.3.13)
The possible configurations of wheel pairs for EOT cranes are either cou-
pled or independent with either movable or fixed wheels. A wheel pair is
considered to be a pair of wheels, opposite each other, one on each end car-
riage, refer to Figure 2.6. Coupled wheel pairs are wheel pairs whose rotation
is either electrically or mechanically synchronised, as in the case of central
wheel drives. Independent wheel pairs are not linked together in any way and
are free to rotate independently of each other. Movable wheels have a lateral
degree of freedom, i.e. they are able to move horizontally transverse to the rail
direction and are thus unable to transfer any lateral forces to the rails. Fixed
wheels do not have this lateral degree of freedom but are horizontally fixed on
their axes.
The combinations of coupled or independent and movable or fixed wheels
which are considered in prEN 1991-3 for the skewing load model are:
• CFF: Coupled, fixed - fixed: the wheels pairs are coupled and none of
the wheels can translate horizontally on their axes.
• CFM: Couple, fixed - movable: the wheel pairs are coupled and the
wheels on one end carriage are fixed on their axes while the wheels on
the other end carriage have a horizontal transverse degree of freedom.
• IFF: Independent, fixed - fixed: the wheel pairs are independent, the
wheels on both end carriages are fixed on their axes.
• IFM: Independent, fixed - movable: the wheel pairs are independent and
the wheels on one end carriage are fixed on their axes while the wheels
on the other end carriage have a horizontal transverse degree of freedom.
Current common practice is to have independent, fixed - fixed wheels [16;

17; 18; 19], however, the load models for the other configurations of wheels will
be presented here as well. Tables 2.12 & 2.13 give the equations for the distance
from the front guidance means to the instantaneous centre of rotation and the
factors for the determination of the guide force and transverse skewing forces.
Figure 2.7 shows the configuration of the skewing loads for a four wheeled crane
guided by wheel flanges for the four different wheel combinations identified in
Table 2.12.
Table 2.12: Distance to instantaneous centre of rotation for skewing

Combination of wheel pairs h
mξ1 ξ2 l2 +Σe2j
Continuous or independent, fixed/fixed (CFF, IFF) Σej
mξ1 l2 +Σe2j
Continuous or independent, fixed/movable (CFM, IFM) Σej
Table 2.13: Force factors for skewing forces

Wheels λs,j λS,1,j,L λS,1,j,T λS,2,j,L λS,2,j,T
Σej ξ1 ξ2 l ξ2 ¡ ej ¢ ξ1 ξ2 l ξ1 ¡ ej ¢
CFF 1− nh nh n 1− h nh n 1− h
Σej ξ2 ¡ ej ¢ ξ1 ¡ ej ¢
IFF 1− 0 n 1− h 0 n 1− h
³ nh ´
Σe ξ1 ξ2 l ξ2 ¡ ej ¢ ξ1 ξ2 l
CFM ξ2 1 − nhj nh n 1− h nh 0
³ ´
Σej ξ2 ¡ ej ¢
IFM ξ2 1 − nh 0 n 1− h 0 0
Where:
h – distance from the front guidance means to the instantaneous centre of

rotation
ej – distance of wheel pair j from the front guidance means
l – span of the crane
m – number of coupled wheel pairs
n – number of wheel pairs

(a) Continuous fixed/fixed (b) Continuous fixed/movable
(c) Independent fixed/fixed (d) Independent fixed/movable
Figure 2.7: Configuration of skewing forces for different wheel combinations
2.3.1.2.4 In-service wind
Cranes on outside gantries will be subject to wind forces during crane ser-
vice. prEN 1991-3, ISO 8686-1:1989, DIN 15018 and AS1418.1-1994 all make
provision for these loads when considering load combinations and refer to the
relevant parts of their set of standards for the calculation of the in-service wind
loads.
2.3.1.2.5 Test Loads
Cranes are required to undergo testing by the lifting of a test load. There are
two types of load tests [16]:
• Tests carried out on cranes after initial installation, refurbishment, up-

grading or modification. The test consists of loading the crane to 125%
safe working load, lifting the load and lowering it with the crane station-
ary.
• Tests carried out on all cranes each year. The test consists of loading the
crane to 110% safe working load and moving the load in the manner of
normal operation of the crane, i.e. all motions: long travel, cross travel
and hoisting.
SABS 0160:1989 and ASCE 7-98 do not mention the test loads.
prEN 1991-3 specifies a dynamic test load of 110% safe working load and a
static test load of 125% safe working load. For design, the dynamic test load is
assumed to be moved by the drives in the way in which the crane will be used
and is multiplied by a dynamic factor φ6 to allow for vertical inertial effects
of lifting the load. The same dynamic factor for the test loads is also used in
ISO 8686-1:1989, AS1418.1-1994 and DIN 15018 and is calculated by:
1 + φ2
φ6 = (2.3.14)
2
ISO 8686-1:1989 specifies that the values of the test loads should be taken
from ISO 4310 and that in the case of dynamic test loads the load should be
multiplied by the dynamic factor φ6 .
AS1418.1-1994 does not state the value of the test load or a reference but
merely states that the value of the load should be appropriate for the crane
and that the dynamic load should be multiplied by the dynamic factor φ6 .
DIN 15018 specifies a small test load of 125% safe working load as a dy-
namic test load to be multiplied by the dynamic factor φ6 and a large test load
of either 133% safe working load for hoist class one or two cranes or 150% safe
working load for hoist class three and four cranes as a static test load.
2.3.1.3 Loads arising from accidental situations
Accidental crane load situations can arise from the movement of the crane or
the hoistload. The various types of accidental loads are discussed below.
2.3.1.3.1 Crane running into the end stops
This situation occurs when the crane is allowed to run into the end stops on the
end of the runway. This is classified as an accidental situation because most
cranes are provided with limit switches on the runway which are designed to
stop the crane before it collides with the end stops. Buffers are provided on
the crane to lessen the impact force. This accidental situation is provided for
in the codes.
SABS 0160:1989 recommends taking for the force on each end stop, assum-
ing that the hoistload is free to swing, the lower value of:
(a) A force equal to the self weight of the crane, i.e. crane bridge and crab.
(b) A force calculated assuming the crane to be travelling at full rated speed,
taking into account the resilience of the end stops and buffers.
prEN 1991-3 and ISO 8686-1:1989 take the same basic approach in the
determination of the buffer forces due to crane movement. They state that
the buffer forces are to be calculated considering the kinetic energy from all
parts of the crane moving at 70% - 100% of the long travel speed, taking into
account the distribution of mass and the buffer characteristics. prEN 1991-3
gives the following equation for the calculation of the buffer forces on each end
stop:
p
HB,1 = φ7 v1 mc SB (2.3.15)
Where:
HB,1 – force on each end stop
φ7 – dynamic factor taking into account non-linearity of buffer
v1 – 70% of the full long travel speed in m/s
mc – mass of the crane and hoistload in kg
SB – spring constant of the buffer in N/m
ISO 8686-1:1989 differs from prEN 1991-3 in that it states that the effects
of hoistloads which are free to swing need not be included.
The value of the dynamic factor φ7 depends on the non-linearity of the
buffer which is represented by the buffer characteristic ξ. ISO 8686-1:1989
refers to the buffer characteristic as the relative buffer energy and defines it
as:
Z umax
1
ξ= Fx du (2.3.16)
Fmax umax 0
Where:
Fmax – buffer maximum force
umax – buffer maximum deflection

R umax
0 Fx du – area under buffer force displacement curve, shown in Figure 2.8
φ Fmax
1.60
1.25
Fx
umax
0.5 1.0 ξ u
Figure 2.8: Buffer force-displacement characteristic curve
prEN 1991-3 recommends values of the dynamic factor φ7 as depicted in Figure

2.8 and given by the following equations:
φ7 = 1.25 if 0 ≤ ξ ≤ 0.5
φ7 = 1.25 + 0.7 (ξ − 0.5) if 0.5 ≤ ξ ≤ 1.0
Where:
ξ – degree of non-linearity of buffer, ratio of the area under the force dis-
placement curve to the product of the maximum force and maximum
displacement. Refer to Figure 2.8 for a graphical representation of the
buffer characteristic ξ.
AS1418.1-1994 and DIN 15018 give the same recommendations for the
calculation of buffer forces due to crane movement. The buffer forces are
calculated assuming the crane to be moving at 85% to 100% of its full long
travel speed except in the case where a reliable system is provided to slow the
crane down, such as limit switches, in which case a speed of 70% of the full
long travel speed may be used. It is stated that hoistloads that are free to
swing need not be taken into consideration.
The buffer forces are calculated on the basis of the kinetic energy of the
crane before collision with the buffer. The same recommendations as in prEN
1991-3 and ISO 8686-1:1989 are given regarding the dynamic factor φ7 .
2.3.1.3.2 Crab running into the end stops
This load situation occurs when the crab runs into the end stops on the end
of the crane bridge in a similar way to the crane running into the ends stops
on the end of the runway.
SABS 0160:1989 does not make provision for this load situation.
prEN 1991-3 specifies that the total buffer force relating to crab movement,
in the case when the hoistload is free to swing, shall be taken as 10% of the
weight of the crab and hoistload.
ISO 8686-1:1989, AS1418.1-1994 and DIN 15018 do not make specific pro-
vision for this load situation but include it under the buffer forces due to crane
movement and recommend that the same procedure be used for both the crane
and crab.
2.3.1.3.3 Crane tilting
The crane can tilt if the hoistload is rigidly fixed, for example on a mast, and
collides with an obstacle.
SABS 0160:1989 and AS1418.1-1994 do not consider this load situation.
prEN 1991-3 and ISO 8686-1:1989 merely state that in the case when tilting
of the crane can occur, the resulting static forces should be considered and the
dynamic effects of the crane falling back onto the rails should be taken into
account.
DIN 15018 gives more guidance on this load situation by stating that a
horizontal load shall be applied at the obstacle level with the crab in the most
unfavourable position.
2.3.1.3.4 Wind blowing crane onto end stops
Cranes on outside gantries or in buildings which are not fully clad are subject to
wind forces. There have been several cases of cranes without an appropriate
anchorage system or without storm brakes activated, where the wind forces
have exceeded either the braking force or the friction force between the rails
and the wheels and the crane has travelled along the runway, through the end
stops and fallen to the ground [18; 20].
Out-of-service wind is mentioned in ISO 8686-1:1989, AS1418.1-1994 and
DIN 15018 but no mention is made of this specific consequence of the wind
loads which could possibly be the reason why many cases of this type of failure
have been observed.
A crane running off the end of the runway has financial and safety impli-
cations and is therefore a highly undesirable occurrence.
Two approaches are possible to prevent this behaviour of the crane. The
first is to anchor the crane so that it is unable to move due to the wind forces.
This anchorage system should be easy to apply to ensure that it is actually

used when the crane is out of service. The second approach is to ensure
that the structural elements which could prevent this crane behaviour have
sufficient resistance to withstand the loads. The structural elements which
resist these loads are the end stops on the ends of the runway and longitudinal
runway bracing. The cost of these elements in comparison to the cost of the
entire gantry is relatively small so a small increase in cost of the crane support
structure would result in a significantly lower risk of financial losses and safety
hazards.
2.3.1.3.5 Hoistload colliding with column
In the case when the hoistload is free to swing and the crab suddenly stops
as it is traversing, due to inertia effects, the hoistload could swing and collide
with a crane column or other structural member. This accidental situation is
not considered in any of the codes.
This type of accidental situation is discussed by Lynch [28] who addresses
it from a crane operator training point of view. This raises the issue of whether
the responsibility for the functioning and safety of the crane support structure
under these loads lies with the crane operator preventing these situations or
with the structural engineer designing a structure strong enough to withstand
loads from these situations. The answer lies in the distinction between ac-
cidental loads and loads due to abuse of the structure. Accidental loads are
loads which should not occur but occasionally due to unforseen circumstances
do occur. Loads due to abuse of the structure occur through the fault of the
occupant of the structure who deliberately uses the structure for a purpose for
which it was not originally intended. It is prudent for structural engineers to
design buildings against accidental loads but the design of a structure for abuse
would in many cases be uneconomical. Loads due to abuse of the structure
are discussed further in the next section.
2.3.1.3.6 Failure of mechanisms or emergency cut-out
In the case where cranes are supplied with emergency brakes as well as service
brakes, failure of a component or an emergency cut-out causes the application
of the emergency brakes. ISO 8686-1:1989 and AS1418.1-1994 recommend
that the provisions for acceleration or braking of the crane or crab be used to
determine these forces.
2.3.1.4 Loads arising from improper construction, lack of main-

tenance or misuse of the crane
There are many loads which can be imposed by the crane on its support
structure resulting from improper construction, lack of maintenance of the
support structure or misuse of the crane.
The main issue in the proper construction and maintenance of a EOT
crane support structures is the alignment of the rails, this consists of two
parts: alignment of the rails with respect to the rest of the support structure
and alignment of the rails with respect to each other and the intended rail
position. In the construction phase the correct alignment starts at floor level
and works upwards with the correct alignment of the column, girders on the
columns and rails on the girders. If the alignment is correctly carried out during
construction then the maintenance of the rail alignment consists of realigning
the rails on the girders with respect to one another and the intended rail
position.
Tolerances on rail alignment in South Africa are given by SAISC [29].
Experience shows that correct rail alignment is difficult to achieve and many
crane support structures have rails which are not aligned to these tolerances.
Misaligned rails cause wear on the tracks and wheel flanges and place stress
on the rail connections.
Misaligned rails may cause an increase in the frequency of skewing of the
crane and misalignment type of crane loads where the crane bridge is either
trying to pull the rails together or force them apart as provided for in SABS
0160:1989.
Rails that are misaligned on the girders cause torsion in the girder and in
severe cases of rails being misaligned with respect to the centre of the web,
rotation of the top flange and bending of the top of the web can result. These
are load effects that are not generally taken into account during the design of
the girders and could therefore cause an ultimate limit state type of failure or
increase the fatigue damage considerably.
Girders which are not properly aligned on the columns or corbels could
cause larger bending moments in the columns than those taken into account
during the design.
There are many ways in which the crane itself could be misused thereby
leading to loads not considered in the design. Examples of such misuse are
oblique hoisting of a load, dragging a load along the ground, routinely running
the crane into the end stops to straighten it, allowing a hoistload to swing
into a column or turning off the overload limit switch to pick up loads larger
than the safe working load [20]. All of these misuses subject the crane support
structure to much larger loads that it was originally designed for and could
cause failures or shorten the fatigue life of the structure.
The question raised is: how much should a structural engineer design an
EOT crane support structure for loads resulting from these types of improper
construction, maintenance and misuse of the crane? This question can be
taken further to ask: how far should design codes go to make provision for this
sort of load?
The initial response to questions like this is that structures are generally
not designed for abuse and the construction of a structure to withstand any
possible abuse that may occur would be prohibitively expensive. The owner
or occupant of a structure should take the responsibility to ensure that the
structure is not subject to these types of abuses. In the case of crane structures
this would entail strict control and monitoring of the construction phase to
ensure that alignment tolerances were met and a routine maintenance plan to
ensure that rails remained aligned within the tolerances. The second aspect
of responsible use of crane support structures is the correct training, and/or
supervision, of crane operators to ensure that the crane is not misused.
Experience shows, however, that correct rail alignment is often not achieved
and that misuse of cranes does occur. In the light of this a structural engineer
may choose to include additional load cases for these situations to be on the
conservative side, however, what should be the response of the code writing
committee? Three options are available:
1. State that the load models presented in the code are based on the as-
sumption that the alignment tolerances are met and that the crane is
used in the manner for which it was intended.
2. Suggest regular inspection and maintenance to ensure rail alignment and

crane operator training to ensure that the crane is not misused.
3. Recommend load models for rails misaligned outside the tolerances or

the load situations occurring from crane misuse.
Of the codes that were considered here, the only code which makes reference
to the correct alignment of the rails to specified construction tolerances is
DIN 15018 which states that unless otherwise specified, the loads shall be
calculated on the assumption that the runway has been properly horizontally
and vertically aligned thereby following option 1.
The question of whether design codes should make provision for lack of
maintenance or misuse of the structure, even in the event that it is common
knowledge that these occur, is a contentious issue and no firm answer has been
reached.
2.3.1.5 Additional loads
Some additional loads can act on crane support structures and are taken into
account by the various codes as shown in Table 2.14.
Table 2.14: Additional loads on crane support structures

Additional load prEN ISO DIN AS
1991-3 8686 15018 1418
Loads due to displacements X X
Temperature variation X X X X
Snow and ice X X X
Off-vertical hoisting X
Seismic loads X X
Construction loads X X
Loads on walkways X X X X
2.3.1.6 Summary of crane loads considered by the codes
Table 2.15 below gives a summary of the basic crane loads that are taken into
account by the six codes that were considered.
Table 2.15: Summary of crane loads considered by the codes

Crane load SABS ASCE prEN ISO DIN AS
0160 7-98 1991-3 8686-1 15018 1418
Vertical
Generic impact X X
Hoisting:
effect on self weight X X
effect on hoistload X X X X
Running on rails X X X X
Release of payload X X X X
Test loads X X X X
Horiz transverse
Generic transverse X
Acceleration of crane X X X X
Acceleration of crab X X X X
Skewing X X X X X
Misalignment X
Crab buffer forces X X X X
Horiz longitudinal
Generic longitudinal X
Acceleration of crane X X X X X
Crane buffer forces X X X X X
2.3.1.7 Combinations of vertical and horizontal crane loads
Crane load combinations consist of vertical crane loads combined with hori-
zontal crane loads to be considered as one characteristic crane action for com-
bination with other actions such as wind.
SABS 0160:1989 gives little guidance on the combinations of crane actions
to be considered, merely stating that the horizontal transverse forces need
not be considered to act together. This results in three ultimate limit state
combinations and one accidental combination:
Ultimate limit state:
1. Vertical with impact + Acceleration or braking of crab + Acceleration

or braking of crane
2. Vertical with impact + Misalignment of wheels or rails + Acceleration

or braking of crane
3. Vertical with impact + Skewing + Acceleration or braking of crane

Accidental limit state:
1. Vertical without impact + End stop forces
ASCE 7-98 does not mention combinations of crane actions but as there is
only one horizontal transverse force and one horizontal longitudinal force it is
assumed that these will be combined resulting in only one load combination:
Ultimate limit state:
1. Vertical with impact + horizontal transverse + horizontal longitudinal
prEN 1991-3 provides a table specifying load combinations resulting in

seven ultimate limit state combinations, two accidental combinations and one
combination for test loads. The load combination table from prEN 1991-3 is
given in Table 2.16. The various individual crane loads are given in the first
column and the load combinations are specified in the succeeding columns. The
φ symbols represent the dynamic factors applicable for each load combination.
Table 2.16: Crane load combinations

Groups of loads
Test Acci-
ULS load den-
tal
1 2 3 4 5 6 7 8 9
1 Self weight of crane φ1 φ1 1 φ4 φ4 φ4 1 φ1 1
2 Hoist load φ2 φ3 - φ4 φ4 φ4 η1 - 1
3 Acceleration of crane φ5 φ5 φ5 φ5 - - - φ5 -
bridge
4 Skewing of crane - - - - 1 - - - -
bridge
5 Acceleration or brak-
ing of crab or hoist - - - - - 1 - - -
block
6 Test load - - - - - - - 1 -
7 Buffer force - - - - - - - - φ7
1 η is the part of the hoist load that remains when the payload
is removed, but is not included in the self weight of the crane
From an inspection of the load combinations recommended by prEN 1991-3,

it was observed that of the combinations given as ultimate limit state combi-
nations, for overhead travelling bridge cranes, only load combinations 1, 5 and
6 are likely to be critical. Load combination 2 considers the case when the
crane releases part of the load where the hoistload value is multiplied by the
dynamic factor φ3 which was shown earlier to be always less than one. Load
combination 3 is the same as load combination 1 except that an unloaded
crane is considered. Load combination 4 considers the case when the crane
is accelerating on the runway with a load, and as φ4 is usually equal to one
and is therefore less than both φ1 and φ2 , this load combination would have
smaller vertical loads than load combination 1. Load combination 7 considers
a stationary unloaded crane with no horizontal forces and would therefore have
smaller loads than the other load combinations.
The conclusion drawn from this observation is that the load combinations
which are not critical for the ultimate limit state, are given for the purposes
of fatigue calculations.
ISO 8686-1:1989 classifies crane loads as regular, occasional or exceptional.

Three types of load combinations are considered: combinations of regular
loads, combinations of regular and occasional loads and combinations of reg-
ular, occasional and exceptional loads. DIN 15018 classifies crane loads into
main loads, additional loads and special loads and load combinations as nor-
mal or special. AS1418.1-1994 takes a similar approach to ISO 8686-1:1989
and DIN 15018 by classifying the crane loads as principal loads, additional
loads and special loads and the load combinations as frequently, infrequently
and rarely occurring combinations.
For ISO 8686-1:1989 and AS1418.1-1994 the regular or frequent load com-
binations consist of combinations of crane loads which result from normal
operation of the crane such as acceleration of the crane and hoist drives, cor-
responding to prEN 1991-3 load combinations 1 - 4, 6 and 7. The occasional
and infrequent combinations include skewing, corresponding to prEN 1991-3
load combination 5. DIN 15018 considers the regular and occasional or fre-
quent and infrequent combinations together as the normal combinations. The
exceptional, special or rare load combinations include test loading, tilting, end
stop forces etc. corresponding to prEN 1991-3 load combinations 8 and 9.
ISO 8686-1:1989, AS1418.1-1994 and DIN 15018 also recommend load com-
binations including the additional loads such as deformations, snow and ice,
temperature etc.
2.3.2 Fatigue
Crane support structures are subject to fatigue due to the cyclic nature of
crane loading as the crane lifts and lowers loads and moves around the building.
Many problems that are encountered with crane support structures in practice
are fatigue problems, indicating that correct fatigue design is an important
consideration.
Fatigue damage is unlike the ultimate limit states where the loading and
resistance of the structure can be treated separately. In the assessment of
fatigue damage a knowledge of both the fatigue loads and the fatigue resistance,
normally in the form of S-N curves, is required.
The question raised then is should crane loading design codes give guidance
on loads to be considered for fatigue or should the onus be on the designer to
ensure that fatigue has been adequately taken into account.
Varying degrees of guidance on fatigue loads for crane support structures
are given in the design codes which have been considered. SABS 0160:1989
and ASCE 7-98 do not give any guidance on the loads to be considered for
fatigue. DIN 15018 discusses fatigue but from the point of view of allowable
stress rather than crane loading. The guidance given in ISO 8686-1:1989 is that
generally the only load combinations taken into account for fatigue verification
are the regular load combinations, which model the hoisting of a load or sudden
release of a load with acceleration of the drives and that, in some cases, in-
service wind, skewing or test loads should be taken into account. AS1418.1-
1994 also states that the fatigue assessment should be carried out considering
only the frequently applied load combinations, which involve hoisting a load,
sudden release of a load and acceleration of drives.
prEN 1991-3 gives a detailed method of fatigue analysis which was derived
from the design code for the design of the crane. This method will be outlined
below.
The basis for the determination of the fatigue loads is the normalisation
of the maximum nominal wheel load resulting in an equivalent constant am-
plitude load for two million cycles. This normalised load is called the ‘fatigue
damage equivalent load’. The normalisation is carried out by multiplying the
maximum nominal wheel load by a factor (λ), as shown below.
Qe = φf at × λ × Qmax (2.3.17)
Where:
Qe – fatigue damage equivalent load
φf at – damage equivalent dynamic impact factor, the factor applied to the

1
crane self weight φf at,1 = 2 (1 + φ1 ) and the factor applied to the
1
hoistload φf at,2 = 2 (1 + φ2 )
Qmax – maximum nominal wheel load
The λ factor is determined from the fatigue class of the crane. Cranes
are classified into fatigue classes depending on the total number of cycles (N)
performed over the lifetime of 25 years and the load spectrum (kQ). One λ
factor is given for each fatigue class for the calculation of normal stresses and
another for the calculation of shear stresses. Table 2.17 shows the λ factors
and Table 2.18 shows the classification table from prEN 1991-3.
Table 2.17: λ factors according to the classification of cranes

Class S0 S1 S2 S3 S4 S5 S6 S7 S8 S9
normal
0.198 0.250 0.315 0.397 0.500 0.630 0.794 1.00 1.260 1.587
stresses
shear
0.379 0.436 0.500 0.575 0.660 0.758 0.871 1.00 1.149 1.320
stresses
The load spectrum definition in prEN 1991-3 is:
X µµ ∆Qi,j ¶m ni,j ¶
kQ = P (2.3.18)
max ∆Qi ni,j
j
Where:
∆Qi,j – load amplitude of range j for wheel i: ∆Qi,j = Qi,j − Qmin,i
max∆Qi – maximum load amplitude for wheel i: max∆Qi = Qmax,i − Qmin,i
ni,j – number of applications of range j for wheel i

P
ni,j – total number of load applications
m – slope of the S-N curve

This is the definition of the load spectrum when fatigue of the crane itself
is being considered. For the fatigue of a simply supported crane girder it was
considered more reasonable to use the values of the wheel loads rather than
the wheel load amplitudes to define the load spectrum. This is because the
load amplitude that is felt by the girder ranges from a load of zero when the
crane is not yet on the girder to a load equal to the wheel load for that given
crane cycle. This will result in a larger load spectrum than that calculated
using the wheel load amplitudes. The suggested load spectrum to take into
account that it is the effect on the girder that is being considered is:
X µµ Qi,j ¶m ni,j ¶
kQ = P (2.3.19)
max Qi ni,j
j
Where:
Qi,j – wheel load of value j for wheel i
maxQi – maximum wheel load for wheel i
The effects of a crane moving along a continuous beam would not be ac-
curately modelled using this definition of the load spectrum. The load effect
in a continuous beam could vary from negative when the crane is on the ad-
jacent span to positive when the crane is on the span being considered. This
highlights the fact that a generic definition of a load spectrum for the fatigue
of a crane support structure element is not as simple as for the crane itself,
due to the variety of possible support structure configurations.
The manner in which the fatigue damage equivalent load is used to assess
the fatigue of an element of the support structure is outlined below.
The fatigue stresses in the support structure element being considered are
calculated by considering one movement of the crane along the girder with the
wheel loads equal to the fatigue damage equivalent load and the stress cycles
causing fatigue are calculated. If the structural element is subject to only one
stress cycle from one pass of the crane, the fatigue resistance is taken as the
stress level at two million cycles on the relevant S-N curve. If more than one
stress cycle is obtained, the parameters of the relevant S-N curve are used to
calculate the accumulative fatigue damage according to Miner’s rule and the
resistance is taken as the fatigue damage at failure equal to one.
Table 2.18: Classification of the fatigue actions from cranes
Class Q0 Q1 Q2 Q3 Q4 Q5
of 0.0313 0.0625 0.125 0.25 0.5
load spectrum kQ ≤ < kQ ≤ < kQ ≤ < kQ ≤ < kQ ≤ < kQ ≤
0.0313 0.0625 0.125 0.25 0.5 1.0
Class of total
number of cycles
U0 N≤ S0 S0 S0 S0 S0 S0
1.6 × 104
1.60 × 104
U1 <N≤ S0 S0 S0 S0 S0 S1
3.15 × 104
3.15 × 104
U2 <N≤ S0 S0 S0 S0 S1 S2
6.30 × 104
6.30 × 104
U3 <N≤ S0 S0 S0 S1 S2 S3
1.25 × 105
1.25 × 105
U4 <N≤ S0 S0 S1 S2 S3 S4
2.50 × 105
2.50 × 105
U5 <N≤ S0 S1 S2 S3 S4 S5
5.00 × 105
5.00 × 105
U6 <N≤ S1 S2 S3 S4 S5 S6
1.00 × 106
1.00 × 106
U7 <N≤ S2 S3 S4 S5 S6 S7
2.00 × 106
2.00 × 106
U8 <N≤ S3 S4 S5 S6 S7 S8
4.00 × 106
4.00 × 106
U9 <N≤ S4 S5 S6 S7 S8 S9
8.00 × 106
The advantage of the fatigue assessment approach given by prEN 1991-

3 is that it is a quick method of carrying out fatigue assessment requiring
the calculation of only one set of stresses and the designer is not required
to consider the various different loads that the crane will lift as well as the
various different crab positions. Being a simplified method though, it should
be conservative.
There are many disadvantages to this approach to fatigue assessment. The
first is the uncertainty in the derivation of the method. As has already been
pointed out in the discussion on the load spectrum, the method still appears
to be based on the fatigue of the crane itself rather than the support structure.
The method is specified to be used with vertical wheel loads and no guid-
ance is given on the method to be used with horizontal wheel loads. No
guidance is given either on the load combinations to be considered for fatigue.
Using this method restricts the freedom of the designer to tailor the fatigue
assessment to a specific crane installation. For example, consider a ladle crane
in a steelworks building moving in a very set pattern, it always lifts a full ladle
at one end of the gantry with the crab at one end of the crane bridge and
travels the full length of the gantry where it brakes and unloads the ladle, it
then moves back the whole length of the gantry and stops in position to refill
the ladle. This crane is always accelerating and braking over the same girders
with the crab at the extreme of its travel and thus a worst-case fatigue situation
is realised for those girders which should be taken into account. In comparison
if a workshop crane is considered, where the movement of the crane is more
random, the girders need not be considered to be subject to the horizontal
loads caused by acceleration of the crane for every crane cycle.
Another disadvantage of this method is that it is less transparent for the
designer and it is difficult to see the relative importance of each different load
case.
2.3.3 Crane - support structure interaction

Most of the load models given in the crane loading codes assume the crane
and the support structure to be rigid. This is not the case in reality and the
issue of crane - support structure interaction could cause additional loads or
different distributions of the wheel loads that are not taken into account by
the codes. It is therefore important that the designer takes possible crane -
support structure interaction into account when carrying out the design, either
in the calculation of the loads or in selecting the structural configuration of

the building.
The structural configuration that would cause the wheel loads to deviate
the most from those given in the codes is one where the runway is not sym-
metrical. If the asymmetry is in the columns, e.g. the columns on one side of
the runway are stiffer than the columns on the other side, the side with the
stiffer columns would be subject to higher lateral loads than the other side.
If the asymmetry is in the girders, e.g. the girder on one side of the runway
has a larger span than the girder on the other side, the difference in vertical
deflections could cause the vertical wheel loads to be distributed unequally. In
the worst case of the girders having unequal deflections, one wheel could lift
off the rails causing an increase in vertical loads on the remaining wheels and
placing large torsion forces on the crane itself.
Crane - support structure interaction causing additional loads on the run-
way can also be caused by the configuration of the crane. In the case where
the crane bridge consists of an I beam supported on top of the end carriage
beams, where the neutral axis of the crane bridge is above the level of the
wheels, when the crane lifts a load, the crane wheels will tend to move out-
wards placing lateral forces on the crane runway.
Most of the codes do not mention the possibility of crane - support struc-
ture interaction, with the exception of SABS 0160:1989 and ASCE 7-98 which
recommend distributing the lateral forces to the crane wheels with regard to
the relative lateral stiffness of the support structure. The onus is therefore
on the support structure designer to make adequate provision for the possible
effects of the crane support structure interaction when distributing the loads
or calculating load effects.
2.3.4 Support structure configurations and details
Buildings housing cranes differ from those without cranes in that they are
subject to dynamic loads and this needs to be taken into account for the de-
sign of the crane girders, columns and connections. Ricker [20] has considered
many aspects of crane support structures, pointing out the crane specific issues
involved in their design and recommending structural configurations to with-
stand these crane specific forces and movements. These issues will be briefly
discussed here.
The correct vertical and horizontal alignment of the girders is an impor-

tant aspect of crane support structures, therefore provision must be made in
the support structure components for the adjustment of vertical and horizon-
tal alignment. Fabrication and construction tolerances need to be taken into
account in allowing for alignment, including inaccuracies in foundations, skew-
ness of columns, rolling or fabrication tolerances of steel members, sweep in
crane beams and fabrication tolerances of the crane itself. The construction
tolerances for South Africa are given by SAISC [29] and adherence to these tol-
erances for all structural members would be critical for the correct alignment
of the crane runway.
The crane girders are the most important part of the crane support struc-
ture and are the members that are most directly loaded by crane loads. The
different types of crane girders are [30]:
1. Wide flange shape, rolled I section
2. Wide flange shape with channel cap
3. Welded wide flange shape with channel as top flange
4. Welded wide flange shape with unequal flanges
5. Welded wide flange shape for narrow clearances
6. Welded wide flange shape with surge plate
7. Welded wide flange shape with surge plate and backup truss
In general simply supported girders are preferable over continuous girders.

Midspan points of continuous girders are subject to stress reversals as the crane
moves over the girder and combinations of maximum bending moments and
shear forces at internal supports which could cause fatigue problems.
In order to select the structural configurations that are appropriate for
crane support structures, an understanding is required of the manner in which
the various crane actions are resisted. The vertical crane loads are resisted
by the vertical crane girder, the transverse crane loads are resisted by the top
flange of the crane girder and the longitudinal loads are resisted by the columns
and the longitudinal runway bracing.
Considering the transverse wheel loads which are resisted by the top flange
of the crane girder. Wide flange shapes with channel caps or unequal flanges
will generally have sufficient resistance for small to medium cranes. Large
cranes will generally require surge plates to resist the transverse crane forces.
The ends of the girders will typically have some type of lateral restraint con-
sisting of a tie-back connection to the column.
The horizontal longitudinal crane loads are applied to the support structure
at the level of the wheel-rail contact. These loads are resisted by the columns
and primarily by the longitudinal runway bracing. The connection of the crane
girder to the column should ensure that the longitudinal forces can be carried
to the bracing system by means of continuity plates on the column cap linking
the bottom flanges of the girders.
The vertical crane wheel loads, which are resisted by the main vertical
crane girder, cause bending of the girder and accompanying rotation of the
girder ends. According to Ricker [20] it is this end rotation which is the major
cause of problems in crane runways because structural configurations which
can accommodate these end rotations are not employed. In order to overcome
these problems, the deflections of the girders should be kept as small as possible
(i.e. short spans and deep beams) and the correct details should be chosen to
handle these deflections.
Tie-back connections from the top flange to the column which have holes
slotted in the longitudinal direction can accommodate the longitudinal move-
ment of the top flange of the girder which results from the vertical deflection.
Details that should be avoided are connections between the ends of adjacent
girders on webs or top flanges and tie-back type of connections from the web
of the girder to the column, both of which will place severe stresses on the
girder due to the end rotations.
Off-centre rails should also be taken into account in the design of the girder.
An off-centre rail causes torsion in the girder and possible top flange rotation
and web bending. The tolerance given by SAISC [29] is that the offset of the
centre of the rail from the centre of the web may not be more than half the
thickness of the web.
The crane columns are also an important part of the crane support struc-
ture in that they provide the support for the crane girders. The different
column configurations are:
1. Bracketed column (with corbel)
2. Step column, solid web

3. Step column, laced web
4. Step column, vierendeel
5. Step column, battened
6. Separate building and crane column
An important point which could easily be overlooked is that in cases where

the crane girders are eccentric to the centre of the columns, the longitudinal
crane loads cause torsion of the columns and column bases.
This is by no means an exhaustive discussion on crane support structure
details and design but highlights a few important issues for the purpose of
demonstrating that special considerations are required for the design of build-
ings containing EOT cranes.
2.3.5 Correct construction
The issue which is the most important when considering construction of crane
support structures is the issue of rail alignment. Misaligned rails could cause
the crane to skew more often and cause more wear on the rails and wheel
flanges. Rails which are badly out of alignment may place additional torsion
on the crane girders and may even cause the crane to get wedged between
the rails thereby placing very large lateral forces on the girders. Misaligned
rails also cause larger lateral forces to be placed on the rail clips. These type
of forces, which are larger than those allowed for in the design codes, could
significantly shorten the fatigue life of the structure.
The correct alignment of the rails starts with the correct alignment of the
column bases and foundations and continues up to the columns and girders and
finally the rails. Allowance should be made in the details of the connections
of the girders to the columns and the rails to the girders for vertical and hori-
zontal alignment. Correct alignment of the rails during the construction phase
and continual maintenance to ensure that the rails stay in alignment could
significantly prolong the lifetime of the structure. Rail alignment tolerances in
the vertical and horizontal directions are given by the SAISC [29].
2.4. Reliability of EOT crane support structures 57
2.4 Reliability of EOT crane support structures

Ensuring sufficient statistical reliability of the EOT crane support structures
is achieved by the application of appropriate partial load factors which are
provided by codes of practice.
The appropriate partial load factors are determined by means of reliability
based code calibration which takes into account the statistical variability in the
loads and resistances, i.e. inherent variability as well as statistical uncertainty
of the calculations.
The calibration of the partial load factors is undertaken by assessing the
experience and judgement based code specifications and carrying out reliability
modelling of the loads and load models.
The objectives of reliability code calibration encompass both concerns of
structural engineering, viz. the safety and economics of a structure. The
selection of appropriate partial safety factors addresses these two concerns by:
• ensuring sufficiently reliable performance of the structure for the ultimate

limit state as well as the serviceability limit state
• obtaining more consistent reliability levels over a range of load situa-

tions, material choices and structural configurations. Unnecessary con-
servatism can be avoided thereby resulting in more economic designs.
The reliability assessment is carried out assuming that all load situations
have been accounted for, the correct structural details have been chosen and
the design has been carried out correctly. These matters are the responsibility
of the designer and a neglect of these issues cannot be compensated for by the
partial safety factors.
As mentioned in Chapter 1, there is no reliability basis for the crane par-
tial load factor applied to crane wheel loads in the South African loading
code, SABS 0160:1989 [1] and the crane partial load factor applied in SABS
0160:1989 differs from that applied in prEN 1991-3.
The difference in crane partial load factor between SABS 0160:1989 and
prEN 1991-3 and the lack of a reliability basis for the crane partial load factor,
indicates that an investigation into the reliability of structures subject to crane
loads is required.
This investigation will comprise an assessment of the crane load models
in prEN 1991-3, an assessment of the level of reliability of structures designed
according to SABS 0160:1989 and prEN 1991-3 and a code calibration exercise
to determine appropriate partial load factors for crane loads.
Chapter 3
Scope of the code
The scope of the new South African crane loading code (SANS 10160 [3]) was
defined as the same as the scope of the current South African crane loading
code, SABS 0160:1989 [1], i.e. electric overhead travelling bridge cranes with
wheels at the same level, therefore excluding portal or semi-portal cranes.
The specific example cranes and support structures chosen for the code
calibration process should be representative of the scope of the updated South
African crane loading code. Due to the fact that the scope of the updated
crane loading code is the same as the current code, the requirement that the
example cranes should cover the scope of the updated crane loading code will
be met by ensuring that they cover the range of cranes currently found in
practice. This would be ensured if the range of the parameters of the example
cranes cover the range of parameters of cranes found in practice.
The crane parameters that affect the loads imposed by cranes were identi-
fied and their ranges and distributions were investigated. Example cranes were
selected to cover the range of parameters. These example cranes will from now
on be referred to as the representative cranes.
3.1 Crane parameters
The crane parameters that affect the loads imposed by cranes were identified
as those that affect the crane wheel loads calculated from the load models in
prEN 1991-3, as discussed in Chapter 2. The ranges and distributions of the
parameters were determined by consultation with a leading crane manufacturer
and an assessment of their database of 563 cranes manufactured since 1956 [18].
59
60 Chapter 3. Scope of the code
The parameters and their ranges and distributions are given below.
3.1.1 Configuration of crane
• Top mounted or underslung
The crane load models that are used to calculate the crane loads are
different for top mounted and underslung cranes. Top mounted cranes
are the focus of this investigation and it can be seen below that this is
reasonable because 95% of the cranes in South Africa are top mounted
cranes.
– Distribution: Overhead - 95% of cranes

Underslung - 5% of cranes
3.1.2 Nominal weights of crane and hoistload
The weights of the crane, self weight of the crane bridge and crab, and the
weight of the hoistload lifted affect all the crane induced loads - vertical, hor-
izontal longitudinal and horizontal transverse loads due to acceleration of the
crane bridge, acceleration of the crab, skewing of the crane bridge in plan and
the buffer forces due to crane movement.
• Weight of crane bridge
– Range: 0.5t - 211t
• Weight of crab
– Range: 0.08t - 112t
• Nominal weight of the hoistload (crane capacity or safe working load

(SWL) of the crane)
– Range: 2t - 630t
– Distribution: 63% of cranes 20t - 130t

42% of cranes 30t - 60t
3.1. Crane parameters 61
3.1.3 Crane geometry

• Span of the crane bridge
The span of the crane bridge has an effect on the vertical loads and
horizontal transverse loads due to acceleration of the crane bridge and
skewing of the crane in plan.
– Range: 4 m - 54 m
• Wheel spacing - distance between the wheels on one end carriage

The wheel spacing affects the horizontal transverse wheel loads due to
acceleration of the crane bridge and skewing of the crane in plan. The
wheel spacing is also required for the calculation of the load effects in the
girder and crane loads to be applied to the column, in that it is needed
for the positioning of the crane wheel loads on the support structure.
– Dependent on bridge span:

span < 15 m: wheelbase = 1.75 m
span > 15 m (4 wheels): wheelbase = span/7
span > 15 m (8 or 16 wheels): wheelbase = span/5
• Minimum distance between hoist and rail

The minimum distance between the hoist and the rail defines the closest
position that the crab, and hence the hoistload, can come to the rail.
This influences the magnitude of the vertical loads and the horizontal
transverse loads due to acceleration of the crane bridge and skewing of
the crane in plan.
– Range: 0.4 m - 5 m
– Distribution: most common 1.5 m - 2.5 m (dependent on span)
• Rail head width

The rail head width is required for the calculation of the skewing angle
for the horizontal loads due to skewing of the crane in plan. The rail
head width is also needed for the calculation of the load effects in the
girder. The vertical wheel load is assumed to act at an eccentricity of
a quarter of the rail head width for the calculation of the torsion in the
girder.
– Range: rails range from 40×40 square bar - DIN A120 rail (depen-
dent on size of crane), rail head width ranges from 40 mm - 120 mm
3.1.4 Travel and hoist speeds
• Long travel speed
The long travel speed is the speed that the crane bridge travels down
the length of the runway. This has an effect on the loads arising from
the accidental load situation of the crane running into the end stops on
the end of the runway.
– Range: 25 m/min - 100 m/min

– Distribution (dependent on length of gantry):
standard cranes: most common 40 m/min
process cranes: most common 60 m/min
• Steady hoisting speed
The hoisting speed of the crane has an effect on the dynamic factor φ2
which takes into account the dynamic effects of lifting a hoist load off the
ground. The larger the hoisting speed, the greater the dynamic effects
will be.
– Range: 2 m/min - 20 m/min
3.1.5 Hoist type and characteristics
• Hoist class
Cranes are classified into four ‘hoist classes’ to allow for the dynamic
effects of hoisting a load from the ground. The hoist class represents the
springing of the hoisting gear which affects the dynamic effects of lifting
a load. The classification is dependent on the use of the crane. The hoist
class has an effect on the dynamic factor φ2 which takes into account the
dynamic effects of lifting a load off the ground.
– Distribution: standard cranes: HC1

process cranes: HC2 - HC4
• Type of load lifting mechanism
The type of load lifting mechanism refers to whether the crane lifts loads
with a hook, grab or magnet or whether it is a ladle crane or coil lifter etc.
This has an effect on the crane loads only in the event of the crane being
equipped with a grab or magnet in the case when the normal operating
behaviour of the crane involves the crane suddenly releasing a part of the
hoistload. In this case, the type of load lifting mechanism has an effect
on the dynamic factor φ3 which takes into account the dynamic effects
of a sudden release of part of the hoistload.
– Distribution: hook - 89% of cranes

ladle - 2% of cranes
grab or magnet - 8% of cranes
other (e.g. coil lifter) - 1% of cranes
• Hoistload free to swing
Whether the hoistload is free to swing or not, determines whether the

crane is prone to tilting due to the hoistload colliding with an obstacle.
– Distribution:
hoistload is free to swing - 97% of cranes
hoistload suspended by mast and not free to swing - 1% of cranes
cables have anti-sway reaving - 2% of cranes
3.1.6 Wheels and wheel drives
• Number of wheels
The number of wheels that a crane has determines the size of the loads
on each wheel as the total load is divided among the individual wheels.
This has an effect on the magnitude of all the vertical and horizontal
transverse loads.
– Distribution: (dependent on size of crane)

4 wheels - 75% of cranes
• Combination of wheel pairs

prEN 1991-3 allows for four configurations of wheel pairs which are com-
binations of either independent or continuous wheels, and fixed or mov-
able wheels.
Continuous wheel pairs are wheels which are joined together between
the end carriages by mechanical or electrical means, e.g. in the case of
central wheel drives. Independent wheels are not linked in any way to the
wheels on the opposite end carriage. The current practice of all the crane
manufacturers in South Africa is to use independent wheels, continuous
wheel pairs were common in the past but are not in use anymore.
Movable wheels are wheels which have a lateral degree of freedom, i.e.
are able to move horizontally along their axis, and fixed wheels are wheels
which do not have this lateral degree of freedom. Cranes with movable
wheels would have fixed wheels on one end carriage and movable wheels
on the other end carriage. Movable wheels affect the distribution of
horizontal transverse forces on the crane as only the fixed wheels can
transmit horizontal transverse forces to the rail. It is the current practice
of all the crane manufacturers in South Africa to use fixed wheels on both
end carriages.
– always independent, fixed wheels
• Type of wheel drive

The type of wheel drive refers to whether the crane is driven by means
of a central wheel drive on continuous wheels or single wheel drives on
independent wheels. This affects the horizontal forces due to the accel-
eration of the crane and skewing of the crane in plan. As mentioned
above, the current practice of crane manufacturers is to use independent
wheels and hence single wheel drives.
– always single wheel drives, no central wheel drives
• Number of wheel drives

The number of wheel drives refers to the number of single wheel drives
on a crane. This affects the horizontal forces due to acceleration of the
crane bridge.
– Distribution:
all standard cranes: 2 out of 4 (2/4) wheels driven
30% of process cranes: 50% driven (i.e. 2/4, 4/8, 8/16)
70% of process cranes: 25% driven (i.e. 2/8, 4/16)
• Smoothness of change of drive forces

The smoothness of change of the longitudinal drive forces affects the
dynamic factor φ5 which takes into account the dynamic effects due to
the acceleration of the crane bridge. Values of φ5 are given in prEN
1991-3 for drives which have smooth changes, e.g. drives with frequency
inverters rather than gears, drives which have sudden changes, i.e. drives
with gears, and drives which have considerable backlash.
– Distribution:
sudden changes - 70% of cranes (mostly standard cranes)
smooth changes - 30% of cranes (mostly process cranes)
3.1.7 Guide means

• Type of guide means
The guide means refers to the way in which the crane is guided horizon-
tally on the rails. The crane is guided by either the flanges of the crane
wheels or by guide rollers. Flanged wheels are the most common guide
means. Guide rollers are typically used in buildings that have furnaces
where one rail is close to the furnace, this rail could deform due to the
heat of the furnace thereby making it unsuitable for flanged wheels.
The type of guide means influences the horizontal transverse forces caused
by skewing.
– Distribution:
guided by wheel flanges - 99% of cranes
guided by guide rollers - 1% of cranes
• Clearance between rail and guide means

The clearance between the rail and guide means refers to the clear space
between the guide means (typically the wheel flange) and the side of
the rail. This distance influences the skewing angle of the crane for the
calculation of the horizontal forces caused by skewing.
– short span cranes - 15 mm

long span cranes - 25 mm
3.1.8 Buffers
• Buffer type
The buffer type has an effect on the accidental load case of the buffer
forces due to the crane running into the end stops. The type of buffer
will define the spring constant and buffer characteristic which are needed
for the calculation of these loads.
– Distribution light cranes - rubber or cellular rubber

heavy cranes - hydraulic
3.1.9 Governing parameters

The two governing parameters were the span of the crane bridge and the
capacity of the crane (SWL); most of the other parameters are dependent
on these two. The spans and capacities of the cranes in the database were
investigated further to determine more information on their distributions.
It was found that the crane bridge spans were Normally distributed with a
mean of 24.5 m. The capacities displayed a Lognormal type distribution. The
mean of the capacities was 70t and the mode was 35t. Figures 3.1 and 3.2 show
histograms of the crane bridge spans and crane capacities for the 563 cranes
in the database. The capacities along the x-axis in Figure 3.2 are shown in a
log10 scale.
Three representative cranes were selected from actual crane installations in
South Africa. A summary of the main parameters of the representative cranes
is given in Table 3.1. A scatter plot of all the cranes in the crane manufacturer’s
database as well as the representative cranes is shown in Figure 3.3 where the
spans are plotted against the capacities of the cranes, the capacities on the
x-axis are again shown in a log10 scale. The small squares represent all the
cranes in the database and the circles show the representative cranes. The
mean of the spans and the mode of the capacities are indicated on the plot.
The representative cranes, having capacities of 5t, 40t and 260t, cover
the full range of capacities, with only 5% of the cranes in the manufacturer’s
database falling outside the 5t - 260t range. The crane bridge spans of the
representative cranes of 19.2 m, 23.8 m and 28.5 m, cover the most likely range
Histogram of crane bridge spans from crane database

200
180
160
140
Number of cranes
120
100
80
60
40
20
0
0 10 20 30 40 50 60
Crane bridge span (m)
Figure 3.1: Histogram of crane bridge spans
Histogram of crane capacities from crane database

180
160
140
120
Number of cranes
100
80
60
40
20
0
3 5 10 19 35 67 127 242 458
Crane capacity (t), log10 scale
Figure 3.2: Histogram of crane capacities

of spans (refer to Figure 3.1) with 44% of cranes having spans in the range of
19.2 m - 28.5 m, and 57% of cranes having spans in the slightly larger range of
19 m - 30 m. This is shown graphically on the scatter plot in Figure 3.3 which
shows that the representative cranes cover the full range of capacities but not
the full range of spans. The implications for the results for cranes with longer
or shorter spans will be investigated by means of parametric studies.
The 40t crane can be considered to be the ‘most representative’ crane
for South Africa because its capacity and span are close to the mode of the
capacities and mean of the spans of the cranes considered, which were produced
by one of the leading South African crane manufacturers [18].
A more detailed description of the representative cranes and their support
structures is given in section 3.3.
Table 3.1: Parameters of representative cranes

Crane Crane Weight Weight Number Outer Hoist
capacity bridge of of of wheel type
(SWL) span bridge crab wheels spacing
t m kN kN m
5 19.2 55 6.4 4 3 hook
40 23.8 298 98 4 4.4 hook
260 28.5 1970 1298 16 12 ladle
Scatter plot of spans and capacities of cranes

60
All cranes
Representative
Spans mean
50 Capacities mode
40
Crane bridge spans
30
20
10
0
3 10 32 100 316 1000
Crane capacity (t), log10 scale
Figure 3.3: Scatter plot of cranes from database

3.2. Range of support structure configurations 69
3.2 Range of support structure configurations

For the example structures to be truly representative, both the crane and the
support structures should be representative of those found in practice. The
crane girders and columns are the elements of the support structure that are
most directly loaded by the crane loads so these elements will be discussed.
The different types of crane girders and columns described by Rowswell [30]
and Ricker [20] that were mentioned in Chapter 2 are outlined below.
3.2.1 Crane Girders

1. Wide flange shape, rolled I section
2. Wide flange shape with channel cap
3. Welded wide flange shape with channel as top flange
4. Welded wide flange shape with unequal flanges
5. Welded wide flange shape for narrow clearances
6. Welded wide flange shape with surge plate
7. Welded wide flange shape with surge plate and backup truss
The different crane girder configurations are suitable for different types of
cranes. Rolled I sections or I sections with channel caps are typically used
for light cranes where the channel cap would be sufficient for resisting the
horizontal loads, this is the crane girder configuration for the 5t crane. Welded
wide flange girders with unequal flanges are generally used for medium cranes,
this is the girder configuration of the 40t crane. Large cranes which impose
large horizontal forces on the crane girders typically have a surge plate with
or without a backup truss. This configuration is the most economical way of
resisting the large horizontal forces. The 260t crane girder is a welded wide
flange shape with a surge plate and backup truss.
The 40t crane girder and the 260t crane girder are simply supported gird-
ers and the 5t crane girder is a two span continuous girder. Simply supported
girders are preferable to continuous girders due to the fatigue problems en-
countered with continuous girders, however, continuous girders are still used
for crane girders.
3.2.2 Crane columns

1. Separate building and crane column
2. Bracketed column
3. Step column, solid web
4. Step column, laced web
5. Step column, vierendeel
6. Step column, battened
Light cranes typically have bracketed columns where the crane girder is
supported by a bracket (corbel) connected to the building column, this is the
configuration of the 5t crane column. Heavier cranes are supported on step
columns. Step columns can have either solid webs, laced webs, vierendeel
truss types webs or battened webs. Heavy cranes in low buildings are typically
supported on crane columns which are separate from the building columns.
The 40t crane column is a step column with a concrete column supporting
the crane girders and a steel roof column extending from the concrete column.
The 260t crane column is also a step column consisting of steel crane columns
and building columns with a laced web. The 260t crane is in a tall building
with the crane girders 24 m from floor level which would make it uneconomical
to use separate building and crane columns.
The support structures for the three representative cranes fall into the
recognised support structure configurations and cover a range of different con-
figurations. The support structures can therefore also be considered to be
representative of those found in practice.
3.3 Representative cranes and their support

structures
Detailed descriptions of the representative cranes and their support structures
are given below. The representative cranes will be referred to by their capaci-
ties.
The crane support structure consists of the crane girders and the building
structure which houses the crane. The girders are an important part of the
3.3. Representative cranes and their support structures 71
support structure as they are the members which are most directly loaded by
crane loads being subject to only their own weight and the crane wheel loads.
3.3.1 5t crane
The 5t crane is one of five cranes in a container making factory. The building
consists of four bays, bays 1, 2 and 4 have one crane each and bay 2 has two
cranes (see Figure 3.4). The 5t crane considered here is in bay 1 on one end
of the building. The container assembly process starts in bay 4 on the far side
of the building and each bay handles a stage of the assembly process which is
completed in bay 1. The 5t crane lifts the almost completed containers for the
final stages of the assembly process.
3.3.1.1 Description of crane
The 5t crane is an overhead, double box girder crane with four wheels and two
wheel drives. The parameters of the crane are given in Table 3.2.
3.3.1.2 Description of support structure
The crane girders for the 5t crane are two span continuous girders, each span
of 6.5 m. The girder section is a compound section consisting of a hot rolled
I section with a channel on the top flange (305×165×41 UB + PFC 220×80).
The channel is welded to the I section with a longitudinal continuous fillet
weld along the edges of the top flange of the I girder. A cross section of the
girder is shown in Figure 3.5.
The four bay building housing the 5t crane is shown in Figure 3.4. For
the purposes of this investigation bay 1 was considered in isolation and the
building was modelled as shown in Figure 3.6. The frame consists of steel
columns (305×165×54 UB) with corbels to support the crane girders and a
truss for the roof member. The corbels are 178×102×19 UB sections.
Table 3.2: Parameters of the 5t crane

Nominal weights of crane and hoistload
Weight of crane bridge 55 kN
Weight of crab 6.5 kN
Hoistload 50 kN
Crane geometry
Span of crane bridge 19200 mm
Minimum distance between hoist and rail 570 mm
Rail type 22 kg/m
Width of top of rail 50 mm
Height of rail 95.3 mm
Travel and hoist speeds
Steady hoisting speed 10 m/min
Long travel speed 20 m/min
Cross travel speed 20 m/min
Hoist type and characteristics
Hoist class HC2
Type of load lifting mechanism hook
Hoistload free to swing Yes
Wheels and wheel drives
Number of wheels 4
Wheel spacing 3000 mm
Type of wheel drive Single
Number of single wheel drives 2
Behaviour of drives Smooth changes
Combination of wheel pairs Independent
Fixed/Fixed
Guide means
Guide rollers present No
Clearance between rail and wheel flange 15 mm
Buffers
Buffer type rubber DPG 80
Buffer characteristic (degree of plasticity) 0.7
Spring constant 750 × 103 N/m
Figure 3.4: Building housing the 5t crane
Figure 3.5: Cross section of 5t crane girder
Figure 3.6: Bay considered for 5t crane

3.3.2 40t crane

The 40t crane is one of two cranes in a two bay mono-pitch portal frame
building which houses a car manufacturing stamping plant. Figure 3.7 shows
a diagram of the structure. The 40t crane is in bay 1 which is the press line,
bay 2 is a maintenance area. The primary purpose of the 40t crane is to move
the stamping press dies for the tool changes. This entails fetching the new
tools from the storage area and lining them up ready for the stamping press
as well as removing the old tools and taking them back to the storage area.
The tool changes happen approximately every hour.
The 40t crane is an overhead, double box girder crane with four wheels and
two wheel drives. The span of the crane is 23.8 m. The parameters of the
crane are given in Table 3.3.
The 40t crane girders are simply supported, 8 m span, mono-symmetric, welded
plate girders. The girders have bearing stiffeners at the supports and interme-
diate stiffeners at a spacing of 2 m. The intermediate stiffeners are welded to
the top flange and web but have a cope around the top flange to web weld of
the girder. There is a clearance of four times the thickness of the web between
the bottom of the stiffener and the bottom flange of the girder. Figure 3.8
shows the cross section of the girder and the intermediate stiffeners. The ends
of the girders are supported laterally by means of tie back connections to the
columns.
The crane runways are braced in the longitudinal direction by means of a
pair of circular hollow sections per runway as shown in Figure 3.9.
The building consists of reinforced concrete crane columns (1500×450) with
steel building columns (356×171×51 UB) and roof members (406×140×39 UB)
with knee braces. Step type columns have been utilised with the crane girders
resting on steel girder ‘chairs’ on top of the concrete columns, see Figure 3.7.

Weight of crab 98 kN
Hoistload 400 kN
Crane geometry
Rail type DIN A100
Height of rail 95 mm
Long travel speed 50 m/min
Cross travel speed 40 m/min
Hoist class HC3
Type of load lifting mechanism hook
Number of wheels 4
Wheel spacing 4400 mm
Fixed/Fixed
Guide means
Buffers
Buffer type hydraulic
PUDZ II φ 63
Spring constant 2 × 106 N/m
(a) Cross section of girder (b) Stiffener dimensions
Figure 3.8: 40t crane girder
Figure 3.9: Crane runway longitudinal bracing for the 40t crane
3.3.3 260t crane

The 260t crane is a ladle crane for a steel making and rolling mill. The building
housing the 260t crane consists of five bays where all except bay 2 contain
cranes, the 260t crane is in bay 3 (see Figure 3.10). The primary purpose of
the 260t crane is to carry molten steel from the furnace to the caster.
The 260t crane is an overhead, double box girder crane with sixteen wheels
and four wheel drives. The span of the crane bridge is 28.5 m. The parameters
of the crane are given in Table 3.4.
The crane girder for the 260t crane is a 42 m span doubly symmetric plate
girder with a 2940×10 mm surge plate and lattice auxiliary girder. A cross
section of the crane girder is given in Figure 3.11 and a diagram of the lattice
auxiliary girder is shown in Figure 3.12.
The building consists of step crane columns and building columns with a
lattice girder roof truss. The step columns are steel laced columns with a cross
girder supporting the building column, see Figure 3.10. The entire building
was modelled for this investigation.


Hoistload 2600 kN
Crane geometry
Rail type DIN A120
Height of rail 105 mm
Long travel speed 81.1 m/min
Cross travel speed 27.6 m/min
Hoist class HC4
Type of load lifting mechanism ladle
Number of wheels 16
Wheel spacing
Outside wheel span: 12000 mm
Individual wheel spacing: 1000/1800/1000/4400/1000/1800/1000 mm
Fixed/Fixed
Guide means
Buffers
Buffer type hydraulic
OLEO 54 MFZ 140 412
Spring constant 1.25 × 106 N/m
3.4. Summary 79
Figure 3.11: Cross section of the 260t crane girder
Figure 3.12: 260t crane auxiliary girder
3.4 Summary
The scope of the new South African crane loading code (SANS 10160 [3]) was
defined as the same as the scope of the current code, SABS 0160:1989 [1],
i.e overhead travelling bridge cranes with wheels elevated at the same level.
As mentioned before, the code calibration procedure is carried out on specific
example cranes which should be representative of the cranes found in practice
i.e. their parameters should cover the range of parameters found in practice.
The ranges and distributions of the crane parameters affecting the crane wheel
loads were investigated. It was found that the two governing parameters were
the crane bridge span and the capacity of the crane.
Three actual crane installations in South Africa were chosen as the repre-
sentative cranes and support structures to be used as example structures for
the code calibration. The representative cranes were shown to cover the full
range of capacities found in practice and the most likely range of spans. De-
scriptions of the cranes and their support structures were given. It was shown
that the support structures are also representative of the range of configura-
tions found in practice.
Chapter 4
Calibration to current practice
The calibration to current practice involves the comparison between the code
provisions in SABS 0160:1989 and prEN 1991-3. The comparison was carried
out with respect to the load situations modelled, the cost of the support struc-
ture and the design effort required. The comparison was carried out on the
girders, columns and longitudinal bracing members of the three representative
cranes discussed in Chapter 3.
4.1 Crane code provisions

The behaviour of cranes and therefore the loads imposed on the support struc-
ture is modelled differently in SABS 0160:1989 and prEN 1991-3. There are
differences in the classification of cranes and the influence it has on the loads,
the load situations modelled and the recommended crane load combinations.
4.1.1 Crane classification
In both SABS 0160:1989 and prEN 1991-3, cranes are classified into four
classes. The methods of classification are similar in that they are both based
on descriptions of the cranes. The influence of the crane classification on the
wheel loads differs vastly between the two codes.
4.1.1.1 Definition of crane classes
In SABS 0160:1989, cranes are classified into four classes according to a de-
scription of the crane usage. prEN 1991-3 divides cranes into four hoist classes
81
82 Chapter 4. Calibration to current practice
also according to the crane use. Table 4.1 gives descriptions of crane usage as
given in the two codes along with the classification of each crane.
Table 4.1: Crane classification
Description of crane usage Class in Hoist class

SABS in prEN
0160:1989 1991-3
Hand cranes C1 HC1
Assembly cranes HC1, HC2
Warehouse cranes - intermittent operation C2 HC2
Power station cranes C2 HC2
Foundry cranes C2 HC2
Casting cranes HC2
General factory and workshop cranes C2 HC2, HC3
Machine shop cranes C2 HC2, HC3
Warehouse cranes - continuous operation C3 HC3
Rolling mill cranes C3
Ladle cranes in steelworks C3 HC3
Scrapyard cranes C3 HC3, HC4
Grab and magnet cranes - intermittent operation C3 HC3, HC4
Grab and magnet cranes - continuous operation C4 HC3, HC4
Soaking pit cranes C4 HC4
Ingot stripping cranes C4 HC4
Furnace charging cranes C4 HC4
Forging cranes C4 HC4
Claw cranes C4 HC4
There is no explanation given in SABS 0160:1989 as to the reasoning behind

the crane classification. The rationale behind the division of cranes into hoist
classes is given in the German code DIN 15018 Part 1 1984 [4] and was given
in Chapter 2 as representing the springing of the hoisting gear and thus the
dynamic effects of lifting a load.
With reference to Table 4.1, the same type of cranes tend to fall in the
same class for both codes, even though the basis for crane classification in
the two codes cannot be shown to be the same and the influence of the crane
classification on the crane loads is vastly different.
4.1. Crane code provisions 83
4.1.1.2 Influence of crane classification in SABS 0160:1989
The class of crane influences the impact factor for the vertical loads and also
the magnitude of the horizontal loads. For example, the vertical wheel loads
including dynamic effects are calculated by multiplying the nominal static
wheel loads by an impact factor which is dependent on the class of the crane.
The horizontal wheel loads caused by misalignment of the crane wheels or rails
are calculated by multiplying the total crane weight by a factor dependent on
the class of the crane. These factors are given in Table 4.2.
Table 4.2: Class dependent factors in SABS 0160:1989

Class of crane Vertical dynamic factor Horizontal factor
Class 1 1.10 0.05
Class 2 1.20 0.12
Class 3 1.25 0.15
Class 4 1.30 0.20
4.1.1.3 Influence of crane classification in prEN 1991-3
The hoist class of the crane influences only the dynamic factor applied to the
hoistload for the calculation of the vertical load with dynamic effects caused by
the hoisting of a load off the ground. The horizontal forces are not influenced
by the hoist class of the crane. The equation for the calculation of the dynamic
factor is given below.
φ2 = φ2,min + β2 × vh (4.1.1)
Where:
φ2 – the dynamic factor applied to the hoistload to model the dynamic effects
of lifting a load off the ground.
vh – the steady hoisting speed of the crane
φ2,min & β2 – class dependent factors given in Table 4.3.

Table 4.3: Hoist class dependent dynamic factors in prEN 1991-3

Hoist class of crane φ2,min β2
HC1 1.05 0.17
HC2 1.10 0.34
HC3 1.15 0.51
HC4 1.20 0.68
4.1.2 Load cases
The modelling of the crane behaviour as the crane travels on the runways,
picks up and sets down loads, results in different load cases for the crane
wheel loads. The loading situations considered by the two codes differ and are
discussed below.
4.1.2.1 SABS 0160:1989
The following load situations are provided for in SABS 0160:1989:
• Vertical wheel loads
– Static wheel loads

– Dynamic wheel loads, impact and other dynamic effects are pro-
vided for by multiplying the static wheel load by a factor depending
on the crane class as given above.
• Horizontal transverse loads due to
– Acceleration and braking of the crab

– Misalignment of rails or wheels
– Skewing of the crane in plan
• Horizontal longitudinal force due to acceleration or braking of the crane.
• End stop forces due to collision of the crane bridge with the end stops
at the end of the runway.
4.1.2.2 prEN 1991-3
prEN 1991-3 makes provision for the following load situations:

• Vertical wheel loads due to
– Lifting a hoistload off the ground

– Part of the hoistload suddenly being released
– The crane travelling along the runway at a constant speed
– Test loads
• Horizontal longitudinal and transverse forces due to
– Acceleration or braking of the crane

– Skewing of the crane in plan
• Horizontal transverse forces due to
– Acceleration or braking of the crab

– Buffer forces due to the crab running into the end stops at the end
of the crane bridge
• Horizontal longitudinal forces due to buffer forces from the crane running
into the end stops at the end of the runway
4.1.2.3 Comparison of load cases
prEN 1991-3 considers in more detail the actual crane behaviour especially
with regards to the vertical loads.
One horizontal load situation which is provided for by SABS 0160:1989
but not by prEN 1991-3 is misalignment of the crane wheels or gantry rails.
The misalignment load case in SABS 0160:1989 takes into account the wheels
of the crane being misaligned in a ‘toe-in’ or ‘toe-out’ manner or a similar
misalignment of the rails. The horizontal loads are due to the horizontal
transverse friction force that is developed due to the rolling direction of the
wheel not coinciding with the longitudinal axis of the rail.
The amount of misalignment which causes these horizontal loads still falls
within the construction tolerances as specified by SAISC [29] and the misalign-
ment loads can thus be classified as ‘load occurring during normal operation
of the crane’. This load case does not take into account the situation where
the rails or wheels are out of misalignment to such an extent that the flanges
of the wheels on each end carriage come into contact with the rails and tend
to either pull the rails together or push them apart. Such a misalignment does
not fall within the construction tolerances and would thus not be classed as
‘normal operation of the crane’.
Perfect alignment of the rails is impossible to achieve so some allowance
in the crane loads for misalignment of the rails within construction tolerances
should be made.
4.1.3 Load combinations

The guidelines for load combinations for each of the codes are discussed. For
comparison purposes, the design wheel loads resulting from each code are
presented here for the 40t representative crane described in Chapter 3 which
is classified as a class 3 crane in SABS 0160:1989 and a hoist class 3 crane in
prEN 1991-3.
4.1.3.1 SABS 0160:1989
The only guidance given on load combinations is that the three horizontal
transverse loads need not be considered to act simultaneously. The resulting
load combinations and forces for the 40t crane are given in Figures 4.1 - 4.5.
The full load calculations are given in Appendix A.
The loads given in Figures 4.1 - 4.5 are design loads that have been factored
using SABS 0160:1989 load factors i.e. 1.6 for ultimate limit state and 1.0 for
accidental loads (end stop forces).
All loads shown are in kN. The horizontal transverse loads may act either
in the directions shown or in the opposite direction.
Figure 4.1: SABS 0160:1989 load combination 1: Vertical + Acceleration of crab +

Acceleration of crane
Figure 4.2: SABS 0160:1989 load combination 2: Vertical + Misalignment + Ac-

celeration of crane
Figure 4.3: SABS 0160:1989 load combination 3: Vertical + Skewing + Acceleration

of crane
Figure 4.4: SABS 0160:1989 load combination 4a: Vertical + End stop force
Figure 4.5: SABS 0160:1989 load combination 4b: Vertical + End stop force
4.1.3.2 prEN 1991-3
prEN 1991-3 gives a table which describes the combinations of crane loads
which are to be considered as one crane action for combination with other
loads. The table was given in Chapter 2, Table 2.16 on page 46.
The load combinations resulting from prEN 1991-3 for the 40t crane are
given below. The ultimate limit state load combinations are shown in Figures
4.6 - 4.11, the serviceability limit state combination is shown in Figure 4.12
and the accidental limit state load combination is shown in Figure 4.13. The
full load calculations are given in Appendix A.
Load combination 2 is not relevant for the 40t crane because its normal
working does not entail sudden release of the hoistload.
The loads shown in Figures 4.6 - 4.13 are design loads which have been
factored using the SABS 0160:1989 load factors i.e. 1.6 for ultimate limit
state, 1.0 for serviceability limit state and 1.0 for accidental loads.
Figure 4.6: prEN 1991-3 load combination 1: φ1 (Self weight) + φ2 (Hoistload) +

φ5 (Acceleration of crane)
Figure 4.7: prEN 1991-3 load combination 3: Self weight + φ5 (Acceleration of

crane)


Skewing)

Acceleration of crab
Figure 4.11: prEN 1991-3 load combination 7: Self weight + η(Hoistload)
Figure 4.12: prEN 1991-3 load combination 8: φ1 (Self weight) + φ6 (Test load) +
Figure 4.13: prEN 1991-3 load combination 9: Self weight + Hoistload + φ7 (Crane
buffer forces)
4.1.4 Fatigue loading
4.1.4.1 SABS 0160:1989
The South African loading code, SABS 0160:1989, does not specify the crane
loads which are to be used for fatigue verification of the crane support struc-
ture, nor how the crane loads are to be applied to the structure for the fatigue
assessment. The South African steel design code, SANS 10162-1:2005 [11], rec-
ommends the use of Miner’s rule of accumulative damage for the assessment
of fatigue for loads not applied at a constant amplitude.
4.1.4.2 prEN 1991-3
prEN 1991-3 gives simplified rules for the assessment of the fatigue of crane
support structures. Cranes are divided into fatigue classes depending on the
range of loads lifted and the total number of cycles performed during the
lifetime. Depending on the fatigue class of the crane, the ‘fatigue damage
equivalent load’ is calculated which is the equivalent constant amplitude load
for two million cycles and is used for the calculation of the fatigue stresses.
The fatigue provisions were discussed in more detail in Chapter 2.
4.1.4.3 Comparison of fatigue loading
Fatigue is an issue where the loads and resistance are traditionally not sepa-
rated. The fatigue damage, which is used as a measure of the fatigue life of a
structure, is a function of both the loads and the resistance of the member in
the form of S-N curves.
SABS 0160:1989 follows the traditional method of not specifying fatigue
loads in the loading code whereas prEN 1991-3 separates the fatigue loads
from the resistances by providing a method of calculating the ‘fatigue damage
4.2. Cost of support structure 91
equivalent load’.
Another approach which could be adopted by a crane loading code is that
taken by the International code ISO 8686-1:1989 [5] where recommendations
are made on the crane load combinations that are to be considered for fatigue.
It is then up to the support structure designer to model the crane behaviour
to determine at what level the loads are applied. The modelling of the crane
behaviour would entail considering the frequency at which each value of hoist-
load is lifted as well as the position of the crab on the crane and the crane on
the runway.
Both the method given by prEN 1991-3 and the alternative method of
modelling the crane behaviour will be considered for the further investigations
of fatigue.
The modelling of the crane behaviour entails a full analysis of the crane
movements and loading to determine the range of stress amplitudes and the
number of times they will be applied over the crane lifetime.
4.2 Cost of support structure

The critical load combinations for the ultimate limit state resulting from each
code were identified as load combinations 2 and 3 from SABS 0160:1989 and
load combinations 1 and 5 from prEN 1991-3. The design (factored) loads for
the 40t crane for these load combinations are shown in Figures 4.14 - 4.17.
The wheel loads have been factored using the South African load factors of 1.6
for ultimate limit state.
The critical load combinations for the accidental limit state of the crane
running into the end stop were load combination 4a and 4b from SABS 0160:1989
and load combination 9 from prEN 1991-3. These load combinations are shown
in Figures 4.18 - 4.20. For the purpose of comparison between the different
methods of calculating the end stop forces in SABS 0160:1989, the forces re-
sulting from both the simplified method and the more detailed method, taking
into account the resilience of the buffer, are shown.
The comparison between the loads from SABS 0160:1989 and prEN 1991-3
could not be carried out directly by considering the crane load combinations
because the configurations of the loads from the two codes are different. The
comparison was therefore carried out on the load effects in the support struc-
tures.
Figure 4.14: SABS 0160:1989 critical load combination 2: Vertical + Misalignment

+ Acceleration of crane
Figure 4.15: SABS 0160:1989 critical load combination 3: Vertical + Skewing +

Acceleration of crane
Figure 4.16: prEN 1991-3 critical load combination 1: φ1 (Self weight) +

φ2 (Hoistload) + φ5 (Acceleration of crane)
Figure 4.17: prEN 1991-3 critical load combination 5: φ4 (Self weight) +

φ4 (Hoistload) + Skewing)
Figure 4.18: SABS 0160:1989 critical load combination 4a: Vertical + End stop
force
Figure 4.19: SABS 0160:1989 critical load combination 4b: Vertical + End stop
force
Figure 4.20: prEN 1991-3 critical load combination 9: Self weight + Hoistload +
φ7 (Crane buffer forces)
4.2.1 Structural elements

The elements of the support structures that were considered for the comparison
were the elements most directly loaded by crane loads i.e. crane girders, crane
columns and longitudinal runway bracing.
The load effects considered for the crane girder were moments about the z
and y axes, shear in the y and z directions, torsion about the x axis and axial
force in the x direction. The axes of the girder are defined in Figure 4.21.
Figure 4.21: Crane girder axes
The load effects considered for the crane columns were axial force and
bending moment. The load effect considered for the longitudinal bracing was
axial force.
4.2.2 Crane class
The class of the crane in SABS 0160:1989 affects the magnitude of the vertical
impact forces as well as the horizontal forces due to acceleration of the crab,
misalignment of the rails or wheels and skewing of the crane in plan. The hoist
class of the crane in prEN 1991-3 affects only the dynamic factor φ2 modelling
the dynamic effects caused by hoisting a load off the ground, which is part of
load combination 1.
A parametric study was carried out for the class or hoist class of the crane
in order to assess the sensitivity of the load effects to the class of crane in
SABS 0160:1989 or hoist class in prEN 1991-3.
The loads and load effects were calculated for each crane assuming the
crane to be, in turn, a class 2, 3 or 4 crane for SABS 0160:1989 and a hoist
class 2, 3 or 4 crane for prEN 1991-3. Class 1 and hoist class 1 crane were
omitted because these classes consist mostly of hand operated cranes which
are not of interest for this study.
The load effects from the SABS 0160:1989 class 2 cranes were compared
with prEN 1991-3 hoist class 2 cranes etc. This is reasonable because it was
shown earlier, with reference to Table 4.1 that the same type of cranes tend to
fall into the same class for each code, e.g. a warehouse crane with intermittent
operation will be classified as class 2 in SABS 0160:1989 and hoist class 2 in
prEN 1991-3.
The 5t crane is classified as a class or hoist class 2 crane, the 40t crane is a
class or hoist class 3 crane and the 260t crane is a class or hoist class 4 crane.
4.2.3 Comparison of load effects
The load effects in the crane girders and columns were calculated for the critical
load combinations. The maximum load effects were identified and the ratio
of load effect from SABS 0160:1989 to the load effect from prEN 1991-3 were
calculated. The load effect ratios for the three representative cranes for class
or hoist class 2 - 4 are given in Table 4.4. The columns containing the ratios
for the actual class of each crane are shown in bold.
Table 4.4: Load effect ratios: SABS 0160:1989/prEN 1991-3
5t crane 40t crane 260t crane

Load effect Crane class or hoist class
2 3 4 2 3 4 2 3 4
Crane girder
Moment about z axis 1.06 1.06 1.06 1.16 1.16 1.16 1.11 1.12 1.13
Shear in y direction 1.06 1.06 1.06 1.16 1.16 1.16 1.11 1.12 1.13
Moment about y axis 0.51 0.64 0.85 0.46 0.58 0.77 1.04 1.30 1.73
Shear in z direction 0.64 0.80 1.07 0.46 0.58 0.77 1.03 1.29 1.72
Torsion 0.81 0.91 1.07 0.76 0.85 0.99 1.16 1.40 1.80
Axial force 2.38 2.38 2.38 3.43 3.43 3.43 6.37 6.37 6.37
Crane column
Moment 0.46 0.55 0.70 0.64 0.73 0.87 1.15 1.39 1.79
Axial force 1.06 1.06 1.06 1.16 1.16 1.16 1.22 1.34 1.53
Bracing
Axial force SABS (a) 2.08 2.08 2.08 1.11 1.11 1.11 2.68 2.68 2.68
Axial force SABS (b) 0.74 0.74 0.74 0.42 0.42 0.42 0.65 0.65 0.65
The load effects that are due solely to vertical loads are for all the girders -
the moment about the z axis and shear in the y direction and for the columns
of the 5t and 40t cranes - the axial force. These load effects are all greater
than one with the largest being 1.16. This shows that the vertical loads due
to SABS 0160:1989 can be up to 16% larger than those due to prEN 1991-3.
The load effects due to the vertical loads only remain the same, or within a
very small range, over the three classes being considered. The maximum ver-
tical wheel loads from both SABS 0160:1989 and prEN 1991-3 are dependent
on the class of the crane, and for both codes, the vertical wheel loads increase
by the same amount as the class of the crane increases.
The load effect ratios influenced by only the horizontal transverse loads are
for all the girders - the moment about the y axis and shear in the z direction.
The load effect ratios which are due to a combination of vertical and horizontal
wheel loads are for all the girders - the torsion, the column for the 260t crane
- the axial force and moment and the columns for the 5t and 40t cranes - the
moment.
The load effect ratios due to horizontal loads only and a combination of
horizontal and vertical loads vary across the different crane classes. The hor-
izontal transverse loads in prEN 1991-3 are not affected by the hoist class of
the crane so this range in load ratios is due to the change in magnitude of the
loads in SABS 0160:1989.
The load effects resulting from both horizontal load only and a combination
of horizontal and vertical loads have ratios that range from 0.46 for the 40t
crane, class 2, moment in the girder about the y axis to 1.80 for the 260t crane,
class 4, torsion in the girder. This means that for a crane configuration like the
40t crane, considering a class 2 crane, the horizontal load effects due to prEN
1991-3 are 2.17 times those due to SABS 0160:1989. For a crane configuration
like the 260t crane, the horizontal load effects from prEN 1991-3 are only 56%
of those given by SABS 0160:1989.
The ratios of the load effects influenced by the horizontal transverse loads
for the 5t crane and the 40t crane are similar but the 260t crane has much
higher ratios indicating that for this crane SABS 0160:1989 gives much larger
horizontal transverse loads than prEN 1991-3. The difference in the ratios be-
tween the 5t crane, 40t crane and the 260t crane can be explained by examining
the load models for the horizontal loads.
The two critical load cases for horizontal loads from SABS 0160:1989 are
misalignment of the rails or wheels and skewing of the crane in plan. The
magnitude of the loads is dependent only on the crane self weight and hoistload,
the number of wheels and the class of the crane.
The two critical load cases for horizontal loads from prEN 1991-3 are ac-
celeration of the crane bridge with the crab eccentric to the centre of mass
and skewing of the crane in plan. Both of these load models are dependent on
the parameters mentioned above as well as the geometry of the crane, i.e. the
span of the crane bridge, the wheel spacing, the minimum distance between
hoist and rail and the clearance between the wheel flange and rail. Both load
cases are most sensitive to the crane bridge span and wheel spacing.
The 5t crane and the 40t crane have similar span to wheel spacing ratios
of 6.4 and 5.4 respectively, whereas the 260t crane has a span to wheel spacing
ratio of 2.4. This could account for the large difference between the load effect
ratios between the 5t and 40t cranes and the 260t crane.
Theoretically, it seems reasonable that the horizontal loads due to acceler-
ation of the crane and skewing are dependent on the span and wheel spacing
because the horizontal loads are caused by moments acting on the crane which
are resolved into horizontal couples. SABS 0160:1989, however, does not take
this into account. The crane load models in prEN 1991-3 are therefore more
sophisticated than the crane load models in SABS 0160:1989 in that they con-
sider the effect of the different crane parameters in more detail and are more
likely to be a true representation of the behaviour of the crane.
Such large differences in the horizontal wheel loads between the two codes,
as is demonstrated by the ratios of load effects influenced by horizontal loads,
highlights the need for experimental data to verify the load models.
The load effect ratio for axial force in the girders is large compared to the
other load effect ratios. This is because SABS 0160:1989 includes the weight
of the hoistload when determining longitudinal forces caused by acceleration
of the crane and prEN 1991-3 does not. In the design of the girder, the effect
of the axial force is very small compared to the moments and torsion so the
large ratio is not of great significance.
The axial force that is carried by the bracing is due to the horizontal
longitudinal forces caused by collision of the crane with the end stops. SABS
0160:1989 does not include the hoistload in the calculation of these forces
whereas prEN 1991-3 does include the hoistload. These ratios remain constant
across the classes of crane because in neither SABS 0160:1989 nor prEN 1991-3
are the end stop forces dependent on the crane class.
The ratios of the axial forces in the bracing calculated using the SABS
0160:1989 method (a) of assuming a maximum deceleration of g, are all greater
than one, and in the cases of the 5t crane and the 260t crane are greater than
two. As expected, this shows that the simplified method in SABS 0160:1989
gives much larger values than a method which takes the buffer characteristics
into account.
The ratios of axial forces in the bracing calculated using the SABS 0160:1989
method (b) where the resilience of the buffers is taken into account are all less
than one. The methods used for the calculation of the end stop forces accord-
ing to SABS 0160:1989 method (b) and the prEN 1991-3 method differ in that
the prEN 1991-3 method specifies a force based on the spring constant of the
buffer, whereas SABS 0160:1989 does not give a specific calculation method
but merely states that the resilience of the buffers should be taken into ac-
count. The method used to calculate the end stop forces for SABS 0160:1989
method (b) was the method used by the crane manufacturers for the selection
of the buffers by considering the absorption of the crane’s kinetic energy.
For the SABS 0160:1989 method (b), the ratios of axial force in the bracing
being less than one can be partially explained by the fact that prEN 1991-3
specifies that the mass of the crane and the hoistload should be considered
whereas SABS 0160:1989 considers only the mass of the crane and not the
hoistload.
The ratios of end stop forces are small for the 40t crane compared to the
other two cranes. The 40t crane has a relatively stiff buffer (SB = 2 × 106 )
compared to the 5t crane (SB = 0.75 × 106 ) and 260t crane (SB = 1.25 × 106 )
causing the prEN 1991-3 calculation method to give larger forces.
End stop forces calculated according to prEN 1991-3 lie between the forces
from the two methods recommended by SABS 0160:1989. Therefore, depend-
ing on which method was selected by the designer previously, using the prEN
1991-3 load model would sometimes result in heavier end stops and bracing
and something in lighter end stops and bracing.
4.2.4 Design effects
As an example, to assess the effect of the combination of the various load

effects, the design of the 40t crane girder was carried out using the loads from
each code. The size of the girder was determined by keeping the proportions of
the cross sectional dimensions the same (as given in Chapter 3) and scaling the
cross section until the design resistance was equal to the design loading. The
cross sectional area of the girder was calculated, Table 4.5 gives the ratios of
the cross sectional areas calculated using loads from SABS 0160:1989 to those
calculated using loads from prEN 1991-3.
Table 4.5: Ratio of areas: SABS 0160:1989/prEN 1991-3
Class or hoist class ratio

2 1.19
3 1.20
4 1.22
The loads calculated from SABS 0160:1989 result in girders approximately

20% larger than those designed using the loads from prEN 1991-3.
The load effect ratios, given in Table 4.4, for the three representative cranes,
indicate that crane girders designed using prEN 1991-3 crane loads would differ
in shape to the those designed using the SABS 0160:1989 crane loads. The
vertical forces from prEN 1991-3 are slightly smaller than those from SABS
4.3. Design effort 99
0160:1989 and this would result in girders which are slightly shallower. For
cranes with a large span to wheel spacing ratio, the horizontal forces from prEN
1991-3 are mostly larger than those in SABS 0160:1989. Using the prEN 1991-
3 crane load models for the design of the girder for these cranes would result in
the girders with larger top flanges. In contrast, for cranes with a small span to
wheel spacing ratio, the horizontal forces from prEN 1991-3 are smaller than
those in SABS 0160:1989. Using the prEN 1991-3 crane load models would
therefore result in smaller top flanges and surge plates for these cranes.
4.3 Design effort

The design effort required for the design of the support structure using the
crane load models in SABS 0160:1989 and prEN 1991-3 was assessed with re-
spect to the amount of crane information it is necessary to obtain, the work
required to determine the loads and the amount of work required for the sub-
sequent design of the support structure.
4.3.1 Information required for design

The information required for the support structure design can be divided into
two parts, the information required to calculate the loads and the information
required to carry out the design of the crane support structure. All the crane
information listed in Table 4.6 is required by prEN 1991-3 for the calculation
of the crane wheel loads. The crane information that is required by SABS
0160:1989 is shown in Table 4.6, classifying the information into that required
for the load calculations and that required for the support structure design.
Slightly more information is required by prEN 1991-3 for the calculation
of the crane loads and design of the support structure than SABS 0160:1989.
The extra information relates mostly to the travel and hoisting speeds, wheel
drives and wheel characteristics and buffers.
Based on Table 4.6, there is slightly more effort involved in obtaining the
crane information required for prEN 1991-3 than for SABS 0160:1989. The
amount of extra effort depends on the availability of the required information
at the stage of the support structure design.
It is often difficult to obtain detailed crane information at the stage of
support structure design. This is because the crane is only installed near the
end of the construction phase of the building and the crane manufacture is
Table 4.6: Crane information required for design of support structure

SABS 0160:1989
Weight of crane bridge Loads
Weight of crab Loads
Hoistload Loads
Crane geometry
Span of crane bridge Support structure
Minimum distance between hoist and rail Support structure
Rail type Support structure
Width of top of rail Support structure
Height of rail Support structure
Steady hoisting speed
Long travel speed
Cross travel speed
Hoist class Loads
Type of load lifting mechanism
Hoistload free to swing Loads
Number of wheels Loads
Wheel spacing Support structure
Type of wheel drive
Number of single wheel drives
Behaviour of drives
Combination of wheel pairs
Guide means
Guide rollers present Loads
Clearance between rail and wheel flange
Buffers
Buffer type
Buffer characteristic (degree of plasticity)
4.3. Design effort 101
scheduled so as to be completed just before installation. This means that the

support structure design is carried out before the crane has been manufactured,
or in some cases even designed, and thus before the details of the crane are
known.
This difficulty in obtaining crane information is more relevant for process
cranes than standard cranes. Process cranes are custom made for a particu-
lar process and are typically the larger cranes, whereas standard cranes are
manufactured by the crane company and bought in an ‘off the shelf’ manner.
The crane parameters of the standard cranes would therefore be available in
advance. The breakdown of cranes sold by a leading crane manufacturer is
30% process cranes and 70% standard cranes. Therefore for 70% of cranes,
the crane information would be readily available in advance.
4.3.2 Work required for load calculations
A rough indication of the amount of work that is required for the calculation of
the loads is the number of pages of hand calculations to obtain the design crane
actions, i.e. factored crane load combinations. Using the SABS 0160:1989
load models, four pages of calculations were required to arrive at factored load
combinations compared to ten pages for the load models in prEN 1991-3.
4.3.3 Work required for support structure design
By inspection of all the ultimate limit state load combinations from SABS
0160:1989 and prEN 1991-3, two critical crane load combinations were identi-
fied for each code.
The initial calculation of the crane loads is only a small part of the total
design effort for the building housing the crane. Because there are the same
number of critical load combinations for each code, the resulting effort for the
subsequent design of the support structure would not be increased by using
prEN 1991-3.
It can be concluded that using prEN 1991-3 for the calculation of the crane
loads would not significantly increase the total work required for the design of
a structure housing an overhead travelling crane.
4.4 Summary
The crane provisions in SABS 0160:1989 were compared to those in prEN
1991-3. The comparison was carried out with respect to the load situations
modelled, the cost of the support structure and the design effort required.
prEN 1991-3 considers in more detail the actual behaviour and geometry of
the crane in the crane load modelling. More guidance is given by prEN 1991-3
on the crane load combinations. prEN 1991-3, however, lacks a load model for
the misalignment of the wheels or rails. There is no guidance given in SABS
0160:1989 on loading to be considered for the fatigue of the support structure
whereas prEN 1991-3 provides a method for the calculation of fatigue loads.
The cost of the support structure was assessed on the basis of the ratios
of load effects obtained from SABS 0160:1989 to those obtained from prEN
1991-3. The load effects were calculated for the crane girders, crane columns
and longitudinal bracing for the support structures of the three representative
cranes.
It was found that the prEN 1991-3 crane load models resulted in smaller
vertical loads than the crane load models from SABS 0160:1989. For cranes
that have a large span to wheel spacing ratio, the prEN 1991-3 load models
result in horizontal load effects that are 217% of the load effects from SABS
0160:1989. In contrast, for cranes with small span to wheel spacing ratios,
the prEN 1991-3 load models result in horizontal loads 56% of the SABS
0160:1989 load models. The resulting girder size was about 20% larger for
SABS 0160:1989 than for prEN 1991-3. The load effects in the bracing cal-
culated using the load model in prEN 1991-3 were smaller than the SABS
0160:1989 simplified method and larger than the SABS 0160:1989 more de-
tailed method.
The design effort required was assessed on the basis of information required
for the design, work required for the calculation of the crane loads and work
required for the design of the support structure. prEN 1991-3 requires more
crane information than SABS 0160:1989 and also more work for the calculation
of the crane loads. The same number of critical crane load combinations result
from each code so the subsequent design of the support structure would entail
the same amount of work.
Chapter 5
Development of limit states

equations
The development of the limit states equations entailed selecting which limit
states to consider, selecting the elements to assess for each limit state, deter-
mining the loading on the element and determining the failure mode to be
considered for the resistance of the element. The identification of the basic
variables for use in the reliability analysis is also part of setting up the limit
states equations.
The limit states that were considered for this investigation were ultimate
limit state, accidental limit state and fatigue. The elements that were chosen
for the assessment of each limit state were individual structural elements from
the three representative structures. The loading scenarios considered were
the various load combinations that include crane loads. For the ultimate limit
state this included combinations with permanent loads, wind and roof imposed
loads. The failure modes that were considered for the resistance of the elements
were the critical failure modes which, in the design, governed the element size.
Each limit state will be discussed below, along with the elements chosen for
assessment, the loading and the resistance.
The two step reliability assessment process consisting of the economic de-
sign of the member and then the reliability analysis, as described in Chapter 1,
results in two equations for each member. Firstly the design equation (includ-
ing all the partial factors) and secondly the limit state equation used as the
performance function for the reliability assessment. Both of these equations
will be outlined in this chapter.
103
104 Chapter 5. Development of limit states equations
5.1 Ultimate limit state

The structural elements considered for the assessment of the ultimate limit
state were the crane girders and those elements in the building frame that were
subject to crane loads. These elements were the crane column, the building
column and in some cases the roof members. The elements that were chosen
from each representative structure are discussed in detail in section 5.1.2.
5.1.1 Loading
The load combinations considered for the ultimate limit state were taken from
SABS 0160:1989 [1] with the addition of the crane load only case. The loading
is discussed separately for the two steps of the code calibration process: firstly
the loads imposed for the design of the elements and secondly setting up the
limit states equations for the reliability analysis.
5.1.1.1 Design
The load combinations considered for the design of the elements are outlined
below, along with the design equations including the partial load factors and
combination factors.
1. Crane load only
a) γC (Crane)
2. Permanent + Crane
a) γP 0 (Permanent)
b) γPu (Permanent) + γC (Crane)
3. Permanent + Crane + Wind
a) γPf (Permanent) + γC (Crane) + ψW C γW (Wind)

b) γPf (Permanent) + ψCW γC (Crane) + γW (Wind)
4. Permanent + Crane + Roof Imposed
a) γPu (Permanent) + γC (Crane) + ψRC γR (Roof Imposed)

b) γPu (Permanent) + ψCR γC (Crane) + γR (Roof Imposed)
5.1. Ultimate limit state 105
The crane related partial load factors and combination factors are the focus
of this investigation and are to be determined in the code calibration exercise.
The values of the remaining partial load factors and combination factors were
taken as given in SABS 0160:1989. The definitions of the partial factors and
combination factors are given in Table 5.1. The values of the partial load
factors from SABS 0160:1989 are given in Table & 5.2.
Table 5.1: Definitions of crane related partial load factors and combination factors
Symbol Definition
γC represents the factoring of the crane load by whichever code
format is being considered
ψW C combination factor for wind load as the accompanying load with
crane load as the leading load
ψCW combination factor for crane load as the accompanying load
with wind load as the leading load
ψRC combination factor for roof imposed load as the accompanying
load with crane load as the leading load
ψCR combination factor for crane load as the accompanying load
with roof imposed load as the leading load
Table 5.2: Definitions of non-crane related partial load factors

Symbol Definition Value
γP 0 partial load factor for permanent load only 1.5
γPu partial load factor for permanent load in unfavourable 1.2
combination with another load
γPf partial load factor for permanent load in favourable 0.9
combination with another load
γW partial load factor for wind load 1.3
γR partial load factor for roof imposed loads 1.6
Example calculations of all the loads for the 40t crane are given in Appen-
dix A.
The crane loads consist of both vertical and horizontal wheel loads. prEN
1991-3 specifies six ultimate limit state crane loading scenarios consisting of
combinations of vertical and horizontal crane wheel loads. The two combi-
nations that were found by inspection to be critical were load combinations
one and five which are described below. From all the ultimate limit state load
combinations shown in Chapter 4, the two critical combinations are shown in
Figures 5.1 & 5.2. The loads shown are characteristic, unfactored wheel loads.
Load Combination 1 : Represents the crane lifting a load off the ground
and accelerating with the crab eccentric to the centre of mass of the
crane.
Figure 5.1: Load Combination 1: φ1 (Self weight) + φ2 (Hoistload) + φ5 (Acceleration

of crane bridge)
Load Combination 5 : Represents the crane travelling along the runway at

constant speed and skewing of the crane bridge occurring.
Figure 5.2: Load Combination 5: φ4 (Self weight) + φ4 (Hoistload) + (Skewing)
Where:
φ1 – dynamic factor applied to the self weight of the crane to take into account
the dynamic effects of lifting a load
φ2 – dynamic factor applied to the hoistload to take into account the dynamic
effects of lifting a load
φ4 – dynamic factor to take into account the dynamic effects induced when
the crane travels along the rails
φ5 – dynamic factor to take into account the dynamic effects of the drive forces
The crane loading for the building frames was calculated by considering the
different possible positions of the crane which gave different loading cases. The
resulting loading cases were either the vertical loads or the horizontal loads at
a maximum with the maximum loads either on the left or right hand side, for
each crane load combination.
The load effects in the crane girders were calculated by simple structural
analysis methods. A structural analysis computer program [31] was used for
the determination of the load effects in the building frames. Linear, two di-
mensional analyses of the frames were carried out.
5.1.1.2 Reliability analysis
The actual loads that the elements were subject to were considered for the re-
liability analysis. The methods of calculating these actual loads were assumed
to be the same methods that are given in the codes i.e. those used for the
economic design of the element. Modelling factors were included to take into
account the uncertainties in the calculation of the loads and load effects. The
modelling factors will be discussed in more detail below.
The load effects in the girders were still determined by simple structural
analysis methods, however, it was not practical to use the structural analysis
program directly for the determination of the load effects in the frame, due
to the iterative nature of the reliability analysis procedure. In order to be
able to determine the load effect in a member, given the load applied to the
frame, influence coefficients were calculated. The basis for the calculation of
the influence coefficients was that a load effect in a linear analysis can be
modelled by the following equation as given by Ellingwood et al. [32]:
Qi = ci θi Ai (5.1.1)
Where:
Qi – load effect
ci – influence coefficient for calculation of load effects
θi – modelling uncertainty in the calculation of the load effect
Ai – applied load
Each load was applied separately to the frame in the structural analysis
program and the influence coefficients were found by:
Qi
ci = (5.1.2)
Ai
Modelling uncertainties for the loading arose from two sources: the uncer-
tainty in the calculation of the loads and the uncertainty in the calculation of
the load effects. The modelling uncertainties involved in the load calculations
affected only the crane loads and were included as:
Avertical wheel load = θv × Calculated vertical crane wheel load

(5.1.3)
Ahorizontal wheel load = θh × Calculated horizontal crane wheel load
The load effect calculations that included modelling uncertainties were the
additional moment due to slenderness effects in the concrete column for the
40t crane, the stress in the girders for the 40t crane and the axial forces and
bending moments in the frame elements from the structural analysis program.
These modelling uncertainties were included in the limit states equations in
the manner shown above.
5.1.2 Resistances
The method of calculating the resistances of the elements for both the economic
design and the reliability analysis were taken from the South African materi-
als design codes (SANS 10162-1:2005 [11]; SABS 0100-1:1992 [10]). Modelling
factors were included in the reliability analysis to account for the uncertain-
ties involved in the resistance calculations. The modelling uncertainties were
included in the limit states equations in a similar manner to the modelling
uncertainties for the loads and load effects.
The structural elements that were considered for each representative struc-
ture are discussed below, along with the failure modes that were considered
for the economic design and reliability analysis.
5.1.2.1 5t crane
5.1.2.1.1 Crane girder
The two span continuous crane girder for the 5t crane was subject to combined
vertical and horizontal bending due to the vertical and horizontal crane wheel
loads. The design equation and the limit state equation were interaction equa-
tions for the combination of vertical and horizontal bending. The different
load cases for the assessment of the 5t crane girder and the specific failure
modes for each load case are given below.
Bending moment diagrams for each load case are given in Appendix B.
1. Maximum positive vertical moment

Three resistance checks for the crane girder were carried out when the
maximum positive vertical moment was considered.
a) Combined vertical and horizontal bending for the top flange in com-
pression in the first span of the girder, i.e. the span subject to the
maximum positive vertical moment. The failure mode considered
for vertical bending was lateral torsional buckling and the failure
mode for horizontal bending was elastic bending.
b) Tension in the bottom flange of the first span of the girder. The
failure mode considered was elastic bending.
c) Accompanying negative vertical moment in the second span of the
girder. The failure mode considered was lateral torsional buckling.
2. Maximum negative vertical moment over the internal support. The fail-
ure mode considered was lateral torsional buckling.
3. Maximum horizontal moment. Combined vertical and horizontal bend-

ing was considered for the top flange in compression. The failure mode
considered for vertical bending was lateral torsional buckling and the
failure mode for horizontal bending was elastic bending.
5.1.2.1.2 Building frame
The structural elements that were considered for the building frame of the 5t
crane were the column, the corbel connection to the column and a roof truss
element. These three elements are discussed separately below:
1. Column
The failure mode considered for the column was a combination of axial
compression and bending about the major axis. Three failure modes
were considered: cross sectional resistance, buckling about the major
axis and lateral torsional buckling.
2. Corbel connection to the column

Two separate connection details were considered for the corbel to column
connection: a bolted connection and a welded connection. The connec-
tion was subject to both a bending moment and a shear force from the
vertical crane wheel loads. The failure mode for the bolts and the weld
was combined shear and tension.
3. Truss element
The truss element selected for analysis was the one subject to the largest
stress. Different load cases and combinations resulted in the truss ele-
ment being subject in some cases to axial compression and in other cases
to axial tension. Both of these failure modes were considered.
5.1.2.2 40t crane
The crane girder was subject to vertical and horizontal crane wheel loads.
Eccentricity of the vertical load combined with the horizontal load caused
torsion in the girder. An analysis method given by Rowswell [30] was used
where axial stresses in the top and bottom flanges, due to the vertical and
horizontal bending and torsion, were considered as the load effects for the
crane girder. The resistance of the flanges was taken as the yield stress.
The structural elements of the building frame that were considered were the
concrete crane column and the steel building column. The effect of the crane
loads on the roof members was very small and they were thus not considered
for analysis.
1. Concrete crane column.
The concrete crane column was subject to combined axial compression

and bending about its major axis. According to the South African con-
crete design code, SABS 0100-1:1992 [10], the column was classified as
slender and an additional moment about the minor axis due to slender-
ness effects was induced. An approximate method of analysis for columns
subject to both axial force and bending moment is given by SABS 0100-
1:1992 [10] in the absence of a more accurate analysis method. It was
decided to use a more accurate analysis method for this investigation. A
method of analysis designed for use with a computer has been given by
Kwan & Liauw [33]. This method combined with general concrete de-
sign principles given by Nawy [34] and the specifications given by SABS
0100-1:1992 was implemented as the more accurate analysis method.
The analysis was an iterative process first using elastic methods to deter-
mine an initial position of the neutral axis following the method given by
Kwan & Liauw [33] and then using the compressive stress block method
prescribed by SABS 0100-1:1992 to find the final position of the neutral
axis and the resistance of the section. The output from the analysis was
the ratio of the resistance of the column to the applied loads.
2. Steel building column
The steel building column was subject to a combination of axial com-

pression and bending about the major axis. As for the column of the
5t crane, three failure modes were considered: cross sectional resistance,
buckling about the major axis and lateral torsional buckling.
5.1.2.3 260t crane
The crane girder for the 260t crane consisted of a main girder with a surge
plate and an auxiliary lattice girder. The main girder supported the vertical
crane loads. A combined girder consisting of the top flange of the main girder,
the surge plate and the top chord of the auxiliary girder carried the horizontal
crane wheel loads. Elastic bending was considered for the resistance of the
girders.
The structural elements that were considered in the building frame were the
steel crane column, the steel building column and two roof truss members.
1. Steel crane column

The crane column was subject to combined axial compression and bend-
ing about the minor axis. The failure modes that were considered were
cross sectional resistance and buckling about the minor axis.
2. Steel building column

The building column was subject to combined compression and bending
about the major axis. As for the column of the 5t crane, three failure
modes were considered: cross sectional resistance, buckling about the
major axis and lateral torsional buckling.
3. Roof truss elements

As with the 5t crane, some load combinations caused tension in the roof
truss member and some caused compression. Both of these failure modes
were considered.
5.1.3 Limit states equations

The general performance function for the ultimate limit states was:
X
g (X) = RθR − ci Ai θL (5.1.4)
Where:
R – resistance of the member
θR – modelling uncertainty in the calculation of the resistance
Ai – applied load
θL – modelling uncertainty in the calculation of the load effects
In cases when the failure mode considered was a combination of axial com-
pression and bending, the limit state equation took the form:
5.2. Accidental limit state 113
µ P P ¶
( ci Ai θL )axial ( ci Ai θL )bending
g (X) = θR − + (5.1.5)
Raxial Rbending
5.2 Accidental limit state

The possible accidental load situations for cranes are outlined below:
1. The crane bridge running into the end stops.
This is considered as an accidental load because there are generally limit

switches on the runways close to the ends which cause the crane to stop
when it passes over them. In the event of the limit switches failing, the
crane will fail to stop and the crane will run at full travel speed into the
end stops. The buffers on the crane are used to lessen this impact force.
2. The crab running into the end stops.
This is a similar situation to the crane bridge running into the end stops.
There are generally limit switches to stop the crab before it reaches the
end stops. This load case is considered for the ultimate limit state in that
prEN 1991-3 considers this case to allow for the effects of acceleration
or deceleration of the crab. In the ultimate limit state this load case is
not critical and it will thus not be considered for the investigation of the
accidental limit state.
3. The hoistload colliding with a column.
This could occur if the crane bridge was between a pair of columns with
the crab travelling such that the line of travel of the crab was directly
towards a column, and the crab suddenly stopped. In this situation the
hoistload could swing and collide with a column.
The case that was chosen for the assessment of the accidental limit state
was the crane bridge running into the end stops because this is the accidental
load that is provided for in prEN 1991-3.
For all the cranes, the longitudinal buffer forces are resisted by the longi-
tudinal runway bracing and all have the same failure mode so the 40t crane
was selected to be representative of all the cranes for the accidental limit state
and the accidental load case has been considered for the 40t crane only.
5.2.1 Loading
The loads caused by the crane bridge running into the end stops are longitudi-
nal loads along the runway. The loads are calculated assuming that the crane
is travelling at 70% of the full rated long travel speed, taking into account the
mass of the crane and hoistload. The load combination is shown in Chapter
4, the load calculations are given in Appendix A.
This method was used for the calculation of the load for the economic
design and the reliability assessment. A modelling factor to take into account
the uncertainty of the load calculation was included in the limit state equation.
5.2.2 Resistance
The longitudinal loads from the crane bridge colliding with the end stops are
resisted by the longitudinal runway bracing. A diagram of the bracing for the
40t crane is given in Figure 3.9 on page 76. The behaviour of the bracing
was modelled as the compressive leg buckling and the tensile leg carrying the
load. The resistance was thus considered as the resistance of one bracing leg
in tension.
5.2.3 Design equation

The design equation for the accidental case is given by:
φR Afy = γA HB (5.2.1)
Where:
φR – partial resistance factor
A – cross sectional area of one bracing leg
fy – yield strength of steel
γA – partial load factor for accidental crane actions
HB – longitudinal load from the crane bridge colliding with the end stops
5.2.4 Limit state equation

The limit state equation for the accidental limit state is given by:
5.3. Fatigue 115
g(X) = Afy − HB θHB (5.2.2)
Where:
θHB – modelling factor to allow for uncertainty in the load calculation
5.3 Fatigue
The procedure for the calculation of fatigue reliability is the same two step
procedure carried out for the other limit states. The first step is the economic
design of the element considered according to the code provisions to ensure that
the resulting element exactly meets the code requirements. The second step is
to carry out the reliability analysis where the loading and the resistance in the
limit state equation represent the actual loads imposed on, and the resistance
behaviour of the element.
The elements that are most susceptible to crane induced fatigue are those
most directly loaded by crane loads i.e. the crane girder, columns and related
connections. The elements chosen for the assessment of fatigue reliability in
this study were taken from two of the three representative cranes and are:
1. 40t crane girder, fatigue at the bottom of the intermediate stiffeners
2. 40t crane girder, fatigue of the top flange to web weld caused by rotation
of the top flange
3. 5t crane, corbel to column welded connection
The 260t crane girder is similar to the 40t crane girder in that they are
both welded plate girders and would therefore display the same behaviour.
The surge plate for the 260t crane girder would to some extent reduce the top
flange rotation, therefore the 40t crane girder represents the worst case. The
5t crane girder is a two span continuous hot rolled I section with a channel on
the top flange. The only weld in a tension region is the weld connecting the
channel to the top flange over the support. This is a longitudinal weld in the
same direction as the stresses and is thus not critical for fatigue.
5.3.1 Economic design method

The two fatigue design methods that were discussed in section 4.1.4 on page
90 were considered for the economic design. The first method is that given in
prEN 1991-3 where the fatigue loads and resistance are separated and prEN
1991-3 specifies the crane fatigue loading. The second method is that assumed
by SABS 0160:1989 where the crane fatigue loading is not specified and it is
the responsibility of the support structure designer to model the behaviour of
the crane. In both cases, the fatigue resistance, in the form of S-N curves, was
taken from the South African steel design code SANS 10162-1:2005 [11].
5.3.1.1 prEN 1991-3 fatigue loads
The method given in prEN 1991-3 of determining the crane loads to be consid-
ered for fatigue is described in detail in Chapter 2 but will be briefly outlined
below.
The basis for the determination of the fatigue loads in prEN 1991-3 is the
normalisation of the maximum nominal wheel load, resulting in the equivalent
constant amplitude load for two million cycles. This normalised load is called
the ‘fatigue damage equivalent load’. The rationale behind the calculation of
this load is that two million crane cycles with the crane wheel loads equal to
the fatigue damage equivalent load, will result in the same amount of fatigue
damage as the normal operating conditions of the crane with the number of
cycles it is likely to perform over its lifetime. The normalisation is carried out
by multiplying the maximum nominal wheel load by a factor (λ):
Qe = φf at × λ × Qmax (5.3.1)
Where:
Qe – fatigue damage equivalent load
φf at – damage equivalent dynamic impact factor, the factor applied to the

1
crane self weight φf at,1 = 2 (1 + φ1 ) and the factor applied to the
1
hoistload φf at,2 = 2 (1 + φ2 )
Qmax – maximum nominal wheel load
The λ factor is determined from the fatigue class of the crane. Cranes
are classified into fatigue classes depending on the total number of cycles (N)
5.3. Fatigue 117
performed over the lifetime of 25 years and the load spectrum (kQ). One λ
factor is given for each fatigue class for the calculation of normal stresses and
another for the calculation of shear stresses.
As discussed in Chapter 2, the load spectrum definition in prEN 1991-3 is
based on the fatigue of the crane itself and not the support structure. The
load spectrum used for this investigation has been modified to more accurately
represent the stress cycles caused by the crane wheel loads on the girder and
is given by:
X µµ Qi,j ¶m ni,j ¶
kQ = P (5.3.2)
max Qi ni,j
j
Where:
Qi,j – wheel load of value j for wheel i
maxQi – maximum wheel load for wheel i
ni,j – number of applications of range j for wheel i

P
ni,j – total number of load applications
The fatigue stresses in the element were calculated by considering one

movement of the crane along the girder with the wheel loads equal to the
fatigue damage equivalent load. The stress cycles causing fatigue were calcu-
lated and in the event of there being more than one stress cycle for one pass
of the crane over the girder, Miner’s rule of accumulative damage was used. A
description of the design procedure for each element is given in section 5.3.2.
5.3.1.2 Resistance
The resistances of the elements were determined using the S-N curves in SANS
10162-1:2005 [11]. The S-N curves in SANS 10162-1:2005 [11] consist of two
slopes, the first in the region of higher stress with a slope of m1 = 3 and the
second in the region of lower stress with a slope of m2 = 5. The equation for
the S-N curves is given by:
A
(S)m = (5.3.3)
N
Where:
S – stress amplitude
A – material parameter
N – number of cycles to failure for a given stress amplitude, S
If only one stress cycle resulted from the stress calculations, the fatigue
resistance was taken as the stress level at two million cycles on the relevant
S-N curve. If more than one stress cycle was obtained, the parameters of the
relevant S-N curve were used to calculate the accumulative fatigue damage
according to Miner’s rule and the resistance was taken as the fatigue damage
at failure equal to one.
5.3.2 Economic design

The economic design was carried out by determining the size of the member
for which the fatigue load effect calculated with the fatigue damage equivalent
load was exactly equal to the resistance.
In order to carry out the economic design, the stress histories for the rele-
vant fatigue stresses in the three elements were required.
The stress history was produced by calculating the stresses at the point
where fatigue was considered as the crane moved incrementally along the
girder. The stress histories for one crane cycle (loaded and unloaded) for the
40t crane girder intermediate stiffener and the 5t corbel to column connection
are shown in Figures 5.3 and 5.5. The stress history for the 40t crane girder
top flange to web weld is shown in Figure 5.4 where only the loaded part of
the cycle is shown. In Figures 5.3 and 5.5 the movement of the crane along the
girder is expressed in terms of time, this was assuming an arbitrary velocity
of the crane of 30 m/min.
A brief description of the specific design procedure for each element is given
below.

The stress causing fatigue at the bottom of the intermediate stiffener
was the longitudinal tension due to global bending of the beam. It was
5.3. Fatigue 119
assumed that there was a stiffener at the point of maximum bending in

the beam i.e. at 5.1m from the support. Figure 5.3 shows the stress
history at this point for the crane travelling over the girder twice, the
first time the crane is travelling from left to right along the girder and
the second time the crane is returning in the other direction. The first
peak in the first pass of the crane is when the leading wheel is at 5.1 m
and the following wheel is at 0.7 m, the second peak is when the following
wheel is at 5.1 m and the leading wheel is off the girder.
Time history of stress in the 40t crane girder at the bottom of the stiffener for a Class 4 crane
70
60
50
Stress (MPa)
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Figure 5.3: Stress history for 40t crane girder at bottom of stiffener
In this case two stress cycles resulted from one pass of the crane over the
girder. A rainflow counting method was used to identify the stress cycles
and their amplitudes. The first cycle was identified as ranging from zero
to 64 MPa (the top of the first peak). The second cycle was identified
as ranging from 40 MPa (the lowest point between the peaks) to 58 MPa
(the top of the second peak) resulting in a stress amplitude of 18 MPa.
The bottom of the stiffener was classified as detail category C1 which is

described as ‘Base metal at the toe of transverse stiffener-to-flange and
transverse stiffener-to-web welds’ according to SANS 10162-1:2005 [11].
Miner’s rule for accumulative damage was used to determine the damage
caused by the two stress cycles. The stresses were calculated using the
fatigue damage equivalent load which is normalised to two million crane
cycles, therefore the damage caused by two stress cycles was multiplied
by two million to give the total damage over the girder lifetime.
The fatigue resistance was taken as the fatigue damage at failure equal
to one.
The method of carrying out the economic design was to scale the di-
mensions of the crane girder, keeping the original proportions, until the
total damage resulting from two million cycles of the fatigue damage
equivalent load was exactly equal to the fatigue damage at failure.
of the top flange
The wheel loads that were considered for this fatigue case were the verti-
cal load and the horizontal transverse load due to the acceleration of the
crane bridge with the crab eccentric to the centre of mass. Considering
the horizontal load as a fatigue load is a conservative but not unreal-
istic assumption. In the case where a crane has very fixed movements
around a building it is possible that it will always accelerate over the
same girder. For example a ladle crane in a steelworks building always
collects the molten steel at one end of the building and runs the full
length of the runway and back again with an empty ladle. In this case
the crane bridge always accelerates in the same place. The configuration
of the horizontal transverse loads caused by acceleration of the crane
bridge are shown in Figure 5.1.
The stresses that caused fatigue in the top flange to web weld were global
and local bending stresses and global and local shear stresses. These
stresses were calculated using the method given in the steel design part
of the Eurocode (ENV 1993-6 [35]) because no guidance is given on the
calculation of these type of stresses in the South African design codes.
The individual stresses are outlined below.
• Compressive longitudinal stress due to global bending

• Global shear due to vertical loads
• Local shear due to vertical loads
• Local vertical compressive stresses due to concentrated wheel loads
• Vertical bending stresses due to the rotation of the top flange caused
by the horizontal wheel load and the eccentricity of the vertical
wheel load. When considering one side of the web, the bending
5.3. Fatigue 121
stress for the first wheel was tensile due to the horizontal and ver-
tical loads causing rotation of the top flange away from the side
of the web being considered. The bending stresses for the second
wheel were compressive because the horizontal wheel loads act in
opposite directions (see Figure 5.1) and the vertical and horizontal
forces from the second wheel tend to rotate the flange towards the
side of the web being considered.
The stress that was considered for the fatigue analysis was the maximum
resulting principal stress. The stress history in terms of the resultant
principal stress from the stresses given above for one pass of the crane
over the girder is shown in Figure 5.4. The stress cycles were broad
banded so a rainflow counting method was used to identify the cycles
and find their amplitudes. Three stress cycles were identified:
• the first cycle ranges from zero to 80 MPa and back to zero.
• the second cycle ranges from zero, up to 20 MPa, down to -50 MPa
and back up to zero resulting in an amplitude of 70 MPa.
• the third cycle ranges from zero up to 8 MPa and back to zero.
Principal stresses at 5100mm, at the top flange to web weld in the crane girder
100
80
60
Principal stresses (MPa)
40
20
−20
−40
−60
0 2000 4000 6000 8000 10000 12000
Position of the leading wheel (mm)
Figure 5.4: Stress history for 40t crane girder at top flange to web weld
The top flange to web weld was classified as fatigue detail B which is de-
scribed as ‘Built up members - Base metal and weld metal in components,
without attachments, connected by continuous full penetration groove

welds with backing bars removed’ according to SANS 10162-1:2005 [11].
Miner’s rule for accumulative damage was used to determine the damage
caused by the three stress cycles and the damage was multiplied by two
million to get the total damage over the lifetime of the girder. The
resistance of the girder was taken as the fatigue damage at failure equal
to one.
The method of carrying out the economic design was the same as that
for the fatigue at the bottom of the intermediate stiffener, viz. to scale
the dimensions of the crane girder, keeping the original proportions, until
the total damage resulting from two million cycles of the fatigue damage
equivalent load was exactly equal to the fatigue damage at failure.
The weld that was considered for the corbel to column welded connection
was the horizontal weld along the top flange of the corbel. The stress
causing fatigue was the tension stress from the moment resulting from the
eccentricity of the crane vertical load. It was assumed that the vertical
welds carried the shear and the horizontal welds carried the moment.
Figure 5.5 shows the stress history for two passes of the crane along
the girder. In this case there is only one stress cycle for each pass of
the crane. The fatigue loading was taken as the amplitude of the stress
cycle.
Time history of tension stresses in the 5t crane corbel weld for a Class 4 crane
0.16
0.14
0.12
0.1
Stress (MPa)
0.08
0.06
0.04
0.02
0
0 20 40 60 80 100 120 140 160
Time (s)
Figure 5.5: Stress history for 5t crane at corbel to column welded connection
5.3. Fatigue 123
The corbel to column welded connection was classified as detail category

C with an additional reduction factor to allow for the fact that it is a
transversely loaded fillet weld, according to SANS 10162-1:2005 [11]. The
resistance of the weld was taken as the stress in the S-N curve for detail
category C at two million cycles, multiplied by the reduction factor.
The economic design was carried out by keeping the size of the column
and the corbel constant and determining the weld thickness required for
the fatigue resistance to be exactly equal to the fatigue loading.
The difference between the fatigue loading calculated according to the

method given in prEN 1991-3 and that calculated by considering the crane be-
haviour was assessed by calculating the damage in the element which had been
economically designed using prEN 1991-3, resulting from simulating the crane
behaviour. The assessment was carried out by simulating the crane behaviour
to obtain a stress history for the element being considered and determining
the resulting fatigue damage using Miner’s rule of accumulative damage.
Two aspects of the crane behaviour were simulated in order to produce a
stress history:
1. Crane movement along the gantry
One crane cycle was defined as the crane lifting a load at one end of the
gantry, travelling along the gantry, setting down the load and travelling
back, unloaded, to the starting point. It was assumed that the crane
travels over the girder being considered, twice for each crane cycle; the
first time carrying a load and the second time with no load.
2. Crane wheel loads
To simulate the crane loading a set of hoistload values and crab posi-
tions were generated. In order to assess the fatigue reliability of the code
provisions, hoistload values were generated to give a specified load spec-
trum value kQ. These load spectrum values were generated considering
the crab always at the extreme of its travel, closest to the wheel being
considered. The generated hoistload values were used to calculate the
crane wheel loads.
5.3.3 Reliability analysis
In the reliability analysis the actual loads applied to the element were modelled
along with the actual resistance of the element.
The actual loading for the reliability analysis was determined by simulating
the crane behaviour to produce a stress history for the member and using
Miner’s rule to determine the accumulative damage. The actual resistance of
the element was taken as the fatigue damage at failure with a mean of one.
5.3.3.1 Loading
The simulation of the crane behaviour was carried out as described in the
previous section. The stress histories for the three elements are shown in
Figures 5.3 - 5.5. In the cases where the stress histories were broad-banded, a
rainflow counting method was used to identify the stress cycles [36] as described
in Section 5.3.2. Miner’s rule for accumulative damage was used to determine
the fatigue damage caused by the crane loads.
The properties of the S-N curves for the reliability analysis were derived
from those in SANS 10162-1:2005 [11]. The statistical models used for the
fatigue analysis are discussed in Chapter 6. The S-N curves have two slopes,
the first in the region of higher stresses with a gradient of m1 = 3, the second in
the region of lower stresses with a gradient of m2 = 5. In order to simplify the
reliability analysis procedure, investigations were also carried out considering
the S-N curve having a single slope of m = 3. The single slope S-N curve was
taken as an extension of the first slope in the double slope curve i.e. instead
of a change of slope the curve continues into the region of lower stresses with
a slope of m = 3.
5.3.3.2 Resistance
The resistance of each element was taken as the fatigue damage at failure with
a mean of one.
5.3.3.3 Limit state equations
For the case when a double slope S-N curve was considered, the limit state
equation is given by:
5.4. Summary 125
 
X (Si )m1 ni X (Sj )m2 nj
g (X) = D −  +  (5.3.4)
A1 N A2 N
i j
Where:
D – fatigue resistance
Si – amplitude of stress cycle i which falls into the higher stress region
Sj – amplitude of stress cycle j which falls into the lower stress region
A1 – fatigue material property from the S-N curve for slope 1
A2 – fatigue material property from the S-N curve for slope 2
m1 – slope of S-N curve in region of higher stresses
m2 – slope of S-N curve in region of lower stresses
ni – number of stress cycles with amplitude Si
nj – number of stress cycles with amplitude Sj
N – total number of stress cycles
In the case where a single slope S-N curve was considered, the limit state
equation is given by:
X (Si )m1 ni
g (X) = D − (5.3.5)
A1 N
i
Where:
Si – in this case is the amplitude of stress cycle i for stresses in the higher and
lower regions.
5.4 Summary
The limit states equations that will be used for the reliability assessment and
code calibration of the crane load models were developed. The three repre-
sentative cranes discussed in Chapter 3 are used as example structures for the
code calibration. Limit states equations were developed considering specific
elements of the support structures for the three representative cranes. Because
specific examples structures are considered for the reliability analysis, the limit
states equations include both the loading on the element and the resistance of
the structural element.
The three limit states that were considered critical for crane support struc-
tures and were considered for the code calibration were ultimate limit state,
accidental limit state and fatigue.
A methodology was demonstrated on how to obtain a design that just
satisfies the code requirements, the so called ‘economic design’. The element so
designed is then used in the reliability analysis. This ensures that the reliability
of the code requirements are assessed without any conservatism included due
to practical rounding of elements sizes.
The crane loads that were considered for the ultimate limit state were the
loads due to normal operation of the crane. The structural elements from the
three representative structures that were considered for the ultimate limit state
were those that were subject to crane loads, i.e. crane girders, crane columns,
building columns and the roof trusses from the 5t and 260t cranes.
The load combinations that were considered for the ultimate limit state
were crane load only, crane with permanent load, crane with permanent and
wind loads and crane with permanent and roof imposed loads. The critical fail-
ure mode, i.e. the failure mode that governed the element size, was considered
for the resistance of the element.
The economic design of the elements was carried out considering the crane
loads from prEN 1991-3, the permanent, wind and roof imposed loads from
the South African loading code SABS 0160:1989 [1] and the resistances from
the South African materials codes.
For the reliability analysis, the actual loads that the element were subject
to and the actual resistance behaviour was considered. The true behaviour of
both the loads and resistances was assumed to be represented by the models
given the loading and materials codes, with the inclusion of modelling uncer-
tainties. The modelling uncertainties were included in the limit states equa-
tions by multiplying the load or resistance, calculated from the code models,
by the relevant modelling factor.
The limit states equations for the ultimate limit state were defined for the
general case when only one load effect was considered, e.g. tension in a roof
truss member, and for the interaction case, e.g. combination of axial load and
bending moment for the columns.
5.4. Summary 127
The situation considered for the accidental limit state was the case that is
provided for in the crane load models in prEN 1991-3, viz. the crane running
into the end stops on the end of the runway. The structural element that was
considered for the accidental limit state was the structural element that resists
the resulting longitudinal force, i.e. the longitudinal runway bracing.
Three fatigue situations were considered as critical for the three crane con-
figurations investigated and were considered as representative of fatigue design
of crane support structures. The three fatigue situations were: 5t crane - cor-
bel to column welded connection, 40t crane girder - bottom of the intermediate
stiffener, 40t crane girder - top flange to web weld.
The loading considered for the economic design was calculated using the
method given in prEN 1991-3 of calculating a fatigue damage equivalent load
which is the constant amplitude crane wheel load normalised to two million
cycles which will result in the same fatigue damage as the normal crane op-
eration. The fatigue resistance for the economic design was determined from
the S-N curves in the South African steel design code [11].
The loading considered for the fatigue reliability analysis was determined
by simulating the crane behaviour to obtain stress histories for the fatigue de-
tail being considered. Two aspects of the crane behaviour were simulated: the
loads lifted by the crane and the movement of the crane around the building.
The resulting stress cycles were counted using a rainflow counting method and
the fatigue damage due to the crane loads was calculated using Miner’s rule
for accumulative damage.
The variables that have been selected from the limit states equations to be
considered as random variables for the reliability analysis are identified and
discussed in Chapter 6.
Chapter 6
Stochastic models
The stochastic models for the material properties, geometric properties, loads
and modelling uncertainties that have been taken from literature are discussed
here. The new stochastic models that were developed for this code calibration
(viz. the stochastic models for the crane hoistload) are discussed in Chapter
7.
6.1 Basis for statistical modelling
There are two types of statistical uncertainties: aleatory uncertainties and

epistemic uncertainties. Aleatory uncertainties are due to the inherent vari-
ability and randomness of basic variables and epistemic uncertainties are due
to lack of knowledge, imperfect representations of the real world or imperfect
predictions.
The two types of uncertainties may be combined in the following manner
[32]:
If X represents a basic variable then estimates of the mean and coefficient
ˆ and δ̂ , can be obtained from an idealised model. Then
of variation of X, X̄ x
δ̂x represents the aleatory uncertainty, resulting from the inherent variability
of X and does not include all sources of uncertainty which contribute to the
total uncertainty in X. The mean and coefficient of variation attributed to the
additional (epistemic) uncertainty can be treated as a bias factor B with mean
and coefficient of variation B̄ and δB . Then the true mean and coefficient of
variation of the basic variable are given by:
129
130 Chapter 6. Stochastic models
ˆ
X̄ = B̄ X̄ (6.1.1)
q
δX = δ̂X 2 + δB 2 (6.1.2)
6.2 Sources of information
Two fundamental sources of information have been consulted on the statistical

modelling of the basic variables:
1. The work that was carried out for the development of the American
National Standard A58, Ellingwood et al. [32] circa 1980. Extensive
investigations were carried out on loads and resistances, both material
properties and resistance models. Due to the importance of the devel-
opment of the code, the work has been subject to extensive peer review.
The references which form part of this study are: [32; 37; 38; 39; 40; 41;
42; 43; 44; 45; 46; 47; 48]
2. The probabilistic model code developed by the Joint Committee on

Structural Safety, JCSS [36], in 2001. This model code was developed
with the intention of compiling the information required to design new
structures or assess existing structures probabilistically. This code has
also undergone extensive peer review.
The remaining references in this chapter do not form a part of either of the
two projects mentioned above.
6.3 Material properties
6.3.1 Structural steel
The steel was assumed to have been drawn from different mills due to the fact
that this investigation deals with a code calibration and the code considers
general structures.
6.3. Material properties 131
6.3.1.1 Yield strength of structural steel, (fy )
fy,nom = 300 MPa µfy = 315 MPa δfy = 10% Distribution = LN
A summary of the statistical properties given in the literature is shown in

Table 6.1. The nominal value of the yield strength has been taken as 300 MPa.
Table 6.1: Statistical parameters of the yield strength of structural steel

Item Mean c.o.v. Distr. Reference
MPa %
Steel members 316.5 7 LN JCSS [36]
Flanges 315 10 N Galambos & Ravindra [37]
Ellingwood et al. [32]
Bjorhovde et al. [38]
Webs and plates 330 11 N Galambos & Ravindra [37]
Bjorhovde et al. [38]
Steel members 321 12 N Ellingwood & Reinhold [39]
Steel members 315 10 N Cooper et al. [40]
Ravindra & Galambos [41]
The distribution that was used was based on that recommended by JCSS
[36]. The main reasons for this are that it is a more recent publication and
that it is accepted as the standard for statistical models. The parameters of
the distribution given by JCSS [36] are similar to those given in the other
references and a Lognormal distribution is thought to be more suitable to
resistance parameters as it cannot go below zero. The coefficient of variation
was increased from 7% to 10% to reflect the additional uncertainty due to the
spread in mean values from the various sources.
6.3.1.2 Moduli of elasticity (E, G)
µE = Enom δE = 3% Distribution = N
µG = Gnom δG = 3% Distribution = N
The statistical properties given above are recommended by JCSS [36]. The
statistical properties recommended by Galambos & Ravindra [37] are the same
except that the coefficient of variation is given as 6%. The models given by
JCSS [36] were used for the same reasons as for the yield strength of structural
steel.
6.3.2 Bolts
The stochastic models for the properties of bolts were taken from Fisher et al.
[48].
6.3.2.1 Ultimate tensile strength (fu )
µfu = 1.2fu,nom δfu = 7% Distribution = N
The statistical parameters of the ultimate tensile strength of bolts were

obtained by Fisher et al. [48] from a number of sets of published experimental
data obtained from tests on high strength bolts. The statistical model for A325
bolts was used. The A325 bolts have a minimum tensile strength of 845 MPa
which corresponds to the Grade 8.8 bolts, considered for the code calibration,
which have a minimum tensile strength of 800 MPa.
τ
6.3.2.2 Ratio of shear strength to tensile strength ( σ
)
³τ ´
= 0.60 µ στ = 0.625 δ στ = 5.3% Distribution = N
σ nom
Design codes specify a ratio of shear strength to tensile strength for the
calculation of the resistance of bolts subject to shear. The statistical para-
meters obtained by Fisher et al. [48] for the ratio of shear to tensile strength
were based on experimental data and are valid for both the A325 bolts and
the higher strength A490 bolts.
6.3.3 Welds
The stochastic models for the strength of the welds were taken from Fisher
et al. [48]. The models were based on experimental data obtained from strength
tests of welds from various electrode types. Longitudinal fillet welds were
considered as they result in the lower bound of strength.
6.3.3.1 Ultimate tensile strength (fuw )
µfuw = 1.05fuw,nom δfuw = 4% Distribution = N
τ
6.3.3.2 Ratio of shear strength to tensile strength ( σ
)
³τ ´
= 0.67 µ στ = 0.84 δ στ = 10% Distribution = N
σ nom
Design codes specify a ratio of weld shear strength to tensile strength for
the assessment of welds subject to tension and shear. Experimental data was
assessed by Fisher et al. [48] to determine the statistical parameters of the
ratio of shear to tensile strength.
6.3.4 Concrete
6.3.4.1 Compressive strength of concrete (fc )
fc,nom = 30 MPa µfc = 32 MPa δfc = 15% Distribution = N
The assumptions made in the statistical modelling of the compressive

strength of concrete were [32]:
1. The quality of construction is average. Because this is a code calibration
exercise, the models should be representative of the overall variability in
construction.
2. The ‘static’ strengths are considered, i.e. assuming slow loading rates.
For concrete compressive strength this relates to testing at a rate of one
hour to failure. The slower the rate of loading, the lower the strength
of the concrete. Considering the static strength is thus a conservative
assumption.
3. Long term changes in the concrete and reinforcing strengths are ignored.
This has been shown to be conservative as the strength of reinforced
concrete increases over time [32; 42].
The statistical model for concrete compressive strength presented by JCSS
[36] is comprehensive but complicated and not suitable or practical for this
application which is concerned primarily with crane loads.
The model that was used was that proposed by Mirza et al. [43]. This
model was also used in reliability analyses carried out by Ellingwood et al.
[32], Grant et al. [44] and MacGregor et al. [42]. Ruiz & Aguilar [49] use a
simpler model suggested by Ellingwood [45]. The model given by Mirza et al.
[43] was preferred because it resulted from a more in-depth study and it was
developed more recently than the model given by Ellingwood [45].
The model for the mean value of the concrete compressive strength is based
on three correction factors and is given by:
µfc = fc,nom rcreal rin−situ rR (6.3.1)
Where:
fc,nom – specified characteristic concrete strength = 30 MPa.
rcreal – corrects the characteristic value to the mean cube strength
rin−situ – corrects the cube strength to the in-situ concrete strength
rR – converts the strength to a ’static’ concrete strength considering failure

in one hour
The South African concrete materials code SABS 0100-2:1992 [50] specifies
that the characteristic strength of concrete is the 0.05 fractile, i.e. only 5%
of the cube tests should fall below the characteristic strength. For a normal
distribution, the value of rcreal is given by:
1
rcreal = (6.3.2)
1 − 1.645δ
Where:
δ – the coefficient of variation of the concrete strength, given as 15% below.
Mirza et al. [43] recommend values of the in-situ and rate of loading cor-
rection factors as:
rin−situ = 0.9 (6.3.3)
rR = 0.89 (1.173 + 0.08 log R) (6.3.4)

Where:
R – is the rate of loading in MPa/s
For failure in one hour (‘static’ concrete strength):
fc,nom 30 MPa 1
R= = = MPa/ s (6.3.5)
3600 3600 s 120
This gives:
µfc = 32.1 MPa
The coefficient of variation of the concrete compressive strength is given as

[43]:
q¡ ¢
δf c = δccyl 2 + 0.0084 (6.3.6)
Where:
δccyl – coefficient of variation of the test cylinders
Ellingwood et al. [32] give the coefficient of variation for average quality
control 5000 psi (≈ 35 MPa) concrete to be 12% which results in:
δfc = 15%
Mirza et al. [43] state that a Normal distribution describes the compressive
strength of concrete where there is good to average quality control whereas a
Lognormal distribution is better for poor quality control. Average quality
control was considered for this investigation, therefore a Normal distribution
was used.
6.3.4.2 Yield strength of reinforcing (fyr )
fyr,nom = 450 MPa µfyr = 510 MPa δfyr = 6% Distribution = N
The reinforcement was assumed to be drawn from many locations due

to the same reasons that the concrete was assumed to have average quality
control, i.e. this is a code calibration exercise and therefore describes the
variability expected for any project.
The ‘static’ yield strength was considered here so as to be consistent with
the concrete rate of loading. The yield strength is based on the nominal cross
sectional area thus the variation in bar diameter is taken into account in the
variation of yield strength.
A summary of the parameters found in the literature for the yield strength
of reinforcing is given in Table 6.2.
Table 6.2: Statistical parameters of the yield strength of reinforcing steel

Mean (MPa) c.o.v. % Distribution Reference
500 8.6 Beta Mirza & MacGregor [46]
cited by:
Grant et al. [44]
MacGregor et al. [42]
Lu et al. [51]
510 5.8 N JCSS [36]
(nom + 2σ) (σ = 30 MPa)
The model used was that given by JCSS [36] for the same reasons as those
given before.
6.4 Geometric properties

The variation in a dimension x, is often best described by considering the
variation of the error y, where:
xactual = xspec + y (6.4.1)
6.4.1 Steel members
The statistical properties of cross sectional dimensions and section properties

of steel members were modelled by Bjorhovde et al. [38] as Normal distributions
with mean values equal to the nominal values and coefficients of variation of
5%. This distribution was used in reliability work carried out by Ellingwood
et al. [32], Ellingwood & Reinhold [39] and Ravindra & Galambos [41].
JCSS [36] base their suggested statistical models on preliminary results
from a study in the Czech Republic on I profiles ranging in size from IPE 80-
200. A Normal distribution is recommended for all section properties. Cross
6.4. Geometric properties 137
sectional dimensions have the Normal parameters given below:

Mean:
−1.0 mm ≤ µy ≤ +1.0 mm
Standard deviation:
σy ≤ 1.0 mm
The mean values of the section properties such as area and section modulus
are given as equal to the nominal values and the recommended coefficients of
variation are given below:
Coefficient of variation for cross sectional area:
δA = 3.2%
Coefficient of variation for section modulus:
δZ = 4%
The statistical model that was chosen for the cross sectional dimensions
was a Normal distribution with the mean value equal to the nominal value.
Based on both suggested models; the coefficient of variation was taken to be
5% with an upper limit of the standard deviation of 1.0 mm.
The statistical model that was chosen for the section properties was also
a Normal distribution with the mean value equal to the nominal value. The
coefficients of variation for the area and section modulus were those given by
JCSS [36] and for all other properties the coefficient of variation was taken as
5%.
6.4.2 Welds
6.4.2.1 Weld throat thickness (a)
µa = anom δa = 15% Distribution = N
The statistical properties of the thickness of the weld were taken from
Fisher et al. [48]. The parameters recommended were not based on quantitative
studies due to a lack of data. With the parameters given, there is a 50% chance
that the shear area of the weld will be within 10% of the specified area, which
was considered to be a conservative assumption [48].
6.4.3 Concrete column
6.4.3.1 Column dimensions
µy = +1.59 mm σy = 6.35 mm Distribution = N
A study on the statistical parameters of rectangular columns with nominal

dimensions from 280 - 760 mm was carried out by Mirza & MacGregor [47].
A Normal distribution for the error on column cross sectional dimensions was
recommended, with the parameters given above.
Ellingwood et al. [32], MacGregor et al. [42] and Grant et al. [44] use the
statistical model given by Mirza & MacGregor [47] for the cross sectional
dimensions. Ellingwood et al. [32] further state that the standard deviations
are independent of the size of the member, therefore the models based on
columns with dimensions 280 - 760 mm can be applied to the concrete column
under consideration here with cross sectional dimensions 450 × 1500 mm.
JCSS [36] suggests a Normal distribution with the range of parameters
given below.
Mean:
0 ≤ µy ≤ 0.003xnom ≤ 3 mm
∴ µy,(h = 1500) = +3 mm µy,(b = 450) = +1.35 mm
Standard deviation:
σy = 4 + 0.006xnom ≤ 10 mm
∴ σy,(h = 1500) = 10 mm σy,(b = 450) = 6.7 mm
This model does not reflect what is stated by Ellingwood et al. [32], i.e.
that the standard deviation is independent of the size of the member and will
thus not be used for this investigation.
The statistical model that was chosen for the cross sectional dimensions of
the concrete column was that given by Mirza & MacGregor [47].
6.5. Loads 139
6.5 Loads
6.5.1 Permanent loads
The self weight of a structure is modelled by JCSS [36] as a function of the

density and the volume of an element:
D = γV
The variability of the permanent load is assumed to be due to the inherent

variability of the density and volume of a given element. Very small coeffi-
cients of variation and bias factors close to one are obtained for this model
(λD = 1.005, δD = 4%). The shortcoming in this model is that it consid-
ers aleatory uncertainty only and does not make any allowance for imperfect
modelling or prediction. Ellingwood et al. [32] state that most designers tend
to underestimate the permanent load of a structure and therefore recommend
a Normal distribution with the bias factor and coefficient of variation given
below. These values allow for both aleatory and epistemic uncertainties.
λD = 1.05 δD = 10%
A set of Normal distributions was used for the various different permanent
loads with the statistical parameters given in Table 6.3. The values of the
bias and coefficient of variation reflect the perceived degree of uncertainty
in, and difficulty of estimation of, the various permanent loads. Values of
the coefficients of variation are taken as less than 10% because the structures
dealt with in this investigation are steel structures where the permanent load
is easier to estimate than for concrete structures.
Table 6.3: Statistical parameters of the permanent loads

Permanent Load Bias c.o.v. %
Sheeting, purlins and utilities, Ds , Dp , Du 1.02 5
Own weight of the structure, Do 1.05 8
Crane bridge weight, Qbr and crab weight, Qcr 1.02 3
6.5.2 Roof imposed loads
6.5.2.1 Lifetime maximum roof imposed load
Rnom
µR = δR = 12.5% Distribution = EV I (Gumbel)
3.75
The statistical model for roof imposed loads was taken from de Villiers
[52]. Roof imposed loads can be classified into those due to maintenance and
those due to construction. The roof imposed load under consideration was
that due to maintenance as it was used in combination with the crane running
under normal operating conditions. The roof loads were further classified as
those imposed on large areas (for frame analysis) and those imposed on small
areas (for purlin analysis). As it was the frame that was under consideration
here, the statistical model used for the roof imposed load was that derived for
maintenance roof imposed loads on large areas.
6.5.2.2 Point-in-time roof imposed load
The point-in-time roof load is assumed to be equal to zero for the combination
with other time varying loads. Maintenance roof loading is not a common
occurrence and the chance of roof loading being present when another time
varying load is at its maximum is negligible.
6.5.3 Wind loads
6.5.3.1 Lifetime maximum wind load
2
µv2 = 0.41vnom δv2 = 52% Distribution = EV I (Gumbel)
The statistical parameters of the wind loads were taken from Ter Haar &
Retief [53]. The statistical model for the wind load was based on the square of
the wind speed because the load effects are linear with respect to this value.
6.5.3.2 Point-in-time wind load
2
µv2 = 0.05vnom δv2 = 108% Distribution = EV III (Weibull)
6.6. Modelling uncertainties 141
As for the lifetime maximum wind load, the parameters for the point-in-
time wind load were taken from Ter Haar & Retief [53].
6.6 Modelling uncertainties
6.6.1 Basis for modelling uncertainties
The response of a structure can be expressed as [36]:
Y = f (X1 , X2 , . . . Xn ) (6.6.1)
Where:
Y – structural response
Xi – basic variables
f – model function
Generally the model function does not completely and exactly describe the
physical process thus some errors are induced in the structural response Y.
The real outcome of an experiment can be expressed as:
Y 0 = θf (X1 , X2 , . . . Xn ) (6.6.2)
Where:
θ – modelling uncertainty
Modelling uncertainties arise because of:
• random effects neglected in the models
• simplification in mathematical relations
• neglecting 3D effects
• inhomogeneities in the material
• simplifications of connection behaviour

Ideally modelling factors should be obtained from many laboratory tests or

measurements on actual structures where the random variables are measured,
then the only uncertainties are the uncertainties inherent in the model. If the
number of measurements is small then there will be some statistical uncertainty
as well. Additional uncertainty could arise due to measurement errors in Y
and X. Ellingwood et al. [32] combine the uncertainties due to measurements
with the modelling uncertainty in the following way:
q
δT /C = δm 2 + δtest 2 + δspec 2 (6.6.3)
Where:
δT /C – variability in the ratios of test values to calculated values.
δm – variability of the model.
δtest – variability in the measurements during the test, i.e. measurement

errors in Y .
δspec – variability in the measurements of the test specimens, i.e. measurement

errors in X.
This gives the modelling uncertainty as:

q
δm = δT2 /C − δtest 2 − δspec 2 (6.6.4)
6.6.2 Distributions for modelling uncertainties
6.6.2.1 Introduction
The different modelling processes that are involved in calculating the loadings
and resistances have been identified and treated separately. This allows the
contribution of each individual modelling process to the overall reliability to be
investigated. The modelling processes were identified as the calculation of the
loads, the calculation of the load effects and the calculation of the resistances.
The methods followed for determining parameters for the modelling uncer-
tainties can be summarised as follows:
1. In the case when information is available, the parameters are based on

an assessment of the information.
2. In the case when no information is available two levels of parameter

selection can be considered:
a) Engineering reasoning and judgement can be employed for the selec-

tion of parameters. This could include consultations with experts.
b) Nominal provision can be made for modelling uncertainties by ap-
plying a modelling factor with no bias and an upper bound of the
coefficient of variation.
The modelling uncertainties that are critical for the reliability should be
identified and more emphasis placed on determining accurate parameters of
these uncertainties.
The three modelling processes are discussed below with their specific ap-
plications. The statistical parameters of the modelling uncertainties are given
for each application.
Lognormal distributions were used for the modelling uncertainties as rec-
ommended by JCSS [36] for modelling uncertainties which are applied by mul-
tiplication, as shown in the equation below from Chapter 5, rather than by
addition.
Qi = ci θi Ai (6.6.5)
Where:
Qi – load effect
θi – modelling uncertainty in the calculation of the load effect
Ai – applied load
The mean value of the modelling uncertainty will be referred to as the

bias, λ, because it indicates the bias with respect to the expected value of the
calculation model.
6.6.2.2 Crane loads
There was no information available on the model accuracy of the calculation

models for the crane wheel loads so the parameters of the modelling factors
were selected on the basis of reasoning.
6.6.2.2.1 Vertical wheel load
λ = 0.95 δ = 5%
The bias factor for the vertical wheel load was taken as less than one
because the dynamic factors model the maximum vertical force that will occur,
which will be greater than the mean values.
The vertical crane loads are calculated using simple vertical equilibrium and
as such the uncertainty in the calculation procedure is small. The assumption
is made, that all the crane wheels are in contact with the rails and each wheel
is treated as a simple support. It is possible for one wheel to lift from the rail
in the case where the girders are misaligned vertically or the girders on each
side of the runway are of different lengths causing unequal deflections. In the
case where one wheel lifts off the rail, the remaining wheels will carry a greater
load. This phenomenon was thought to be most critical for cranes with four
wheels because cranes with eight or sixteen wheels have spreader mechanisms
between the wheels to distribute the load more evenly.
An assessment was carried out of four crane configurations (outlined be-
low) to determine the percentage increase of the maximum wheel load from
removing the support at one wheel or lowering the wheel by a given amount.
1. Single I girder crane, 8 m span
2. Single box girder crane, 8 m span
3. Double box girder crane, 8 m span
4. Double box girder crane, 24 m span
The cranes were modelled in a finite element analysis with the crab at the
extreme of its travel carrying the hoistload. One of the wheels on the far side
of the crane, furthest away from the hoistload, was either removed or lowered
by 5 or 10 mm. A displacement of 10 mm was considered to be a reasonable
maximum because the tolerances of vertical girder alignment given by SAISC
[29] state that the maximum relative vertical misalignment of crane girders
is 10 mm. The percentage increase of the maximum wheel load and, where
applicable, the displacement of the wheel, were noted. The results are given
in Table 6.4 below.
Table 6.4: Investigation into vertical wheel load modelling uncertainty

Crane Wheel support removed Wheel lowered by
configuration 5 mm 10 mm
% increase displacement % increase % increase
1 N/A crane too flexible 0.02 0.03
2 25 41.28 mm 3 6
3 33 27.5 mm 6 12
4 20 47.0 mm 4.4
As can be seen from Table 6.4, the stiffer the crane, the greater the increase
of the maximum wheel load. The maximum percentage increase of 33% was
for the stiffest crane (double box girder, 8 m span). This was accompanied by
a 27.5 mm displacement of the wheel which was considered unreasonably large.
If the worst case is considered to be a displacement of 10 mm and the
crane configuration taken as a double box girder short span crane, then the
percentage increase of 12% can be considered the upper limit. A coefficient of
variation of 5% for the vertical load would mean that the worst case would lie
2.4σ away from the mean and would have a 1.09% chance of occurring which
was judged to be reasonable.
6.6.2.2.2 Horizontal wheel loads
λ = 1.0 δ = 15%
The load models for the horizontal loads are much more complex and less
transparent than the simple equilibrium used for the calculation of the vertical
loads. Therefore it was felt that there was greater uncertainty in the calculation
models for the horizontal loads than the vertical loads and this was reflected
in the larger coefficient of variation.
6.6.2.2.3 Horizontal longitudinal loads due to buffer forces
λ = 0.75 δ = 10%
A series of tests on end buffer loads was carried out by Kohlhaas [54]. The
tests were carried out on a crane with cellular rubber buffers.
Although the buffers considered for the end stop forces for the code calibra-
tion are hydraulic buffers and the tests were carried out with cellular rubber
buffers, this test data is the best available data and so will be used for the
development of the modelling uncertainty parameters.
Two end stop situations were considered, firstly the end stops being in
alignment so the buffers strike the end stops at approximately the same time
and secondly, the end stops 20 mm out of alignment in the longitudinal direc-
tion so the buffer on one end carriage will strike the end stop before the other
side.
Three hoistload positions were considered, firstly the crab in the centre
with the hoistload 150 mm from the ground, secondly the crab in the centre
with the hoistload 800 mm from the ground and thirdly the crab at one end of
the crane bridge with the hoistload 890 mm from the ground.
The force on both end stops, one on each line of rails, were measured. It
was found that the end stop forces increased with the height of the hoistload.
With the crab on one side of the crane bridge, as expected, the end stop forces
on one side of the runway were greater than those on the other side. The
same trend was observed when the end stops were offset in the longitudinal
direction.
The maximum end stop force, from the forces on each buffer, for each crane
run, were analysed to obtain the modelling factor.
The bias values obtained considering the various different hoistload posi-
tions are given in table 6.5.
Table 6.5: Bias values for end stop forces

Hoistload position bias value
crab in centre, hoistload height = 150 mm 0.59
crab in centre, hoistload height = 800 mm 0.61
crab off centre, hoistload height = 890 mm 0.70
As was mentioned before and as can be seen from Table 6.5, as the height
of the hoistload increases, the bias increases. The crab positioned on one side
of the crane bridge results in a larger bias than the crab positioned at the
centre of the crane bridge.
Because of the configuration of the experimental setup, the hoistload could
not be lifted to very close under the crane bridge. For this reason, the bias
values in the table were considered to be less than the maximum possible bias
values and a bias value of 0.75 was selected for the modelling uncertainty in
the calculation of the end stop forces.
The coefficient of variation considering all the end stop forces was 8.77%.
The coefficient of variation has been increased to 10% for this analysis, to
take into account additional uncertainty included for the fact that the buffers
considered for the code calibration are hydraulic buffers and the test data was
gathered for cellular rubber buffers.
6.6.2.3 Load effects
6.6.2.3.1 Additional moment due to slenderness in concrete column
λ = 1.0 δ = 5%
The method of calculating the additional moment in the concrete column

due to second order effects is given in the South African concrete design code
SABS 0100-1:1992 [10]. There was no data available on the accuracy of the
method so the parameters of the distribution were estimated.
6.6.2.3.2 Axial stress in flanges of crane girder
λ = 1.0 δ = 5%
For the 40t crane the normal stresses in the girder flanges due to vertical
and horizontal crane loads were calculated. The uncertainty in the stresses
is mostly due to the uncertainty in the section properties of the girder com-
bined with the elastomeric pad and the rail. The distribution parameters were
estimated because no information was available.
6.6.2.3.3 Axial forces from an analysis model
λ = 1.0 δ = 5%
These parameters were taken from JCSS [36].

6.6.2.3.4 Bending moments from an analysis model
λ = 1.0 δ = 10%
6.6.2.4 Resistances
6.6.2.4.1 Concrete column - biaxial bending and axial force
λ = 1.10 δ = 15%
A bias of 1.20 is suggested by Holický [55] for the resistance of concrete

columns. This value was reduced to 1.10 because a more accurate calculation
method than the code method was used. The coefficient of variation given by
Holický [55] was adopted.
6.6.2.4.2 Steel column - bending and axial force
λ = 1.10 δ = 15%
Table 6.6 gives a summary of the modelling uncertainty parameters in the

literature.
Table 6.6: Modelling uncertainty of resistance of steel columns

Failure mode Bias c.o.v. % Reference
In plane strength 1.107 16.4 Kotoguchi et al. [56]
Out of plane strength 1.18 14 Kotoguchi et al. [56]
In plane strength 0.98 8 Rojiani & Woeste [57]
Out of plane strength 0.88 4 Rojiani & Woeste [57]
In plane strength 1.07 10 Bjorhovde et al. [38]
Bending and compression 1.02 10 Bjorhovde et al. [38]
Bending and compression 1.20 10 Holický [55]
The bias ranges from 0.88 to 1.20 and the coefficient of variation from 4%
to 16.4%. The parameters given by Rojiani & Woeste [57] were neglected
because of their vast difference from the rest of the parameters. The value
of the mean selected as 1.10 was considered to be representative of the range

given. The value of 15% selected for the coefficient of variation is on the upper
side of the values given, in order to reflect the additional uncertainty due to
the wide range in the mean values.
6.6.2.4.3 Crane girder - elastic bending
λ = 1.0 δ = 5%
The parameters were taken from JCSS [36] for bending moment capacity
of steel members.
6.6.2.4.4 Truss members - axial tension
There is no modelling uncertainty in the resistance of a member to axial tension

because the tension model is the same as the tension test coupon so all the
variability is assumed to be in the material and the dimensions (Ellingwood
et al. [32]).
6.6.2.4.5 Truss members - axial compression
λ = 1.10 δ = 10%
The parameters of the modelling uncertainty for the truss members in axial
compression was based on the modelling uncertainty for steel columns. The
coefficient of variation was reduced from 15% to 10% because of the simpler
load case of only axial load, no bending moment.
6.6.2.4.6 Resistance of bolts
λ = 1.25 δ = 15%
6.6.2.4.7 Resistance of welds
λ = 1.15 δ = 15%
6.7 Fatigue
The method used for the fatigue analysis was the S-N curve approach. This
is one of the methods recommended by JCSS [36] for the consideration of
fatigue. Unless otherwise stated, the statistical models for fatigue are those
recommended by JCSS [36].
6.7.1 Fatigue resistance
6.7.1.1 Fatigue damage at failure (D)
Dnom = 1 µD = 1 δD = 30% Distribution = LN
6.7.1.2 S-N curve material property (A)
Anom
µA = δA = 58% Distribution = LN
∆2
The S-N curves that are given in SANS 10162-1:2005 [11] are given in terms
of mean minus two standard deviations on the fatigue life [58]. The method
for determining the mean value of the material parameter from a mean minus
two standard deviations S-N curve was given by Maddox [59] as:
A
a= (6.7.1)
∆d
Where:
A – the mean minus two standard deviations material factor

¡1¢
∆ – is related to the standard deviation on the fatigue life by: σ = log ∆
d – number of standard deviations of log(N ) away from the mean, in this

case d = 2.
Values of ∆ have been given by Maddox [59] for various fatigue details
corresponding to different welded joint configurations. The fatigue detail cat-
egories given by Maddox [59] were compared to the fatigue detail categories
6.7. Fatigue 151
for each crane girder fatigue situation as classified by the South African steel
design code and the relevant values of ∆ were selected as described below.
The bottom of the stiffener was classified as detail category C1 which

is described as ‘Base metal at the toe of transverse stiffener-to-flange
and transverse stiffener-to-web welds’ according to SANS 10162-1:2005
[11]. This corresponds to detail category E according to Maddox [59],
described as: ‘Welded attachments on the surface or edge of a stressed
member - attachment to girder webs in regions of combined bending and
shear’. The value of ∆ for this detail category is:
∆E = 0.561
of the top flange
The top flange to web weld was classified as fatigue detail B which is de-
scribed as ‘Built up members - Base metal and weld metal in components,
without attachments, connected by continuous full penetration groove
welds with backing bars removed’ according to SANS 10162-1:2005 [11].
This corresponds to detail category F2 according to Maddox [59], which
is described as ‘web-to-flange fillet weld in crane girder’. The value of ∆
for this detail category is:
∆F2 = 0.592
The corbel to column welded connection was classified as detail category

C with an additional reduction factor to allow for the fact that it is a
transversely loaded fillet weld according to SANS 10162-1:2005 [11]. This
detail is also classified as detail category F2 according to Maddox [59],
described as ‘fillet welded T joint between plates or built up members,
transverse to direction of stress’. The value of ∆ for this detail category
is given above.
6.7.1.3 Stress level for slope change in S-N curve (So )
µSo = So,nom δSo = 30% Distribution = N
The slopes of the S-N curve were taken as deterministic, the slope of the
first part of the curve m1 = 3, and the slope of the second part of the curve
m2 = 5. The nominal values of the material properties for both parts of the
curve are given in SANS 10162-1:2005 [11] and the nominal value of the slope
change in the S-N curve was calculated in the following manner:
µ ¶ 1
A2 m2 −m1
So,nom = (6.7.2)
A1
During the reliability analysis, the material property for the second part
of the S-N curve was calculated in the following manner:
A2 = A1 So(m2 −m1 ) (6.7.3)
6.7.2 Fatigue loading
6.7.2.1 Modelling factors for stress calculations (Bglob and Bloc )
µBglob = 1 δBglob = 10% Distribution = LN
µBloc = 1 δBloc = 25% Distribution = LN
Bglob and Bloc are the modelling factors for the uncertainties associated
with the global and local stress calculations respectively.
6.7.2.2 Fatigue stresses (S)
µS = Snom δS = 25% Distribution = LN
The nominal values of the fatigue stresses in the girder were calculated by
simulating the crane running over the gantry and obtaining a stress history for
the girder. The stress process obtained was broad-banded therefore a rainflow
counting method was used to determine the stress cycles as recommended by
JCSS [36]. There was no statistical data given for the stresses obtained by a
6.8. Summary 153
rainflow counting method but it was judged that there was some uncertainty
in the stresses, therefore a coefficient of variation of 25% was assigned to the
stress values. A Lognormal distribution was used in order to prevent the
stresses becoming negative.
Sensitivity studies on the effect of the statistical parameters of the fatigue
stresses on the fatigue reliability showed that they have a limited effect. There-
fore the parameters were not investigated further and were taken as those given
above.
6.7.2.3 Number of stress cycles (N )
µN = Nnom δN = 20% Distribution = N
There was no statistical model given by JCSS [36] for the number of stress
cycles but it was judged that this was an uncertain parameter so a coefficient
of variation of 20% and a Normal distribution were assumed.
As with the fatigue stresses, sensitivity studies showed that the statistical
parameters of the number of stress cycles have a limited effect on the fatigue
reliability therefore the parameters were taken as those given above.
The fact that the sensitivity studies on the fatigue stresses and number of
cycles showed that these variables have a limited effect on the fatigue reliability
could explain why no statistical models are provided in JCSS [36] for these
variables.
6.8 Summary
A summary of all the statistical models is given in Tables 6.7 - 6.11.
Table 6.7: Statistical models for material properties
Material property Nominal Mean c.o.v. % Distr.

Yield strength of structural steel 300 MPa 315 MPa 10 LN
Modulus of elasticity of steel 200 GPa Nom 3 N
Shear modulus of elasticity of steel 70 GPa Nom 3 N
Bolt ultimate tensile strength 800 MPa 1.2 Nom 7 N
Bolt shear/normal strength 0.60 0.625 5.3 N
Weld ultimate tensile strength 480 MPa 1.05 Nom 4 N
Weld shear/normal strength 0.67 0.84 10 N
Compressive strength of concrete 30 MPa 32 MPa 15 N
Yield strength of reinforcement 450 MPa 510 MPa 6 N
Table 6.8: Statistical models for geometric properties

Geometric property Mean c.o.v. % Distr.
Dimensions of steel members Nominal 5, σ < 1 mm N
Cross sectional area of steel members Nominal 3.2 N
Section moduli of steel members Nominal 4 N
Section properties of steel members Nominal 5 N
Weld throat thickness Nominal 15 N
Error in concrete column dimension +1.59 mm σ = 6.35 mm N
Table 6.9: Statistical models for loads

Load Bias c.o.v. % Distr.
Permanent load (sheeting, purlins, utilities) 1.02 5 N
Permanent load from structure self weight 1.05 8 N
Crane self weight (bridge and crab) 1.02 3 N
Lifetime maximum roof imposed load 1/3.75 12.5 EV I
Point-in-time roof imposed load 0
Lifetime maximum wind load 0.41 52 EV I
Point-in-time wind load 0.05 108 EV III
6.8. Summary 155
Table 6.10: Statistical models for modelling uncertainties

Model Bias c.o.v. % Distr.
Vertical crane wheel loads 0.95 5 LN
Horizontal crane wheel loads 1.00 15 LN
Longitudinal crane buffer forces 0.75 10 LN
Second order moment in concrete column 1.00 5 LN
Axial stress in crane girder flanges 1.00 5 LN
Axial forces from an analysis model 1.00 5 LN
Bending moments from an analysis model 1.00 10 LN
Concrete column resistance 1.10 15 LN
Steel column resistance 1.10 15 LN
Crane girder resistance to bending 1.00 5 LN
Roof truss member resistance to tension 1.00 0
Roof truss member resistance to compression 1.10 10 LN
Bolt resistance 1.25 15 LN
Weld resistance 1.15 15 LN
Table 6.11: Statistical models for fatigue analysis

Property Mean c.o.v. % Distr.
Miner’s sum at failure 1 30 LN
S-N curve material property Anom /∆2 58 LN
40t crane intermediate stiffener ∆ = 0.561
40t crane top flange to web weld ∆ = 0.592
5t crane corbel to column weld ∆ = 0.592
Slope change in S-N curve Nominal 30 N
Global stress calculation uncertainty 1 10 LN
Local stress calculation uncertainty 1 25 LN
Fatigue stresses in the girder Nominal 25 LN
Number of stress cycles Nominal 20 N
Chapter 7
Stochastic modelling of crane

hoistload
The new stochastic models that were developed for the code calibration were
the models that describe the hoistload lifted by the crane. The stochastic
models that are required for the hoistload are those that represent ‘arbitrary
point-in-time’ models as well as extreme value type models.
7.1 ‘One cycle’ distribution
A crane cycle is defined as the crane lifting a load, moving to the position where
it puts down the load, unloading and moving to the position where it will lift
the next load. The ‘one cycle’ distribution models the probability distribution
of the size of the hoistload for one lift of a load, i.e. one crane cycle. The ‘one
cycle’ distribution can be considered as the frequency distribution of all the
loads lifted during the crane lifetime.
Hoistload distributions for grab cranes lifting granular material have been
presented by Köppe [13] as a combination of Normal distributions with an
upper limit of 1.2 times the safe working load (SWL). Cranes were divided into
four classes depending on their intensity of work from light to very heavy and a
hoistload distribution was provided for each class of crane. The philosophy in
the development of the upper region of the distributions presented by Köppe
[13] was that if too large a load was lifted, some of the granular material could
be released until the load was just of a size where it could be lifted without
triggering the overload limit switch. This caused a concentration of loads in
157
158 Chapter 7. Stochastic modelling of crane hoistload
the region of the overload limit.

Most of the overhead travelling cranes in South Africa have hooks as their
load lifting mechanism (see Chapter 3) and the characteristics of loads lifted by
a hook differ from loads in the form of granular material lifted by a grab in that
they are of a set size and cannot be adjusted, as with a granular material, to
be just within the overload limit. Distributions of loads lifted by hook cranes
would therefore not have such a concentration of loads around the overload
limit. It was also unclear whether the value of 1.2 SWL for the upper limit
was reasonable for hook cranes in South Africa.
For these reasons, the hoistload distributions presented by Köppe [13] were
not considered to be suitable for the purposes of the calibration process and
new models were developed.
In the development of the new hoistload distributions, information was
obtained from crane manufacturers and operators about the upper limit and
the distribution of loads lifted. Cranes were again divided into four classes, as
given by Dunaiski et al. [26], described as:
1. Light – rarely lifts the SWL and normally only small proportions of the
SWL
2. Medium – fairly frequently lifts the SWL and normally moderate pro-
portions of the SWL
3. Heavy – frequently lifts the SWL and normally high proportions of the
SWL
4. Very heavy – regularly lifts the SWL
The class one and two cranes are generally workshop and maintenance
type cranes where the level of control over the usage of the crane is relatively
low. Class three cranes are generally process cranes and class four cranes are
typically ladle cranes in metal works where the level of control over the usage
of the crane is higher.
7.1.1 Type of distribution

In the modelling of the loads lifted by cranes, it was judged to be reasonable
to use a distribution with upper and lower limits. The hoistload lifted cannot
be lower than zero and as loads lifted by cranes have a mechanical origin the
7.1. ‘One cycle’ distribution 159
upper tail of the distribution cannot tend to infinity as with loads of natural
origin.
One hoistload distribution was required to model the distribution of hoist-
loads lifted by each of the four classes of crane. There were therefore four
different shaped distributions to be developed and the distribution type se-
lected for the hoistload distribution was required to be able to easily deal with
the different shapes.
The statistical distribution that best fulfilled these requirements was the
beta distribution because of the ease of generating the required shape and
because the beta distribution has lower and upper bounds.
7.1.2 Upper limit of distribution

All the crane manufacturers and operators stated that the overload limit
switches on cranes are set to 10% overload. Overloading of cranes does, how-
ever, occur and it is possible to turn off the overload limit switch and to pick
up a load greater than 1.1 SWL. This is more likely to occur with the class
one and two cranes than the class three and four cranes.
The load that is used as the test load for the commissioning of the crane
after installation is 1.25 SWL. The crane operators therefore know that the
crane is capable of lifting this load and would be confident to overload the
crane to this level but would probably be unwilling to lift loads larger than
1.25 SWL. For this reason, the upper limit of the distribution was chosen as
1.25 SWL.
7.1.3 Shape of distribution

Information was gathered from crane manufacturers and operators in the form
of expert opinion giving descriptions of the usage of the crane and histograms
of loads lifted. These were used, together with the descriptions of the classes
of the cranes, to develop the shapes of the ‘one cycle’ distributions. Appendix
C contains all the information gathered from the manufacturers and operators,
i.e. the histograms of loads lifted and descriptions of usage of the cranes.
7.1.3.1 Information gathered from crane operators
The histogram data obtained from the crane operators was based on produc-
tion records and in one case on measurements of loads lifted. All the histograms
are given in Appendix C. The histograms have been normalised with respect
to the SWL.
A class one crane was investigated at a boiler making factory as part of
a study carried out by Van der Walt [60]. Instrumentation was fitted to the
crane and the loads that were lifted over a four week period were recorded.
Four weeks has been determined by Pasternak et al. [14] to be the period over
which the lifting history of a crane is representative. The histogram for this
boiler making crane is shown in Figure C.3.
Three cranes were investigated at a car manufacturing plant, two class two
cranes on press lines and a class one maintenance crane. The main duty of
the press line cranes is to carry the press tools from the storage area to the
press and back again for the tool changes. The maintenance crane is used for
moving the tools around the maintenance area.
The histograms were developed by assessing the production information
combined with the weights of the tools. For one crane there were 38 different
car parts that were manufactured in the press line and for the other crane there
were 28. The production data was assessed to determine how many times each
car part was manufactured in a given year. It was assumed that the crane
lifted the tools required for the manufacture of each part twice for each time
the part was manufactured, once to take the tool to the press, and once to
take it back to storage. The weights of the tools for each part were known
therefore it was possible to determine how often the cranes lifted a particular
weight. The histograms for the two press line cranes are given in Figures C.4
& C.5.
The histogram for the maintenance crane was developed based on the as-
sumption that each tool requires an equal amount of maintenance. Each tool
consists of two parts, an upper part and a lower part. It was assumed that
the maintenance crane lifted the entire tool and each part of the tool, an equal
number of times. The weights of all the tools were known and these were used
to develop the histogram. The histogram for the maintenance crane is given
in Figure C.1.
Production data was obtained on a class three slab handling crane in a
steel works. The data was in the form of delivery records of slabs that had
been transferred. The histogram for this crane is shown in Figure C.11.
The other histograms given in Appendix C were generated based on de-
scriptions of the loads that the various cranes lift, obtained from the crane
operators. This is the type of information that would be given to the struc-
tural engineer for the design of the support structure. For process cranes with
a fixed operating routine, this type of information is more accurate than for
maintenance and general workshop cranes.
7.1.3.2 Distribution parameters
Histogram data was available for the class one to three cranes but not for the
class four, ladle cranes. The hoistload distributions for classes one to three
were therefore based on both histogram data and descriptions given by the
crane operators of the loads lifted, whereas the distribution for the class four
cranes was based on the range of ladle weights given by the operators.
The parameters of the beta distributions for the class one to three cranes
were determined by fitting the shape of the beta distribution to the shape of
the histograms of loads lifted and the description of the crane classes. The
parameters of the beta distributions for the class four crane were determined
by fitting a beta distribution to the information given by the crane operators
on the ladle weights for the class four cranes, combined with the description
of the crane class. Each class will be discussed in turn below.
Figures 7.1 - 7.3 show histograms of loads lifted and the beta distributions
for crane classes one to three. The distributions and histograms have been
normalised with respect to the SWL.
Class 1: The definition of class one cranes is that they rarely lift the SWL and
normally only small proportions of the SWL. The peak of the class one
distribution was therefore chosen as a small proportion of the SWL, taken
as 16% SWL, which corresponded well to the peak of the histogram. The
upper tail of the distribution was modelled so as to reflect the fact that
class one cranes will rarely lift the SWL. The probability given by the
chosen beta distribution of the class one crane lifting a load equal to or
greater than the SWL is 1.25%, which was judged to represent rarely
lifting such large loads.
An example of an histogram of loads lifted for a class one crane is shown
in Figure 7.1. Most loads that were lifted were in the region of 0.1 to 0.6
SWL which corresponds well with the beta distribution for this class.
Class 2: The definition of the class two cranes is that they fairly frequently
lift the SWL and normally moderate proportions of the SWL. The peak
of the class two distribution was taken as 56% SWL. The probability of
the class two cranes lifting the SWL is higher than the class one cranes
and this is modelled in the upper tail of the beta distribution.
An histogram of loads lifted by a class two crane is shown in Figure 7.2
along with the beta distribution for the class two cranes. The histogram
is symmetrical with the peak at 0.6 SWL which corresponds well to the
beta distribution.
Class 3: The definition of the class three cranes is that they frequently lift
the SWL and normally high proportions of the SWL. The peak of the
class three distribution, representing ‘high proportions of the SWL’ was
taken as 80% SWL.
An histogram of loads lifted by a class three crane is shown in Figure
7.3 together with the beta distribution. Most of the loads lifted are in
the region of 80% to 100% SWL which corresponds well with the beta
distribution.
Class 4: The definition of the class four cranes is that they regularly lift
the SWL. The information given by two crane operators about the loads
lifted by the class four, ladle cranes was that the ladle weights range from
98% SWL to 110% SWL with most ladle weights being equal to the safe
working load. The beta distribution was fitted to the descriptions by the
peak being at the SWL and setting the lower limit at 90% SWL. From
the beta distribution, the probability that a class four crane would lift
loads equal to or greater than the SWL is 58%.
The parameters of the beta distributions, for the ‘one cycle’ distributions,
are summarised in Table 7.1.
The ‘one cycle’ distributions for classes one to three are shown in Figure
7.4(a). The ‘one cycle’ distribution for class four cranes is given in Figure
7.4(b). The distributions have been normalized with respect to the SWL.
Figure 7.5 shows an enlargement of the upper tails of the ‘one cycle’ distri-
butions for all four classes of crane. The class one and two cranes are typically
more general use cranes, such as maintenance or workshop cranes, rather than
fixed process cranes as for the class three and four cranes. The class one and
two cranes therefore have less control over their usage and it is more likely
than for the class three or four crane for the overload limit switch to be turned
off and to lift loads higher than SWL. The upper tails model the greater level
of control over the usage of the class three and four cranes than the class one
and two cranes.
Table 7.1: Parameters of ‘one cycle’ beta distributions

Shape factors
lower bound upper bound q r
Class 1 0 1.25 1.3 3
Class 2 0 1.25 3.2 3.7
Class 3 0 1.25 8 4.9
Class 4 0.9 1.25 3.5 7
Histogram of hoistloads for 20t maintenance crane ’One cycle’ hoistload distribution, class 1 (light) , q = 1.3, r = 3
25 1.8
1.6
20
1.4
Probability density
1.2
Frequency %
15
1
0.8
10
0.6
0.4
5
0.2
0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL Fraction SWL
(a) Histogram of hoistloads (b) Beta distribution
Figure 7.1: Class 1 crane ‘one cycle’ distribution

Histogram of hoistloads for 25t press line crane ’One cycle’ hoistload distribution, class 2 (medium) , q = 3.2, r = 3.7
18 1.8
16 1.6
14 1.4
Probability density
12 1.2
Frequency %
10 1
8 0.8
6 0.6
4 0.4
2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Histogram of hoistloads for 80t slab handling crane ’One cycle’ hoistload distribution, class 3 (heavy) , q = 8, r = 4.9
20 2.5
18
16 2
14
Probability density
Frequency %
12 1.5
10
8 1
4 0.5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2

’One cycle’ hoistload distributions ’One cycle’ hoistload distribution, class 4 (very heavy) , q = 3.5, r = 7
2.5 8
Class 1
Class 2
Class 3 7
2
6
Probability density
Probability density
5
1.5
1
3
2
0.5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
(a) Classes 1 to 3 (b) Class 4
Figure 7.4: ‘One cycle’ distributions for all crane classes
’One cycle’ hoistload distributions

0.01
Class 1
0.009 Class 2
Class 3
Class 4
0.008
0.007
Probability density
0.006
0.005
0.004
0.003
0.002
0.001
0
1.18 1.19 1.2 1.21 1.22 1.23 1.24 1.25
Fraction SWL
Figure 7.5: Upper tails of ‘one cycle’ hoistload distributions

7.2 Extreme hoistload distributions

Two possible types of extreme value distributions were identified: either an
extreme value distribution for the wheel load or an extreme value distribution
for the hoistload.
The method of calculating the vertical wheel load for a four wheel crane is
given in Equation 7.2.1
· ¸
1 Qbr L−y
Qmax = + (Qcr + Qh ) (7.2.1)
2 2 L
Where:
Qmax – maximum vertical wheel load
Qbr – weight of the crane bridge
Qcr – weight of the crab
Qh – weight of the hoistload being lifted
y – distance of the crab from the wheel being considered
L – span of the crane bridge
The four parameters that can be considered as basic variables are the
weights of the crane bridge, crab and hoistload and the position of the crab
on the bridge. The same four parameters were identified in the calculation of
the horizontal wheel loads. The methods of developing the two possible types
of extreme value distributions are discussed below.
1. Extreme value distribution for the wheel load. This extreme distribu-
tion could be developed by simulating ‘one cycle’ hoistloads, self weight
values and crab positions to generate histories of crane wheel loads. A
distribution could be fitted to the extreme ‘observed’ wheel loads and the
wheel loads would be considered as random variables for the reliability
analysis.
2. Extreme value distribution for the hoistload. The reliability analysis

would be carried out considering the crane self weight and hoistload as
basic variables. The position of the crab would be considered determin-
istic at some point along its travel. The wheel loads would be calculated
during the reliability analysis process.
7.2. Extreme hoistload distributions 167
The second approach was adopted for this code calibration procedure. The
motivation for using this approach is outlined below.
1. The second approach enables the sensitivity of the reliability analysis to

the individual variables to be assessed. The first approach of extreme
value distributions for the wheel loads loses the information about the
relative importance of the separate variables.
2. Developing extreme value distributions for the wheel loads combines the
crane parameters into one basic variable. This would require new models
to be developed for each crane considered. Developing an extreme value
distribution for the hoistload only, results in more generic models which
can be used for any crane. The extreme value hoistload distributions
could be used for future code calibration or reliability assessment work
on crane loading.
The position of the crab was treated as a deterministic variable and was
taken conservatively as at the extreme of its travel closest to the wheel being
considered.
7.2.1 Basis for development

The basis for the development of the extreme hoistload distributions was that
structural reliability is always considered with reference to a given time period.
The extreme distributions model the probability distribution of the largest load
that will occur in the time period considered and are therefore dependent on
the given time period. When considering crane loads, the extreme hoistload
distributions are dependent on the number of crane cycles carried out during
the given time period.
The extreme hoistload distributions are referred to by the number of cycles
considered (Nextr ), e.g. a ’100 cycle’ hoistload distribution is the probability
model for the size of the largest load that will be lifted in 100 crane cycles, i.e.
Nextr = 100.
The value of Nextr depends on the limit state and load combination being
considered. For the ultimate limit state when crane loads are the leading
load, Nextr would be the total number of cycles a crane would perform during
its lifetime. When the crane load is the accompanying load for combination
with wind, Nextr would be the number of cycles that the crane would perform
during the extreme wind event. In the case of the accidental load situation of
the crane running into the end stops, the number of times during the crane
lifetime that this will occur is estimated and that number of cycles is used as
Nextr for determining the hoistload distribution for the accidental limit state.
In this way it is relatively simple to obtain appropriate hoistload distributions
for any limit state and for the combination of crane loads with other loads or
combinations of more than one crane.
7.2.2 Method of development

A simulation technique was used for the development of the extreme hoistload
distributions. A set of hoistload values was generated from the ‘one cycle’
distribution for a given number of cycles, Nextr , and the largest value from
that set was selected as the extreme value, Hextr . This process was repeated
En times to obtain a set of En number of extreme values.
For Nextr less than 1000, there was a greater spread in the extreme values
and a greater En was required to determine the extreme hoistload distribution
parameters than for Nextr greater than 1000. For Nextr less than 1000, the
number of extreme values generated was En = 3000 and for Nextr greater than
1000, the number of extreme values were generated was En = 200.
Histograms of extreme values generated for a class one crane for Nextr =
100, 1000 and 3000 are shown in Figures 7.6 - 7.8 where it was observed that
as Nextr increases, the mean value of the extreme hoistload increases and the
standard deviation decreases.
A linear relationship between the log of the cycles and the log of the stan-
dard deviations was obtained. Polynomials were fitted to the curves for the
trend of the mean values. The trend lines allow interpolation and extrapo-
lation of the mean values and standard deviations for a given value of Nextr .
Extrapolation is valid up to two million cycles.
The trends of the mean values and standard deviations are shown for the
four crane classes in Figures 7.9 & 7.10. The trend lines were plotted from ex-
treme values generated for Nextr ranging from 30 to 300 000 and extrapolation
was carried out to two million cycles. Equations for the trends of the mean
values and standard deviations are given in Appendix C.
The extreme hoistloads generated were plotted on probability paper so as
to investigate the best distribution type to use for the extreme hoistloads. The
coefficients of skewness of the sets of extreme hoistloads were all very close to
zero indicating that a Normal distribution could be considered. Figure 7.11

shows extreme hoistload values for Nextr = 3000 for a class one crane plotted
on Normal probability paper and it was observed that there is a good fit to a
Normal distribution.
The ‘one cycle’ distributions have an upper bound at 1.25 SWL which
means that the extreme hoistloads have the same upper bound. The final
distribution type that was chosen for the extreme hoistload distributions was
a truncated Normal distribution with an upper limit of 1.25 SWL.
Histogram of extreme hoistloads for Nextr = 100 for a class 1 crane

25
20
15
Frequency %
10
0
0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15 1.2
Hextr as a fraction of SWL
Figure 7.6: Histogram of extreme hoistloads for Nextr = 100
Histogram of extreme hoistloads for N = 1000 for a class 1 crane

extr
20
18
16
14
12
Frequency
10
0
0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15 1.2

Histogram of extreme hoistloads for N = 3000 for a class 1 crane

extr
18
16
14
12
Frequency %
10
0
0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15 1.2
Trend of mean values of extreme hoistloads

1.25
1.2
1.15
Mean of exteme hoistloads
1.1
1.05
0.95
0.9
0.85
0.8 Class 1
Class 2
0.75 Class 3
Class 4
0.7
0 1 2 3 4 5 6
10 10 10 10 10 10 10
N
extr
Figure 7.9: Trend of the mean values of the extreme hoistload distributions
0
Trend of standard deviations of extreme hoistloads
10
Class 1
Class 2
Standard deviation of exteme hoistloads
Class 3
Class 4
−1
10
−2
10
−3
10
0 1 2 3 4 5 6
10 10 10 10 10 10 10
N
extr
Figure 7.10: Trend of the standard deviations of the extreme hoistload distributions
Extreme hoistload values for N = 3000 for a class 1 crane

extr
3
theory
data
2
Standard normal variate
−1
−2
−3
1.14 1.16 1.18 1.2 1.22 1.24
Fraction SWL
Figure 7.11: Normal probability plot of extreme hoistload values

The cumulative distribution functions of the extreme hoistload distribu-

tions are shown in Figure 7.12 for all the crane classes for various values of
Nextr .
CDF of extreme hoistload distribution for a class 1 crane CDF of extreme hoistload distribution for a class 2 crane
1 1
2 2
Nextr = 10 Nextr = 10
0.9 Nextr = 103 0.9 Nextr = 103
0.8 Nextr = 105 0.8 Nextr = 105
0.7 0.7
0.6 0.6
CDF
CDF
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0.95 1 1.05 1.1 1.15 1.2 1.25 0.95 1 1.05 1.1 1.15 1.2 1.25
N N
extr extr
(a) Class 1 (b) Class 2

CDF of extreme hoistload distribution for a class 3 crane CDF of extreme hoistload distribution for a class 4 crane
1 1
0.9 Nextr = 103 0.9 Nextr = 103
0.8 Nextr = 105 0.8 Nextr = 105
0.7 0.7
0.6 0.6
CDF
CDF
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0.95 1 1.05 1.1 1.15 1.2 1.25 0.95 1 1.05 1.1 1.15 1.2 1.25
N N
extr extr
(c) Class 3 (d) Class 4
Figure 7.12: Cumulative distributions functions for extreme hoistload distributions
7.3 Summary
The stochastic models of the loads lifted by a crane were discussed. Both
arbitrary point-in-time models as well as extreme value type models were de-
veloped.
7.3. Summary 173
Cranes were divided into four classes defined as light, medium, heavy and
very heavy. The definitions of the classes were given in terms of the loads that
are lifted by the crane.
The first hoistload distribution developed was the ‘one cycle’ distribution
which is the probability distribution of the size of the hoistload for one load
lifted, i.e. one crane cycle. The ‘one cycle’ distribution can be considered to
be the frequency distribution of all the loads lifted by a crane over its lifetime.
Information was obtained from crane manufacturers and operators on the
loads that are lifted by cranes as well as the level of the overload limit switch.
The information was either in the form of production data which could be used
to obtain histograms of loads lifted, or descriptions of the loads that are lifted
by specific cranes.
The development of the ‘one cycle’ distributions for the four crane classes
was based on the information given by crane manufacturers and operators and
the descriptions of the crane classes. Beta distributions were used for the ‘one
cycle’ hoistload distributions.
The extreme value type distributions are the probability distributions of
the largest load lifted in a given number of crane cycles (Nextr ). The extreme
hoistload distributions were developed from the ‘one cycle’ distributions using
a simulation technique. Trends of the mean values and standard deviations
of the extreme hoistload distributions with respect to Nextr were observed.
These trends allow the parameters of the extreme hoistload distributions to be
determined given a value of Nextr . A truncated Normal distribution was shown
to be appropriate for the extreme hoistload distributions. The appropriate
extreme hoistload distribution can be obtained by using a suitable value of
Nextr for various limit states and load combinations.
Chapter 8
Code calibration method
The method of carrying out the code calibration was to perform a reliability
analysis, with assumed crane partial load factors, and then to vary the load
factors until the reliability met the specified target reliability.
Individual structural elements of the representative structures were con-
sidered for the code calibration. Different failure modes for the elements were
considered separately, resulting in a value of reliability for each element for
each failure mode considered. The probability of failure of the structure was
assumed to be equal to the maximum probability of failure of the individual
elements.
8.1 Reliability analysis method

The reliability analysis was carried out in two steps: firstly an economic design
of the structural element under consideration and secondly the reliability as-
sessment of the economically design member. These two steps will be discussed
in detail below.
8.1.1 Economic design
This step entailed designing the member under consideration to exactly comply
with the code requirements. Because this is a code calibration exercise, the
reliability of the code provisions are of interest and no conservatism in the
design of the member can be included.
Conservatism in the design of the member would lead to unconserva-
tive code calibration results because the member would be larger (and thus
175
176 Chapter 8. Code calibration method
stronger) than that specified by the code and its reliability would be greater
than the code reliability. The reliability of the member is assumed to represent
the reliability of the code and the partial load factors are calibrated so that
the reliability of the member is equal to the target reliability. If the calibration
was carried out on a member which included some conservatism, the calibrated
partial load factors would be lower than if the calibration was carried out on a
member which didn’t include any conservatism, thereby causing the reliability
of the code itself to be lower than the target reliability.
The loads imposed on the element for the economic design were the factored
design loads and load combinations as specified by the code. Crane loads were
taken from prEN 1991-3 [2] and the permanent, wind and roof imposed loads
were taken from the South African loading code SABS 0160:1989 [1]. Partial
load factors for the crane loads and combination factors for the crane loads
and their accompanying loads were assumed. During the calibration process,
these factors were varied until the target reliability was achieved.
The resistances of the members were determined by the methods given in
the South African steel and concrete design codes, SANS 10162-1:2005 [11]
and SABS 0100-1:1992 [10].
The method of carrying out the economic design was to take the member
section as it was built in the actual structure and to scale it down, keeping
the proportions the same, until its design resistance was exactly equal to the
applied design loads. This resulted in impractical member sizes but as this
was a theoretical exercise this was not of significance. Practical rounding of
the member sizes would have incorporated some conservatism in the member
design which is not acceptable as discussed above.
The size of the member obtained from the economic design was used as
the nominal size for the reliability assessment. The statistical properties of the
member dimensions in relation to the nominal sizes are discussed in Chapter
6.
8.1.2 Reliability assessment
The reliability analysis method was a first order reliability method (FORM)
using the Rackwitz-Fiessler procedure of equivalent normal distributions. The
FORM analysis was implemented in Matlab using the matrix method given
by Nowak & Collins [15].
8.2. Definition of code objective 177
Use of FORM is recommended by the Eurocode basis of design prEN 1990

[61] for a Level II type calibration where Level I is a semi-probabilistic method
and Level III is a full probabilistic calibration. The initial development of the
Eurocodes had an historical and empirical basis and the further development
has been based on Level II type calibration [61]. This is thus a recognised and
accepted reliability method for use in code calibration.
The reliability analysis was carried out on the economically designed mem-
ber by considering the actual loads that it was subject to and the real resistance
behaviour. The loads and resistances considered for the reliability analysis are
discussed in Chapter 5.
8.2 Definition of code objective

The code objective chosen for this code calibration process was the achievement
of a target reliability. The target reliabilities vary for the different limit states
considered.
8.2.1 Ultimate limit state

A table of fifty year target reliabilities for the ultimate limit state from JCSS
[36] is given in Table 8.1. These target reliabilities are based on a cost benefit
analysis taking into consideration the relative cost of safety measures and the
consequences of failure of the structure.
Table 8.1: 50 year target reliabilities

Relative cost of Consequences of failure
safety measure Small Moderate Large
Large 1.7 2.0 2.6
Moderate 2.6 3.2 3.5
Small 3.2 3.5 3.8
The fifty year target reliability used for calibration of the Eurocodes is
given in the Eurocode basis of design, prEN 1990 [61] as βT = 3.8 where it
is stated that the achieved reliabilities should be as close as possible to the
target reliability.
The approach taken in the calibration of the South African codes differs
from the Eurocodes in that the target reliability is seen as the minimum level of
reliability. The target reliability used in the calibration of the South African
codes is given by Kemp et al. [9] and Milford [62] as βT = 3.0 for ductile
failure modes. This target reliability was obtained from the implicit reliability
of existing acceptable practice at the introduction of limit states design codes
of practice in South Africa.
Due to the fact that this code calibration is for the determination of partial
load factors for the crane loads in the proposed SANS 10160, the target relia-
bility that has been selected is that used for calibration of the South African
codes of βT = 3.0.
8.2.2 Accidental limit state
The accidental load case considered was that of the crane running into the end
stops on the end of the runway. The end stop, or buffer, forces are classified
in prEN 1991-3 as accidental actions.
It is estimated that, in South Africa, the case of the crane running into the
end stops could occur 10 times during the lifetime of the crane. This is too high
a probability of occurrence for this load case to be treated as an accidental
load situation for which the probability of occurrence should be very small.
The load situation of the crane running into the end stops, though classified
as an accidental action in prEN 1991-3, will be treated, for the calibration, as
an ultimate limit state due to its relatively high probability of occurrence.
When selecting a target reliability for the accidental load situation, the
cost of the structural elements which resist the loads and the consequences
of failure were considered. The elements that resist the longitudinal buffer
forces are the end stops and the longitudinal crane bracing. The cost of these
structural elements is only a small proportion of the total cost of the crane
and the support structure. Failure of the end stops or longitudinal bracing
due to the crane running into the end stops could have large safety or financial
consequences due to damage to the support structure or crane and interruption
of the industrial process.
The table of target reliabilities from JCSS [36], which relates the relative
cost of safety measures to consequences of failure for a 25 year design life,
is given in Table 8.2. From Table 8.2, the recommended target reliability
for a situation where the relative cost of safety measures is small and the
consequences of failure are large is βT = 4.0. This is the value that will be
8.2. Definition of code objective 179
used for the calibration of the partial load factors for the buffer forces due to
the crane running into the end stops.
Table 8.2: 25 year target reliabilities

Relative cost of Consequences of failure
safety measure Small Moderate Large
Large 2.0 2.3 2.8
Moderate 2.8 3.4 3.6
Small 3.4 3.6 4.0
8.2.3 Fatigue
Guidelines are given in the Eurocode basis of design prEN 1990 [61] on the
target reliability for fatigue. The 50 year target reliability for office and indus-
trial type structures is given as ranging from βT = 1.5 to 3.8 depending on the
degree of inspectability, reparability and damage tolerance. These lower and
upper limits of fatigue reliability correspond to the 50 year target reliabilities
for serviceability limit state and ultimate limit state respectively. This implies
that if the structural element under consideration is easily inspectable, repara-
ble and has a high damage tolerance then fatigue is considered a serviceability
limit state, however in the opposite case fatigue is considered an ultimate limit
state.
Gulvanessian et al. [63] divide structures into damage-tolerant and damage-
intolerant with respect to fatigue and state that if a structure is damage-
tolerant (i.e. inspectable and reparable) then fatigue may be considered as a
serviceability limit state.
Suggestions are given in SANS 2394:2004 [64] on target reliability levels
for fatigue. The suggested values range from βT = 2.3 to 3.1 depending on
the possibility of inspection. The lower value is higher than the lower target
reliability value given by prEN 1990. A target reliability of βT = 1.5 is given for
irreversible serviceability limit states by SANS 2394:2004. The case of elements
subject to fatigue being inspectable and reparable cannot be considered an
irreversible limit state because with correct maintenance the fatigue damage
is reversible, however, it does have maintenance consequences similar to an
irreversible serviceability situation.
The elements subject to fatigue in crane support structures are primarily

the crane girders and columns, including corbel connections. These elements
are generally accessible for inspection and repair therefore in this investigation
fatigue will be considered as a serviceability limit state with a target reliability
of βT = 1.5.
8.3 Definition of code format

The code format is defined by Faber & Sørensen [8] as the number and type of
partial load factors, characteristic values of loads, load combination rules and
number and type of load combination factors.
Characteristic values of loads are defined in the prEN 1990 [61] as the most
representative value of the load and may be taken as a mean value, an upper or
lower value or a nominal value which is not statistically based. The character-
istic values of the weights of the crane and hoistload in the new proposed code
will be kept the same as those specified in SABS 0160:1989. The characteristic
values of the crane self weight (bridge and crab) are taken as the nominal val-
ues obtained from the crane manufacturer. The most representative value of
the hoistload is the nominal capacity of the crane, i.e. the safe working load.
Cranes in practice are always referred to by their capacity and so to use any
other value as the characteristic value would be confusing for the designer.
The partial load factors, load combination rules and load combination fac-
tors will be discussed separately for the different limit states.

The load combination rules and number and type of load combination factors
are specified by SABS 0160:1989.
The definition of the code format here will concentrate on the number and
type of crane partial load factors. Four code formats were considered and are
discussed below in increasing order of complexity.
1. One partial load factor applied to the calculated characteristic wheel

loads
This is the most basic format and constitutes current practice for crane
loads, being used in both SABS 0160:1989 and prEN 1991-3. The ad-
vantage of this format is its simplicity and familiarity for designers.
8.3. Definition of code format 181
2. Two partial load factors, one applied to the crane self weight and one
applied to the hoistload
The next level of complexity in the code formats was to factor the self
weight and hoistload separately before the calculation of the design wheel
loads. This format takes into account the difference in variabilities of the
self weight and hoistload and the higher bias of the hoistload.
3. As for format 1 but with an additional factor applied to the calculated

design horizontal loads
The wheel loads are calculated as for format 1 and the characteristic
vertical wheel load is factored by the crane partial load factor while
the characteristic horizontal wheel load is factored by the product of
the crane partial load factor and the additional partial factor for the
horizontal load.

design horizontal loads
The crane self weight and hoistload are factored separately as for format
2 and then the design wheel loads are calculated. The horizontal wheel
load is then factored further by the additional partial factor.
Formats 3 and 4 take into account the greater uncertainty in the calcu-
lation of the horizontal loads than the vertical loads.
With increasing complexity in the code format, greater consistency is ex-

pected in the level of reliability. For practical purposes, this greater consistency
should be weighed up against the increased complexity for the designer. One
measure that could be implemented to decrease the perceived complexity for
formats 3 and 4, would be to include the additional partial factor for the hori-
zontal loads as part of the wheel load calculation procedure instead of treating
it separately and explicitly as a partial load factor. The disadvantage in this
approach is that it decreases the rationality and transparency of the code.
The partial load factors for each code format are defined in Table 8.3.

The two code formats that were considered for the accidental limit state are
the same as code formats 1 and 2 for the ultimate limit state.
Table 8.3: Definitions of crane partial load factors
Symbol Definition
γC Crane partial load factor applied to wheel loads for formats 1 &
3
γCsw Crane partial load factor applied to crane self weight for formats
2&4
γCh Crane partial load factor applied to hoistload for formats 2 & 4
γH Additional partial load factor applied to horizontal wheel loads
for formats 3 & 4
8.3.3 Fatigue
Two fatigue assessment methods have been considered in this investigation, the
method recommended in prEN 1991-3 of calculating a fatigue damage equiva-
lent load and the method implied by SABS 0160:1989 where the responsibility
rests on the support structure designer to simulate the crane behaviour for the
calculation of the fatigue damage.
In the case where both alternatives were given in the code, it would be
preferable for the code format to be the same for both methods.
One way of achieving this would be to apply a fatigue partial load factor
to the characteristic crane wheel loads. This design wheel load would then be
used either for the calculation of the fatigue damage equivalent load, which
is directly proportional to the crane wheel load, or for the calculation of the
fatigue damage directly in the case when the crane behaviour is simulated. In
this way only one partial load format for fatigue need be specified in the code.
The disadvantage in this approach is that currently the fatigue assessment
is carried out with unfactored loads, implying a partial load factor for fatigue
of γF = 1.0. Changing this could meet with resistance from designers.
An alternative approach, in the case when the fatigue design is carried
out using the method given in prEN 1991-3, would be to include the partial
load factor in the λ factors given for the calculation of the fatigue damage
equivalent load. This would result in a less transparent code but would retain
the familiar code format.
8.4. Calibration method 183
8.4 Calibration method

The procedure that was followed for the code calibration was that given by
Ter Haar & Retief [53]. The procedure is carried out in two steps:
1. Calibration of the crane partial load factors

A single load case is considered for the calculation of partial load factors
to achieve the target reliability. For this calibration the crane only load
case was considered first for the determination of the crane partial load
factors.
2. Calibration of the combination factors

The partial load factors calculated for the single load case are used in
combinations of time dependent loads for the calculation of combination
factors to achieve the target reliability. The time dependent load combi-
nations considered here were crane with wind loads and crane with roof
imposed loads.
8.4.1.1 Calibration of crane partial load factors
The calibration of the crane partial load factors was carried out considering the
crane only load case. The elements that were considered were those elements
that are subject to predominantly crane loads, i.e. the crane girders and crane
columns.
Parametric studies were carried out to determine the effect of important
parameters on the reliability. The practical range of each parameter was de-
fined and the value that resulted in the lowest reliability was identified for
use in the calibration of the partial load factors and combination factors. The
parametric studies for the crane only calibration are outlined below.
1. Class of the crane

The four classes of crane as allowed for in prEN 1991-3 and as modelled
in Chapter 7 were considered.
2. The number of crane cycles considered for the extreme hoistload distri-
bution (Nextr ).
Nextr was considered in the range of 100 to 106 cycles. One million cycles
was considered a reasonable maximum of the number of cycles a crane is
likely to perform over its lifetime. This relates to the crane performing
8 cycles an hour, 16 hours a day, 6 days a week, 52 weeks a year for 25
years.
3. The ratio between the weight of the hoistload and the total weight of the
crane (hoistload and self weight).
In the selection of the representative cranes, the capacities covered the
full range of capacities, however, the spans of the representative cranes
covered only the most likely values of crane bridge spans with the shortest
span being that of the 5t crane at 19.2 m and the longest span being that
of the 260t crane at 28.5 m.
One of the crane properties that is affected by the span of the crane is
the ratio of hoistload to total crane weight. The longer the span, the
stronger, and hence heavier, the crane bridge needs to be to support the
hoistload.
A database of cranes that includes small standard cranes and large
process cranes obtained from various crane manufacturers and opera-
tors showed the ratio of hoistload to total crane weight to range from 0.3
to 0.85. The list of cranes together with the ratios of hoistload to total
crane weight are given in Appendix D.
An inspection of the ratios and the crane data shows that as expected, the
cranes with higher ratios of hoistload to total crane weight have relatively
short spans compared with cranes having a lower ratio of hoistload to
total crane weight. The cranes with higher ratios of hoistload to total
crane weight are all standard cranes which would probably be class one
or two cranes.
4. The load effects caused by the horizontal and vertical crane wheel loads.
For the investigation of the relative effects of the horizontal and vertical
crane wheel loads, different elements were considered that were normally
subject to various ratios of horizontal and vertical wheel loads. An el-
ement subject only to vertical crane loads was the column for the 5t
crane for the load case of crane load combination 1 with maximum verti-
cal loads. An element subject only to horizontal loads was the auxiliary
girder for the 260t crane. This parametric study was of interest for the
calibration of code formats 3 and 4 only.
All the elements mentioned above for all the crane only load cases were
considered for the parametric studies. Due to the fact that the constraint for
the calibration process was that the reliability should always be greater than
the target reliability, the scenarios that resulted in the lowest reliability were
identified as the critical cases. Crane partial load factors were determined
that resulted in the reliability of the critical cases being equal to the target
reliability.
8.4.1.2 Calibration of combination factors
The load combination method that was used was Turkstra’s rule [65]. The
probability distribution of the maximum value of the leading load was con-
sidered together with the probability distribution of the maximum value of
the accompanying load for the time period when the leading load is active.
For example in the case of wind as the leading load and crane loads as the
accompanying load, the number of cycles taken for Nextr for the hoistload dis-
tribution is the number of cycles that the crane is likely to perform during an
extreme wind event, e.g. a storm.
The two load combinations that were considered were crane with wind
loads and crane with roof imposed loads. The elements that were considered
for these load combinations were the frame elements (columns and roof trusses)
which were subject to crane, wind and roof imposed loads. Permanent loads
were included in the load combinations.
The effect of different resistance modes, and hence different bias and un-
certainty values, was investigated by carrying out reliability analyses on the
various elements with the same loading conditions. The element with the resis-
tance mode which resulted in the lowest reliability was identified as the critical
element and together with the load case which resulted in the lowest reliability,
was considered the critical case for the code calibration process.
The values of the parameters for the crane only load case that resulted in
the lowest reliability were used for the calibration of combination factors. The
additional parametric studies that were carried out for each of the combinations
are discussed below.
1. The ratio between the load effect caused by permanent load and the total
load effect (e.g. permanent, crane and wind).
This parametric study was carried out to assess the affect of the perma-
nent load on the reliability. Initially a study was carried out considering
only crane and permanent load. From the results of this study, the ratio
of permanent load to total load for the combinations was selected.
2. The ratio between the load effect caused by crane load and the load effect
caused by crane and wind loads.
3. The ratio between the load effect caused by crane load and the load effect
caused by crane and roof imposed loads.
8.4.1.3 Verification of partial load factors
The code calibration was carried out on specific selected critical elements and
load cases. This resulted in partial load factors and combination factors for
which the reliability of the critical element was equal to the target reliability.
In order to verify these partial load factors, reliability analyses were carried
out on all the selected elements from the representative structures for all the
load cases.
The verification of the partial load factors had two main aims. Firstly to
ensure that the correct elements and load cases had been identified as critical,
i.e. that the reliability of all the elements was above the target reliability.
Secondly to investigate the degree of conservatism induced by calibration to
the critical cases and to select the optimal code format.
The degree of conservatism was assessed by carrying out a cost analysis.
This was carried out for the crane only load case for the crane girders and
columns. Both the columns and the girders for the 5t and 260t cranes were
steel so the costs of the elements were expressed in terms of their volume. For
the 40t crane where the columns are concrete and the girders are steel, the
actual monetary costs of the elements were calculated based on the current
rates for steel, concrete and reinforcing given below.
The method followed in calculating the cost for each element was to firstly
design the element for each load case and to determine which load case resulted
in the largest element (i.e. to find the critical load condition). The reliability
of the element subject to the critical load case was assessed. The cost was
expressed in relative reliability terms and the resulting cost has been termed
the ‘excess cost’. The excess cost was calculated by multiplying the volume
(or monetary cost) by the difference between the reliability and the target
reliability i.e. Volume × (β − βT ) or Cost (R) × (β − βT ).
The costs were calculated for three ratios of hoistload to total crane weight,
r = 0.3, r = 0.5 and r = 0.85. The three cost values were summed to get the
total cost for the element.
Rates for steel, concrete and reinforcing
Concrete R750/ m3 based on 30 MPa concrete with 19 mm stone
Reinforcing R7/kg
Structural steel R18 500/ton = R145 000/ m3

The accidental load situation that was considered was the crane running into
the end stops on the end of the runway. The loads resulting from this load
case are horizontal longitudinal forces which are resisted by the longitudinal
runway bracing.
8.4.2.1 Parametric studies
The parametric studies that were carried out for the accidental limit state
were:
1. The number of cycles considered for the extreme hoistload distribution

(Nextr )
The extreme hoistload distribution models the probability distribution
of the largest load that will be lifted in a given number of cycles. The
number of cycles selected for Nextr was the number of times that the crane
is estimated to run into the end stops during its lifetime. The extreme
hoistload distribution then models the largest load that the crane is likely
to be lifting when it runs into the end stops.
2. Class of the crane
A parametric study was carried out on the class of the crane, to deter-
mine which class results in the lowest reliability for the number of cycles
selected for Nextr for the accidental load situation.
3. The ratio of the weight of the hoistload to the total crane weight
The motivation for this parametric study is the same as for the ultimate
limit state.
The longitudinal bracing elements resist the end stop forces and are thus
the elements considered for the accidental limit state.
8.4.3 Fatigue
8.4.3.1 Parametric studies
Two parametric studies were carried out for the fatigue reliability analysis.
1. The assessment of the effect of using a single slope versus a double slope
S-N curve.
Reliability analyses were carried out considering both a single slope and
a double slope S-N curve and the differences between the results were
assessed.
2. The load spectrum class.
Load spectrum classes Q2 to Q5 were considered. A comparison between

the different classes was made to determine which class resulted in the
lowest reliability.
8.4.3.2 Calibration of fatigue partial load factors
The fatigue provisions in prEN 1991-3 were assessed by calculating the bounds
of reliability for a given load spectrum class Q. With reference to Table 2.18 on
page 51, the combination of the upper bound for both the load spectrum and
the number of cycles gives the lower reliability bound whereas the combination
of the lower bound for both the load spectrum and the number of cycles gives
the upper reliability bound. The combination of the upper bound for the load
spectrum and lower bound for the number of cycles, or vice versa, results in
the intermediate values of reliability.
In this investigation fatigue is considered as a serviceability limit state.
Calibration of serviceability limit states is carried out with the average values
of reliability rather than the minimum values as for ultimate limit states.
This is because serviceability is less of a safety issue and more an economic
consideration resulting in a softer performance target for serviceability, where
the requirements are satisfied with sufficient performance on average.
The intermediate values of reliability can be considered as the average
values as will be explained with reference to Figure 8.1 which depicts one
block from the table of fatigue classes, Table 2.18 on page 51.
kQ
U A
N Si
A L
Figure 8.1: Combinations of load spectrum, kQ, and number of cycles, N

pdf of reliability
L A U Reliability
Figure 8.2: Distribution of reliability
If the load spectrum, kQ, and the number of cycles, N, are increasing in
the directions shown then the point marked ‘U’ in the top left hand corner
represents the upper bound of reliability with kQ and N at a minimum. Simi-
larly the point marked ‘L’ in the lower right hand corner represents the lower
bound of reliability. The two points marked ‘A’ in the upper right hand corner
and the lower left hand corner represent the combinations of the minimum of
kQ and maximum of N and vice versa. These two points give the intermediate
levels of reliability which are the same, or very close, for both points. The
broken line joining these two points consists of combinations of kQ and N that
also result in this intermediate level of reliability. If the distributions of kQ
and N are considered to be uniform then the most likely combinations fall
along the dotted line joining the two ‘A’ points as shown in Figure 8.2. These
are therefore the most likely (or average) combinations and will be used for
the calibration process. All combinations of kQ and N which are between the
diagonal line and the point marked ‘U’ will have reliabilities between the inter-
mediate line and the upper bound and all combinations between the diagonal
line and the point marked ‘L’ will have reliabilities between the intermediate
line and the lower bound.
The method of calibration was to apply a partial load factor to the max-
imum wheel load before calculating the fatigue damage equivalent load and
to assess the reliability of the resulting element. The partial load factor was
varied until the average value of reliability was equal to the target reliability.
The calibration of the fatigue partial load factor was also carried out for
the second method of fatigue design of simulating the behaviour of the crane.
As discussed in Chapter 5, the economic design of the element was carried out
according to the prEN 1991-3 fatigue design method and the damage resulting
from the simulated crane behaviour was calculated. Using this damage and the
calibration results from the prEN 1991-3 fatigue method, the fatigue partial
load factor was determined for the fatigue analysis method of simulating the
crane behaviour.
8.5 Summary
The two step method of carrying out the reliability analysis was discussed. The
first step is the economic design of the member under consideration to ensure
that the member is designed to exactly complies with the code requirements
with no conservatism included. The second step is the reliability analysis of
this economically designed member.
The code objective was defined as meeting a given target reliability. The
recommended target reliabilities were considered.
8.5. Summary 191
For the ultimate limit state the target reliability was chosen as βT = 3.0
as is the norm for the calibration of the South African codes. The accidental
limit state was treated as an ultimate limit state because of the relatively high
probability of occurrence of the accidental load case, given in prEN 1991-3, of
the crane running into the end stops. The target reliability was based on the
fact that the cost of the elements which resist the accidental loads is small in
comparison to the consequences of failure. A target reliability of βT = 4.0 was
selected. For these two limit states the constraint was that the reliability must
be greater than or equal to the target reliability.
Fatigue was considered as a serviceability limit state with a target reliability
of βT = 1.5, because crane girders and columns are generally inspectable and
reparable. As for serviceability limit states, which are more of an economic
consideration than a safety issue, the constraint was that the average reliability
for the fatigue should be equal to the target reliability.
The code format was discussed, the characteristic values of the loads and
load combination rules have been taken as in SABS 0160:1989. The number
and type of partial load factors and combination factors to be considered for
the calibration have been defined.
Four code formats were considered for the ultimate limit state considering
applying one partial load factor to the calculated characteristic wheel load or
two separate factors to the hoistload and self weight before the design wheel
loads are calculated. An additional factor applied to the horizontal wheel loads
was also included.
The code formats for the accidental limit state consisted of either applying
one partial load factor to the calculated characteristic longitudinal load or
applying separate partial load factors to the crane self weight and hoistload
before calculating the design longitudinal load.
The code format for fatigue consisted of applying a partial load factor
to the characteristic crane wheel loads before the calculation of the fatigue
damage equivalent load or the simulation of the crane behaviour.
The parametric studies that were carried out for each limit state and load
combination were discussed. The parametric studies were carried out to deter-
mine the critical values of various parameters to be used in the code calibration.
The parameters that will be assessed for the ultimate limit state for the
calibration of the crane partial load factors, are the class of the crane, the
number of cycles considered for Nextr for the determination of the extreme
hoistload distribution, the ratio of hoistload to total crane weight to take into
account shorter or long crane bridge spans and varying ratios of vertical and
horizontal wheel loads.
The calibration of the combination factors for the ultimate limit state com-
binations of permanent and crane with wind or roof imposed loads considered
the following parameters for investigation: the ratio of permanent load to total
load and the ratio of crane load to crane and wind or crane and roof imposed
loads, in order to cover the range of possible loading situations.
of the crane running into the end stops were the class of crane, the number of
cycles considered for Nextr and the ratio of hoistload to total crane weight.
The parameters that were investigated for the reliability assessment of
fatigue were the use of either a single or double slope S-N curve and the effect
of different load spectrum classes as given by prEN 1991-3.
The two step procedure for the calibration of the partial load factors and
combination factors for the ultimate limit state was discussed. The first step
consists of determining the crane partial load factors by considering the crane
only load case. In the second step the combination factors are determined
using the calibrated crane partial load factors.
Chapter 9
Code calibration results
Results are presented here of the parametric studies and code calibration of
the ultimate limit state, accidental limit state and fatigue.
9.1 Ultimate limit state

The calibration of partial load factors and combination factors for the ultimate
limit state was carried out in two stages as discussed in Chapter 8. The first
stage was the calibration of the crane partial load factors considering the crane
only load case. The second stage was the calibration of the combination factors
for crane loads with wind or roof imposed loads using the crane partial load
factors determined in stage one.
9.1.1 Crane load only
The elements that were considered for the crane only load case were the crane
girders and crane columns. The results of the parametric studies that were
outlined in Chapter 8 are given, followed by the calibration results.
9.1.1.1 Results of parametric studies
The first set of parametric studies that was carried out investigated the effect of
the class of the crane and the number of cycles taken for the extreme hoistload
distribution (Nextr ). The variation of reliability with Nextr was plotted for the
four different classes of crane. Values of the partial load factor were assumed
for this parametric study. Investigations were carried out for all the crane
193
194 Chapter 9. Code calibration results
girders and columns and the results were found to be consistent for all the
cases. Figure 9.1 shows a typical graph with results from this investigation for
the 5t crane column considering crane load combination 5 which includes both
vertical and horizontal crane wheel loads. The results of this investigation are
not influenced by the code format, the value of the crane partial load factor or
the ratio of the hoistload to total crane weight. Code format 1 was chosen for
simplicity, the value of the crane partial load factor was taken as the value in
SABS 0160:1989 of γC = 1.6 and the ratio of hoistload to total crane weight
was taken as r = 0.85.
Crane column, LC5, maximum horizontal and vertical forces, γC = 1.6, r = 0.85
2.9
Class 1
Class 2
Class 3
2.8 Class 4
2.7
Reliability index, β
2.6
2.5
2.4
2.3
2 3 4 5 6
10 10 10 10 10
Nextr
Figure 9.1: Parametric study into class of crane and Nextr
It can be seen from Figure 9.1 that the reliability decreases as Nextr in-
creases as expected from the trends of the extreme hoistload distributions.
The lowest reliability is achieved with a class one crane and Nextr equal to one
million cycles. These are the values of the parameters that will be used for the
code calibration process.
The second parametric study carried out was to investigate the effect of
the ratio of hoistload to total crane weight, i.e. the crane self weight and the
hoistload. The code format does have an effect in this investigation with a
difference in results observed between the code formats that factor the calcu-
lated characteristic wheel loads (1 & 3) and the code formats that factor the
hoistload and self weight separately (2 & 4). The inclusion of an additional
partial load factor for the horizontal loads, as in code formats 3 & 4 does not
have a significant effect on the results of this parametric study. For the sake
of simplicity, the results are presented here for code formats 1 and 2 only.
The reliability curves for the horizontal loads from crane load combina-
tion 1 (LC1) differed from the curves for the vertical loads from the same load
combination and the horizontal and vertical loads from crane load combina-
tion 5 (LC5). This is shown in Figures 9.2 & 9.3. Figure 9.2 shows the results
for the 260t crane auxiliary girder subject to only horizontal wheel loads from
LC1. Figure 9.3 shows the results for the 260t crane main girder for LC5.
Figure 9.3 is typical for elements subject to vertical loads from either crane
load combination or horizontal loads from LC5. Values of the crane partial
factors were assumed for these investigations and are shown in the figures.
The difference between the shapes of the graphs for the auxiliary girder
and the main girder is due to the different relative sensitivities of the reliability
analysis to the self weight components of the crane and the hoistload. Figure
9.4 shows the variation of the sensitivity factors with the ratio of hoistload to
total crane weight for the 260t auxiliary girder and main girder with the same
values of the parameters as in the previous figures.
The shape of Figure 9.3 is typical of most of the elements and load combi-
nations. The elements that did not display this behaviour are those dominated
by horizontal loads from LC1 and are listed below:
1. 5t crane column with LC1 maximum horizontal loads
2. 40t crane girder with LC1 maximum horizontal loads, stress in the top
flange
3. 40t crane girder with LC1 maximum vertical loads, stress in the top
flange
4. 40t crane concrete column with LC1
5. 260t crane auxiliary girder with LC1
None of these elements were the critical elements for the calibration so the
parametric study focussed mainly on the more typical cases as shown in Figure
9.3.
6
260t auxiliary girder, LC1, class = 1, N = 10
extr
3.2
γ = 1.4
C
γ = 1.5
3.1 C
γC = 1.6
3
2.9
2.8
2.7
2.6
2.5
2.4
2.3
2.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(a) Code format 1

6
260t auxiliary girder, LC1, class = 1, N = 10 , γ = 1.2
extr Csw
1.8
γCh = 1.35
γCh = 1.45
γCh = 1.55
1.75
1.7
1.65
1.6
1.55
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(b) Code format 2
Figure 9.2: 260t crane auxiliary girder, LC1

6
260t crane girder, LC5, class = 1, N = 10
extr
5.5
γ = 1.4
C
γ = 1.5
C
γC = 1.6
5
4.5
3.5
2.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(a) Code format 1

260t crane girder, LC5, γ = 1.2
Csw
3.6
γCh = 1.35
γCh = 1.45
3.4 γCh = 1.55
3.2
2.8
2.6
2.4
2.2
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(b) Code format 2
Figure 9.3: 260t crane main girder, LC5

6
260t auxiliary girder, sensitivity factors, class = 1, N = 10
extr
0.14
Q
br
Qcr
0.12 Q
h
0.1
Sensitivity factor, α
0.08
0.06
0.04
0.02
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(a) 260t crane auxiliary girder, LC1

6
260t crane girder, sensitivity factors, class = 1, N = 10
extr
0.14
Q
br
Qcr
0.12 Q
h
0.1
Sensitivity factor, α
0.08
0.06
0.04
0.02
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(b) 260t crane main girder, LC5
Figure 9.4: Sensitivity factors for LC1 & LC5

Considering code format 1 and referring to Figure 9.3(a) it was observed

that the reliability index decreases as the ratio of hoistload to total crane
weight increases. The practical range of ratios is r = 0.3 to 0.85. The lowest
level of reliability is obtained at a ratio of r = 0.85 for a class one crane.
The table of crane descriptions with the ratios of hoistload to total crane
weight (given in Appendix D) was assessed to determine whether there are
class one cranes in the region of higher ratios of hoistload to total crane weight
in order to ensure that the combination of class one crane and r = 0.85 is not
unnecessarily conservative. It was observed that the cranes in the region of
higher ratios of hoistload to total crane weight were all small standard cranes
which are likely to be class one or two cranes therefore the combination of a
class one crane with r = 0.85 is likely in practice and is thus not over conser-
vative. The value of r = 0.85 is therefore the critical value of this parameter
which will be considered for the calibration.
Considering code format 2 and referring to Figure 9.3(b) it was observed
that depending on the values of the partial load factors applied to the self
weight and hoistload, the reliability either increases or decreases as the ratio
of hoistload to total crane weight increases. It can also be seen that it is
possible to choose values of the partial load factors so that the reliability
remains constant over the range of ratios. This is an advantage of code format
2 in that it is possible to fulfil the code calibration objective of attaining a
consistent reliability over a range of parameters.
The next parametric study carried out was to investigate the relative ef-
fects of vertical and horizontal crane wheel loads. The element chosen for this
investigation was the 5t crane column because of the load situations it is sub-
ject to. Two loading situations were considered, LC1 maximum vertical loads
(no horizontal loads) and LC5 vertical and horizontal loads. Results of this
investigation are shown in Figures 9.5 & 9.6 for code formats 1 and 2 respec-
tively. The value of the reliability index at r = 0.85 is shown on the graphs in
Figure 9.5.
It can be seen clearly from Figures 9.5 & 9.6 that the inclusion of horizontal
loads leads to a decrease in reliability. For code format 1 from Figure 9.5 the
critical level of reliability, with γC = 1.5, for the vertical only load case is
β = 2.68 compared to the vertical and horizontal load case where β = 2.03.
The same is seen for code format 2 in Figure 9.6 where the vertical load only
case results in β = 2.35 compared to the horizontal load case where β = 1.78.
5t crane column, LC1, max vertical, class = 1, Nextr = 106

4
γC = 1.4
γ = 1.5
C
γ = 1.6
C
3.5
3.04
3
2.68
2.5
2.30
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Qh/(Qbr + Qcr + Qh)
(a) Vertical loads only

5t crane column, LC5, max horz & vert, class = 1, Nextr = 106
4
γC = 1.4
γ = 1.5
C
γ = 1.6
C
3.5
2.5
2.32
2.04
2
1.75
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Qh/(Qbr + Qcr + Qh)
(b) Vertical and horizontal loads
Figure 9.5: Influence of vertical and horizontal loads, code format 1

5t crane column, LC1, max vert, class = 1, Nextr = 106, γCsw = 1.2
2.8
γCh = 1.35
γ = 1.45
Ch
2.6 γ = 1.55
Ch
2.4
2.2
1.8
1.6
1.4
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(a) Vertical loads only

5t crane column, LC5, max horz & vert, class = 1, Nextr = 106, γCsw = 1.2
2.8
γCh = 1.35
γ = 1.45
Ch
2.6 γ = 1.55
Ch
2.4
2.2
1.8
1.6
1.4
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(b) Vertical and horizontal loads
Figure 9.6: Influence of vertical and horizontal loads, code format 2

This reduction in reliability with the inclusion of horizontal wheel loads is

expected because of the higher modelling uncertainty for the calculation of the
horizontal loads than the vertical loads. Code formats 3 and 4 (as discussed
in Chapter 8) have been included in this investigation because of this decrease
in reliability with increasing horizontal loads.
9.1.1.2 Reliability of current practice
An assessment of the reliability of current practice was carried out on the

crane loading provisions in SABS 0160:1989 [1] and the Eurocode crane loading
provisions in prEN 1991-3 [2].
The assessment of the reliability of current practice was carried out with
the 40t crane which can be considered to be the ‘most representative’ crane in
South Africa as discussed in Chapter 3. The element that was considered was
the crane girder subject to its own weight and crane loads.
The economic design of the girder was carried out for each code considering
the critical crane load combinations from that code, the critical load combi-
nations are given in Chapter 4. The partial load and resistance factors were
taken as given in each code and are shown in Table 9.1.
Table 9.1: Current partial safety factors in SABS 0160:1989 and prEN 1991-3
Partial load or resistance factor SABS 0160:1989 prEN 1991-3
Partial load factor for permanent loads 1.2 1.35
Partial load factor for crane loads 1.6 1.35
Partial resistance factor for steel 0.9 1/1.1
For the reliability analysis, the actual crane loads imposed on the girder
were assumed to be represented by the prEN 1991-3 crane load models with
the inclusion of the modelling uncertainties discussed in Chapter 6, as is the
case for all the reliability analyses.
The assessment was carried out for SABS 0160:1989 crane classes 2 - 4
and prEN 1991-3 hoist classes 1 - 4. SABS 0160:1989 class 1 cranes have been
omitted as they are hand operated cranes and this investigation deals only
with electric overhead travelling cranes.
The results from the assessment of current practice are shown in Figure
9.7. The variation of the reliability with respect to the number of cycles taken
for the extreme hoistload distribution (Nextr ) is shown in Figure 9.7(a) with a
ratio of hoistload to total crane weight of r = 0.5. This is the actual ratio of
the 40t crane. The variation of reliability with the ratio of hoistload to total
crane weight is shown in Figure 9.7(b).
40t crane girder, r = 0.5

6.5
SA class 2
SA class 3
6 SA class 4
Euro HC 1
Euro HC 2
5.5 Euro HC 3
Euro HC 4
4.5
3.5
2.5
2 3 4 5 6
10 10 10 10 10
N
extr
(a) Variation with Nextr

6
40t crane girder, Nextr = 10
7
SA class 2
SA class 3
6.5 SA class 4
Euro HC 1
6 Euro HC 2
Euro HC 3
Euro HC 4
5.5
4.5
3.5
2.5
2
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8
Qh/(Qbr + Qcr + Qh)
(b) Variation with ratio of hoistload to total crane weight
Figure 9.7: Assessment of reliability of current practice

Considering Figure 9.7(a), the trend of the reliability is the same as is

observed in Figure 9.1 in that the reliability decreases as Nextr increases.
It can be seen that SABS 0160:1989 has a higher reliability than prEN
1991-3 for this case of the 40t crane girder.. The level of reliability of SABS
0160:1989 at Nextr = 106 cycles, ranges from β = 4.2 for the class 2 crane
to β = 5.75 for the class 4 crane whereas prEN 1991-3 has a reliability of
β ≈ 2.8 to 3.0.
The level of reliability of prEN 1991-3 is lower than is expected and could
be partly due to two conservative assumptions that have been made regarding
the loading. The first assumption is the upper limit of 1.25 SWL for the ‘one
cycle’ distribution which results in the mean of the extreme hoistload distrib-
ution at Nextr = 106 cycles being close to 1.25 SWL. The second conservative
assumption is that the position of the crab is assumed to be deterministic at
the extreme of its travel closest to the wheel being considered resulting in the
maximum possible wheel load always being obtained.
The large difference between the classes for SABS 0160:1989 in Figure 9.7
is partly due to the method of calculating the wheel loads for the economic
design. The vertical and horizontal wheel load are dependent on the class of the
crane as is explained in Chapter 4 and shown in the calculations in Appendix
A. In contrast, the prEN 1991-3 hoist class affects only the dynamic factor for
the vertical loads for LC1.
The second reason for the large difference in reliability between the classes
for SABS 0160:1989 is the failure mode for the reliability analysis is sometimes
dominated by horizontal loads and sometimes by vertical loads. The class 2
crane and the class 3 crane at a ratio of hoistload to total crane weight of r
= 0.85 in Figure 9.7(b) have failure modes dominated by horizontal loads and
therefore have a lower reliability. This can be seen from the sensitivity factors
for the class 2 and 3 cranes at Nextr = 106 cycles and r = 0.5 as shown in
Figures 9.8 & 9.9.
Where:
mf V – represents the modelling uncertainty in the calculation of the vertical

wheel loads
mf H – represents the modelling uncertainty in the calculation of the horizon-

tal wheel loads
6
Sensitivity factors for significant basic variables, Class 2 crane, N = 10 , r = 0.5
extr
0.8
0.6
0.4
Sensitivity factor α
0.2
−0.2
−0.4
−0.6
−0.8
Fy Zyt Zxt Qh Qcr Qbr mf sVt mf V mf sHt mf H
Basic variables
Figure 9.8: Sensitivity factors for class 2 crane, SABS 0160:1989
Sensitivity factors for significant basic variables, Class 3 crane, N = 106, r = 0.5
extr
0.4
0.2
0
−0.2
−0.4
−0.6
−0.8
−1
Fy Zxb Zyb Qh Qcr Qbr mf H mf sHb mf sVb mf V
Basic variables
Figure 9.9: Sensitivity factors for class 3 crane, SABS 0160:1989
It can be seen in Figures 9.8 & 9.9 that the class 2 crane is much more
sensitive to the modelling uncertainty for the calculation of the horizontal
wheel loads than the class 3 crane.
9.1.1.3 Calibration of partial load factors
The critical elements that resulted in the lowest reliability were identified for
use in the code calibration procedure. Of the crane girders and columns for
the three representative cranes, the element with the lowest reliability was the
5t crane column. The complete set of results of the calibration process are
given in Appendix E where it can be seen that the 5t crane column has the
lowest reliability. Two load cases were identified for the code calibration:
1. 5t crane column subject to LC1 maximum vertical loads. This load case
consists of only vertical loads.
2. 5t crane column subject to LC5. This load case consists of both vertical
and horizontal loads.
The calibration of the partial load factors for the different code formats
was carried out as outlined below.
1. Calibration of the partial load factors for code formats 1 and 2

These load factors were calibrated considering LC5 consisting of vertical
and horizontal loads.
2. Calibration of the partial load factors for code formats 3 and 4
a) The crane partial load factors (γC , γCsw , γCh ) for code formats 3
and 4 were calibrated considering LC1 consisting of only vertical
loads
b) The load factors for the horizontal loads for code formats 3 and
4 were calibrated using the partial load factors from step (a) and
considering LC5 consisting of vertical and horizontal loads.
The calibration was carried out considering the critical values of the pa-
rameters determined above, i.e. class 1 crane, Nextr = 106 cycles, r = 0.85.
The partial load factors were varied until the reliability was equal to the target
reliability of βT = 3.0.
The results of the code calibration are given in Table 9.2. Figures 9.10 -
9.12 show the reliability graphs for the critical elements with the calibrated
partial load factors.
Table 9.2: Calibrated partial load factors

Code format γC γCsw γCh γH
1 1.907
2 1.617 1.963
3 1.590 1.261
4 1.354 1.623 1.271
5t crane column, LC5, max horz & vert, class = 1, N = 106

extr
4
γC = 1.8
γC = 1.907
3.8 γC = 2
3.6
3.4
3.2
3.20
3
3.00
2.8 2.77
2.6
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(a) Code format 1

6
5t crane column, LC5, max horz & vert, class = 1, NExtr = 10 , γCsw = 1.617
3.25
γCh = 1.85
γCh = 1.963
3.2
γCh = 2.05
3.15
3.1
3.05
2.95
2.9
2.85
2.8
2.75
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(b) Code format 2
Figure 9.10: Crane only calibration, code formats 1 and 2

6
5t crane column, LC1, max vert, class = 1, Nextr = 10
4.2
γC = 1.5
γ = 1.59
4 C
γ = 1.7
C
3.8
3.6
3.4 3.37
3.2
3
3.00
2.8
2.68
2.6
2.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(a) Code format 3

6
5t crane column, LC1, max vert, class = 1, Nextr = 10 , γCsw = 1.354
3.3
γCh = 1.5
γCh = 1.623
3.2 γCh = 1.7
3.1
2.9
2.8
2.7
2.6
2.5
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(b) Code format 4
Figure 9.11: Crane only calibration, code formats 3 and 4 step (a)
5t crane column, LC5, max horz & vert, class = 1, Nextr = 106, γH = 1.261
4
γC = 1.5
γ = 1.59
C
3.8 γ = 1.7
C
3.6
3.4
3.28
3.2
3
3.00
2.8 2.77
2.6
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(a) Code format 3

6
5t crane column, LC5, max horz & vert, class = 1, NExtr = 10 , γCsw = 1.354, γH = 1.271
3.3
γCh = 1.5
γCh = 1.623
3.2 γCh = 1.7
3.1
2.9
2.8
2.7
2.6
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(b) Code format 4
Figure 9.12: Crane only calibration, code formats 3 and 4 step (b)
9.1.1.4 Verification of partial load factors
The final step in the crane only calibration was to verify the results. This
consisted of ensuring that the reliability of all the elements was not below the
target reliability and assessing the degree of conservatism.
The reliability graphs for all the crane girders and columns from the three
representative cranes were plotted using the calibrated partial load factors and
it was observed that in all cases the reliability was above the target reliability.
The minimum reliabilities for the various elements and load cases ranged from
β = 3.00 - 5.55. The reliability graphs for all the girders and columns with
the calibrated partial load factors are given in Appendix E.
The second step in the verification of the partial load factors was to assess
the degree of conservatism for the support structures as a whole. The excess
costs of the elements for the three representative cranes were calculated as
described in Chapter 8. Figures 9.13 - 9.15 give the results of the cost analyses.
Costs of 5t crane support structure

1.4
Girder
Column
Total
1.2
Excess cost = volume x (β − β )
T
0.8
0.6
0.4
0.2
0
1 2 3 4
Code format
Figure 9.13: Excess cost of the 5t crane support structure

5
x 10 Costs of 40t crane support structure
5
Girder
Column
4.5 Total
4
Excess cost = cost x (β − β )
T
3.5
2.5
1.5
0.5
0
1 2 3 4
Code format
Costs of 260t crane support structure

120
Girder
Column
Total
100
Excess cost = volume x (β − βT)
80
60
40
20
0
1 2 3 4
Code format
It can be clearly seen from Figures 9.13 - 9.15 that the degree of conser-
vatism of the support structures reduces with increasing complexity of the code
format. Two progressions of the increase in complexity of the code format were
identified:
1. Code formats 1 → 2 → 4
The lowest level of complexity is code format 1 with one partial load
factor applied to the calculated characteristic wheel load. The next level
of complexity is code format 2 where two separate partial load factors
are applied to the crane self weight and hoistload before the calculation
of the design wheel loads. The third level of complexity is code format
4 where an additional partial load factor, to code format 2, is applied to
the horizontal wheel loads.
2. Code formats 1 → 3 → 4
This progression starts and ends the same as the previous one but the
intermediate step is seen as applying one partial load factor to the calcu-
lated characteristic wheel loads and an additional factor to the horizontal
wheel loads.
For both of these progressions, there is a decrease in the degree of conser-

vatism as indicated by the excess cost of the support structure. Code format 4
has the least conservatism and can thus be considered the code format which
best meets the code calibration objective.
9.1.2 Combinations of time varying loads

The element that was selected for the investigation into crane loads combined
with other time varying loads was the column for the 5t crane with LC5 max-
imum horizontal load. This was the element with the lowest reliability.
The code format did not have an effect on the results of these investigations
so code format 4 has been chosen. The results are shown with the calibrated
values of the crane partial load factors. The critical values of the parameters
discussed previously have been used here, i.e. ratio of hoistload to total crane
weight r = 0.85; Nextr = 106 cycles and class 1 crane in the case of the crane
load being the leading load.
9.1.2.1 Results of parametric studies
The first parametric study that was carried out for the combination of crane
loads with wind or roof imposed loads was an investigation into the influence
of permanent loads. In order to carry out this investigation a combination of
crane and permanent loads only was considered.
SABS 0160:1989 gives two load combinations for the crane and permanent
loads:
1. 1.2(Permanent) + γC (Crane)
2. 1.5(Permanent)
Both of these load combinations were considered in this investigation. Fig-

ure 9.16 shows the variation of the reliability index with varying ratios of crane
load to crane and permanent load.
5t crane column, LC5, max horz, r = 0.85, γCsw = 1.354, γH = 1.271

3.6
γ = 1.5
Ch
γ = 1.623
3.4 Ch
γCh = 1.7
1.5 Perm
3.2
3
2.8
2.6
2.4
2.2
1.8
0 0.2 0.4 0.6 0.8 1
Crane/(Crane + Permanent)
Figure 9.16: Crane and Permanent loads
As can be seen from Figure 9.16, the reliability decreases as the ratio of
permanent load increases. The assessment of the code provisions for permanent
loads is not within the scope of this project, therefore the ratio of permanent
loads to total loads will be kept very small (p = 0.01), i.e. ratio of crane load
to crane and permanent load close to one, in the following investigations.
The second parametric study that was carried out was an investigation into
the class of crane and the number of cycles considered for Nextr for the crane
load as the accompanying load.
From Figure 9.1 it was observed that the class 1 crane has the lowest
reliability for Nextr > 5 000 cycles. In the case where crane loads are the
accompanying loads for wind or roof imposed loads, Nextr would be chosen to
be the number of cycles the crane is likely to perform during an extreme wind
event or roof loading which is unlikely to be more than 5 000. From Figure 9.1
it was observed that for Nextr < 5 000 cycles, a class 4 crane gives the lowest
reliability. Therefore for the assessment of the reliability of crane loads as the
accompanying load combined with other time varying loads, a class 4 crane
will be considered.
The number of cycles that a crane is likely to perform during an extreme
wind event (e.g. a storm) or roof loading event (e.g. roof maintenance) would
be in the range of 10 - 100. An assessment of the effect of this range in
values of Nextr has been carried out by considering a combination of wind and
crane loads for the wind load dominant case. A nominal value of the crane
combination factor has been chosen as ψCW = 0.6. Figure 9.17 shows the
results of this parametric study in the form of a graph of reliability versus the
ratio of crane load to crane and wind load. It can be seen that the difference
in Nextr has a very small effect on the reliability. A variation in Nextr has a
larger effect on the class 1 - 3 cranes but some conservatism will be included
for these classes due to the class 4 crane being considered as the calibration
case. A similar, validating, parametric study will be carried out for the class
1 crane after the calibration has been carried out to ensure that the reliability
is above the target reliability.
5t column, LC5, max horz, r = 0.85, p = 0.01, γCsw = 1.354, γCh = 1.623, γH = 1.271
3.5
Nextr = 10
Nextr = 20
Nextr = 50
Nextr = 100
3
2.5
1.5
1
0 0.2 0.4 0.6 0.8 1
Crane/(Crane + Wind)
Figure 9.17: Effect of Nextr for crane load as accompanying load
A value of Nextr = 20 cycles was considered reasonable for the number of

cycles a crane will perform during an extreme wind or roof loading event and
has been used for the calibration.
9.1.2.2 Combination of crane and wind loads
The two load combinations that were considered for the combination of crane
and wind loads and the partial load factors for permanent and wind loads are
given below.
1. Crane dominant: γP (Permanent) + γC (Crane) + ψW C 1.3(Wind)
2. Wind dominant: γP (Permanent) + ψCW γC (Crane) + 1.3(Wind)
Where:
γP – partial load factor for permanent loads

γP = 1.2 when the permanent loads are in unfavourable combination
with crane and wind loads
γP = 0.9 when the permanent loads are in favourable combination with
the crane and wind loads
γC – represents conceptually the factoring of the crane loads by whichever

code format is being considered
In the case where the crane load was the accompanying load, the crane
partial load factors (γC , γCsw , γCh ) were multiplied by the crane combination
factor.
The reliability was determined as a function of the ratio of crane load to
crane and wind load. The ratios are expressed in terms of the unfactored,
characteristic values.
Figure 9.18 shows the reliability of the column with varying ratios of crane
load to crane and wind load. Figure 9.18(a) shows the case where wind is
the leading load and Figure 9.18(b) shows the case where the crane load is
the leading load. It can be seen from these graphs that the values of the
combination factors which keep the reliability constant, in the pertinent range
of ratios, for each case are:
1. Combination factor applied to wind loads: ψW C = 0
2. Combination factor applied to crane loads: ψCW = 0.65
Figure 9.19 shows the combined graph for the crane dominant and the wind
dominant cases with these combination factors.
5t column, LC5, max horz, r = 0.85, p = 0.01, γ = 1.354, γ = 1.623, γ = 1.271

Csw Ch H
3.5
ψC = 0.55
ψC = 0.65
ψC = 0.75
3
2.5
1.5
0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
(a) Wind dominant

5t crane column, LC5, max horz, r = 0.85, p = 0.01, γCsw = 1.354, γCh = 1.623, γH = 1.271
4
ψW = 0
ψW = 0.1
3.5
ψ = 0.2
W
3
2.5
1.5
0.5
−0.5
−1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
(b) Crane dominant
Figure 9.18: Calibration of combination factors for crane and wind loads
5t column, LC5, max horz, r = 0.85, p = 0.01, γ = 1.354, γ = 1.623, γ = 1.271

Csw Ch H
3.5
ψC = 0.65
ψW = 0
3
2.5
2
1.5
0.5
−0.5
−1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure 9.19: Calibrated combination factors for crane and wind
The investigation into the sensitivity of a class one crane to the number of
cycles considered for Nextr is shown in Figure 9.20 with ψCW = 0.65. Compar-
ing this graph with Figure 9.17 it was observed that the class 1 crane is more
sensitive to Nextr but that sufficient conservatism is incorporated in the com-
bination factor so that the reliability does not fall below the target reliability.
Therefore no further investigation is required into the exact number of cycles
to be considered for Nextr .
9.1.2.3 Combination of crane and roof imposed loads
The two load combinations that were considered for the combination of crane
and roof imposed loads and the partial load factors for permanent and roof
imposed loads are given below:
1. Crane dominant: 1.2(Permanent) + γC (Crane) + ψRC 1.6(Roof imposed)
2. Roof dominant: 1.2(Permanent) + ψCR γC (Crane) + 1.6(Roof imposed)
In the same way as for the combination of crane and wind loads, γC repre-
sents conceptually the crane partial load factors from whichever code format
is being considered.
5t column, LC5, max horz, r = 0.85, p = 0.01, γCsw = 1.354, γCh = 1.623, γH = 1.271
3.6
Nextr = 10
Nextr = 20
3.4 Nextr = 50
Nextr = 100
3.2
Reliability index, β 3
2.8
2.6
2.4
2.2
1.8
0 0.2 0.4 0.6 0.8 1
Figure 9.20: Sensitivity of class 1 crane to Nextr
The crane combination factor ψCR was applied to the crane loads by mul-
tiplying the crane partial load factors (γC , γCsw , γCh ) by the combination
factor.
The reliability of the column was determined as a function of the ratio of
crane load to crane and roof imposed load. The ratio is expressed in terms of
the unfactored, characteristic loads.
Results are shown in Figure 9.21 for the calibration of the combination
factors for crane and roof imposed loads. Figure 9.21(a) shows the full relia-
bility graph and Figure 9.21(b) shows the more relevant range of reliabilities
between zero and five.
The reliability curves which relate to the crane dominant case are the
mostly horizontal curves and are labelled on the figures, the reliability curves
which relate to the roof dominant case are the curves with the steeper gradients
and are also labelled on the figures.
The roof dominant curve could not be presented for ratios of crane load to
crane and roof loads of less than 0.35 because the reliability was too large and
the numerical FORM analysis method broke down. The trend of the curves is
clearly increasing though, into the region of reliability greater than 5 which is
not in the critical range.
5
ψC = 0
ψC = 0.1
Crane dominant ψR = 0
0 ψR = 0.05
−5 Roof dominant
−10
−15
−20
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Crane/(Crane + Roof)
(a) Full range of reliability

5
ψC = 0
Roof dominant ψC = 0.1
4.5
ψR = 0
4 ψR = 0.05
3.5
3
Crane dominant
2.5
1.5
0.5
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Crane/(Crane + Roof)
(b) Relevant range of reliability
Figure 9.21: Calibration of combination factors for crane and roof imposed loads
From Figure 9.21, for ratios of crane load to crane and roof imposed load
less than 0.4, the roof dominant case has a higher reliability than the crane
dominant case, even with a crane combination factor of ψCR = 0. This can be
explained from the parameters of the statistical model for the lifetime maxi-
mum roof imposed loads. The mean value of the lifetime maximum roof loads
is:
Rnom
µR =
3.75
With such a large bias in the roof imposed loads, design of members ac-
cording to these loads is over conservative, hence the large reliability for small
ratios of crane to crane and roof loads.
Considering Figure 9.21, for ratios of crane load to crane and roof load
greater than 0.4, the crane dominant case has the higher reliability, even in
the case where the combination factor for roof loads ψRC = 0. Again, the
statistical model for the roof imposed loads explains this phenomenon. The
point-in-time roof imposed load for combination with other time varying loads
is taken to be zero. This means that any combination factor for roof loads
greater than zero is conservative and essentially results in loads taken into
account in the design which are not present for the reliability analysis.
These results indicate that the combination of crane load with roof imposed
load is not a rational load combination and the two load cases should rather
be treated separately as:
1. Crane dominant: 1.2(Permanent) + γC (Crane)
2. Roof dominant: 1.2(Permanent) + 1.6(Roof imposed)
In other words, the appropriate values of the combination factors are:
1. Combination factor applied to roof loads: ψRC = 0
2. Combination factor applied to crane loads: ψCR = 0
9.2 Accidental limit state

The accidental limit state considered the crane load case which are classified
as accidental loads in prEN 1991-3. This accidental load case consists of the
crane running into the end stops on the end of the runway causing longitudinal
forces which are resisted by the longitudinal runway bracing. The 40t crane
was taken to be representative for this load case.
9.2.1 Results of parametric studies

The first two parametric studies that were carried out were investigations into
the class of crane that resulted in the lowest reliability and the number of
cycles that was considered for Nextr .
The code format did not have an effect on the results of these parametric
studies so for the sake of simplicity, code format one has been selected.
The number of cycles considered for Nextr is the number of times that this
action is likely to occur over the lifetime of the crane. The extreme hoistload
distribution then models the probability distribution of the largest load that
will be lifted when the crane runs into the end stops. This load case is not
classified under crane loads arising from normal operation, being classified as
an accidental load situation in prEN 1991-3, and therefore should not occur
very often. Low values of the number of cycles have been considered, in the
range 5 ≤ Nextr ≤ 35.
Figure 9.22 shows the reliability analysis curves for the four classes of crane
for the range of Nextr given above, for code format one, with the accidental
partial load factor, γC = 1.0.
40t crane longitudinal bracing, γC = 1, r = 0.85

3.5
Class 1
Class 2
3.4 Class 3
Class 4
3.3
3.2
3.1
2.9
2.8
2.7
2.6
5 10 15 20 25 30 35
N
extr
Figure 9.22: Effect of class and cycles for accidental limit state
From Figure 9.22, it was observed that the class four crane gives the lowest
reliability and that, as expected, the reliability decreases with increasing Nextr .
The decrease in reliability from Nextr = 5 cycles to Nextr = 35 cycles is not
large, ranging from β = 2.81 for Nextr = 5 to β = 2.68 for Nextr = 35.
The number of cycles that was selected for Nextr for the calibration of the
accidental limit state was Nextr = 10. The reliability implications of calibration
with Nextr = 10 in the event of the actual number of cycles being larger will
be investigated after the calibration has been carried out.
9.2.2 Calibration of partial load factors

Figure 9.23 shows the reliability curves with the calibrated partial load factors
for code formats one and two.
It was found that the partial load factors that resulted in reliabilities equal
to the target reliability of βT = 4.0 were:
• Code format 1:
– γC = 1.196
• Code format 2:
– γCsw = 1.344
– γCh = 1.448
The partial load factors applied separately to the crane self weight and
hoistload before the calculation of the design forces, for code format 2 are
significantly larger than the partial load factor for code format 1 because the
buffer forces are proportional to the square root of the weights of the crane
and hoistload.
9.2.3 Verification of partial load factors

As mentioned above, the number of cycles selected for Nextr for the calibration
was 10. The number of cycles was selected to be small due to the fact that
the crane running into the end stops should only occur infrequently, however,
10 cycles was an arbitrary selection.
Figure 9.24 shows the effect on the reliability of selecting different numbers
of cycles for Nextr . Code format one is shown, applying the calibrated partial
load factor.
40t crane longitudinal bracing, class 4, Nextr = 10

5
γC = 1.1
γ = 1.196
4.8 C
γ = 1.3
C
4.6 4.57
4.4
4.2
4
4.00
3.8
3.6
3.42
3.4
3.2
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(a) Code format 1

40t crane bracing, class 4, γCsw = 1.344, Nextr = 10
4.25
γCh = 1.35
γCh = 1.448
4.2
γCh = 1.55
4.15
4.1
4.05
3.95
3.9
3.85
3.8
3.75
0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
(b) Code format 2
Figure 9.23: Calibration of partial load factors for accidental limit state
40t crane longitudinal bracing, class = 4, γC = 1.196

4.25
Nextr = 5
Nextr = 10
4.2 Nextr = 25
Nextr = 35
Reliability index, β 4.15
4.1
4.05
3.95
3.9
3.85
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
Figure 9.24: Effect of different values of Nextr on accidental limit state reliability,
class 4
It can be seen from Figure 9.24 that the reliability falls below the target
reliability of βT = 4.0 for the larger values of Nextr for ratios of hoistload to
total crane weight greater than 0.6. It has been noted before that the cranes
with higher ratios of hoistload to total crane weight tend to be class one cranes.
Figure 9.25 shows the effect on the reliability of selecting different numbers of
cycles for Nextr for a class one crane using the partial load factors calibrated
for the class 4 crane.
From Figure 9.25, for the class one crane, the reliability remains above the
target reliability of βT = 4.0 for all ratios of hoistload to total crane weight
for all values of Nextr .
It is thus concluded that calibrating the partial load factors for a class four
crane with Nextr = 10 cycles ensures that the reliability remains above the
target reliability of βT = 4.0 for all relevant ranges of parameters.
9.3 Fatigue
The results of the reliability assessment of fatigue are discussed first for the
fatigue loading method given in prEN 1991-3 and these results are then ap-
plied to the alternative method of assessing fatigue where the detailed crane
behaviour is considered.
9.3. Fatigue 225
40t crane longitudinal bracing, class = 1, γC = 1.196

4.6
Nextr = 5
Nextr = 10
4.55 Nextr = 25
Nextr = 35
4.5
4.45
4.4
4.35
4.3
4.25
4.2
4.15
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
Figure 9.25: Effect of different values of Nextr on accidental limit state reliability,
class 1
The fatigue provisions in prEN 1991-3 were assessed by calculating the

bounds of reliability for a given load spectrum class Q, as discussed in Chapter
8. An upper bound, lower bound and average value of reliability were obtained
for each load spectrum class. The reliability was determined as a function of
the number of cycles the crane will perform over its lifetime.
9.3.1 Results of parametric studies

The first parametric study carried out for fatigue was the assessment of the
effect of using a single or a double slope S-N curve. A typical graph is shown
in Figure 9.26 which gives the reliability of the 40t crane girder fatigue at the
bottom of the intermediate stiffener, for load spectrum class Q5 .
Comparing the graph obtained considering a single slope S-N curve with
that considering a double slope S-N curve, it is seen that there is a tendency
for the reliability to increase in the region of higher cycles for the double slope
S-N curve, most markedly so for the lower bound. This is due to the fact
that the designs carried out with higher cycles would have a higher λ factor
resulting in a larger beam. When the actual crane movement is simulated, a
larger proportion of the stresses would fall into the lower region than those for
a smaller beam. With a double slope S-N curve these lower stresses result in
less damage than for a single slope S-N curve resulting in a higher reliability.
40t crane girder, bottom of stiffener, Q5, γF = 1

1.8
N , kQ
min min
Nmin, kQmax
1.6 Nmax, kQmin
N , kQ
max max
1.4
1.2
0.8
0.6
0.4
0.2
4 5 6 7
10 10 10 10
Number of crane cycles in girder lifetime, N
(a) Single slope S-N curve

1.8
N , kQ
min min
Nmin, kQmax
1.6 Nmax, kQmin
N , kQ
max max
1.4
1.2
0.8
0.6
0.4
0.2
4 5 6 7
10 10 10 10
(b) Double slope S-N curve
Figure 9.26: Effect of single or double slope S-N curve

9.3. Fatigue 227
The second phenomenon that is observed is that when a double slope S-N
curve is considered, there is a dip in reliability in the region of 2 × 105 cycles,
more pronounced in the upper bound. This is due to the fact that there are
more uncertainties involved in using a double slope S-N curve. This can be
observed from Figures 9.27 - 9.29 which give the sensitivity factors for reli-
ability analyses carried out on the 40t crane girder bottom of intermediate
stiffener for load spectrum class Q5 . Figure 9.27 shows the sensitivity factors
for a reliability analysis considering a single slope S-N curve at 2 × 106 cycles.
Figures 9.28 & 9.29 show the sensitivity factors for reliability analyses consid-
ering a double slope S-N curve, Figure 9.28 is at 1.25 × 105 cycles and Figure
9.29 is at 2 × 106 cycles. Figures 9.27 & 9.28 are almost identical because
with 1.25 × 105 cycles the stresses are in the upper region of the S-N curve
and only the first slope is considered with m = 3 as for the single slope S-N
curve. Figure 9.29 shows the additional uncertainty which is included for a
double slope S-N curve in the variable So which is the level of stress at which
the gradient of the S-N curve changes.
For simplification purposes, a single slope S-N curve has been considered for
the further investigations as the double slope S-N curve does not significantly
change the trend of the reliability.
Sensitivity factors for significant basic variables, single slope S−N curve, kQ , cycles = 2x106
min
0.6
0.4
0.2
−0.2
−0.4
A1 D hw cycles Bglob Stress Bloc

Basic variables
Figure 9.27: Sensitivity factors for a single slope S-N curve at 2 × 106 cycles
Sensitivity factors for significant basic variables, double slope S−N curve, kQ , cycles = 1.25x105
min
0.6
0.4
0.2
−0.2
−0.4
A1 D hw So cycles Bglob Bloc Stress

Basic variables
Figure 9.28: Sensitivity factors for a double slope S-N curve at 1.25 × 105 cycles
6
Sensitivity factors for significant basic variables, double slope S−N curve, kQ , cycles = 2x10
min
0.6
0.4
0.2
−0.2
−0.4
So A1 D hw cycles Bglob Bloc Stress

Basic variables
Figure 9.29: Sensitivity factors for a double slope S-N curve at 2 × 106 cycles
9.3. Fatigue 229
The second parametric study carried out was to assess the difference in
reliability between the load spectrum classes. Figure 9.30 gives the reliability
bounds for the 40t crane girder bottom of the intermediate stiffener. Figure
9.30(a) shows results for load spectrum class Q2 and Figure 9.30(b) shows
results for load spectrum class Q5 .

2.4
N , kQ
min min
2.2 Nmin, kQmax
Nmax, kQmin
2 N , kQ
max max
1.8
1.6
1.4
1.2
0.8
0.6
0.4
0.2
4 5 6 7
10 10 10 10
(a) Load spectrum class Q2

2.4
N , kQ
min min
2.2 Nmin, kQmax
Nmax, kQmin
2 N , kQ
max max
1.8
1.6
1.4
1.2
0.8
0.6
0.4
0.2
4 5 6 7
10 10 10 10
(b) Load spectrum class Q5
Figure 9.30: Effect of different load spectrum classes

An initial steep downwards slope is observed in the graph for load spectrum
class Q2 . This can be explained by referring to Table 2.18 on page 51 and
considering load spectrum class Q2 . The first four fatigue classes are all S0
and therefore for a number of cycles up to 1.25 × 105 the same λ factor is
used for the design. This results in over design of elements subject to a lower
number of cycles and hence a greater reliability. After the initial downwards
slope, horizontal lines are obtained where the level of reliability is the same
for class Q2 and class Q5 .
The same trend of the results was observed for the 5t crane corbel to
column welded connection, however, a different trend was observed for the 40t
crane girder top flange to web weld. Figure 9.31 shows results for the 40t crane
girder top flange to web weld, again comparing load spectrum classes Q2 and
Q5 .
The difference between the graphs obtained for the fatigue at the bottom
of the stiffener and the fatigue at the top flange to web weld can be explained
by considering the calculation of the λ factors. The λ factors for the normal
stresses have been derived considering a single slope S-N curve with m = 3 [2].
This results in the constant reliability observed for the graphs generated for a
single slope S-N curve for the 40t crane girder at the bottom of the stiffener
which displays the same behaviour as the 5t crane corbel connection. The
graphs for the 40t crane girder considering the fatigue of the top flange to
web weld are not horizontal as with the other elements. This is because the
stresses involved are a combination of shear and normal stresses. The stress
considered for the fatigue analysis was the resulting principal stress and the
S-N curve for the normal stress was used with the slope given previously of
m = 3. The lambda factor for the calculation of the shear stresses in the design
was derived considering a single slope S-N curve with a slope of m = 5 [2].
This would result in lower reliability for the higher cycles if the reliability was
assessed with an S-N curve with a slope of m = 3. The trend of the reliability
values for the upper and lower bounds and the average value is not significantly
affected with the values centered on those for the other elements.
The 40t crane girder, fatigue at the bottom of the intermediate stiffener
was considered for the calibration process. The load spectrum class considered
was Q5 with a single slope S-N curve.
9.3. Fatigue 231
40t crane girder, top flange to web weld, Q2, γF = 1

2.5
Nmin, kQmin
N , kQ
min max
Nmax, kQmin
2 Nmax, kQmax
1.5
0.5
0
4 5 6 7
10 10 10 10
(a) Load spectrum class Q2

40t crane girder, top flange to web weld, Q5, γF = 1
2.5
Nmin, kQmin
Nmin, kQmax
Nmax, kQmin
2 Nmax, kQmax
1.5
0.5
0
4 5 6 7
10 10 10 10
(b) Load spectrum class Q5
Figure 9.31: 40t crane girder, top flange to web weld

9.3.2 Calibration of partial load factors

Considering the 40t crane girder bottom of the intermediate stiffener, in Figure
9.30(b), the upper bound of the reliability is βU ≈ 1.6, the lower bound is
βL ≈ 0.5 and the average value is βA ≈ 1.0. It was found that applying a
partial load factor of γF = 1.25, the average value of the reliability reached
the target reliability of βT = 1.5. The resulting reliability for the three elements
for load spectrum class Q5 are shown in Figures 9.32 - 9.34.
5t crane corbel to column welded connection, Q5, γ = 1.25

F
2.4
N , kQ
min min
Nmin, kQmax
2.2
Nmax, kQmin
N , kQ
2 max max
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10
Figure 9.32: Calibration results for 5t crane corbel to column connection
9.3.3 Simulation of crane behaviour

The second method of assessing fatigue that could be included in the code is
recommending crane load combinations to be considered for fatigue and the
responsibility is then on the support structure designer to simulate the crane
behaviour to determine the fatigue damage in the element.
This method of considering fatigue was assessed by carrying out the eco-
nomic design of each element using the prEN 1991-3 method and then simu-
lating the crane behaviour and assessing the resulting fatigue damage in the
element.
This was carried out for each element for the average level of reliability
combination of maximum number of cycles and minimum load spectrum for
load spectrum class Q5 . The average damage for all the elements for each
9.3. Fatigue 233
40t crane girder, bottom of stiffener, Q5, γF = 1.25

2.4
N , kQ
min min
Nmin, kQmax
2.2 Nmax, kQmin
N , kQ
max max
2
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10
Figure 9.33: Calibration results for 40t crane girder bottom of stiffener
40t crane girder, top flange to web weld, Q5, γF = 1.25

2.4
Nmin, kQmin
Nmin, kQmax
2.2
Nmax, kQmin
Nmax, kQmax
2
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10
Figure 9.34: Calibration results for 40t crane girder top flange to web weld
point on the reliability curve was calculated and taken to represent the fatigue
damage resulting from the true crane behaviour on the economically designed
elements.
Considering the case when the fatigue partial load factor was γF = 1.0, the
average fatigue damage resulting from the simulated crane behaviour on the
elements which had been economically design according to the prEN 1991-3
method was 0.851. This indicates that there is some conservatism built into
the prEN 1991-3 method.
Considering the case when the fatigue partial load factor was γF = 1.25,
which results in a reliability equal to the target reliability, the average fatigue
damage resulting from the simulated crane behaviour was 0.419.
A fatigue design equation in terms of fatigue damage can be considered in
the form of:
D≥d (9.3.1)
Where:
D – fatigue resistance
d – fatigue damage resulting from simulated crane behaviour
In this case, the fatigue damage at failure is treated as the resistance and
the fatigue damage resulting from the simulated crane behaviour is treated as
the loading.
To achieve a target reliability of βT = 1.5, the actual, nominal, fatigue
damage in the elements was 0.419. When designing according to prEN 1991-3,
the design value of the fatigue damage resulting from simulated crane behav-
iour was 0.851.
The partial factor which should be applied to the fatigue damage to obtain
the target reliability is therefore:
ddes 0.851
γd = = = 2.03 (9.3.2)
dnom 0.419
Partial factors are traditionally applied to loads rather than to load effects,
such as fatigue damage, so the partial factor obtained above will be reformu-
lated so as to relate it to the crane wheel load rather than the fatigue damage
resulting from the wheel load.
9.3. Fatigue 235
For an S-N curve with a constant slope of m = 3, the fatigue damage is

calculated using the equation given below:
X (Si )3 ni
d= (9.3.3)
A N
i
The damage is therefore proportional to the cube of the stress. Assuming

that the stress is linear with respect to the load, the partial load factor which
should be applied to the crane wheel load to achieve a target reliability of
βT = 1.5 is:
r
√ ddes
γF = 3 γd = 3
dnom
r
3 0.851 (9.3.4)
=
0.419
= 1.26
This value of the fatigue partial load factor of γF = 1.26 corresponds well
with the value of γF = 1.25 obtained considering the reliability graphs.
The above discussion was based on a structural element that had been
economically designed according to the fatigue design method given in prEN
1991-3. The damage in the economically designed element due to the simu-
lation of the crane behaviour was calculated and was found to be 0.851 for a
fatigue partial load factor of γF = 1.0.
The second method of fatigue assessment entails carrying out the design of
the girder by simulating the crane behaviour to determine the fatigue damage
and taking the fatigue resistance as the fatigue damage at failure equal to 1.0.
In this case the design method is the same as the method that was used to
assess the fatigue damage in the girder designed according to the prEN 1991-3
method. Whereas for design according to prEN 1991-3, with a fatigue partial
load factor of γF = 1.0 the damage from simulating the crane behaviour is
0.851, carrying out the design by simulating the crane behaviour would result
in a corresponding fatigue damage of 1.0.
The required fatigue partial load factor applied to the crane loads in the
case where the design is carried out by simulating the crane behaviour is given
by:
r
3ddes
γF =
d
r nom
3 1 (9.3.5)
=
0.419
= 1.34
The partial load factor is larger than that required for design according
to the prEN 1991-3 method because of the in built conservatism in the prEN
1991-3 method which is not present when considering the crane behaviour for
the design.
9.4 Summary
The partial load factors and combination factors that resulted from the code
calibration exercise are given in Tables 9.3 & 9.4.
Partial load factor Symbol Calibrated

value
Crane partial load factor, code format 1 γC 1.907
Self weight partial load factor, code format 2 γCsw 1.617
Hoistload partial load factor, code format 2 γCh 1.963
Crane partial load factor, code format 3 γC 1.590
Horizontal partial load factor, code format 3 γH 1.261
Self weight partial load factor, code format 4 γCsw 1.354
Hoistload partial load factor, code format 4 γCh 1.623
Horizontal partial load factor, code format 4 γH 1.271
Partial load factor for accidental load, code format 1 γC 1.196
Partial load factor for accidental load applied to γCsw 1.344
crane self weight, code format 2
Partial load factor for accidental load applied to γCh 1.448
hoistload, code format 2
Partial load factor for fatigue loads for prEN 1991-3 γF 1.25
design method
Partial load factor for fatigue loads for design by con- γF 1.34
sidering the crane behaviour
9.4. Summary 237
Table 9.4: Calibrated combination factors

Combination factor Symbol Calibrated
value
Combination factor for crane loads when combined ψCW 0.65
with wind loads
Combination factor for wind loads when combined ψW C 0
with crane loads
Combination factor for crane loads when combined ψCR 0
with roof imposed loads
Combination factor for roof imposed loads when ψRC 0
combined with crane loads
Chapter 10
Discussion of results
The sensitivity factors for the reliability analyses showed that the modelling
uncertainties in the calculation of the vertical and horizontal crane wheel loads
were the dominant parameters. These were also the basic variables for which
there was the least information available for the stochastic modelling. Sensi-
tivity studies on the parameters of the modelling uncertainties are discussed in
this chapter. The further work which could be carried out on the investigation
of code calibration for crane support structures is discussed.
10.1 Sensitivity factors

The selection of the 5t crane column for the code calibration procedure was
due to it having the lowest reliability of all the structural elements considered.
The sensitivity factors from the reliability analysis were assessed to investigate
the reason for this.
The sensitivity factors for a reliability analysis on the 5t crane column with
LC5 for a class 1 crane with the ratio of hoistload to total crane weight r =
0.85 and Nextr = 106 cycles for code format 4, are given in Figure 10.1.
It was observed from the sensitivity factors that the modelling uncertainties
have the most significant effect on the reliability. The reason why the 5t crane
column has a lower reliability than the other elements is that the dominant
loading is bending due to the horizontal crane wheel loads. The modelling
factors for the uncertainty in the calculation of the horizontal loads and the
bending moments obtained from the structural analysis program have large
variations in comparison with the other variables, resulting in a low reliability.
239
240 Chapter 10. Discussion of results
Sensitivity factors for significant basic variables

0.6
0.4
0.2
0
−0.2
−0.4
−0.6
−0.8
mf R fy Zplx Qh Qbr e mf V mf Bend mf H
Basic variables
Figure 10.1: Sensitivity factors for the 5t crane column
Where:
mf R – modelling factor for the uncertainty in the resistance of the column
fy – yield stress of steel
Zplx – plastic section modulus of the column
Qh – weight of the hoistload
Qbr – weight of the crane bridge
e – eccentricity of the vertical wheel load
mf V – modelling uncertainty in the vertical wheel load calculations
mf Bend – modelling uncertainty in the bending moments obtained from the

structural analysis program
mf H – modelling uncertainty in the calculation of the horizontal wheel loads
It was the general trend for all of the elements for the modelling factors
to be the dominant variables, especially the modelling factors allowing for
uncertainties in the calculation of the wheel loads. The parameters of these
modelling factors were estimated because there was no literature available,
as discussed in Chapter 6. The parameters of these modelling factors are
therefore uncertain. In order to investigate the effect of different coefficients
10.1. Sensitivity factors 241
of variation for these modelling factors, a sensitivity study was carried out on
the reliabilities of the crane girders and columns.
10.1.1 Sensitivity of reliability to modelling uncertainty

parameters
Four sets of coefficients of variation were considered for the modelling uncer-
tainties for the calculation of the wheel loads:
cov1 δV = 5% δH = 10%
cov2 δV = 5% δH = 15%
cov3 δV = 10% δH = 15%
cov4 δV = 10% δH = 20%
The code calibration as presented in Chapter 9 was based on cov2.

In order to assess the sensitivity of the code calibration results to variations
in the modelling uncertainties, the reliability of the girders and elements were
calculated in a similar manner to the cost analysis for the verification of the
results presented in Chapter 9. Following the same method of finding the
critical load case for each element and assessing its reliability with respect to
that load case, the minimum level of reliability of each element was assessed.
The calibrated crane partial load factors were used for this assessment and it
was observed that the code format did not have an effect on the trend of the
results so code format 4 has been considered here. The minimum reliability
for each of the crane girders and columns for each of the sets of modelling
uncertainty parameters are shown in Figure 10.2.
As expected, the reliability decreases as the coefficients of variation of
the modelling uncertainties increase. The difference in decrease between the
different sets of coefficients of variation is due to the different influences of the
vertical and horizontal wheel loads on each element. For example, the 5t crane
column is loaded primarily by horizontal wheel loads, and there is therefore
only a very small change in the reliability between cov2 and cov3 where there
is only an increase in the coefficient of variation of the modelling uncertainty
in the calculation of the vertical wheel loads. Similarly for the 5t crane girder
which is loaded primarily by vertical wheel loads there is only a small change
in reliability between cov1 & cov2 and cov3 & cov4.
Sensitivity of reliability to c.o.v. of modelling uncertainties

5
cov1
cov2
4.5 cov3
cov4
4
3.5
2.5
1.5
0.5
0
5t girder 5t column 40t girder 40t column 260t girder 260t column
Structural element
Figure 10.2: Sensitivity of reliability to c.o.v. of modelling uncertainties
The element chosen for the code calibration was the element which had the
lowest reliability, i.e. the 5t crane column. The resulting conservatism for the
other crane girders and columns ensures that as the coefficients of variation of
the modelling uncertainties increase, though the reliabilities of these elements
decrease they do not fall below the target reliability of βT = 3.0. The reliability
for the critical element, however, does fall below the target reliability with a
reliability of β = 2.77 for cov4.
This sensitivity analysis shows that, though the majority of the elements
have sufficient conservatism to meet the code calibration requirements for a
range of coefficients of variation of the modelling uncertainties, more inves-
tigation is needed into the modelling uncertainties in the calculation of the
crane wheel loads. The information that would assist in the improvement of
the modelling uncertainty parameters would be a comparison between mea-
sured wheel loads to calculated wheel loads, both vertical and horizontal. The
information would be most valuable if the comparison was carried out over
a range of parameters similar to the parameters considered in this study but
also including the stiffness of the crane and support structure.
10.1.2 Sensitivity of partial load factors to modelling

uncertainty parameters
The modelling uncertainty for the calculation of the horizontal wheel loads was
the modelling factor that had the greatest influence on the crane partial load
factors and was the modelling factor for which there was the least information
available to determine statistical parameters.
The crane partial load factors which are influenced by the modelling un-
certainty for the calculation of the horizontal loads are:
γC – partial load factor applied to characteristic crane wheel loads in code

format 1
γCsw – partial load factor applied to crane self weight in code format 2
γCh – partial load factor applied to hoistload in code format 3
γH – additional partial load factor applied to horizontal wheel loads for code
formats 3 & 4
A sensitivity study was carried out on the partial load factors mentioned
above. The bias and coefficient of variation of the modelling uncertainty of
the calculation of the horizontal wheel loads were varied and the partial load
factor required to obtain a reliability equal to the target reliability of βT = 3.0,
was calculated.
Only partial load factors γC , applied to the characteristic wheel load, and
γH , the additional factor applied to the horizontal wheel load are considered
for the sensitivity study. The results from the sensitivity studies for these two
partial load factors are similar and were judged to be representative of the
sensitivity of the load factors to the modelling uncertainty parameters.
The results of the sensitivity study are shown in Figures 10.3 & 10.4. The
sensitivity of the crane partial load factor, γC , for code format 1, to the mod-
elling uncertainty parameters is shown in Figure 10.3. The sensitivity of the
additional partial load factor applied to the horizontal wheel loads, γH , is
shown in Figure 10.4, the values of γH do not vary significantly from code
format 3 to 4, therefore Figure 10.4 shows the results for code format 3.
The bias of the modelling uncertainty, λmf H , is defined as the ratio of
mean horizontal wheel load to calculated characteristic horizontal wheel load:
Variation of γC with modelling uncertainty parameters, code format 1

3
λmfH = 1.2
λ = 1.1
mfH
2.8 λ =1
Partial factor for crane wheel loads, γC
mfH
λmfH = 0.9
2.6 λ = 0.8
mfH
2.4
2.2
1.8
1.6
1.4
5 10 15 20 25 30 35
Coefficient of variation of horizontal wheel load modelling uncertainty (%)
Figure 10.3: Sensitivity of γC to horizontal wheel load modelling uncertainty para-

meters
Variation of γH with modelling uncertainty parameters, code format 3

2.2
λmfH = 1.2
λmfH = 1.1
Partial factor for horizontal wheel loads, γH
2 λmfH = 1
λmfH = 0.9
λmfH = 0.8
1.8
1.6
1.4
1.2
0.8
5 10 15 20 25 30 35
Coefficient of variation of horizontal wheel load modelling uncertainty (%)
Figure 10.4: Sensitivity of γH to horizontal wheel load modelling uncertainty para-

meters
H̄
λmf H = (10.1.1)
Hc
In other words, a bias greater than one indicates an unconservative model
which consistently underestimates the value of the wheel loads and conversely
a bias less than one indicates a conservative model which consistently overes-
timates the value of the wheel loads.
The trends of the partial load factors are:
1. As the bias increases, the required partial load factor to achieve a target
reliability of βT = 3.0 also increases.
2. As the coefficient of variation increases, the partial load factor also in-
creases, with the curve becoming steeper in the region of larger coeffi-
cients of variation.
The parameters of the modelling uncertainty for the horizontal loads that
were used for the code calibration were:
λmf V = 1.0 δmf V = 15%
These values resulted in partial load factors of:
γC = 1.907
γH = 1.261
Load models which have been rationally developed, such as the crane load
models in prEN 1991-3, are unlikely to have a high bias (i.e. be unconservative)
combined with a large coefficient of variation. Therefore, considering Figures
10.3 & 10.4, the curve for a bias of 1.2 is unlikely to realistically go beyond a
coefficient of variation of modelling uncertainty of 5% and for a bias of 1.1, the
coefficient of variation is unlikely to be larger than 15%. This trend suggests
a reasonable upper limit of γC = 2.3 and γH = 1.5.
This sensitivity study confirms the fact that the results of the code cal-
ibration are indeed sensitive to the parameters of the modelling uncertainty
for the calculation of the horizontal wheel loads and that more investigation
is required into these parameters.
10.2 Code format
Four code formats were considered for the ultimate limit state and have been
discussed in section 8.3. They are briefly outlined below:
1. One partial load factor applied to the calculated wheel load
2. Two partial load factors, one applied to the crane self weight and one
applied to the hoistload

horizontal loads

horizontal loads
It can be clearly seen from the cost analyses for the girders and columns
presented in Figure 9.13 - 9.15 that the code format which results in the
lowest cost support structure is code format 4. This code format allows for the
different bias and variability between the crane self weight and hoistload by
applying different partial load factors to each. It also allows for the higher level
of uncertainty in the calculation of the horizontal wheel loads than the vertical
wheel loads by applying an additional partial load factor to the calculated
horizontal wheel loads.
Code format 4 best fulfils the requirements of the code calibration exercise
in that it achieves the most consistent level of reliability over a range of load-
ing conditions and crane parameters resulting in the most economic support
structures.
The disadvantage of code format 4 is that it entails more partial load factors
than the current code format and is therefore a more complicated code format
for designers. One method of ameliorating this problem would be to include
the additional partial load factor for the horizontal loads in the calculation
model for the wheel loads. This would result in a code format more consistent
with current practice at the expense, however, of transparency.
10.2. Code format 247
Characteristic loads are used in SABS 0160:1989 for the assessment of acciden-
tal load situations, this implies a partial load factor of 1.0. In order to achieve
the target reliability of βT = 4.0 for the accidental limit state the following
partial load factors were required:
• Code format 1:
– γC = 1.196
• Code format 2:
– γCsw = 1.344
– γCh = 1.448
In order to keep the familiar code format, one manner in which these factors
may be included in the code is to incorporate them into the dynamic factor
φ7 .
These partial load factors have been calibrated on the assumption that the
crane will collide with the end stops 10 times during the life of the structure,
resulting in this load situation being treated as an ultimate limit state. If
sufficiently reliable measures were in place (e.g. limit switches) to prevent the
crane colliding with the end stops, the choice may be made available to the
designer to reduce the partial load factor or to follow current practice of using
characteristic loads. If this assumption was made and the crane did occasion-
ally collide with the end stops, this would lead to a reduction in reliability of
the support structure. In order to assess the level of reliability for this situation
an investigation was carried out into the effect of retaining the current practice
of using characteristic values for the assessment of the accidental limit state.
Figures 10.5 to 10.8 show the reliability curves for all four classes of crane
with various values of Nextr and ratios of hoistload to total crane weight, using
characteristic loads for the design.
For the practical range of ratios of hoistload to total crane weight, the
reliabilities range from β = 2.68 for the class four crane to β = 3.43 for a
class one crane. These values of reliability are all much lower than the target
reliability of βT = 4.0.
40t crane longitudinal bracing, class = 1, γ = 1

C
3.5
Nextr = 5
3.4 Nextr = 10
Nextr = 25
Nextr = 35
3.3
3.2
3.1
2.9
2.8
2.7
2.6
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(a) Class 1
Figure 10.5: Reliability of bracing elements for accidental limit state, Class 1

C
3.5
Nextr = 5
3.4 Nextr = 10
Nextr = 25
Nextr = 35
3.3
3.2
3.1
2.9
2.8
2.7
2.6
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(a) Class 2
10.2. Code format 249

C
3.5
Nextr = 5
3.4 Nextr = 10
Nextr = 25
Nextr = 35
3.3
3.2
3.1
2.9
2.8
2.7
2.6
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(a) Class 3

C
3.5
Nextr = 5
3.4 Nextr = 10
Nextr = 25
Nextr = 35
3.3
3.2
3.1
2.9
2.8
2.7
2.6
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh)
(a) Class 4
10.2.3 Fatigue
The current partial load factor for fatigue in SABS 0160:1989 [1] is 1.0, i.e.
characteristic loads are used for the fatigue analysis. The code calibration
results showed that when using the fatigue loading method in prEN 1991-3,
a partial load factor of 1.25 applied to the crane wheel loads is required to
obtain the target reliability of βT = 1.5. This would be a deviation from
current practice and could lead to opposition from designers. This could be
dealt with in a similar manner to the horizontal load partial factor in the
ultimate limit state, in that the fatigue partial load factor could be included in
the λ factors which are used to calculate the fatigue damage equivalent load.
The code would then keep the current format with the disadvantage of a loss
of transparency.
Very little information is given in the prEN 1991-3 as to the development
of the λ factors so it could be argued that this fatigue design method is not
transparent and the inclusion of a partial load factor into the λ factors could
therefore not decrease the transparency.
10.3 Further work
10.3.1 Multiple cranes
This code calibration exercise dealt only with loads imposed by a single crane.
Many industrial buildings consist of more than one bay with a crane in each
bay or more than one crane in one bay, therefore the combined actions of more
than one crane are of interest.
Due to the spatial and temporal variability of crane loading, a time depen-
dent approach is recommended for the assessment of the combination of more
than one crane.
A study of the combination of two cranes for column loading has been
carried out by Pasternak et al. [14] where the crane vertical loads on the col-
umn were modelled as intermittent trapezoidal wave renewal processes. An
upcrossing approach was used to determine the mean upcrossing rate func-
tion and the cumulative distribution function of the maximum value of the
combined loading for a given time period.
When considering combinations of intermittent wave renewal processes the
following items need to be determined (Wen [66]):
10.3. Further work 251
1. Load occurrence rate
2. Intensity variation
3. Duration of each pulse
4. Simultaneous occurrence of different loads
The intensity variation of the pulses was modelled by Pasternak et al. [14]
considering two random variables, namely the hoistload and the position of
the crab. The hoistload models given by Köppe [13] were used which have
been shown in Chapter 7 to be unsuitable for hook cranes in South Africa. In
future work on the combination of more than one crane in South Africa the
hoistload models developed in Chapter 7 could be used. Two distributions were
considered by Pasternak et al. [14] for the distribution of the crab position,
viz. a uniform distribution (modelling the crab working evenly across the
crane bridge) and a left triangular distribution (modelling the crab working
predominantly to the left side of the crane bridge). Further crab position
distributions which could be considered are a centre triangular distribution
(modelling the crab working predominantly in the centre of the crane bridge)
and a ‘concentrated distribution’ where the crab is always at the extreme of
its travel as is the case with many ladle cranes.
The load occurrence rate and duration of each pulse would depend on the
speed of the crane, the length of the runway and the characteristics of the
crane operation.
Combination factors for two cranes are given by Pasternak et al. [14], ex-
pressed as a given percentile of the maximum combined load divided by the
sum of the given percentiles for the individual loads.
Further work is required, however, on the combination of more than one
crane for codification due to the fact that the results given by Pasternak et al.
[14] use the hoistload distributions developed by Köppe [13] and that the
results are given for time periods of only up to one year whereas the lifetime
of a crane support structure is usually 25 years.
A further limitation of the work carried out by Pasternak et al. [14] is that
only vertical loads on columns have been considered. The combination of hor-
izontal loads should also be considered for the full assessment of combinations
of cranes.
10.3.2 Probabilistic optimisation

The target reliabilities selected for the code calibration for the three limit
states considered, were based on current practice and recommendations given
by JCSS [36], prEN 1990 [61] and SANS 2394:2004 [64]. The target reliabili-
ties were treated as the minimum allowable value of reliability for all practical
ranges of parameters. This stringent condition could be moderated by car-
rying out a probabilistic cost optimisation exercise to determine the optimal
reliability level taking into account the risks associated with crane support
structures.
In order to carry out the probabilistic optimisation, a risk assessment of
crane support structures would be required to identify possible unfavourable
situations and to determine their consequences. Ultimate limit state type fail-
ures could be considered as well as serviceability limit states and fatigue. One
aspect that could be considered is the frequency of inspection and maintenance
required for fatigue considerations.
The four classes of crane should be considered separately for the probabilis-
tic optimisation because the characteristics of the risks for the four different
classes are likely to differ.
Chapter 11
Conclusions
Electric overhead travelling cranes are used in industrial applications for mov-
ing loads around the industrial area. Cranes often form an integral part of the
industrial process and any down time can have severe financial implications for
the owner. Cranes move over the industrial area lifting heavy loads, therefore
any failure of the crane or the support structure which causes the load or crane
to fall could become a serious safety hazard.
Due to the importance of the smooth running and safety of cranes, the cor-
rect design of the crane support structure is a very important issue when de-
signing an industrial building which will contain cranes. Many problems have
been encountered with crane support structures, particularly fatigue problems.
Codes of practice on loadings on buildings provide load models to calculate
the wheel loads that cranes impose on their support structures. Two aspects of
the crane loading provisions in the South African code of practice on loadings
on buildings, SABS 0160:1989 [1], were identified for investigation. These two
aspects were:
1. Crane load models
A comparison between the crane load models in SABS 0160:1989 [1] and
those in prEN 1991-3 [2], ISO 8686-1:1989 [5], DIN 15018 Part 1 1984
[4] and AS1418.1-1994 [6] indicated that the SABS 0160:1989 crane load
models are over-simplistic.
Updating of the South African crane load models by adoption of the

prEN 1991-3 load models is currently under consideration by the South
African loading code committee, and for this, an assessment of the prEN
253
254 Chapter 11. Conclusions
1991-3 crane load models is required to determine their suitability for

inclusion into the updated South African loading code, proposed SANS
10160.
2. Reliability of crane support structures

Reliability based code calibration has not previously be applied to crane
support structures are there is therefore no reliability basis for the crane
partial load factor in SABS 0160:1989. The load factor which is applied
to the crane loads is the imposed load factor, which has been calibrated
for floor loads in office buildings, of γL = 1.6. In contrast the crane
partial load factor in prEN 1991-3 is the same as the load factor applied
to permanent loads, of 1.35.
The difference between the partial load factors in the two codes and
the lack of a reliability basis for the crane partial load factor in SABS
0160:1989, indicated a need for a reliability assessment of the crane load
models.
The aims of the reliability assessment of the crane load models were
to assess the level of reliability of current practice and, if necessary, to
determine suitable partial load factors to achieve the required level of
reliability.
Both of these issues were dealt with as a code calibration problem, where
the code that was being considered, was the crane loading code. The code
calibration consisted of two parts, firstly calibration to current practice and
secondly the reliability calibration.
Both of these calibration procedures were carried out on specific example
structures. An investigation was carried out into the crane parameters which
affect the loads as calculated by prEN 1991-3 and the range and distribution
of these parameters for cranes in South Africa. The critical parameters were
found to be the crane bridge span and the capacity of the crane.
Three representative cranes were selected to cover the range of the parame-
ters. The full range of capacities was covered but the spans covered only the
most likely range. The effect of shorter or longer spans was taken into account
in the parametric study on the ratio of hoistload to total crane weight, that
was carried out for the reliability calibration. The representative cranes were:
• 5t crane, 19.2 m span

11.1. Calibration to current practice 255
The support structures were also shown to be representative of those found

in practice.
The selection of the representative cranes and the calibration to current
practice has been discussed in Warren et al. [67].
11.1 Calibration to current practice

The aims of the calibration to current practice were:
1. To assess the crane load models from prEN 1991-3 for their suitability
for adoption into proposed SANS 10160.
2. To assess the implications of adopting the crane load models from prEN
1991-3 on the cost of the support structure and the design effort required
for the crane support structure.
11.1.1 Crane load models from prEN 1991-3

The crane load situations that are allowed for in SABS 0160:1989 and prEN
1991-3 were compared and are summarised in Table 11.1.
The crane load models in prEN 1991-3 model in more detail the different
vertical loads, taking into account specific situations for the dynamic effects of
lifting loads, running on rails etc.
A comparison between the horizontal load cases shows that both codes
consider acceleration of the crane, skewing and crane buffer forces but that
the load case for misalignment of the crane wheels or rails is not covered by
prEN 1991-3.
The misalignment load case in SABS 0160:1989 takes into account the
situation where either the rails or wheels are misaligned in a ‘toe-out’ or ‘toe-in’
manner causing horizontal transverse wheel loads, which tend to pull the rails
together or push them apart. The misalignment forces are due to transverse
friction on the rails and can occur when the alignment of the wheels and rails
are both within specified alignment tolerances.
The misalignment load case is one of the critical load cases in SABS
0160:1989 and its inclusion in the SABS 0160:1989 crane loading provisions
Table 11.1: Summary of crane loads considered by the codes

Crane load SABS 0160:1989 prEN 1991-3
Vertical
Generic dynamic amplification X
Hoisting a load off the ground X
Running on rails X
Release of payload X
Test loads X
Horizontal transverse
Acceleration of crane X
Acceleration of crab X
Skewing X X
Misalignment X
Crab buffer forces X
Horizontal longitudinal
Acceleration of crane X X
Crane buffer forces X X
implies that it was deemed to be necessary to consider this load situation.

Therefore, when the crane load models are updated, a misalignment load model
should be included.
An assessment of the prEN 1991-3 crane load models showed that they
were more sophisticated than the SABS 0160:1989 models in that more crane
parameters were taken into account. The comparison between the load effects
resulting from horizontal loads from prEN 1991-3 and SABS 0160:1989 demon-
strated the greater sophistication of the prEN 1991-3 load models by the fact
that the load effect ratios due to horizontal loads depended on the span and
wheel spacing of the crane.
prEN 1991-3 load models also take into account in more detail the dynamic
factors for different load situations (e.g. hoisting a load, acceleration of the
crane), thereby more accurately modelling the mechanism behind the load.
Considering all the above points, it is likely that the prEN 1991-3 crane
load models more accurately model the crane behaviour than the crane load
models from SABS 0160:1989.
11.1.2 Fatigue loading in prEN 1991-3

SABS 0160:1989 does not specify crane loads to be considered for fatigue
load assessment of the crane support structure. The onus is therefore on the
designer to model the crane behaviour, i.e. the load cases to be considered for
fatigue, the range of hoistloads, the crab positions and the movement of the
crane along the runway.
The approach of prEN 1991-3 is to normalise the maximum crane wheel
load resulting in a constant amplitude load for two million cycles which is
then used to calculate the fatigue stresses. The normalisation is carried out by
applying a factor to the maximum crane wheel load depending on the number
of cycles that the crane will perform over its lifetime and the load spectrum of
the crane.
The method of normalising the wheel load was derived from the fatigue
design procedure of the crane itself and the applicability of this method to the
support structure is questionable.
An issue of concern that was identified was the definition of the load spec-
trum. The definition of the load spectrum in prEN 1991-3, considers the
change in wheel load that the crane will experience, from the minimum load
to the wheel load being considered for a particular load being lifted. For fa-
tigue design of the support structure it would be more reasonable to consider
the change in load that the support structure would experience, from zero,
before the crane moves onto the particular girder or column, to the value of
the wheel load, when the crane is on the particular girder or column.
A modified load spectrum definition was presented that considered the
change in load that the support structure would experience rather than the
change in load that the crane would experience. This load spectrum definition,
however, does not cover the case of a continuous girder which could experience
a negative load effect when the crane is on the adjacent span and a positive
load effect when the crane is on the span being considered. This highlights
the fact that it is more difficult to apply this fatigue design method to the
support structure than to the crane because of the variety of support structure
configurations.
The advantage of the fatigue assessment method given in prEN 1991-3 is
that it is simpler to use than a full simulation of the crane behaviour. It
should be conservative though, as with all simplified methods. The degree
of conservatism included in this method will be discussed in Section 11.2 on
reliability of the support structures.

An alternative method of including crane fatigue loading recommendations
in proposed SANS 10160 is to identify the crane load combinations which are
to be considered for the fatigue assessment. The support structure designer is
then responsible for modelling the crane behaviour for the fatigue design.
The crane load combinations given in prEN 1991-3 include combinations
which will not be critical for ultimate limit state but which could be used
for a fatigue assessment. The load combinations to be considered for fatigue
would be those that consist of frequently occurring actions such as hoisting,
acceleration of the crane bridge, acceleration of the crab. Skewing of the crane
in plan is viewed as an infrequently occurring load situation (ENV 1993-6 [35])
and therefore should not be included in the fatigue assessment.
The advantage of this method is that it allows the designer to assess the
effect of each crane load on the fatigue damage and to be able to tailor the
crane behaviour modelling to fit each crane installation.
11.1.3 Implications of adopting crane load models from

prEN 1991-3
Two issues were investigated in assessing the implications of adopting the crane
load models from prEN 1991-3 into proposed SANS 10160: the cost of the
support structure and the design effort required for the design of the support
structure.
11.1.3.1 Cost of the support structure
The costs of the support structure designed considering crane wheel loads from
the models in SABS 0160:1989 and prEN 1991-3 were compared by considering
the load effects in the crane girders, columns and longitudinal bracing of the
three representative cranes. In order to assess the effect of the crane class on
the load effects, the comparison was carried out assuming that the cranes were
either class or hoist class 2, 3 or 4. The class 2 cranes were compared with the
hoist class 2 cranes etc.
The load effects due to vertical crane wheel loads showed that, on average,
the vertical loads due to the prEN 1991-3 load models were 10% smaller than
those due to the SABS 0160:1989 load models. The ratios of the load effects
due to the vertical loads did not vary across the crane classes.
The load effects influenced by horizontal transverse loads ranged from

the prEN 1991-3 load models resulting in loads 56% to 217% of the SABS
0160:1989 loads. The difference depended on the geometry of the crane, specif-
ically the span and the wheel spacing. The loads from prEN 1991-3 were the
smallest when considering the 260t, crane which has a span to wheel spac-
ing ratio of 2.4, and the largest when considering the 40t crane, which has
a span to wheel base ratio of 5.4. The ratios of horizontal transverse loads
(SABS 0160:1989/prEN 1991-3) were different for the various classes of crane,
increasing with increasing class. The horizontal loads from prEN 1991-3 do
not change with the class of crane as in SABS 0160:1989, so the variation in
ratios was due to the change in magnitude of the SABS 0160:1989 loads.
The buffer forces calculated from prEN 1991-3 resulted in forces with mag-
nitudes in between those obtained from the simple method and the more de-
tailed method in SABS 0160:1989. The simple method gives the largest forces
as would be expected from a simplified method. Both prEN 1991-3 and the
more detailed method in SABS 0160:1989 consider the resilience of the buffers
but prEN 1991-3 results in larger loads because it includes the weight of the
hoistload whereas SABS 0160:1989 does not.
The effect on the cost of the support structure of adopting the prEN 1991-
3 crane load models into proposed SANS 10160 would depend on the class
and geometry of the crane. The horizontal forces in SABS 0160:1989 depend
on the class of the crane and those in prEN 1991-3 do not. The horizontal
forces in prEN 1991-3 take into account the geometry of the crane, the span
and wheel spacing, unlike those in SABS 0160:1989 which take into account
only the crane self weight, hoistload and number of wheels. For lower class
cranes with a large span to wheel spacing ratio, the prEN 1991-3 load models
give larger horizontal forces than the SABS 0160:1989 load models. For higher
class cranes and cranes with small span to wheel spacing ratio, the prEN 1991-
3 load models result in smaller horizontal forces than the SABS 0160:1989 load
models. The vertical loads from prEN 1991-3 would on average result in lighter
girders.
Considering the longitudinal bracing, design using the prEN 1991-3 load
models would result in heavier bracing members if the detailed method in
SABS 0160:1989 had previously been used, or lighter bracing if the simplified
method had previously been used.
No general trend can be identified for the change in the cost of the support
structure by design according to the prEN 1991-3 crane load models rather
than the SABS 0160:1989 crane load models due to the difference in the ap-
proach taken to calculate the horizontal transverse loads.
11.1.3.2 Design effort
The design effort required for the design of the support structure was consid-
ered by dividing the design process into the calculation of the crane loads and
the subsequent design of the support structure.
It was found that slightly more crane information is required for the cal-
culation of the crane loads according to prEN 1991-3 than SABS 0160:1989.
Calculating the crane loads according to prEN 1991-3 also required more pages
of calculations than according to SABS 0160:1989.
Two critical load combinations were identified for each code, therefore the
same design effort would be required for the subsequent design of the support
structure, which in itself entails vastly more design effort than the calculation
of the loads.
Slightly more design effort is required for the prEN 1991-3 crane load
models than the SABS 0160:1989 models, however, with the use of computer
spreadsheets this would probably be a once-off effort and is therefore insignif-
icant.
11.2 Reliability calibration
The reliability code calibration was carried out on individual elements of the
support structures of the representative cranes. The elements that were con-
sidered were those that were influenced by crane loads, i.e. the crane girders,
crane columns, roof columns, roof truss members and longitudinal runway
bracing.
The limit states considered for the reliability assessment of the crane sup-
port structures were ultimate limit state, accidental limit state and fatigue.
The calibration did not follow the convention for general code calibration
of separating loads and resistances but, because specific structures were being
considered, the limit states equations were developed including both loads and
resistances.
11.2. Reliability calibration 261
11.2.1 Stochastic modelling

The stochastic models for most of the basic variables were obtained from lit-
erature. There was no literature available on the basic variables relating to
crane loading so these were developed for this investigation.
The crane wheel loads were modelled by considering the crane self weight
and hoistload as basic variables and including a modelling uncertainty factor
to take into account the uncertainty in the calculation of the wheel loads.
11.2.1.1 Modelling uncertainties
There was no literature available on modelling uncertainties in the calculation

of the crane loads; therefore the statistical parameters for these modelling
uncertainties were developed for this investigation
Three crane load modelling uncertainties were included to take into account
the various crane load models.
1. Uncertainty in the calculation of the vertical wheel loads
2. Uncertainty in the calculation of the horizontal transverse wheel loads
3. Uncertainty in the calculation of the horizontal longitudinal loads due to

the crane colliding with the end stops.
The modelling uncertainty for the vertical wheel loads was investigated by
carrying out a numerical assessment of various crane configurations to deter-
mine the increase in vertical loads due to one wheel losing contact with the rail.
The modelling uncertainty for the horizontal longitudinal loads was assessed
from experimental data available on end stop forces. However, there was no
information available on which to base the parameters of the modelling uncer-
tainty for the horizontal transverse loads, so the parameters were estimated
using judgement.
The sensitivity factors from the reliability analysis and the sensitivity study
carried out on the partial load factors showed that the modelling uncertainty
for the horizontal transverse loads was a critical variable. In the light of this,
more information should be gathered to improve the parameters for this mod-
elling uncertainty. Difficulties arise in that there is no available experimental
or numerical modelling data. One option would be to conduct an expert sur-
vey but here too, difficulties are encountered. The experts in crane support
structure design in South Africa are currently unfamiliar with the prEN 1991-3
crane load models and horizontal crane wheel loads are inherently difficult to
assess because of the lack of observational data, unlike roof or floor imposed
loads where the loads are due to the weight of visible objects.
A solution to the problem of determining the parameters of the modelling
uncertainty in the horizontal transverse loads is to obtain data on actual hori-
zontal loads that cranes impose on their support structures. This data could be
obtained from laboratory experiments or numerical modelling of crane struc-
tures. In order to be comprehensive, a range of crane parameters should be
considered.
11.2.1.2 Stochastic modelling of the hoistload
There were no crane load models, suitable for the code calibration, presented
in the literature. Although Pasternak et al. [14] and Köppe [13] have devel-
oped crane load models they are not suitable for the code calibration of crane
loads for use in South Africa. Therefore stochastic models for crane loads
were developed for this investigation in the form of statistical models for the
hoistloads lifted by cranes.
The stochastic models for the hoistload that were developed for the code
calibration were both arbitrary point-in-time and extreme value type distribu-
tions. The stochastic modelling of the hoistload has been discussed in Warren
et al. [68].
For the purpose of this study, cranes were divided into four classes accord-
ing to a description of the loads that are lifted and hoistload distributions
developed for each class.
‘One cycle’ distributions were developed which model the probability dis-
tribution of the load lifted by a crane for one crane cycle, i.e. one load lifted.
This distribution can be considered as the frequency distribution of the loads
that a crane lifts over its lifetime. The ‘one cycle’ distributions were developed
from data and descriptions of loads lifted obtained from crane operators as well
as the descriptions of the class of crane.
Beta distributions were used for the ‘one cycle’ hoistload distributions be-
cause they have lower and upper bounds and because of the ease of fitting the
required shape.
The upper bound of the hoistload distributions was set higher than the
overload limit switch level of 110% overload because, according to the crane
operators, the overload limit switches do get turned off to lift larger loads. The
level chosen for the upper limit was 125% overload which is the value of the
test load that the crane lifts after installation. This value can be considered
to be an extreme upper bound for the upper limit of the distribution.
The extreme distributions were developed on the principle that reliability is
always expressed in terms of a time period and the extreme load distributions
model the largest load that is likely to occur during that time period. When
considering cranes, a given time period can be related to a number of crane
cycles and the extreme hoistload distribution can be considered to be the
probability distribution of the largest load that the crane will lift in a given
number of cycles, termed Nextr .
The extreme hoistload distributions were developed from the ‘one cycle’
distributions using a simulation technique. Curves were fitted to the observed
trends in the mean values and standard deviations which allow interpolation
and extrapolation to obtain the parameters of the extreme hoistload distribu-
tion for Nextr ≤ 2 × 106 cycles.
Extreme hoistload distributions can be obtained for various limit states
and load combinations by selecting the appropriate number of cycles.
The hoistload distributions that were developed are not dependent on the
specific representative cranes but are general distributions which could be used
for future reliability assessment of crane support structures, including time
variant assessments of one crane or combinations of more than one crane.
11.2.2 Code calibration
The reliability assessment for the code calibration was carried out in two steps.
The first step was the economic design of the member being considered where
the design was carried out so that the member exactly complied with the code
requirements. The second step was to carry out the reliability analysis of the
economically designed member. This procedure ensured that the reliability of
the code itself was being assessed.
11.2.2.1 Ultimate limit state
The objective for the code calibration was the achievement of a target relia-
bility. The target reliability selected for the ultimate limit state was the value
used for the calibration of the South African codes of βT = 3.0, with the
constraint that the reliability obtained must be greater than or equal to this
target reliability.
Four partial load factor formats were considered for the ultimate limit state.
The first and simplest code format considered was the current practice for both
SABS 0160:1989 and prEN 1991-3. The remaining three code formats were
more complex formats which were included for this investigation to take into
account the different characteristics of the crane self weight and hoistload, and
the vertical and horizontal forces. The effect of the different code formats on
the consistency of the reliability over a range of parameters was investigated.
The four code formats are discussed in detail in Section 8.3 on page 180
and are outlined briefly below:
1. One partial load factor, γC , applied to the calculated characteristic wheel

loads.
2. Two partial load factors, one applied to the crane self weight, γCsw , and
one applied to the hoistload, γCh .

design horizontal loads, γH .

design horizontal loads, γH .
As with all design codes, increased sophistication of the code format leads
to greater consistency in the reliability which must be weighed up against the
increased complexity of the code.
The calibration of the ultimate limit state was carried out in two steps:
firstly, calibration of the crane partial load factors considering a crane load
only case and secondly, calibration of the combination factors considering com-
binations of crane loads with other time varying loads.
The calibration was carried out on the critical structural element which
had the lowest reliability which was found to be the 5t crane column.
Parametric studies were carried out to determine the values of critical pa-
rameters which resulted in the lowest reliability, these critical values were used
for the calibration. The parametric studies that were carried out, and the
critical values of these parameters, were:
1. Class of crane: the lowest reliability was obtained with a class one crane
2. Number of cycles considered for the extreme value distribution: Nextr =

106 cycles. This can be considered to be a reasonable upper limit of the
number of cycles that a crane is likely to perform over its lifetime.
3. Ratio between hoistload and total crane weight: r = 0.85. This ratio
takes into account shorter or longer crane bridge spans than the spans
of the representative cranes.
It was also found that the higher the ratio of horizontal load effect to
vertical load effect, the lower the reliability.
The calibrated partial load factors that were obtained for the crane only
load case are given in Table 11.2. The partial load factors were calibrated
using the extreme hoistload distributions that were developed for the code
calibration, with the assumption that the crane could be overloaded to 125%
SWL, the modelling uncertainties discussed above and considering the crab
always at the extreme of its travel.

1 1.91
2 1.62 1.96
3 1.59 1.26
4 1.35 1.62 1.27
Verification of the calibrated partial load factors was carried out to ensure
that the reliability of all the elements was above the target reliability and to
assess the conservatism over a range of parameters for the four code formats.
It was found that code format 4 resulted in the most consistent level of
reliability and hence the minimum conservatism. This would thus be the code
format that best meets the code calibration objective of achieving a consistent
target reliability. As mentioned above, consistent reliability and code simplic-
ity are two mutually opposing goals and one has to be traded off against the
other when it comes to deciding on the code format for the design code.
Two combinations of crane loads with other time varying loads were con-
sidered. The load combination equations are given below:
1. Crane and wind loads
a) Crane dominant: γP (Permanent) + γC (Crane) + ψW C 1.3(Wind)

b) Wind dominant: γP (Permanent) + ψCW γC (Crane) + 1.3(Wind)
2. Crane and roof imposed loads
a) Crane dominant: γP (Permanent) + γC (Crane) + ψRC 1.6(Roof im-

posed)
b) Roof dominant: γP (Permanent) + ψCR γC (Crane) + 1.6(Roof im-
posed)
Where:
γC – represents conceptually the factoring of the crane loads by whichever

code format is being considered
The crane partial load factors that were calibrated for the crane only case
were used in the calibration of the combination factors.
The combination factors for the combination of crane and wind loads were
selected so as to result in constant reliability over the pertinent range of ratios
of crane load to crane and wind load. The appropriate value of the combi-
nation factor applied to the crane loads was found to be ψCW = 0.65 and
the appropriate value of the combination factor applied to the wind loads was
found to be ψW C = 0.
The combination of crane and roof imposed loads resulted in very high
reliability levels for the roof dominant case when designing the element for roof
loads only. Constant levels of reliability equal to the reliability of the crane
only load case, were obtained for the crane dominant case when designing for
crane loads only. The combination of crane loads and roof imposed loads is
therefore not a rational one.
The load combinations that resulted from this investigation are thus:
1. γP (Permanent) + γC (Crane)
2. γP (Permanent) + 0.65γC (Crane) + 1.3(Wind)
3. γP (Permanent) + 1.6(Roof imposed)

11.2.2.2 Accidental limit state
The accidental limit state that was considered was the crane running into
the end stops causing horizontal longitudinal forces which are resisted by the
longitudinal bracing. This load situation is classified as an accidental load in
prEN 1991-3, though it was estimated to occur 10 times during the lifetime of
the crane. This load situation was therefore considered as an ultimate limit
state because of its relatively high probability of occurrence.
The target reliability selected was βT = 4.0 because of the low cost of the
longitudinal bracing elements and end stops in proportion to the consequences
of failure.
The code formats considered for the accidental limit state were code for-
mats one and two as given for the ultimate limit state.
and the critical values of the parameters were:
1. Class of crane: the lowest reliability was obtained with a class 4 crane
2. Number of cycles considered for the extreme hoistload distribution: Nextr

= 10 cycles
The calibrated partial load factors were:
• Code format 1:
– γC = 1.196
• Code format 2:
– γCsw = 1.344
– γCh = 1.448
These partial load factors are all greater than 1.0 indicating that the cur-
rent practice of using the characteristic loads for the accidental limit state is
unconservative.
The variation of the reliability with the ratio of hoistload to total crane
weight was small, therefore code format 1 does not include significant conser-
vatism in the lower ratios of hoistload to total crane weight. Code format 1
is a simpler format than code format 2 and, since it does not include signifi-
cant over conservatism, would be the most suitable code format the accidental
loads.
The calibration of the accidental load situation of the crane running into
the end stops was carried out on the assumption that it would occur 10 times
during the lifetime of the crane and was therefore considered as an ultimate
limit state. In the event of the crane having sufficiently reliable systems to
stop the crane before collision with the end stops, the choice could be left to
the designer to reduce the partial load factor, or follow the current practice of
using characteristic loads.
11.2.2.3 Fatigue
Fatigue was considered as a serviceability limit state with the objective that
the average value of reliability should meet a target reliability of βT = 1.5
because of the general inspectability of fatigue effects.
The code format that was considered for fatigue was code format one, as
given for the ultimate limit state, consisting of applying one partial load factor,
γF , to the characteristic crane wheel loads.
The fatigue reliability was assessed by determining the bounds of reliability
for a given load spectrum class. The parametric studies that were carried out
for fatigue were:
1. Considering a single slope or double slope S-N curve: the trend of the
results did not differ significantly between the two curves so for simplicity,
a single slope S-N curve was considered.
2. The load spectrum class considered: the load spectrum class was taken as
Q5 which was representative of the reliability for all the classes without
the initial conservative area included in the other load spectrum classes.
The fatigue reliability was assessed for the prEN 1991-3 method of nor-
malising the maximum wheel load and the results used to determine a partial
load factor for the method of simulating the crane behaviour. The partial load
factors are given in Table 11.3.
The partial load factor for fatigue assessment using the method given in
prEN 1991-3 is less than that for the simulation of the crane behaviour. This
11.3. Recommendations 269
Table 11.3: Partial load factors for fatigue

Fatigue assessment method Partial load factor, γF
prEN 1991-3 1.25
Simulating crane behaviour 1.34
indicates that the method given in prEN 1991-3 is slightly conservative, but
not sufficiently so to be able to use the characteristic loads for the design.
Current practice is to use the characteristic loads for the fatigue assess-
ment and the inclusion of a fatigue partial load factor in the code could meet
with some resistance from designers. If the option to include only the fa-
tigue assessment method given by prEN 1991-3 was taken, the fatigue partial
load factor could be included in the factors used for the normalisation of the
maximum wheel loads. This could be seen to be reducing the rationality and
transparency of the code, though the fatigue method given by prEN 1991-3 is
not completely transparent.
If it was decided to include both the fatigue assessment method from prEN
1991-3 and the method of simulating the crane behaviour; including two fa-
tigue partial load factors, one for each method, would be undesirable from
the perspective of code complexity. A method of simplifying the code format
would be to specify only the fatigue partial load factor calibrated for simula-
tion of the crane behaviour of γF = 1.34. This would result in designs using
the method from prEN 1991-3 being conservative but this is appropriate for a
simplified method.
11.3 Recommendations
The prEN 1991-3 crane load models have been shown to be more sophisticated
and thus more likely to accurately model the crane wheel loads, than the crane
load models in SABS 0160:1989. The crane load models from prEN 1991-3
should thus be included in proposed SANS 10160, with the addition of a load
model for misalignment of the crane wheels or rails.
Of the four code formats considered for the crane partial load factors for
the ultimate limit state, code format 4 resulted in the most consistent level of
reliability and should be adopted in the code. The different partial load factors
for the crane self weight, hoistload and horizontal loads, model the different
sources of uncertainty in the crane wheel loads and this is therefore the most
transparent code format.
Code format 4 would result in an increase in code complexity from the
current practice of applying one partial load factor to the characteristic wheel
loads. The advantages of a more consistent reliability with less conservatism,
resulting in more economic structures, outweigh the increased complexity of
the code. With the use of computers and spreadsheets for load calculations,
the increased complexity of the code format could be easily incorporated into
the design process.
The issue of code complexity versus consistent reliability requires consid-
eration though, and a less complex code format may be deemed to be more
suitable for a code of practice. One method of simplifying code formats 3 or
4 for inclusion into the design code would be to include the additional partial
load factor for the horizontal loads into the horizontal load models.
The calibrated crane partial load factors are reproduced in Table 11.4.

1 1.91
2 1.62 1.96
3 1.59 1.26
4 1.35 1.62 1.27
These partial load factors are all larger than the factors in SABS 0160:1989
and prEN 1991-3 indicating that the current level of reliability for crane sup-
port structures is inadequate.
The calibration of the combination factors for combinations of crane loads
with wind loads or roof imposed loads resulted in only three rational load
combinations that should be included in the design code:
1. γP (Permanent) + γC (Crane)
2. γP (Permanent) + 0.65γC (Crane) + 1.3(Wind)
3. γP (Permanent) + 1.6(Roof imposed)
A partial load factor should be included in the code for the accidental load
situation of the crane colliding with the end stops. Because the structural
11.4. Further work 271
elements which resist the accidental loads have a low cost in proportion to
the total cost of the structure and because of the serious safety and financial
implications of failure, partial load factors greater than 1.0 should be included
in proposed SANS 10160 for accidental loads. In the event that a crane has
reliable measures to prevent collision with the end stops, an option should be
available for the designer to reduce the partial load factor or follow current
practice of using the characteristic loads for the design of these elements.
In order to keep the current code format of using characteristic loads for
the assessment of the accidental situations, the partial load factors could be
included in the dynamic factor φ7 .
Both the fatigue methods considered, the method in prEN 1991-3 of nor-
malising the wheel load and the method of recommending load cases and for
the support structure designer to model the crane behaviour, should be in-
cluded into proposed SANS 10160.
The current method of using characteristic loads for the fatigue assessment
was shown to be unconservative. The fatigue partial load factors that were
obtained are reproduced in Table 11.5.
Table 11.5: Partial load factors for fatigue

Fatigue assessment method Partial load factor, γF
prEN 1991-3 1.25
Simulating crane behaviour 1.34
If both methods are to be included in the code, the recommendation would

be to include only one fatigue partial load factor of γF = 1.34. This would
result in the fatigue assessment method from prEN 1991-3 being conservative,
which is appropriate for a simplified method.
11.4 Further work

No literature was available on the modelling uncertainties for the calculation
of the crane wheel loads therefore these models were developed for use in
the code calibration. The modelling uncertainty for the calculation of the
horizontal transverse loads was the variable for which the least information
was available and the statistical parameters were therefore estimated. This
modelling uncertainty was identified by the sensitivity factors and sensitivity
studies to be a critical variable requiring further investigation. Experimental

or numerical investigations into the horizontal wheel loads that cranes impose
on the support structures would provide the necessary information to be able
to derive more accurate modelling uncertainties. The reliability procedures
developed here can then be used to update and improve the partial factors
derived in this study. Since the results presented here necessarily have to
be on the conservative side to reflect the uncertainty in the true modelling
uncertainty, the validated modelling factor variables are expected to result in
more economical partial load factors.
In the code calibration, only single cranes were considered. Many industrial
buildings consist of multiple bays with one or more cranes in each bay. An
extension of the code calibration could be the calibration of combinations of
more than one crane, using the stochastic models of hoistloads presented here.
A risk assessment and probabilistic cost optimisation could be carried out
to determine the optimal levels of reliability for ultimate limit state, accidental
limit state and fatigue.
References
[1] SABS 0160:1989: Code of Practice, The general procedures and loadings to be
adopted in the design of buildings. The Council of the South African Bureau of
Standards, 1990.
[2] prEN 1991-3: Eurocode 1991: Actions on structures, Part 3: Actions induced by
cranes and other machinery. prEN 1991-3. European Committee for Standardi-
sation, CEN TC250/SC1, 2002.
[3] SANS 10160: Working document on SANS 10160: Basis of structural design and
actions for buildings and industrial structures, Section 10: Actions induced by
cranes and machinery. Personal communication with Working Group 61M of
STANSA.
[4] DIN 15018 Part 1 1984: Deutsche Norm: Cranes, steel structures, verification
and analysis. Beuth Verlag GmbH, Berlin, nov 1984.
[5] ISO 8686-1:1989: International Standard: Cranes - Design principles for loads
and load combinations, Part 1: General. International Organisation for Stan-
dardization, Geneva, Switzerland, nov 1989.
[6] AS1418.1-1994: Cranes (including hoists and winches) Part 1: General require-
ments. Standards Australia, 1994.
[7] ASCE 7-98: Minimum design loads for buildings and other structures. American
Society of Civil Engineers, jan 1998.
[8] Faber, M.H. and Sørensen, J.D.: Reliability based code calibration. Joint Com-
mittee on Structural Safety, Workshop on Code Calibration, Zurich, mar 2002.
[9] Kemp, A.R., Milford, R.V. and Laurie, J.A.P.: Proposals for a comprehensive
limit states formulation for South African structural codes. The Civil Engineer
in South Africa, vol. 29, no. 9, pp. 351–360, sep 1987.
[10] SABS 0100-1:1992: Code of Practice, The structural use of concrete, Part 1:
Design. The Council of the South African Bureau of Standards, 1992.
273
274 References
[11] SANS 10162-1:2005: Code of Practice, The structural use of steel, Part 1: Limit-
states design of hot-rolled steelwork. The Council of the South African Bureau
of Standards, 2005.
[12] Hansen, P.F. and Sørensen, J.D.: Reliability-based code calibration of partial
safety factors. Joint Committee on Structural Safety, Workshop on Code Cali-
bration, Zurich, mar 2002.
[13] Köppe, U.: Nutzlastkollektive von kranen. Hebezeuge und Fördermittel, vol. 21,
no. 2, pp. 36–39, 1981.
[14] Pasternak, H., Rozmarynowski, B. and Wen, Y.-K.: Crane load modelling. Struc-
tural Safety, vol. 17, pp. 205–224, 1996.
[15] Nowak, A.S. and Collins, K.R.: Reliability of Structures. McGraw-Hill, 2000.
[16] Crane Aid Morris Cranes: Personal communication. jul 2005.
[17] Verlinde Cranes: Personal communication. jul 2005.
[18] Demag Cranes and Components: Personal communication. mar 2003.
[19] Condra Cranes: Personal communication. jul 2005.
[20] Ricker, D.T.: Tips for avoiding crane runway problems. American Institute of
Steel Construction Engineering Journal, pp. 181–205, Fourth quarter 1982.
[21] Lobov, N.A.: Calculation of dynamic loads on an overhead travelling crane.

Russian Engineering Journal, vol. 56, no. 1, pp. 44–48, 1976.
[22] Lobov, N.A.: Loads of an overhead travelling crane caused by transverse and
rotatory motions of the bridge girder. Soviet Engineering Research, vol. 62, no. 6,
pp. 31–35, 1982.
[23] Lobov, N.A.: Overhead travelling crane loads when track-wheel flanges contact
the rails. Soviet Engineering Research, vol. 64, no. 7, pp. 22–26, 1984.
[24] Lobov, N.A.: Loads on an overhead travelling crane when it moves with a con-
stant skew setting of the girder. Soviet Engineering Research, vol. 66, no. 12,
pp. 13–17, 1986.
[25] Spitsyna, D.N. and Anoskin, I.V.: Investigation into the dynamics of the metal
structures of foundry cranes when accelerating. Soviet Engineering Research,
vol. 66, no. 5, pp. 25–30, 1986.
References 275
[26] Dunaiski, P.E., Barnard, H., Krige, G. and Mackenzie, R.: Review of provision of
loads to structures supporting overhead travelling cranes. In: Zingoni, A. (ed.),
Proceedings of the International Conference on Structural Engineering Mechanics
and Computation, pp. 1321–1328. Elsevier, apr 2001.
[27] EN 13001-2: Crane safety - General design - Part 2: Load effects. European
Committee for Standardisation, dec 2004.
[28] Lynch, L.: Keeping out of trouble. Hoist, feb 2001.
[29] SAISC: South African Structural Steelwork Specification for Construction. 1st
edn. The Southern African Institute of Steel Construction, Johannesburg, 2000.
[30] Rowswell, J.C.: Crane Runway Systems. Master’s thesis, Department of Civil
Engineering, University of Toronto, aug 1987.
[31] Prokon: Structural analysis computer program, version W2.1.09. Prokon Soft-
ware Consultant Ltd, dec 2004.
[32] Ellingwood, B., Galambos, T.V., MacGregor, J.G. and Cornell, C.A.: Develop-
ment of a probability based load criterion for American National Standard A58.
NBS Special Publication 577, National Bureau of Standards, US Department of
Commerce, Washington DC, jun 1980.
[33] Kwan, K.H. and Liauw, T.C.: Computer aided design of reinforced concrete
members subjected to axial compression and biaxial bending. The Structural
Engineer, vol. 63B, no. 2, pp. 34–40, jun 1985.
[34] Nawy, E.G.: Reinforced Concrete, a Fundamental Approach. 3rd edn. Prentice
Hall, New Jersey, 1995.
[35] ENV 1993-6: Eurocode 1993: Design of steel structures, Part 6: Crane support-
ing structures. Draft ENV 1993-6. European Committee for Standardisation,
CEN TC250/SC3, 1999.
[36] JCSS: Joint committee on structural safety - Probabilistic model code. JCSS
internet publication: http://www.jcss.ethz.ch/, 2001.
[37] Galambos, T.V. and Ravindra, M.K.: Properties of steel for use in LRFD. ASCE
Journal of the Structural Division, vol. 104, no. ST9, pp. 1459–1468, sep 1978.
[38] Bjorhovde, R., Galambos, T.V. and Ravindra, M.K.: LRFD criteria for steel
beam-columns. ASCE Journal of the Structural Division, vol. 104, no. ST9, pp.
1371–1387, sep 1978.
276 References
[39] Ellingwood, B. and Reinhold, T.A.: Reliability analysis of steel beam columns.
ASCE Journal of the Structural Division, vol. 106, no. ST12, pp. 2560–2564, dec
1980.
[40] Cooper, P.B., Galambos, T.V. and Ravindra, M.K.: LRFD criteria for plate
girders. ASCE Journal of the Structural Division, vol. 104, no. ST9, pp. 1389–
1407, sep 1978.
[41] Ravindra, M.K. and Galambos, T.V.: Load and resistance factor design for steel.
ASCE Journal of the Structural Division, vol. 104, no. ST9, pp. 1337–1353, sep
1978.
[42] MacGregor, J.G., Mirza, S.A. and Ellingwood, B.: Statistical analysis of resis-
tance of reinforced and prestressed concrete members. ACI Journal Proceedings,
vol. 80, pp. 167–176, may-jun 1983.
[43] Mirza, S.A., Hatzinikolas, M. and MacGregor, J.G.: Statistical descriptions of

strength of concrete. ASCE Journal of the Structural Division, vol. 105, no. ST6,
pp. 1021–1037, jun 1979.
[44] Grant, L.H., Mirza, S.A. and MacGregor, J.G.: Monte carlo study of strength
of concrete columns. American Concrete Institute Journal, vol. 75, no. 8, pp.
348–358, aug 1978.
[45] Ellingwood, B.: Statistical analysis of RC beam-column interaction. ASCE Jour-

nal of the Structural Division, vol. 103, no. ST7, pp. 1377–1388, jul 1977.
[46] Mirza, S.A. and MacGregor, J.G.: Variability of mechanical properties of rein-
forcing bars. ASCE Journal of the Structural Division, vol. 105, no. ST5, pp.
921–937, may 1979.
[47] Mirza, S.A. and MacGregor, J.G.: Variations in dimensions of reinforced concrete
members. ASCE Journal of the Structural Division, vol. 105, no. ST4, pp. 751–
766, apr 1979.
[48] Fisher, J.W., Galambos, T.V., Kulak, G.L. and Ravindra, M.K.: Load and
resistance factor design criteria for connectors. ASCE Journal of Structural
Engineering, vol. 104, no. ST9, pp. 1427–1441, sep 1978.
[49] Ruiz, S.E. and Aguilar, J.C.: Reliability of short and slender reinforced-concrete
columns. ASCE Journal of Structural Engineering, vol. 120, no. 6, pp. 1850–
1865, jun 1994.
[50] SABS 0100-2:1992: Code of Practice, The structural use of concrete, Part 2:
Materials and execution of work. The Council of the South African Bureau of
Standards, 1992.
References 277
[51] Lu, R., Luo, Y. and Conte, J.P.: Reliability evaluation of reinforced concrete
beams. Structural Safety, vol. 14, no. 4, pp. 277–298, jul 1994.
[52] de Villiers, P.J.: Imposed Loads for Inaccessible Roofs of Light Industrial Steel
Buildings. Master’s thesis, Civil Engineering, University of Stellenbosch, apr
2003.
[53] Ter Haar, T.R. and Retief, J.V.: Development of a methodology for structural
code calibration. ISI Report R00/1, University of Stellenbosch, dec 2000.
[54] Kohlhaas, S.: Impact forces on end stops for overhead travelling crane support
structures. Tech. Rep., University of Stellenbosch, dec 2004.
[55] Holický, M.: Conventional probabilistic models of basic variables. Tech. Rep.,
Klokner Institute, Czech Technical University in Prague, jun 2002.
[56] Kotoguchi, H., Leonard, J.W. and Shiomi, H.: Statistical evaluation of steel
beam-column resistance. Engineering Structures, vol. 7, no. 2, pp. 143–147, apr
1985.
[57] Rojiani, K.B. and Woeste, F.E.: A probabilistic analysis of steel beam-columns.
Engineering Structures, vol. 4, no. 4, pp. 233–241, oct 1982.
[58] Personal communication with The Southern African Institute of Steel Construc-
tion: Commentary on SANS 10162: Part 1: Limit-States Design of hot-rolled
steelwork. 2005.
[59] Maddox, S.J.: Fatigue strength of welded structures. 2nd edn. Abington Pub-
lishing, 1991.
[60] Van der Walt, S.J.: Reliability based determination of load models for EOTC
support structures. Current Masters study, Civil Engineering, University of
Stellenbosch.
[61] prEN 1990: Eurocode - Basis of structural design: Draft prEN 1990. European
Committee for Standardisation, jul 2001.
[62] Milford, R.V.: Target safety and SABS 0160 load factors. The Civil Engineer in
South Africa, vol. 30, no. 10, pp. 475–481, oct 1988.
[63] Gulvanessian, H., Calgaro, J.-A. and Holický, M.: Designer’s guide to EN 1990,
Eurocode: Basis of structural design. Thomas Telford Publishing, London, 2002.
[64] SANS 2394:2004: General principles on reliability for structures. The Council of
the South African Bureau of Standards, 2004.
278 References
[65] Melchers, R.E.: Structural Reliability Analysis and Prediction. 2nd edn. John
Wiley and Sons, New York, 1999.
[66] Wen, Y.-K.: Statistical combination of extreme loads. ASCE Journal of the
Structural Division, vol. 103, no. ST5, pp. 1079–1093, may 1977.
[67] Warren, J.S., Dunaiski, P.E. and Retief, J.V.: A comparison between crane
induced load effects from SABS and Eurocode. In: Proceedings of the Second In-
ternational Conference on Structural Engineering, Mechanics and Computation.
Balkema, jul 2004.
[68] Warren, J.S., Retief, J.V. and Dunaiski, P.E.: Reliability models of overhead
traveling crane loading for code calibration. In: Proceedings of the Ninth Inter-
national Conference on Structural Safety and Reliability. Millpress, jun 2005.
List of Figures
2.1 Main components of an EOT bridge crane . . . . . . . . . . . . . . 12

2.2 Crane support structure . . . . . . . . . . . . . . . . . . . . . . . . 15
(a) Crane bay . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
(b) Crane runway . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 Dynamic factor φ2 values for four design codes . . . . . . . . . . . 24
2.4 Dynamic factor φ2 values for four design codes for 0 ≤ vh ≤ 30 m/min 24
(a) Hoist Class 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 24
(b) Hoist Class 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 24
(c) Hoist Class 3 . . . . . . . . . . . . . . . . . . . . . . . . . . 24
(d) Hoist Class 4 . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.5 Wheel loads due to acceleration of crane bridge . . . . . . . . . . . 29
2.6 Configuration of skewing loads . . . . . . . . . . . . . . . . . . . . 33
2.7 Configuration of skewing forces for different wheel combinations . . 36
(a) Continuous fixed/fixed . . . . . . . . . . . . . . . . . . . . . 36
(b) Continuous fixed/movable . . . . . . . . . . . . . . . . . . . 36
(c) Independent fixed/fixed . . . . . . . . . . . . . . . . . . . . 36
(d) Independent fixed/movable . . . . . . . . . . . . . . . . . . 36
2.8 Buffer force-displacement characteristic curve . . . . . . . . . . . . 39
3.1 Histogram of crane bridge spans . . . . . . . . . . . . . . . . . . . 67

3.2 Histogram of crane capacities . . . . . . . . . . . . . . . . . . . . . 67
3.3 Scatter plot of cranes from database . . . . . . . . . . . . . . . . . 68
3.4 Building housing the 5t crane . . . . . . . . . . . . . . . . . . . . . 73
3.5 Cross section of 5t crane girder . . . . . . . . . . . . . . . . . . . . 73
3.6 Bay considered for 5t crane . . . . . . . . . . . . . . . . . . . . . . 73
3.7 Building housing the 40t crane . . . . . . . . . . . . . . . . . . . . 76
3.8 40t crane girder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
279
280 List of Figures
(a) Cross section of girder . . . . . . . . . . . . . . . . . . . . . 76

(b) Stiffener dimensions . . . . . . . . . . . . . . . . . . . . . . 76
3.9 Crane runway longitudinal bracing for the 40t crane . . . . . . . . 76
3.10 Building housing the 260t crane . . . . . . . . . . . . . . . . . . . . 77
3.11 Cross section of the 260t crane girder . . . . . . . . . . . . . . . . . 79
3.12 260t crane auxiliary girder . . . . . . . . . . . . . . . . . . . . . . . 79
4.1 SABS 0160:1989 load combination 1 . . . . . . . . . . . . . . . . . 86

4.4 SABS 0160:1989 load combination 4a . . . . . . . . . . . . . . . . . 87
4.5 SABS 0160:1989 load combination 4b . . . . . . . . . . . . . . . . 87
4.6 prEN 1991-3 load combination 1 . . . . . . . . . . . . . . . . . . . 88
4.14 SABS 0160:1989 critical load combination 2 . . . . . . . . . . . . . 92
4.15 SABS 0160:1989 critical load combination 3 . . . . . . . . . . . . . 92
4.16 prEN 1991-3 critical load combination 1 . . . . . . . . . . . . . . . 92
4.18 SABS 0160:1989 critical load combination 4a . . . . . . . . . . . . 92
4.19 SABS 0160:1989 critical load combination 4b . . . . . . . . . . . . 93
4.21 Crane girder axes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.1 Load Combination 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 106

5.2 Load Combination 5 . . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.3 Stress history for 40t crane girder at bottom of stiffener . . . . . . 119
5.4 Stress history for 40t crane girder at top flange to web weld . . . . 121
5.5 Stress history for 5t crane at corbel to column welded connection . 122
7.1 Class 1 crane ‘one cycle’ distribution . . . . . . . . . . . . . . . . . 163

(a) Histogram of hoistloads . . . . . . . . . . . . . . . . . . . . 163
List of Figures 281
(b) Beta distribution . . . . . . . . . . . . . . . . . . . . . . . . 163

7.4 ‘One cycle’ distributions for all crane classes . . . . . . . . . . . . . 165
(a) Classes 1 to 3 . . . . . . . . . . . . . . . . . . . . . . . . . . 165
(b) Class 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
7.5 Upper tails of ‘one cycle’ hoistload distributions . . . . . . . . . . . 165
7.6 Histogram of extreme hoistloads for Nextr = 100 . . . . . . . . . . 169
7.9 Trend of the mean values of the extreme hoistload distributions . . 170
7.10 Trend of the standard deviations of the extreme hoistload distrib-
utions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
7.11 Normal probability plot of extreme hoistload values . . . . . . . . . 171
7.12 Cumulative distributions functions for extreme hoistload distributions172
(a) Class 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
(b) Class 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
(c) Class 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
(d) Class 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
8.1 Combinations of load spectrum, kQ, and number of cycles, N . . . 189

8.2 Distribution of reliability . . . . . . . . . . . . . . . . . . . . . . . 189
9.1 Parametric study into class of crane and Nextr . . . . . . . . . . . . 194

9.2 260t crane auxiliary girder, LC1 . . . . . . . . . . . . . . . . . . . . 196
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 196
(b) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 196
9.3 260t crane main girder, LC5 . . . . . . . . . . . . . . . . . . . . . . 197
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 197
(b) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 197
9.4 Sensitivity factors for LC1 & LC5 . . . . . . . . . . . . . . . . . . . 198
(a) 260t crane auxiliary girder, LC1 . . . . . . . . . . . . . . . 198
(b) 260t crane main girder, LC5 . . . . . . . . . . . . . . . . . . 198
282 List of Figures
9.5 Influence of vertical and horizontal loads, code format 1 . . . . . . 200

(a) Vertical loads only . . . . . . . . . . . . . . . . . . . . . . . 200
(b) Vertical and horizontal loads . . . . . . . . . . . . . . . . . 200
9.6 Influence of vertical and horizontal loads, code format 2 . . . . . . 201
(a) Vertical loads only . . . . . . . . . . . . . . . . . . . . . . . 201
(b) Vertical and horizontal loads . . . . . . . . . . . . . . . . . 201
9.7 Assessment of reliability of current practice . . . . . . . . . . . . . 203
(a) Variation with Nextr . . . . . . . . . . . . . . . . . . . . . . 203
(b) Variation with ratio of hoistload to total crane weight . . . 203
9.8 Sensitivity factors for class 2 crane, SABS 0160:1989 . . . . . . . . 205
9.9 Sensitivity factors for class 3 crane, SABS 0160:1989 . . . . . . . . 205
9.10 Crane only calibration, code formats 1 and 2 . . . . . . . . . . . . 207
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 207
(b) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 207
9.11 Crane only calibration, code formats 3 and 4 step (a) . . . . . . . . 208
(a) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 208
(b) Code format 4 . . . . . . . . . . . . . . . . . . . . . . . . . 208
9.12 Crane only calibration, code formats 3 and 4 step (b) . . . . . . . . 209
(a) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 209
(b) Code format 4 . . . . . . . . . . . . . . . . . . . . . . . . . 209
9.13 Excess cost of the 5t crane support structure . . . . . . . . . . . . 210
9.14 Excess cost of the 40t crane support structure . . . . . . . . . . . . 211
9.15 Excess cost of the 260t crane support structure . . . . . . . . . . . 211
9.16 Crane and Permanent loads . . . . . . . . . . . . . . . . . . . . . . 213
9.17 Effect of Nextr for crane load as accompanying load . . . . . . . . . 214
9.18 Calibration of combination factors for crane and wind loads . . . . 216
(a) Wind dominant . . . . . . . . . . . . . . . . . . . . . . . . . 216
(b) Crane dominant . . . . . . . . . . . . . . . . . . . . . . . . 216
9.19 Calibrated combination factors for crane and wind . . . . . . . . . 217
9.20 Sensitivity of class 1 crane to Nextr . . . . . . . . . . . . . . . . . . 218
9.21 Calibration of combination factors for crane and roof imposed loads 219
(a) Full range of reliability . . . . . . . . . . . . . . . . . . . . . 219
(b) Relevant range of reliability . . . . . . . . . . . . . . . . . . 219
9.22 Effect of class and cycles for accidental limit state . . . . . . . . . 221
9.23 Calibration of partial load factors for accidental limit state . . . . . 223
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 223
List of Figures 283
(b) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 223

9.24 Effect of different values of Nextr on accidental limit state reliability,
class 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
9.25 Effect of different values of Nextr on accidental limit state reliability,
class 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
9.26 Effect of single or double slope S-N curve . . . . . . . . . . . . . . 226
(a) Single slope S-N curve . . . . . . . . . . . . . . . . . . . . . 226
(b) Double slope S-N curve . . . . . . . . . . . . . . . . . . . . 226
9.27 Sensitivity factors for a single slope S-N curve at 2 × 106 cycles . . 227
9.28 Sensitivity factors for a double slope S-N curve at 1.25 × 105 cycles 228
9.29 Sensitivity factors for a double slope S-N curve at 2 × 106 cycles . 228
9.30 Effect of different load spectrum classes . . . . . . . . . . . . . . . 229
(a) Load spectrum class Q2 . . . . . . . . . . . . . . . . . . . . 229
(b) Load spectrum class Q5 . . . . . . . . . . . . . . . . . . . . 229
9.31 40t crane girder, top flange to web weld . . . . . . . . . . . . . . . 231
(a) Load spectrum class Q2 . . . . . . . . . . . . . . . . . . . . 231
(b) Load spectrum class Q5 . . . . . . . . . . . . . . . . . . . . 231
9.32 Calibration results for 5t crane corbel to column connection . . . . 232
9.33 Calibration results for 40t crane girder bottom of stiffener . . . . . 233
9.34 Calibration results for 40t crane girder top flange to web weld . . . 233
10.1 Sensitivity factors for the 5t crane column . . . . . . . . . . . . . . 240

10.2 Sensitivity of reliability to c.o.v. of modelling uncertainties . . . . 242
10.3 Sensitivity of γC to horizontal wheel load modelling uncertainty
parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
10.4 Sensitivity of γH to horizontal wheel load modelling uncertainty
parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
10.5 Reliability of bracing elements for accidental limit state, Class 1 . . 248
(a) Class 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
(a) Class 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
(a) Class 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
(a) Class 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
A.1 Building outline for wind calculations . . . . . . . . . . . . . . . . 308

284 List of Figures
A.2 Surface pressure coefficients on walls . . . . . . . . . . . . . . . . . 309

(a) Wind along . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
(b) Wind across . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
A.3 Surface pressure coefficients on roof . . . . . . . . . . . . . . . . . . 309
(a) Wind along . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
(b) Wind across from tall side . . . . . . . . . . . . . . . . . . . 309
(c) Wind across from short side . . . . . . . . . . . . . . . . . . 309
B.1 5t crane girder, maximum positive vertical moment, LC1 . . . . . . 313

(a) Vertical moments . . . . . . . . . . . . . . . . . . . . . . . . 313
(b) Horizontal moments . . . . . . . . . . . . . . . . . . . . . . 313
B.2 5t crane girder, maximum negative vertical moment, LC1 . . . . . 314
B.3 5t crane girder, maximum horizontal moment, LC1 . . . . . . . . . 314
B.4 5t crane girder, maximum positive vertical moment, LC5 . . . . . . 314
B.5 5t crane girder, maximum negative vertical moment, LC5 . . . . . 315
B.6 5t crane girder, maximum horizontal moment, LC5 . . . . . . . . . 315
C.1 20t class 1 crane (produced from data) . . . . . . . . . . . . . . . . 317

C.2 3.2t class 1 crane (produced from descriptions) . . . . . . . . . . . 318
C.6 5t class 2 crane (produced from descriptions) . . . . . . . . . . . . 320
List of Figures 285
E.1 LC1 maximum positive vertical moment, code formats 1 & 3 . . . 329
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 329
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 329
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 330
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 330
E.3 LC1 maximum negative vertical moment, code formats 1 & 3 . . . 330
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 330
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 330
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 330
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 330
E.5 LC1 maximum horizontal moment, code formats 1 & 3 . . . . . . . 331
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 331
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 331
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 331
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 331
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 331
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 331
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 332
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 332
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 332
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 332
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 332
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 332
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 333
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 333
286 List of Figures
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 333

(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 333
E.13 LC1 maximum horizontal, code formats 1 & 3 . . . . . . . . . . . . 334
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 334
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 334
E.14 LC1 maximum horizontal, code formats 2 & 4 . . . . . . . . . . . . 334
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 334
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 334
E.15 LC1 maximum vertical, code formats 1 & 3 . . . . . . . . . . . . . 335
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 335
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 335
E.16 LC1 maximum vertical, code formats 2 & 4 . . . . . . . . . . . . . 335
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 335
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 335
E.17 LC5, code formats 1 & 3 . . . . . . . . . . . . . . . . . . . . . . . . 335
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 335
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 335
E.18 LC5, code formats 2 & 4 . . . . . . . . . . . . . . . . . . . . . . . . 336
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 336
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 336
E.19 LC1 maximum horizontal moment, bottom flange, code formats 1
&3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 336
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 336
&4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 337
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 337
E.21 LC1 maximum horizontal moment, top flange, code formats 1 & 3 337
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 337
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 337
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 337
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 337
&3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
List of Figures 287
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 338

(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 338
&4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 338
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 338
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 338
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 338
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 339
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 339
E.27 LC1 maximum vertical moment, bottom flange, code formats 1 & 3 339
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 339
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 339
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 339
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 339
E.29 LC1 maximum vertical moment, top flange, code formats 1 & 3 . . 340
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 340
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 340
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 340
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 340
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 340
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 340
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 341
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 341
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 341
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 341
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 341
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 341
288 List of Figures
E.35 LC1, code formats 1 & 3 . . . . . . . . . . . . . . . . . . . . . . . . 342

(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 342
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 342
E.36 LC1, code formats 2 & 4 . . . . . . . . . . . . . . . . . . . . . . . . 342
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 342
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 342
E.37 LC5, code formats 1 & 3 . . . . . . . . . . . . . . . . . . . . . . . . 343
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 343
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 343
E.38 LC5, code formats 2 & 4 . . . . . . . . . . . . . . . . . . . . . . . . 343
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 343
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 343
E.39 LC1 maximum horizontal force, code formats 1 & 3 . . . . . . . . . 344
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 344
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 344
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 344
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 344
E.41 LC1 maximum vertical force, code formats 1 & 3 . . . . . . . . . . 345
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 345
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 345
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 345
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 345
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 345
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 345
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 346
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 346
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 346
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 346
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 346
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 346
List of Figures 289

(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 347
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 347
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 347
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 347
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 348
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 348
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 348
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 348
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 349
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 349
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 349
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 349
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 350
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 350
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 350
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 350
E.55 LC5, code formats 1 & 3 . . . . . . . . . . . . . . . . . . . . . . . . 350
(a) Code format 1 . . . . . . . . . . . . . . . . . . . . . . . . . 350
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 350
E.56 LC5, code formats 2 & 4 . . . . . . . . . . . . . . . . . . . . . . . . 351
(a) Code format 2 . . . . . . . . . . . . . . . . . . . . . . . . . 351
(b) Code format 3 . . . . . . . . . . . . . . . . . . . . . . . . . 351
E.57 5t crane corbel to column connection, load spectrum class Q2 . . . 351
E.61 40t crane girder intermediate stiffener, load spectrum class Q2 . . . 352
290 List of Figures
E.65 40t crane girder top flange to web weld, load spectrum class Q2 . . 354
List of Tables
2.1 Classification of cranes in SABS 0160:1989 . . . . . . . . . . . . . . 18

2.2 Classification of cranes in DIN 15018 and prEN 1991-3 . . . . . . . 19
2.3 Classification of cranes into hoist classes in AS1418.1-1994 . . . . . 20
2.4 Dynamic factors for vertical loads from SABS 0160:1989 . . . . . . 21
2.5 Dynamic factors for vertical loads from ASCE 7-98 . . . . . . . . . 21
2.6 Factors for the calculation of φ2 . . . . . . . . . . . . . . . . . . . . 23
2.7 Determination of φ1 in AS1418.1-1994 . . . . . . . . . . . . . . . . 26
2.8 Determination of φ1 in DIN 15018 . . . . . . . . . . . . . . . . . . 27
2.9 Values of the dynamic factor φ5 in prEN 1991-3 . . . . . . . . . . . 30
2.10 Factors from SABS 0160:1989 for the loads due to acceleration and
braking of the crab . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.11 Factors from SABS 0160:1989 for the loads due to misalignment of
crane wheels or gantry rails . . . . . . . . . . . . . . . . . . . . . . 32
2.12 Distance to instantaneous centre of rotation for skewing . . . . . . 35
2.13 Force factors for skewing forces . . . . . . . . . . . . . . . . . . . . 35
2.14 Additional loads on crane support structures . . . . . . . . . . . . 44
2.15 Summary of crane loads considered by the codes . . . . . . . . . . 45
2.16 Crane load combinations . . . . . . . . . . . . . . . . . . . . . . . . 46
2.17 λ factors according to the classification of cranes . . . . . . . . . . 49
2.18 Classification of the fatigue actions from cranes . . . . . . . . . . . 51
3.1 Parameters of representative cranes . . . . . . . . . . . . . . . . . . 68

3.2 Parameters of the 5t crane . . . . . . . . . . . . . . . . . . . . . . . 72
3.3 Parameters of the 40t crane . . . . . . . . . . . . . . . . . . . . . . 75
3.4 Parameters of the 260t crane . . . . . . . . . . . . . . . . . . . . . 78
4.1 Crane classification . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
291
292 List of Tables
4.2 Class dependent factors in SABS 0160:1989 . . . . . . . . . . . . . 83

4.3 Hoist class dependent dynamic factors in prEN 1991-3 . . . . . . . 84
4.4 Load effect ratios: SABS 0160:1989/prEN 1991-3 . . . . . . . . . . 95
4.5 Ratio of areas: SABS 0160:1989/prEN 1991-3 . . . . . . . . . . . . 98
4.6 Crane information required for design of support structure . . . . . 100
5.1 Definitions of crane related partial load factors and combination

factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.2 Definitions of non-crane related partial load factors . . . . . . . . . 105
6.1 Statistical parameters of the yield strength of structural steel . . . 131

6.2 Statistical parameters of the yield strength of reinforcing steel . . . 136
6.3 Statistical parameters of the permanent loads . . . . . . . . . . . . 139
6.4 Investigation into vertical wheel load modelling uncertainty . . . . 145
6.5 Bias values for end stop forces . . . . . . . . . . . . . . . . . . . . . 146
6.6 Modelling uncertainty of resistance of steel columns . . . . . . . . . 148
6.7 Statistical models for material properties . . . . . . . . . . . . . . . 154
6.8 Statistical models for geometric properties . . . . . . . . . . . . . . 154
6.9 Statistical models for loads . . . . . . . . . . . . . . . . . . . . . . 154
6.10 Statistical models for modelling uncertainties . . . . . . . . . . . . 155
6.11 Statistical models for fatigue analysis . . . . . . . . . . . . . . . . . 155
7.1 Parameters of ‘one cycle’ beta distributions . . . . . . . . . . . . . 163
8.1 50 year target reliabilities . . . . . . . . . . . . . . . . . . . . . . . 177

8.2 25 year target reliabilities . . . . . . . . . . . . . . . . . . . . . . . 179
8.3 Definitions of crane partial load factors . . . . . . . . . . . . . . . . 182
9.1 Current partial safety factors in SABS 0160:1989 and prEN 1991-3 202
9.2 Calibrated partial load factors . . . . . . . . . . . . . . . . . . . . . 207
9.4 Calibrated combination factors . . . . . . . . . . . . . . . . . . . . 237
11.1 Summary of crane loads considered by the codes . . . . . . . . . . 256

11.3 Partial load factors for fatigue . . . . . . . . . . . . . . . . . . . . . 269
11.5 Partial load factors for fatigue . . . . . . . . . . . . . . . . . . . . . 271
List of Tables 293
A.1 Crane information required for SABS 0160:1989 . . . . . . . . . . . 295

A.2 Crane load combinations . . . . . . . . . . . . . . . . . . . . . . . . 299
Appendix A
Load calculations
Example load calculations for the 40t representative crane described in Chapter
3 are given below.
A.1 Crane loads according to SABS 0160:1989
Diagrams showing the load combinations are in Figures 4.1 - 4.5 on page 86.
The information about the crane which is required for the calculation of
the load according to SABS 0160:1989 [1] is given in Table A.1.
Table A.1: Crane information required for SABS 0160:1989

Crane parameter
Nominal weight of hoistload 400 kN
Total number of wheels 4
Maximum vertical wheel load 335 kN
Minimum vertical wheel load 75 kN
Buffer type Hydraulic PUDZ II φ 63
The crane is a process crane in a car stamping plant which is under contin-
uous use and is classified according to SABS 0160:1989 [1] as a class 3 crane.
295
296 Appendix A. Load calculations
A.1.1 Vertical loads
A.1.1.1 Static vertical loads
The static vertical loads are as given by the crane manufacturer:
Vs,max = 335 kN
Vs,min = 75 kN
A.1.1.2 Dynamic vertical loads
The impact factor for the dynamic loads is 1.25 for a class 3 crane.
Vd,max = 335 × 1.25 = 419 kN
Vd,min = 75 × 1.25 = 94 kN
A.1.2 Horizontal transverse loads
A.1.2.1 Acceleration or braking of the crab
The factor for the calculation of the horizontal transverse loads due to accel-
eration or braking of the crab is 0.15 for a class 3 crane.
Ha = 0.15 × (Qcr + Qh )
= 0.15 (98 + 400)
= 18.7 kN
A.1.2.2 Misalignment of rails or wheels
The factor for the calculation of the horizontal transverse loads due to the
misalignment of the rails or wheels is 0.15 for a class 3 crane.
Qbr + Qcr + Qh
Hb = 0.15 ×
Total number of crane wheels
298 + 98 + 400
= 0.15 ×
4
= 29.9 kN
A.1. Crane loads according to SABS 0160:1989 297
A.1.2.3 Skewing of the crane in plan
Hc = 1.5 × Hb = 44.8 kN
A.1.3 Horizontal longitudinal load

The horizontal longitudinal load is caused by the acceleration and braking of
the crane bridge.
HL = 0.1 × (sum of maximum static wheel loads on rail)

= 0.1 × 2 × 335
= 67.0 kN
A.1.4 End stop forces

Method (a): The force on each end stop caused by the crane bridge running
into the end stops on the end of the runway is given by:
He = Qbr + Qcr = 396 kN
Method (b): Taking the resilience of the buffers into account, assuming the
ends stops are rigid, and the crane travelling at full rated speed, taking only
the weight of the crane bridge and crab.
Ekin = 0.5mv 2
= 0.5 (29800 + 9800) (0.833)2
= 13739J
Ekin
He = k
s
13739
= 1.1 × 10−3
0.1
= 151 kN
Where:
k – buffer characteristic factor, degree of plasticity

s – buffer stroke length
A.2 Crane loads according to prEN 1991-3
The crane loads have been calculated using the models in prEN 1991-3. Dia-
grams showing the factored load combinations are shown in Figures 4.6 - 4.13
on page 88.
The information about the crane that is required to calculate the loads is
given in Table 3.3 on page 75.
A.2.1 Dynamic factors
φ1 – vibrational excitation of the crane structure due to lifting of the hoistload

off the ground.
φ1,min = 0.9
φ1,max = 1.1
φ2 – dynamic effects of transferring the hoistload from the ground to the

crane.
φ2 = φ2 min + β2 vh
The crane is HC 3 therefore φ2 min = 1.15 and β2 = 0.51.
φ2 = 1.15 + (0.51)(10/60) = 1.235
φ4 – dynamic effect induced when travelling on rail tracks or runways.

The rails are welded and ground smooth therefore:
φ4 = 1.0
φ5 – effects caused by drive forces

Forces change smoothly therefore 1 ≤ φ5 ≤ 1.5
φ5 = 1.25
φ6 – dynamic effects of test load.

A.2. Crane loads according to prEN 1991-3 299
φ6,dyn – when test load moved by the drives in the way the crane is
used.
φ6,dyn = 0.5(1 + φ2 ) = 0.5(1 + 1.235) = 1.118
φ6,stat – when a test load is not moved by the drives
φ6,stat = 1.0
φ7 – elastic effects of impact on buffers
φ7 = 1.25 + 0.7(ξ − 0.5) = 1.25 + 0.7(0.9 − 0.5) = 1.53
A.2.2 Load combinations
The crane load combinations as specified by prEN 1991-3 are given in Table
A.2.
Table A.2: Crane load combinations

Groups of loads
Test Acci-
ULS load den-
tal
1 2 3 4 5 6 7 8 9
1 Self weight of crane φ1 φ1 1 φ4 φ4 φ4 1 φ1 1
2 Hoist load φ2 φ3 - φ4 φ4 φ4 η1 - 1
3 Acceleration of crane φ5 φ5 φ5 φ5 - - - φ5 -
bridge
4 Skewing of crane - - - - 1 - - - -
bridge
5 Acceleration or brak-
ing of crab or hoist - - - - - 1 - - -
block
6 Test load - - - - - - - 1 -
7 Buffer force - - - - - - - - φ7
1 η is the part of the hoist load that remains when the payload
is removed, but is not included in the self weight of the crane

A.2.3 Vertical loads
A.2.3.1 Load combination 1
φ1 (Self weight) + φ2 (Hoistload)

Maximum wheel load:
· ¸
1 Qbr l − emin
Qr,max = φ1 + (φ1 Qcr + φ2 Qh )
2 2 l
· ¸
1 298 23800 − 1650
= 1.1 + (1.1 × 98 + 1.235 × 400)
2 2 23800
= 362 kN
Minimum wheel load:

· ¸
1 Qbr emin
Qr,min = φ1 + (φ1 Qcr + φ2 Qh )
2 2 l
· ¸
1 298 1650
= 1.1 + (1.1 × 98 + 1.235 × 400)
2 2 23800
= 103 kN
A.2.3.2 Load combination 3 and 7
(Self weight) + η(Hoistload)

Maximum wheel load:
· ¸
1 Qbr l − emin
Qr,max = + (Qcr )
2 2 l
· ¸
1 298 23800 − 1650
= + (98)
2 2 23800
= 120 kN
Minimum wheel load:

· ¸
1 Qbr emin
Qr,min = + (Qcr )
2 2 l
· ¸
1 298 1650
= + (98)
2 2 23800
= 78 kN
A.2.3.3 Load combination 4, 5 and 6
φ4 (Self weight) + φ4 (Hoistload)
Maximum wheel load:

· ¸
φ4 Qbr l − emin
Qr,max = + (Qcr + Qh )
2 2 l
· ¸
1 298 23800 − 1650
= + (98 + 400)
2 2 23800
= 306 kN
Minimum wheel load:

· ¸
φ4 Qbr emin
Qr,min = + (Qcr + Qh )
2 2 l
· ¸
1 298 1650
= + (98 + 400)
2 2 23800
= 92 kN
A.2.3.4 Load combination 4, 5 and 6
(Self weight) + (Hoistload)
Maximum wheel load:

· ¸
1 Qbr l − emin
Qr,max = + (Qcr + Qh )
2 2 l
· ¸
1 298 23800 − 1650
= + (98 + 400)
2 2 23800
= 306 kN
Minimum wheel load:

· ¸
1 Qbr emin
Qr,min = + (Qcr + Qh )
2 2 l
· ¸
1 298 1650
= + (98 + 400)
2 2 23800
= 92 kN
A.2.4 Horizontal loads
A.2.4.1 Longitudinal and transverse loads caused by acceler-

ation and braking of crane
Horizontal longitudinal force:
1
HL = φ5 K
nr
Where:
nr – is the number of rails
Drive force K:
X
K=α Qr,min
= αmw Qr,min (mw = number of single wheel drives)
= (0.2)(2)(78)
= 31.2 kN
Therefore:
1
HL = (1.25)(31.2) = 19.5 kN
2
Horizontal transverse forces:
M
HT,1 = φ5 ξ2
a
M
HT,2 = φ5 ξ1
a
Position of center of mass of crane:
P
Qr,max
ξ1 = P
Qr
(2)(306)
=
298 + 98 + 400
= 0.769
ξ2 = 1 − ξ1 = 0.231
Moment caused by drive forces not acting at the center of mass:
M = Kls = K (ξ1 − 0.5) l

= (31.2)(0.769 − 0.5)(23.8)
= 199.75 kNm
Therefore:
µ ¶
199.75
HT,1 = (1.25)(0.231) = 13.1 kN
4.4
µ ¶
199.75
HT,2 = (1.25)(0.769) = 43.6 kN
4.4
A.2.4.2 Horizontal loads and the guide force caused by skew-

ing
Guide force:
X
S = f λs,j Qr
Horizontal longitudinal forces:

X
Hs,i,j,L = f λs,i,j,L Qr
Horizontal transverse forces:

X
Hs,i,j,T = f λs,i,j,T Qr
Where:
i – is the rail number
j – is the wheel pair number
L – stands for longitudinal forces
T – stands for transverse forces
Non positive factor:
f = 0.3 (1 − exp (−250α)) ≤ 0.3

Skewing angle:
α = αF + αv + αo ≤ 0.015
0.75x
αF = (x = clearance between rail and wheel flange)
a
0.75(25)
= = 4.26 × 10−3
4400
y
αv = (y = wear of rail and guide means ≥ 0.10b)
a
0.10(100)
= = 2.27 × 10−3
4400
αo = 0.001
∴ α = 7.53 × 10−3
¡ ¡ ¢¢
f = 0.3 1 − exp −250 × 7.53 × 10−3 = 0.254
Combination of wheel pairs - Independent, fixed/fixed.

Distance to instantaneous slide pole:
P 2
ej 44002
h= P = = 4400 mm
ej 4400
force factors:
P
ej 4400
λs,j = 1 − =1− = 0.5
nh 2 × 4400
λs,i,j,L = 0
ξ2 ³ ej ´ 0.231
λs,1,1,T = 1− = = 0.115
n h 2
ξ1 ³ ej ´ 0.769
λs,2,1,T = 1− = = 0.385
n h 2
λs,i,2,L = 0
Where:
n – is the number of wheel pairs
ej – is the distance of wheel pair j from the guide means

Guide force:
X
S = f λs,j Qr
= (0.254)(0.5)(298 + 98 + 400) = 101.1 kN
Transverse forces:
X
Hs,1,1,T = f λs, 1, 1, T Qr = (0.254)(0.115)(796) = 23.25 kN
X
Hs,2,1,T = f λs, 2, 1, T Qr = (0.254)(0.385)(796) = 77.80 kN
A.2.4.3 Horizontal forces caused by acceleration or braking of

the crab
The forces are assumed by the code to be less than the buffer forces related to
movement of the crab and are not calculated.
A.2.4.4 Buffer forces related to movements of the crab
The payload is free to swing, therefore:
HB2 = 0.10 (Qcr + Qh )

= 0.10(98 + 400) = 49.8 kN
Force on one wheel = 24.9 kN.
A.2.4.5 Test loads
Dynamic test load – 10% overload:
QT,dyn = φ6,dyn 110%Qh = 1.118(1.10)(400) = 492 kN
Static test load – 25% overload
QT,stat = φ6,stat 125%Qh = 1.0(1.25)(400) = 500 kN
Maximum test load is static load.

Vertical wheel loads - Load Combination 8: φ1 (Self weight) + φ6 (Test load)
Maximum wheel load:

· ¸
1 Qbr l − emin
Qr,max = φ1 + (φ1 Qcr + φ6 QT )
2 2 l
· ¸
1 298 23800 − 1650
= 1.1 + (1.1 × 98 + 500)
2 2 23800
= 365 kN
Minimum wheel load:

· ¸
1 Qbr emin
Qr,min = φ1 + (φ1 Qcr + φ6 QT )
2 2 l
· ¸
1 298 1650
= 1.1 + (1.1 × 98 + 500)
2 2 23800
= 103 kN
A.2.4.6 Buffer forces related to crane movement
The equation given for the calculation of the end stop forces in prEN 1991-3
is:
p
HB,1 = φ7 u mc SB
Where:
φ7 – dynamic factor taking into account plasticity of buffer
u – long travel speed of crane taken as 70% of maximum speed
mc – mass of crane and hoistload
SB – spring constant of buffer
Buffer maximum forces Fmax = 200 kN and maximum deflection smax =

100 mm.
Fmax
SB =
smax
= 2 × 106 N/m
A.3. Roof Imposed loads 307
p
HB,1 = φ7 u mc SB
p
= (0.53)(0.583) (29800 + 9800 + 40000) 2 × 106
= 356 kN
A.3 Roof Imposed loads

The roof imposed loads were calculated using the method given in Clause 5.4.4
SABS 0160:1989 [1].
(5.4.4.3) Inaccessible roof – a uniformly distributed load applied to the roof
with a value of:
µ ¶
15 − A
ω= 0.3 + kN/m2 ≥ 0.3 kN/m2
6
Where:
A – is the tributary area of the member being considered in m2

For one frame:
A = 45 × 8 = 360 m2
∴ ω = 0.3 kN/m2
Force per unit length of roof:
q = ω × 8 = 2.4 kN/m
This force is multiplied by the tributary length of the purlin to get the
force per purlin/frame connection.
A.4 Wind loads

The wind loads were calculated using the method given in Clause 5.5 in SABS
0160:1989 [1]. An outline of the building as considered for the calculation of
wind loads is given in Figure A.1.
Regional wind speed = 40 m/s
Mean return period = 50 years
Terrain category 3 - built up industrial
Height of structure = 18400mm on tall side, (14190mm on short side)
Figure A.1: Building outline for wind calculations
(Table 5 SABS 0160:1989 [1]) Wind speed multiplier:
kz,tall = 0.814
kz,short = 0.769
Characteristic wind speed:
Vz,tall = kz,tall V = 32, 56 m/s
Vz,short = kz,short V = 30.76 m/s
Velocity pressures:
qz = kp Vz 2
kp = 0.6 (0m above sea level)
qz,tall = 0.6(32.56)2 = 636 N/m2
qz,short = 0.6(30.76)2 = 568 N/m2
Building plan ratio:

b 104
= = 2.31
w 45
Building height ratio:
h 184
= = 0.41
w 450
The surface pressure coefficients were taken from Tables 6 and 8 in SABS
0160:1989 [1]. Figures A.2 & A.3 show the surface pressure coefficients.
A.4. Wind loads 309
(a) Wind along (b) Wind across
Figure A.2: Surface pressure coefficients on walls
(a) Wind along (b) Wind across from tall side (c) Wind across from short side
Figure A.3: Surface pressure coefficients on roof
The wind load per purlin/frame or girt/frame connection:
Wn = (frame spacing) × (pressure coefficient)

×(velocity pressure) × (purlin/girt spacing)
A.4.0.7 Load Case 1 - Wind along
Walls:
Tall : Wn = (8)(0.5)(0.636)sg = 2.544sg kN
Short : Wn = (8)(0.5)(0.568)sg = 2.272sg kN
Roof:
Wn = (8)(1.0)(0.636)sp = 5.088sp kN
Horizontal component : Wn,h = Wn sin 4◦ = 0.355sp kN
Vertical component : Wn,v = Wn cos 4◦ = 5.076sp kN

A.4.0.8 Load Case 2a - Wind across from the tall side
Walls:
Tall : Wn = (8)(0.7)(0.636)sg = 3.562sg kN
Short : Wn = (8)(0.25)(0.568)sg = 1.136sg kN
Roof (windward):
Wn = (8)(1.0)(0.636)sp = 5.088sp kN
Roof (leeward):
Wn = (8)(0.5)(0.636)sp = 2.544sp kN
A.4.0.9 Load Case 2b - Wind across from the short side
Walls:
Tall : Wn = (8)(0.25)(0.636)sg = 1.272sg kN
Short : Wn = (8)(0.7)(0.568)sg = 3.181sg kN
Roof (windward):
Wn = (8)(1.0)(0.636)sp = 5.088sp kN
Roof (leeward):
Wn = (8)(0.5)(0.636)sp = 2.544sp kN

A.5. Permanent loads 311
A.5 Permanent loads

The permanent loads on the frame consisted of:
1. Sheeting and utilities

Weight of sheeting, assume 6 kg/m2
Weight of utilities, assume 1 kg/m2
Force per purlin/frame connection:
Fsu = (6 + 1)(8)(9.81) × 10−3 sp = 0.55sp kN
2. Purlins
Purlin section: 225 × 75 × 20 × 2.5 CFC, weight = 7.78kg/m.
Force per purlin/frame connection:
Fp = (7.78)(8)(9.81) × 10−3 = 0.61 kN
3. Monitors
Frame section: 150 × 65 × 20 × 2.5 CFC, weight = 5.92 kg/m.
Length of frame = 1500 × 2 + 4490 = 7490 mm
One frame per purlin, force per purlin = (7.49)(5.92)(9.81) × 10−3 =
0.43 kN
Two false rafters 45 × 45 × 5L, weight = 3.38 kg/m
Two girts 45 × 45 × 5L, weight = 3.38 kg/m
Force per unit length of roof = 4(3.38)(9.81) × 10−3 = 0.133 kN/m
Force per purlin/frame connection from frames, rafters and girts:
Fm = 0.43 + (0.133)(2.15) = 0.716 kN
4. Self weight
The self weight of the structural elements is calculated by Prokon and
included in the ‘Permanent’ load case for analysis.
Appendix B
Load Effects
B.1 5t crane girder
Moments in the 5t crane girder, leading load at 5.5 m Moments in the 5t crane girder, leading load at 5.5 m
−60 −10
−44.82 −8.57
−40
Moment due to crane wheel loads (kNm)
−20
−5
20
0
40
60
76.13 4.535
80 5
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Distance along beam (m) Distance along beam (m)
(a) Vertical moments (b) Horizontal moments
Figure B.1: 5t crane girder, maximum positive vertical moment, LC1
313
314 Appendix B. Load Effects
Moments in the 5t crane girder, leading load at 8 m Moments in the 5t crane girder, leading load at 8 m
−50 −49.56 −10
−8.996
−8
−40

−6
−30 −4
−2
−20
−10
2
0 4
6
10
8
8.996
17.9
20 10
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Figure B.2: 5t crane girder, maximum negative vertical moment, LC1
−30 −6
−28.66
−4.602
−20 −4
−10 −2
0
0
10
2
20
4
30
6
40
8
50
60 10
10.47
65.2
70 12
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Figure B.3: 5t crane girder, maximum horizontal moment, LC1
Moments in the 5t crane girder, leading load at 5.5 m Moments in the 5t crane girder, leading load at 5.5 m
−40 −4
−38 −3.769
−3
−20
−2
0 −1
0
20
1
40 2
3
60
64.54 4
80 5 4.976
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Figure B.4: 5t crane girder, maximum positive vertical moment, LC5

B.1. 5t crane girder 315
−50 −6
−4.925
−42.01
−40 −4

−30 −2
−20 0
−10 2
0 4
10 6
15.18
7.346
20 8
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Figure B.5: 5t crane girder, maximum negative vertical moment, LC5
−30 −6 −5.696
−24.29
−4
−20
−2
−10
0
0
2
10
4
20
6
30
8
40
10
50 12
55.28 12.96
60 14
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Figure B.6: 5t crane girder, maximum horizontal moment, LC5

Appendix C
Graphs for stochastic modelling

of hoistload
C.1 Class 1 cranes
Histogram of hoistloads for 20t maintenance crane

25
20
Frequency %
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL
Figure C.1: 20t class 1 crane (produced from data)
C.1.0.10 Maintenance crane
For 80% of the cycles the crane lifts less than half the SWL. Crane does
sometimes lift the SWL.
317
318 Appendix C. Graphs for stochastic modelling of hoistload
Histogram of hoistloads for 5t container making crane

35
30
Frequency % 25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL
Figure C.2: 3.2t class 1 crane (produced from descriptions)
Histogram of hoistloads for 20t boiler making crane

50
45
40
35
Frequency %
30
25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL

C.2. Class 2 cranes 319
C.2 Class 2 cranes
Histogram of hoistloads for 25t press line crane

18
16
14
12
Frequency %
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL

18
16
14
12
Frequency %
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL
C.2.0.11 Coil handling crane
Coils range from 5% SWL to 70% SWL. Most coils are in the range 40% SWL
to 50% SWL.

30
25
20
Frequency %
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL
Figure C.6: 5t class 2 crane (produced from descriptions)
Histogram of hoistloads for 80t caster crane in steelworks

50
45
40
35
Frequency %
30
25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL
Histogram of hoistloads for 35t coil handling crane

50
45
40
35
Frequency %
30
25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL

C.3 Class 3 cranes

40
35
30
Frequency %
25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL

25
20
Frequency %
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL

Histogram of hoistloads for 80t slab handling crane

20
18
16
14
Frequency %
12
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction SWL

C.4 Class 4 cranes

C.4.0.12 Ladle crane at a steelworks
Full ladles range from 98% SWL to 110% SWL. The empty ladle is 32% SWL.
C.4.0.13 Ladle crane at a steelworks
Always lifts SWL, the empty ladle is 27% SWL.
C.5 Extreme hoistload distributions

Equations are given below for the means and standard deviations of the ex-
treme hoistload distributions.
C.5.0.14 Class 1
( ¡ ¢
0.0143Nextr 3 − 0.1575Nextr 2 + 0.6225Nextr + 0.3188 Nextr ≤ 104
µ1 = ¡ ¢
−0.0861 · 10−3 Nextr 3 − 0.0052Nextr 2 + 0.0749Nextr + 0.9953 Nextr ≥ 104
(C.5.1)
log10 (σ1 ) = −0.3334 log10 (Nextr ) − 0.4614 (C.5.2)
C.5.0.15 Class 2
( ¡ ¢
−0.0220Nextr 2 + 0.1923Nextr + 0.7850 Nextr ≤ 104
µ2 = ¡ ¢ (C.5.3)
−0.0065Nextr 2 + 0.0805Nextr + 0.9864 Nextr ≥ 104
log10 (σ2 ) = −0.2922 log10 (Nextr ) − 0.6874 (C.5.4)
C.5.0.16 Class 3
( ¡ ¢
µ3 = ¡ ¢
(C.5.5)
log10 (σ3 ) = −0.2346 log10 (Nextr ) − 0.9835 (C.5.6)

C.5.0.17 Class 4
( ¡ ¢
µ4 = ¡ ¢
(C.5.7)
log10 (σ4 ) = −0.1675 log10 (Nextr ) − 1.3613 (C.5.8)

Appendix D
Ratios of hoistload to total

crane weight
Weight Weight Crane

r Crane description of bridge of crab Capacity
kN kN kN
0.30 Standard 1t 15m 22.37 0.83 10
0.35 Charging crane 160t 28.5m 2000 1000 1600
0.37 Warehouse crane 35t 30m 468 120 350
0.38 Standard 1t 12,5m 15.47 0.83 10
0.40 Coil handling crane 45t 30m 545 139 450
0.43 Standard 1t 10m 12.67 0.83 10
0.44 Ladle crane 260t 28.5m 1970 1298 2600
0.45 Container making crane 5t 19.2m 55 6.4 50
325
0.45 Standard 2t 15m 23.28 1.12 20
0.45 Roll shop crane 40t 24.5m 353 133 400
0.48 Standard 10t 20m span 100 10 100
0.48 Caster crane 80t 31.75m 670.1 204.5 800
0.48 Standard 3,2t 16,5m 31.5 2.8 32
0.50 Standard3,2t 15m 29 3.5 32
326 Appendix D. Ratios of hoistload to total crane weight
Weight Weight Crane

kN kN kN
0.51 Standard2,5t 15m 23.29 1.21 25
0.52 Standard 3,2t 15,5m 26.7 2.8 32
0.52 Standard2,5t 12,5m 21.59 1.21 25
0.53 Standard15t 20m span 118 15 150
0.54 Standard 1t 7,5m 7.67 0.83 10
0.55 Rolling mill crane 16t 20m 114 17 160
0.57 Press line crane 25t 22.86m 157 33 250
0.57 Standard 20t 20m 128.5 20 200
0.57 Standard 2,5t 11m 17.29 1.21 25
0.58 Standard 3,2t 11,2m 20 2.8 32
0.59 Standard 10t 15m span 31 3.5 50
0.60 Standard 2t 9m 12.08 1.12 20
0.61 Standard 3,2t 10m 17.7 2.8 32
0.61 Mould shop crane 65t 21m 300 116 650
0.62 Press line crane 20t 19m 99 25 200
0.62 Tool Repair crane 20t 18,76m 99 25 200
0.62 Roll shop crane 80t 14.5m 353 133 800
0.62 Standard 5t 14,5m 27.4 2.8 50
0.63 Standard 2t 7,5m 10.38 1.12 20
0.65 Standard 6,3t 14m 29 4.6 63
0.66 Standard 10t 15m span 43.4 7 100
0.67 Standard 3,2t 8m 13.1 2.8 32
0.67 Standard 5t 12.5m 22 2.8 50
0.67 Standard 2.5t 8m 11.09 1.21 25
0.67 Standard 6.3t 13m 25.8 4.6 63
0.68 Maintenance crane 10t 14.1m 37 9 100
0.71 Standard 5t 9.7m 17.8 2.8 50
0.71 Standard 6.3t 11m 20.8 4.6 63
0.73 Standard 3.2t 6m 9.2 2.8 32
0.74 Standard 8t 11.5m 23.7 4.6 80
327
Weight Weight Crane

kN kN kN
0.75 Standard 5t 8.3m 14.1 2.8 50
0.75 Standard 6.3t 8.5m 16.1 4.6 63
0.75 Standard 2.5t 5.5m 6.99 1.21 25
0.77 Standard 8t 9.5m 19 4.6 80
0.78 Standard 10t 10m 22.9 4.9 100
0.79 Standard 5t 6m 10.5 2.8 50
0.79 Standard 6.3t 6.5m 11.9 4.6 63
0.80 Standard 8t 8m 15.5 4.6 80
0.81 Standard 10t 8.5m 19.3 4.9 100
0.85 Standard 10t 6m 13.2 4.9 100
Appendix E
Code calibration results
E.1 Ultimate limit state - crane only

E.1.1 5t crane girder
5t crane girder, LC1, max pos vert moment, γH = 1 5t crane girder, LC1, max pos vert moment, γH = 1.261
7 5.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
6.5 5
6 4.5
5.5 4
5 3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Qh/(Qbr + Qcr + Qh) Qh/(Qbr + Qcr + Qh)
(a) Code format 1 (b) Code format 3
Figure E.1: LC1 maximum positive vertical moment, code formats 1 & 3
329
330 Appendix E. Code calibration results
5t crane girder, LC1, max pos vert moment, γCsw = 1.617, γH = 1 5t crane girder, LC1, max pos vert moment, γCsw = 1.354, γH = 1.271
5.7
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 γCh = 1.7
5.5
5.4 4
5.3
5.2
5.1
3.5
5
4.9
4.8
4.7 3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
5t crane girder, LC1, max neg vert moment, γH = 1 5t crane girder, LC1, max neg vert moment, γH = 1.261
7 5.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
6.5 5
6 4.5
5.5 4
5 3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure E.3: LC1 maximum negative vertical moment, code formats 1 & 3
5t crane girder, LC1, max neg vert moment, γCsw = 1.617, γH = 1 5t crane girder, LC1, max neg vert moment, γCsw = 1.354, γH = 1.271
5.7
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 γCh = 1.7
5.5
5.4 4
5.3
5.2
5.1
3.5
5
4.9
4.8
4.7 3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
E.1. Ultimate limit state - crane only 331
5t crane girder, LC1, max horz moment, γH = 1 5t crane girder, LC1, max horz moment, γH = 1.261
7 5.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
6.5 5
6 4.5
5.5 4
5 3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure E.5: LC1 maximum horizontal moment, code formats 1 & 3
5t crane girder, LC1, max horz moment, γCsw = 1.617, γH = 1 5t crane girder, LC1, max horz moment, γCsw = 1.354, γH = 1.271
5.7
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 γCh = 1.7
5.5
5.4 4
5.3
5.2
5.1
3.5
5
4.9
4.8
4.7 3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
5t crane girder, LC5, max pos vert moment, γH = 1 5t crane girder, LC5, max pos vert moment, γH = 1.261
7 6
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
5.5
6.5
5
4.5
5.5
4
5
3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
5t crane girder, LC5, max pos vert moment, γCsw = 1.617, γH = 1 5t crane girder, LC5, max pos vert moment, γCsw = 1.354, γH = 1.271
5.7
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 γCh = 1.7
5.5
5.4 4
5.3
5.2
5.1
3.5
5
4.9
4.8
4.7 3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
5t crane girder, LC5, max neg vert moment, γH = 1 5t crane girder, LC5, max neg vert moment, γH = 1.261
7 5.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
6.5 5
6 4.5
5.5 4
5 3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
5t crane girder, LC5, max neg vert moment, γCsw = 1.617, γh = 1 5t crane girder, LC5, max neg vert moment, γCsw = 1.354, γh = 1.271
5.7
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 γCh = 1.7
5.5
5.4 4
5.3
5.2
5.1
3.5
5
4.9
4.8
4.7 3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
5t crane girder, LC5, max horz moment, γH = 1 5t crane girder, LC5, max horz moment, γH = 1.261
7 5.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
6.5 5
6 4.5
5.5 4
5 3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
5t crane girder, LC5, max horz moment, γCsw = 1.617, γH = 1 5t crane girder, LC5, max horz moment, γCsw = 1.354, γH = 1.271
5.7
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 γCh = 1.7
5.5
5.4 4
5.3
5.2
5.1
3.5
5
4.9
4.8
4.7 3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

E.1.2 5t crane column
5t crane column, LC1, max horz, γH = 1 5t crane column, LC1, max horz, γH = 1.261
4.6 4.6
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
4.4
4.4 γC = 2 γC = 1.7
4.2
4.2
4
4
3.8
3.8 3.6
3.4
3.6
3.2
3.4
3
3.2
2.8
3 2.6
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure E.13: LC1 maximum horizontal, code formats 1 & 3
5t crane column, LC1, max horz, γCsw = 1.617, γH = 1 5t crane column, LC1, max horz, γCsw = 1.354, γH = 1.271
4.5 3.5
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
γCh = 2.05 3.4 γCh = 1.7
3.3
4
3.2
3.1
3
3.5
2.9
2.8
3 2.7
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Figure E.14: LC1 maximum horizontal, code formats 2 & 4

5t crane column, LC1, max vert, γH = 1 5t crane column, LC1, max vert, γH = 1.261
5 4.2
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
4
γC = 1.7
3.8
4.5 3.6
3.4
3.2
4 3
2.8
2.6
3.5 2.4
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure E.15: LC1 maximum vertical, code formats 1 & 3
5t crane column, LC1, max vert, γCsw = 1.617, γH = 1 5t crane column, LC1, max vert, γCsw = 1.354, γH = 1.271
4.4 3.3
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
4.3 γCh = 2.05 3.2 γCh = 1.7
3.1
4.2
3
4.1
2.9
4
2.8
3.9
2.7
3.8 2.6
3.7 2.5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Figure E.16: LC1 maximum vertical, code formats 2 & 4
5t crane column, LC5, max horz & vert, γH = 1 5t crane column, LC5, max horz & vert, γH = 1.261
4 4
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
3.8 γC = 2 3.8 γC = 1.7
3.6 3.6
3.4 3.4
3.2 3.2
3 3
2.8 2.8
2.6 2.6
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure E.17: LC5, code formats 1 & 3

5t crane column, LC5, max horz & vert, γCsw = 1.617, γH = 1 5t crane column, LC5, max horz & vert, γCsw = 1.354, γH = 1.271
3.25 3.3
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
3.2 γCh = 1.7
γCh = 2.05 3.2
3.15
3.1
3.1
3.05
3
3
2.9
2.95
2.9 2.8
2.85
2.7
2.8
2.75 2.6
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
40t crane girder, LC1, max horz, stress in bottom flange, γH = 1 40t crane girder, LC1, max horz, stress in bottom flange, γH = 1.261
7.5 6
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
7 5.5
6.5 5
6 4.5
5.5 4
5 3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
Figure E.19: LC1 maximum horizontal moment, bottom flange, code formats 1 &
3
40t crane girder, LC1, max horz, stress in bottom flange, γCsw = 1.617, γH = 1
40t crane girder, LC1, max horz, stress in bottom flange, γCsw = 1.354, γH = 1.271
5.9
γCh = 1.85 4.8
γ = 1.5
γCh = 1.963 Ch
5.8 γ = 1.623
γCh = 2.05 Ch
4.6 γCh = 1.7
5.7
4.4
5.6
5.5 4.2
5.4 4
5.3
3.8
5.2
3.6
5.1
3.4
5
4.9 3.2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
4
40t crane girder, LC1, max horz, stress in top flange, γH = 1 40t crane girder, LC1, max horz, stress in top flange, γH = 1.261
6.4 7
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
6.2 6.8
γC = 2 γC = 1.7
6.6
6
6.4
5.8
6.2
5.6
6
5.4
5.8
5.2
5.6
5
5.4
4.8 5.2
4.6 5
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
Figure E.21: LC1 maximum horizontal moment, top flange, code formats 1 & 3
40t crane girder, LC1, max horz, stress in top flange, γCsw = 1.617, γH = 1 40t crane girder, LC1, max horz, stress in top flange, γCsw = 1.354, γH = 1.271
6 5
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.8 γCh = 2.05 4.8
γCh = 1.7
5.6 4.6
4.4
5.4
4.2
5.2
4
5
3.8
4.8
3.6
4.6 3.4
4.4 3.2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
40t crane girder, LC5, max horz, stress in bottom flange, γH = 1 40t crane girder, LC5, max horz, stress in bottom flange, γH = 1.261
7.5 6.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
7 6
6.5 5.5
6 5
5.5 4.5
5 4
4.5 3.5
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
3
40t crane girder, LC5, max horz, stress in bottom flange, γCsw = 1.617, γH = 1
40t crane girder, LC5, max horz, stress in bottom flange, γCsw = 1.354, γH = 1.271
5.9
γCh = 1.85 5
γCh = 1.5
γCh = 1.963
5.8 γCh = 1.623
γCh = 2.05
4.8 γCh = 1.7
5.7
5.6 4.6
5.5
4.4
5.4
5.3 4.2
5.2
4
5.1
3.8
5
4.9 3.6
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
4
40t crane girder, LC5, max horz, stress in top flange, γh = 1 40t crane girder, LC5, max horz, stress in top flange, γh = 1.261
5.6 5.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
5.4
γC = 2 γC = 1.7
5.2 5
5
4.8 4.5
4.6
4.4 4
4.2
4 3.5
3.8
3.6 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
40t crane girder, LC5, max horz, stress in top flange, γCsw = 1.617, γH = 1 40t crane girder, LC5, max horz, stress in top flange, γCsw = 1.354, γH = 1.271
4.5 4.4
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
γCh = 2.05 4.3
4.4 γCh = 1.7
4.2
4.3
4.1
4
4.2
3.9
4.1
3.8
4 3.7
3.6
3.9
3.5
3.8 3.4
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
40t crane girder, LC1, max vert, stress in bottom flange, γH = 1 40t crane girder, LC1, max vert, stress in bottom flange, γH = 1.261
7.5 6
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
7 5.5
6.5 5
6 4.5
5.5 4
5 3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
Figure E.27: LC1 maximum vertical moment, bottom flange, code formats 1 & 3
40t crane girder, LC1, max vert, stress in bottom flange, γCsw = 1.617, γH = 1
40t crane girder, LC1, max vert, stress in bottom flange, γCsw = 1.354, γH = 1.271
5.9
γCh = 1.85 4.6
γCh = 1.5
γCh = 1.963
5.8 γCh = 1.623
γ = 2.05
Ch
4.4 γCh = 1.7
5.7
5.6 4.2
5.5
4
5.4
5.3 3.8
5.2
3.6
5.1
3.4
5
4.9 3.2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
40t crane girder, LC1, max vert, stress in top flange, γH = 1 40t crane girder, LC1, max vert, stress in top flange, γH = 1.261
7 6.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 6 γC = 1.7
6.5
5.5
6
5
4.5
5.5
5
3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
Figure E.29: LC1 maximum vertical moment, top flange, code formats 1 & 3
40t crane girder, LC1, max vert, stress in top flange, γCsw = 1.617, γH = 1 40t crane girder, LC1, max vert, stress in top flange, γ = 1.354, γ = 1.271
Csw H
6 5
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
γCh = 2.05 4.8
γCh = 1.7
4.6
4.4
5.5
4.2
3.8
5
3.6
3.4
3.2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
40t crane girder, LC5, max vert, stress in bottom flange, γH = 1 40t crane girder, LC5, max vert, stress in bottom flange, γH = 1.261
7.5 6.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
7 6
6.5 5.5
6 5
5.5 4.5
5 4
4.5 3.5
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
40t crane girder, LC5, max vert, stress in bottom flange, γCsw = 1.617, γH = 1
40t crane girder, LC5, max vert, stress in bottom flange, γ = 1.354, γ = 1.271
5.9 Csw H
γCh = 1.85 5
γCh = 1.5
γCh = 1.963
5.8 γCh = 1.623
γCh = 2.05
4.8 γCh = 1.7
5.7
5.6 4.6
5.5
4.4
5.4
5.3 4.2
5.2
4
5.1
3.8
5
4.9 3.6
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
04t crane girder, LC5, max vert, stress in top flange, γH = 1 04t crane girder, LC5, max vert, stress in top flange, γH = 1.261
6 5.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
5.5 5
5 4.5
4.5 4
4 3.5
3.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
40t crane girder, LC5, max vert, stress in top flange, γCsw = 1.617, gammaH = 1
40t crane girder, LC5, max vert, stress in top flange, γCsw = 1.354, gammaH = 1.271
4.7
γCh = 1.85 4.5
γCh = 1.5
γCh = 1.963
γCh = 1.623
4.6 γ = 2.05 4.4
Ch γCh = 1.7
4.3
4.5
4.2
4.4
4.1
4.3 4
4.2 3.9
3.8
4.1
3.7
4
3.6
3.9 3.5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Q /(Q + Q + Q ) Q /(Q + Q + Q )
h br cr h h br cr h
40t crane column, LC1, γH = 1 40t crane column, LC1, γH = 1.261

4.8
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
4.6
γC = 2 γC = 1.7
4.4
5
4.2
4
3.8
4.5
3.6
3.4
3.2
4 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
40t crane column, LC1, γCsw = 1.617, γH = 1 40t crane column, LC1, γCsw = 1.354, γH = 1.271
4.7 4
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
4.6 3.9
γCh = 2.05 γCh = 1.7
4.5 3.8
4.4 3.7
4.3 3.6
4.2 3.5
4.1 3.4
4 3.3
3.9 3.2
3.8 3.1
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

40t crane column, LC5, γH = 1 40t crane column, LC5, γH = 1.261

4.5 4.5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
4 4
3.5
3 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
40t crane column, LC5, γCsw = 1.354, γH = 1.271

40t crane column, LC5, γ = 1.617, γ = 1
Csw H 3.8
3.6 γCh = 1.5
γCh = 1.85 γCh = 1.623
γ = 1.963 γCh = 1.7
3.55 Ch
γ = 2.05
Ch
3.5
3.6
3.45
3.4
3.35
3.4
3.3
3.25
3.2
3.2
3.15
3.1 0 0.2 0.4 0.6 0.8 1

0 0.2 0.4 0.6 0.8 1
Qh/(Qbr + Qcr + Qh)
Qh/(Qbr + Qcr + Qh)

260t crane girder, LC1, maximum horizontal, γH = 1 260t crane girder, LC1, maximum horizontal, γH = 1.261
6.4 5.2
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
6.2 5
γC = 2 γC = 1.7
4.8
6
4.6
5.8
4.4
5.6
4.2
5.4
4
5.2
3.8
5
3.6
4.8 3.4
4.6 3.2
0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8
Figure E.39: LC1 maximum horizontal force, code formats 1 & 3
260t crane girder, LC1, maximum horizontal force, γCsw = 1.617, γH = 1 260t crane girder, LC1, maximum horizontal force, γCsw = 1.354, γH = 1.271
5.7
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 γCh = 1.7
5.5
5.4 4
5.3
5.2
5.1
3.5
5
4.9
4.8
4.7 3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

260t crane girder, LC1 maximum vertical force, γH = 1 260t crane girder, LC1 maximum vertical force, γH = 1.261
7 6
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
5.5
6.5
5
6
4.5
5.5
4
5
3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure E.41: LC1 maximum vertical force, code formats 1 & 3
260t crane girder, LC1, maximum vertical force, γCsw = 1.617 260t crane girder, LC1, maximum vertical force, γCsw = 1.354
5.7
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 γCh = 1.7
5.5
5.4 4
5.3
5.2
5.1
3.5
5
4.9
4.8
4.7 3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
260t crane girder, LC5, maximum horizontal force, γH = 1 260t crane girder, LC5, maximum horizontal force, γH = 1.261
7 6
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
5.5
6.5
5
4.5
5.5
4
5
3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

260t crane girder, LC5, maximum horizontal force, γCsw = 1.617, γH = 1 260t crane girder, LC5, maximum horizontal force, γCsw = 1.354, γH = 1.271
5.7 4.6
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 4.4 γCh = 1.7
5.5
5.4 4.2
5.3
4
5.2
3.8
5.1
5 3.6
4.9
3.4
4.8
4.7 3.2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
260t vrane girder, LC5, maximum vertical force, γH = 1 260t vrane girder, LC5, maximum vertical force, γH = 1.261
7 6
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
γC = 2 γC = 1.7
5.5
6.5
5
4.5
5.5
4
5
3.5
4.5 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
260t crane girder, LC5, maximum vertical force, γCsw = 1.617, γH = 1 260t crane girder, LC5, maximum vertical force, γCsw = 1.354, γH = 1.271
5.7 4.6
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
5.6
γCh = 2.05 4.4 γCh = 1.7
5.5
5.4 4.2
5.3
4
5.2
3.8
5.1
5 3.6
4.9
3.4
4.8
4.7 3.2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

E.1.6 260t crane auxiliary girder
260t auxiliary girder, LC1, maximum horizontal force, γH = 1 260t auxiliary girder, LC1, maximum horizontal force, γH = 1.261
4.5 5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
4.4
γC = 2 4.8 γC = 1.7
4.3
4.2 4.6
4.1
4.4
4
4.2
3.9
3.8 4
3.7
3.8
3.6
3.5 3.6
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Auxiliary girder, LC1, maximum horizontal force, γ = 1.617 γ = 1 Auxiliary girder, LC1, maximum horizontal force, γ = 1.354 γ = 1.271
Csw H Csw H
3.65
γCh = 1.85 γCh = 1.5
3.35
γ = 1.963 γ = 1.623
Ch Ch
γ = 2.05 γ = 1.7
Ch Ch
3.6
3.3
3.55
3.25
3.5
3.2 3.45
3.4
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

260t auxiliary girder, LC5, maximum horizontal force, γH = 1 260t auxiliary girder, LC5, maximum horizontal force, γH = 1.261
4.4 4.8
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
4.6
4.2 γC = 2 γC = 1.7
4.4
4
4.2
3.8
4
3.6
3.8
3.4
3.6
3.2
3.4
3 3.2
2.8 3
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
260t auxiliary girder, LC5, maximum horizontal force, γCsw = 1.617, γH = 1 260t auxiliary girder, LC5, maximum horizontal force, γCsw = 1.354, γH = 1.271
3.6
γCh = 1.85 3.8
γCh = 1.5
γCh = 1.963 γ = 1.623
Ch
3.5 γCh = 2.05 3.7 γCh = 1.7
3.6
3.4
3.5
3.3
3.4
3.2
3.3
3.1
3.2
3
3.1
2.9 3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

260t crane column, LC1, max horz, γH = 1 260t crane column, LC1, max horz, γH = 1.261
5.6 5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
5.4
γC = 2 γC = 1.7
5.2 4.5
5
4.8 4
4.6
4.4 3.5
4.2
4 3
3.8
3.6 2.5
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
260t crane column, LC1, max horz, γCsw = 1.617, γH = 1 260t crane column, LC1, max horz, γ = 1.354, γ = 1.271
Csw H
4.5 3.4
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
3.3
4.4 γCh = 2.05 γCh = 1.7
3.2
4.3
3.1
4.2
3
4.1
2.9
4
2.8
3.9
2.7
3.8 2.6
3.7 2.5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

260t crane column, LC1, max vert, γH = 1 260t crane column, LC1, max vert, γH = 1.261
5.6 4.4
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
5.4 4.2
γC = 2 γC = 1.7
5.2 4
5 3.8
4.8 3.6
4.6 3.4
4.4 3.2
4.2 3
4 2.8
3.8 2.6
3.6 2.4
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
260t crane column, LC1, max vert, γCsw = 1.617, γH = 1 260t crane column, LC1, max vert, γCsw = 1.354, γH = 1.271
4.5 3.3
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
4.4 γCh = 2.05 3.2 γCh = 1.7
4.3 3.1
4.2 3
4.1 2.9
4 2.8
3.9 2.7
3.8 2.6
3.7 2.5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
260t crane column, LC5, max horz & vert, γH = 1 260t crane column, LC5, max horz & vert, γH = 1.261
5.2 5
γC = 1.8 γC = 1.5
γC = 1.907 γC = 1.59
5
γC = 2 γC = 1.7
4.5
4.8
4.6
4
4.4
4.2
3.5
4
3.8
3
3.6
3.4 2.5
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

E.2. Fatigue 351
260r crane column, LC5, max horz & vert, γCsw = 1.617, γH = 1 260r crane column, LC5, max horz & vert, γCsw = 1.354, γH = 1.271
4.3 3.7
γCh = 1.85 γCh = 1.5
γCh = 1.963 γCh = 1.623
4.2 γCh = 2.05 3.6 γCh = 1.7
4.1 3.5
4 3.4
3.9 3.3
3.8 3.2
3.7 3.1
3.6 3
3.5 2.9
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
E.2 Fatigue
E.2.1 5t crane corbel to column welded connection

F
Nmin, kQmin
Nmin, kQmax
2.2
Nmax, kQmin
Nmax, kQmax
2
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10
Figure E.57: 5t crane corbel to column connection, load spectrum class Q2

F
Nmin, kQmin
Nmin, kQmax
2.2
Nmax, kQmin
Nmax, kQmax
2
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10


F
Nmin, kQmin
Nmin, kQmax
2.2
Nmax, kQmin
Nmax, kQmax
2
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10

F
Nmin, kQmin
Nmin, kQmax
2.2
Nmax, kQmin
Nmax, kQmax
2
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10
E.2.2 40t crane girder bottom of intermediate stiffener

2.2
Nmin, kQmin
Nmin, kQmax
2 Nmax, kQmin
Nmax, kQmax
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10
Figure E.61: 40t crane girder intermediate stiffener, load spectrum class Q2
E.2. Fatigue 353

2.2
Nmin, kQmin
Nmin, kQmax
2 Nmax, kQmin
Nmax, kQmax
1.8
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10

2.2
Nmin, kQmin
Nmin, kQmax
2 Nmax, kQmin
N , kQ
max max
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10

2.2
Nmin, kQmin
Nmin, kQmax
2 Nmax, kQmin
Nmax, kQmax
1.8
1.6
1.4
1.2
0.8
0.6
4 5 6 7
10 10 10 10
E.2.3 40t crane girder top flange to web weld

2.5
Nmin, kQmin
Nmin, kQmax
Nmax, kQmin
2 Nmax, kQmax
1.5
0.5
0
4 5 6 7
10 10 10 10
Figure E.65: 40t crane girder top flange to web weld, load spectrum class Q2

2.5
Nmin, kQmin
Nmin, kQmax
Nmax, kQmin
2 Nmax, kQmax
1.5
0.5
0
4 5 6 7
10 10 10 10

2.5
Nmin, kQmin
Nmin, kQmax
Nmax, kQmin
2 Nmax, kQmax
1.5
0.5
0
4 5 6 7
10 10 10 10
E.2. Fatigue 355

2.5
Nmin, kQmin
Nmin, kQmax
Nmax, kQmin
2 Nmax, kQmax
1.5
0.5
0
4 5 6 7
10 10 10 10

Www.sefindia.org

Uploaded by

Copyright:

Available Formats

Www.sefindia.org

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Www.sefindia.org

Uploaded by

Copyright:

Available Formats

Reliability based codification for the design of overhead

travelling crane support structures

Juliet Sheryl Dymond

Dissertation presented for the degree of

Doctor of Philosophy in Engineering

Department of Civil Engineering

Reliability based codification for the design of overhead

Dissertation: PhD (Engineering)

Reliability based codification for the design of overhead

Proefskrif: PhD (Ingenieurswese )

Elektriese oorhoofse brugkrane vorm ’n ge-integreerde deel van baie nywer-

• My supervisors, Prof PE Dunaiski and Prof JV Retief for their academic

• My friend Michele van Rooyen, for all her encouragement

• My husband, Jacques, for always believing in me and supporting me and

2 Design of crane support structures 11

3 Scope of the code 59

4 Calibration to current practice 81

5 Development of limit states equations 103

6 Stochastic models 129

6.7.1 Fatigue resistance . . . . . . . . . . . . . . . . . . . . . 150

7 Stochastic modelling of crane hoistload 157

8 Code calibration method 175

9 Code calibration results 193

9.2.1 Results of parametric studies . . . . . . . . . . . . . . . 221

10 Discussion of results 239

List of Figures 279

List of Tables 291

A Load calculations 295

B Load Effects 313

C Graphs for stochastic modelling of hoistload 317

D Ratios of hoistload to total crane weight 325

E Code calibration results 329

E.1.6 260t crane auxiliary girder . . . . . . . . . . . . . . . . . 347

1.1 Crane load models

1.2 Reliability of crane support structures

1.3 Code calibration of crane load models

1. Crane load models

The crane load models in SABS 0160:1989 are over-simplistic, so it is

An assessment of the prEN 1991-3 crane load models is required to in-

2. Reliability of the crane load models.

A reliability assessment of the crane load models is required to determine

Both of these aspects of crane induced loads that require investigation

1.3.1 Calibration to current practice

2. Cost of support structure

The amount of work required to calculate the factored load combinations

1.3.2 Reliability calibration

1. A reliability assessment of the two different sets of load models.

2. An investigation into different code formats.

3. Determination of partial load factors and combination factors.

1. Definition of the scope of the code

2. Definition of the code objective

assessment of recommended target reliabilities is carried out and suitable

3. Definition of the code format

4. Identification of typical failure modes

runway bracing. The design of these elements is carried out according

5. Development of stochastic models

6. Determination of the optimal partial load factors

7. Verification of partial load factors

Design of crane support