Design by Ansys
Design by Ansys
2001/001
CONTENTS
1.
Introduction
1.1
Background
1.2
Objectives
1.3
Report Layout
1
1
1
1
2.
3
3
3
4
4
4
4
3.
Loading Conditions
3.1
Load Cases
3.2
Load Combinations
5
5
6
4.
9
9
9
9
9
10
10
10
11
11
11
5.
13
13
17
18
6.
23
23
23
23
25
25
25
26
30
7.
Comparison of 5-5A and 5-5B Allowable Stresses and Implied Safety Factors
7.1
Basis of Comparison
7.2
Axial Compression
7.3
Bending
7.4
Combined Compression and Bending
7.5
Implied Safety Factors
31
31
31
34
39
43
(i)
8.
45
45
45
57
9.
Conclusions
9.1
Conclusions
61
61
REFERENCES
65
TABLES
FIGURES
(ii)
1. INTRODUCTION
1.1 BACKGROUND
This project relates to the document titled Practice for Site Specific Assessment of Mobile
Jack-Up Units Working Stress Design (WSD)(1). The issue status of this document is
reported First Edition May 1999, although within the document it is referred to as both Rev 0,
September 1999 and Rev 0, October 1999. In the remainder of this report, this document is
referred to as 5-5B.
This document offers a working stress design approach to the Site Specific Assessment of
Mobile Jack-Up Units as an alternative to the Load and Resistance Factor Design (LRFD)
approach adopted in the SNAME Bulletin 5-5 Guidelines and Bulletin 5-5A Recommended
Practice for Site Specific Assessment of Mobile Jack-Up Units(2).
In addition, reference is made to the application of SNAME Bulletins 5-5 and 5-5A to the North
Sea as reflected in the North Sea Annex to Bulletin 5-5A(3).
1.2 OBJECTIVES
The main objectives of this study were to compare the response and performance of a typical
jack-up rig deployed in the North Sea when initially assessed in accordance with the proposed
SNAME WSD Practice (5-5B), and then reassessed to both the existing SNAME LRFD Practice
(5-5A) and to the North Sea Annex to SNAME Bulletin 5-5A (NSA).
The jack-up platform considered in this project was the Marathon LeTourneau 116C. This
platform was employed at a typical southern/central North Sea location selected on the basis of
water depth and the availability of suitable environmental and soils data. The assessments were
performed on the basis of real location-specific data, provided by Shell Expro. This ensured
that the results were particularly pertinent to the North Sea and were consistent with earlier
work. Dynamic loading was treated in a quasi-static manner and the maximum spudcan fixity
was determined on the basis of specific soil parameters.
A secondary objective of this study was to determine the basis of the reliability analysis and
calibration process performed during the Joint Industry Study (JIS) that led to the partial load
and resistance factor format contained within the existing LRFD 5-5A Recommended Practice.
1.3 REPORT LAYOUT
The report is formatted as follows. Section 2 contains basic design data for the 116C, including
a general description of the jack-up platform and water levels, environmental data and soils data
at the selected North Sea location. Design load cases are presented in Section 3. This section
describes the platform weight and centre of gravity summaries for fixed, variable and live loads.
Description of the computer model and the analysis procedure are given in Section 4. All
aspects of the leg, hull, the leg-hull connection and the foundation modelling are discussed.
Section 5 presents a detailed critique of 5-5A, 5-5B and NSA practices. They are reviewed for
consistency and accuracy. Analysis results are presented in Section 6, covering natural
frequency, displacements, base shears and overturning moments and member utilisation ratios
for each assessment procedure considered. In Section 7, 5-5A and 5-5B strength formulations
have been compared and discussed on a one-to-one basis, accounting for all appropriate factors.
Section 8 presents a review of the jack-up reliability analysis and partial factor determination
work performed by Noble Denton Consultancy Services (NDCS) in connection with the Joint
Industry Study (JIS) on Jack-Up Assessment Methods. Conclusions are given in Section 9,
followed by references, tables and figures.
E = 200,000 MPa
G = 80,000 MPa
= 7,850 kg/m3
= 0.3
The density of seawater was taken to be 1025 kg/m3. For wind load calculations, the density of
air was taken as 1.22 kg/m3.
3. LOADING CONDITIONS
3.1 LOAD CASES
This section defines the loads considered for the assessment of the jack-up members and
foundations. Loading conditions related to extreme storm and preload were considered in this
assessment, covering:
dependent on whether the analysis is in accordance with 5-5A, 5-5B or the NSA.
Overturning moment checks were performed according to Section 8.2 of 5-5A. In Section 3.2.2
of this Practice, it is stated that for checks requiring minimum elevated weight, 50% of the
variable load may be applied. For overturning checks the corresponding factor, 2, was therefore
set to 0.5.
and stiff sections on the elements P- curve it was possible to accurately model the gaps
existing in the upper and lower guide shoes. From Figure 8a it is seen that the leg chord was
attached to three non-linear springs in order to allow sliding from front to back and from side to
side within the guide shoe. The non-linear springs were in turn connected to coincident stiff
beam members, representing the guide shoe that connected to the hull. The chord was free to
slide vertically within the guide shoe. The stiff beam members were given an axial stiffness
approximately 100 times the stiffness of a chord. This is consistent with the recommended
approach in 5-5A that states the guides should be relatively stiff, with freedom introduced by
means of gaps. The depth of the upper and lower guides was taken into account by using a
double prong connection at both the upper and the lower guide. Figure 8b shows a line diagram
of the guide arrangement.
4.5 MODELLING OF THE JACKING SYSTEM
Figure 9 shows a line diagram of the rack, pinion and gear unit case. Each pinion was modelled
by two non-linear spring elements. Both elements were given a nominal low stiffness in
tension, in order to achieve numerical stability. The upper pinion element was given a linear
compressive stiffness of 174 MN/m, based on data received from LeTourneau. Defining a
region of low stiffness in the material properties for the lower non-linear spring included the
backlash of three millimetres. The pinions were inclined at a pressure angle of 25o, the rack was
offset from the chord centroid and the gear unit case was offset from the edge of the yoke. It is
important to include these details to correctly capture the line of action of the forces and
associated moments. The gear unit case was modelled as a box section with properties based on
its construction. Similarly, the associated bracing was modelled as I-sections and orientated
correctly. Figure 10 shows the gear unit case, bracing and upper guide.
4.6 SPUDCAN AND FOUNDATION MODELLING
Typical soils data from the selected site was used to determine the values for parameters used in
modelling the spudcan fixity. Calculations showed that it was reasonable to assume a leg
penetration of 3.0 m. The vertical, lateral and rotational stiffnesses of the soil were calculated
based on the undrained shear strength of 200 kPa and a G/Su ratio of 50. By using spring to
ground elements it was possible to apply vertical, lateral and rotational stiffness values to the
node at the tip of each spudcan. The spudcans were modelled as beams, with a stiffness of
approximately 100 times the stiffness of the chord.
4.7 AIRGAP
An airgap of 16.52 m (measured from the nominal LAT to the underside of the hull) was
calculated in accordance with 5-5A Section 3.7.5. It is defined as the sum of the distance of the
extreme still water level above LAT, the extreme wave crest height and 1.5 m clearance.
In 5-5A it is recommended that two positions of the lower guide relative to the chord should be
investigated, one at a node and the other at midspan of a leg bay. The FE model hull was
elevated above the minimum airgap until the first node on the chord was located at the centre of
the lower guide shoe. A second FE model was generated with the hull elevated further so that
the centre of the lower guide shoe was located at the midspan of a bay.
The NSA requires airgap to be calculated using the more critical of 100-year return period
individual maximum wave, taken with a clearance of 1.5 m, and 10,000-year return period
individual extreme wave, taken with no clearance. In the case under investigation, the latter was
critical. The airgap of the second FE model, with guide shoe located at midspan of a bay
satisfied the 10,000-year criteria and so this model was used to investigate the NSA
requirements.
10
11
12
5.
SNAME Recommended Practice for Site Specific Assessment of Mobile Jack-up Units,
Revision 1 (hereafter referred to as 5-5A)
The review includes typographical errors, section layout, clarifications, corrections and
technical consistency with other documents. The findings are presented in the order in which
they appear in the document and are cross-referenced using page, section and equation numbers.
1. The issue status of 5-5B is reported First Edition May 1999, although within the document
it is referred to as both Rev 0, September 1999 and Rev 0, October 1999.
2. Section 1.2.1 The current guideline references .(typographical)
3. Contents -Step 3, Displacement check through to 8.7 Structure Condition Assessment are
not included in this document, and it is assumed that these remain unchanged from 5-5A.
4. Page 1 Section 8 Horizontal deflections (Section 8.4), The condition of the unit
(Section 8.7). are not contained in this document and are assumed to be the same as in 55A.
5. Page 1 .. allowable unity check .. calculated unity check are not correct phrases.
Unity check is defined as actual stress divided by allowable stress, and for code
compliance, this ratio should be less than or equal to unity. These phrases should be
deleted.
6. Page 1 recognised standard such as API RP 2A sections 2.3.6 c4 and 2.c.6e. The latter
of these sections should read 2.3.6e.
In 5-5B, this is the first time that API RP 2A is mentioned, so it is necessary to clarify the
version and whether LRFD or WSD. It is noted that this is in fact done on the following
page in Section 8.1.1.1.
7. Page 3 Figure 8.1 the flowchart commences with Member Strength Assessment using
Factored Loads. There are no factored loads.
13
8. Page 4 Section 8.1.2.1 a) i) The limits for R/t are in error and should read:
R/t > 1650/Fy or R/t > 30
{Metric: 11375/Fy}
Table B5.1 of AISC 9th Edition and Section 3.2.2a of RP 2A-WSD present the tubular
limits in terms of the diameter to thickness ratio (D/t). It is clear that the error has resulted
from converting the D/t ratio to a radius to thickness (R/t) ratio.
9. In Section 8.1.2.2 5-5B states Typically all the components forming parts of chord sections
may be regarded as stiffened. It is not clear why the outstands of the back plate component
should be treated as stiffened along both edges, as opposed to stiffened along one edge
only. Similarly, the rack component ought to be considered as stiffened along one edge
only. It is noted that this clause also appears in the 5-5A.
10. Page 6 Section 8.1.4.1a in the Note states that when local element P-Delta effects are
included in a three dimensional analysis of the entire jack-up, the Euler modification in
H1-1 may be assumed equal to 1.0 for chord elements only. Other elements shall include
the Euler modification unless independent analysis is used to demonstrate that local node
to node Euler effects have been included in the element load level.
A similar clause is given in 5-5A Section 8.1.4.5 ii). Here it is stated that where the
individual member loads are determined from a second order analysis, ie the equilibrium
conditions were formulated on the elastically deformed structure so that local P- loads
were also included in the analysis B = 1.0, where
In the latter case, Cm is also set to unity, provided the conditions are satisfied. However, 5-
B=
Cm
PU
1
PE
5B permits only the Euler modification to be set to unity. Furthermore, this code clearly
states that this is permitted for chord sections only. For all other elements the Euler
modification must generally be included. This disparity gives rise to an inconsistent
approach between the two practices.
11. Page 6 Section 8.1.4.1a 2). In the definition of Fb and Fe ksi units are specified. Metric
units may also be used to determine these parameters so ksi should be deleted.
12. Page 6 Axial Tension and Bending Prismatic Section. This section has been positioned
within Section 8.1.4.1a Axial Compression and Bending Prismatic Sections. It would be
clearer to move this section to after Equation (3.3.1-4) on the following page, or provide a
separate section number to Axial Tension and Bending.
13. Page 7 Note: Where tubular elements are uses(d) as .(typographical).
This note also permits the Euler amplification to be set to unity under the same conditions
discussed in item 10, above. The arguments of 10 apply equally here.
14. Section 8.1.4.2 Allowable Axial Strength of a . Replace strength with stress. Also, the
word tensile should have a leading capital and the t in Ft should be a subscript.
15. Section 8.1.4.3 In the title, the a in Fa should be a subscript.
16. Section 8.1.4.3 iii) Change 11,380 to 11,375 for consistency with Section 8.1.2.1 iii).
14
17. Section 8.1.4.3 iv) The limit for R/t is in error and should read:
R / t 30
18. Page 10 Section 8.1.4.3b) 2) 2. Inelastic local Buckling Stress. The equation given here
(API 3.2.2-4) is fundamentally flawed in that it is not correctly non-dimensionalised. To
achieve this, Fy and D/t need to be combined in a manner not dissimilar to that found in the
corresponding equation (see 3.2.3-1) on page 13 of FyD / Et. This has been done more
rigorously for the local buckling check in the proposed ISO fixed steel structures
formulations. Here the equations are given as:
Fxc = Fy
= Fxe
12 2 E
F =
23( Kl b / rb ) 2
'
e
21. 8.1.4.5 ii) This refers to a second order analysis in which local P- loads are included in
the analysis. This section on compression is not consistent with the corresponding
paragraph on tension. A weakness was noted in 19. with respect to the requirements for
tension. A similar weakness exists here in that it refers to local P- loads but not local
loading. In both cases, global and local P- effects as well as local loading are all to be
included.
22. Page 12 Section 8.1.4.6 The limiting equation for lateral torsional buckling is incorrect. It
should read:
Lb / ry 25860
JA
Mp
Similarly in the metric equivalent, the square root should be applied only to the numerator.
23. Page 12, 2nd sentence under Check: Local buckling Insert , the between control and
allowable.
24. In the second paragraph to this section, the need to differentiate between flange and web
elements is indicated. Descriptions follow as to how this might be applied, for example, to
split tubular flanges and tear drop chords. The problem with this is that the descriptions
only relate to bending about one axis (the major axis). It is also necessary to consider
minor axis bending at which case, the webs and flanges interchange. Such distinction
would be necessary if the slenderness (b/t) was to be limited on the grounds of which stress
15
pattern the component was subjected. For example, a web subjected to shear only will
normally have a different slenderness limit to one subjected only to flexure. The limits on
slenderness that follow are not limited on this basis, but on the basis that they form part of
conventional hybrid girders. This is discussed below but the chords of jack-up legs are not
conventional hybrid girders.
25. Page 13 Section 8.1.4.6 a ii) The metric version of the equation is incorrect. It should read:
bi / t i 170 /
(F )
yi
60
167
80
70.5
100
36.1
These limits have been determined for unfactored loads and resistances. Smaller depths
clearly apply once partial factors are introduced. A statement is required concerning
consideration of hydrostatic pressure.
30. Pages 15 and 16 Section 8.2.3 At two locations within this section it is noted that factored
loads are used. There are no factored loads in 5-5B.
16
3.
Page 5 C3.5.2 Freak should not be used. It requires clear acknowledgement that the
Most Probable Maximum (MPM) wave used can, for example, be exceeded 63% of the
time.
4.
Page 7 Section 2.3.2 confirms the requirement of 50 year one hour wind see comment 1.
above.
5.
Page 7 Section 3.2.1: Should this be constructed in terms of crest height rather than wave
height because one has to allow for storm surge, tides, etc. Is it clear where wave height
fits with LAT? Some specific advice is given in Section 3.5.2 of the Practice.
6.
Page 8 Section 4.1.1 notes three environmental loading options for North Sea applications
(note, however, that a fourth option is to use the 5-5A requirements, as noted in Section 1.2
of the NSA). One of these options covers the case where the wind loads exceed the
wave/current plus dynamic loads. This implies that some form of prior load analysis is
required to determine whether the wind loads are in excess of those from
wave/current/dynamic loads.
Section 3.5.1.2 quantifies the kinematics reduction factor. In the absence of suitable
software, the factor can instead be applied to reduce the considered wave height. The
results of these two approaches might not be dissimilar if the wetted heights are the same.
7.
17
8.
Page 12 Section 3.5.2 refers to 5-5A Section 4.4.1 for determining an appropriate wave
theory for the determination of the wave crest elevation. It is noted that Figure 4.1 of 5-5A
has an error in one of the y-axis value (0.05 should read 0.005).
9.
Page 12 Section 3.5.2 presents a relationship between the 10000-year Hsrp and the 100-year
Hsrp. A factor of 1.31 is quoted. This factor has been confirmed by using the formulations
given in Table 11.8 of the, now superseded, UK Department of Energy Offshore
Installations: Guidance on design, construction and certification, Fourth Edition.
In Rev 0 of the 5-5B code, changes from the 5-5A assessment practice are proposed in Sections
8.3.1 to 8.3.3, while no amendments are proposed to any part of Section 6, Section 8.3.4 Step 3
Displacement check, or Section 8.3.5 Punch-through.
Sections 8.3.1 to 8.3.3 cover the following aspects of foundation assessment:
Section 8.3.1 Step 1 - Preload and Sliding Checks
Section 8.3.2 Step 2a - Capacity Check - pinned foundation
Section 8.3.3 Step 2b - Capacity Check - with foundation fixity
Details of the differences between the 5-5A and 5-5B codes are detailed and discussed in
Section 5.3.2 of this report.
5.3.1 Foundation Assessment Methodology
Prior to describing the differences between the two versions of the jack-up code, it is necessary
to review the approach adopted for assessing foundation integrity.
There are three areas of geotechnical concern. Of these three areas, only the second area: jackup foundation stability after preloading differs between the 2 versions of the code and will be
considered further.
5.3.1.1 Prediction of Footing Penetrations During Preloading
The assessment of spudcan load/penetration behaviour generally follows three steps:
Compute the vertical bearing capacity of the footing at various depths below the seabed
using closed form bearing capacity solutions and plot as a curve.
Enter the vertical bearing capacity versus footing penetration curve with specified
maximum preload and read off the predicted footing penetration.
18
19
The effect of removing these factors is illustrated in Figure 12. It should be noted that on the
leeward leg it is assumed that the vertical forces due to environmental effects always act in the
same (compressive) manner as the forces due to dead and live loads.
It can be seen that where there is no environmental or inertial load (E + Dn = 0, 0% on the xaxis), the 5-5A version is 1.00/0.90 = 11% more conservative than the 5-5B version of the code
due to the inclusion of this resistance factor. This percentage could theoretically increase to a
maximum of 1.25/0.90 = 39% for no dead and live load (100% on the x-axis).
5.3.2.2 Step 1b: Sliding Resistance Windward Leg(s)
Section 8.3.1.5
Removal of load factors from horizontal leg reaction, QH (3 = 1.25, 4 = 1.0)
in association with QV (1 = 2 = 4 = 1.0, 3 = 1.25)
Removal of resistance factor for horizontal foundation capacity:
Hfc = 0.80 effective stress sand drained
Hfc = 0.64 total stress clay undrained
The effect of removing these factors is illustrated on Figure 13. It should be noted that on the
windward leg it is always assumed that the vertical forces due to environmental & inertial (E +
Dn) effects always act in the opposite (tensile) manner to the forces due to dead and live loads.
It can be seen that there is a non-linear relationship between the proportion of E + Dn loading
and the degree of conservatism.
For a sand foundation, where there is no E + Dn load in the vertical direction (0% on the xaxis), the 5-5A version is 1.25/0.80 = 56% more conservative than the 5-5B version of the code.
The 1.25 in this case relates to the relative increase in the horizontal load, QH = 1.25 x (HE +
1.00.HDn), which applies to all QH values irrespective of the percentage horizontal E + Dn load.
The equivalent degree of conservatism for clay soils for no environmental or inertial load in the
vertical direction is 1.25/0.64 = 95%.
For the case where the E + Dn load is 40% of the total absolute load on the windward leg, this
percentage would increase to 213% in sand and 291% in clay soil. The curve tends to infinity
as the (D + L) (E + Dn) =0, i.e. at 50% E + Dn. Although these percentage differences
appear large, they reflect the fact that the tensile environmental & inertia forces are becoming
equivalent to the compressive dead and live loads, giving a combined total load tending to zero.
e.g.
VD
VL
VE
VDn
= 25
= 5
= -15
= -5
MN
MN
MN
MN
VE + VDn
20MN
=
= 40% E + Dn
VD + VL + VE + VDn
50MN
Qv_5-5A = 25 + 5 + 1.25 x ((-15) + (-5)) = 5.0
= 10.0
Qv_5-5B = 25 + 5 + (-15) + (-5)
Difference in Q H :
LRFD
1.25
=
= 3.13 (in sand = 0.8)
WSD 0.5x 0.8
20
jack-up code. The one area of potential confusion being associated with the high percentage
differences between the codes on the windward leg(s) as the proportion of Environmental +
Inertial load approaches 50% and the total load approaches zero.
Therefore, for a specific foundation model, it is relatively easy to follow the path through the
three steps to verify the acceptability of the foundation stability after preload, as illustrated on
Figure 11. This path is summarised on Table 7.
However, because of the large number of variables in both load and resistance and the steps
taken towards foundation verification, it is difficult to combine these differences at each step to
obtain an overall generalised description of the difference between the two versions of the code.
Such a comparison could be achieved by performing Monte-Carlo simulations for a range of
possible loads, jack-up platform types and soil conditions. This exercise would require an
investigation into the range and frequency of such variables so that appropriate probability
distributions can be attributed to each variable. Due to the considerable effort required to
adequately describe these distributions, such an exercise is considered to be outside the scope of
the current project.
5.3.3 Conclusions
A comparison between the 5-5A and 5-5B versions for the jack-up code has been performed
with regard to verifying the acceptability of the foundations.
The only difference in the 5-5B version of the code from the 5-5A version is the removal of the
load and resistance factors. Since these factors are 1.0 on the loads and 1.0 on the resistance
terms the 5-5A will always be more conservative than the 5-5B code for loads acting in the
same direction. For the situation where the environmental and inertial loads are acting in the
opposite sense to the dead and live loads the situation becomes more complex as the load factor
of 1.25 on the environmental and inertial loads can lead to a reduction in the total load. In such
instances the 5-5B code can be more conservative than the 5-5A code, depending upon the
resistance factor applied.
The analysis procedure covers several steps of increasing complexity in the foundation model
with these steps employing different loads (e.g. horizontal, vertical, and the load vector with and
without the associated can moment). Differences between the codes are relatively easy to
quantify at each step. However, the large number of variables in both load and resistance, and
the uncertainty in the number of steps taken towards foundation verification, make it difficult to
generalise these differences to obtain an overall description of the difference between the two
versions of the code. This could be achieved by performing a Monte-Carlo analysis, although
such an exercise would require considerable effort to determine appropriate probability
distributions to the numerous variables.
22
see Figure 3
see Figure 3
324 mm diameter x 19.1 mm wt
324 mm diameter x 19.1 mm wt
229 mm diameter x 9.5 mm wt
324 mm diameter x 25.4 mm wt
324 mm diameter x 25.4 mm wt
The structural analysis software used in this study was a non-linear large displacement package,
and took into account the global and local P- loads. Therefore in line with the 5-5A practice,
the coefficient B, which accounts for the load reduction factor (Cm) and the Euler amplification,
was set to unity. In the working stress design practice 5-5B, a similar clause permits setting the
Euler amplification to unity, but not Cm. For a consistent comparison in this study, Cm for all
members was also set to unity. In the various checks for compactness, no tube members were
classed slender.
The chord elastic and plastic section properties were determined using standard techniques.
Account was taken of the different yield strength of the rack components. In 5-5A it is stated
that all components for a chord may be assumed to be stiffened along both edges. However, this
was considered inappropriate for outstands of the guide plate and for the rack component. For
these components slenderness checks were based on b/t limits for a component stiffened along
one edge only. Both 5-5A and 5-5B include this statement, and therefore consistency was
maintained.
23
For the reinforced chord, in addition to conducting local buckling checks for each component,
for the stiffened side plates it was considered prudent to check the combined section comprising
side plate and side plate reinforcement for local buckling. To perform this check, an equivalent
plate thickness was calculated on the basis of equivalent second moment of area and the
buckling formulations for stiffened plates. This resulted in an equivalent plate thickness of 40.9
mm.
The code check formulations for each assessment were prepared for tube and chord sections.
Where appropriate, load and resistance factors were applied in accordance with the practice
requirements. In particular, for working stress design checks, a one-third increase in allowable
stresses was incorporated for the storm load combinations.
Every leg element was code checked for each combination. Every wave direction was
considered together with the phase angle for a full wave cycle. For each assessment, code
checks were made for the position of the lower guide adjacent to the leg node and for the lower
guide at midspan of the chord section. The variations noted above resulted in a very large
number of code check results. The critical utilisation ratios (UR) were identified for each major
group and these are reported in Tables 9 to 11. Table 9 covers the results for 5-5A, Table 10 for
5-5B. The tables also present the ratio of maximum UR for lower guide at midspan to the UR
for lower guide at midspan. The results for the NSA are presented in Table 11. In the case of
the NSA, only the analysis related to the lower guide adjacent to midspan was conducted, based
on the results of the earlier work.
The results for checks to 5-5A (Table 9) show that the maximum UR for the reinforced chord, in
the region of the lower guide, is 1.86, considerably in excess of unity. Table 9 also indicates that
all other element groups exhibit maximum URs below unity. It is evident that the position of
the lower guide in relation to the chord section can make a difference of up to 6% in UR. In
some cases the midspan location is critical, in others, the node position. Therefore, both need to
be checked for an accurate assessment of the structure. Figures 19 and 20 show the locations of
the critical members in each group, for the lower guide position adjacent to the chord node and
adjacent to the chord midspan, respectively.
Results for assessment to the working stress design practice, 5-5B, are presented in Table 10.
Examination of the results in this table shows a similar pattern to the results for 5-5A although
the largest utilisations are marginally lower than for 5-5A. The midspan to node results ratios
are similar to those for 5-5A with the largest variation amounting to 7%.
The North Sea Annex results are presented in Table 11. A comparison between the 5-5A and
the NSA results shows that the 100 year joint probability environmental loading specified in the
NSA results in, on average, a 5% reduction in UR compared to the LRFD assessment procedure
of 5-5A.
In the case of 5-5A and NSA, chord stress checks have been performed for a number of
variations of the value of , the exponent for biaxial bending.==In the case of 5-5A, = 1 has
been used as the default value, resulting in a maximum UR of 1.86. By exploiting the
information presented in the Commentary to 5-5A, a minimum value for of 1.45 can be used.
This leads to a reduced UR of 1.50. From the Recommended Practice, a procedure for
determining a more accurate value for is provided. Following this procedure leads to an
value of 2.25, and a corresponding maximum UR of 1.27. A similar pattern of reducing UR is
apparent when the NSA is used.
It is noted that values of increasingly greater than unity lead to dramatic reductions in
utilisation. For the present 5-5A case, increasing from 1.0 to 2.25 leads to a reduction in
utilisation from 1.86 to 1.27, equivalent to some 32%.
For the reinforced chord checks to 5-5B, using the yield value for the rack, which coincides with
the minimum section modulus, and following the instruction given in Figure 8.2 of 5-5B to
24
check the stress at the extreme rack fibre (ignoring the teeth) leads to a UR value of 1.74. The
corresponding UR for the unreinforced chord is 0.74.
Clearly, the effects of on the maximum UR are significant for those chord members subjected
to high bending moments. It is known that significant bending occurs in the leg sections
particularly close to the lower guides. Accurate determination of is important since an underestimation leads to an unnecessarily conservative utilisation ratio, whilst an over-estimation
quickly leads to non-conservative utilisation ratios.
Table 12 presents the comparisons of maximum URs for each group of members. Examination
of the results of this table show that in general, the working stress design assessment procedure
leads to UR some 10% lower than the LRFD assessment procedure.
6.4 LOAD FACTOR APPLICATION TO LOAD EFFECTS OR LOAD PATTERNS
In this context, load effects are defined as the internal member forces arising from the analysis.
Load patterns are defined as load sets to be applied to the analysis. It is not clear from 5-5A
whether load factors are to be applied to load effects or load patterns. The impression is that the
factors are applied to load effects (reference Commentary to 5-5A Section C8.0.1). However,
this contrary to normal LRFD practice and is considered incorrect. Significant differences in the
distribution of permanent, variable and environmental load effects can be expected when
unfactored load patterns are applied compared with when factored load patterns are applied to
the structural model. These differences are a direct result of the non-linearities in response.
The method of application of load factors was investigated during the early stages of the study,
using environmental data valid at the time. During the course of the project, minor amendments
were made to the environmental data. However, the comparison of load effects and load
patterns was not repeated for the revised loading, since the principles and the conclusions
remain unchanged. Maximum member utilisation ratios for the comparison of application of
load factors to load effects and load patterns are presented in Table 13. This comparison was
made for the case of assessment to 5-5A only, and for the condition where the lower guides
were adjacent to the chord midspan. One wave direction (30 degrees) was examined. The
results clearly indicate a large variance between the two methods of load factor application.
Load factors applied to load effects give member URs significantly lower than the case where
the load factors are applied to the load patterns.
6.5 PRELOAD
The results of maximum URs are presented in Table 14. It is seen that the maximum URs are
considerably lower than those obtained from storm load analysis, since there is little or no
loading from the environment. Assessment to 5-5B was performed to the basic allowable
stresses, without the one-third increase permitted when storm loading is considered.
6.6 OVERTURNING CHECK
Overturning moment checks were performed in accordance with each of the assessment
procedures. The checks for 5-5A indicate that, after application of the appropriate load and
resistance factors, the critical overturning moment exceeds the sum of the stabilising moments
by a factor of 1.93. The equivalent checks in accordance with NSA, result in a factor of 1.84.
For overturning stability, 5-5B states that the ratio of the sum of the stabilising moments to the
sum of critical overturning moments must exceed 1.1. For direct comparison with 5-5A and
NSA, the results of the WSD overturning check lead to a factor of 1.57.
25
5-5B
= 0.112 MN
= 6.629 MN
5-5A
= 0.112 MN
= 8.286 MN
Vertical loads
Dead + Live Load
VD+L
Environmental + Inertial Load VE+Dn
= 30.10 MN
= 18.62 MN
= 30.10 MN
= 23.28 MN
Total loads
Horizontal Load
Vertical Load
Vector Load
= 6.741 MN
= 48.72 MN
= 49.18 MN
= 8.426 MN
= 53.38 MN
= 54.03 MN
QH
QV
QVH
i.e. the Environmental + Inertial Load comprises around 37% of the total load and the vector
load QVH is around 10% larger in the 5-5A version of the code than in the 5-5B version.
From the soil data supplied by Shell(4), from the structural drawings and assuming a spudcan
penetration of 3m, the following variables are defined:
Soil = undrained clay type
26
D
po'
B
A
As
= 0.00 m
= 0.00 MPa
= 14.02 m
= 154.38 m
= 14.38 m
Cu
Nc
sc
dc
= 0.18 MPa
= 5.14
= 1.19
= 1.00
Rotational + Vertical
Horizontal
(both codes)
Equations linking the vectorial capacity (FVH), the horizontal capacity (FH) and the vertical
capacity (FV), are as follows:
FVH
= A.(Cu.Nc.sc.dc.(1-1.5FH*/(Nc.A.Cu)))
= FH* +As(Cuo + Cul)
FH
= (FVH 2-FH2)
FV
But the ratio of FV to FH should be the same as that for QV to QH
Solving these formulations give the bearing capacity envelope presented in Figure 21. The
following loads and capacities are specified on this figure:
Still water load
SWL
= 30.10 MN
QVH_5-5B
FVH_5-5B
UFVH_5-5B
= 48.72 MN
= 146.8 MN
= 0.332
QVH_5-5A
FVH_5-5A
UFVH_5-5A
= 53.38 MN
= 121.5 MN
= 0.439
Nb. The 5-5A resistance factor for foundation penetration sufficient to mobilise maximum
bearing area hfc = 0.85.
Consequently, for the maximum leeward leg bearing capacity, both the 5-5B and 5-5A codes
have utilisations less that 1.00.
Thus the ratio of the utilisation factors is: 5-5A(LRFD)/5-5B(WSD) = 0.439/0.332 = 32%. This
is in-line with expectations for environmental load being around 40% of the total load, see
Figure 14.
6.7.3 Sliding Capacity Check of the Windward Leg(S)
The largest sliding capacity relates to minimising the downward force from the hull and
maximising uplift on the leg due to environmental loads. These vertical loads should be
combined with the maximum horizontal load due to environmental conditions. From the
structural analysis, the largest unfactored loads for a windward leg relate to the environmental
loads coming from the 60 direction, i.e. the starboard leg is the windward leg.
The unfactored (5-5B) and factored (5-5A) loads on the bow leg are as follows:
27
Horizontal loads
Environmental + Inertial Load HE+Dn
5-5B
= 7.812 MN
5-5A
= 9.765 MN
Vertical loads
Dead + Live Load
VD+L
Environmental + Inertial Load VE+Dn
= 27.22 MN
= -27.03 MN
= 27.22 MN
= -33.78 MN
Total loads (note load factor = 1.25 for E + Dn loads in 5-5A code)
Horizontal Load
QH
= 7.812 MN
= 9.765 MN
= 0.190 MN
= -6.568 MN
Vertical Load
QV
= 7.814 MN
= 11.768 MN
Vector Load
QVH
i.e. the Environmental + Inertial Load comprises over -50% of the total load, therefore there is
an overall uplift on the starboard windward leg for this loadcase. This is particularly significant
for the 5-5A code where the 25% increase in environmental load leads to a 650% larger uplift,
QV , than the 5-5B version.
It should be noted that uplift on this leg for this loadcase would lead to a failure in the jack-up
overturning moment (OTM) check. Consequently, under such circumstances, the foundation
check would not be undertaken.
The key variables are as for the capacity check above with:
Vector Cu based on load ratio V:H
Cu
= 0.063 MPa
Vector Cu based on load ratio V:H
Cu
= 0.116 MPa
Effective overburden pressure due to back-flow F0'
Submerged unit weight of soil
g'
Volume of soil displaced
V
The horizontal sliding capacity is given by:
= A.Cul +As(Cuo + Cul)
FH
On this basis:
= 11.60 MN
FH
= 21.30 MN
FH
= 0.00 MPa
= 0.01 MPa
= 94.40 m3
(5-5B)
(5-5A)
(Assumed)
(5-5B)
(5-5A)
The Go-By document associated with Bulletin 5-5A, includes limits for the application of the
above formula based on values of FVH and Qu (this may be a mis-typing of Qv). However, the
document then applies a different range of applicability in the described example. No reference
to this limit is made in Section 6 of either 5-5A or 5-5B. The results of this analysis should be
treated with caution as the starboard leg is described to be in a state of uplift while significant
horizontal force is applied to the leg.
Overall, including the 5-5A resistance factor for foundation sliding in undrained clay soil fc =
0.64, the utilisation factors due to sliding are as follows.
5-5B Horizontal load
5-5B Horizontal capacity
5-5B Utilisation factor
QH_5-5B
FH_5-5B
UFH_5-5B
= 7.812 MN
= 11.60 MN
= 0.674
QH_5-5A
FH_5-5A
UFH_5-5A
= 9.765 MN
= 13.63 MN
= 0.716
Consequently, for the maximum leeward leg bearing capacity, both the 5-5B and 5-5A codes
have utilisations less that 1.00. The 5-5A code has a utilisation factor 0.861/0.696 = 6% larger
than the 5-5B code. This value is less than predicted for an environmental load of around 50%
of the total load, due the different values of undrained cohesive shear strength (Cu) that were
28
adopted. This was not expected to vary between the analyses and only does so due to having
different values in the Shell soil report for the vertical and horizontal directions.
6.7.4 Yield Surface Check
The yield surface check is based on an iterative procedure that considers the combinations of
vertical, horizontal and overturning moment forces at each spudcan. These force combinations,
factored in the 5-5A version of the code, are compared to the yield surface as determined from
the yield interaction equation given in Section 6.3.4 of the SNAME Bulletins.
For extreme wave analyses it is anticipated that the forces will lie outside the yield surface and
this will require repeated analyses at reduced rotational stiffness. Failure of any spudcan is
indicated if the moment is reduced to zero, yet the force combination still lies outside the yield
surface.
Further loadcases will need to be analysed to determine preload forces and yield interaction
forces, prior to performing the iterative assessment of the yield surface. Initial indications
suggest that the difference between the forces determined from the 5-5B and 5-5A versions of
the code will be minimal. In addition, it does not appear that any resistance factors are applied
to the capacities FVHM, FHM and FM, employed in the assessment.
Consequently, it is considered that there will be minimal difference between the 5-5A and 5-5B
versions of the jack-up codes for the yield surface check.
6.7.5 Conclusions
For the 116C jack-up, foundation checks have been performed under extreme wave loading
conditions, in accordance with Bulletin 5-5A and 5-5B. The North Sea Annex to Bulletin 5-5A
follows that for the main Bulletin, with some further aspects of the design to be considered.
These additional items, such as the possibility of scour, do not affect the basic code checks
required.
For the maximum 50-year wave loading on the jack-up, it was shown that the bearing capacity
of the leeward leg was satisfactory to meet the maximum design loads. The utilisation of the
leeward (Bow) leg being less than 50% irrespective of the code used. However, as previously
estimated in Section 5.3.2.3 of this report, the 5-5A code yields a utilisation factor 32% larger
than for the equivalent 5-5B code, based on the environmental load contributing around 40% of
the total load. The bearing capacity envelope has been described and the load and resistance
values plotted accordingly, see Figure 21.
The sliding check on the windward (starboard) leg gave higher utilisations than the bearing
check for the maximum wave loading combined with minimum hull weight. In fact, for both
factored and unfactored loads, the uplift on the leg due to environmental loads exceeds the
structural weight contribution on this leg. It should be noted that a foundation check would not
normally be performed in such a case where there is uplift on a leg and the overturning moment
check indicates a significant over-utilisation. Therefore, concern is expressed as to the validity
of the quoted formulae and utilisations quoted for this scenario. The 5-5A code gave a
utilisation around 6% larger than the equivalent utilisation from the 5-5B code. This difference
being less than previously anticipated (see Section 5.3.2.2), due to a significant variation in the
value of the undrained cohesive shear strength (Cu) that was applied to each assessment. This
parameter was not expected to vary between the codes and does so due to having significantly
different horizontal and vertical components according to the supplied data from Shell.
A full yield surface check was not performed in this study. This check requires an iterative
assessment of all spudcans to be performed and the combined vertical, horizontal and
overturning moment forces to be compared to the yield surface. The rotational stiffness of each
spudcan was adjusted following each iteration. It is anticipated that the 5-5A and 5-5B versions
29
of the code will give identical yield surfaces, with only slightly increased loads for the 5-5A
version of the code.
6.8 SUMMARY OF RESULTS
A summary of results for the assessments conducted in accordance with 5-5A, 5-5B and the
NSA is presented in Table 15. For the chord stress checks, these have been expanded to
include, in the case of 5-5A and NSA, more appropriate values of and, in the case of 5-5B,
other interpretations of how and where stress checks have been performed.
In the case of 5-5A, = 1 has been used as the default value. By exploiting the information
presented in the Commentary to 5-5A, a minimum value for of 1.45 can be used. From the
RP, a procedure for determining a more accurate value for is provided. Following this
procedure leads to an value of 2.25. The corresponding reductions in utilisation are listed for
both 5-5A and NSA. Values of increasingly greater than unity lead to dramatic reductions in
utilisation. For the present 5-5A case, increasing from 1.0 to 2.25 leads to a reduction in
utilisation from 1.86 to 1.27, equivalent to some 32%. Similar improvements occur with the
NSA.
For 5-5B, using the yield value for the rack, which coincides with the minimum section
modulus, and following the instruction given in Figure 8.2 of 5-5B to check the stress at the
extreme rack fibre (ignoring the teeth) leads to the listed value of 1.74.=
30
where c
K
l
r
Fyeff
E
for c Q 1.5
[Eq. A-B5-11]
[Eq. A-B5-13]
and where Q is determined according to the slenderness of the cross-section and, for a tubular,
takes the values
Q=1
= 3790/[Fy (R/t)] + 2/3
[A-B5-9]
in which R is the outer radius of the tubular and t the tubular wall thickness.
7.2.2 5-5A Prismatic sections
For prismatic sections, the slenderness limits are defined by the width to thickness ratio b/t of
the rectangular element under consideration and whether the element is supported on one or two
edges, ie, along the sides of the element parallel to the direction of the stress. The following
31
summarises the criteria for these two types of elements in which Fr is defined as residual stress
for which a value of 114 MPa is to be assumed.
b/t slenderness limits for rectangular elements of prismatic sections
Slenderness range
Stiffened along two edges
Stiffened along one edge
Compact and Noncompact
625/(Fy - Fr)
250/Fy
Fr = residual stress to be taken as 114 MPa
The source of the residual stress is presumably welding and its effects are independent of the
thickness of the welded elements or of the size of the weld. Although these limits are exactly in
line with those given in the source document, AISC LRFD, the basis and application here is
questioned on several fronts. Firstly, as the level of residual stress increases, so does the b/t
limit. Effectively then, as more welding is applied to an element, the more robust it becomes
because where it might have been classified as slender it can, after more welding, be described
as noncompact. This is contrary to the behaviour of plate panels subjected to welding in that
their stiffness when compressed decreases roughly in proportion to the amount of welding and
they may also suffer a loss in strength. Secondly, the level of welding residual stress is
independent of the size of the weld and of the thickness of the welded elements. The magnitude
of welding residual stresses demonstrates an almost linear relationship with the weld size which
is often a function of the thicknesses of the elements being welded but not necessarily.
In AISC LRFD, for elements stiffened on two edges, cover plates are included in the
description. It would be helpful if the same description was included in 5-5A so that the plates
used to reinforce at least two main elements of typical chords which are the side plates and racks
of tear drop chord sections were clearly included. Admittedly, cover plates in AISC LRFD
terms are normally applied to flanges. For rack chords, in 5-5A parlance, they are applied to
webs. Nevertheless, they serve the same function in relation to webs as cover plates do to
flanges.
In the case of rack chord sections, Q for use in [Eq. A-B5-11] and [Eq. A-B5-13] is taken as
unity for compact and noncompact cross-sections or determined in accordance with 5-5A
clauses 8.1.4.3 b) i) or ii) for slender cross-sections.
For use in the axial -bending interaction equation, the above stresses are converted to forces (Pn)
by simple multiplication by the cross-sectional area A.
7.2.3 5-5B Tubular sections
The definitions on cross-sectional slenderness in 5-5B follow the general lines of 5-5A in that
compact and noncompact sections are dealt with via the same set of equations and only slender
sections receive independent treatment.
For tubular sections, the R/t limit for compact and noncompact sections based on AISC LRFD is
11,380/Fy, ie, D/t 22,760/Fy. For a yield stress of 587 MPa, the R/t = 19.4, ie, D/t = 38.8.
However, the succeeding compressive strength formulations for tubulars are based on API RP
2A-WSD for which the corresponding R/t ratio is 30 (and not 120 as printed), ie, D/t = 60. This
major difference between the AISC LRFD and AP RP-2A limitations is a consequence of AP
RP-2A retaining a formula for tubular local buckling compressive strength that is not properly
non-dimensionalised. The equation (API 3.2.2-4) is given below from which it is seen that the
controlling parameter is D/t. It should at least be FyD/Et as adopted in the corresponding AP
RP-2A formulation for bending or, more generally, Fy/Fxe where Fxe is the local elastic buckling
criterion for tubular sections.
In the following table, a comparison is presented between the 5-5A and 5-5B slenderness
limitations indicating, in the case of 5-5B, the source of the two sets of limitations.
32
Slenderness range
5-5A
38.8
5-5B
AISC LRFD
API RP 2A
38.8
60
For 5-5B, the equations used to determine overall axial compressive (column) strength are those
denoted [E2-1] and [E2-2], ie,
( Kl / r ) 2
Fye
1
2 Cc2
Fa =
5 3 Kl / r ( Kl / r )3
+
3 8 Cc
8 Cc3
for Kl / r Cc
[E2-1]
Fa =
12 2 E
for Kl / r > C c
23 (Kl / r ) 2
[E2-2]
where Cc = (2 E/Fye)
K = effective length factor taken as 1.0 for rack chord members and 0.8 for K-braces
and span breakers
l
= unbraced length
r
= radius of gyration
Fye = smaller of minimum specified yield stress Fy or 5/6 of ultimate tensile strength
E = elastic modulus for which the recommended value is 200,000 MPa.
In the case of tubulars, Fye is the smallest of the values of the nominal yield stress, the elastic
local buckling stress Fxe or the inelastic local buckling stress Fxc. The buckling stresses are
defined as follows:
for D/t 60
for D/t 60
Fxc = Fy
= Fy [1.64 - 0.23 (D/t)1/4]
where D
t
C
(API 3.2.2-3)
(API 3.2.2-4)
= outer diameter
= tubular wall thickness
= elastic critical buckling coefficient for which a value of 0.3 is recommended to
consequences of initial geometric deformations that lie within the limits
specified in API RP 2B.
These definitions for tubular sections apply across the range of cross-sectional slenderness,
namely, compact, noncompact and slender.
7.2.4 5-5B Prismatic sections
The criteria on cross-sectional slenderness are tabulated below.
b/t slenderness limits for rectangular elements of prismatic sections
Slenderness range
Compact and Noncompact
33
These are compared below with the corresponding criteria in 5-5A for a yield stress of 460 MPa
remembering that for 5-5A for elements supported on two edges a residual stress of 114 MPa is
relevant.
Comparison of 5-5A and 5-5B b/t slenderness limits for rectangular elements
of prismatic sections for a yield stress of 460 MPa
Slenderness range
As noted above, the presence of residual stresses allows more slender elements to be treated as
compact than is otherwise the case.
Compact and noncompact prismatic members are designed for overall compression using the
same equations as for tubular sections, ie, [E2-1] and [E2-2]. No specific information is
provided for prismatic members comprised of slender elements. For elements subjected to
bending, the recommendation is not to use such sections. [The wording of the relevant text Section 8.1.4.6c) - does not in fact say this but it appears to be the intent of the text.]
7.3 BENDING
7.3.1 5-5A Tubular sections
Sections in bending are assumed on a cross-sectional basis to be compact, noncompact or
slender.
Compact sections are permitted to reach their full plastic moment capacity. For tubular
sections, this requires the D/t ratio to be less than 14,270/Fy. [Note that the printed equation for
the limit p is incorrect, the sign should be replaced by =. This same error also appears in
the Imperial version of the equation, in the Note to 8.1.4.6 a), and in the equivalent equation for
noncompact section 8.1.4.6 b), both in the SI and Imperial versions.] In the present work, full
plastic moment capacity has been determined using the expression:
Mp = Fy [D3 - (D - 2t)3]/6
The approximation Fy D2t gives very similar results provided D is taken as the mean diameter of
the tubular.
Noncompact sections are allowed to reach yield but not the full plastic moment capacity.
Tubulars with D/t ratios up to 61,850/Fy but beyond the compact limit are defined as
noncompact. The equation defining the moment capacity of noncompact sections is
Mn = {2068/(R/t) + Fy} S
where S is the elastic section modulus.
Note that the equation as printed is defined as Mr, ie, the noncompact section limiting buckling
moment which, by implication is inserted into [Eq. A-F1.3] which is
p
M n = M p (M p M r )
r p
34
[Eq. A-F1.3]
in which defines the slenderness of the element. For tubulars, = 2R/t; the definition of for
rectangular elements is discussed below. This substitution is to determine Mn on a proportional
basis between Mp and Mr. However, in the case of tubulars, this is incorrect as a check with the
source document AISC LRFD Table A-F1.1 note (f) confirms. Figure 22 plots the correct
moment capacity versus the incorrect one from which it can be seen that, fortunately, the error is
not that significant. The maximum error is 9.1% when D/t = 60. There is no discrepancy in the
range D/t less than p. when the fully plastic moment capacity prevails.
For slender sections, the moment capacity is limited by the local buckling stress of the
compressed face of the tubular. This is given by
Fcr = 33,610/(R/t) MPa
exactly as in AISC LRFD. The moment capacity is found by multiplying this expression by the
section modulus of the tubular. That this equation provides a smooth transition with the
noncompact criterion can be appreciated from Figure 22 where the plots can be seen to extend
beyond r
7.3.2 5-5A Prismatic sections
Compact and noncompact sections are expected to satisfy a lateral torsional buckling criterion.
If they do not, they are classified as slender.
For prismatic sections, the slenderness limits are defined by the width to thickness ratio (b/t = )
of the rectangular element under consideration and whether the element is supported on one or
two edges, ie, along the sides of the element parallel to the direction of the stress. The following
summarises the criteria for these two types of elements in which Fyw is defined as the yield
stress of the web.
b/t slenderness limits for rectangular elements of prismatic sections
Slenderness range
Stiffened along two edges
Stiffened along one edge
Compact p
500/Fy
170/Fy
Noncompact r
625/(Fy - Fr)
278/(Fyw - Fr)
Fr = residual stress to be taken as 114 MPa
Two aspects about these limits are immediately apparent. Firstly, the slenderness limits for
noncompact sections are a function of the level of residual stress present. Secondly, the
criterion for noncompactness for the element stiffened along one edge is a function of the yield
stress of the web.
The first of these issues was addressed earlier. In relation to the criterion for noncompactness
for the element stiffened along one edge, which is a function of the yield stress of the web, this
was implemented in AISC LRFD for application to flanges of hybrid girders. Hybrid girders
are girders constructed with webs of lower yield stress than that of the flange. The reason for
the criterion being developed as a function of the yield stress of the web, although the element
under consideration is a flange element, is that the restraint being sought is a rotational restraint
of the flange. This is to inhibit local torsional buckling of the flange with respect to its line of
connection to the web. It is the properties of the web that control this degree of restraint so it is
the value of web yield stress that is of relevance as far as the criterion is concerned.
In the context of tear drop chords, both the rack and the side plates are described as webs (see 55A Section 8.1.4.6 Check: Local buckling). In this application, it is clear which yield stress is
of relevance provided one knows the reason for the selection of the yield stress in the first place.
However, for a designer not necessarily fully conversant with local buckling requirements, it
may not be obvious which value to choose.
35
The preceding discussion has been confined to major axis bending in respect of the rack and the
side plates being identified as webs. For minor axis bending, however, these elements
effectively become flanges and the back plates webs. As far as the rack is concerned, the
yield stress of the back plates is the relevant value of yield stress to insert in the equation under
discussion. Such a selection is unlikely to arise from the definitions currently extant in 5-5A.
The criterion needs to be rewritten in a form that refers to any element supported on one edge as
the choice of the yield stress being the lower of the two connected elements.
However, even more care is required in rewriting such a criterion. Consider the back plate of a
tear drop chord under major axis bending. The back plate has two (albeit small) outstands
which, by definition, are elements that are supported on one edge. If the yield stress of the side
plates is smaller than that of the back plate, the yield stress of the side plate would not be the
value to insert in this criterion. The reason for this is that the back plate extends beyond the side
plate with a thickness equal to that of the outstand. The restraint on local torsional buckling is
controlled by the stiffness of all the elements restraining the outstand. Even for a very weak
side plate, the middle portion of the back plate will restrain the outstand so that the yield stress
of the side plate is of no relevance in the context of this criterion.
As an element, the back plate serves to highlight major weaknesses in the definitions provided
for local buckling checks in 5-5A. This can be traced back to the source document AISC LRFD
which does not deal at all with an element created by extending an element supported on two
edges beyond the line of the supports. These cantilever elements are potentially likely to
suffer the same type of local torsional buckling as flanges of I and channel sections. However,
the rotational restraint provided via the internal section of the base plate and the connection to
the side plate is substantially stiffer than that provided by traditional webs to the flanges of I and
channel sections. Thus, these cantilever elements of the tear drop chord can be subject to more
relaxed requirements than the flanges of I and channel sections.
Some of the problems with the above definitions on element slenderness become more apparent
when the equations for determining the moment capacity of noncompact sections are
considered. These have been extracted directly from AISC LRFD and allow for both major axis
and minor axis bending. The equations are as follows:
major axis bending:
Mr = (Fywj - Fr) S
(flange buckling)
Mr = Re Fyfj S
(web buckling)
=
=
=
=
=
=
(flange buckling)
These equations have been extracted directly from AISC LRFD in which they are designated as
appropriate for channels and doubly and singly symmetric hybrid I girders subjected to major
axis bending or channels and doubly symmetric I girders subjected to minor axis bending.
Almost by definition, they do not apply to tear drop chords but, of more relevance, is the
inappropriateness of applying them to such sections for the following reasons:
36
i)
under major axis bending, with the rack (a web) in compression, what value of Fyfj
should be used when considering the rack
ii)
under major axis bending, with the rack in tension, is the rack to be checked for possible
tension yield when flange buckling of the back plate is being addressed. For
asymmetric sections such as the tear drop chord, the neutral axis is normally closer to
the back plate than to the rack. The longer lever arm to the extreme fibre of the rack
compared with that applicable to the back plate generates higher stresses in the rack
than in the back plate. Thus, it is possible for yield of the rack to precede buckling of
the back plate. Of course, the back plate may not in fact be undergoing flange buckling,
it may be compact enough to yield before buckling in which case tensile yielding of the
rack could clearly occur. On the other hand, the definition for S, the section modulus, is
as the minimum for the plane of bending under consideration. This tends to ensure
the design is safe but it may be unnecessarily conservative as the following results
indicate.
For the reinforced chord section of the 116C, the elastic moduli for major axis bending
are:
rack chord:
back plate:
574.2 MPa
460.0 MPa.
Substituting into the major axis bending flange buckling criteria above in which Fr =
114 MPa gives for the back plate
Mr = (460.0 - 114) 1.395 x 107 = 4.827 x 109 Nmm.
By inspection, this is the smallest Mr that will be determined because the back plate is
the only flange element, the yield stresses everywhere else are at the same as that of
the back plate or greater, and Re can be taken as unity.
However, if the moment capacity of the section was to be more rationally determined,
the following set of calculations would be conducted.
For the back plate, flange buckling governs, so using the relevant back plate properties
Mr = (460.0 - 114) 2.000 x 107 = 6.920 x 109 Nmm.
For the rack, only web buckling is applicable for which, according to the above
criterion, a flange yield stress should be used. No flange is attached to the rack to
inhibit local torsional buckling but the side plates serve the same purpose. The yield
stress of the side plates should thus be used in the equation so that
Mr = 460.0 x 1.395 x 107 = 6.417 x 109 Nmm.
The lesser of these is the latter, ie, 6.417 x 109 Nmm, which can now be compared
directly with the moment capacity determined above in accordance with the 5-5A
provisions of 4.827 x 109 Nmm. This has resulted in an under-estimate of the correct
value of Mr of the rack chord of some 33%. This is significant but its impact on the
37
flexural capacity of the chord Mn will be reduced because the slenderness of the chord
elements are not at their limit r for which Mr is only applicable. Nevertheless, the
result reflects a serious weakness in the calculation procedure implicit in the 5-5A
methodology.
The flexural capacity of slender prismatic cross-sections is determined via equations extracted
directly from AISC LRFD. However, these appear to only apply to channels and doubly and
singly symmetric I-shaped beams subjected to major axis bending and to channels and doubly
symmetric I-shaped beams subjected to minor axis bending. This does not seem to include Tsections which, from the perspective of the rack of a tear drop chord, the chord section most
closely resembles. This issue is being promoted for further discussion and needs to be
considered in more detail.
7.3.3 5-5B Tubular sections
The requirements given in the WSD broadly follow those of 5-5A but they differ in detail.
For tubular sections, the cross-sectional slenderness limits are 2R/t 10,300/Fy for compact
sections and 2R/t 300 for noncompact sections. The following table compares the two sets of
requirements for a yield stress of 587 MPa.
Comparison of 5-5A and 5-5B D/t slenderness limits for
tubular sections for a yield stress of 587 MPa
Slenderness range
Compact
Noncompact
5-5A
24.3
105.4
5-5B
17.5
300
The limits on slenderness differ significantly between the two sets of requirements although that
on compactness may be of more relevance.
In 5-5B, the bending capacity of sections is expressed in terms of stresses rather than in terms of
moments as adopted in 5-5A. Stresses are to be preferred since they are independent of
sectional dimensions. For compact sections, bending stress is limited to 0.75 Fy. For
noncompact sections, the bending stress is defined by a bilinear relationship involving a nondimensional cross-sectional slenderness parameter Fy D/Et as follows:
2R/t 20,680/Fy
Fb = [0.84 - 1.74 Fy D/Et] Fy
2R/t 300
Fb = [0.72 - 0.58 Fy D/Et] Fy
These have been extracted directly from API RP 2A-WSD.
7.3.4 5-5B Prismatic sections
The cross-sectional slenderness limits for rectangular elements are:
b/t slenderness limits for rectangular elements of prismatic sections
Slenderness range
Stiffened along two edges
Stiffened along one edge
Compact p
500/Fy
170/Fy*
Noncompact r
625/Fy
249/Fyw
* the printed version is 65/Fy exactly as for the Imperial version and thus needs correction
38
These are identical to the 5-5A requirements in respect of compact elements but differ in respect
of the noncompact elements mainly through the omission of the residual stress term. The
criterion for noncompactness for elements stiffened along one edge is, as in 5-5A, expressed in
terms of the yield stress of the web. This is, as discussed above, not appropriate.
The following table compares the two sets of requirements for a yield stress of 460 MPa.
Comparison of 5-5A and 5-5B b/t slenderness limits for rectangular elements
of prismatic sections for a yield stress of 460 MPa
Slenderness range
Compact
Noncompact
As noted above, the requirements are identical in respect of slenderness limits for compact
elements but the 5-5A requirements for noncompactness are more relaxed than those of 5-5B.
For both major and minor axis bending, the stress limits are:
compact sections:
Fb = 0.66 Fy
noncompact sections:
Fb = 0.6 Fy
In determining the moment capacity of sections including hybrid sections, 5-5B requires
standard techniques shall be applied to obtain a section modulus based on individual
components. A little more direction is needed here because, although no reference is made
here to Figure 8.2 which contains directions as to the stress points to be checked, these
directions are inadequate except for cross-sections composed of one material. If the side plates
in Figure 8.2 are of lower yield stress than that of the rack, in principle it is necessary to check
the stress in the side plate at the point of its connection to the rack.
This more fundamental approach adopted in 5-5B for the determination of relevant bending
section properties is preferred to that of 5-5A. 5-5A applies formulae developed for channel and
I-section girders (including hybrid girders) to cross-sections for which their application at least
needs confirmation and seems to do so, as discussed above, unnecessarily conservatively.
7.4 COMBINED COMPRESSION AND BENDING
7.4.1 Tubular sections
The interaction equations between axial compression (and tension) and bending in 5-5A are
for Pu/aPn > 0.2
M uey
8 M uex
Pu
+
+
a Pn 9 b M nx
b M ny
else
39
1.0
[Eq. H1-1a]
M
M uey
Pu
uex
+
+
b M nx
b M ny
2 a Pn
1.0
[Eq. H1-1b]
where Pu
= applied axial load (this is found under factored loading conditions)
Muex, Muey = effective bending moments in the x, y planes to be determined in
accordance with 5-5A Section 8.1.4.4 (for tension) and 8.1.4.5 (for
compression) (these are found under factored loading conditions)
= axial load resistance factor = 0.85 for compression and 0.90 for tension
a
= bending resistance factor = 0.90
b
fa
+
0.6 Fy
f bx2 + f by2
Fb
(3.3.1-1)
1 .0
(3.3.1-2)
40
fa
+
Fa
f bx2 + f by2
Fb
1.0
(3.3.1-3)
If different Cm and Fe values apply in the x and y planes, then (3.3.1-1) has to be modified to
fa
+
Fa
C mx f bx
f
1 a'
Fex
C my f by
+
f
1 a'
Fye
Fb
1 .0
(3.3.1-4)
5-5B notes in Section 8.1.4.1b that when local element P-Delta effects (it is assumed it is
meant to say global) are included in a structural analysis of an entire jack-up, the Euler
modifications in the above equations may be taken as unity for chord elements only. Further,
that if the local P- effects are included in the local member response, the Euler amplification
may be set to unity for the corresponding members. No advice is given as to how to treat Cm.
For the present assessment, in keeping with 5-5A, Cm has been taken as unity when applying the
above 5-5B equations.
Figure 23 presents a comparison between the 5-5A and 5-5B requirements for the axial
compressive stress capacities for tubulars as a function of length to diameter (L/D) ratio when
subjected to purely environmental loading conditions. These capacities are, in the case of 5-5A,
the loads before the load factors are applied whilst, in the case of 5-5B, they include the onethird increase in allowable stresses permitted in the presence of environmental loading. For 55A, the load factor was taken as 1.25. The D/t of the plotted tubular is 12.8 and its effective
yield stress Fye is 574.2 MPa, the minimum of the nominal value of yield stress of 587 MPa and
5/6 of the ultimate tensile strength of 689 MPa.
The range of L/Ds included in the figure encompasses those values relevant for tubular
members of the legs of the 116C. These are indicated in the figure covering the horizontal and
diagonal braces and the span breakers. The corresponding L/D values are 13.0 for the
horizontals, 16.8 for the diagonals and 26.2 for the span breakers. In keeping with the
recommendations of both sets of requirements, an effective length factor K of 0.8 was used in
all cases. The 5-5B requirements can be seen to provide significantly enhanced capacity
throughout the L/D range. In the case of the 116C components, the differences are 14% for the
horizontals, 15% for the diagonals and 19% for the span breakers. This can be appreciated in
more detail through an examination of the equations in question.
Consider a 116C diagonal 324 mm x 19.1 mm wall thickness spanning 5443 mm having a
nominal value of yield stress of 587 MPa and an ultimate tensile strength of 689 MPa. The L/D
is 16.8, the D/t is 17.0, and the effective yield stress Fye is 574.2 MPa. For a K of 0.8, the value
of c in 5-5A is 0.688. The value of Cc in 5-5B is 82.9 which can be found by exploiting the
relationship between the slenderness terms used in the two sets of requirements which can be
shown to be c/2 = (Kl/r)/Cc.
From a cross-sectional slenderness viewpoint, the section is compact so that for 5-5A Q = 1.
From [Eq. A-B5-11], Fcr = 0.820 Fye which, after introducing a load factor of 1.25 and a
resistance factor of 0.85, leads to a compressive capacity of (0.85/1.25) 0.820 Fye = 0.558 Fye.
41
For 5-5B, from equation [E2-1], the basic strength is 0.882 Fye (from the numerator) whilst the
safety factor (the denominator) is 1.834. Allowing for the one-third increase under
environmental loading conditions, the compressive capacity is (4/3) 0.882 Fye/1.834 = 0.641 Fye.
The ratio of 5-5B to 5-5A compressive capacities is 0.641/0.558 = 1.15 which coincides with
the value found above from Figure 23. From these results, it is seen that the source of the
difference is as follows:
Figure 24 presents a comparison between the 5-5A and 5-5B requirements for the bending
capacities for tubulars as a function of diameter to thickness ratio (D/t) ratio when subjected to
purely environmental loading conditions. The range of D/ts included in the figure encompasses
those values relevant for tubular members of the legs of the 116C plus a value of 40 for
comparison purposes.. The 116C D/ts are indicated in the figure, the corresponding values
being 12.8 and 17.0 for the diagonals, 17.0 for the horizontals and 24.2 for the span breakers.
For this loading condition, the two sets of requirements lead to very similar results although
5-5B is seen to be marginally bigger than 5-5A.
The origin of the differences can be found through examination of the diagonal considered
above when subjected to pure bending under environmental loading conditions. The section is
compact in accordance with both sets of requirements. For 5-5A, the section can achieve full
plastic moment capacity, ie, Mn = Z Fye where Z is the plastic section modulus. Introducing the
relevant load and resistance factors, the unfactored moment capacity = (0.9/1.25) ZFye.
Converting this to an equivalent stress by dividing by the elastic section modulus S gives
(0.9/1.25) (Z/S) Fye. For this section, Z/S = 1.349 so that the allowable equivalent bending
stress = 0.971 Fye.
For 5-5B, the basic allowable bending stress = 0.75 Fye which can be increased by one-third in
the presence of environmental loading conditions. The allowable bending stress is thus Fye.
Both this and the corresponding 5-5A value are seen to coincide with the values plotted in
Figure 24 given the effective yield stress is 574.2 MPa. The source of the difference between
the pair of results is as follows:
The comparison between the two sets of requirements over the full range of combined
compression and bending is presented in Figure 25. It can be seen to cover four D/t ratios, those
relevant for the 116C tubular diagonals, horizontals and span breakers as well as a D/t of 40 for
comparative purposes, and seven L/D ratios. The L/D ratios extend reasonably uniformly over
the range up to the maximum applicable to that of the span breakers. To help distinguish the
curves relating to the L/D ratios relevant in the present assessment, these are shown in bold.
In the left hand set of diagrams, which relate to 5-5A, the bilinear relationship implicit in the
pair of equations [Eq. H1-1a and b] can be clearly seen. That the two sections of the curve are
basically linear is a consequence of the structural analysis being fully non-linear so that the
Euler amplification term is set to unity. Similarly for the 5-5B equations (3.3.1-1) and (3.3.1-2)
where, because both the Cm and Euler terms are equal to unity, equation (3.3.1-1) governs at all
occasions. This arises because Fa in equation (3.3.1-1) is always less than or equal to 0.6 Fy,
both of course being increased by one-third in the presence of environmental loading conditions.
For a practical range of tubular geometries, which in the context of Figure 25 is all the
considered geometries with L/D ratios in excess of 8.5, the average ratio of 5-5B to 5-5A
allowable stresses is 1.06. The maximum ratio is 1.19 whilst the minimum is 1.0.
42
43
documents for most of the structural, stability and foundation limit states addressed in Section 8
of 5-5A.
The 5-5A safety factors are simply determined by combining the corresponding load and
resistance factors. For example, for cross-sections subjected to compression the load factor is
1.25 and the resistance factor 0.85. The safety factor listed in the table is found from 1.25/0.85
= 1.47.
The 5-5B factor is found by taking the allowable stress, enhancing it by the one-third increase,
and determining the margin with respect to yield or component strength if less than yield. Thus
for columns of zero slenderness, the basic allowable stress is Fy/(5/3) = 0.6 Fy. Incorporating
the one-third increase gives (4/3) x 0.6 Fy = 0.8 Fys leading to a safety factor of 1/0.8 = 1.25.
When considering preload and foundation capacity, the proportion of gravity load to
environmental load has also been considered. Safety factors are given for the two extremes, ie,
zero environmental load and zero gravity load, as well as for a ratio of environmental to gravity
load of 2/3 a typical ratio in the present case.
In all cases, the 5-5A factors exceed those of 5-5B, and by a significant margin for many of
them. On this simple basis, the major differences appear inappropriate. However, the adequacy
or not of a safety factor cannot be assessed in the absence of knowledge of the degree of
conservatisms with all aspects of the assessment. In the comparisons presented in Figures 25
and 28, for example, in connection with the allowable stresses in tubular and prismatic section
beam-columns, account was taken of the different strength formulations. However, in neither
case was the accuracy of either formulation checked against test or other appropriate data. In
the case of foundation checks, no differences exist between the basic formulations but
differences do exist between the load combinations under which the checks are effected. In
principle, the 5-5A checks are conducted using environment load analysis results that include a
1.25 factor on loads induced by environmental forces. Because these are amplified by large
deflection effects, they will be more onerous than a simple ratio of results from analyses
excluding the 1.25 factor would suggest.
In seeking to determine the relative adequacy of the safety implicit in 5-5A and 5-5B, a number
of important factors need to be taken into account. Some of these have been independently
addressed as described in the next section. However, some of these are only considered in
relative terms whereas a degree of absoluteness is required to properly determine safety levels.
In addition, structural redundancy does not seem to have been addressed at all. In the wider
context, strong indicators of the adequacy of implied safety levels are the industrys experience
and societys expectations. An appropriate balance of these is always required.
44
wave height
drag coefficient
45
wave period
current velocity
wind velocity.
Chapter 7 defined the four limit states on which the reliability analysis would proceed. They
were:
The standard models and methodologies by which the reliability analysis was to be effected
were described in a step by step manner. They involved equivalent leg models for determining
equivalent leg drag loading, a detailed leg only model for determining equivalent leg structural
properties, and a detailed leg model with supporting structure for checking hull-leg interaction
and for quantifying stiffnesses between the legs and the guides and pinions allowing for gaps
and offsets. The output of these models was used as input to an equivalent leg large deflection
FE time-domain analysis model so that in principle it included appropriate leg, hull and leg-hull
interaction stiffnesses.
Chapter 8 identified the jack-ups units proposed for analysis to be:
The water depth (chart datum), tide and surge, airgap (from chart datum), maximum wave
height and current and wind velocities for use in the analyses of the three units were listed
together with the factors of safety determined from the standard analyses and methodologies
prescribed in Chapter 7. At first sight, the factors of safety appear too low in several instances.
They were claimed to be acceptable following a consideration of metocean directional data, a
detailed foundation study, and rig orientation. Part of the reason for the selection of the Mod VI
(Galaxy) was that it was to be monitored during in-place operations and that, should such data
become available, it was expected to be helpful in defining modelling uncertainties.
The probability density functions to be adopted in the reliability analysis were listed for:
These were determined from a literature survey, the sources of which were identified. Fixity
was also identified as a special variable but, at this stage of the work, only pinned foundations
were considered.
The means and COVs of the basic loading variables were determined using the UK Department
of Energy Guidance Notes. However, the methods by which COVs, which cannot be
determined directly from the Guidance Notes, were quantified were not described.
The means and COVs of interest are listed in Table 17. Some of the COVs seem inappropriate,
for example, 13% for the annual extreme significant wave height and up to 53% for the tidal
current. Wind speed and wind driven current were indicated to be conditional on significant
wave height. Wave period was also to be conditional on wave height but to have its own
distribution.
46
The value of COV proposed for yield stress was not listed.
Noble Denton Consultancy Services Ltd, Jack-Up Assessment Criteria. Statement of
Variable Selection, L15670/NDCS/RVA, Rev 0, 12th September 1991(6)
Reference 6 aimed to provide a basis by which the variables and their distribution parameters to
be used in the reliability analysis were determined. All of the relevant metocean variables were
discussed. Joint probability of wind/wave/current and collinearity was dealt with by splitting
the current into wind-driven and tidal components. Wind and wind-driven current were
assumed correlated with the seastate whilst the tidal component was considered as an
independent variable. The one-minute wind speed was taken as a function of significant wave
height Hs and location water depth. No gust was allowed for. Significant wave height annual
maxima were assumed Gumbel distributed. Wave period was conditional on the seastate and
both its mean and standard deviation were linear functions of Hs. The JONSWAP spectrum
was adopted and spreading was ignored.
Wind-driven current was taken as a constant proportion of wind speed whilst the tidal current
was an independent variable determined from location specific data. A Wheeler-like definition
of current stretching was adopted. Horizontal water particle velocities were assumed accurately
determined from wave harmonics whilst vertical water particle velocities were ignored. Surface
roughness was accounted for in the choice of drag coefficient and marine growth through an
increase in member dimensions. Anodes, etc, were accounted for through the choice of drag
coefficient. The drag coefficient was based on NDCS practice and its variability in accordance
with the literature survey. Nothing was stated in relation to the inertia coefficient. Shielding
and vortex shedding were assumed covered by the selected drag coefficient parameters.
Self weight was assumed deterministic whilst the variable load (ballast, drilling consumables
and live load) took the average, i.e., 75%, of the extremes used in practice of 100% when it acts
together with the environmental forces and leg strength has to be considered and 50% when
variable load opposes environmental forces and overturning is to be checked. Buoyancy was
taken as deterministic.
At this stage, pinned conditions were adopted so soil variability was ignored and penetration
was taken from its deterministic calculation.
For yield stress, the distribution parameters were adopted relevant to the grade of steel.
Ultimate tensile strength was not explicitly considered but was assumed to be addressed in the
LRFD strength checks. Geometric variability was ignored.
Uncertainties arising from FE model layout and analysis errors, jackhouse stiffnesses, hull-leg
interaction stiffnesses, clamping and guide modelling and P- effects were ignored. With
respect to uncertainties in the dynamic simulation, regular versus irregular waves, simulation
time and selection and extrapolation to extremes, irregular time domain analysis was judged as
most appropriate, a simulation time of 3 hours was indicated to be necessary to achieve stability
in the 3rd and 4th moments (of the distribution), the peak was found according to Winterstein and
thus uncertainty in the extreme value was explicitly accounted for. Morisons equation was
taken to give a deterministic estimation of wave/current forces: this is believed to be nonconservative. Damping was assumed to be accounted for through the relative velocity
formulation adopted in the analysis. Possibly important aspects that were ignored included leg
non-verticality and sag due to hull self-weight.
On stress analysis issues, variability in Euler and local plate buckling, joint punching shear and
principal stress checks was ignored. The appropriateness of punching shear as a joint strength
check was questioned and, given more appropriate joint strength checks, these are known to
demonstrate a degree of variability. Further, since triangular chords not infrequently have side
plates that fail the compactness check and, therefore, are controlled by buckling, the
deterministic approach to plate buckling was also questioned.
47
To help in the selection of the variables to be incorporated within the reliability analysis, each of
the variables considered was assigned two weightings, one to reflect its significance in relation
to jack-up behaviour and the other to reflect its likely level of uncertainty. The weighting
factors were based on the findings reported in Reference 5 together with those determined in the
preparation of Reference 6. The subjective nature of this process was recognised by the authors.
The weightings varied from 0 to 10 in increasing order of importance and, when listed, those
where the total was more than 50 were highlighted. The assigned values of some of the
weightings are questioned by this review.
Although some of the variables received combined weighting of more than 50, the lack of
appropriate information to adequately define them was concluded as a just reason for ignoring
them at this point in time. The reliability analysis was thus concluded to be concerned with
wave height, drag coefficient, current, wave period, randomness of the dynamic simulation of
extremes, and yield stress.
Noble Denton Consultancy Services Ltd, Jack-Up Reliability Study. Procedural and
Input Study - Draft 1, 19th September 1991(7)
In Reference 7, a brief review was presented of existing LRFD codes and how these could be
exploited in the present application. The list of reviewed codes seemed limited in the sense that
only the completed codes themselves were apparently reviewed, but not the corresponding
background documents. Perhaps not surprisingly, the review was concerned with offshore
codes only. However, since much of the UK practice in limit state codes was pioneered with
BS5400, it is a little surprising that this at least was not included. Reference was made in the
text and in a comparison table to API-LRFD-NS. However, this document was not cited in the
list of reviewed documents.
The load factors were categorised in accordance with the API RP 2A-LRFD groupings, as
follows:
D1
D2
L1
L2
W
Dn
E
dead load due to self-weight of structure, permanently mounted equipment and other
fixtures, hydrostatic and buoyancy forces.
Loads due to equipment moveable between one mode of operation and another or which
can be removed from the platform but otherwise remains constant for long periods of
time.
Loads due to weights of consumables and fluids in pipes and tanks.
Loads due to short duration operations such as drilling, crane lifting, helicopter
landings, etc, but which are not expected to occur during extreme environmental
conditions.
Static component of extreme environmental loading.
Inertial component of dynamic response.
Seismic loading.
A summary table was presented listing the load factors adopted in the reviewed documents.
Only one of the considered documents related to jack-ups and, as a consequence, was the only
document that provided for load factors covering preload and overturning conditions. For
overturning, the need to consider load factors significantly less than unity was noted. APILRFD-NS was noted for its large factor on W of 1.85.
A further table summarised the resistance factors adopted by the various codes for tension,
compression and bending only. A brief summary was presented of the main strength
formulations for compression, and combined compression and bending. However, no
comparisons were presented between the different formulations.
The need to take care when exploiting the reviewed safety formats was noted because of their
relevance for jacket structures in which dynamic effects were not normally of real interest. The
need to use AISC LRFD in order to handle non-tubular sections was noted despite a preference
48
for API LRFD because of its specific application to offshore structures. A particular preference
was recorded for using the API LRFD approach to dealing with combined axial and bending
loadings (cosine interaction) instead of the linear form implicit in AISC LRFD. This
unfortunately failed to recognise the importance of the beam-column equation in controlling, in
particular, tubular member design as confirmed in a recent study for UK HSE(8).
The report concluded that because the strength formulations would not be addressed as part of
the described study, the resistance factors associated with the strength formulations proposed for
adoption from API LRFD and AISC LRFD would be unchanged.
Noble Denton Consultancy Services Ltd, Jack-Up Assessment Criteria. Reliability
Analysis of M116C in 55 m Water Depth, F&G Mod V in 97 m Water Depth, F&G Mod
VI in 118 m Water Depth Operating in the North Sea, L15709/NDCS/RVA, Rev 2, 27th
February 1992(9)
Reference 9 presented a comprehensive review of the work done to determine the load factors.
It began with a useful summary of reliability analysis fundamentals. This included a glossary
and definitions of limit state, safety margin, basic variables, response surface, safety index
(reliability index), sensitivity factors, design point values, and partial safety factors. These were
explained in the context of a mean Level II reliability analysis. This is helpful but the advanced
Level II reliability analysis (FORM) adopted in the work is not explained anywhere. Some
examples were given to demonstrate some of the principles including safety index derivation
and the relation between safety index and central safety factor. The central safety factor is the
ratio of the means of the resistance and loading distributions. This form of safety factor was
noted, quite rightly, to be of little use in design.
The jack-up units selected for the reliability assessment and partial factor derivation were listed
as:
Particulars for each of these units were listed in Chapter 5 together with the values of the
variables used in the deterministic assessment. In the deterministic study, 50% of the variable
load was used when considering overturning and 100% when considering preload and strength
unity checks. Pinned spudcans were also assumed. The values of the variables listed included
the 50-year return period metocean conditions together with the results of existing analyses that
identified the critical headings and safety factors with respect to overturning, preload and stress
exceedance, and natural periods in sway, surge and yaw. Some of the 50-year return period
values listed did not coincide with those presented in Reference 5 and summarised in Table 17:
no explanation was given for the changes. The safety factors were based on the most probable
maximum (MPM) value for each of the corresponding check conditions: these are summarised
in Table 18.
From the table, it can be seen that the safety factors are generally reasonable except for those
relating to preload for which the margin differs little from unity and, in one case, appears
unacceptable.
Not very helpfully, MPM was not defined anywhere despite the fact that much simpler and
perhaps more complex parameters were defined as noted above. Notwithstanding, a procedure
was described in Appendix F by which the MPM value from a time domain simulation could be
derived. It was based on the Winterstein 1988 method for transforming a non-linear, nonGaussian and finite bandwidth process into a solvable probability function via a Hermite
polynomial fit.
49
The reliability analyses of the three units were described in detail in Chapters 6, 7 and 8 with
details in three appendices. The topics covered were:
Four basic loading variables were considered in the analyses. They were significant wave
height, peak period, tidal current, and drag coefficient-diameter. Four of the six variables noted
in Reference 6 were required for inclusion in the reliability analysis. In the reliability analyses
for overturning, 75% of the variable load was adopted. However, no uncertainty was assigned
to this load or any other gravity loads.
The approach adopted was, using the jack-up model, to conduct sufficient reliability analyses in
the potential failure space for the jack-up so that a response surface for as a function of the
basic variables could be created. Having done this, the response surface could be further
investigated using the reliability analysis rather than performing further reliability analyses on
the model itself. The surface then had to be de-conditioned so as to eliminate the dependency of
the surface on other than the most important variable, namely, significant wave height. Once
this process was achieved, the partial factor on loading was evaluated. This was repeated for
each limit state - see Table 18 - and a range of appropriate load partial factors. Included in the
reliability assessment of the overturning limit state was consideration of the limit state for
lifting/sliding of the windward leg.
Chapter 9 summarised the reliability analysis results, both in tabular and graphical forms.
Average safety factors were determined given a reference of 2.5 and 3.0 and then the spread
of reliabilities given these average safety factors was described. There is a concentration on the
partial factors attributable to each of the basic variables, which is of little use when only the
factor on environmental loading is of real interest. For each limit state, the discussion then
concentrated on the impact of a of 2.5 for which the average safety factors were 1.27 for
overturning (reported as 1.31 in the Executive Summary and Conclusions), 1.17 for preload, and
1.19 for strength utilisation check. A reliability index of 2.5 was then compared with the results
of the API LRFD deliberations from which it was concluded that it is broadly similar to the
findings of the API study. The study concluded that on the basis of the reported work the
present findings could be carried forward to the next phase of work. Further, the lack of
sensitivity to tidal current and peak period suggested that both of these could be treated as
deterministic.
In the reliability analysis of overturning and preload, no uncertainty is attached to the resistance
side of the equation. These analyses, therefore, do not provide probability of failure estimates
but only probability of exceedance estimates. In the case of overturning of the 116C, from the
reported results, it is possible to exploit a mean level 2 reliability analysis to examine the impact
of the introduction of an uncertainty to the resistance side of the equation, ie, to the righting
movement RM. Further, the impact of different safety factors can them be examined.
The reported results for the 116C are:
Safety Factor
(SF)
1.33 (25*)
1.59 (Fig .E-1)
Safety Index
()
2.31 (33)
2.74 (Fig E-1)
50
ln (RM m OTM m )
2
2
VRM
+ VOTM
(8.1)
where OTM is the overturning moment, m indicates the mode of the distribution and V is
coefficient of variation.
The design (d) values of RM and OTM are related to the modes by
RM m = X RM RM d
(8.2a)
(8.2b)
(8.3)
ln (SF X RM X OTM )
(8.4)
2
2
VRM
+ VOTM
Introducing the pairs of values for SF and listed above into (8.4) and adopting XRM = 1, since
value of RM is reportedly appropriate, and VRM = 0 (an assumption in the reported work), leads
to XOTM = 0.510 and VOTM = 0.415. The results indicate that the MPM is just over half the value
used in design and that the total uncertainty associated with loading is relatively large.
If 50% of the variable load had been used in the reliability analysis rather than 75%, the righting
moment would have been factored by 0.945. Treating this as a bias, ie, XRM = 0.945, and
introducing an uncertainty of 10% on the estimation of restoring moment, ie, VRM = 0.10, (8.4)
gives = 2.11. If the safety factor adopted in 5-5B of 1.1 (Table 16) is adopted instead, =
1.67 corresponding to a probability of failure (not now exceedance since RM is random) of 4.7
x 10-2.
However, because in the reliability analyses, the RM was determined using 75% of the variable
load, strictly the value of the safety factor should be similarly determined. Thus, the safety
factors listed above should be factored by 1/0.945. This results in, from resolving (8.4), XOTM =
0.537 and VOTM = 0.417. for the 50% variable load RM remains at 2.17 and 2.11 for VRM of
0.0 and 0.10 whilst, for the 5-5B safety factor of 1.1, it reduces to 1.54.
The 116C stress exceedance limit state reliability analysis can be examined in a similar manner.
Based on a safety factor of 1.32 (from page 25 of the report), using the results for of 1.0 and
1.5 of of 2.13 and 3.09 respectively (see page 45), and assuming initially that is
deterministic, as in the report, then XUC the bias associated with the loading, is found as 0.537
and the associated uncertainty VUC as 0.422. These are quite similar to the values found for
overturning.
If, as considered in the report, is assigned random values of X = 1.19 and V = 0.129 (=
0.154/1.19 on page 46 of the report), then is found as 2.43 which can be compared with value
of 2.41 reported on page 47 of the report.
When assessing the values for stress exceedance, it is important to note that the AISC
equation as adopted in the report included the AISC resistance factors of 0.85 for compression
and 0.9 for flexure see page 130 of the report. However, it is unlikely that these resistance
factors can be inferred as effectively increasing the safety factor associated with the calculated
51
because the bias used for of 1.19 is considered too large to merely reflect the bias of the test
data with respect to the adopted AISC strength equation. The 1.19 is likely to include most if
not all of the safety inherent in these resistance factors which can be estimated as the average of
their inverses, namely, 0.5 (1/0.85 + 1/0.9) = 1.14.
On the other hand, the report indicates (see page 138) that, of the two governing equations listed
on page 130, only the first was incorporated within reliability assessment because of the
problem of selecting an appropriate condition when the axial load rather was 0.2. Thus, in
principle, members that were dominated by bending were taken to be stronger than should have
been the case, leading to over-estimates of their safety indices.
One of the reported conclusions was that the damping due to fluid-structure interaction was
some 6% of critical. This was taken to indicate soil fixity was highly sensitive to this parameter.
It is not clear that this issue was addressed in this particular report.
Noble Denton Consultancy Services Ltd, Jack-Up Assessment Criteria. Executive
Summary and Conclusions of L15709/NDCS/RVA, Rev 2, 27th February 1992(10)
This reference appeared to be a fuller Executive Summary and Conclusions to Report L15709
than that which appeared in the report itself.
Noble Denton Consultancy Services Ltd, Jack-Up Assessment Criteria: Status Summary
on Stage 2 of Reliability Analysis, 24 August 1992 (Preliminary)(11)
The work documented in Reference 11 was reportedly based on the 11th Draft of the RP (i.e.,
Recommended Practice). This is the first time that the version of the RP draft being considered
in any of the reported work has been clearly identified. This version of the report was also
preliminary which is clearly reflected in the quality of some of the work and in its
completeness.
The input data used for each jack-up in Stage 2 was summarised in Attachment 1 of the report.
Unfortunately, it was only input and did not include a summary of results as presented in
Chapter 5 of Reference 9. Attachment 1 also included input data for a Marathon LeTourneau
Gorilla but without any introduction or other comment in the main text.
Chapter 2 presented a summary of the differences between the results of the Stage 2 work (this
report) and of those of Stage 1 as presented in References 9 and 10. The differences, apparently
mainly related to the 116C OTM (overturning moment), are summarised here and commented
upon in turn.
(a) Mean OTM increased by about 22% due to increased wind area.
The wind areas were apparently increased in Stage 2 but by nearly 29%, 1832 m2
compared with 1424 m2. However, the 1-minute mean wind speed was reduced from 40
m/s to 29 m/s. Since wind forces are a function of (velocity)2, the wind force has in fact
reduced by the ratio
1832 x 292/1425 x 402 = 0.676.
This is a reduction of 32.4% which is in total contradiction to the above conclusion.
Further, it was not just the wind loading that had changed. For Stage 1, the surface
current considered was 0.9 m/s whereas for Stage 2 it was 0.69 m/s, the result of
applying a current blockage factor of 0.77 to the Stage 1 values. Also, the percentage of
critical damping used in the analysis had increased from 2% at Stage 1 to 4% at Stage 2.
This decreased the single degree-of-freedom (SDOF) dynamic amplification factor
52
(DAF) leading to a smaller inertial load set. In addition, the equivalent drag loading
CdD on the leg had altered. For Stage 1, a value of 10.2 m was used whereas, for Stage
2, two values were considered, 9.43 in the absence of marine growth and 11.42 with
marine growth. The net effect of this change is difficult to estimate but it may, for the
values used, not make a significant difference to the OTM. However, the changes in
current loading and in the inertial load set lead to a reduced OTM, again raising
questions about the source of claimed increase in OTM. No other parameters appear to
have changed. The natural frequencies are the same so no changes in mass nor in fixity
have occurred.
(b) The standard deviations of the Stage 1 and Stage 2 responses were within 0.3%.
This seems to be borne out by the computerised output (O/P) presented in Attachment 2.
(c) The reliability (safety) index conditioned on seastate had decreased from 1.45 at Stage 1
to 1.04 at Stage 2. It was considered that this was partly due to the increased mean and partly
due to the increase in dynamics in the RP.
In the O/P in Attachment 2, five safety index values were listed. The quoted numbers
related to the finite peak of peaks values: no explanation was given for selecting this
value as distinct from any of the others.
The justification for the decrease in as due partly to the increase in OTM is not, as was
examined above, consistent with the input parameters. As for the effects of dynamics in
the RP, this cannot be even considered let alone compared in the absence of relevant
documentation.
(d) The assessment safety factor on OTM decreased from 1.33 to 1.22.
Again, it is unclear as to how the OTM has increased that would lead to a reduced safety
factor. Further, the presented calculations in Attachment 2 appear to be in error. This
can be readily seen by inserting in (a) of Sheet No. 1 of 2 the values for MPMax (OTMRM) from the second page of O/P instead of the values from the first page of O/P that
has been done in the presented calculations.
(e) The final safety index decreased from 2.31 to 2.06.
The basis for this is not given nor is it consistent with the documented input as noted
above.
(f) Calculations showed that the use of the factored Hs as recommended in the RP decreased
by 0.35. The increased mean also contributed to a reduction in of 0.10. However, these
effects were offset by increased critical damping (from 2 to 4%) and reduced CdD when
compared to the design point value of CdD in Stage 1. Thus the discrepancy was reduced to
0.25 on final .
The use of a factor on Hs is presumably, as indicated in the Attachment, to account for
the fact that a stochastic analysis is being performed. Such analyses employ linear
wave theory and therefore under-estimate crest heights compared with Stokes V and
similar theories. The basis for the adopted factor of 1.07 is not given.
The second sentence of (f) refers to an increase in the mean. The basis for this is
unclear because no indications are reported concerning, for example, additional storm
53
surge or any other parameters that might lead to an increase in the mean sea level as
considered in the Stage 2 analysis.
Because of the lack of, and uncertainty in, the input parameters adopted in the analysis,
the effects on the final cannot be estimated.
The final observation on Stage 1 versus Stage 2 concerned the Friede & Goldman Mod VI
OTM. This apparently showed an increase in the safety factor from 1.23 to 1.54. Three reasons
were given two of which were confirmed by the presented information, namely, reduced wind
area and increased damping. The third was that the RM (restoring moment) had been calculated
using 75% of the variable load. Whilst this may be justified in this particular case, it is not
consistent with the assumptions made elsewhere where 50% of the variable load is normally
adopted in OTM calculations.
Chapter 3 addressed the question of which of two criteria to use when deciding whether to adopt
the relative velocity form of Morisons equation or not. This does not seem to be tackled
rationally. Both were rewritten to be consistent with the other following which the similarity
between one and that used by NPD was noted with interest. Irrespective, the relative form was
used in all Stage 2 analyses, seemingly whether the criteria were satisfied or not. (The relative
form of the Morison equation was adopted in the Stage 1 work.)
Chapter 4 simply presented the results (in Attachment 3) for the three rigs at four locations and
for three limit states ostensibly in both graphical and tabular form: only the plots ( versus
safety factors) were contained in the report. The fourth location purportedly referred to a 116C
located in 65 m of water. However, input details were presented in Attachment 1 for a MLT
Gorilla in 106.7 M of water but not for a 116C in 65m.
The reasons for including this extra location are unknown, but of some concern. Having
increased the water depth by 10 m, the presented plots indicated (where they can be interpreted)
that the safety index increased by at least 0.2 and up to 0.4. It can only be concluded that the
input data for the 116C in 65 m is significantly different from that of the 116C in 55m.
Chapter 5 addressed the question of storm loading parameters in the Gulf of Mexico (GOM)
compared with those adopted in the remainder of the analysis which related to the North Sea
(NS). Sets of 10-year and 100-year return period values for both wind and wave were listed
together with their corresponding annual means and COVs: the source(s) of the data was not
given. The average COV on waves increased from 13% for the NS to 27% for the GOM. The
COV on GOM wind averaged 0.51.
Because the COV on GOM wave height almost doubled, the mean value adopted in the
reliability assessment had to be reduced to compensate: the basis for this was not described.
The results were simply presented in graphical form of versus safety factor but were not
discussed in any way and still demonstrated that the 116C in 65 m of water was more reliable
than when it was in 55 m.
Chapter 6 examined the sensitivity of time domain solutions to the size of the time step used in
the analysis. The question was posed as to whether the RP recommendation of using the smaller
of the zero-upcrossing period Tz or the natural period Tm both divided by 20 could be relaxed.
The answer was given in terms that both 0.25 s and 0.5 s time steps gave similar results so that a
0.5 s time step could be adopted. The choice of 0.25 s was made on the basis that it was
approximately 1/20 of the smallest natural period determined over all the rig/water depth
combinations. The choice of 0.5 s was NDCS standard practice.
54
The result confirmed the finding reported in Appendix C of Reference 9. However, it did not
seem to lead to a conclusion that the RP requirement of the smaller of Tz/20 or Tm/20 could be
relaxed.
Chapter 7 presented some brief conclusions and Chapter 8 a procedure for calibration involving
the use of different partial load factors on the static and dynamic components of environmental
loading.
Chapter 9 referred to the use of the SDOF approach presented in the 11th Draft of the RP to deal
with dynamic effects and a comparison between results obtained using the approach and
detailed analysis that was presented in an attachment to the report. No assessment was
presented.
The final Chapter described four methods by which the calibration procedure summarised in
Chapter 8 could be exploited. Results for two of these methods were presented and described.
Noble Denton Consultancy Services Ltd, Jack-Up Assessment Criteria (Stage 2).
Reliability Analysis of 116C in 55 m Water Depth, F&G Mod V in 97 m Water Depth,
F&G Mod VI in 118 m Water Depth and 116C in 65 m Water Depth in the Elevated
Condition, Report No: L16268/NDCS/RVA Rev No 1 dated 8 January 1993(12)
This document reported completion of the Stage 2 reliability analysis work. As such, it updated
the preliminary Stage 2 results presented in Reference 11.
The introduction to reliability analysis presented in Reference 9 was repeated in full. The
particulars of the considered rigs were presented but in an expanded format compared with that
adopted in the earlier reference. In the case of the 116C, the source of the error in Reference 11
referred to above could be identified. The 1-minute mean wind speed adopted in Stage 2 was in
fact the same as that for Stage 1, i.e., 40 m/s. The main reason for the 116C in 65 m water depth
being safer than when in 55 m water depth, as noted above, was identified. The maximum
wave height considered was 15.6 m instead of 18.4 m. Also, a reduced 1-minute mean wind
speed was used 35.6 m/s compared to 40 m/s.
The Stage 1 and 2 results were compared. The deterministic safety factors were broadly similar.
The exceptions were that the stress check for the 116C in 55 m water depth was reduced by 14%
whilst the overturning stability for the F&G Mod VI was increased by 25%. These differences
were more or less reflected in the reported safety indices.
The increased variability in the Gulf of Mexico environment was noted to lead to a flattening
of the corresponding safety index versus safety factor curves. The consequential challenge for
improving safety levels in such conditions compared with those areas of smaller variability, i.e.,
the North Sea, was noted.
The relatively small changes in safety indices realised between Stages 1 and 2 were noted
except in the case of stress checks where the introduction of leg inclination moments led to a 5%
reduction.
The SDOF approach briefly described in Reference 11 was presented in detail. The results
indicated that the RP was not precise enough in defining the parameters to be used in such an
assessment and that this had to be addressed.
Similarly the calibration methodology introduced in Reference 11 was presented in more detail.
The results were presented and discussed concentrating on the approach whereby target safety
indices were determined based upon the average achieved for each limit state. The determined
55
target reliabilities fell in the range 1.5 to 2.65. Partial factors were determined and resulting
indices compared for a single factor and separate static and dynamic factors. Relatively large
dynamic factors were determined (1.35 to 2.2) remembering in the safety format adopted that
these are multiplied by the static factor (1.20 to 1.25) to obtain the total factor on the dynamic
component of loading.
The factors were noted to be load effect factors, i.e., they were applied after an analysis. This is
inconsistent with other LRFD practice.
Joint Industry Jack-Up Committee, Summary of the Invitation Meeting of 11th, 12th
March 1993 at Noble Dentons Offices in London (including attachments)(13)
The document summarised the results of detailed assessments undertaken by the contractor as a
result of a number of questions concerning the results reported to that time that had been raised
at a meeting about one month earlier. The questions related to:
differences in the determined dynamic load component between the RPs SDOF and
detailed time domain analyses.
the trend of the dynamic load component from the detailed time domain analyses.
regular wave and SDOF load component results both of which conflicted with those of other
studies.
using 3 hour rather than 1 hour time series in time domain simulations.
adopting an improvement to the Winterstein method that matched exact results and
reduced MPM values by up to 15%. It resulted in a reduced static component and an
increased DAF. Safety indices were apparently little affected.
the conflicts with other results seemed largely a result of different inputs and assumptions.
a consequence of the change in the proportion of static to dynamic load components was a
reduction in the dynamic load partial factor. A value of unity was recommended.
56
terms. It was argued that the usefulness and reduced risk of using a single factor more than
outweighed the increased reliability spread.
The safety formats proposed for each of the considered limit states (overturning, preload and
stress) were summarised.
No attempt was made here to repeat the mean level 2 reliability analysis assessment reported
above when discussing Reference 9. The reason for this is that not all of the details necessary
for such an exercise are included in Reference 15. The exercise could have been conducted in
connection with Reference 12 but, as noted above, the results therein have been corrected and
not all the relevant details are presented in the later document.
Notwithstanding, the same weaknesses recorded in relation to Reference 9 still exist. Thus, in
relation to overturning and preload, the treating of the resistance terms as deterministic implies
that the resulting safety indices only relate to probability of exceedance and not probability of
failure. Thus the plots shown in Figures 1 and 2 of the report need to be lowered. The extent of
this reduction is, in the case of overturning of the 116C, equivalent to a decrease in of some
0.2.
Jack-Up Site Assessment Procedures. Establishment of an International Technical
Guideline Seminar of 23rd September, 1993 at City University, London(16)
and
Joint Industry Project. Jack-Up Site Assessment Procedures Establishment of an
International Recommended Practice. Seminar(17)
These documents contained the minutes and copies of overheads for the Seminar. The material
appeared to be an expanded version of that contained in Reference 14.
In the presentation on Structural Modelling, it was noted that the load factors were to apply to
load effects and not loads.
8.3 SUMMARY
A survey of previous offshore component reliability studies demonstrated a range of of 2.1 to
3.7 on a 20-year return period exposure basis. A of 3.0 was concluded to be an appropriate
annual component reliability.
A deterministic loading sensitivity study on a Marathon LeTourneau 116C found it most
sensitive to wave height, drag coefficient, wave period, current velocity and wind velocity.
The jack-ups used in the study were the 116C (55 m water depth) plus the Friede and Goldman
L780 Mods 5 (97 m water depth) and 6 (118 m water depth). The results of deterministic
analyses for these units at their chosen locations were reported. They exploited equivalent leg
models to determine drag loading, and leg, hull and leg-hull interaction stiffness properties for
input to a large deflection time-domain equivalent leg model. Spudcan fixity was ignored
during the early phases. Some of the resulting safety factors appeared to be too low.
The probability density functions adopted for the random variables were listed based on the
results of a literature survey. Coefficients of variation for these loading parameters were
reportedly determined from HSE Guidance. It is not clear how this was done and further
explanation is required. Some of the values seem inappropriate.
Wind and wind-driven current were assumed correlated with the seastate whilst the tidal current
was treated independently. The one-minute wind speed was a function of significant wave
height. Wave period was conditional on seastate with both its mean and standard deviation
linear functions of Hs. The JONSWAP spectrum was adopted and spreading ignored.
57
Wave particle kinematics and related coefficients were treated in a standard linear manner
consistent with the time-domain solutions that were effected. Inertia loading appeared to be
ignored. A value of 75% of the variable load was used as a mean for this loading.
Uncertainties associated with FE modelling and analysis and P- effects were ignored.
A second parameter sensitivity study involving weightings on the expected influence on jack-up
response and on the likely level of inherent uncertainty led to an unsatisfactory conclusion
regarding the choice of variables to be included in the reliability analysis. The basic list of
random variables was unchanged except for the addition of yield stress.
The safety format proposed for use was similar to that adopted in the API LRFD study.
Because the resistance formulations to be used were unchanged from their source documents,
API LRFD and AISC LRFD, the resistance factors existing therein were adopted unaltered. The
absence of any consideration as to the accuracy or otherwise of the strength formulations is of
concern.
When considering variable loading for the reliability analysis, 75% was used. Some 50-year
return period values used in the reliability analysis differed from those concluded initially as
appropriate. The MPM value of each of overturning, preload and unity check was used as the
design value of response. A response surface was generated for each failure mode as a function
of the loading variables and their products. The surfaces were simplified to facilitate
computation seemingly without significant loss of accuracy. A review of the safety index
findings and comparisons with the API LRFD findings led to the conclusion that = 2.5 was an
appropriate target.
Some of the presented Stage 2 preliminary results were in error or lacked adequate reporting.
Certain errors have been readily identified by reference to later work. A fourth rig, a 116C in 65
m water depth, was included in the Stage 2 assessment.
An extension of the work to encompass GOM environments was included simply by modifying
the mean and COV of the environmental loading uncertainty. The choice of time step was
examined but not satisfactorily resolved.
The Stage 2 results were ostensibly similar to those of Stage 1. This finding lacks transparency
because it is only during Stage 2 that the draft version of the RP being used was identified (11th
Draft) and changes to previous drafts were not reported. Further, the critical damping was
increased from 2% to 4%, and a factor of 1.07 was adopted to account for the use of a stochastic
analysis. In later Stage 2 work, a review of results led to modifications to the method of
determination of MPM values resulting in smaller quasi-static components and increased DAFs.
A consequence of this was a considerably reduced partial factor on the dynamic component of
loading compared with the Stage 1 findings.
Both the Stage 1 and 2 reliability assessments relating to overturning and preload suffer from
treating the resistance term as deterministic. The presented results thus relate only to probability
of exceedance and real safety indices have been over-estimated. The extent of this has been
determined to be equivalent to a reduction in of 0.2 in the case of overturning of the 116C. It
has been further shown, using a mean level 2 reliability analysis, that the safety factor on
overturning of 1.1 in 5.5 B corresponds to a safety index for the 116C of some 1.54 equivalent
to a probability of failure of 6 x 10-2.
The final recommendation on partial factors was unity for the dynamic component and 1.25 for
the total environmental loading. The final values were noted not to produce the smallest spread
of reliabilities. However, the ease of use and reduced risk from adopting a single partial factor
for all limit states apparently outweighed the increase in spread that was apparently not
significant in absolute terms.
58
The load partial factors were noted in at least two separate documents to apply to load effects
and not to loads.
59
60
9. CONCLUSIONS
9.1 CONCLUSIONS
9.1.1 Comparison of Analysis Results
A Marathon LeTourneau 116C was subjected to site specific assessments in accordance with
current SNAME (LRFD) practice, denoted 5-5A, a working stress design (WSD) alternative,
denoted 5-5B, and a North Sea Annex to 5-5A that exploits 100-year joint probability
environmental load events in place of the combined 50-year return period extremes explicit in 55A. The unit was positioned in the North Sea in 59 m water depth and subjected to
environmental parameters determined from data provided by Shell Expro. A quasi-static
structural analysis approach was adopted to determine member forces. Spudcan fixity was
considered on the basis of the 5-5A provisions determined for soil information also provided by
Shell Expro.
A detailed leg and hull finite element model was constructed for the analysis. Pinion-induced
loading is of importance in the 116C: the offsets associated with these were accurately
modelled. The guides were also correctly represented using non-linear springs, the lower guide
being located at the two positions of concern, ie, the brace-rack chord intersection and mid-way
between these nodes. The structural analysis package included both global and local large
deflection (P-) effects.
The natural periods were determined as 8.3 s in sway, 8.2 s in surge and 6.8 s in yaw.
Maximum member utilisation ratios (UR) were generally found with the guide mid way
between nodes. The maximum values were in the reinforced chords just below the lower guides
where URs of the order of 1.8 were recorded. These largest values arose when using 5-5A for
the checks. The leg diagonals and the leg horizontals within the depth of the hull were the next
most heavily utilised groups with maximum URs around 0.85. The maximum recorded UR for
the unreinforced chord was 0.75. The least utilised member groups were the increased thickness
horizontals and the span breakers, with URs of 0.5 and 0.3, respectively.
The 5-5B URs averaged some 10% less than those found using 5-5A. The NSA checks were on
average 5% less than those found using 5-5A.
When load factors are applied to load effects following an analysis rather than to loads prior to
effecting an analysis, significantly smaller URs result. Preload URs were low. None of the
practices (5-5A, 5-5B and the NSA) satisfied the overturning check by considerable margins.
The soil bearing capacity URs were found to be less than 0.5, although the 5-5B UR was over
20% larger than the 5-5B value. Similar differences were found in relation to the sliding check
but the larger UR here was 0.72. Uplift was occurring at the leg in question but, while neither
5-5A nor 5-5B suggest this condition is outside their range of applicability, such a condition will
fail the overturning moment check.
9.1.2 Critique of Proposed 5-5B Practice
The proposed WSD alternative to the SNAME LRFD requirements was critically reviewed for
consistency with its base documents accuracy, both technical and editorial.
Many editorial errors were found: these have been identified. A considerable number of
technical issues have been raised. The important ones are:
61
For a non-linear analysis, 5-5A sets the Cm beam-column parameter to unity. 5-5B provides
no guidance as to its appropriate value.
The adopted tubular local buckling check is based on API RP 2A. This is flawed in that it
has not been properly non-dimensionalised. A well-documented accurate alternative is
proposed.
The use of a definition on slenderness for rectangular elements supported on one edge that
involves the web yield stress is questioned since it does not seem to apply to some rack
chord configurations.
The locations at which stress checks need to be performed on a tear drop chord section are
inadequate because they overlook the possibility that the section may be composed of
elements having different yield stresses.
The use or not of load factors in the foundations checks can be expected to generate very
large differences in utilisations. This is particularly the situation when the environmental
and inertial loads are large compared to the dead and live loads. Utilisations to 5-5A will
always be greater than those to 5-5B provided the environmental and inertial loads are
acting in the same direction (compressive) as the dead and live loads.
The relative severities of the permitted metocean parameter combinations (100-year return
joint probabilities, 50-year return combination of extremes) needs demonstrating.
The equivalence of a kinematics reduction factor applied to wave height instead of to wave
particle kinematics needs confirmation in view of the likely significant difference in wetted
structure.
9.1.3 Comparison of 5-5A and 5-5B Allowable Stresses and Implied Safety
Factors
A one-to-one comparison of allowable axial compression, bending and combined compression
and bending stresses has been conducted. Findings from this are:
In 5-5A, the use of definitions on slenderness for rectangular elements supported on one or
two edges that involve a measure of residual stress leads to inconsistent definitions in that a
slender element may be classified one class more compact than a more stocky element.
Cover plates should be added to the group of elements supported on two sides included in
the 5-5A and 5-5B definitions in order to formalise their use as stiffening elements of some
rack chord cross-sections.
The definition on slenderness for axial compressed tubulars adopted in 5-5B is totally
inconsistent with corresponding definitions found elsewhere for reasons noted above.
The definition for the moment capacity of noncompact tubular sections given in 5-5A is in
error. The maximum error is demonstrated to be some 9% around the range of D/t =60.
Some of the definitions for slenderness in both 5-5A and 5-5B are in terms of the web yield
stress. This is not necessarily appropriate for rack chord sections because the definition has
been extracted from those relevant to hybrid I-girders. It is an important issue since some of
the definitions relate to the local buckling of elements supported on one edge, the buckling
of which can be quite catastrophic. Relevant requirements must be derived with caution and
recognise that rack chord sections cannot be necessarily equated to hybrid I-girders in all
respects.
Exploitation of the hybrid I-girder analogy is carried further by 5-5A to include the
definition of bending strength. The immediate application to some rack chord sections is
questioned particularly as they may ignore the onset of tensile yield since asymmetry of the
cross-section is not treated adequately. On the other hand, the requirement to use the
minimum section modulus irrespective of direction of bending leads to an under-estimation
62
by some 33% of the 116Cs leg chord section. This inconsistent treatment needs
rationalisation.
The slenderness limits for noncompactness in the presence of bending are significantly more
generous under 5-5A than under 5-5B. This is a direct result of including a residual stress
term in the 5-5A definition, the consequences of which were anticipated above.
The more fundamental approach adopted in 5-5B for the determination of bending strength
rather than trying to adapt existing formulations for hybrid I-girders as adopted in 5-5A is to
be strongly preferred since it provides for a direct opportunity to check the safety of the
cross-section. Notwithstanding, more information is needed than provided in 5-5B and its
structural checks extended to account fully for plastic capacities if its approach is to provide
the appropriate balance in jack-up assessment between safety and economy.
A comparison of allowable axial compressive stresses for typical 116C tubular sections
demonstrates that the 5-5B requirements range from 14 to 19% greater. This is
demonstrated to be approximately equally due to differences in the basic strength curve and
in the safety factors, ie, the load and resistance factors in the case of 5-5A and the one-third
increase in the case of 5-5B coupled with a typical WSD safety factor. Similar benefits are
found for 116C rack chord cross-sections.
In the case of bending of tubulars, the differences are only marginally in favour of 5-5B.
This arises from a much lower basic strength curve in 5-5B being offset by the effects of
safety factors. In contrast, rack chord cross-sections benefit from being designed to 5-5A by
up to 48% for compact sections. This is a direct consequence of allowing plastic moments
to fully develop in prismatic sections and accounting correctly for their interaction.
Under combined compression and bending, over a practical range of geometries, the
average allowable 5-5B stress is 1.06 times that of 5-5A. The maximum ratio is 1.19 whilst
the minimum is 1.0. For rack chord cross-sections, the 5-5A allowable stresses average
1.36 times the 5-5B average value with individual values varying from 0.92 to 1.78.
A simple comparison between 5-5A and 5-5B safety factors under extreme environmental
conditions is presented. All of the 5-5A factors are larger than their 5-5B counterparts,
frequently by a significant margin. The need to account fully for the inherent parameter
uncertainties in any assessment is noted if the adequacy of the implicit safety margins is to be
determined. The absence of any evaluation of redundancy in such an assessment is noted.
9.1.4 Basis of 5-5A Reliability Analysis
The considered reliability analyses work was performed in two stages.
In Stage 1, a survey of previous offshore reliability analysis and partial factor derivation work
was concluded to demonstrate that a of 3.0 represented an appropriate annual target index.
However, as a result of studies on three jack-up units, this was reduced to 2.5. The three units
involved were a Marathon LeTourneau 116C and Friede and Goldman L780 Mods 5 and 6. The
water depths selected for the analyses were 55 m, 97 m and 118 m respectively.
The basic random variables selected for consideration were wave height, wave period
conditional on seastate, wind speed correlated with seastate, tidal current with the wind-driven
component correlated with seastate, leg (simplified) drag loading and an unspecified LRFD
code check equation. Modelling uncertainties of all descriptions were ignored.
The analysis proceeded on the basis of the Most Probable Maximum (MPM) value for
overturning, sliding, preload and UR in a 3-hour storm. Non-Gaussian responses were
converted to near-Gaussian responses through Wintersteins Hermite polynomial transformation
process. Approximations were used in the derivation of some of the relevant parameters: these
were not necessarily transparent.
63
was determined using a response surface approach, the response surface being defined in
terms of the random loading variables and their cross-products. A conditioning process was
used to reduce the complexity of the surfaces. Again it is not clear how this was implemented.
When considering overturning and preload, the resistance term in the reliability equation was
treated as deterministic. The resulting therefore represents probability of exceedance and not
probability of failure. In the case of overturning of the 116C, the reliability index corresponding
to failure was determined using a mean level of 2 analysis to be 0.2 less than the represented
value. Using the 5-5B safety factor on overturning suggests the corresponding failure
probability is some 6x10-2 per annum. In the case of the structural reliability assessment, only
one of the two controlling equations on beam-column strength has been used. This will
overestimate the reliability of members dominated by bending: the extent of this has not been
questioned.
The Stage 2 work included a 116C in 65 m of water but with reduced wave heights. An
extension included the Gulf of Mexico environment on a simplistic basis. The determined
reliabilities in Stage 2 were reportedly similar to those of Stage 1: this is not readily apparent in
the light of the changes made to the analyses between the two stages.
In at least two of the considered documents, it is recorded that the load factors be applied to load
effects, not loads. This is contrary to traditional limit state (LRFD) practice.
64
REFERENCES
1.
IADC
Practice for Site Specific Assessment of Mobile Jack-up Units Working Stress Design
(WSD)
First Edition, May 1999
2.
SNAME
Recommended Practice for Site Specific Assessment of Mobile Jack-up Units
T&R 5-5A, Revision 1, May 1997
3.
SNAME
North Sea Annex to SNAME T&R Bulletin 5-5A Draft for Industry Comment
Revision D5, September 1999
4.
SHELL EXPRO
Metocean Data for Selected North Sea Fields
December, 1999
5.
6.
7.
8.
P A F A CONSULTING ENGINEERS
Implications for the Assessment of Existing Fixed Steel Structures of Proposed ISO
13819-2 Member Strength Formulations
C031-002-R Rev 1, February 2000
9.
66
Table 1
Water levels at selected site
50-year individual
Nominal LAT (m)
Storm surge (m)
Tidal rise MHWS (m)
Max SWL ref. LAT (m)
Crest elevation (m)
Total water level
(max SWL + Crest elevation) (m)
100-year
individual
59
1.60
2.49
4.09
11.2
59
1.53
2.49
4.02
not available
15.3
Table 2
Environmental data for selected site
50-year
individual
100-year
individual
100-year joint
33.5
39.5
34.5
40.7
see discussion
in Section 2.4
10.6
11.1
18.6
14.2
10.8
11.2
19.0
14.3
Current velocity
Depth-mean velocity (m/s)
Surface current velocity (m/s)
0.95
1.02
0.99
1.06
Table 3
Current velocity profile
Proportional height
above sea-bed
1.00d
0.75d
0.50d
0.25d
0.15d
0.05d
0.01d
Ratio of velocity to
depth-mean velocity
1.07
1.07
1.07
0.97
0.90
0.77
0.61
0.56
0.60
Table 4
Soil profile
Description
very soft CLAY
sandy CLAY Su = 200 kPa
medium dense SAND
Table 5
Dead load summary
Description of Loading
Hull basic and fixed load
Cantilever assembly
Bow leg and spudcan
Port leg and spudcan
Starboard leg and spudcan
Total
Weight (MN)
42.7
8.1
8.7
8.7
8.7
76.9
LCG (m)
42.82
60.23
18.29
57.61
57.61
45.23
TCG (m)
0.24
-0.29
0.00
-21.64
21.64
0.10
LCG is measured aft of hull bow and TCG is measured off the centreline
of the hull, taken as positive towards starboard.
Table 6
Load factors for assessment in accordance
with 5-5A and NSA
Limit
State
ULS
Dead Loads
Variable Loads
(D)
1
1.0
(L)
2
1.0
Environmental
Loads
(E)
3
1.25
Dynamic
Loads
(Dn)
4
1.0
Table 7
Foundation stability assessment possible paths through the foundation check
(see Figure 11)
Path
Number
through
Figure 11
1
2
3
4
5
6
7
8
9
10
11
Step 1a
Preload
Check
Leeward
Leg
Step 1b
Sliding
Check
Windward
Leg
Step 2a
Capacity
Check
Leeward
Leg
Step 3
Displacement
Check
All
Legs
Pinned
Footing
Step 2b
Capacity
Check
Windward &
Leeward
Legs
Foundation
Fixity
Pinned
Footing
Pinned
Footing
Non-linear
Foundation
Foundation
Acceptance
Note: The and symbols represent the foundation assessment passing or failing each step of the
foundation stepwise check procedure.
Table 8
Typical base shear and overturning moment magnitudes
factored (5-5A & NSA) and unfactored (5-5B)
Hdet (m)*
Period (s)
Surface current (m/s)
Reference wind speed (m/s)
Wave + current base shear (MN)
Wind base shear (MN)
Inertial loadset base shear (MN)
Total base shear (MN)
Total overturning moment (MNm)
5-5A
17.0
14.2
1.02
39.5
18.6
2.6
6.5
27.7
1895
5-5B
17.0
14.2
1.02
39.5
15.0
2.1
4.5
21.5
1475
NSA
17.3
14.3
0.6
34.5
17.0
2.0
5.6
24.6
1806
Table 9
Maximum member utilisation ratios for 5-5A
Group
Reinforced
chord (=1.00)
Reinforced
chord (=1.45)
Reinforced
chord (=2.25)
Unreinforced
chord
Guide at node
Wave
Direction Element
(deg)
Max
UR
Guide at midspan
Wave
Max
Direction Element UR
(deg)
Midspan UR
/Node UR
60
98
1.76
60
2209
1.86
1.06
60
98
1.41
60
2209
1.50
1.06
60
98
1.22
60
2209
1.27
1.04
30
150
0.80
30
150
0.75
0.94
Horizontal
30
195
0.92
187
0.87
0.95
Span breaker
60
656
0.30
60
654
0.31
1.03
Diagonal
30
402
0.80
402
0.84
1.05
30
2066
0.47
2081
0.47
1.00
30
2105
0.79
30
2105
0.74
0.94
Increased thk.
Horizontal
Increased thk.
diagonal
Table 10
Maximum member utilisation ratios for 5-5B
Group
Reinforced
chord, rack
Reinforced
chord, guide
Unreinforced
chord, rack
Unreinforced
chord, guide
Guide at node
Wave
Direction Element
(deg)
Max
UR
Guide at midspan
Wave
Max
Direction Element UR
(deg)
Midspan UR
/Node UR
30
98
1.63
30
2209
1.74
1.07
30
102
1.13
30
2208
1.13
1.00
30
150
0.75
30
150
0.74
0.99
30
150
0.75
30
150
0.74
0.99
Horizontal
30
195
0.85
30
195
0.84
0.99
Span breaker
60
654
0.25
60
654
0.25
1.00
Diagonal
30
402
0.70
30
402
0.75
1.07
30
2066
0.44
30
2066
0.45
1.02
30
2105
0.73
30
2105
0.74
1.01
Increased thk.
Horizontal
Increased thk.
diagonal
Table 11
Maximum member utilisation ratios for the NSA
Guide at node
Wave
Direction Element
(deg)
Group
Max
UR
Reinforced
chord (=1.00)
Reinforced
chord (=1.45)
Reinforced
chord (=2.25)
Unreinforced
chord
Guide at midspan
Wave
Max
Direction Element UR
(deg)
60
2209
1.64
60
2209
1.31
60
2209
1.13
30
150
0.73
Horizontal
187
0.88
Span breaker
60
654
0.28
Diagonal
402
0.85
2081
0.45
30
2105
0.71
Increased thk.
Horizontal
Increased thk.
diagonal
Midspan UR
/Node UR
*Note that guide at node analysis was not performed for the North Sea Annex
in view of the results for 5-5A and 5-5B
Table 12
Member utilisation ratio comparisons
Group
Reinforced chord rack ( = 1.00)
Unreinforced rack chord
Horizontal
Span breaker
Guide at node
5-5B/5-5A
NSA/5-5A
0.93
*
0.94
*
0.92
*
0.83
*
Diagonal
0.88
0.94
0.92
Guide at midspan
5-5B/5-5A
NSA/5-5A
0.94
0.88
0.99
0.97
0.97
1.01
0.81
0.90
0.89
1.01
0.96
1.00
0.96
0.96
Table 13
Comparison of utilisation ratios for load factors
applied to load effects and to load patterns
Load pattern
UR
1.91
1.52
1.27
0.90
1.00
0.56
1.02
0.61
1.08
Group
Reinforced chord (=1.00)
Reinforced chord (=1.45)
Reinforced chord (=2.25)
Unreinforced Chord
Horizontal
Span breaker
Diagonal
Increased thickness horizontal
Increased thickness diagonal
Load effect
UR
1.26
1.09
0.99
0.53
0.92
0.46
0.82
0.57
0.98
Table 14
Maximum utilisation ratios for preload condition
Group
Reinforced chord
(=1.00)
Reinforced chord
(=1.45)
Reinforced chord
(=2.25)
Unreinforced Chord
Horizontal
Span breaker
Diagonal
Increased thickness
horizontal
Increased thickness
diagonal
5-5A
5-5B
UR
0.37
Element
1405
UR
0.59
Element
1405
0.45
1405
0.57
1405
0.41
0.12
0.09
0.07
821
1543
2140
1750
0.60
0.30
0.16
0.11
821
1544
2141
1750
0.12
2080
0.23
2080
0.08
2137
0.17
2137
Table 15
Summary of results
5-5A
5-5B
NSA
Hdet (m) *
17.0
17.0
17.3
Period (s)
14.2
14.2
14.3
1.02
1.02
0.6
39.5
39.5
34.5
18.6
15.0
17.0
2.6
2.1
2.0
6.5
4.5
5.6
27.7
21.5
24.6
1895
1475
1806
1.86 (=1.00)
1.74 (rack)
1.64 (=1.00)
1.50 (=1.45)
1.13 (guide)
1.31 (=1.45)
1.27 (=2.25)
1.13 (=2.25)
0.75
0.74
0.73
Max UR horizontal
0.87
0.84
0.88
0.31
0.25
0.28
Max UR diagonal
0.84
0.75
0.85
0.47
0.45
0.45
0.74
0.74
0.71
0.37
0.59
0.41
0.60
0.12
0.30
0.09
0.16
0.07
0.11
0.12
0.23
0.08
0.17
0.44
0.33
0.72
0.67
1.93
1.57
1.84
Overturning check
##
#
##
Table 16
Comparisons of 5-5A and 5-5B safety factors under extreme environmental loading
Criterion
Structural Strength
Overturning
moment
Preload
Sliding
Capacity check Pinned foundation
Condition
Bending dominated
Compact I-section
Compact tubular
Axial dominated
Short column
(compact section)
Long column ( = 1)
Displacement effects treated as reduction in
stabilising moment
Leeward leg
GL only
EL only
EL/GL = 2/3
Windward leg
Sand/drained
Clay/undrained
Maximum bearing
GL only
area not mobilised
EL only
EL/GL
= 2/3
(== 0.90)
Maximum bearing
GL only
area mobilised
EL only
EL/GL = 2/3
(== 0.85)
5-5A
1.39
1.39
1.47
1.47
1.32
5-5B
1.30
1.35
1.25
1.42
1.10
1.11
1.39
1.22
1.56*
1.95*
1.11
1.39
1.22
1.18
1.47
1.29
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
Table 17
Determined distribution parameters for the important basic variables
Basic variable
Significant wave
height (annual)
Wind speed 1 hour
mean (annual)
Tidal current
Random wave drag
coefficient (CdD)
Distribution
parameter
Mean
COV
Mean
COV
Mean
COV
Mean
COV
Marathon
LeTourneau 116C
8.05 m
13%
27.7 m/s
13%
0.17
53%
10.2 m
20%
Friede &
Goldman Mod V
10.5 m
13%
27.7 m/s
13%
0.15
43%
4.90 m
20%
Friede &
Goldman Mod VI
11.6 m
13%
28.5 m/s
13%
0.14
24%
6.35 m
20%
Table 18
Factors of safety for considered units under loading conditions relevant
for reliability analysis
Jack-up Unit
Marathon LeTourneau 116C
Friede & Goldman Mod V
Friede & Goldman Mod VI
Overturning moment
1.33
1.29
1.23
Preload
0.97
1.00
1.06
Figure 1
Jack-up layout
104.632m
Tip of can
Z
Y
Figure 2
Isometric of jack-up
Guide plate
711 x 51
Rack plate
213 x 127
(Fy = 586 MPa)
Side plate
466 x 19
(a) Unreinforced
Rack plate
213 x 127
(Fy = 586 MPa)
Guide plate
711 x 51
Side plate
466 x 19
Notes:
1. All dimensions in mm
2. Fy = 483 MPa, unless noted
(b) Reinforced
Figure 3
Reinforced and unreinforced chord
PORT
CL
30
180
BOW
150
STBD
60
120
90
Figure 4
Wave directions used in analysis
Z
Y
Figure 5
Full finite element model
Figure 6
Detailed leg model
FEMGV 5.1-01.C
P A F A Consulting Engineers
Z
X
Figure 7
Deck bulkheads
2 MAR 2000
Figure 8a
Connection point for guide arrangement
Yoke
Nonlinear
spring
elements
Chord
centroid
Indicates
connection between
nonlinear springs
and stiff beam
elements
2
Upper
guide
Stiff
beam
elements
Lower
guide
Figure 8b
Connection point for guide arrangement
Figure 9
Line diagram of rack and pinion
Figure 10
Gear unit case, bracing and upper guide
Figure 11
Foundation stability assessment flow diagram
3.0
2.5
1.5
1.0
0.5
0.0
0%
20%
40%
60%
80%
100%
Figure 12
Step 1a: Preload check of the leeward leg - pinned foundation
3 .0
2 .5
c la y
2 .0
LRFD/WSD
LRFD/WSD
2.0
1 .5
san d
1 .0
0 .5
0 .0
-4 0 %
-3 5 %
-3 0 %
-25%
-20%
-1 5 %
-10%
-5 %
P e r c e n t a g e o f V e r t ic a l L o a d f r o m E n v ir o n m e n t a l & D y n a m ic L o a d s
Figure 13
Step 1b: Sliding check of the windward leg - pinned foundation
0%
3.0
2.5
LRFD/WSD
2.0
Max. bearing
area m obilised
1.5
Max. bearing
area not
m obilised
1.0
0.5
0.0
0%
20%
40%
60%
80%
100%
Figure 14
Step 2a : Capacity check of the leeward leg pinned fixity
3.0
2.5
LRFD/WSD
2.0
1.5
1.0
0.5
0.0
-40%
-20%
0%
20%
40%
60%
80%
100%
Figure 15
Step 2b : Capacity check of the leeward and windward leg foundation fixity
Figure 16
Jack-up eigenvalues, period = 8.27 seconds
Figure 17
Jack-up eigenvalues, period = 8.15 seconds
Figure 18
Jack-up eigenvalues, period = 6.81 seconds
FEMGV5.1-01.C
2104
98
656
150
2083
403
195
PAFAConsultingEngineers
2104
Figure 19
Utilisation ratios for lower guide adjacent to leg chord node
656
98
403
195
150
2083
3MAR2000
FEMGV 5.1-01.C
2104
406
654
187
403
2209
150
2083
2115
2104
P A F A Consulting Engineers
654
406
2209
Figure 20
Utilisation ratios for lower guide adjacent to leg chord midspan
187
403
2083
150
3 MAR 2000
2115
180
160
FVH_WSD = 146.8 MN
140
120
FVH_LRFD = 121.6 MN
Unfactored capacity
100
80
60
QVH_LRFD = 54.0 MN
QVH_WSD = 49.2 MN
Factored capacity
40
SWL = 30.1 MN
20
0
0
10
20
30
40
50
60
Figure 21
Bearing capacity envelope for leeward (bow) leg wave from 0
70
M n (Nmm)
1.E+10
Correct Mn
Incorrect Mn
8.E+09
6.E+09
p
4.E+09
r
2.E+09
0.E+00
0
25
50
75
100
125
Figure 22
Comparison of 5-5A correct and incorrect tubular Mn moment capacities
fa (N/mm)
500
400
300
5-5A
200
Horizontal
5-5B
100
Diagonal
Span breaker
0
0
10
15
20
25
L/D
30
Figure 23
Comparison of 5-5A and 5-5B tubular column axial compression capacities
fb (N/mm)
600
500
Diagonal
400
300
5-5A
200
5-5B
100
0
0
10
20
30
40
D/t
Figure 24
Comparison of 5-5A and 5-5B tubular bending capacities
50
Figure 25
Comparison of 5-5A and 5-5B tubular column compression and bending interaction
capacities following a full non-linear structural analysis and assuming K = 0.8
400
fa (N/mm)
300
200
5-5A
100
5-5B
Fy = 460 MPa
0
0.00
0.50
1.00
1.50
2.00
Figure 26
Comparison of 5-5A and 5-5B prismatic column compression
capacities for compact sections
fb (N/mm)
600
500
400
300
5-5A
200
5-5B
100
0
0
14
21
28
b/t
Figure 27
Comparison of 5-5A and 5-5B prismatic section
bending capacities for compact/noncompact sections
35
Figure 28
Comparison of 5-5A and 5-5B prismatic column compression and bending interaction
capacities for compact sections following a full non-linear structural analysis and
assuming K = 1.0
ISBN 0-7176-1980-X
OTO 2001/001
20.00
9 780717 619801