Вы находитесь на странице: 1из 5

Typical Reliability Analysis of Jacket Structure

General

Pushover analysis is used as a measure of reserve strength in the determination of the


probability of failure and the reliability index of a platform. In a non-linear static analysis, the
lateral loads are proportionally increased until total structural collapse (the ultimate capacity)
is reached. The ratio of the ultimate lateral loads to the design lateral loads is called the
Reserve Strength Ratio (RSR). The target reliability is decided in conjunction other hazards
threatening a structure. There are nine category of hazard whose combined effect on impairing
the TR should fall below 10 3 . Structural failure is not a major contributor, as its level is kept
reasonably low.
The target failure probability for new and high consequence structure in the UK sector of
North Sea is around 104 . For the Gulf of Mexico such target level is one order of magnitude
less. For the example platform the probability of structural failure is than 10 5 .

The target failure probability can be related to a target RSR as shown below.

Target Reliability

The target safety level is expressed by probability of failure, p ft , or reliability index, t . These
values are taken as the weighted safety level inherent in the existing platforms in its
geographical location, or in platforms in other regions of the world. The target values can
either be related to annual or lifetime risks. In establishing the target values, the use of notional
probability of failure, p ft , is preferred over the reliability index, t , since the weighted
averaging of the former is safer than the latter. This allows us to bias the calculation towards
the higher risk failure. In the re-qualification of existing structures, the target reliability level is
chosen based on judgement or with reference to the implied reliability level of a code. For
platforms whose failure is of low consequence, a reduced level of reliability is assumed.

An offshore structure faces various demands on its capacity. There are a large number of
events (e.g. storm, seismic, ice) which could initiate the failure of the structure. It is generally
adequate to assume that the probability of failure due to multiple initiating events is equal to
the sum of the individual probabilities. Furthermore, the failure probability due to the union of
such causes is approximately equal to the probability of the initiator with the largest
probability of failure. This is due to the fact that it is rare that the failure probabilities of two or
more initiating events to be of similar magnitude. Consequently, it is adequate for design
purposes to assign each primary initiator (e.g. each major load combination) the target
reliability. Hence it is not necessary to allocate a portion of the total reliability budget to each
case. These rules can become meaningless however if the load cases are defined too finely, for
instance an extreme wave is defined for each day of the year.

The current attempt to apply reliability methods to jacket structures uses the weakest mode of
failure as the only mode of failure. At least for the important case of extreme global loads such
as extreme wave plus current and wind, it is sufficient to assume that the reliability of the
structure is equal to that of its most likely failure mode. In most cases this is effectively its
deterministically weakest mode of failure. In this context the pushover load becomes useful as
a measure of structural capacity for its most significant failure initiator.

Reliability Index

As stated at the beginning of this paper, pushover load is used as a measure of reserve strength
for an intact structure, or a measure of residual strength for a damaged structure. It is assumed
that the structures analysed are not dynamically sensitive. This measure simultaneously
accounts for structural redundancy, the ability of members to allow the formation of hinges
before local buckling or column buckling, and the ability of the structure to turn into a
mechanism before global buckling. This measure is known as the Reserve Strength Ratio
(RSR) expressed as the ratio of ultimate lateral loads to the design lateral loads, i.e.
Q
ul ult (1)
Qd
where Qult is the ultimate base shear, and Qd is the design base shear.
The system reliability can be assessed by treating the whole platform as one equivalent
element[3,5 and 9]. Defining Z R L ,where R is the resistance (load carrying capacity) and
L is the load effect. Failure occurs when Z 1, i.e. when the load effect exceeds the
resistance. Assuming lognormal distributions, it can be shown that the safety index is given by
ln m R m L


ln 1 2R 1 2L
(2)

where
m R and m L are the medians of R and L respectively
R and L coefficients of variation (COV) of resistance and loads respectively.
The total lateral load (base shear), Q , is assumed to have an empirical relationship with wave
height, h , of the form Q ch . The deterministic constants c and can be determined from
the base shear values calculated for different wave heights, preferably in the region of the
collapse wave height. In this calculation the best industry practice, e.g., the latest edition of
API RP2A, should be used. The appropriate values of wave period, current and wind
associated with the wave height should be used to minimise the load uncertainties.

A probabilistic model for the annual extreme environmental loads, Qannual , may be written as:
Qannual cH (3)
where H is the annual extreme wave height, is the load modelling error, and c and are
defined as before. Parameters Qannual , H , and are random variables, and c and are
deterministic constants. It is assumed that has a log normal distribution with 10 . and
covariant of 0.20 [2]. In most simplified reliability analysis uncertainty in wave
description is ignored. The distribution of wave height is assumed to follow Gumble
distribution with parmeters given by BP, see the attached excel sheet. Consequently the
distribution of Q ( Qannual ) can also be found from the same distribution, and the median
annual base shear is given:
Q c H cH (4)
where H is the median of H a ,; the annual maxima. Assuming H a , and to be
independent, then the COV of Qannual is given by:
Q 2 2H 2 (5)

2
where H is the coefficient of variation of H a .
The median ultimate strength of the structure is given by
m R ult chd = ult Qd (6)
h Q
where d is the design wave height, and d is the design base shear. In the context of current
practice hd is equal to 100-year load, i.e. h100 .
The relationship between QH and Qd can be written as:

h
Qd QH d (7)
H
Substitution of Eq (7) into Eq (6) gives:

hd
m R ult Qd ult QH (8)
H
Substitution of Eq (8) into Eq. (2) gives:
hd


ln ult
H (9)


ln 1 2R 1 2L
Here is assumed
m L = Q c H cH , and
L Q 2 2
H 2
(10)
This is the safety index for annual extreme environmental loads. Note that RSR is expressed in


terms of design base shear, and h H to account for the difference between the design
d
wave height and the annual wave height.
The ratio hd H can be expressed as ( for lognormal model):
hd
H

exp ln 1 2H 1 1 1 rd (11)
where rd is return period of design wave hd . Substituting Eq (11) into Eq (9) gives the annual
reliability index in terms of the design wave return period:





ln ult exp ln 1 2H 1 1 1 rd

(12)

ln 1 2R 1 2L
The annual failure probability, Pf , can be calculated from Pf .

The wave height at which the lateral load becomes equal to the ultimate strength, Eq (7), is
given by:
hult hd ult
1
(13)

Application

3
Following table shows the reliability index, probability of structural failure and the wave height
associated as a function of . The value of depends on the water depth, wave environment
and associated current, marine growth, and the type of the structure. The value of for the
example jacket is around 1.7. From this table, the wave height associated with the collapse
generates wave-in-deck, which was included in the calculation of RSR. If the waves hit the
topside, the associated loads should be included in the base shear assessment. In such cases,
the RSR would be lower and the target safety index would not be achieved. This phenomenon
should be allowed for when determining .

Description ult = 2.2 ult = 2.2 ult =2,2 ult =2.2 ult =2.2
=1.5 =1.6 =1.7 =1.8 =1.9
hd = h100 (m) 15.7 15.7 15.7 15.7 15.7
H Annual extreme wave height 9.4 9.4 9.4 9.4 9.4
Annual 4.6 4.5 4.4 4.3 4.25
Pf 2.56x 3.91x 5.71x 8.02x 1.09x
106 10 6 106 10 6 10 5
Ult. Wave height hult 26.6 24.96 24.9 23.4 23.3
Results of Reliability Analysis for the jacket structure with different value. It is
assumed that in Eq 12, (see references 7,8 and 9), R =0.1, =0.1, and H =0.2.
Note: The maximum wave height distribution is assumed to be lognormal, which gives
conservative estimates.

9. Conclusions

The minimum calculated RSR is 2.2, by assuming the maximum wave height is equally likely
for all platform direction. The failure probability for the jacket structure is less than 105 - see
column for =1.7.
The comparable value for the UK sector of the North Sea is around 10 4 . In the Gulf of
Mexico higher failure rate is targeted.

References
1. Moses, F., "Development of Preliminary Load and Resistance Factors for Design
Document for Fixed Offshore Platforms, APR-PRAC Project 85-22, Final Report to the
American Petroleum Institute, Dallas, 1986.
2. Ang, Alfredo H-S and Cornell, C.A., Reliability Bases of Structural Safety and Design,
ASCE, Jour. of Strut. Div., Sept. 1974 ST9, pp1755-1769.
3. Digre, K. A., Puskar, F. J, Aggarwal, R. K., Irick, J.T., Krieger, W.F. and Petrauskas, C.,
Modification to and Applications of the Guidelines for Assessments of Existing Platforms
Contained in Section 17 of API RP2A, OTC 7779, Proc. Offshore Technology
Conference, Houston, 1995.
4. Petrauskas, C., Finnigan, T.D., Heideman, J.C., Vogel, M. Santala, M., and Merek, G.P.,
Metocean Criteria/Loads for Use in Assessment of Existing Platforms, OTC7484,Proc.
Offshore Technology Conference, Houston, 1994.
5. Marshal, P.W. and Bea R.G., Failure Modes of Offshore Platforms, BOSS76, pp579-635,
Trondheim, Norway, 1976.
6. Yasseri, S. F. and Fowler, D.F. Reassessment of Offshore Structures, Offshore
Mediterranean Conference, OMC95 Rvena, Italy, March 1995.

4
7. Zettlemoyer, N.,ISO Harmonization of Offshore Guidance on Strength of Tubular Joints,
Proceeding of Sixth (1996) International Offshore and Polar Engineering Conference, Los
Angeles, USA 1996. Vol. IVPP 9-14.
8. Gilber, R. T. and Tang W. H., Model Uncertainty in Offshore Geotechnical Reliability,
Offshore Technology Conference 1995, OTC007757. pp 557-567.
9. Yasseri, S. F. Uncertainty and Modelling Errors in Pushover Analysis and their Effect on the
Target Reliability, ERA Conference 96: Major Hazard Offshore, 1996, pp

Вам также может понравиться