You are on page 1of 218

PETRONAS TECHNICAL STANDARDS

DESIGN AND ENGINEERING PRACTICE

MANUAL (SM)

DESIGN FOR OFFSHORE STEEL


STRUCTURES EXPOSED TO
ACCIDENTAL LOADS

PTS 20.072
AUGUST 1988
PREFACE

PETRONAS Technical Standards (PTS) publications reflect the views, at the time of publication,
of PETRONAS OPUs/Divisions.

They are based on the experience acquired during the involvement with the design, construction,
operation and maintenance of processing units and facilities. Where appropriate they are based
on, or reference is made to, national and international standards and codes of practice.

The objective is to set the recommended standard for good technical practice to be applied by
PETRONAS' OPUs in oil and gas production facilities, refineries, gas processing plants, chemical
plants, marketing facilities or any other such facility, and thereby to achieve maximum technical
and economic benefit from standardisation.

The information set forth in these publications is provided to users for their consideration and
decision to implement. This is of particular importance where PTS may not cover every
requirement or diversity of condition at each locality. The system of PTS is expected to be
sufficiently flexible to allow individual operating units to adapt the information set forth in PTS to
their own environment and requirements.

When Contractors or Manufacturers/Suppliers use PTS they shall be solely responsible for the
quality of work and the attainment of the required design and engineering standards. In
particular, for those requirements not specifically covered, the Principal will expect them to follow
those design and engineering practices which will achieve the same level of integrity as reflected
in the PTS. If in doubt, the Contractor or Manufacturer/Supplier shall, without detracting from his
own responsibility, consult the Principal or its technical advisor.

The right to use PTS rests with three categories of users :

1) PETRONAS and its affiliates.


2) Other parties who are authorised to use PTS subject to appropriate contractual
arrangements.
3) Contractors/subcontractors and Manufacturers/Suppliers under a contract with
users referred to under 1) and 2) which requires that tenders for projects,
materials supplied or - generally - work performed on behalf of the said users
comply with the relevant standards.

Subject to any particular terms and conditions as may be set forth in specific agreements with
users, PETRONAS disclaims any liability of whatsoever nature for any damage (including injury
or death) suffered by any company or person whomsoever as a result of or in connection with the
use, application or implementation of any PTS, combination of PTS or any part thereof. The
benefit of this disclaimer shall inure in all respects to PETRONAS and/or any company affiliated
to PETRONAS that may issue PTS or require the use of PTS.

Without prejudice to any specific terms in respect of confidentiality under relevant contractual
arrangements, PTS shall not, without the prior written consent of PETRONAS, be disclosed by
users to any company or person whomsoever and the PTS shall be used exclusively for the
purpose they have been provided to the user. They shall be returned after use, including any
copies which shall only be made by users with the express prior written consent of PETRONAS.
The copyright of PTS vests in PETRONAS. Users shall arrange for PTS to be held in safe
custody and PETRONAS may at any time require information satisfactory to PETRONAS in order
to ascertain how users implement this requirement.
CONTENTS

1. INTRODUCTION
1.1 Purpose
1.2 Application
1.3 Background
1.4 Overview
1.5 References

2. DESIGN PHILOSOPHY
2.1 Introduction
2.1.1 General
2.1.2 Aims of the design
2.1.3 Definitions

2.2 Design approach


2.2.1 General
2.2.2 Event control
2.2.3 Indirect design
2.2.4 Direct design

2.3 Safety evaluations


2.3.1 General
2.3.2 Safety evaluation principles
2.3.3 Acceptance criteria
2.3.4 Safety evaluation methology
2.3.5 Use of results from Concept Safety Evaluations

2.4 Code requirements


2.4.1 General
2.4.2 Progressive Collapse Limit States (PLS)
2.4.3 Load coefficients and combinations
2.4.4 Characteristic loads
2.4.5 Design resistance

2.5 Commentary to code requirements


2.5.1 General
2.5.2 Determination of characteristic loads
2.5.3 Safety aspects

2.6 References
3. DESIGN GUIDANCE FOR STRUCTURES EXPOSED SHIP COLLISIONS
3.1 Introduction
3.1.1 General
3.1.2 Definitions
3.1.3 Notations

3.2 Design strategy


3.2.1 General
3.2.2 Event control
3.2.3 Indirect design
3.2.4 Direct design

3.3 Collision scenarios


3.3.1 Categorization
3.3.2 Impact geometry

3.4 Characteristic collision loads


3.4.1 General
3.4.2 Vessel size
3.4.3 Added mass
3.4.4 Ship collision velocity
3.4.5 Impact geometry

3.5 Collision mechanics


3.5.1 Fundamental principles
3.5.2 Conservation of momentum
3.5.3 Conservation of energy
3.5.4 Dissipation of strain energy

3.6 Energy dissipation in vessel


3.6.1 General
3.6.2 Supply vessels
3.6.3 Tankers and other ships
3.6.4 Barges

3.7 Design methods for unstiffened tubulars


3.7.1 General
3.7.2 Beam deformation
3.7.3 Local damage
3.7.4 Bending capacity of dented cross-section
3.7.5 Local buckling
3.7.6 Capacity of joints
3.7.7 Critical strains
3.7.8 Mechanisms involving multiple members
3.7.9 Compression members
3.8 Design methods for stiffened cylindrical shell (columns)
3.8.1 General
3.8.2 Resistance to large penetrations
3.9 Design methods for pontoons

3.10 Global integrity during impact


3.10.1 General
3.10.2 Partial safety coefficient
3.10.3 Global integrity check

3.11 References
3.12 Illustrative examples

4. DESIGN GUIDANCE FOR STRUCTURES EXPOSED TO FALLING OBJECTS


4.1 Introduction
4.1.1 General
4.1.2 Notations

4.2 Design strategy


4.2.1 General
4.2.2 Direct design

4.3 Falling object scenarios


4.3.1 General
4.3.2 Categorization of loads

4.4 Characteristic dropped object loads


4.4.1 Probability level for the individual events
4.4.2 Object behaviour under impact
4.4.3 Support structures
4.4.4 Buoyancy structures
4.4.5 Subsea installations

4.5 Impact mechanics


4.5.1 General
4.5.2 Fundamental principles
4.5.3 Conservation of momentum
4.5.4 Conservation of energy
4.5.5 Dissipation of strain energy

4.6 Plugging capacity of steel plates


4.6.1 General
4.6.2 Impact phenomena
4.6.3 Design model
4.6.4 Design recommendations

4.7 References
4.8 Illustrative example
Appendix 4A
4A 1 Static strain energy
4A 1.1 Introduction
4A 1.2 Static force-displacement curve

5. DESIGN GUIDANCE FOR STRUCTURES EXPOSED TO EXPLOSIONS


5.1 Introduction
5.1.1 General
5.1.2 NPD regulations and guidelines
5.1.3 Definitions

5.2 Design strategy


5.2.1 General
5.2.2 Direct design
5.2.3 Combination with fire

5.3 Explosion scenarios


5.3.1 General
5.3.2 Detonations
5.3.3 Deflagrations
5.3.4 Pressure explosions

5.4 Characteristic explosion loads


5.4.1 General
5.4.2 Unconfined gas cloud explosions
5.4.3 Confined explosions
5.4.4 Partly confined explosions

5.5 Structural response


5.5.1 General
5.5.2 Structural resistance
5.5.3 Dynamic amplification
5.5.4 Elasto-plastic response

5.6 Examples
5.6.1 Structures exposed to explosions General
5.6.2 Explosion in a module
5.6.3 Recommendations to reduce the build-up pressure
5.6.4 Explosion in the derrick area
5.7 References
6. DESIGN GUIDANCE FOR STRUCTURES EXPOSED TO FIRE
6.1 Introduction
6.1.1 General
6.1.2 Definitions

6.2 Design strategy


6.2.1 General
6.2.2 Event control
6.2.3 Indirect design
6.2.4 Direct design
6.2.5 The need for a interdisciplinary approach

6.3 Protection systems


6.3.1 General
6.3.2 Active protection systems
6.3.3 Passive protection systems

6.4 Fire scenarios and characteristic heat loads


6.4.1 Classification of fires
6.4.2 Fire in superstructure
6.4.3 Fire at sea level
6.4.4 Relevant heat load for pool fires
6.4.5 Calculations of the thermal response

6.5 Structural response at elevated temperature


6.5.1 Material properties
6.5.2 Thermal expansion
6.5.3 Beam-columns
6.5.4 Plates

6.6 References
6.7 Illustrative example

7. DESIGN GUIDANCE FOR STRUCTURES EXPOSED TO EARTHQUAKE

7.1 Introduction
7.1.1 General
7.1.2 Earthquake hazard
7.1.3 NPD regulations and guidelines
7.1.4 API RP2A guidelines
7.1.5 Post-elastic structural behaviour
7.1.6 Structural configuration
7.1.7 Definitions
7.2 Design earthquake loads
7.2.1 Site specific study
7.2.2 Response spectrum
7.2.3 Time history
7.2.4 Multicomponent earthquake excitation

7.3 Response analysis of the substructure and the deck


7.3.1 General
7.3.2 Soil-structure interaction analysis
7.3.3 Response spectrum analysis

7.4 Response analysis of the topside main structures


7.4.1 General
7.4.2 Discussion of alternative analysis methods
7.4.3 Response spectrum analysis

7.5 Response analysis of the equipment and the secondary structure


7.5.1 General
7.5.2 Quasi-static analysis
7.5.3 Simplified response spectrum method

7.6 Strength and ductility evaluation


7.6.1 General
7.6.2 Equipment classification
7.6.3 Equipment-structure interaction
7.6.4 Use of the behaviour factor q
7.6.5 Code checking

7.7 Examples and guidelines for structural configuration


7.7.1 Structural configuration
7.7.2 Brace configuration in jacket
7.7.3 Tubular steel members
7.7.4 Response spectrum analysis
7.7.5 Articulated columns
7.7.6 Tension leg platforms

7.8 Reference
7.9 Illustrative example
8. DESIGN GUIDANCE FOR RESIDUAL STRENGTH ANALYSIS
8.1 Introduction
8.1.1 General
8.1.2 Definitions
8.1.3 Notations

8.2 Design strategy


8.3 Capacities of damaged elements
8.3.1 Dented unstiffened tubulars
8.3.2 Cracked unstiffened tubulars
8.3.3 Stiffened members with initial deformations
8.3.4 Plated members with permanent deflections
8.3.5 Built-up members with initial deformations

8.4 System analysis


8.4.1 Linear methods
8.4.2 Extended linear methods
8.4.3 Nonlinear techniques
8.4.4 Cyclic degradation of platform stiffness

8.5 References
8.6 Illustrative example
1. INTRODUCTION

1.1 Purpose
The purpose of this document is to give guidance and to serve as a design tool on how to
obtain and document required strength of offshore steel structures against abnormal
environmental and accidental loads.

1.2 Application
The design guidance applies to offshore steel load-bearing structures. The safety format and
design basis given in this document comply with the requirements in "Regulation for structural
design of load-bearing structures intended for exploitation of petroleum resources",
Norwegian Petroleum Directorate (NPD), 1984 /1/ and "Guidelines for safety evaluation of
platform conceptual design", 1981 /2/. Reference is also made to NPD "Guidelines for the
determination of loads and load effects", 1987 /3/.
The design philosophy given in Chapter 2 is also relevant for structures made of other
materials.
In addition, information is provided for assessing the residual strength of damaged structures.
This part will also be useful during reassessment of strength during operation.
It is emphasized that the design against accidental loads is not a straight forward task and,
consequently, the user of this guidance should spend some time to get familiar with the text.

1.3 Background
It is a fundamental requirement that offshore platforms shall have adequate strength against
all effects influencing their structural integrity. Traditionally structures are designed against the
effects of ordinary loads like dead load, live load and environmental loads. With the activity
related to oil exploitation by offshore platforms, the structural integrity may be threatened by
incidents caused by malfunction of equipment, improper operation or similar unexpected
events. Even by taking all reasonable measures to reduce the risk of such incidents it will be
difficult to eliminate such incidents totally. This is reflected in the recorded statistics of
offshore accidents. Realizing the threats to the integrity of the platform from these accidents,
guidelines are needed to account for accidental loads.
There exist procedures for dealing with ordinary loads like dead loads, live loads and
environmental loads. Procedures for accidental loads are not generally available. For this
reason there has been a considerable research effort to determine the nature of accidental
loading and the corresponding structural response and residual strength. The knowledge
gained from this research has not been implemented in the design codes, and the research
results are not always easy accessible to the designer. The accidental loads confront the
designer with unconventional design situations where large deformations and inelastic
material behaviour must be taken into account. This guidance is meant to be a tool for the
designer to achieve sufficient strength and ductility against abnormal environmental and
accidental loads.
1.4 Overview
The guidance contains recent research results in a condensed form and is intended to
represent the state of the art.
The following accidental and abnormal environmental events have been given a dedicated
chapter in the guidance:
- Ship collisions
- Falling objects
- Explosions
- Fires
- Earthquakes
In addition one chapter deals with the general philosophy for design against abnormal
environmental and accidental loads, and one chapter contains guidance for assessing the
residual strength.
This document should provide a general understanding of the philosophy for the design
against, and the physical phenomena connected with abnormal environmental and accidental
loads.
Guidance Is given on how to establish characteristic loads and load effects.
Design strategies and recommended design practices for selected types of structures are
presented and relevant limit states are defined. Design tables, formulas and procedures for
assessment and rational calculation of structural resistance are given together with
calculation examples. Finally, each topic includes references to recommended literature.
Detailed discussion of abnormal wave, current and wind loads are excluded here because
they are well treated in other documents. Accidental loads related to blow-out, helicopter
collision, loss of intended pressure difference, and unintended distribution of ballast, may be
relevant accidental conditions (see /1/ and /2/), but are not covered in this document.

1.5 References
/1/ NPD:
"Regulation for structural design of load-bearing structures intended for exploitation of
petroleum resources", Norwegian Petroleum Directorate, 1984.

/2/ NPD:
"Guidelines for safety evaluation of platform conceptual design", 1981.

/3/ NPD:
"Guidelines for the determination of loads and load effects", Norwegian Petroleum
Directorate, 1987.
2. DESIGN PHILOSOPHY

2.1 Introduction
2.1.1 General
The objective of this chapter is to give a general overview of the design philosophy and
principles with respect to the design against abnormal environmental and accidental loads
and to explain the requirements and guidelines given by the Norwegian Petroleum
Directorate, (NPD).
Section 2.2 deals with design against abnormal environmental and accidental loads through
event control, indirect and direct design. Section 2.3 outlines the role of safety analyses in
connection with design against accidental loads Section 2.4 gives the specific requirements
stated by NPD. Section 2.5 lists references to recommended literature.

2.1.2 Aims of the design


The main requirement for the design against abnormal environmental and accidental loads is
to ensure that the structure has adequate safety against an unintended event developing into
significantly greater damages than caused by the initial event.
The major aim of the design is to avoid:
- loss of life
- significant pollution
- considerable economical losses
The design should provide a sufficient level of safety both during the accidental situation and
for a specified period of time after the accident has occurred. This time will depend on
necessary time for evacuation of personnel, risk of significant pollution etc.
NPD may permit a control of the limit states for progressive collapse to be disregarded
provided a comprehensive assessment demonstrate that collapse will not cause the above
consequences.

2.1.3 Definitions
Load-bearing structure: That part of the facility whose main function is to
transfer loads.
Accidental Event: An undesired incident or condition which, in
combination with other conditions (e.g.: weather
conditions, failure of safety barrier, etc.), determines
the accidental effects.
Accidental Effect: The result of an accidental event, expressed in terms
of heat flux, impact force and energy, acceleration,
etc. which is the basis for the safety evaluations.
Design Accidental Event (DAE): An accidental event which results in effects that the
platform should be designed to sustain.
Residual Accidental Event (RAE): An accidental event which results in effects that the
platform is not assumed capable to sustain.
Acceptance criteria: Functional requirements which are concerned with
the platforms, resistance to accidental effects. This
should be in accordance with the authority's
definition of acceptable safety levels.
Active protection: Operational actions and mechanical equipment
which are brought into operation when an accident is
threatening or after the accident has occurred, in
order to limit the probability of the accident and the
effects thereof, respectively. Some examples are
safety valves, shut down systems, water drenching
systems, working procedures, drills for coping with
accidents, etc.
Passive protection: Protection against damage by means of distance,
location, strength and durability of structural
elements, insulation, etc.
Event control: Implementation of measures for reducing the
probability and consequence of accidental events,
such as changes and improvements in equipment,
working procedures, active protection devices,
arrangement of the platform, personnel training, etc.
Indirect design: Implementation of measures for improving structural
ductility and resistance without numerical
calculations and determination of specific accidental
effects.
Direct design: Determination of structural resistance, dimensions,
etc. on basis of specific design accidental effects.
Load: Any action causing load effect in the structure.
Characteristic load: Reference value of a load to be used in
determination of load effects when using the partial
coefficient method or the allowable stress method.
Load effect: Effect of a single load or combination of loads on the
structure, such as stress, stress resultant (internal
force and moment), deformation, displacement,
motion, etc.
Resistance: Capability of a structure or part of a structure to
resist load effect.
Characteristic resistance: The nominal capacity that may be used for
determination of design resistance of a structure or
structural element. The characteristic value of
resistance is to be based on a defined percentile of
the test results.
Design life: The time period from commencement of construction
until condemnation of the structure.
Limit state: A state where a criterion governing the load-carrying
ability or use of the structure, is reached.

2.2 Design approach


2.2.1 General
Design against abnormal environmental and accidental loads influences the arrangement of
topside facilities as well as the choice of load-bearing systems, materials, etc. The type and
magnitude of accidental loads are to be determined in the conceptual design stage. If the
design check involving abnormal environmental and accidental loads is accomplished early in
the design assessment, required safety can often be achieved at negligible extra cost.
The approaches for design against abnormal environmental and accidental loads may be
categorized as:
- event control
- indirect design
- direct design

2.2.2 Event control


By event control is meant the implementation of measures to reduce the probability and
consequence of accidental events. This may be made by changes and improvements in e.g.:
- equipment
- working procedures
- active protection devices
- arrangement of the platform
- structural configuration
- personnel training
The design work should be initiated by a study of the operations planned in each individual
area of the platform and during each phase of the platform life. The potential hazards involved
should be identified and the magnitude of accidental effects at the specified low probability
level, should be obtained by use of safety evaluation methods, see Section 2.3.
On the basis of the accidental effects established during the safety analysis, the adequacy of
the design may be verified as described in the following chapters. If compliance with criteria is
not achieved, the situation may be improved by strengthening the structure or by event
control measures as given above, e.g. by changing the load-bearing system.
Normally the adequacy of the load-bearing system and platform arrangement has to be
verified prior to the detailed design of the equipment. In such cases reasonable design values
may simply be chosen by the designer. Later, during the detail design, it is then to be verified
that the equipment and passive protection fulfill the assumptions made in the earlier design
phases.

2.2.3 Indirect design


By indirect design is meant implementation of measures for improving structural ductility and
resistance without numerical calculations and determination of the specific accidental effects.
Indirect design measures reduce and may in some cases eliminate the amount of direct
design work.
Many accidental loads are of the impact type. The structure has to absorb a certain amount of
impact energy rather than transfer a specific load. Among such accidents are ship collisions,
earthquakes, dropped objects, etc. By providing redundancy in the structure such that
alternate load paths may be developed, the systems resistance is improved without directly
considering the accidental loads in question, i.e. indirect design.
Energy absorption requires the structure to behave in a ductile manner. Measures to obtain
adequate ductility are:
Provide connections of primary members with a strength exceeding that of the parent
member.
Provide redundancy in the structure such that alternate load distributions may be developed.
Avoid dependence on energy absorption in struts with a sharply decreasing post buckling
capacity.
Avoid pronounced weak sections and abrupt change in strength or stiffness.
Select materials with sufficient fracture toughness properties.
Select materials such that the ultimate tensile strength is higher than the yield strength to
avoid rupture to occur at first yield even if the fracture toughness is high.
Design joints and members to maintain their capacity through substantial concentrated
inelastic deformation, to meet compact section requirements.

2.2.4 Direct design


In the direct design approach the structural resistances, dimensions, etc., are determined on
basis of specific design accidental effects which are obtained from the safety analysis. It is
prudent to strike a balance between the requirements of strength on the one hand and the
need for flexibility on the other to take maximum advantage of ductility.
Guidance for direct design against the accidental loads from events listed in Section 1.4 is
given in Chapters 3 to 8.

2.3 Safety evaluations


2.3.1 General
The intention of this section is to give a brief introduction and background for the principles
and methods applied by the safety analyst.
One main objective of the safety analysis is to define specific design accidental effects to be
used In the direct design. Thus safety assessments represent the basis for the accidental
load analyses. It is therefore important that the structural engineer is familiar with such
methods.
Project safety evaluations may be divided into a number of analysis stages during the phases
of a project.
The most important safety analyses in a project will occur in the first 4 phases (feasibility
study, concept study, pre-engineering and detail engineering phases), where the influence on
safety matters is great.
Figure 2.1 /5/ gives a survey of the sequence of 10 analysis stages pointed out for a project,
their position in the project process, and the analysis type to which they belong. The analyses
are described in more detail in /5/.

In the following the methodology for concept safety evaluations (analysis no. 5 in Figure 2.1)
Is briefly discussed. This methodology may also be used in the feasibility stage as well as in
the engineering phases.
FIGURE 2.1 : SAFETY ANALYSES IN THE PROJECT /5/

2.3.2 Concept safety evaluation principles


The principles for Concept Safety Evaluations (CSEs) given in this section are based on the
requirements expressed in "Guidelines for safety evaluation of platform conceptual design"
(NPD), 1981 /1/.
These Guidelines are in the process of being revised. An interim draft in this process is
officially known as Draft 11, 14.6.84 /11/. Where there are difference in phrasing between the
two versions, the 1984 version is used, as this is believed to be closer to what the future
requirements will be.
The principles on which to base the evaluation are:
(1) The platform concept is analysed in order to identify possible accident scenarios,
taking into account:
- possible initiating events
- possible failures of safety systems
- environmental conditions
Based on this analysis, a number of possible accidental events are defined.

(2) Based on evaluation of consequences the Design Accidental Events (DAE) are
selected amongst the possible accidental events. This selection shall be based on
quantified accidental effects.
(3) For the Design Accidental Events the platform concept shall be compared to
qualitative acceptance criteria (see Section 2.3.3) in order to verify that the platform
concept has an acceptable safety level.
(4) Accidental events which violate the acceptance criteria, are called Residual
Accidental Events (RAE). According to the revised guidelines each of these RAEs
shall be evaluated carefully to determine whether risk reducing measures can be
implemented. Such risk reduction may be reduction either of accidental effect, or
frequency of occurrence, or a combination of these two. In the first case the
accidental event may be transformed into a Design Accidental Event, if impairment
can be avoided. In the case of frequency reduction, the event will still be a RAE, but a
less critical event.
(5) Maximum allowable frequency for Residual Accidental Events is stated to be
10-4/year per accident type and per safety function.
These principles are schematically shown in Figure 2.2.

FIGURE 2.2 : MODEL FOR CONCEPT SAFETY EVALUATION

Figure 2.2 /5/ presents the overall model for the Concept Safety Evaluation. Based on
concept risk analysis the accidental events may be grouped in three categories:
- The first category contains the events that are not critical for the platform concept
because of passive and active safety systems, and inherent resistance of the structure.
These events may have any, low or high, probability of occurrence, but they result in
acceptable accidental effects. These events are therefore disregarded, and are termed
Non-critical Accidental Events (NAE).
- The second category contains the events that have very low probabilities, but may
produce unacceptable consequences. It may be impossible or infeasible to design
against these consequences. The maximum allowed cummulative probability of
occurrence of the accidental events of the same group that may be disregarded is 10-4.
These events are termed Residual Accidental Events (RAE).
- The third category contains the events that may cause unacceptable effects, but which
cannot be disregarded due to the frequency cut-off limit. The events will for the different
safety functions be defined as the Design Accidental Events (DAE).
Design measures must be taken against the accidental loads of this category to maintain
the platform safety functions. Evidently, for the completed design DAEs are becoming
NAEs.
2.3.3 Acceptance criteria
The acceptance criteria as given by NPD are built on two dimensions; consequence and
frequency:
1) Consequence
In the NPD guidelines /1/ the overall acceptability criterion for Design Accidental
Events states that the personnel consequences shall be limited to personnel in the
immediate vicinity of the accident.
This criterion concerns personnel risk only. There are no corresponding criteria for
environmental pollution risk and economical risk.
The overall personnel risk is broken down into requirements for three safety
functions, called the Main Safety Functions /1/ (/11/):
(a) At least one escape-way from central positions which may be subjected to an
accident shall, during a Design Accidental Event, be usable until safe escape
can be accomplished.
(b) Shelter areas shall be usable during a calculated accidental event until safe
evacuation can be performed.
(c) Depending on the platform type, function and location, when exposed to the
Design Accidental Event, the main support structure must maintain its load-
carrying capacity until necessary emergency actions can be accomplished.
These requirements will lead to establishment of design accidental loads for the
protective walls (and roof) of shelter areas, escape ways, as well as for the main
support structure.
In the NPD regulations /2/ the functional requirements (a) to (c) above are replaced
by specific requirements to the load-bearing structure:
(d) It is to be demonstrated that the structure experiences purely local damage,
when exposed to abnormal effects.
(e) After a local damage as defined in (d), or after a defined local damage, the
structure shall resist defined environmental conditions without extensive
failure, free drifting, capsizing or sinking.
In addition it is stated in /2/ that NPD may permit a check against the accidental or
abnormal environmental loads to be disregarded provided a comprehensive
assessment demonstrates that collapse will not cause:
- loss of lives
- significant pollution
- considerable economical losses
2) Frequency
-4
The cut-off limit for allowable frequencies of disregarded events (RAEs) is 10 per
platform year, for each safety function and for each type of event (i.e. for each
hazard).

2.3.4 Safety evaluation methology


The methology given below includes the following tasks:
- Hazard Identification:
- Analysis of Accidental Events
- Assessment of Accidental Effects
- Selection of Design Accidental Effects
- Assessment of Frequency for Residual Accidental Events
- Hazard Identification:
The hazard identification should be carried out systematically to assure that all hazards
including combinations of different types are identified. The hazards could be related to
personnel, environmental pollution damage, major production stops, as well as damages to
equipment and structures.
However, the hazards of interest in the Concept Safety Evaluation are those that have the
potential to cause severe or global damages.
The hazards specifically treated in this guideline are:
- ship collisions
- falling objects
- fires
- explosions
- earthquakes
Also, when relevant, the following hazards should be identified:
- blow-out
- loss of buoyancy
- unintended ballast distribution
- abnormal environmental conditions
The hazards identified should be described in terms of possible events that could initiate
accidents. A rather wide selection of such events should be made in order not to omit any
significant hazard.
A general rule for hazard identification is that one should attempt to relate hazards as closely
as possible to different parts or systems of the Installations being assessed.

- Analysis of Accidental Events:


Spectra of Accidental Events are defined on basis of hazard identification as well as analyses
of possible accident scenarios. This analysis of Accidental Events shall primarily reflect the
possibilities of causing significant damage to personnel, environment or equipment and
structures, as well as the production and transport functions.
The accidental events may be divided into two categories:
- Events related to leakages of hydrocarbon fluids, i.e. those accidental events that may
cause explosion or fire damage.
- Events that may cause structural impact directly without fire or explosion.
With respect to structural events, two categories should always be defined:
- Events that are dependent on the activity in the area (like vessel traffic density)
- Events that depend on errors and omissions made in connection with the structure
(design, fabrication, operation of the structure)
Event trees are often used to systemize the different accident scenarios that may take place.
Such event trees should be constructed to reflect the actual platform details and systems as
closely as possible. On the other hand there will always be a need to standardize the event
trees in order to ease the computational work.
An example of an event tree is shown in Figure 2.3. In this example the initiating event is
hydrocarbon release from process system. For this event the first factor may be immediate
ignition. This defines the first branching point and, under the right hand branches, all possible
developments of the event when hydrocarbon release are immediately ignited are described.
The branches to the left describes the possible developments of releases that are not
immediately ignited. The next factor may be failure/success of the emergency shut down
system, and so forth.
The event tree is developed until all major scenarios are described. Each ending point now
represents an accident sequence described by the initiating event and the factors that have or
have not come into play down to the ending point (end event).

FIGURE 2.3 : Example on selection of Design Accidental Events (DAE) for the safety function
SUPPORT STRUCTURE from analysis of accidental events by use of event tree
and through assessment of accidental effects

- Assessment of Accidental Effects:


The assessment of Accidental Effects are carried out for those accidental events considered
in the event trees. This assessment of accidental effects shall be directed towards possible
damages to the safety functions defined in Section 2.3.3.
The accidental effects are normally expressed as physical characteristics for the different
phenomena, such as:
- impact energy
- explosion, over pressure and impulse
- heat loads, smoke effects and duration of fires
- accelerations
- static and dynamic load variations
According to the regulations, detailed calculations of accidental effects are not required at an
early project stage. An engineering evaluation on the basis of damage potential, experience
and possible historical data, is described as normally satisfactory. Verification of such
evaluations may be necessary at a later stage.
Nevertheless, it is recommended to carry out detailed calculations as early as possible,
especially for those accidental events that define the design basis for the platform.

- Selection of Design Accidental Events:


The assessment of accidental effects is the basis on which the selection of Design Accidental
Events are carried out.
In general terms a Design Accidental Event is an event which cause effects that do not impair
(or damage) any of the safety functions defined for the concept. These safety functions are
essentially the functions defined in Section 2.3.3 above, but may also include others reflecting
environmental pollution safety and economical safety.
A more precise definition of a design accidental event than given by the NPD guidelines may
be:
A Design Accidental Event is the accidental event with
maximum accidental effect relative to one lement (one failure
mode) of the main functions when disregarding all Residual
Accidental Events.

By this definition there will only be one DAE for each element of a safety function, whereas
the NPD guidelines seem to imply that all events that cause no damage to safety functions
are Design Accidental Events. When calculating the design loads, only the event giving the
highest design load with respect to the considered mode of failure will be of interest.
Therefore Design Accidental Events are limited to the events with the maximum accidental
loads.
In order to give a better understanding of the distinction between Design Accidental Events
and Residual Accidental Events a theoretical procedure for selection of Design Accidental
Events is illustrated in Figure 2.3. The procedure is described in the following 4 steps:
1) Consider for each main safety function all accidental events that may threaten this
safety function or failure mode. All accidental events that only have effects which the
safety function can survive may at this stage be omitted from consideration (see
NAEs in Figure 2.2).
2) Sort the relevant accidental events in decreasing order in terms of accidental effects.
3) Sum up the probabilities of the N first accidental events of this order, as long as the
-4
annual sum of probabilities is less than 10 .
4) Disregarding these N accidental events, the next event that applies to the considered
safety system element in the sequence of the remaining accidental events, is then the
Design Accidental Event for this element.
In this procedure, it is assumed that one main safety function may contain several physical
elements that must be treated separately because they are influenced by different accidental
situations.
In Figure 2.3 the accidental events, all associated with explosion over pressures are sorted in
decreasing order with respect to accidental effects.
-4
If events with a maximum annual frequency up to 10 can be disregarded, event no. 3 will be
the Design Accidental Event for the safety function in question. The corresponding design
accidental load will in this example be an over pressure load of 0.7 bar.
The given example is rather simplistic, and does not show all aspects which are of
importance.
-4
Note that the cut-off limit 10 is not an exact limit, as is implied in the example above. This is
further discussed below.

- Assessment of Frequencies for Residual Accidental Events:


For those accidental events for which it is considered unfeasible to prevent damage to the
safety functions an assessment of frequencies is required. The cut-off limit defined by the
authorities is directed towards the total probability of these disregarded events, but
categorized per safety function and per accident type.
It has been emphasized by the authorities that the cut-off limit is not to be regarded as a strict
limit, but should be regarded more as an order of magnitude limit. Hence, it may possibly
under certain conditions be accepted conditions which imply a total frequency higher than the
limit. On the other hand, even though an actual platform presents a sum of frequencies
slightly below the limit, there is no justification for not considering further risk reduction.
2.3.5 Use of results from Concept Safety Evaluations
The results from Concept Safety Evaluations include:
A) Specification of Design Accidental Loads for:
- fire and explosion protective walls
- main support structure
- protective shielding against falling objects
- escape way protection
- process equipment and supports
These specifications will obviously in every case at east be based on the minimum
requirements of the authority regulations.
B) Documentation which verifies that the concept satisfies the acceptance criteria
regarding main safety functions of the platform, assuming that the design is based on
the above load specifications.
C) All assumptions used in analysis and verification, which should be stated as premises
for future design of safety systems and other systems and equipment.
The results should be reported and documented, preferably in standardized formats.
It is of great importance that the safety analysis reports are made available to the structural
engineers prior to the accidental load design work and that the engineers are familiar with the
methods used and the assumptions made in the safety evaluations.
During the design work it may become necessary to refine/update the safety analyses.

2.4 Code requirements


2.4.1 General
The code requirements given below are according to the formulations in the NPD regulations
/2/. The requirements by NPD are based on the partial coefficient method (similar to the
proposed approach of the API/AISC Load and Resistance Factor Design) which may be
considered to represent level 1 of structural reliability methods.
The NPD makes use of the limit state design format. A structure, or part of a structure is
considered not to satisfy the design requirements when it exceeds a particular state, beyond
which it violates one of the criteria governing its performance or use. This state is called a
limit state, by which the target safety level is implicitly obtained by applying partial coefficients
to characteristic values of loads and material or structural resistances.
Reliability methods may be used in the safety analyses to account for the uncertainties in
loads and resistances. The design codes allow for use of such methods, if properly
documented.

2.4.2 Progressive Collapse Limit States (PLS)


The requirements with respect to accidental or abnormal environmental loads are presented
as checks for the limit states of progressive collapse.
The limit states of progressive collapse are defined in relation to the danger of extensive
failure, free drifting, capsizing or sinking of the structure when subjected to abnormal load
effects, and in relation to the damaged structure when subjected to normal loads.
The PLS check includes the identification and quantification of design loads with a given
annual probability of being exceeded. The design accidental loads are obtained through the
safety analysis while the abnormal environmental loads are obtained from wave statistics,
seismic statistics, etc.
A PLS check should be carried out in the following two steps:
First Step: The effects caused by design abnormal environmental or accidental loads on
the intact structure should be determined. All relevant failure modes
corresponding to the defined design loads should be identified and the design
loading effect should be verified not to exceed the design structural
resistance for any of the failure modes. The extent of possible damage
should be assessed.
Second Step: If the load-carrying capacity is reduced, it should be demonstrated that the
damaged structure can sustain defined environmental conditions without
extensive failure, free drifting, capsizing or sinking. The criteria and
combinations of design loads are given in the following.

2.4.3 Load combinations and coefficients


The Progressive Collapse Limit States should be checked for two combinations of design
loads. These are:
Intact Structure: Design loads including accidental loads or abnormal environmental
loads.
Damaged Structure: Design loads excluding accidental/ abnormal environmental loads.
The load coefficient is to be 1.0 for all loads.
Accidental loads are normally not combined with environmental loads when checking the
intact structure because the corresponding probability of the combined event is low. If the
actual accidental event is correlated with bad weather conditions (e.g. boat collision during a
storm), then a combination of the accidental load and the environmental loads shall be
considered.
The environmental loads do come into consideration when checking the damaged structure.
Earthquake loads need not be combined with other environmental loads.

2.4.4 Characteristic loads


The characteristic values of accidental loads should be obtained from the safety analysis.
The characteristic values of abnormal environmental loads are determined from statistics of
wave, wind, etc.
The definition of characteristic loads for the PLS check is dependent on whether they
represent:
- temporary design condition, such as loads in construction, transportation, installation,
retrieval or modification phases
- normal operation conditions
- intact or damaged condition
For temporary design conditions, the characteristic value is specified depending on
operational requirements, as given in Table 2.1. The loads should be specified so that the
safety level (or probability of failure) per time unit corresponds to the NPD acceptance level
for normal operation conditions.
Characteristic values for normal operation are summarized in Table-2.2.
It is recognized that available data will seldom permit an accurate estimate of probabilities of
accidental events. The number is rather meant as a target for an order of magnitude. An
engineering evaluation based on experiences, historical data and estimates of possible
damage potential will normally be accepted, see discussion in e.g. /8/.
Where several loads of the categories deformation (D), environmental (E) or accidental (A)
act simultaneuosly, the probabilities apply to the combination of these loads. Guidance on
combination of different environmental loads for use on the Norwegian continental shelf is
given in /3/.
TABLE 2.1 : Basic for selection of characteristic loads in PLS.
Temporary Design Conditions.

Notes : See Table 2.2.

TABLE 2.2 : Basic for selection of characteristic loads in PLS.


Normal Operating Design Conditions.
Note 1): The probability of exceedance applies to the combination of these loads
2): Expected value : Expected value (mean value, best estimate) during the period
considered
3): Specified value : Minimum or maximum value during the period considered. This value
may take into account operational requirements stated and measures taken such that
the required safety level is obtained.
4): Probability of being exceeded : The characteristic load expressed as the value with a
prescribed probability of exceedance and corresponding to the most severe load
effect.
5): Prescribed probability : For accidental loads the prescribed probability represents the
sum of annual probabilities for each safety function and for each type of event (hazard)
for the accidental events which are disregarded in the design.

2.4.5 Design resistance


The design resistance is to be determined based on a characteristic resistance divided by a
material coefficient. The characteristic resistance shall be determined statistically according to
recognized regulations for material testing and design or as a specified value. The
th th
characteristic value of strength or resistance is normally based on the 5 or the 95 percentile
of the test results, whichever is the most unfavourable.
th
The 95 percentile shall be used in such cases when high resistance is unfavourable for the
safety. This is relevant e.g. when designing relief vents (weak links) which have the function
of reducing the pressure build-up in rooms exposed to accidental explosion loads.
For the progressive collapse limit states the material coefficient is 1.0.

2.5 Commentary to code requirements


2.5.1 General
The accidental loads are in general not well defined and significant uncertainties exist both for
their magnitude and also for their resulting structural response. It is therefore necessary to a
large extent to use engineering judgement to supplement the numerical analyses. Thus it is a
need of simple procedures to accommodate an effective design work. Nevertheless it is
important to understand the theoretical ideal calculation methods in order to assess the
significance of the conservatism caused by the simplification. The code requirements is
therefore discussed in the following.

2.5.2 Determination of characteristic loads


Ideally all design accidental events (DAEs) for each hazard category should be considered in
the design process. In practical design work only the most unfavourable event for each failure
mode need to be considered and the corresponding accidental effect may then be taken as
the characteristic load. The following formulation is therefore proposed to be basis for
determination of characteristic loads:

The characteristic load is the load from that particular accidental


event among the DAEs which leads to the largest structural
consequences.

Since the accidental event leading to the largest structural consequences will vary from failure
mode to failure mode the characteristic loads will also vary. Furthermore it may be difficult to
select for each failure mode the event that will lead to the largest structural consequences.
The reason for this may be that the problem is non-linear or that the load is defined by more
than one parameter such as mass and velocity.
This is illustrated in the following example:

Example 1: Explosion
For a cantilevered blast wall the two following failure modes are considered:
Failure mode 1: Local failure of cladding
Failure mode 2: Failure in bending at support for the stiffened frames
Further it is assumed that two Design Accidental Events (DAEs) are
identified:
DAE no. 1: Deflagration of gas cloud with varying pressure over
the wall area.
DAE no. 2: Explosion in process area resulting in high localized
pressure.
In this situation two values for the characteristic loads should be defined:
1) The value to be used for check against local failure of cladding is
taken as the maximum of the varying pressure from the gas cloud
and the pressure from the explosion in the process area.
2) The value to be used for check against failure in bending at the
support is taken as an average value of the gas cloud pressure.
In practical design work it is necessary to make simplifications and it is sought to define
one value of the characteristic loads.
In cases where the consequence of the accidental loads are of importance for the cost of the
structure it is necessary to balance the desire of having a simplified design against the
possible overdesign by using a maximum value for all failure modes all over the structure. In
such cases the designer could use the safety analysis as basis for the possible establishment
of more detailed and accurate characteristic loads based on risk measures.
An example of this is shown below:
Example 2: Ship Impact
The probability of hitting an offshore platform for a specific ship type i is :
P(Hit) = N • Pi = 365 • 24 • D • vi • ρi •Pi /year
where N = no. of ships with course against platform per year
vi = ship speed (km/hr)
ρi = ship density (ship/km ) (random course/distribution)
2

Pi = probability of navigation error when on collision course


D = circumference of platform
For each ship type, an energy Ei defined by the ship speed and mass can be
found, and a histogram drawn, see Figure 2.4, and from this a cumulative
distribution maybe established. Disregarding impact angle and uneven
strength distribution, the design energy, Edesign, is given in Figure 2.5.
FIGURE 2.4 HISTOGRAM (FREQUENCY DISTRIBUTION)

FIGURE 2.5 CUMULATIVE DISTRIBUTION

With uneven strength distribution and difference impact angles, a more


detailed analysis is required. However, all RAE must be summed.
Assuming that the ultimate strength can be calculated for different parts j on
the circumference of the platform and for different impact sectors k from 0 to
90 degrees, the design criterion is :

(2.1)
where lj = length of sector j along the circumference of the platform.
θk = hit direction sector k (i.e. relative to the considered member
or platform face).
The minimum energy, Ej,k, for a specific direction and sector to be used for
the design can be found from eg. (2.1), P(Hit(Ej,k,)) is shown in Figure 2.6.
FIGURE 2.6 CUMULATIVE DISTRIBUTION

2.5.3 Safety aspects


The safety format used in codes will always be a compromise between simplicity in order to
be suitable for design work on the one hand, and the wish of giving structural design with
uniform probability of failure for different type of structures and loading on the other hand.
For the safety format given by NPD several assumptions are made. It should be kept in mind
that the overall goal for the design is to have adequate safety. This may not be fulfilled by the
simplified design format if some of the basic assumptions are not valid. The safety format by
NPD may be visualized as shown in Figure 2.7.
FIGURE 2.7 PROBABILITY OF FAILURE

It is seen from this figure that the probability of failure is not only given by the value of the
characteristic load, but also of the probability distribution of the load effect (and resistance).
The lower figure with the long "tail" on the load probability density function will obviously lead
to a larger probability of failure (which is the probability of R < S integrated over all events)
than the upper one. It may therefore be necessary to consider also accidental events with
-4
probability less than 10 in cases where the probability density function of the load effect are
abnormal. For cases where the accidental loads are of large importance and have abnormal
probability distribution it may be justified to calculate the resulting probability of failure by
means of probabilistic methods in order to overcome the errors inherent in the safety format.
The following is an outline of a generalization of the PLS check to better represent consistent
safety criteria.
The NPD design format implies a systems failure probability of

corresponding to the hazard A, where the annual probability of exceedance QA of A is taken


-4
to be 10 • P FS A is calculated by using structural reliability densities of loads and
resistances and the fact that characteristic values are applied in the design check.
In reality, PFS is determined by

and events with higher probabilities than , may contribute most to PFS. A
generalization of the NPD check may therefore be envisaged: If the target level for the
systems failure probability is PT, the structural design check, given a damage Ai may be
carried out to ensure that:

a set of relevant A i for hazard A,

where is the exceedance probability for A i.


Obviously, a refined design approach is only warranted if it consistently improves the safety
measure. Adequate data for the accidental loads are then crucial.

2.6 References
/1/ NPD:
"Guidelines for safety evaluation of platform conceptual design", Norwegian
Petroleum Directorate, 1981.
/2/ NPD:
"Regulation for structural design of load-bearing structures intended for exploitation of
petroleum resources", Norwegian Petroleum Directorate, 1984.
/3/ NPD:
"Guidelines for the determination of loads and load effects", Norwegian Petroleum
Directorate, 1987.
/4/ Det norske Veritas TNA 101:
"Design Against Accidental Loads", 1981. (New issue. RP D 204)
/5/ J.E. Vinnem, B. Hope:
"Offshore Safety Management", TAPIR, Trondheim, 1987.
/6/ Andersen et al:
"Risk Analysis in Offshore Development Projects", SINTEF Report STF18 A83503,
1983.
/7/ T. Moan:
"On Hazard-Tolerance Criteria for Marine Structures", Proc. ICASP5, Vancouver,
1987.
/8/ T. Moan:
"Loads and Safety for Marine Structures, Accidental Loads", Seminar, Norwegian
Society of Chartered Engineers, NTH, January 1985.
/9/ S. Fjeld:
"Offshore Oil Production and Drilling Platforms. Design Against Accidental Loads",
BOSS'79, London, 1979.
/10/ B. Hope, P.A. Johannesen:
"Safety Management In offshore Development Projects", Tanum-Norli, Oslo, 1983.
/11/ NPD:
"Regulations for safety evaluations", Draft 11, 14.6.84.
3. DESIGN GUIDANCE FOR STRUCTURES EXPOSED TO SHIP COLLISIONS

3.1 Introduction
3.1.1 General
The objective of this chapter is to give guidelines for the direct design of offshore structures
exposed to ship collision.
This is generally a very complex problem due to pronounced inelastic material and nonlinear
geometric behaviour, dynamic effects (inertia, strain rate) and finite material ductility. At the
present state of the art it is virtually impossible to perform rigorous analyses. Hence, it is
necessary to apply simplified techniques.
A significant part of the collision energy is dissipated as strain energy. Except for global
deformation modes the contribution from elastic straining can normally be neglected. In such
cases rigid-plastic methods of analysis have proven to be useful. Most of the guidance given
below is based on this approach. The methods recommended for calculation of energy
dissipation are often rather crude as compared to conventional elastic design. However, in
view of the large uncertainties associated with characterization of the collision load itself, the
level of accuracy is considered to be adequate.
A review of methods for collision damage assessment can be found for example In Refs /1,
2/. In Ref /3/ historical data are also investigated.

3.1.2 Definitions
Characteristic buckling strength: buckling capacity with a defined probability of not
being smaller
Load coefficient: prescribed coefficient to be multiplied with the
characteristic load in order to obtain the design load
Material coefficient: coefficient by which the characteristic material
strength is divided to obtain the design material
strength

3.1.3 Notations
B Ship breath
C Coefficient
CB Ship block coefficient
D Ship draught
E Elastic modulus, energy
EP Strain energy platform
ES Strain energy ship
HS Significant wave height
J Mass moment of inertia
L Element length
M Bending moment
MP Plastic bending moment
MU Ultimate bending moment capacity
N Axial force
NP Plastic axial force
NU Ultimate axial force capacity
P Lateral load, collision load
PO Reference collapse load
PP Collision force platform
PS Collision force ship
Q Parameter
d Tube diameter
do Leg diameter
fE Euler stress
fy Yield stress
g Acceleration of gravity
h Effective height of contact area
k Spring stiffness
lh Length of plastic hinge
m Mass
ma Added mass
ms Mass of ship including added mass
mP Mass of platform including added mass
t Tube thickness, time
tc Duration of collision
ti Thickness of insert pile
t1 Leg thickness
v Velocity
vP Platform velocity
vs Ship velocity
z Distance from effective pivot point to point of collision
∇ Ship displacement
α Load position
δ Deformation
δd Local deformation of cross-section
δe End shortening
δ max Maximum beam displacement at the point of collision
δp Platform deformation
δs Ship deformation
δ tot Total deformation at the point of collision
ε max Maximum (fracture) strain

 fy 
λ Reduced slenderness parameter  
 fE 
 
θ Angle
η Non-dimensional plastic hinge length
µ Reduction factor for plastic moment capacity due to local damage
ϒ Reduction factor for plastic moment capacity due to local buckling

3.2 Design strategy


3.2.1 General
An acceptable safety against ship collision can be obtained by different means (see Section
2.2)
- event control
- indirect design
- direct design

3.2.2 Event control


The probability of collision is affected by several factors such as
- operational procedures/limitations
- field layout
- reliability of ship system (machinery, manoeuvering system)
- quality of mooring system
- navigational aids
- traffic regulations
- traffic survey
- stand by boats
- location with respect shipping lanes

3.2.3 Indirect design


Indirect design principles are discussed in Section 2.2.3.

3.2.4 Direct design


The principles for direct design against ship collision can be summarized as follows:
- Define collision situations and collision energy (mass and speed) (Section 3.3 and 3.4)
- Determine the fraction of collision energy that remains as kinetic energy after impact. The
rest is to be absorbed as strain energy in vessel and platform. (Section 3.5)
- Assume a deformation mechanism and calculate the energy absorbed by elastic and
plastic deformation of the actual platform element(s) and the vessel by use of load-
deformation curves and formulas established in Sections 3.6, 3.7, 3.8 and 3.9.
- Check that the element joint capacity is sufficient to transfer the impact forces. (Section
3.7.6 and Figure 3.3)
- Check that the surrounding elements have sufficient capacity to sustain the collision load
effects and the permanent platform loads. Otherwise, modify the deformation mechanism
and repeat calculations of the energy dissipation.
- Check that global integrity is not impaired during collision (Section 3.10)
- Check if the damaged platform can resist functional loads and environmental loads with a
specified return period (see Section 8)
3.3 Collision scenarios
3.3.1 Categorization
Collision events may be categorized according to the type of marine traffic as follows:
- Authorized vessels servicing the installation
- Tankers for offshore loading
- By-passing ships and fishing vessels
- Floating accommodation- or service platforms
- Drifting objects
The nature of causes as well as consequences may be quite different for the various groups.
Another way of classification is according to the total amount of collision energy involved.
a) Low energy collision (≈1MJ): Caused by a smaller vessel (e.g. supply vessel) at a
speed related to normal operations of vessel approaching or leaving the structure
-4
(frequency > 10 per year).
b) Accidental collision: often assumed caused by a vessel (e.g. supply vessel) drifting
out of control in the worst sea state in which it may operate close to the platform
-4
(frequency = 10 per year).
c) Catastrophic collision: Caused by a ship or other type of objects in which the mass or
the impact velocity or a combination or both results in a magnitude of kinetic energy
-4
which causes total platform collapse (frequency < 10 per year).
Traditionally, supply vessels and other dedicated vessels operating at moderate speed have
been assumed to represent the highest probability of low energy and accidental collisions.
However, recent Investigations indicate that high energy impacts with by-passing vessels and
supply vessels may contribute significantly to the collision risk.
This section deals with accidental collisions (case b above).
3.3.2 Impact geometry
Impact situations should be chosen on the basis of a thorough consideration of the most
unfavourable events together with their probability of occurrence.
The type of contact is a function of location of the elements to be analysed as indicated in
Table 3.1.
Accounting for the waves and draught of the vessel, the impact should be assumed to occur
any place between levels 10 m below LAT and 13 m above HAT [1] for fixed platforms and 8
m below LAT and 11 m above HAT for floating platforms. Impact from vertical motion of ship
in waves should be considered for horizontal elements positioned between these levels.
TABLE 3.1 POSSIBLE COLLISION SITUATIONS

3.4 Characteristic collision loads


3.4.1 General
The work associated with determining appropriate design accidental collision loads may
easily become very comprehensive due to the large number of possible accident scenarios,
involving
- many types and sizes of striking vessel and
- different parts of the striking vessels that come in contact
Bearing in mind the immense complexity of the problem, the insufficiency of available data
and the intrinsic dependence of ship/platform collisions on human factors, some
simplifications are usually introduced into the analysis. In the following guiding figures and
methods most commonly applied are presented. Further background material may be found
in the references given.
The collision load is governed by the impact energy which depends upon:
- size of vessel
- added mass
- collision velocities
-4
The design collision energy should, at best estimate, not exceed a probability level of 10 per
year. The conditional probability of the location of contact area during collision should also be
considered (Section 3.4.5).

3.4.2 Vessel size


The sizes of supply vessels classed by Veritas are presented in Figure 3.1, /4/. However, a
trend towards an increased size. of supply vessels is recognized. This trend should be
carefully observed and steps taken to adjust the statistics when necessary.
FIGURE 3.1 SUPPLY VESSELS CLASSED WITH VERITAS 1982

The size of tankers and other by-passing vessels may vary considerably from the size of a
supply vessel and up to 100 000 – 150 000 tons. Table 3.2 summarizes typical sizes in a
collision study /5/.

TABLE 3.2 SHIP CHARACTERISTICS

* Tug/Supply vessel

3.4.3 Added mass


The total mass of a vessel includes both the mass of the vessel and the added mass.
The added mass depends on many factors such as for instance: submerged shape and
draught of the vessel and duration of impact. The following simplied expression for added
mass in sway motion may be used /6,7/.
where
ma = added mass
∇ = displacement of the ship
g = acceleration of gravity
tc = duration of collision
D = draught of ship
B = breadth of ship
CB = block coefficient
Added mass for sideway motion for typical supply vessel geometrics is presented in Figure
3.2 /6, 7/.

Assuming relatively short duration of collision (i.e tc g / D ) < 1) the added mass may be
determined by

FIGURE 3.2 SIDEWAY ADDED MASS FOR SUPPLY VESSELS AS A FUNCTION OF


DURATION OF COLLISION, tC , AND VESSEL BREADTH TO DRAUGHT RATIO,
B/D
3.4.4 Ship collision velocity
Frequently, the collision velocity used in the design is related to a drifting supply vessel. For
this case the following relationship is quoted /8/
vs = 0.5 • Hs (max) [m/s]
where
vs = impact velocity [m/s]
Hs (max) = maximum significant wave height for operation close to the platform [m]
Often a design collision of 2.0 m/s is used, which implies Hs (max) = 4 m.
Characteristic impact velocities for drifting and manoeuvring attendant vessels have been
estimated in Ref /9/.
According to a recent investigation /10/, frequencies of collisions with powered visiting vessels
as well as with passing vessels are significant if the platform is located in the vicinity of a
typical shipping lane. In such cases the collision velocity may be assumed equal to the
operating speed of the vessel.

3.4.5 Impact geometry


Collision frequencies refer often to the probability of all possible contact situations, based on
the effective diameter of the ship and the platform. In such cases the conditional probability of
collision for a specific contact location (leg/bracing) should be considered.

3.5 Collision mechanics


3.5.1 Fundamental principles
Analysis of collision mechanics is generally to be based upon the solution of the differential
equations of the dynamic equilibrium. The collision force is a function of the relative
indentation of the ship and the platform. Thus, an incremental solution procedure is required.
The problem is greatly simplified if the collision duration is considerably smaller than the
natural period of the governing motion. This assumptions is often valid for relevant rigid body
notions of floating and articulated platforms. Then the solution can be based upon a
quasistatic solution using the principles of
- conservation of momentum
- conservation of energy
In this way the determination of impact kinematics and energy transfer during collision can be
decoupled from the analysis of strain energy dissipation in the colliding objects (collision
effects).
For collision durations significantly longer than the natural period of the governing motion, a
static solution applies.
For jackets at medium water depths (North Sea), the ratio between the collision duration and
the natural period of vibration for leg impacts may be such that significant dynamic effects are
involved /26/. This has been investigated to a very little extent. Normally static analysis is
considered appropriate, but possible dynamic magnification should be evaluated.

3.5.2 Conservation of momentum


In the following, the energy to be dissipated as strain energy is determined considering
translatory motions only. Better accuracy may be obtained by taking into consideration more
motion components (platform and vessel rotations), however, at the expense of more
complex derivations /10/. It is always conservative to use the formulas given in Section 3.5.3.
FIGURE 3.3 Collision between supply vessel and semi-submersible platforms

The conservation of momentum for a central collision between a ship and platform moving in
the same direction (see Figure 3.3) is expressed by
msvs + mpvp = (ms + mp) vc
where
vc = common velocity after impact
vs = velocity of ship vs > vp
vp = wave induced velocity of platform
ms = mass of ship including added mass
mp = mass of platform including added mass
The common velocity is thus defined by:
m s v s + mp v p
vc =
m s + mp

3.5.3 Conservation of energy


In a central collision all kinetic energy dissipates as strain energy (elastic or plastic) in ship
(Es) and platform (Ep).
In case of an eccentric collision some of the kinetic energy will remain as rotational energy of
the vessel or the platform after collision.
The equation of energy conservation assuming central collision reads
1 1 1
ms v 2s + mp v p2 = (m s + mp ) v c2 + E s + Ep
2 2 2
where
Es = strain energy dissipated by the ship
Ep = strain energy dissipated by the platform
Combining this equation with the equation for the common velocity vc the following expression
for the strain energy dissipation emerges
vp
(1 − )2
1 vs
E s + Ep = m s v 2s
2 m
1+ s
mp
Often the wave induced platform motion is small compared to the ship velocity and can be
neglected. Otherwise, the characteristic velocities should be based upon a stochastic
evaluation of the relative collision velocity between the ship and the platform (vs -vp)
accounting for the phase lag between the respective motions, Ref /11/.
For design purposes the expressions in Table 3.3 for the total strain energy dissipation may
be used.

TABLE 3.3 TOTAL STRAIN ENERGY DISSIPATION

* Provided that duration of collision significantly smaller than fundamental period of vibration
where
J = mass moment of inertia of the column (including added mass) with respect to
the effective pivot point
z = distance from the effective pivot point to the point of contact. See Figure 3.4.

FIGURE 3.4 IDEALIZATION OF COLLISION BETWEEN A SUPPLY VESSEL AND AN


ARTICULATED COLUMN

3.5.4 Dissipation of strain energy


The dissipation of strain energy is a matter of structural analysis where the force
displacement curves for the hit and striking object must be established. For a given load level
the area under each curve represents the absorption of energy. The distribution and extent of
damage result from the condition of equal collision force acting on the structures and that the
sum of absorbed energies equals the portion of kinetic energy dissipated as strain energy.
where
Ps, Pp = force deformation relationship for ship and platform, respectively
δs, δp = deformation of ship and platform, respectively
The procedure is illustrated in Figure 3.5

FIGURE 3.5 PROCEDURE FOR CALCULATING STRAIN ENERGY DISSIPATION

3.6 Energy dissipation in vessel


3.6.1 General
Analytical tools for assessing energy dissipation in ships are relatively few. Some methods for
determination of the crushing resistance of bows are based upon discretization of the cross-
section into basic elements, for which the deformation characteristics are known, and then
sum up the contribution from each basic element, Refs /12-15/. The empirically based
Minorski's method /16/ has also been used to some extent. Due to the large uncertainty in the
low-energy range of interest, this approach is not recommended.
The methods described above may be considered relevant for collisions against large
diameter columns, (D > 10 m). For collisions against bracings the bow may normally be
considered rigid.
Simplified calculation models for broadside collision can be found in Refs /17-19/. A modified
non-linear FEM approach is used in Ref /20/ and Ref /21/, the latter being the more recent
and up-to-date approach.
In lieu of more accurate methods the load indentation relationships specified in the following
sections may be used.

3.6.2 Supply vessels


Load-deformation curves for bow, stern and broadside collision with an infinitely stiff element
are given in Figure 3.6 /4/.
To determine the contact area the ship geometries given in Figure 3.7 /4/ may be used. The
theoretical areas should be reduced by multiplying with a factor of 0.4 for broadside collision
and 0.7 for bow and stern collision to account for uneven load distribution .
FIGURE 3.6 LOAD-DEFORMATION CURVES FOR SUPPLY VESSELS
FIGURE 3.7 TYPICAL 5000 t DISPLACEMENT SUPPLY VESSEL GEOMETRIES
3.6.3 Tankers and other ships
In lieu of more accurate data the following load-deformation relations for broadside collision
may be used /5/
-3 -3
P = (0.18 • 10 ∇ + 26) δ + 0.12 • 10 ∇ + 18
where
P = collision force [MN]
δ = deformation [m]
∇ = ships displacement [tons]
In lieu of more accurate information the “effective” height of the ship side (contact area) may
be determined by
-3
h = 0.03 • 10 ∇ + 5
where
h = effective contact height [m] of the ship during collision

3.6.4 Barges
In lieu of more accurate information, the following load-deformation relations for broadside
collision may be used /5/
P = 6 • δ + 4
where
P = collision force [MN]
δ = deformation [m]
The above formula is based on data for a barge with a height of 6.0 m and should not be
applied to barges with significantly different height.

3.7 Design methods for unstiffened tubulars


3.7.1 General
The collision response of tubular members of space frame type platforms can conveniently be
split into the following deformation components:
- Local deformation of tube wall at the point of impact.
- Beam deformation of bracing element.
- Global deformation of platform.
The first two modes involve considerable plastic energy absorption while the global platform
response is mainly elastic, possibly with some dynamic effects involved. Often the
contribution to energy absorption from global deformation is neglected.
The transition between the different deformation modes is difficult to define in practice. The
response of a tube to lateral load can be described as follows, (Figure 3.8 and 3.9).
- In the initial stages of deformation the response is governed by bending effects which are
affected by and interacts with local denting under the load. In this stage the bending
capacity will also be reduced if local buckling on the tube wall occurs on the compression
side towards the ends. As the beam undergoes finite deflections the load carrying
capacity may increase considerably due to the development of membrane tension forces.
This depends on the ability of the adjacent structure to keep the tubular joints fixed in
space. Provided that the adjacent structure does not fail, the energy absorption capacity
is restricted either by excessive straining of the tube or joint failure.
It is very difficult to account for all the behavioral effects simultaneously by analytical
means. A useful approach is to study local deformation and beam deformation separately
and to construct the interaction curve on the basis of this information, as illustrated in
Figures 3.8 and 3.9

FIGURE 3.8 LOAD-DISPLACEMENT CHARACTERISTICS FOR BEAM WITH AXIALLY FREE


ENDS
FIGURE 3.9 LOAD-DISPLACEMENT CHARACTERISTICS FOR BEAM WITH AXIALLY
RESTRAINED ENDS
3.7.2 Beam Deformation
The beam deformation may be determined by assumption of a three hinge mechanism. With
no axial restraint the plastic collapse load in bending is given by
Pu = Po • Cu
where

2 Mp
Po =
α(1 − α ) L
= plastic collapse load in bending for a beam with rotationally fixed ends
2
Mp = fyd t
= plastic moment capacity
αL = distance from element support to point of impact
L = element length
fy = yield stress
t = tube thickness
d = tube diameter
δ = lateral deformation
Cu is defined in Figure 3.10 and is a factor which accounts for the detrimental effect of local
damage (Section 3.7.3). At the beam ends the smaller of the plastic bending/torsional
capacity of the adjacent beams in the node shall be used.
In case of axially restrained ends the load carrying capacity increases considerably beyond
the plastic collapse load in bending due to development of membrane forces. For the pure
membrane phase the lateral force is given by
π δ δ
Pu = P0 • • for < 0.15
2 d L
FIGURE 3.10 FACTOR FOR PLASTIC COLLAPSE LOAD IN BENDING

M
µ = Reduced moment capacity due to local damage at load application
Mp
Ref. 3.7.4
M
γ = Reduced moment capacity due to local buckling at beam ends Ref. 3.7.5
Mp
For legs with inset pile the effective thickness can be taken as

t = t 2l + t 12
where t l = thickness of leg

t 1 = thickness of insert pile


In case of rotationally free ends (type 1) or rotationally fixed ends (type 4) the transition from
pure bending to pure membrane behaviour is given by

 cδ 
2
cδ 
  cδ  cδ
Pu = Po C u 1 −   + ar sin   , ≤1
  d d  d  d
 
where

C = 2
for rotational ly free ends
 for rotationna lly fixed ends
1
For different boundary conditions the interpolation may be used as shown in Figure 3.11a.
From Figure 3.11a-b it appears that local damage only affect the bending contribution to the
load bearing capacity. For intermediate and small d/t ratios it is often appropriate to disregard
possible local buckling in case of axially fixed ends.
A simple approach is to use the membrane solution over the entire deformation range
disregarding the effects of bending, local damage and local buckling (Figure 3.9b). This will
normally be conservative.
A necessary condition for assuming the pure membrane phase to develop is that the adjacent
members can support the fully plastic tension force, Np, magnified by a strain hardening factor
N = 1.3Np = 1.3 π fy dt
The most infavourable brace deformation state shall be considered.
Analysis models for members with partial end restraint are described in Refs /23, 24/. The
axial restraint can be evaluated by means of a finite element analysis as follows:
- remove the element and substitute its load effects due to functional loads by concentrated
loads at the support.
- Simulate the membrane force by applying unit forces at the support
- Evaluate the spring stiffness, k, with respect to the membrane force.

If the spring stiffness k satisfies the condition

πLtf y
k>
d
full axial resistance can be assumed.
If

πLtfy
k < 0.05
d
free ends should be assumed.
For intermediate values of k the membrane phase is delayed as compared with the fully
restrained case. The membrane effect may still be taken into account provided that the
member can undergo large lateral displacements without initiation of fracture (which possibly
is accelerated by local buckling in the bending phase).
FIGURE 3.11 PLASTIC LOAD-DISPLACEMENT RELATIONSHIP
3.7.3 Local damage
Local damage of the tube wall at the point of impact for an element depends on the nature of
impact. A head-on collision gives more concentrated force than a broadside impact and
results in larger local energy absorption for a given mass and velocity of the vessel.
A comparison of different calculation models for local damage is given in Ref /25/.
The non-dimensional load-deformation curves presented in Figure 3.12 /26, 27/, give fairly
good agreement with experimental results for δd/D < 0.5.
For legs with insert piles the equivalent tube thickness can be taken as /37/

t = t 12 + t i2
where
t1 = thickness of leg
t i = thickness of insert pile

FIGURE 3.12 LOAD-DEFORMATION CURVES FOR LOCAL DENTING


3.7.4 Bending capacity of dented cross-section
The bending capacity at the point of impact is reduced due to local damage. The model
presented in Ref /27/ may be applied. Conservatively, the dented section may be assumed
noneffective, (see Figure 3.13). This yields:
θ 1
Mu = Mp (cos − sin θ)
2 2
where
2
Mp = fyd t
θ = arccos (1-2δd/d)
δd = depth of dent as defined in Figure 3.13

FIGURE 3.13 PLASTIC BENDING CAPACITY OF DENTED CROSS-SECTION

3.7.5 Local buckling


The ultimate moment capacity of the cross-section at the beam ends is reduced once local
buckling is initated. This should be considered for beams with little axial restraint.
A fully plastic bending moment may be assumed initially, provided that the reduced cross-
sectional slenderness parameter, λ, satisfies the condition

where
For higher slenderness the maximum moment should be taken as follows /28/

where

The deformation limit at which local bucling is initiated can be estimated by

where

The check should be carried out for both ends. The smaller value of δ/d is the governing one.
For compression members the corresponding end shortening, δe, can be taken as

where δ/d is defined above.


The moment capacity is still not exhausted after local buckling. However, very little
information is available on the behaviour in the post-buckling range.

3.7.6 Capacity of joints


The tubular joints have to provide sufficient capacity to transfer the element reaction forces.
Formulas for tubular joint capacity may be found in various design codes.
Axial force capacities for different type of tubular joints are given in Figure 3.14 /29/, using the
notations defined in Figure 3.15.
The bending moment capacity of tubular joints may be determined as follows /30/ :
where

The notations are defined in Figure 3.15


For tubular joints the combined effect of axial force and bending moment may be defined by
the following interaction formula /31/
N M
+ ( )1.2 ≤ 1
Nu Mu
where
Nu = Ultimate axial force capacity given in Figure 3.14
Mu = Ultimate bending moment capacity defined above
During development of membrane tension forces in the tube-element the combined effect of
the membrane tension force and the bending moment have to be considered. The
combination of axial force and bending moment for a tubular cross-section is defined by :

M  πN 
= cos  
Mp  2N 
 p 
where
N, F = actual element force and moment
2
Mp = fyd t
Np = πfydt
FIGURE 3.14 CAPACITY OF TUBULAR JOINTS
FIGURE 3.15 NOTATIONS FOR TUBULAR JOINTS

A practical check of the tubular joint capacity is to verify that the entire axial force-bending moment
interaction curve lies outside the interaction curve for the tubular element, as illustrated in Figure 3.16.
FIGURE 3.16 INTERACTION CURVES FOR TUBULAR JOINTS AND TUBULAR MEMBERS

3.7.7 Critical strains


Due to combined bending and axial load large strains will be developed in the member.
Hence, if the capacity of the joints as such satisfies the criterion in Section 3.7.6, tension
rupture of the tube wall governs the maximum energy absorption capability of the member.
This effect should be carefully considered and documented.
Is it generally not possible to assess the total strains accurately by analytic means.
Furthermore the critical strain of the material is highly uncertain. On the basis of a fracture
mechanics approach evaluations are performed for ship plating in Ref /20/.
In lieu of more accurate information the maximum deformation in the beam mode should not
exceed
for axially fixed ends, and

for axially free ends.

εmax = Ultimate strain in wide plate tests. In lieu of more accurate information εmax =
0.15 can be used.
The effect of abnormal cracks is not included in the above expressions. The maximum
deformations quoted for beams with axially free ends are valid until initiation of local buckling
at the beam ends (Section 3.7.5). If calculations are carried out beyond this deformation level
a model for the nominal strain contribution in the post-buckling range must be included.

3.7.8 Mechanisms involving multiple members


The methods described in Section 3.7.1 - 3.7.7 apply directly to energy absorption in single
members subjected to lateral collision load. However, depending on the structural layout
several members, including compression members, may be subjected to plastic deformations.
Alternative deformation modes are suggested for two example cases in Figure 3.17.
A useful approach is to assume that the deformation mode with the lower limit collapse load in
bending is activated initially, and then check whether a potential combined mechanism is
developed if the load bearing capacity increases due to membrane forces.
The capacity of compression members should be taken as the characteristic buckling strength
(i.e. partial safety factors = 1).
Plastic analysis as described above may become cumbersome and of accuracy limited.
Alternatively, nonlinear finite element analysis may be performed, Ref /32/. However, the
importance of local damage, local buckling and finite ductility should be evaluated and the
numerical results adjusted if these effects are not included in the finite element analysis.
FIGURE 3.17 ALTERNATIVE MECHANISM MODELS
3.7.9 Compression members
Plastic methods are also applicable to analysis of energy absorption in compression
members. The approximate plastic axial force-end shortening relation ship (Figure 3.18) can
be determined by the equations

where
δs = end shorthening

FIGURE 3.18 LOAD-END SHORTENING CURVE FOR COMPRESSION MEMBER

It is often appropriate to assume M = Mp.


The elastic load-end shortening curve is given by
where
δ o = initial lateral displacement
L e = effective buckling length

NE = Euler buckling load


Thin-walled members are susceptible to local buckling, which may reduce the energy
absorption capability significantly. The initiation of local buckling should be checked
according to the procedure described in Section 3.7.4.

3.8 Design methods for stiffened cylindrical shell (columns)


3.8.1 General
Columns of typical semisubmersibles with diameters in the range of 5-10 m are strengthened
by vertical stringers and horizontal ring stiffeners. To minimize the flooding the columns are
partitioned by watertight compartments. In the waterline area an inner shell is also often
provided.
Extensive tests and theoretical studies have been conducted on the response of stiffened
cylindrical shells subjected to lateral loading, see for example Refs /33, 34, 35/. However, so
far this work has not resulted in simple design formulas.
On the basis of experimental observations, it is concluded that the energy is dissipated mainly
by membrane stretching of shell plating and longitudinal stiffeners (if present) and radial
deformation of ring frames. Inspired by these observations, rigid-plastic mechanism models
have been developed. The major factor of uncertainty is related to the range frame behaviour
in the large deflection range.
It should be observed that such models can be very unconservative if fracture occurs,
especially in the shell plating. This is demonstrated in Ref /10/. Thus, the calculations should
be accompanied by an assessment of critical strains, so that the membrane contributions is
reduced once fracture is initiated. Conversely, if the column is provided with decks and
bulkheads in the collision zone, the strength may be significantly under predicted by these
models /35/.
The nonlinear FEM of Refs /20, 21/ will provide more accurate evaluation for plated structures
in general. It has been successfully used in several occasions on ship-column /38/ and ice-
column collision /39/.
On the basis of a parametric study, using the approach described In Refs /33, 34/, it is
concluded that an energy dissipation of 8-9 MJ in typical columns will produce indentations in
the order of 0.7-1.1 m /3/. This is in agreement with the predictions of Ref /10/.

3.8.2 Resistance to large penetrations


For major collisions the load-displacement relationships presented in Figure 3.19 may be
used. They are based upon the model described in Refs /10/.
FIGURE 3.19 LATERAL LOAD-INDENTATION RELATIONSHIP FOR STIFFENED COLUMN

3.9 Design methods for pontoons


The structural configuration of pontoons resembles those of a ship hull. This suggests that
load deformation relations for the pontoon may be based on the load-deformation formulas
proposed for tankers or barges in Section 3.6.3 or 3.6.4 or analyzed by methods discussed in
Section 3.6.1.

3.10 Global integrity during impact


3.10.1 General
Local damage is accepted provided that a progressive collapse mechanism does not occur. In
the previous sections, local damage and impact force due to ship collision are estimated,
based on the energy dissipation in the ship and a single or a few platform elements.
This section gives guidelines and procedures which may be used to check that a global
progressive collapse mechanism is not developed during impact,
It is noted that if the platform complies with the residual strength requirement (Section 8) is
often unlikely that it will turn into a global collapse mechanisme under the direct collision load.
(For example: the collision load for a brace impact is an order of magnitude smaller than the
total design wave load for conventional jackets.)
The influence of dynamic effects has been studied to a very limited extent. According to Ref
/26/ the dynamic displacement made may differ significantly from the static mode for a jacket
at medium water dept.
However, most often the global strength check is carried out by means of static methods to
which the guidance given below applies. Linear analyses is appropriate provided that no yield
- or buckling criterion is violated outside the collision zone. Otherwise, nonlinear finite element
analysis should be performed to verify global integrity (see Section 8).

3.10.2 Partial safety coefficient


The load and material coefficients to be used for the limit state of progressive collapse are
given in Section 2.4.

3.10.3 Global integrity check


The global integrity check is often carried out for the maximum collision load. if this does not
coincide with the maximum damage state, possible unfavourable combinations of collision
loads and damage states should be considered.
Elements undergoing membrane deformations (for example braces) should be assumed to
have no strength. The collision load can then be represented by the corresponding load
effects at member ends as indicated in Figure 3.20a.
Elements acting predominantly in bending may be assumed to be partly effective. The effect
of lateral deformation may be modelled by subdivision of the original element into two new
elements with a node offset corresponding to the permanent beam displacement as illustrated
in Figure 3.20b.
FIGURE 3.20 IDEALIZATION OF DAMAGE STATE AND COLLISION LOAD IN THE GLOBAL
INTEGRITY CHECK
3.11 References

/1/ Ellinas, C.P. and Valsgard, S.:


th
"Collision and Damage of Offshore Structures. A State-of-the-Art." 4 Int. Symp. on
Offshore Mechanics and Arctic Engineering, OMAE-ASME, Dallas, Texas, February,
1985.
/2/ Amdahl, J.:
"Offshore Ship Collision and Dropped Objects", NIF-course "Last og sikkerhet for
marine konstruksjoner", Trondheim, January, 1985.
/3/ "Study on Offshore Installations Protection Against Impact - Background Report": J.P.
Kenny, June, 1987.
/4/ DnV TN 202. 1981-05-01
/5/ DNV impact and Collisions. Offshore Progress Report No 8, 1979-10-20
/6/ Pedersen, M.J.:Dynamics of Ship Collision. The Technical University of Denmark,
Report no 185, July 1980.
/7/ Petersen, J.P. and Pedersen, M.S.:Collisions between Ships and Offshore Platforms,
OTC 4134, Houston 1981.
/8/ Rules for Classification of Mobile offshore Units, Det norske Veritas, Hovik, March,
1981.
/9/ Collision of Attendant vessels with offshore Installations, Part 1 and 2, National
Maritime Institute report no 84208 and 84209.
/10/ Risk Assessment of Buoyancy Loss - Assessment of MODU Collision Frequencies,
Draft Report Technics, London, 1987.
/11/ Amdahl, J.:"Probabilities of Occurrence and Impact Velocities for Impacts in the North
Sea", Veritas Report No. 80-0687, Hovik, 1980.
/12/ Gerard, G.:"Compressive Strength of Flat Stiffened Panels", NACA, TN3785, 1957.
/13/ Minorsky, V.U.:
"Bow Loading Values (Bulbous Bow)" U.S. Mar. Adm., Dept. of Commerce,
Washington, Report MA-RD-920-78035, 1977.
/14/ Wierzbicki, T. et al.:
"Crushing analysis of Ship Structures with Particular Reference to Bow Collisions",
Det norske Veritas, Progress Report No. 16 on Impacts and Collisions Offshore,
Report No. 82-079, 1982.
/15/ Amdahl, J.:
"Energy Absorption in Ship-Platform Impacts", Norwegian Institute of Technology,
Division of Marine Structures, Report No. UR-83-34, Trondheim, 1983.
/16/ Minorsky, V.U.:
"An Analysis of Ship Collisions with Reference to Protection of Nuclear Power
Plants", Journal of Ship Research, pp 1-4, Oct. 1959.
/17/ "Tanker Structural Analysis for Minor Collisions",
M Rosenblatt & Son. Inc., N.Y. Dep. of Transp. United States Coast Guard,
Washington 1975.
/18/ Hysing, T.:
"Analysis of Penetration of Hull", Progress Report No. 4 on Impacts and Collisions
Offshore", Veritas Report No 78-433, Hovik 1978.
/19/ Low, H.Y. and Morley, C.T.:
"Indentation Tests of Simplified Models of Ship Structures", Proc. IABSE Colloquium
on Ship Collision with Bridges and Offshore Structures, Preliminary Report,
Copenhagen, pp 213-220, 1983.
/20/ Pettersen, E. and Valsgard, S.:
"Collision Resistance of Marine Structures", In Structural Crashworthiness, Edited by
N Jones and T Wierzbicki, Butterworths, London, pp 338-370, 1983.
/21/ Valsgard, S. and Jorgensen, L.:
"Evaluation of Ship/Ship Collision Damage Using a Nonlinear Finite Element
nd
Procedure", 2 Int. Symp. on Practical Design in Shipbuilding, Tokyo & Seoul, Oct.
17-22, 1983 (Also as Veritec paper series no. 001, Jan. 1985).
/22/

/23/ Oliveira, J.G. de:


"Simple Methods of Estimating the Energy Absorption Capability of Steel Tubular
Members Used in Offshore Structures", The Norwegian Institute of Technology,
Division of Marine Structures, Trondheim, Report No SK/R50, 1979.
/24/ Oliveira, J. G. de:
"The Behaviour of Steel Offshore Structures Under Accidental Collisions, OTC 4136,
1981.
/25/ "Boat Impact Study",
LLoyds Register of Shipping, Offshore Technology Report, OTH 85224, 1985.
/26/ Amdahl, J.:
Impact Capacity of Steel Platforms and Tests on Large Deformations on Tubes Under
Transverse Loading. DnV Progress Report No. 10 on Impacts and Collisions
Offshore, Report No. 80-0036, 1980.
/27/ Ellinas, C.P. and Walker, A.C.:
"Damage of Offshore Tubulars Bracing Members", proc. IABSE, Colloquium on Ship
Collision with Bridges and Offshore Structures, Preliminary Report, Copenhagen, pp.
253-261. 1983.
/28/ Sherman, DR.:
"Inelastic Flexural Buckling of Cylinders". Steel Structures, Recent Research
Advances and Their Applications to Design, Elsevier Appl. Science Publ., 1986.
/29/ Wardenier, J. and Starke, J.V.B.:
"The Static Strength of Welded Lattice Girder Joints in Structural Hollow Sections",
Stevin report 6-78-4, March 1978.
/30/ Yura, J.A. et al:
Ultimate Capacity Equations for Tubular Joints, OTC 3690, 1980.
/31/ Hoadly, P.W., Clarkson, U. and Yura, J.A.:
Ultimate Strength of Tubular Joints Subjected to Combined Loads. OTC 4854, 1985.
/32/ Moan, T. and Taby, J.:
"Strength Assessment of Damaged Tubular Offshore Structures", Brazil Offshore '87,
Rio de Janeiro, August, 1987.
/33/ Dowling, P.J., Ronalds, B.F., Onoufriou, A. and Harding, J.E.:
rd
"Resistance of Buoyant Columns to Vessel Impact", 3 Int. Symp. on
Practical Design of Ships and Mobile Units, PRADS '87, Trondheim, June, 1987.
/34/ Onoufriou, A., Elnashai, A.S., Narding, J.E. and Dowling, P.J.:
th
"Numerical Modelling of Damage to Ring Stiffened Cylinders", 6 Int. Symp.
on Offshore Mechanics and Arctic Engineering, OMAE-ASME, Houston, 1987.
/35/ Zayas, V.A. and Daog B.V.:
"Experimental and Analytical Comparisons of Semisubmersible Offshore Rig
Damage Resulting From a Ship Collision", OTC 4888, 1985.
/36/ Risk Assessment of Buoyancy Loss - Case Study 1 -
Deep Sea Bergen and Treasure Scout, Sictec/SINTEF, Trondheim, July 1987.
/37/ Amdahl, J.:
"Local Denting of Grouted Tubes", Veritas Report No 81-0032, Hovik, 1981.
/38/ Jorgensen, L.:
"Platform - Ship Impact Analysis of Column for GVA 3000", Veritas Report No 83-
1299 (Proprietary), Hovik, 1983.
/39/ Lindebjerg, K., and Andersen, L.:
th
"Ice Impact on Semi-Submersibles", 6 Int. Symp. on Offshore Mechanics and Arctic
Engineering, OMAE-ASME, Houston, 1987.
3.12 Illustrative example

Example 1:

a) Boundary conditions
This example shall illustrate a central impact on a semi-submersible platform brace 5m
above operation draft level. Calculate the energy absorption in the brace considering the
effects of local damage and local buckling.
Data : Length of brace L = 39.6 m
Brace dimension dxt = 1.543 x 0.0215 m
Impact point αL = 0.71 L
Material strength fy = 235 N/mm²
This evaluate the spring stiffness of the brace/deck connection a computer analysis of a
H3.2 rig is performed. The brace is removed from the model and replaced by its load
effects. Further, a unit force of 1 MN is introduced. This unit load gives joint displacement
as indicated below

Resultant spring stiffness in the brace direction


k1 = N / (0.0085+0.0151)cos45° = 60 MN/m
The spring stiffness criterion according to Section 3.7.2
K > (πLtfy) / d = π39.6 • 0.0215 • 235) / 1.543 = 407 MN/m
è k 1= 0.15(πLtfy) / d
Conclusion: Full axial restraint cannot be assumed!
For simplicity free ends are used in the following.
b) Energy absorption by local dent
The energy absorption in the brace is done by a combination of local damage and beam
deformation. Thus the load – deformation relationship for these two mechanisms must be
determined.
Load – deformation relationship for local damage is calculated according to Fig. 3.12 and
Ellinas/Walker formula. The d/t ratio used is 71.7.

δd / d 0.1 0.2 0.3 0.4 0.5


P / P0 0.53 0.75 0.92 1.06 1.19

P0 =
2
(Mp / L ) = 2.95MN and
2
Mp = fyd t = 12.03 MNm
α (1 − α )
Load – deformation relationship for plastic collapse in bending fig 3.0, load case 4 accounting
for local denting according to fig 3.13

δd / d 0.00 0.05 0.10 0.15 0.20 0.25


P / P0 1.21 1.07 1.00 0.95 0.91 0.87

The two curves intersect for a δd / d value of 0.225. Energy absorption in the local dent is the
area under the load – deformation curve for δd / d from 0.0 to 0.225. Numerical integration
yields:
El = 0.51 MNm
c) Energy absorption due to beam bending prior to local buckling
From the intersection point the beam deformation is the governing energy absorption
mechanism. The bending capacity at the impact point, Fig. 3.13, is
Mul = 0.46 Mp for δd / d = 0.225
The bending capacity at the brace ends before local buckling is initiated, Section 3.7.5
assuming rotationally fixed ends.
The deformation limit at which local buckling is initiated can be estimated according to
Section 3.7.5 assuming rotationally fixed ends.

This end deformation causes a beam deflection at midpoint of


δ1 = 0.12d = 0.19 m
Energy absorption in the brace due to beam deformation prior to the initiation of local buckling
is then

d) Energy absorption after local buckling


After local buckling is initiated the diameter is here assumed reduced by 50%, thus the
moment capacity at the beam ends is as follows
2
Mp’ = fy(d/2) t = 0.25 Mp
Maximum plastic strain before local buckling is initiated, Fig 3.17, load case 4, axially free
ends
ε1 = δ1 / (2αLη) = 0.19 / (2 • 0.71 • 39.6 • 1.0) = 0.0033
Assuming a critical strain εcr = 0.05 the available plastic strain is
ε2 = εcr - ε1 = 0.05 – 0.0033 = 0.0467
Which gives a plastic beam deformation of
δ2 = 2αLηε2 = 2 • 0.71 • 1 • 0.0467 • 39.6 = 2.62 m
Energy absorption after local buckling
E3 = (0.46Mp + 0.25Mp) δ2 / L (1/0.71 + 1/0.29)
E3 = 2.74 MNm

e) Total energy absorption


Local denting E1 = 0.51 MNm
Beam bending prior to local buckling E2 = 0.40 MNm
Beam bending after local buckling E3 = 2.74 MNm
Total Etot = 3.65 MNm
Example 2:

A horizontal jacket brace is exposed to a central impact load. Calculate the maximum energy
absorption in the brace considering first the pure membrane method and secondly the
combined bending/membrane method accounting for local denting.
Data Length of brace L = 20 m
Brace dimension dxt = 0.9 x 0.025 m
Impact point αL = 0.5L
Material strength fy = 340 N/mm²

a) Boundary conditions
To make sure that the brace supports are axially restrained an evaluation of the spring
stiffness is necessary. A simplified way of doing this is to calculate the axial stiffness in
the adjacent members disregarding the leg shear stiffness and axial stiffness of out-of-
plane members.

Axial spring stiffness, EA/L


Length of 1 and 3, L1 = 28.3m
Length of 2 L2 = 20.0

The member tension force, N, is assumed to be supported by axial compression in the


members 1 and 3.
2
Spring stiffness; k = 2(EA/L1)cos 45° = 525 MN/m
This spring stiffness must satisfy the criterion
K > (πL2tfy)/d = (π20 • 0.025 • 340) / 0.9 = 593 MN/m
Consequently the axial spring stiffness does not fulfill the above criterion. However,
considering the omission of shear stiffness and axial stiffness of out-of-plane members, we
assume that the members is fully axially restrained.
b) Energy absorption
Method 1:
The beam model for the pure membrane phase is given below.

Maximum plastic beam deformation according to Fig. 3.17 and ultimate impact force
according to Fig. 3.11 are respectively (εcr = 0.05 is assumed, η = 1)

The total energy absorption in the brace is the area under the load deformation curve in
Fig. 3.11
E1 = 0.5 • 2.34 • 1.49 • P0d = 4.32 MNm
Method 2:
The bending/membrane load – deformation relationship is determined using Fig. 3.11
assuming a pure membrane behaviour for δd/d > 1.0. In addition the effect of local damage
must be accounted for as indicated in Fig. 3.9C.
Load – deformation relationship for local damage according to Fig. 3.12 and a d/t ratio of 36.
The Ellinas/Walker formula is used:

δd / d 0.1 0.2 0.3 0.4 0.5


P / P0 0.92 1.30 1.60 1.84 2.06
Load – deformation relationship for plastic collapse in bending Fig 3.10, load case 4,
accounting for local denting according to Figure 3.13:

δd / d 0.0 0.1 0.2


P / P0 1.0 0.83 0.75

The resulting load – deformation curves are plotted below.


The maximum beam deformation, δ/d, according to Fig. 3.17 load case 4 is

which gives a P/P0 ratio of 1.3 in the figure below.

The total energy absorption is then the sum of E1 and E2 where


E1 = 2/3 •0.18 • 1.3P0d = 0.38 MNm
E2 = 0.5 (0.88 + 1.30) 0.4P0d = 1.08 MNm
E1 + E2 = 0.38 + 1.08 = 1.46 MNm
From the above calculations we conclude that for the brace dimensions used in this example
the variation in energy absorption from method 1 to 2 is significant. The large difference
seems to occur in the maximum strain calculation.
c) Tubular joint capacity
The check of the tubular joint capacity is done according to Section 3.7.6 and Fig. 3.15
through 3.17.
Considering a T-joint in tension:
2 2
Nu = fyt (5.1 + 23.5β ) t = 75 mm
Nu = 25.93 MN β = d/d 0 = 0.6
Qβ = 1.0
2
f y dt 0 d
Mu = 0.8 ( 3 .4 + 7 )Q β
sin θ d0
Mu = 10.5 MNm
Capacity of the tube element:
Np = πfydt = 24.0 MN
2
Mp = fyd t = 6.9 MNm
The interaction curve for tubular KT-joint is
1.2
(N / Nu) + (M / Mu) = 1.0
1.2
(N / 1.08Np) = (M / 1.52Mp) = 1.0
and the interaction curve for a tube element is
M / Mp = cos(πN / 2Np)
The joint capacity is sufficient it its interaction curve lies outside the entire axial force –
bending moment interaction curve of the tube element. As seen from the figure below the joint
capacity is sufficient in this case.
Example 3:

A horizontal jacket x-brace is exposed to a central impact load. Calculate the maximumenergy
absorption in the brace assuming a deformation mechanism involving several members.
Data: Length of brace L = 28.3 m
Brace dimension dxt = 0.9 x 0.025 m
Impact point αL = 0.2L
2
Material stength fy = fy = 340 N/mm

a) Boundary conditions
In order to check whether the brace is axially restrained or not the joint stiffness has to be
evaluated. The calculation is done in the same simplified way as shown in example 2.

Length of 1 and 2, L1 = 20.0 m


Length of 3 L2 = 28.3 m

The member tension force, N, is assumed to be supported by axial compression in the


members 1 and 2. The spring stiffness is
2
k = (EA1/L1 + EA2/L1)cos 45°
k = π (E/20)(1.5 • 0.035 + 0.9 • 0.025)0.5 = 1237 MN/m
The spring stiffness must satisfy the criterion
K > (π L 2tfy)/d = (π 28.3 • 0.025 • 340)/0.9 = 840 MN/m
Conclusion: Full axial restraint can be assumed!

b) Energy absorption
Initially, the impact load can cause either of the two basic deformation modes as indicated
below.
Mode 1:

Mode 2:

The first step is to calculate the plastic collapse – load for mode 1 and 2 to find the governing
beam deformation mechanism.
Mode1: P1δ1 = 2Mp(δ1/0.4L + δ1/0.6L)
P1 = 8.33Mp/L
Mode 2: P2δ2 = 2Mp(δ2/0.4L + δ2/0.6L + 2δ3/L)
Where δ3 = δ2/1.6
P2 = 8.75Mp/L
Mode 1 has the lower plastic collapse load. The reaction force in the intersection of the two
diagonal braces is then found to be
Px = 0.4 • P1 = 3.33Mp/L
We assume an increasing capacity of the brace in mode 1 due to membrane forces. The
capacity limit is reached when the reaction force,Px , is exceeding the plastic bending capacity,
P0, of the intersecting diagonal brace.
P0 = 8Mp/2L = 4Mp/L = Px
The corresponding inpact load is then
P1 = Px/0.4 = 10Mp/L
The load – deformation relationship could be determined using fig 3.11 and assuming a pure
membrane behaviour for δd/d > 1.0.
For Px > P0 a combination of mode 1 and mode 2 will start. A simplified way of calculating the
load carrying capacity in the large deflection regime is to assume pure membrane behaviour
and establish the reaction forces from vertical equilibrium. The horizontal equilibrium is
neglected.

1) P1 – Np(δ1/0.4L + (δ1-δ2)/0.6L) = 0
2) Np(δ1-δ2)/0.6L - Npδ2/L – Px = 0
The load carrying capacity due to membrane force in the intersecting brace is
Px = 2N pδ2/L
From equation 2 we then get δ2/δ1 = 0.36
This yields the following solution for the combined mode in the membrane phase
1) P1 = 3.75Npδ1/L = 11.2Mpδ1/Ld
The load – deformation curve for mode 1 and mode 2 could then be plotted in the same
diagram as shown in the figure below. The transition from mode 1 to mode 2 is drawn by
engineering judgement.
To determine the δ1/d ratio in rupture we have to calculate the maximum plastic strain which
is assumed to be

Assuming ε max = εcr = 0.05 and a reduction in effective diameter due to local damage of 50%
we get from the above equation the δ1/d ratio in rupture
δ1 = 1.12ε cr L = 1.12 • 0.05 • 20 = 1.12
è δ1/d = 1.12/0.9 = 1.24
The total energy absorption is then
E3 ≈ [ 0.5(10 + 8.33) 0.6 + 0.5(10 + 13.9)0.64] Mpd/L
Where Mp = fyd² t = 6.89 MNm
E3 = 2.92 MNm

c) Tubular joint capacity


A check of the tubular joint capacity is done according to Section 3.7.6 and Figs. 3.15 through
3.17.
Considering a Y-joint in tension:

The above axial and bending moment capacity values are larger than those calculated in
example 2, thus the tubular joint capacity is sufficient.
Example 4:

A jacket K-brace is exposed to a central impact load. Calculate the maximum energy
absorption in the brace assuming a deformation mechanism involving compressive members.
Data : Length of external brace L = 20.0 m
Brace dimension dxt = 0.9 x 0.025 m
Impact point αL = 0.5L
Material strength Fy = 340 N/mm²
In order to check whether the brace is axially restrained or not, the joint stiffness has to be
evaluated. The calculation is done in the simplified way as shown in example 2.

Length of 1 L1 = 20.0 m
Length of 2 L2 = 22.4 m

The member tension force, N, is assumed to be supported by axial compression in member 2.


2
Spring stiffness: k = (EA/L2) cos 60° = 166 MN/m
The spring stiffness must satisfy the criterion
k > (πL1tfy)/d = 593 MN/m
Conclusion: Full axial restraint cannot be assumed!
For simplicity free ends are presupposed.
b) Energy absorption of the external brace starts as a beam deformation until the impact
load becomes larger than the characteristic buckling strength of the adjacent out-of plane
members. The deformation mechanism is shown in the figure below.

The first step is to calculate the energy absorption in the brace due to beam deformation
taking into account the reduction in moment capacity due to local damage and local buckling.
The reduced cross-sectional slenderness is according to 3.7.5.

thus no initial reduction in moment capacity due to a local buckling.


The external brace in this example is similar to the one in example 2 except for the difference
in joint thickness, 40mm vs. 25mm. Consequently, the bending capacity of the dented cross-
section will somewhat higher in this example. For simplicity Mu is set to 90% of Mp.
The maximum plastic beam deformation according to Fig. 3.17 is
δmax = 2αLηεcr = 2 • 0.5 •20.0 • 1 • 0.05 = 1.0 m
The energy absorption due to beam bending of the external brace is then
2
E1 = 2(Mp + 0.9 Mp)2δmax/L where Mp = fyd t = 6.89 MNm
E1 = 2.62 MNm

c) Energy absorption in internal braces


The next step is to calculate the energy absorption in the compressive members. The
maximum beam deformation of the external brace will give a corresponding end shortening of
approximately
δe ≈ δmax cosφ/2
The plastic load – end shortening relationship for compressive members is according to
Section 3.7.9.

Where C and M are assumed to be 2 and Mp respectively.


The load – end shortening curve is plotted for the following values.

-3
δe /L • 10 1 5 10 20 30 40 50 60
N (MN) 21.9 9.8 6.9 4.9 4.0 3.5 3.1 2.9

The limits must be defined before calculating the energy absorption. The upper limit of N is
assumed to be the characteristic buckling strength, while rupture of the extended brace gives
the lower limit.
Calculation of characteristic buckling strength is here taken from NPD, Volume 2, Appendix 3,
section 3.2.1:

Assuming rupture in the external brace for the calculated maximum beam deformation the
corresponding compressive end shortening is 0.71m and the δe /L ratio is 0.05. This δe /L ratio
is assumed to be the rupture limit.
The compressive energy absorption is then according to the figure below.

d) Total energy absorption

Energy absorption external brace E1 = 2.62 MNm


Energy absorption internal braces E2 = 7.88 MNm
Total Etot = 10.50 MNm
e) Tubular joint capacity
The total energy absorption is calculated assuming that the tubular joints provide that
sufficient capacity to transfer the member reaction forces. This has to be checked further.
The ultimate axial force for a K-joint according to Fig. 3.15

The critical axial load is the characteristic buckling strength of 18.6 MN, consequently the joint
capacity is sufficient.
Check of punching shear capacity:

The punching shear capacity is sufficient.


4. DESIGN GUIDANCE FOR STRUCTURES EXPOSED TO FALLING OBJECTS.

4.1 Introduction
4.1.1 General
The scope of this chapter is to give guidelines for the direct design and documentation of
strength of offshore structures exposed to dropped objects.
The chapter contains the following areas of concern:
- strategical aspects of resisting/avoiding impact loads from dropped objects.
- typical accident scenarios
- determination of characteristic loads
- basic impact mechanics
- a practical design path for verification of various types of structures
Mainly this chapter concentrates on actions and load effects caused by objects dropped
during lifting operations (platform crane operations).
However, similar loads may also be caused by equipment falling down due to local structure
failures, but normally these events will not be considered as the probability will be too low.
Since the probability level for the design events will be low the actual loading will be
determined by the most frequent equipment handling operations. This again will in practice
limit the scope of such load investigations to topside structures, especially covering structures
such as roofs, or dedicated protection means.
Some special considerations are given to impacts from falling objects directly onto platform
support structures (jacket framing, buoyancy compartments, tethers, risers) and subsea
installations (templates, subsea pipelines).
Determination of impact energies related to a TLP tether which fractures, is described In
ref. /14/.

4.1.2 Notations

E Energy
Ei Strain energy dissipated in the impact structure
Ek Kinetic energy
Eo Strain energy dissipated in the falling object
Fs Applied force
Fs o Mechanism force
Fs u Critical force
Kt e Elastic stiffness
Kt p Membrane stiffness
Mp Mass Of projectile
Mt Mass of target
*
Mt Generalized mass of target
P (δ) Force/deformation relationship
Po Force/deformation relationship for object
Pi Force/deformation relationship for structure
T Kinetic energy
Tp u Critical impact energy
Us u Critical strain energy
dp Outer diameter of projectile
fu o Ultimate stress
fy o Yield stress
g Acceleration of gravity
h Height
ht Plate (target) thickness
lt Plate (target) side length
mi Effective mass of impacted structure
mo Mass of falling object, plastic yield moment per unit length
ro smallest distance from point of impact to plate boundary
v velocity
vc common velocity after impact
vi velocity of projectile at impact
vo velocity of falling object
w plate displacement
wc plate displacement at the edge of the projectile
wc o plate displacement at mechanism force
δo deformation of object
δi deformation of structure
β mass participation factor
ρt plate (target) density
τ critical shear stress

4.2 Design strategy


4.2.1 General
An acceptable safety against dropped objects can be obtained by the following three design
approaches (see Section 2.2):
- Event control
- Indirect design
- Direct design

4.2.2 Direct design


The principles for direct design against accidental loads from dropped objects can be
summarized as follows:
1. Detect the most likely progressive collapse mechanism which may be endangered by the
falling object.
2. Check whether the object has sufficient energy to trigger the collapse mechanism when
no barrier-structures are present.
3. In case of barrier-structures, check whether the object is sufficiently halted/retarded by
the barrier to avoid to trigger the collapse mechanism.
4. Check if the damaged structure can resist the functional loads and the environmental
loads with a specified return period (see section 2.4 and chapter 8).
Figure 4.1 gives an example of the direct design process for a structure exposed to dropped
objects.
FIGURE 4.1 - FALLING OBJECT – DESIGN FLOW SHEET

4.3 Falling object scenarios


4.3.1 General
The amount of damage produced by an object falling onto a structure may vary drastically as
function of various parameters such as target area location and degree of local collapse. The
appearance of these parameters will have their own probability distribution, which also must
be taken into account when calculating the total event probability.
This means that one must spend unnecessary efforts on events belonging in the category
with a rather "frequent" scenario in order to sort out "residual accidental events". It is therefore
necessary to check that input from other disciplines really reflects this aspect.
As the investigation is limited to "frequent" scenarios one may already state that events like
equipment or structures falling down by themselves (due to e.g. corrosion, blown off
foundations, vibrations) will, in combination with the consequencial damage probability,
belong in the "residual" domain, or they may be regarded as covered by the standard codes.
This means that only scenarios containing physical human controlled "activities" will be dealt
with here.
Typical scenarios are given in Table 4.1.

TABLE 4.1 - TYPICAL DROPPED OBJECT SCENARIOS


On basis of the above operations one will through examining of the frequency of the various
operations and the exposed areas involved be able to concentrate on those which may be of
interest to further investigations.
Typically, one would be inclined to believe that all operations involving use of crane would
create hazard.
This is specially the case when studying the revolving platform cranes.
Revolving pedestal cranes will in the operational phase of the platform lift a majority of all
items going to and from the platform, or even from one platform area to another. Further the
cranes are situated above most of the platform structures and may, due to the long booms,
reach considerable heights above the deck with its lifting hooks.
In a majority of cases operations involving the platform revolving cranes will govern the
accidental event definition for falling objects from two simple reasons:
1) Frequencies of operations
2) Lifting height potential
In most cases an investigation will show that it will be the "daily use" type frequency
operations which will be detrimental for the structural design, in other words, rather small lift
weights with a variety in shapes and drop heights.
The more specialized overhead cranes ("gantry cranes") have much less potential drop
heights and are also less frequently in use. Only exception would be overhead cranes located
above hatchways extending several stories down under.
Trucks and trolleys may be falling objects themselves, but will in most cases be less critical.
For subsea installations one may have difficulties to define the PLS check scenario, as
required by NPD, since the installation normally will be unmanned. However, the pollution
aspect is present, as well as potential economical losses, which both may serve as a basis for
design measures.
4.3.2 Categorization of loads
Impact loads may be categorized in different ways. In this context there will be distinguished
between two categories:
1) Impacts directly leading to a structural global collapse of the main structure or parts
thereof, e.g.:
- Damage directly onto main load-bearing structure resulting in progressive collapse of
the same. (mobilization of global collapse mechanism in load carrying structure).
- Puncture of buoyancy chambers and possible capsizing of vessel.
2) Impacts onto local "barrier" type structures with potential to escalate directly or indirectly
to a global collapse of the structure, e.g.:
- Drillpipe punching hole in deck hitting high pressure manifold resulting in blowout and
subsequent loss of installation.
Load of category 1 will normally be of the large mass/limited drop height type. The energy
involved may be considerable, and typically these events will activate the global deformation
mechanism rather than local ones. This is caused by the fact that large masses usually have
a substantial contact area and will immediately engage a major part of the structure, thus
contact pressures do not reach punching shear magnitudes.
Load of category 2 is the more "missile" type damage: Significant velocities / small contact
area and local denting/punching hole of the structure with the possibility to proceed and
damage equipment/structures below.
In most cases the damage will be limited to plates or plates/stiffeners failing very locally
(contact area) with only limited deformation of surrounding structures. Typically this may
occur when the drop height is high and the contact area small.
An extreme case (often used as a design case) is the dropping of a 3 tones mass drill collar
from 15 - 20 meters height hitting a plated deck in a straight vertical manner with its "pin" end
(outer diameter 146 mm). In such a case the chance for penetrating the plate is present with
subsequent damage to equipment down under.
Under certain circumstances the structural system would behave independent of for instance
plate field span and properties of the surrounding structure as defined by beam
spacing/grillage pattern and associated stiffness. In such cases the inertia forces (due to high
velocity and high density of the plate field mass to be accelerated) produced by acceleration
of plate mass in the vicinity of the impact would in it self be sufficient to produce penetration of
the plate. However, in most practical cases there will be influence from the surrounding
boundary conditions like panel stiffness, mass distribution and resistance against buckling
phenomena.
4.4 Characteristic dropped object loads
4.4.1 Probability levels for the individual events
Table 4.2 shows typical item and masses and indicative lift frequencies for lifting operations
by the platform revolving cranes.

TABLE 4.2 – LIFTING DISTRIBUTION AND FREQUENCY – EXAMPLE

From such tables one may determine the most likely object to be dropped from a platform
crane. By investigation of the lifting frequency and the associated exposure of platform areas
during the given lift operation, one should be able to calculate approximate probabilities for
hitting a certain part of the structure with various types of objects.
Ultimately one could also produce probability density distribution diagrams (2D/3D
presentation overlayed the platform plan drawings), see Figure 4.2.
Normally this type of evaluation is prohibitive due to lack of data for future lifting operations.
Usually the design event will be determined among the "daily" use category lifting operations,
that Is, lifting of relatively small objects. In practise this would limit the reach/height-capacity
curve of the crane whip hoist, usually limited to max. 7-8 tones lifting operations, depending
on size of crane and lifting radius.
For determination of the drop height normal practise has been to use the reach/height-
capacity curve of the crane/whip hoist, or max. lifting heights given by effective lifting
procedures or limited by automatic "key-locks", overruling the crane operator.

FIGURE 4.2 – DROPPED OBJECT IMPACT DISTRIBUTION – EXAMPLE

The input premise is then adequately defined in terms of type of objects and drop heights in
the different areas of the platform and the kinetic energies may be calculated.

4.4.2 Object behaviour under impact


The nature of the object like physical composition, size, orientation at the time of drop and
eventual rotation of the object as a function of sling breakage etc. makes the analytical part of
the impact calculation rather complicated. The contact force may create a moment about the
object center of gravity leading to changes in "effective retarded mass" of the falling object
(angular acceleration instead of object retardation).
However, in most cases these facts may simply be neglected and a "worst case" behaviour of
the object may conservatively be established ( i.e. rigid body behaviour - straight translation).
According to accepted principles the force Imposed onto the target area from a dropped
object also acts onto the object itself. This means that a part of the total energy (kinetic
energy) involved may be used to plastically deform the object itself. This can be the dominant
part of energy dissipation in some cases:
- helicopter crashing against helideck/platform roof
- ordinary ISO type containers dropped during lifting (verified in full scale tests)
Other objects may behave more like rigid non-deformable bodies:
- drillpipe/drillcollars
- heavy spareparts (castings)
- machinery
In most cases the design accidental load results from objects behaving dominantly as the
latter type. In practice this means that all kinetic energy will have to be absorbed by the target
area structure and its surroundings as strain energy and plastic deformation work, eventually
with residual kinetic energy of the object if the object passes through the structure.

4.4.3 Support structures


By the term support structure is ment Jacket or gravity base structure (GBS).
Normally a falling object would originate from activity at the surface and most likely activities
of the same nature as described in the previous part of this chapter.
What may be different is the actual velocity, the trajectory of the object, and the effect of the
added mass of object and target structure respectively.
Regarding velocity one may in the literature find examples on terminal velocity for certain
objects in water (in most cases reached after 10 to 20 meter travel from surface), see
Table 4.3 /10/.

TABLE 4.3 – MASS AND TERMAL VELOCITY FOR OBJECTS IN WATER – EXAMPLE

The direction of travel may be close to horizontal, especially near the surface. It is therefore
recommended to consider all applicable directions.
It Is recommended to consider all structural elements which finds itself within a 20 degree
cone starting at the surface intercept of the object. At 150 m water depth this would represent
a horizontal travel of about 50 to 60 m in any direction, which obviously represents a large
part of any sub-structure.
The proposed cone is based on stochastic simulations of various items and may be
considered as the maximum expected value.
4.4.4 Buoyancy structures
On floating platforms critical situations may arise in case of punctuation and subsequent
flooding of components in pontoons or columns.
In principle the problem definition is equivalent to the jacket/GBS as discussed above, and will
not be subjected to any detailed discussion here. One could however be aware that the
added mass effect involved for submerged stressed skin type structures (standard ship hull
composition structure with liquid on one or both sides of the plate) may drastically change the
dynamic behaviour of panels compared to their behaviour in air. In such cases model tests
may have to be carried out in order to justify results.

4.4.5 Subsea installations


Subsea installations will usually be covered in the same manner as jacket structures. A
vulnerable part may be riser systems. In the case of a tension leg platform, damage from
tethers (fractured parts or dropped elements) must be considered.

4.5 Impact mechanics


4.5.1 General
Most dropped object problems can be treated as single degree of freedom problems, i.e.
unidirectional translation/deformation. In the following single degree of freedom systems are
assumed. It is also assumed that the target structure has zero velocity before impact, that is,
all energy included in the problem is the kinetic energy of the falling object.
Generally, an impact can be divided into two phases:
i) Acceleration of target structure/retardation of falling object to a common velocity
ii) Retardation of object and target structure by elastic/plastic deformation work

4.5.2 Fundamental principles


The kinetics energy for the falling object is
2
Ek = ½ movo (4.1)
Where
mo = mass of the falling object
vo = velocity of object at impact moment
For impact in air
2
vo = 2gh (4.2)
For impact against underwater platform structures reference is made to /12/.
If the impact duration is small the impact and the dynamic behaviour of the impacted structure
can be treated as two separate problems. The solution of the problem can be based upon the
principles of
- Conservation of momentum
- Conservation of energy
4.5.3 Conservation of momentum
Conservation of momentum during a central impact for a dropped object is expressed by the
equation
movo = (mo + mi) vc (4.3)
where
vc = common velocity after impact
mi = effective mass of the impacted structure
Using the velocity vo = 2gh for drop in air the common velocity after impact is expressed by

mo 2gh
v= (4.4)
mo + mi

4.5.4 Conservation of energy


The kinetic energy before impact dissipates as strain energy in the falling object and in the
impacted structure. The equation of energy conservation assuming central impact is
2 2
½ movo = ½ (mo + mi)vc + Eo + Ei (4.5)
where
Eo = strain energy dissipated in the falling object
Ei = strain energy dissipated in the impacted structure
Using Eqs. (4.2) and (4.4) the conservation of energy is expressed by

 mo 
E o + E i = mo gh1 −  (4.6)
 mo + mi 

4.5.5 Dissipation of strain energy


Generally, dissipation of strain energy is expressed by
dE = P(δ)dδ
where
dE = dissipated energy
P(δ) = force/deformation relationship fir the structure.
For a given load level, the absorption of energy will be represented as the area under the P-δ
curvature.
For dropped object problems this may be expressed as
∆o ∆i

Eo + Ei = ∫ Po (δo )dδo + ∫ Pi ( δi )dδi (4.7)


0 0

Where
Po, Pi = force/deformation relationship for the object and the
impacted structure, respectively
∆o, ∆i = deformation of the object and the impacted structure,
respectively
4.6 Plugging capacity of steel plates
4.6.1 General
The response of a steel plated structure subjected to a dropped drill-collar consists of a global
deformation of the plate, combined with local shear deformation at the impact point. This
combined mechanism results in a difficult design situation, and existing design formulas
/15/,/16/ show poor agreement with dropped object tests. Furthermore, formulas developed
for other mass and velocity regimes /3/ are not directly applicable to the problem, as the
physical behaviour during impact is quite different.
Due to the lack of knowledge about penetration mechanics related to large mass projectiles in
the low velocity range, an experimental model study has been carried out at the Division of
Structural Engineering, The Norwegian Institute of Technology /4/ to develop a simplified
analytical model for the plugging capacity of steel plated structures subjected to a dropped
drill-collar. The following recommendations are based on this study.

4.6.2 Impact phenomena


When a projectile impacts a steel plate a very complex process occurs due to the transfer of
kinetic energy from the projectile to the plate. This happens, as observed by Langseth /4/; in
two phases which are separated by the time when the plate supports are activated.
Figure 4.3 shows a typical force-time curve where the two phases, called the transient phase
and the global mode phase, are given. In the transient phase only inertial forces are present
in the plate. In the global mode phase the supports are fully activated and the target behaves
almost quasistatically. The plugging failure always occurs in the latter phase and at maximum
force.

FIGURE 4.3 – PLATED STRUCTURE SUBJECTED TO A DROPPED OBJECT

The deformation pattern in the transient phase is characterized by a circumferential “hinged” moving
in the radial direction towards the plate boundary as shown in Figure 4.4. Here, region ¬ is
elastic-plastic with both membrane and bending stresses present, while - is purely elastic.

FIGURE 4.4 – DEFORMATION PATTERN IN THE TRANSIENT PHASE


4.6.3 Design model
A design model must take into account what happens both in the transient phase and the
global mode phase including the plugging failure. In the following such a model is derived by
using an energy approach, taking the drill-collar as a rigid flat ended body, and neglecting
strain rate effects in the target material.
Figure 4.5a and b show a square plate impacted by a projectile at its center. The projectilehas
a mass Mp, and impact velocity vi, is flat ended and has an outer diameter dp. The target has
density ρt, thickness ht, side length lt, and has arbitrary boundary conditions.

FIGURE 4.5 – PLATED STRUCTURE SUBJECTED TO A DROPPED OBJECT

The calculation of the dissipated in the transient phase is difficult to carry out, but may be
avoided by means of an energy balance at the end of the phase. The transient phase of the
impact is for simplicity considered as terminated when the hinge reaches the boundary, and
the supports are activated, Fig 4.5c.
An axissymmetric deformation pattern is assumed at the end of the transient phase on the
form
w(r,t) = wc(t) • f(r) (4.8)
Where wc(t) is the time dependent plate displacement at the edge of the projectile, and f(r) a
shape function.
The corresponding velocity field is

(4.9)

where is the velocity of the plate centre and the projectile at time t - tB when
the hinge reaches the support.
Based on the velocity field in Eq. (4.9), an energy balance is set up, where the kinetic energy
of the projectile and target is assumed to be dissipated as strain energy during the global
mode phase. The strain energy of the plate at incipient plugging is assumed equal to the
corresponding static critical strain energy Usu, where Usu is a function of the target geometry,
support conditions and the mechanical properties of the material.
Taking the total kinetic energy T at a time tB as starting point, this gives
T(tB) = ½ Mp v B + ½ Mt • v
2 * 2
B = Us u (4.10)
*
Mt is the generalized mass of the target, and is determined from

(4.11)
The mass participation factor β is defined by

(4.12)
and is found by combining Eqs. (4.9) and (4.11)

(4.13)
Mt is here the plate mass enclosed by the hinge circle at time tB which means that the target
mass enclosed by the nose of the projectile is neglected.
The velocity vB of the projectile at the time when the hinge reaches the support is determined
on the basis of linear momentum and experimental observations, /4/, and may be expressed
as

(4.14)
where α = 0.48 is an experimental determined factor.
Combining Eqs. (4.10), (4.12) and (4.14) and solving for the critical impact energy Tpu of the
projectile yields:

(4.15)
or on dimensionless form:

(4.16)
4.6.4 Design recommendations
Based on the tests carried out and the outlined model the plugging capacity of a plated
structure subjected to a flat ended dropped object can be found by using the formula

(4.17)
where
fyo = static yield stress of the target material
ro = smallest distance from point of impact to plate boundary
The formula is valid for a plated structure without stringers, with the following limitations:
Steel quality: St 52-3N or steel with similar strain rate sensitivity.

Mass ratio:

Scaled span length: l t /d p ≤ 14.5


Note:
No restrictions are given to b t
Scaled target thickness: ht/dp ≤ 0.270
Comment:
The factor 1/(1+β Mt/Mp) in Eq. (4.16) is omitted as tests have shown that it takes on
values ranging between 0.94 and 0.99.
The static strain energy Us u may be calculated by means of a nonlinear finite element
program based on the actual geometry of the target and its boundary conditions, the
simplified method given in Appendix A or any other available method.
The failure criteria for plugging may be based on a mean critical shear stress found
from the static test /4/ as follows:
τc r = fu o • (0.41• ht /dp + 0.42) (4.18)
The critical plugging force, Fs u, is given by
Fs u = πdphtτcr (4.19)
The experimental study was restricted to square plates subjected to a central impact.
However, Eq. (4.17) will also apply to situations where the impact takes place at other
locations and to rectangular targets when the smallest span length lt is less then
14.5 • dp. Figure 4.6 shows how the target mass Mt, represented by the shaded
areas, should be calculated for such cases. No restrictions are given to bt.
FIGURE 4.6 – CALCULATION OF TARGET MASS

Eq. (4.17) may also be used for targets with stringers when neglecting the target mass, i.e
r0= 0 ⋅ Usu is then calculated based on the actual geometry. However, the critical shear failure
criteria given above should be used only away from the stingers.

4.7 References
/ 1/ Wenger,A. ,Edvardsen,G. ,Olafsson,S. ,Alvestad,T., Taylor,T.:
"Design for impact of dropped objects", OTC 4471, 1983
/ 2/ Sereide,T. ,Amdahl,J.:
"Deformation characteristic of tubular members with reference to impact load and
dropped objects", Norwegian Maritime Research, 2/1982.
/ 3/ Neilson,A.J.:
"Empirical equations for the perforation of mild steel plates", International journal of
Impact Engineering, 1985, Vol. 3, no. 2, pp. 137-142.
/ 4/ Langseth,M.:
"Dropped objects. An Experimental Investigation of the Plugging Capacity of Steel
Plates", Dr.ing. Thesis. Division of Structural Engineering, Norwegian Institute of
Technology, Trondheim 1988.
/ 5/ Johnson,W.:
"Impact strength of materials", Edward Arnold publisher, London 1972.
/ 6/ Clough,R.W.,Penzien,J.:
"Dynamics of structures", Mc. Graw Hill 1982.
/ 7/ Langen,I., Sigbjoernson,R.:
"Dynamisk analyse av konstruksjoner", Tapir 1979.
/ 8/ Blodgett,O.W.:
"Design of structures", Ohio 1966.
/ 9/ Dallard,P.R.B., Miles,J.C.:
"Design tools for impact engineers", International journal of impact engineering.
/10/ Det Norske Veritas TNA 101:
"Design against accidental loads", 1981. (New issue: RP D204).
/11/ NPD:
"Guidelines for the determination of loads and load effects", Norwegian Petroleum
Directorate, 1987.
/12/ Amdahl,J.:
"Offshore ship collision and dropped object", Last og sikkerhet for marine
konstruksjoner, Seminar, Norw. Soc. of Chartered Eng., The Norwegian Institute of
Technology, Trondheim, 1985.
/13/ Palm,E.:
"ME-1 - Forelesninger i dynamikk", Matematisk institutt, Universitet i Oslo, 1970.
/14/ Moan,T., Webster,W.C.:
"Consequence Analysis of Tether Failure", Proc. Fifth BOSS, NTH Trondheim, 1988.
/15/ Edvardsen,G., Marley,M.J.:
"Dropped Objects. A summary report on damage assessment and energy
absorption", Project report no. 8, Det Norske Veritas, Oslo 1983.
/16/ GJeldsvik,A.:
"Mobile Statfjord A Platform. Dropped Object Study", Aker Engineering A/S, Oslo,
1977.
/17/ Timoshenko,S.P., Woinowsky-Krieger,S.:
"Theory of Plates and Shells", Second Edition. McGraw-Hill International Book
Company, 1959.

4.8 Illustrative example


A drill collar, diameter 9.5” and mass 3 tons is falling from 10 meter height to a simply
supported deck plate. The deck plate is quadratic, lt = bt = 3m.
Material constants for the deck plate:
Density ρt = 7.85 tons/m³
Yield stress fy o = 350 MPa
Ultimate stress fu o = 450 MPa
= 2.1 • 10 MPa
5
Youngs modulus E
Assume deck plate thickness ht = 15mm
According to Eq. (4.17) the critical impact energy of the deck plate is
Tp u = 1.284 • Us u
where Us u is the static critical strain energy
w cu

Us u = ∫F
0
s • dw c

where wcu is the deflection corresponding to the critical plugging force, Fs u, which obtained
from Eq. (4.19)
Fs u = πdphtτcr = 2279 kN
Where the critical shear stress is calculated according to Eq. (4.18)
τcr = (0.41ht /dp + 0.42) fsu = 201 MPa
The force-displacement curve is, according to Appendix 4A, given as

According to formulaes given in Appendix 4A:


The mechanism force Fs o = 136 kN
Membrane stiffness Kt p = 14239 kN/m
Elastic stiffness Kt e = 622 kN/m
Displacement at mechanism force wco = 0.031 m
The force-displacement curve is drawn in Figure 4.7
From the force-displacement relationship

Static critical strain energy is then calculated as

The plugging capacity then becomes


Tp u = 236 kJ
Energy of the object
Mpgh = (3 • 9.81 • 10) = 294 kJ
I.e. the plugging capacity Tp u < Mpgh è the plate thickness should be increased

Try plate thickness ht = 20 mm:


Then Tp u = 1.387 • Us u
τcr = 204 Mpa
Fs u = 3097 kN
Mechanism force Fs o = 241 kN
Membrane stiffness Kt p = 18985 kN/m
Elastic stiffness Kt e = 1473 kN/m
Displacement at mechanism force wc o = 0.035 m
From the force-displacement relationship
wc u = wc (Fs u) = 0.0186 m
and the static critical strain energy is calculated to
Us u = 255.6 kJ
Plugging capacity Tp u = 355.2 kJ > Mpgh è capacity ok

Check of limitations:
Span length: l t /d p = 12.5 < 14.5 è ok

Target thickness: ht /dp = 0.0625 (ht = 15mm) < 0.27 è ok


= 0.0833 (ht = 20mm) < 0.27 è ok
Mass ratio:
FIGURE 4.7 – FORCE DISPLACEMENT CURVE
APPENDIX 4A

4A.1 Static strain energy


4A.1.1 General
Having computed the critical force Fs u and determined the forced-displacement relationship,
the strain energy Us u is given as the area under the load-deflection curve.

4A.1.2 Static force-displacement curve


Figure 4A.1 shows a typical force-displacement curve for a plated structure. At initial loading
the plate behaves in a purely elastic way with an elastic stiffness Kt e. With increasing force
yielding takes place in the plate and the force is carried by a combination of bending and
membrane stresses.

FIGURE 4A.1 – FORCE-DISPLACEMENT CURVE FOR PLATE STRUCTURE

Above the mechanism force Fs o, or beyond the displacement wc o, the stiffness is assumed to
be constant and equal to the membrane stiffness Kt p. The mechanism force Fs o is defined as
the force which gives a plastic mechanism in the plate when assuming small deflection
theory.
The force-displacement curve is developed on the basis of the principle of virtual work when
assuming
1) A deformation pattern in the plate
2) Virtual strains based on the nonlinear strain displacement relationships.
3) A stress field in the plate.
4) A rigid-perfectly plastic material with no strain hardening. A correction is subsequently
included to account for elastic deformations.

Deformation pattern.
Figure 4A.2 shows a quadratic plate loaded at its center by a punch with diameter dp. The
plate has a span length lt and a thickness ht. The boundary conditions are either clamped or
freely rotating. Based on tests and numerical simulations the deformation pattern is assumed
axisymmetric as shown in the figure.
FIGURE 4A.2 – QUADRATIC PLATE LOADED BY A CIRCULAR PUNCH

Virtual strains
The virtual strains δεr r and δεφ φ in the radial and circumferential directions respectively are
derived from the nonlinear strain-displacement relationships /17/

(4A.1)

Here, is the displacement in the radial direction when


neglecting the displacement uo of the middle plan and w is the normal displacement, Eq.
(4A.1) is valid for moderately large displacements and rotations.

Stress field
A stress field corresponding to the virtual strain pattern is assumed in the plate. In the
circumferential direction only bending stresses are included. The stresses in the radial
direction are a combination of both bending and membran stresses that depend on the
magnitude of the deformations. However, for force levels above the mechanism load Fs o the
membrane stresses dominate, and a pure membrane stress field is hence used.
The assumed stress field must satisfy the yield criterium for material. For the present
application the material is taken to be elastic-perfectly plastic, and for the sake of simplicity
the field criterium of maximum normal stress is used.

Virtual work
The relationship between the applied force Fs and the displacement wc is found using the
principle of virtual displacement. Applying a virtual displacement δw from an equilibrium state
yields

F • δw c = ∫ σ r r • δε r r • dV + ∫ σ φ φ • δε φ φ • dV + πl t • mo • δθ (4A.2)
V V
where
F • δwc = external virtual work

∫σ
V
r r • δε r r • dV = internal virtual work in the radial direction

∫σ
V
φφ • δε φ φ • dV = internal virtual work in the circumferential direction

π • l t • mo• δθ = internal virtual work on the rim for a clamped support


condition. The term is omitted when having a freely
rotating support.
= ¼ • h t • f y o is the plastic yield moment per unit length
2
mo
Based on Eq. (4A.2) the force-displacement relationship may be written as
Fs = K t p • wc + Fs o (4A.3)
In this simplified calculation, the only contribution to Fso come from bending in the
circumferential direction and from the plastic moment on the rim for a clamped target.
Furthermore, the stiffness Kt p is independent of the support conditions.
Figure 4A.3 shows a comparison between Eq. (4A.3) and the desired force-displacement
curve. Here, no elastic deformations have been considered. However, this is taken care of by
assuming a transition curve in the region 0 ≤ Fs ≤ Fs o.

FIGURE 4A.3 – FORCE-DISPLACEMENT CURVES

The following formulas can be used to establish the force displacement relationship.
1. General formulas

Constants:

Displacement at mechanism force:


Force-displacement curve:

2. Freely rotating plate

Mechanism force:

Membrane stiffness:

3. Clamped plate

Mechanism force:

Membrane stiffness:

Comments:
The elastic stiffness Kt e can be found in textbooks and is calculated on the basis of
the actual geometry and support conditions.
The force-displacement relationship can also be used to situations where the load is
applied away from the center and to rectangular plates. In such cases the span length
lt is taken equal to 2ro where ro is the smallest distance to support as shown in
Figure 4.6. This will give a conservative design.
5. DESIGN GUIDANCE FOR STRUCTURES EXPOSED TO EXPLOSIONS

5.1 Introduction
5.1.1 General
The objective of this chapter is to give guidance for design of offshore steel load bearing
structures exposed to explosion loads. This chapter gives guidance in accordance with NPD
"Regulations for structural design of load bearing structures etc." /2/ regarding accidental
explosion loads.

5.1.2 NPD regulations and guidelines


Explosion is classified as an accidental load in NPD "Regulation for structural design of load-
bearing structures etc." /2/. The explosion loads and explosion areas should in general be
determined on the basis of a safety analysis as described in Chapter 2.
All offshore installations on the Norwegian continental shelf shall be classified in terms of
explosion risk, and divided into zones according to degree of risk /l/. Installation lay-out aim at
minimizing the consequences of fire and explosion. Systems and components shall be
designed and used to minimize the probability of fire and explosion.
Requirements and guidelines regarding safety evaluations and area classifications are further
given by NPD in /3/, /4/ and /5/. The hazardous areas are divided into 3 zones dependent on
probability of the release of flammable gases and vapours which could result In an explosive
atmosphere being formed.
With reference to NPD "Guidelines for determination of loads and load effects" /6/, account
should be taken of the following types of explosions:
- ignited gas clouds
- explosion in enclosed rooms
Structural designs should be in such a manner that the effects of explosions are reduced.
Further the changes in the properties of materials caused by fire and explosion may be taken
into account. Fires and explosions should be assumed to occur at the same time. Regarding
fire, reference is made to Chapter 6.

5.1.3 Definitions
Auxiliary systems: Auxiliary equipment for use in connection with operation of
the production installation and other equipment installed to
prevent or restrict accidents and damage including pollution
in connection with operation of the production installation.
Deflagration: An explosion or rapid combustion with flame front
propagation well below the speed of sound (~ 300 m/s).
Detonation: An explosion propagating at supersonic combustion velocity
(~ 2000 m/s) resulting in shock pressure.
Hazardous areas: A space, or part of a space or other area where there exists
or may exist an explosion risk.
Source of Ignition: Ignition point releasing a quantity of energy sufficient to ignite
an explosive atmosphere.
Ductility factor: Ratio of the actual structural displacement to the
displacement at first yield.
5.2 Design strategy
5.2.1 General
An acceptable safety against explosions can be obtained by different means. Section 2.2
specify event control, indirect design and direct design.
Almost every hydrocarbon fire and explosion follows an unexpected leak of inflammable
liquids or gases. A proper level of safety against fire and explosion could therefore by event
control be achieved by:
- reducing the probability of leaks
- reducing the consequences of any leak
Indirect and direct design as briefly described in Chapter 2 includes:
- improvement of structural resistance and capacity without directly considering the
explosion load in question. This implies structural ductility and redundancy with energy
absorption.
- structural design to demonstrate sufficient safety and structural dimensions for the
explosion events and effects obtained from the safety analyses.
Hazardous areas or rooms shall to the extent possible be so built that the load-bearing
structures roof and walls will not collapse or be blown away by an explosion in the area. Relief
of the explosion pressure shall be ensured by proper design e.g. by explosion vent openings
or blow-out panels /3/.
Due consideration should be given to the direction of explosion venting, possibility of flying
objects, drag forces from the vent flow, and the flame "gust" extending out of the openings.
When dimensioning and locating fire walls the consequences of explosion must be
considered.

5.2.2 Direct design


The direct design for load-bearing structures subjected to explosions loads can be
summarized as follows:
i) define explosion loads, areas and structural elements to be considered as a result of
safety analysis (Chapter 2). Note the effects of explosion vent openings, and
probability and consequences of leaks, to reduce explosion loads
ii) determine load effects during the explosion
iii) calculate the ultimate load bearing capacity (residual strength)
iv) check that load effects are below ultimate load bearing capacity, and evaluate the
extent of local damage
v) design pressure relief systems. Note the influence of explosion venting to keep
explosion loads below the load-bearing capacity of the structure (see Section 5.4.4)
vi) assess residual strength of damaged elements
vii) assess structural load-bearing capacity in damaged condition
Effects on the material strength such as strain hardening, high temperature due to fire, and
loading rate, should be carefully evaluated in order to avoid non-conservative assumptions
and results, see Section 5.5.
Design guidances for damaged structures are given in Chapter 8.
5.2.3 Combination with fire
According to NPD guidelines /6/, fires and explosions should be assumed to occur at the
same time. For practical purposes, the structure can normally be designed to withstand
explosion loads alone as a first event. Thereafter the structure can be designed for the effect
of a subsequent fire. Fires will change the steel material properties, dependent on
temperature. Regarding changes in steel strength and elasticity, reference is made to
Chapter 6.

5.3 Explosion scenarios


5.3.1 General
An accidental release of hydrocarbons at or near an offshore installation represents a major
threat to the people working there, the installation itself, and the environment. The main
hazards are linked to the formation and possible ignition of premixed gas/air mixture within
certain concentration limits of inflammability, and possible succeeding fire.
The complete sequence of fire and explosion events has to be studied and the limitations and
uncertainties of each separate step considered. As a result-of such studies, explosion areas
and explosion loading should be given for the proper dimensioning of load-bearing structures.
There are three main groups of explosions. These are:
- detonations (i.e. a supersonic process)
- deflagrations (i.e. a subsonic combustion)
- pressure explosions (sudden energy release e.g. from a pressure vessel rupture)

5.3.2 Detonations
If a stable detonation occurs in a large hydrocarbon vapour cloud, the explosion side on
2
pressure will be about 1800 kN/m , and will be disastrous to the platform. It is not practically
to design a platform to withstand such pressure so as to withstand a full stable detonation
once it is established.
Detonations are very unlikely in unconfined accidental vapour clouds unless they are
deliberately ignited. Hence this group of explosion event are normally not considered as a
design explosion load.

5.3.3 Deflagrations
When igniting a combustible gas mixture the hot burned gases expand, resulting in a
pressure rise across the flame front and within the whole confined area in the case of a
confined explosion.
This expansion will also set up a gas flow which, in turn, may stretch and fold the flame,
produce turbulence, and initiate combustion instabilities. All these phenomena contribute to
an acceleration of the flame propagation, and hence to an increase in the energy-release rate
and the pressure build-up.
This coupling mechanism between flame acceleration and gas dynamics is the key problem in
gas explosions. Reference is given to /8/.
Deflagration is a heat and mass transfer explosion. Such explosion may be divided into three
groups:
- confined explosions (closed compartments)
- partly confined explosions (vented compartments)
- unconfined gas cloud explosions
Confined or partly confined explosions can occur in vessels, rooms or in other configurations
which may confine the gas-mixture. The rate at which the pressure. Is increased and the
maximum pressure from the explosion depends on the percentage of the total volume
occupied by the flammable gas mixture, the area of pressure relief openings, the location of
the ignition source and structures and equipment within the confinement. The latter can lead
to considerable flame accelerations.
Damage outside the confinement in case of an explosion depends on whether the
confinement ruptures or not.' If the confinement totally ruptures a pressure wave propagates
radially from its former position. In addition, fragments can also cause damage. If total rupture
does not take place, the pressure wave will emanate from the relief apertures, and the blast
outside the confinement will depend on the position relative to these sources.
Unconfined deflagrations may form blast waves of different strengths, depending on the flame
velocity. Transition from deflagration to detonation may happen if very intensive flame
accelerations occur.
Section 5.4 give guidance for first approximative values of confined, unconfined and partly
confined deflagration. These guidances are only applicable for deflagrations in saturated
hydrocarbon gases mixed with air at initial atmospheric pressure. More detailed calculations
applying recognized analysis methods should be carried out for special geometries and
arrangements.

5.3.4 Pressure explosions


Pressure explosions may be due to rupture of vessels or piping due to malfunctioning of
safety valves or safety equipment elsewhere in the process. A rupture can also occur due to
fatigue, corrosion etc. in the process equipment. The pressure wave from a pressure
explosion can cause damage to nearby structures. Flying objects from the exploding
confinement may cause the most severe damage.
The maximum pressure to be used in the design of load-bearing structures are normally a
part of the safety analyses. Often structural failures are only expected in secondary structural
elements, hence analyses for Progressive collapse Limit State (PLS) according to NPD /2/
may in such cases be waived.

5.4 Characteristic explosion loads


5.4.1 General
Characteristic explosion loads due to deflagrations are given below for unconfined, confined
and partly confined explosions. References is made to Det norske Veritas TNA 101 /7/. These
guidances are only approximative values, and more detailed calculations are normally carried
out for each specific geometry and arrangement.

5.4.2 Unconfined gas cloud explosions


For a completely unconfined gas cloud of radius Rc prior to ignition, a pressure build-up of the
order of magnitude of 0.1 bar is expected within a distance of x = 2 Rc from the cloud centre.
Outside x = 2 Rc the pressure will drop inversely proportional (1/x) to the distance from the
cloud centre, see figure 5.1
FIGURE 5.1 – PRESSURE LOADING FROM UNCONFINED GAS EXPLOSION /7/

Note that for open construction elements such as e.g. rolled H beams inside an unconfined
gas cloud, the built-up pressure is equal at both sides of the element web and flanges, and
the resulting pressure loading is approximately zero. For closed construction elements such
as e.g. plate box girders, the built-up pressure will occur as lateral local pressure on plate
walls.

5.4.3 Confined explosions


In a completely closed compartment, and with the whole volume filled with an explosive gas
mixture, the theoretical maximum deflagration pressure increase will be approximately 7 bar.
In practice, the maximum pressure will be determined by the weakest part of the structure
causing a pressure relief.
For closed compartments partly filled with explosive gas (at initial atmospheric pressure), the
maximum pressure can be estimated as:
∆p = ε • 7 (bar) (5.1)
where ε is the fraction of total volume containing explosive gas. If the total volume contain
explosive gas, ε = 1.0. For partly filled compartments, ε < 1.0
Warning: similar reduction in the maximum pressure is not applicable for vented
compartments (partly confined explosions).

5.4.4 Partly confined explosions


The characteristic explosion load in vented compartments, ∆p, can be found from Figure 5.2
where the explosion over pressure is plotted for different sized volumes and vent areas. The
vent area Av can be taken as the sum of free opening areas and blow-out panel areas (i.e.
weather claddings) provided:
2∆ps t a t < ∆pc (5.2)
where
∆ps t a t = static opening pressure used for design of the blow-out panels
∆pc = 0.1 bar; pressure build-up in centre of an unconfined gas cloud, see
Figure 5.1
When the blow-out panels start to open at ∆ps t a t the total vent openings are not fully
effective, and it take some time to accelerate the panels away from the opening. During this
time the explosion pressure within the confinement continues to increase. The factor 2
accounts for this dynamic behaviour.
FIGURE 5.2 – EXPLOSION PRESSURE FOR VENTED COMPARTMENTS /7/

If 2∆ps t a t > ∆pc for blow-out panels or walls foreseen for explosion venting, the characteristic
pressure ∆p should be increased to 2∆ps t a t.
Figure 5.2 should only be used for compartments where:
L/D<3
Where

D= A
A; the smaller cross-sectional area
L; the greater dimension of the compartment
If L / D > 3, the compartment should be divided in section of L / D = 3 and the vent areas for
each section determined according to Figure 5.2.
Note that if venting is provided through a compartment on both sides of the explosion source,
a vent area of only one side and half volume of gas cloud should be applied for use in
Figure 5.2. Pressure are then assumed symmetrical.
With respect to other walls than blow-out panels the dynamic amplification for the structural
response is discussed in Section 5.5.3.
For explosion propagation from compartment to compartment and for tunnels and chutes
where explosion venting can be foreseen at one end area only, detailed investigation should
be carried out.
5.5 Structural response
5.5.1 General
The structural elements and equipment subjected to an explosion can be damaged from
pressure loading and, as a secondary effect, suffer damage from flying object and fragments.
The mode of failure of interest is ultimate collapse and excessive deformation. Hence, the
purpose of the analyses will be to predict an extreme response value, including dynamic
amplification effects.
The damage resulting from an explosion should be assessed by comparing the load effects
(structural response) to the capacity of the actual structure.

5.5.2 Structural resistance


Recognized elastic and plastic theories are normally used to determine the ultimate capacities
of load-bearing elements.
The ultimate capacity of a structural element considering plastification may in certain cases
be regarded as "acting forces" on other connected elements. Hence, to avoid non-
conservative results, the capacities should in certain cases account for effects such as:
- strain hardening of steel materials
- reduction of the characteristic strength due to heating, and variation of the temperature
with distance from the heat source
- increase of characteristic strength (yield, tensile strength) of the structural steel material
due to the effect of loading rate
Strain hardening may give 30% increase in ultimate strength compared to yield strength for
typical offshore structural steels. The plastic capacity of the elements taking part in the energy
absorption mechanism may for particular cases be increased with a factor 1.3. Normally a
factor 1.0 is applied.
If fire occur prior to the explosion, the strength reduction due to heating should be accounted
for, ref. Chapter 6. In most cases, however, the fire is initiated by the explosion, which means
no significant change in material properties due to heat during the explosion event.
Tests of increased material strength due to short loading rate show a wide scatter of results.
2
An increase in yield strength in the order 170-200 N/mm is reported in /9/. Since the absolute
increases are found to be essentially the same for all structural steels, the percentage
increase is much larger for the low-strength steels. The strength increase due to loading rate
may for particular cases be increased with a factor up to 1.5. Normally a factor 1.0 is applied.

5.5.3 Dynamic amplification


The structure can be considered as a statically loaded system for load impulses with long rise
time and long duration compared to the natural period of the structural element in question,
see e.g. Figure 5.3, loading type a and b. In such cases the structural response is determined
by the maximum pressure amplitude without any dynamic amplification. The duration of the
pressure build-up and decrease from a deflagration will depend on the size of the explosion
(e.g. volume of gas cloud). As an indication on order of magnitude the duration of a load
impulse for deflagrations which might occur on platforms will be 0.1 - 0.2 seconds. Hence, the
natural period of vibration should be checked for each particular case to determine the
dynamic amplification.
The shape of pressure vs time for deflagrations may be taken as loading type a In Figure 5.3.
The shape of pressure vs time for detonations may be taken as loading type c in Figure 5.3
with duration (t1) in the range up to 0.02 seconds.
For assessment of the dynamic amplification of the response, the natural periods of vibration
for the actual structural elements are needed, see e.g. /10/ and /11/. Figure 5.3 gives
examples of dynamic amplification factors based on linear elastic theory for different loading
types.

FIGURE 5.3 – DYNAMIC AMPLIFICATION FACTOR FOR FOUR DIFFERENT TYPE OF


LOADINGS. T IS THE FIRST NATURAL PERIOD OF VIBRATION OF THE
STRUCTURAL ELEMENT /10/

In general the dynamic amplification factor depends on parameters such as :


- elastic versus plastic structural response
- type of loading
- natural period of the structure
- damping

5.5.4 Elasto-plastic response


The energy which can be absorbed elastically is only a fraction of the total energy absorption
capacity of a structure. To avoid unnecessary conservative result, inelastic capacity should be
in the design for explosion loads.
A measure of inelastic response is the ductility factor, µ, defined as the ratio between the
actual displacement and the displacement at first yield. The influence of the ductility factor on
the ratio required resistance to applied load, Rm/F1, is presented in Figure 5.4 on basis of a
single-degree-of-freedom system subjected to triangular load pulses with zero rise time.
FIGURE 5.4 – MAXIMUM RESPONSE OF ELASTO-PLASTIC ONE-DEGREE-OF-FREEDOM
SYSTEMS (UMDAMPED) DUE TO TRIANGULAR LOAD PULSES WITH ZERO
RISE TIME /13/

For structural elements with material and cross sectional properties allowing a ductility factor
µ = 3, the required resistance can be obtained by a quasi-static linear elastic analysis using a
dynamic amplification factor DAF = Rm/F1 = 1.2. The design resistance may be determined by
plastic theory.
The procedure cover all t1/T ratios, i.e. it is valid for both denotations and deflagrations. For
load pulses with rise time larger than zero (deflagration), it will give results to the conservative
side.

5.6 Examples
5.6.1 Structures exposed to explosions - General
The following list give examples of structural items that normally will be designed against
accidental explosion loads:
- Module Support frame (MSF):
MSF of plate girder type should be assessed for explosion loads resulting from confined,
partly confined or unconfined explosions. MSF of truss type with open areas between the
trusses will experience equal build-up pressure at both sides of the beam-columns, with
resulting lateral pressure equal to zero. Local lateral pressure on hollow sections such as
e.g. welded box girders should however be considered, see Section 5.42.
- Blow-out panels
Direct design should be applied to demonstrate maximum strength of blow-out panes in
compartments/modules intended as relief vent openings.
- Living quarters
Living quarters shall (according to NPD /3/) be located in non-hazardous area and safely
separated or adequately protected from the drilling and production area. In addition to
local strength of panels etc., sufficient strength and ductility of living quarter support due
to global explosion pressure should be demonstrated.
Structural elements and components may be allowed to experience local damages when
exposed to the accidental explosion event provided structural damage and failure will not
cause:
- loss of life
- significant pollution
- reduced evacuation possibilities
- increased probability of gas leakage, fire and explosion.

5.6.2 Explosion in a module


As a first approximation to find maximum explosion pressure in a module, Figure 5.2 may be
applied. The determination of build-up pressure ∆p is illustrated by the following example
(see also Section 5.4.4):
Example : Consider a module with L x B x H equal 20m x 10m x 5m.
Hence: V = 1000 m³
A = 10 x 5 m² = 50 m²

D = A = 7.1 m
L/D = 20 / 7.1 = 2.8 < 3
Parts of the walls are light blow-out panels. The blow-out panels cover an
area Av = 100 m². The static opening pressure for the blow-out panels is
calculated to be 0.1 bar (10 kN/m²). According to Section 5.4.4 the explosion
pressure built-up should at least be 2∆ps t a t, i.e. 0.2 bar.

FIGURE 5.5 – EXAMPLE ON PRESSURE BUILD-UP IN VENTED COMPARTMENT /7/


2/3
A/V = 1.0,

Which gives a pressure build-up according to Figure 5.5, i.e.:

∆p = 0.24 bar

Remarks:

1. The blow out panels will start to open for a pressure approximately 0.1 bar

2. Due consideration should be given to the direction of explosion venting, possibility of


flying objects, drag forces from vent flow, and the flame “gust” extending out of the
openings.

3. The other module wall shall be designed to resist the pressure ∆p, considering
recommendations for structural resistance and dynamic amplification as given in Section
5.5

5.6.3 Recommendations t reduce the build-up pressure


In addition to increasing the vented area In modules, the designer may obtain reduction of
build-up pressure by the following recommendations:
a) Vented areas/relief vents should be located close to possible sources for ignition, see
Figure 5.6. The sources for ignition may e.g. be:
- open flames
- hot surfaces
- electric mechanical sparks
- pressure waves
b) The sources for ignition should be positioned close to openings (relief walls), see
Figure 5.6. Further, the walls openings/relief walls should preferably be the longer wall of
a module.
c) Equipment and vent openings should as far as possible be arranged such that the burned
gas and flame front propagation moves in opposite directions, see Figure 5.7.
d) Equipment should e positioned away from the vent openings, see Figure 5.8.
e) Equipment should, if possible, be positioned such that the vent flow is parallel to the
longitudinal direction of the equipment. Further, repeated obstructions in the direction of
the flame front propagation should be avoided see Figure 5.8.
FIGURE 5.6 – ILLUSTRATION OF LOCATION OF VENT AREAS IN A ROOM /12/

FIGURE 5.7 – VENT OPENING POSITION /12/


FIGURE 5.8 – ILLUSTRATION OF LOCATIONS OF OBSTACLES (EQUIPMENT) IN RELATION
TO VENT LOCATIONS /12/

5.6.4 Explosion in the derrick area


Around the lower part of the derrick there are normally light walls, which may form a
compartment without roof. This area may be filled with an explosive gas-mixture. Due to lack
of roof and normally very light wall panels, his compartment may be regarded as close to
2
unconfined, resulting in a build-up pressure of approximately 0.1 bar(10 kN/m ). Such an
explosion will not affect the main load-bearing elements of the derrick and its connections.
The wall, however, are expected to fail. It is recommended to install such walls with one
strong boundary line connection (acting as a hinge), and with the other boundaries acting as
relief openings. This in order to reduce possible damages from flying objects and fragments.

5.7 References
/1/ NPD:
"Regulation concerning safety, etc. relating to the Act concerning petroleum activities
in the Norwegian Continental Shelf", Norwegian Petroleum Directorate, 1985.
/2/ NPD:
"Regulation for the structural design of load-bearing structures intended for
exploitation of petroleum resources", Norwegian Petroleum Directorate 1984.
/3/ NPD:
"Regulation for production- and auxiliary systems on production installations etc.",
Norwegian Petroleum Directorate, 1978.
/4/ NPD:
"Guidelines for safety evaluation of platform conceptual design", Norwegian
Petroleum Directorate, 1981.
/5/ NPD:
"Guidelines for area classification", Norwegian Petroleum Directorate, 1983.
/6/ NPD:
"Guidelines for the determination of loads and load effect", Norwegian Petroleum
Directorate, 1987.
/7/ Det norske Veritas:
TNA 101 Technical Note, "Design against accidental loads" (1981) (New issue RP
D24)
/8/ D.M. Solberg:
"Industrial as Explosion Problems", Plant/ Operations Progress Vol. 1, No. 4, (pp 243-
248), The American Institute of chemical Engineers (AIChE), October 1982.
/9/ J.M. Barsom:
"Material Consideration in Structural Steel Design." Engineering Journal, American
rd
Institute of Steel Construction, Inc. 3 Quarter, 1987/Volume 2, No. 3
/10/ P.G. Bergan el. al.:
"Svingning av konstruksjoner" (In Norwegian), TAPIR, Trondheim 1981.
/11/ R.B. Blevins:
"Formulas for Natural Frequency and Mode Shape", Van Nostrand Reinhold
Company Inc., New York, 1979.
/12/ J.A. Pappas:
"Gas Explosions. Venting of Large Scale Volumes", Veritas Paper no. 83-P028, Det
Norske Veritas, Oslo, 1983.
/13/ R.P. Kennedy,T.E. Blejwas, D.E. Bennett:
"Capacity of Nuclear Power Plant Structures to Resist Blast Loadings", Prepared for
US. Nuclear Regulatory Commission, NUREG/CR-2462, SAND83-1250, 4, RP,
September 1983.
6. DESIGN GUIDANCE FOR STRUCTURES EXPOSED TO FIRE

6.1 Introduction
6.1.1 General
The objective of this chapter is to provide guidance on how to obtain and document sufficient
strength of offshore steel structures exposed to accidental fire conditions. The primary aim is
to satisfy the requirement of the Norwegian Petroleum Directorate (NPD):
- "Regulations for structural design of load-bearing structures intended for exploitation of
petroleum resources", 1984 /1/.
For this purpose reference is also made to the following NPD Guidelines:
- "Guidelines or safety evaluation of platform conceptual design" 1981 /2/.
- "Guidelines for the determination of loads and load effects", 1987 /3/.
The general philosophy is that the structure shall be designed to preclude progressive
collapse when exposed to the design accidental events. he NPD guidelines do not explicitly
define any design accidental events related to fire, but /3/ does state that account should be
taken for fires caused by:
- burning blow-outs
- fires from leaks in risers or processing equipment
- burning oil on the sea
- fire in equipment or electrical installations
- fire on the helicopter deck
- fire in the living quarters.
It is also noted in /3/ that fires and explosions should be assumed to occur at the same time.
Reference is made to Chapter 5 for a discussion of explosions.
Section 6.2 presents design strategies, while Section 6.3 briefly discusses active and passive
fire protection methods.
The design of fire resistant structures includes two elements, namely the heat exposure
model and the structural model. Heat exposure considerations are discussed in Section 6.4,
while material and structural behaviour are presented in Section 6.5. References are listed in
Section 6.6.

6.1.2 Definitions
Jet fires: Fires driven by impulse of a gas or liquid jet released from process
equipment under high pressure.
Pool fires: Fires in a pool of liquid with no air restriction and the buoyancy forces as
driving forces of the fire process.
Fire balls: Fires as results of BLEVE - Boiling Liquid Expanding Vapour Explosion. The
fire is a ball of flames, burning as air is entrained in the gas/liquid mix.
6.2 Design strategy
6.2.1 General
There are three general approaches in the design against accidental loads (see Section 2.2),
viz:
- event control
- indirect design
- direct design

6.2.2 Event control


For accidental fires, event control may be categorized as systems or measures to:
- prevent the occurrence of a fire by, e.g., limiting the release of hydrocarbons or by
precluding sources of ignition when release has occurred.
- limit the extent and duration of a fire by, e.g., emergency shut down procedures and by
active detection and suppression systems.
The most effective event control comprises fire protection systems designed to interact rapidly
to minimize the leakage of hydrocarbons and to facilitate the early and rapid detection of such
leakages into hazardous areas. Well designed protective safety systems should maximize
personnel and equipment safety at all times.
Event control relies heavily on active systems (which may be mechanical, electrical,
electronic, etc.). Such active systems, by their very nature, are not completely reliable. Hence
redundant system design is generally needed.
Moreover, rational analysis of the effectiveness of event control must employ methods which
account for the reliability of the systems and components. Reference is made to Section 2.3
for a description of such methods.
The types and reliabilities of the event control systems are primary determinant of the severity
of the design accidental event.

6.2.3 Indirect design


Since fires and explosions should be considered to occur simultaneously /3/, indirect design
methods relevant for impact type loading should be implemented. These include using
connections with strength similar to or greater than the connected members, joints and
members capable of sustaining concentrated inelastic deformation, and materials with
sufficient ductility and fracture toughness.

6.2.4 Direct design


The principles for direct design against accidental loads from fires can be summarized as
follow:
1) Determine fire scenarios and acceptance criteria (location, extent and duration, heat flux,
evacuation time or time to extinguish the fire) (Section 2.3)
2) Determine the temperature of the structural elements, Ti, after fire duration equal to time
determined in step 1. (Section 6.4.5)
3) Determine material properties and characteristic resistance based on the temperature Ti,
(Section 6.5)
4) Determine the load effects. This step must include load effects due to thermal expansion
as well as change in load effects due to stiffness reduction.
5) Check that the structural elements have sufficient capacity to resist the load effects.
(Section 6.5)
6) If a structural element fails, determine if the surrounding structure can support the excess
load. If not, then progressive collapse is indicated and design revisions are required.
Possible revisions may include:
- use of event control/active protection measures to reduce the probability of the
considered accidental event.
- use of passive protection methods (insulation or water-filling of the member) or active
methods (e.g., water deluge) to reduce the temperature rise due to the design
accidental event.
- increase dimensions of the considered structural element, or of the surrounding
structure to provide greater redundancy.

6.2.5 The need for an interdisciplinary approach


The complete design against fire requires an inter-disciplinary approach. Table 6.1 gives
some examples of areas where design decisions or analyses are required and which
disciplines should carry out the design:

TABLE 6.1 – DESIGN DECISIONS/ANALYSES VS RESPONSIBLE DISCIPLINES


6.3 Protection system
6.3.1 General
Unprotected steel structures are vulnerable to fires. Tests have shown that columns and
beams loaded to normal design limits will collapse when the steel temperature reaches about
500°C. In the case of unprotected columns or beams, collapse can occur after only
10-15 minutes exposure to fire, ref. /4/. This is shorter than the required time for which escape
routes to shelter areas must be available, and is much shorter than the time needed for
evacuation of an off-shore platform. Hence, critical exposed steel members must be
protected. This protection can be provided by several approaches.
Protection systems are divided in two categories:
- active protection systems
- passive protection systems

6.3.2 Active protection systems


Active protection systems comprise detection and alarm systems, fire suppression systems
(water, foam and Halon) and ventilation systems. For further detail reference may be given to
/6/ and /7/.

6.3.3 Passive protection systems


Passive fire protection is the use of a protective barrier to prevent the immediate exposure of
the protected elements to the fire and to prevent or retard heat transfer into the element. A
wide range of products is available to form this barrier.
The main classifications of passive protection systems used offshore include the following /B/:
- Panels
- Organic Coatings
- Inorganic Coatings
An additional method of passive fire protection system which has been effectively employed
on offshore structures is the water-filling of the structural members themselves. Typical
applications are the deck leg tubulars and the MSF truss work members.
Unlike external water spray systems, the use of water-filled structures fully employs the
thermal properties of steel, water and steam. The system as a whole is therefore very efficient
both in minimizing the amount of water used to provide protection for a specified duration and
in maintaining the temperature below a safe limit.
There are two types of water-filled structure systems - the replenished and unreplenished
systems. The unreplenished system is one in which the water lost as steam during a fire is
not replaced. It is a simple, inexpensive system but provides only a limited fire-protection
period.
In the replenished system (which is actually an active protection system) the water is
replenished from a storage tank or water source. The duration of protection in this case can
be unlimited and it therefore offers a good solution to the problems of protracted fire
protection in offshore structures.
For either type of system, consideration must be given to the effects of explosion. Care
should be taken to ensure that loss of some structural elements does not render the water
system inoperable leaving primary structural elements unprotected.
6.4 Fire scenarios and characteristic heat loads
6.4.1 Classification of fires
It is necessary to divide fire into different situations or types of fire, and treat them separately.
This is due to the difference in "driving forces", and hence the methods of calculation. The
different fire situations are described in Table 6.2:

TABLE 6.2 – FIRE SITUATIONS

The factors needed to be calculated or estimated could be:


- heat release rates
- temperature within flames
- flame heights and flame extensions
- heat fluxes onto objects within and outside flame
- burning period
- time-temperature history of enclosures
- air supply need of fires
- smoke movement
6.4.2 Fires in the superstructure
Any of the fires described in the previous section may conceivably occur in the deck area or
modules of a platform which produces or processes hydrocarbons. The type and location of
the possible fires depends on the fuel, release rates, release location, dispersion rates and
ventilation /5/.
Calculation of the development of fires within a module is largely based on experimental data
from fires in compartments simulating rooms in onshore structures /9/. It is often not possible
and normally not necessary to carry out detailed analyses of the development of a
hydrocarbon fire.
Based on a defined accident scenario the fire characteristics must be determined.
Rough estimates of the fire conditions may often be adequate for assessing the effects of the
fire.
For structural members exposed to fires of more than a very brief duration or intensity,
passive protection is required. The type and thickness of the fire protection Is dependent on
the heat load. For a specific structural member It makes no difference if the member is
exposed to a "small" or "large" fire as long as the flame size is sufficient to surround the
member with an "optically-thick" flame (approx. 1-2 m for hydrocarbon fires). Thus a "small"
fire of sufficient duration can be the dimensioning case for the fire protection design.
Generally, collapse will not follow the loss of a single element. A fire critical to the main
support structure must be of sufficient size to encompass enough members to constitute a
collapse mechanism.

6.4.3 Fire at sea level


The size and duration of a pool fire on the sea is dependent on leakage rate, type of fuel and
combustion rate. For normal hydrocarbon liquids the combustion rate is approximately

and the size of a circular stabilized pool fire will be

The height of the flames can be determined by the Thomas’ equation /10/.

(6.2)
where L = flame height (m)
ρair = mass density of air; may be taken as 1.225 kg/m³ for dry air
g = gravity constant = 9.81 m/s²
Size and diameter as a function of release rate for stable circular pool fires are shown in
Figure 6.1
FIGURE 6.1 – SIZE OF FIRE VS HYDROCARBON RELEASE RATE

6.4.4 Relevant heat load for pool fires


Typical hydrocarbon pool fires will have temperatures in the range 1050-1150°C, and will emit
heat radiation as a black body.
For the assessment of the consequences of an accidental cool fire on the sea exposing
structural members to the fire, it is recommended that a constant temperature from the offset
of the fire is employed. This is considered to be realistic as such fires do not require
significant time to develop. Using a constant heat load will result in a conservative
assessment and will also simplify the calculations. Medium and large pool fires (D > 5 m) on
the sea should be selected as basis for the calculations. Smaller fires will normally be of short
duration because:
- the fuel source is a moderate process leak which can be controlled
- the fire can be controlled by means of fire extinguishers and is therefore not considered a
threat to practical structures
The characteristic heat loads are obtained from the safety analysis (see Section 2.3), and are
based on the following parameters:
- Temperature T = 1150°C
- Radioactive heat load as black body
- Convective heat load typical for highly turbulent gases
- Exposure on all sides of structural member (fire can occur anywhere and size is
significantly larger than typical structure member).
Some indicating figures are given in Table 6.3:
TABLE 6.3 – INDICATIVE FIGURES OF FIRE LOADS

6.4.5 Calculation of the thermal response


Analytical solutions to the heat transfer problem are not tractable for any but the simplest of
geometries. Realistic cases (multiple materials, e.g. an insulated and/or water-filled member,
non-linear boundary conditions such as radiation, etc.) require numerical integration or
drastically simplified analytical solutions. Many practical cases can be reasonably modeled as
one-dimensional heat flow problems. The finite difference method is well suited for the
solution of such cases.
There are a number of computer program available for the calculation of the temperature
gradient and its variation with time, see e.g. /11/ and /16/.

6.5 Structural Response At Elevated Temperature


6.5.1 Material properties
The material properties which are relevant are the yield stress and modulus of elasticity for
steel.
Based on numerous tests /12/ the following relationships have proposed for the yield stress
and modulus of elasticity, respectively :

  T  
−1

f y (T ) = 1 + T •  767 • ln   • fy (6.3)
  1750  

E(T) = (1 – 17.2 • 10 T + 11.8 • 10 T – 34.5 • 10 T + 15.9 • 10 T ) • E (6.4)


-12 4 -9 3 -7 2 -5

Where T = temperature in °C
fy = yield stress at 20°C
E = modulus of elasticity at 20°
The equations are applicable for T < 600°C. The decrease in E and fy with increasing
temperature are shown in Figure 6.2. The above relationships have been adopted in the Norwegian
Standard for Steel Structures, NS 3472 /13/ and in NS 3478 “Design Rules for Structural Members for
Fire Resistance” /14/.
FIGURE 6.2 – VARIATION IN E AND fy WITH TEMPERATURE

6.5.2 Thermal Expansion


Like most other materials, steel and concrete expand with increasing temperature. The
following coefficients of thermal expansion, α, may generally be applied:
α = 1.2 • 10 /°C
-5
Steel:
α = 1.0 • 10 /°C
-5
Concrete
It is generally assumed that the difference in the coefficient of thermal expansion for concrete
and steel is small enough such that no differential stresses result in a composite member
when that member has a uniform temperature through the cross-section. This is valid whether
the concrete is present structural material or whether the concrete is used solely as a fire
proofing for the encased steel member.
Due to the thermal expansion, an end restrained member exposed to fire experiences
increasing compressive axial loads. It is generally difficult to assess these stresses due to
uncertainties in thermal gradient and in the end conditions.
It should be noted that quite small temperature increases will result in excessive stresses for
an isolated member with fixed ends: for steel with fy = 350 MPa, a temperature rise of only
about 140°C will result in compressive stresses equal to the yield stress. However, several
mitigating factors are normally present:
- the considered member is rarely fully restrained against axial expansion
- the end-restraining members are generally also exposed to some temperature rise and
thus expand, thereby reducing the effective end restraint.
- stresses resulting from thermal expansion are self-equilibrating and hence redundant
structures may remain capable of resisting external loads, but possibly by an alternative
load path (this capability is diminished if members in the structure have steeply degrading
post-buckling capacities).
6.5.3 Beam-columns
An estimate of the critical load for a centrally loaded column, subjected to a uniform elevated
temperature along its length, is given by the ECCS column curve where the reduced
slenderness ratio is modified to account for the change in material properties /15/:

and fy(T) and E(T) are taken according to Eqs. 6.3 and 6.4.
Since both E and fy decreases in a similar fashion, NS 3472 allows a simplification where the
reduced slenderness can be based on that at normal temperature (20°C). This simplification
gives a slightly conservative estimate (up to 12% too high) for temperatures below 550°C, but
is unconservative at 600°C. The variation of the reduced slenderness with temperature is
shown in Figure 6.3.
Due to a lack of experimental data for the buckling of columns exposed to fire conditions,
researchers /12/ have recommended that only ECCS column curve C should be used to
check capacity at elevated temperatures. This is believed to be conservative assumption for
tubular cross sections. However, use of curve C is required for fire conditions by NS 3472
/13/.
Curve C is given by:

(6.6)
where

using Eqs. 6.3 and 6.4 for the variation of fy and E with temperature, it is a straight forward
task to develop a column strength curve for any temperature. Curves for temperatures
ranging from 20°C to 550°C are shown in Figure 6.4. Note that the curves shown in the figure
take account of the change in the reduced slenderness ratio with temperature.
FIGURE 6.3 – COLUMN REDUCED SLENDERNESS, , VARIATION WITH TEMPERATURE

For a beam-column subjected to axial force and bending moments, the capacity may be
checked by the following interaction equation:
N M 1
+ • ≤1 (6.7)
Nk fy(T ) • W N
1−
NE
where N = axial force
Nk = fk(T)A = characteristic buckling force

π 2E(T )A
NE = = Eular buckling force
λ2
M = bending moment
W = sectional modulus
A = cross sectional area
Alternatively interaction formulae in NS 3472 /13/ may be used. It is generally acceptable to
use plastic moment capacity provided the cross section complies to compact section
requirements, see NS 3472.
FIGURE 6.4 – COLUMN BUCKLING CURVES

6.5.4 Plates
NS 3478 /14/ specifies that the capacity of steel plates are to be based on NS 3472 /13/ with
the reduced values for the yield strength and modulus of elasticity as given by Eqs. 6.3 and
6.4.

6.6 References
/1/ NPD:
"Regulations for structural design of load-bearing structures intended for exploitation
of petroleum resources". 1984
/2/ NPD:
"Guidelines for safety evaluation of platform conceptual design". 1981.
/3/ NPD:
"Guidelines for the determination of loads and load effects". 1987
/4/ Blake, A.C.:
"Structural Protection and Fine Protection Claddings" in 'Development in Fire
Protection of Offshore Platforms', R.G. Gowar editor, Applied Science Publishers,
1978.
/5/ Andersen, R.S., et. al:
"Risk Analysis in Offshore Development Projects", SINTEF Report No. STF18-
A83503, 1983.
/6/ Goodwin, B.A.:
"An Outline of the Fire Risks on Offshore Installations" in Development in Fine
Protection of Offshore Platforms, op. cit.
/7/ Vervalin, C.H.:
"Fine Protection Manual for Hydrocarbon Processing Plants", Gulf Publishing Co.
1985.
/8/ Bone, D.:
"The Protection for Structural Integrity", Proc. of Certification and Safety of Offshore
Installation, London 1981.
/9/ Odegaard, E. and Foyn, T.E.:
"Ulykker og Ulykkelaster", Veritas Paper No. 84-P007, 1984 (in Norwegian).
/10/ Lees, F.P.:
"Loss Prevention in the Process Industries", Butteworth and Company Ltd. London,
1974.
/11/ Skramstad, E.:
"Analysis of Temperature Bulks in the Steady and Unsteady State". User's Manual for
NV 613, Veritas Report 74-46-M.
/12/ Janss, I.; and Minne R.:
"Buckling of Steel Columns in Fire Conditions", Fire Safety Journal, Vol. 4, 1981/82.
/13/ Norwegian Standard 3472:
nd
"Steel Structures - Design Rules", 2 Ed., 1984.
/14/ Norwegian Standard 3478:
"Design Rules for Structural Members for Fire Conditions", 1979.
/15/ Aasen, B:
"An Experimental Study on Steel Columns Behaviour of Elevated Temperatures", Div.
of Steel Structures, Univ. of Trondheim, 1985
/16/ Soreide, T.H. and Amdahl, J.:
"A Computer Program for Ultimate Strength Analysis of Frame Offshore Structures -
Theory Manual", SINTEF Report No. STF71 A86049, Trondheim, 1986.
6.7 Illustrative example
The following example considers a four-legged platform exposed to a pool fire on the sea.
-4
Assume that, for a annual probability level of 10 , it has been calculated that the fire will be
uncontrolled for a two hour period. The problem is to assess the capacity of the jacket leg
after the fire duration of 2 hours. It is assumed that loss of one leg implies collapse of
platform, i.e., no redistribution possible.
For a four-legged jacket, analyses have shown that axial deformation in one deck leg does
not alter the load distribution significantly. Further, it is assumed that the diagonals bracing
the deck leg are exposed to the same temperatures as the leg. Hence, for this example, there
is no increase in load due to thermal expansion.
The following leg geometry is considered:
Main leg :D = 1500 mm
:t = 35 mm
Insert pile :D = 1330 mm
:t = 65 mm
Groute thickness : tg = 50 mm
Length between braces : L = 20 m
Exposure to the pool fire quickly (perhaps 10-15 min) increases the temperature of the outer
tubular to well above 600°C, and capacity of the main leg is essentially lost. Due to the
insulative effects of the grout, the insert pile retains considerable load carrying capacity. The
problem is to determine if this capacity is sufficient to prevent progressive collapse. This is
checked by following the principles for direct design given in Section 6.2.4:
1) Required time equal to 2 hours (given in problem definition).
2) Determine the temperature of the insert pile. This is based on methods outlined in
Section 6.4.5. For the given geometry, output from NV 613 /11/ is plotted in Figure 6.5. At
time equal to 2.0 hours, the temperature of the insert pile is calculated to be 400°C.
3) Determine Nk from Figure 6.4.
Calculate slenderness of insert pile at normal temperature:
= 2.522 • 10 mm²
5
A
= 4.821 • 10 mm
10 4
I

i = I = 437 mm
A
From Figure 6.2:
fy (400°C) = .65(355) = 201 Mpa
E (400°C) = .83(210000) = 174000 Mpa
Hence

4) Assume that the applied loads in the accidental condition are:


N = 30 MN, M = 3.5 MNm
5) Interaction equation to be checked (Eq. 6.7) is:
N M 1
+ • ≤1 (6.7)
Nk fy(T ) • W N
1−
NE
Inserting above values:
30 3.5 1
+ • = 0.93 < 1
46 14.5 30
1−
207
Hence capacity is shown to be acceptable. Note that if the thickness of the grout had been
100 mm rather than 50 mm, the temperature after a two hour exposure to fire would have
been only about 150°C (see Figure 6.5). In this case the interaction ratio is calculated to be
only 0.65. This shows that the jacket leg resistance to buckling in fire conditions may be
improved significantly by increasing the grout thickness, and it indicates the typical
improvements which may be gained by applying insulative fire-proofing.
FIGURE 6.5 – EXAMPLE CALCULATION : MEMBER TEMPERATURE
7. DESIGN GUIDANCE FOR STRUCTURES EXPOSED TO EARTHQUAKE

7.1 Introduction
7.1.1 General
The objective of this chapter is to give guidance related to seismic design of offshore
structures in order to achieve a safe design against earthquake loads, for the platform as well
as for the topside modules and equipment.
The safety format and the design basis herein comply with the requirements and guidelines
issued by the Norwegian Petroleum Directorate, NPD /1/ and /2/.
The recommendations given by API RP2A /3/ regarding seismic design of piled steel jacket
platforms are referred to. The API recommendations are primarily given for the US. coast, but
the principles and procedures described in API are also applicable for seismic design in other
parts of the world. Therefore, the API RP2A can preferably be used as a basic reference
document in earthquake engineering of offshore structures.
The approach for evaluating the response to earthquake motion involves four main steps:
i) Specification of mudline motions.
ii) Soil-structure interaction analysis with use of soil stiffness and soil damping for
computation of the motions at the platform base.
iii) Earthquake response analysis and check of plat-form main structure.
iv) Analysis and check of platform components, i.e. modules, towers, flare and
equipment.
The experience from numerous earthquake response analyses of platforms and topside
structures in the North Sea indicate mostly linear elastic behaviour when the structure is
subjected to what NPD denotes abnormal earthquake excitations, i.e. the characteristic
-4
earthquake loads with an annual probability of exceedance equal 10 . Response analyses
are discussed in Sections 7.3 to 7.5.
For certain platform types and topside structures, analyses have also shown that limited
plastic behaviour may occur. In this context it is important to be aware of the conservatism
and uncertainties inherent in the design approach currently adopted in practice. The
responses from the so-called "response spectrum" analysis have to date been used as the
basis for code checking. As the sign and phase information is lost in this method, e.g. with the
consequence that the bending moment distribution along an element is unknown, the code
checking may be very conservative when accounting for permanent and variable functional
loads. It is therefore emphasized that utilization factors from an ordinary linear elastic
earthquake response spectrum analysis, accounting for static loads, must only be regarded
as a basis for further strength and ductility evaluations. With reference to the NPD
Regulations regarding limit state for progressive collapse (PLS), the plastic behaviour should
be focused rather than strengthening. Strength and ductility evaluation including code
checking are further discussed in Section 7.6.
With reference to above, main emphasis is given to the uncertainties, the assumptions, and
the conservatism associated with the current design methods, as well as to recommend
specific guidelines for performance of such analyses and for evaluation of the results.
Emphasis is also made to the plastic and the post-buckling capacity of structural details
which, in a qualitative manner, will ensure ductile behaviour.
The earthquake free-field pseudo-acceleration spectra developed for North Sea projects show
a marked peak in the period range 0.3-0.5 s. The element forces are directly dependent on
the product of the modeshape matrix, the modal loadfactor (participation factor), and the
pseudo-acceleration value. This product can conveniently be used to visualize that significant
dynamic amplification can be expected for short period structures, and that higher modes will
contribute significantly for conventional platforms. Hence, when considering the energy
content in earthquake loads and wave loads, correct simulation of the dynamic behavior will
be more important for earthquake response analyses. In general, a stiffer platform will be
more dynamic sensitive to earthquake than a flexible structure. Further, depending on the
filtering process through the platform and depending on the actual local platform modes (e.g.
cantilever effects), the topside structures as modules, flare boom and derrick may respond
dynamically to earthquake excitation.
As the Chapter is concerned with practical application, the theoretical background for dynamic
analyses in general, and soil-structure interaction analysis and earthquake response analysis
in particular, is not repeated in detail. For description of the methods and their theoretical
basis, references is made to relevant textbooks such as /14/, /15/ and /16/.

7.1.2 Earthquake hazard


Most of the world's earthquakes occur in narrow zones along tectonic plate boundaries. The
Norwegian continental shelf is well removed from the nearest plate boundary and is therefore
characterized by relatively low earthquake activity. Distinct zones of significant seismic activity
may, however, be identified, with the Norwegian vest coast being among the most active area
in Norway, see /4/ and /5/. Figure 7.1 shows that relatively large earthquakes (Richter
magnitude above 5) have occurred regularly in these areas over the last 500 years. For the
time period after 1950, when Instrumented coverage was available for the offshore areas, the
locations of these earthquakes indicate that the earthquake hazard offshore Norway is
significant. The earthquake hazard for the Norwegian offshore sector is also higher than for
those sectors belonging to the other North Sea countries.

FIGURE 7.1 – EARTHQUAKES WITH INTENSITY > 5 (RICHTER) IN DIFFERENT TIME


PERIODS, /5/

7.1.3 NPD regulations and guidelines


Earthquake motion is classified as an environmental load in the NPD "Regulation for
structural design of load-bearing structures intended for exploitation of petroleum resources",
/1/. Reference is given to Chapter 2 and references /1/ and /2/ regarding earthquake loads
and load combinations for the different limit states. Earthquake motion should be applied for
the following categories of limit states:
- Serviceability Limit States (SLS)
- Ultimate Limit States (ULS)
- Progressive collapse Limit States (PLS).
Table 7.1 gives guidance for determination of characteristic environmental loads. Earthquake
loads should not be combined vith other environmental loads such as wind, wave and current,
see ref. /2/.

TABLE 7.1 – CHARACTERISTIC ENVIRONMENTAL LOADS FOR DIFFERENT LIMIT STATES .


EARTHQUAKE ANALYSES

The PLS check for intact structure is covered in this Chapter. Reference is made to Chapter 8
regarding PLS check for damaged structure. The PLS check is to ensure that the platform has
adequate capacity to prevent collapse under a rare intense earthquake with a recurrence
Interval of 10 000 years. This is to prevent:
- loss of life
- significant pollution
- considerable economic losses
The SLS and ULS are not covered in this document.

7.1.4 API RP2A guidelines


API RP2A /3/ applies in general to all fixed platform types. Most of the recommendations are,
however, typical for piled steel jacket platforms. The principles and procedures given in API
are summarized below.
The API's seismic design recommendations are based on a two level design approach, see
Table 7.1. These are:
- Strength Requirements.
The platform is designed for a severe earthquake which has a reasonable likelihood of
not being exceeded during the platform life (typical return period hundreds of years,
Strength Level Earthquake SLE).
- Ductility Requirements.
The platform is then checked for a rare earthquake with a very low probability of
occurrence (typical return period thousands of years, Ductility Level Earthquake DLE).
The objective of the strength requirements is to prevent significant interruption of normal
platform operations after exposure to a relatively severe earthquake. Response spectrum
method or time history approach are normally applied.
The objective of the ductility requirements Is to ensure that the platform has adequate
capacity to prevent total collapse under a rare intense earthquake. Member damage such as
in-elastic member yielding and member buckling are allowed to occur, but the structure
foundation system should remain stable under the loads imposed. This means that the
platform behaviour should be ductile under severe earthquakes, such that it absorbs the
imposed energy. The energy absorbed by the foundation is expected to be mostly dissipated
through non-linear behaviour of the soil.
For some typical jacket structures, both strength and ductility requirements are by API
considered satisfied if the below listed provisions are implemented in the strength design of
these platforms:
- Strength Requirements for strength level earthquake loads (SLE) are in general
documented
- Strength Requirements are documented for jacket legs, including enclosed piles, using 2
times the strength level earthquake loads (i.e. 2*SLE)
- rare, intense earthquake ground motion is less than 2 times the earthquake ground
motions applied for documentation of strength level requirements (i.e. DLE < 2*SLE)
- geometrical and ultimate strength requirements for primary members and their
connections as given in API are satisfied. These requirements concern number of legs,
jacket foundation system, diagonal bracing configuration in vertical frames
(see Section 7.7.2), horizontal members, slenderness and diameter/thickness ratio of
diagonal bracings, and tubular joint capacities. Further reference is made to /3/ and /9/.

TABLE 7.2 – EARTHQUAKE DESIGN PHILISOPHY, API RP2A

By comparing the NPD and API design philosophy for earthquake loading the NPD
requirements for Progressive collapse Limit State (PLS) earthquake can be considered to
correspond to the API Ductility Level Earthquake (DLE).
If the total API earthquake design philosophy (SLE and DLE) are used it may normally be
considered that operators own requirements and NPD other requirements (SLS, ULS) also
are satisfied.

7.1.5 Post-elastic structural behaviour


The capability of a structural system to resist seismic loads in the post-elastic range can in
many cases be assessed for by Introducing a behaviour factor q. This parameter takes into
account the energy dissipation capacity through ductile behaviour.
Hence the parameter q corresponds to the ratio between the earthquake excitation resulting
in a collapse of the structural system and the excitation leading to the attainment of the elastic
limit.
When applying this procedure, the acceleration values given by the response spectrum can
be divided by the behaviour factor q, and the specified structures or structural elements can
be with these reduced accelerations such that the elastic limit is attained.
Reference is given to Eurocode No. 8 /18/ and ECCS /19/ where these principles are
introduced. Similar principles are used in API RP2A /3/ for particular elements and
configuration of Jackets using the behaviour factor q = 2.0, see Section 7.1.4.
This procedure with use of the behaviour factor q = 2.0 is recommended to be applied for
Class Ib and Class II equipment (see Section 7.6.2) and module support frames for elements
satisfying provisions given in Section 7.6.4. Note that this procedure may imply two load
combinations for code checking as described in Section 7.6.5.

7.1.6 Structural configuration


For North Sea structures, the environmental loads waves, current and wind will combined with
gravity loads normally be decesive for the structural configuration. However, the principles
listed below can be of value regarding structural configuration.
Redundancy:
For structures to be located designed in earthquake active areas, redundancy should
in general be provided in the structure and foundation so that alternate load paths
may be developed in the event of individual member failures. Thus, failure of an
individual element should not greatly impact the ultimate strength.
Load-deflection characteristics of structural members are illustrated in Figure 7.2 with
brittle, buckling, plastic, and hardening failure modes. These effects must be
considered when assessing the structural strength and ductility of members and
further the structural redundancy.
It has been demonstrated that an "X" bracing configuration provides better
redundancy and post-failure behaviour than conventional "K" bracing or diagonal
bracing. When properly designed, the "XI' bracing configuration increases the
redundancy and thus improves the post-elastic (ductility) behaviour of the structure.
This bracing scheme does not necessarily require more steel or cost more than an
alternative design. Examples are steel jackets and module support frames of truss
type.

Uniformity in stiffness and strength:


Uniformity in stiffness and strength along the length of the structure should be
achieved to the extent possible. Abrupt changes in stiffness and strength result in
excessive concentration of stresses and Intensify the so called soft story effect.
Uniformity may be difficult to achieve, particularly in the transition zone between the
deck and the substructure, and for support of topside modules and equipment.
Special attention should be given to such zones.

Geometrical symmetry:
It is desirable to maintain symmetrical designs. Primary truss frames should utilize
symmetrical bracing patterns and member sizes. Eccentric mass and/or stiffness
tends to induce torsional loads which add to the shear loads. Torsional modes are
therefor important to account for in the analyses.
FIGURE 7.2 – TYPES OF MEMBER FAILURE MODES

7.1.7 Definitions
Ductility: Ratio between total (elastic + plastic) displacement and elastic
displacement. Ductility gives a measure of the ability to absorb
energy.
Epicenter: Projection of the hypocenter to the surface.
Earthquake: Shaking of the ground by different types of waves generated by
volcanic activity or tectonic movements. An earthquake is initiated
when the accumulated tectonic stresses at any one point in the
ground becomes greater than the strength at the point. Release of
stress at one point may increase the stresses nearby, and result in a
progressive rupture which may propagate for several hundred
kilometers.
Fourier spectra: Plot of amplitudes versus frequency of the harmonic motions used to
describe a given time history (Fourier transformation).
Free field motion: Earthquake motion unaffected by structures. When modeling the free
field, account should be taken of the dynamic and cyclic behavior of
the soils and of hysteretic and radiation energy dissipation.
Frequency domain: Any time-variable can be represented by a sum of harmonic notions
vith different frequencies, amplitudes and phase angles through a
Fourier transformation. In frequency domain analyses where the
responses are computed for each of the harmonic motions instead of
for the original time history, i.e. the variable "time" is exchanged with
the variable "frequency".
Ground motion: The vibration movement of the ground resulting from an earthquake.
Motion at any point is uniquely described in terms of either
acceleration, velocity, or displacement time history.
Hypocenter: The point where the earthquake started, also called focus.
Intensity: A measure of the ground shaking at a given site based on effects of
the earthquake such as how the earthquake was felt, damage to
structures, how people reacted, soil or rock slides, etc. Several
different intensity scales are presently available. The table below
illustrates the intensity (Richter):
Characteristic effects of shallow shocks Approximate
in populated areas magnitude
Damage nearly total 8.0
Great damage 7.4
Serious damage, rails bent 7.0-7.3
Considerable damage to buildings 6.2-6.9
Slight damage to buildings 5.5-6.1
Felt by all 4.9-5.4
Felt by many 4.3-4.8
Felt by some 3.5-4.2
Not felt but recorded 2.0-3.4
PLS: Progressive collapse Limit States, see Chapter 2.
Response spectrum: The response spectrum is defined as the maximum response to a
ground motion of a series of single degree of freedom oscillators
having different natural periods but the same degree of Internal
damping. The response may be maximum absolute acceleration,
relative velocity or relative displacement. (The response spectrum
method should not be confused with conventional spectral analysis
methods for stochastic response).
Seismicity: Same as earthquake activity, i.e. frequency of occurrence and
locations of earthquakes.
Time history: Time history is a continuous record over time of ground motion or
response
Tsunami: Long period water waves in the ocean caused by movement of the
sea floor or volcanic eruption. Tsunamis may travel with speeds of
several hundred kilometers per hour, and wave lengths may be
several kilometers. The wave height is, however, only a few meters in
the open ocean. It is when the tsunami breaks in coastal waters that
it may become destructive.

7.2 Design earthquake loads


7.2.1 Site specific study
For platforms sited in seismically active offshore regions, the first step in the design process is
to define the seismicity of the proposed site. In order to provide useful input to the process of
selecting earthquake design criteria, the evaluation of site-specific seismicity must lead to a
quantitative description of both the severity and the likelihood of occurrence of possible future
ground motions at the site. Furthermore, the description of seismicity must be extended to
characterize earthquake ground motions in a format which is suitable for a structural response
analysis. The characteristics of earthquake ground motions, which are most important to
structural response are the peak ground acceleration, the frequency content and the duration
of excitation. The frequency content is important because the response of a structure is
generally dictated by the amount of input energy at or near the frequency of its vibration
modes. In terms of damage potential, the response of a structure may depend not only on the
amplitude and frequency content but also on the cumulative number of shaking
The processes which govern the generation, transmission, and local modulation of
earthquake ground motions are such that significant differences in the severity of shaking can
occur at sites which are relatively close. Thus, evaluation of earthquake hazard potential is
often best treated as a site-specific problem. Site-specific studies should be particularly
considered for sites in areas of high seismicity or in any location where earthquake loading is
anticipated to significantly influence structural design.
The Commentary in API RP2A, /3/, has an extensive description of the recommended
framework for site-specific seismicity studies.
A site specific study provides free field (mudline) response spectra and corresponding
generated time histories for different levels of probability of exceedance.
Alternatively, such a study may recommend the application of real time histories measured in
other geographical areas, with similar seismotectonic and geological characteristics, i.e. the
peak ground acceleration and the frequency content. It is important to be aware that the
seismic motions at mudline are greatly influenced by the local filtering and may vary
drastically within a small area.
Note that site specific studies in the northern North Sea show that the horizontal earthquake
motion at mudline may be amplified by a factor 1.5 to 3.0, in the frequency range of 0.3 Hz to
2 Hz, with respect to rock outcrop motions.
Several earthquake response analyses if North Sea platforms indicate that the effect of soil-
structure interaction can be significant, both for fixed pile structure and, particularly for gravity
platforms.

7.2.2 Response spectrum


As a result of a site specific study, the response spectrum and the corresponding time
histories for the particular site are established for different probability levels.
The response spectrum given in Figure 7.3 may be applied for the Norwegian Continental
Shelf in early phases of the design for areas vith good soil conditions, that is on stiff clay and
medium to stiff sand or silt (/2/, /4/).
FIGURE 7.3 RESPONSE SPECTRUM
(normalized to a peak ground acceleration of 10 m/s²), 5% Damping. Maximum
acceleration, velocity and displacement is for curve:
-4 2
a) earthquake with annual probability of exceedance 10 , 30m/s , 2.0 m/s and 0.8 m
-2 2
b) earthquake with annual probability of exceedance 10 , 30m/s , 1.5 m/s and 0.5 m
The response spectrum can be used for locations with good soil conditions, but not
for locations with soft clay and medium to loose sand or silt.

The response spectra in Figure 7.3 are normalized with respect to a peak ground acceleration
of 10 m/s² at high frequencies, and the response spectra can be used together with the
accelerations given in Figure 7.4 in early phases of the design /2/. If the acceleration for the
actual location is e.g. 2.5 m/s² according to Figure 7.4, the response spectrum is obtained by
multiplying the normalized spectrum in Figure 7.3 with 0.25.
FIGURE 7.4 – SEISMIC ZONING MAP. THE BOLD LINE CONFINE THE BOUNDARIES WHERE
INFORMATION ABOUT SEISMIC IS ANALYSED BY RINGDAL ET AL (1982) /4/.

The spectrum is based on a total damping of 5% of critical, with energy loss to the soil
included. The spectrum has to be adjusted if the damping is different from this value. If other
values of modal damping are justified (n in percent of critical damping), the following factor D
may be used to multiply the response ordinates obtained from the response spectra
(Figure 7.3):
ln(100 / η)
D= (7.1)
ln( 20 )
The factor D is appropriate for values of damping between η = 2% and η = 10 %.
The spectrum can be scaled with regard to the peak ground acceleration (acceleration at high
frequencies) or with respect to the level of expected velocity at 1 Hz. The peak ground
accelerations can be selected according to Figure 7.4.
Regarding structural and hydrodynamic damping to be applied with soil-structure interaction
analysis, see Section 7.3.

7.2.3 Time history


As an alternative to the use of response spectra, the response of the structure can be
analysed for at least three sets of time histories which may be either artificially generated or
real (measured).
The mean value of the largest response from these time history analyses should be used as a
basis for the load calculations (/2/ and /3/). The time histories should have a representative
peak ground acceleration, magnitude and frequency content for the site.

7.2.4 Multicomponent earthquake excitation


The vector process describing earthquake-induced ground motions at a site is normally
resolved into component processes convenient for design purposes. The rotational
components are usually neglected, and the translational are given in two horizontal and one
vertical directions. In general, the translational components of the ground motions are
correlated processes. However, Penzien et al /6/ have shown that there exists a set of
orthogonal directions along which the components of ground motion may be considered
uncorrelated. The three translational earthquake components can, for practical purpose, be
assumed to act at the same time, and considered to be statistical independent. The
orthogonal horizontal component can be chosen equal to the horizontal component and the
vertical component 50 % of the horizontal component referred to bedrock.
Hence, the following combination of ground motion acceleration components may be applied:
- Main horizontal axis: 100%
- Orthogonal horizontal axis: 100%
- Vertical axis: 50%
The main horizontal axis should act along one principal axis in the structure.
NPD /2/ recommends that this approach is applied unless more accurate calculations are
performed. This approach require only one analysis.
In case a site specific study is performed the free field vertical and horizontal components are
normally generated separately. Note that if a soil-structure interaction analysis is
subsequently carried out, the vertical exitation will in general differ from 50 % of the main
component.

7.3 Response analysis of the substructure and the deck


7.3.1 General
Several recognized methods are available to evaluate the response of structures to
earthquake ground motions. The choice of method to be used is usually dictated by the type
of analysis being undertaken and the degree of precision appropriate for the available data.
Earthquake response analyses of North Sea plat-forms are normally performed by the
response spectrum method.
The main objective of the earthquake analysis is to document that the structure and
foundation have adequate strength and ductility for dissipation of the earthquake energy.
Simplified analysis methods are often suggested in building codes for onshore structures. In
Eurocode 8, ref. /18/, two groups of simplified methods are defined, i.e. "the simplified
response spectrum modal analysis" and "the quasi-static analysis". It is important to be aware
of that the application of these methods assumes that the fundamental frequency mode
dominates the response, and that the structure has a regular distribution of mass and
stiffness. Due to the complexity of an offshore platform, involving irregular mass and stiffness
distribution as well as important torsional and cantilever effects, higher modes will certainly be
important to account for. This is particular the case for topside main structures, which can be
excited by higher platform modes. Hence, simplified methods can not in general be
recommended for earthquake analysis of an offshore platform and its topside main structures,
such as modules, living quarters, derrick and flare boom.
The response spectrum method is derived from the mode superposition principle of solving
the dynamic equation of equilibrium. The dynamic solution is, in fact, partly available by the
existence of the response spectrum. The computational effort is limited to a free vibration
analysis followed by a relative simple post-processing. By this method, which applies only to
linear elastic models, the dynamic problem is solved in an approximate manner. Due to the
loss of the sign and the phase information, only an approximate estimate of the maximum
value of each response is computed. The method has found wide practical application in
earthquake engineering during the last decades. Reference is made to /16/ for description of
the theoretical background.
It is important to be aware of the following effects when code checking (post-processing)
results obtained from the response spectrum method:
- the sign of each stress resultant is lost
- only the maximum value of each stress resultant is computed (maximum values occur at
different positions in time)
- the shape of the bending moment diagram along the element is unknown
Consequently, the code check must assume the worst combination of dynamic element
forces/moments for each element as well as the most conservative combination of the
dynamic and the static element forces/moments. A systematic quantification of the
conservatism associated vith the final code check results for different type of structures is
presently not available in the literature.
The above analysis strategy, which is adopted for all North Sea installations (Norwegian
sector), has for the PLS checks often resulted in a slight exceedance of an elastic failure
mode criterion for a limited number of elements. The designers are, however, faced with the
problem to document proper behaviour during the abnormal earthquake excitation. For the
purpose of this documentation It is important to be aware of:
- the conservatism associated with the code checking
- that plastic behaviour is allowed, i.e. evaluate the post-yield and/or post-buckling capacity
- the very short duration of the peak in the ground motion response
- that the response will reduce in case of plastification, i.e. increased damping
With basis in the uncertainties and the conservatism associated with the response spectrum
approach, it appears attractive to apply time-history analysis (or frequency domain analysis)
of the platform and the topside main structures to keep control of the sign and the phase
information. Further, time history analysis can be applied to both linear elastic and non-linear
structural analyses. The main reasons for the continued use of the response spectrum
method for the structural response analyses of North Sea platforms can be related to:
- earthquake load effects leading to structural responses normally within the elastic range
- simplicity of the response spectrum method
- large uncertainties In definition of the excitation criteria
- time history analysis is expensive and requires significant data handling
- artificially generated time histories are often rather conservative and the application of
them has been subject to discussions
- in the case of real measured earthquakes are used, at least three sets of time histories
must be applied, i.e. three time histories in each translational direction
- impractical to specify time histories in design specifications, particularly for the topside
structures
- uncertainty in selection of damping values
- the problem of handling the frequency dependent soil damping and soil stiffness
(impedance) from a linear frequency domain soil-structure interaction analysis
This Chapter is therefore concerned with providing guidelines for application of the response
spectrum method and code checking based on results from this method. The following two
main items are focused:
1) guidelines for the performance of the response spectrum method analysis
2) guidelines for the combination its static loads and the code checking, and the evaluation
of the plastic capacity and the energy absorption capability
For the platform, i.e. the substructure and the deck, item 1 is discussed in Sections 7.3.2 and
7.3.3.
For the topside main structures, i.e. modules, living quarter, derrick and flare boom, item 1 is
discussed in Section 7.4
For the equipment and the secondary structures, item 1 is discussed in Section 7.5.
Equipment classification and item 2 are discussed in Section 7.6.
It has been common practice, both for piled structures and gravity structures, to perform soil-
structure interaction (SSI) analysis based on the free field excitation criteria developed
through a site specific study. The results from such an analysis serve as basis for establishing
excitation criteria, soil springs and soil damping. The main steps in the soil-structure
interaction analysis procedure adopted for North Sea installation are described in
Section 7.3.2.
In conceptual studies and for locations which satisfy certain soil property requirements (i.e.
stiff soil condition), the free field earthquake exciation criteria as proposed in the NPD
"Guidelines for determination of loads and load effects" /2/ can be adopted. Reference is also
made to Section 7.2.

7.3.2 Soil-structure interaction analysis (SSI)


According to the NPD Regulations, ref. /1/ sec. 4.1; "the effect of changes in excitation by
propagation through the soil, and by interaction between soil and structure, shall be
considered”. For North Sea projects this is today normally accounted for by a soil-structure
interaction (SSI) analysis. The theoretical basis for SSI-analyses is rather complex, and the
method used for North Sea installations is only briefly mentioned here.
The filtering process associated with the propagation through the combined soil/platform
system will change the frequency content of the free field spectrum, with the tendency to filter
out higher frequency components. The shape and the ordinates of the horizontal and vertical
response spectra at the base of foundation for gravity type platforms (at the top of the piles
for-jacket type platforms), will depend on the soil stiffness relative to the foundation stiffness.
This filtering process is illustrated in Figure 7.5 for a four-shafted gravity platform. The
horizontal and the vertical base of foundation spectra are plotted along with the corresponding
free field spectra. The base of foundation spectra are developed for both an upper and a
lower soil stiffness. It can be noted that the peak moves toward higher periods (compared to
the free field spectrum), and that the variation in the soil stiffness has a significant effect.

FIGURE 7.5 – BASE FOUNDATION SPECTRA PLOTTER ALONG WITH THE FREE FIELD
SPECTRA FOR AN UPPER AND LOWER SOIL STIFFNESS. (Example four-
shafted gravity platform).

a) Horizontal b) Vertical

A complete and true soil-structure interaction analysis involves theoretically the following
aspects:
- three dimensional representation of the structure and the soil
- linear modeling of the soil and representation of the soil as an unbounded medium
- non-linear three dimensional modeling of the structure with appropriate degree of
detailedness
An analysis method which fulfills all these requirements is not yet practical available, and is
normally not required.
Some SSI-analysis methods, referred to as "direct methods", are restricted to a two
dimensional representation of the soil-structure system, but allow to model the soil as a non-
linear medium. The "direct methods" involve a large number of degrees of freedom in the soil
region. For practical application the "direct method" is currently limited to two dimensional and
axisymmetric analyses.
The other type of SSI-analysis methods, referred to as "substructure methods", allows a three
dimensional representation of the soil-structure system with the soil as an unbounded
medium. The substructure method is restricted to linear behaviour of the soil. This method,
which is applied for a number of North Sea projects, treats the soil and the structure
separately. The soil-structure interaction analysis is linear and carried out in the frequency
domain. The non-linearity of the soil, i.e. the variation of the shear modules and the damping
with the strain level, is studied separately in a one-dimensional non-linear free field
deconvolution analysis. In this way appropriate dynamic soil characteristics for input to linear
frequency domain analysis are determined.
The procedure for linear soil-structure interaction analysis in the frequency domain
("substructure method") with subsequent structural response analysis and code checking is
indicated in the general flow diagram in Figure 7.6.
FIGURE 7.6 – GENERAL FLOW DIAGRAM FOR EARTHQUAKE RESPONSE ANALYSIS OF AN
OFFSHORE PLATFORM BASED ON A FREQUENCY DOMAIN SOIL-STRUCTURE
INTERACTION ANALYSIS (“SUBSTRUCTURE METHOD”)

Figures 7.7 and 7.8 illustrate the main steps associated with a soil-structure interaction
analysis for a piled platform and for a gravity type platform, respectively. These steps are:
Step 1A : The kinematics interaction analysis provides the consistent base of
foundation motions (or consistent motion at pile top). The motions are
consistent with the quantities computed in step 1B.
Step 1B : The computation of foundation impedances provides soil springs and soil
dampers, which are frequency dependent.
Step 2 : A three dimensional detailed dynamic model is established and a
conventional dynamic analysis is performed based on excitations computed
in Step 1A and with springs and dampers from Step 1B.
The analyst can carry out a frequency domain analysis (i.e. based on energy input spectra
and transfer functions), a time history analysis, or a response spectrum analysis. Although a
frequency domain analysis seems attractive due to the frequency dependent foundation
impedances, i.e. soil springs and soil dampers, the response spectrum method is normally
used. The computational errors that may be introduced when the frequency dependency is
neglected is examined in /20/.
The "substructure method" is attractive as the three dimensional FEM computer model of the
platform generated for other in-place design conditions can be directly adopted in the
earthquake response analysis. Further the method is attractive as the SSI-analysis and the
structural response analysis are two separate activities.
Due to the relative high fundamental period of offshore structures, the importance of higher
modes is significant. Further, the unsymmetry in the mass distribution and in the stiffness
demands a detailed model of the platform. In particular, reliable estimation of the response of
the topside main structures, which is often governed by local platform modes, may require the
inclusion of a large number of modes.
The theoretical background for the “substructure method” is given in ref. /7/. For a
comprehensive description of the soil structure interaction problem, reference is made to
ref. /8/.

FIGURE 7.7 – PROCEDURE FOR A SOIL-STRUCTURE INTERACTION ANALYSIS OF AN


OFFSHORE PILED STRUCTURE

FIGURE 7.8 – PROCEDURE FOR A SOIL-STRUCTURE INTERACTION ANALYSIS OF AN


OFFSHORE GRAVITY STRUCTURE
7.3.3 Response spectrum analysis
Step 2 in the "substructure method" for solving the soil-structure interaction problem, see
Figures 7.7 and 7.8, involves the structural response analysis of the three dimensional
platform model. This analysis can, based on the results from the SSI analysis, be performed
according to three alternative solution methods for the dynamic problem:
1) Response spectrum analysis
2) Time history analysis (modal analysis/ direct solution)
3) Direct frequency domain analysis
In practical design the response spectrum method will be used. This method, assumptions
and limitations, will be described and discussed in this section. Some illustrative practical
examples are given.
For the alternatives 2) and 3), which represent the "accurate" basic methods for solution of
the dynamic equation of equilibrium, reference is made to the textbooks, e.g. /16/, /21/, and
/22/.
The soil-structure interaction (SSI) analysis provides the horizontal and vertical response
spectra to be applied for the earthquake response analysis of the platform (i.e. substructure
and deck) and the topside main structures. The motions associated with these spectra are
often referred to as consistent motions, as they are consistent with the foundation
impedances, also computed in the SSI analysis. These impedances are complex values (due
to the frequency domain approach), and provides frequency dependent values for the soil
springs and the soil damping, see Section 7.3.2 .
The results, in terms of response spectra, from a soil-structure interaction analysis of a four
shafted concrete platform are presented in Figures 7.10 and 7.11. Floor spectra are given for
the positions indicated in Figure 7.9. Figures 7.10 and 7.11 show the development of the
earthquake excitation from the free field to selected locations for the horizontal and vertical
motions, respectively. The position of the range of important modes is also plotted in the
relevant spectrum for the platform itself and the topside structure in question. Response
spectra for both upper and lover soil stiffness estimates are shown.
In this example the effect of the Soil-structure interaction is significant, and the following
aspects are noticed:
- The peak (occurring at a period of approximately 0.3 s) in the free field horizontal and
vertical response spectra is moved towards higher periods.
- A significant increase in the acceleration spectrum ordinates is generally observed. This
effect will in particular influence the response of the topside main structures.
- The selection of soil stiffness will in this example influence the responses dramatically.
- The fundamental period of the platform is approximately 2 s. The base of foundation
spectra to be used for the platform response analysis suggests the importance of
including a large number of modes. It is seen that the 70 first modes will approximately
cover the peak area for both the horizontal and the vertical spectra. As higher platform
modes will excite topside main structures and equipment, effort should be made to ensure
that a sufficient number of modes is accounted for. In Figure 7.12 the positions of the 70
platform modes are indicated, for all three translational base of foundation spectra.
- Cantilever effects are important for topside structures, located at the ends of the deck
structure.
The floor spectra in Figures 7.10 and 7.11 are developed with a detailed three dimensional
model of the platform, and rather detailed models of the top-side main structures are
incorporated in the global platform model to account for Interaction effects. It is emphasized
that no general conclusion should be made from this example. The development of the
excitation throughout the platform and the deck area is closely dependent on the actual soil
conditions, the skirt and penetration depth, and the platform type. The example is included to
illustrate the complexity of the subject, and that large accelerations and structural responses
(element forces/moments) can be induced, particularly in the upper part of the platform.
FIGURE 7.9 – SELECTED LOCATIONS FOR CALCULATING RESPONSE SPECTRA (EXAMPLE)
FIGURE 7.10 – CHANGE IN HORIZONTAL EARTHQUAKE MOTIONS FROM THE FREE FIELD
TO DIFFERENT LOCATIONS IN THE PLATFORM (EXAMPLE).
The change is due to the filtering process associated with the propagation through
the combined soil/platform system. Floor spectra for upper and lower soil stiffness
estimates are indicated together with the first important period of motion modes.
FIGURE 7.11 – CHANGE IN VERTICAL EARTHQUAKE MOTIONS FROM THE FREE FIELD TO
DIFFERENT LOCATIONS IN THE PLATFORM (EXAMPLE).
The change is due to the filtering process associated with the propagation through
the combined soil/platform system. Floor spectra for upper and lower soil stiffness
estimates are indicated together with the first important period of motion modes.
FIGURE 7.12 – 70 FIRST PLATFORM MODES AND BASE OF FOUNDATION SPECTRA, UPPER
SOIL STIFFNESS (EXAMPLE)

It is important to be aware of the following aspects of the response spectrum method:


1) Only the maximum value of each response is estimated, see e.g. /16/, /21/, /22/.
2) The summation of the modal maxima must be performed for the actual response quantity
in question, e.g. element forces. Consequently, quasi-static analysis based on maximum
accelerations from a response spectrum analysis is not consistent with the theoretical
basis, and will give misleading results.
The importance of this is demonstrated and illustrated with an example in Section 7.7.4. This
example includes the concept of the response spectrum method. See also Section 7.4.3
regarding these aspects.
Hence the maximum total computed acceleration in each node will occur at different time
instances, and the deformation pattern associated with the application of the corresponding
inertia forces will not reflect an instantaneous deformation pattern. To ensure consistency with
the mode superposition principle, the response quantity wanted must take part in the
summation process.
To summarize, for a complex three-dimensional structure as an offshore platform and the
topside main structures, for which higher modes definitely will contribute significantly, the
maximum modal element forces associated with the deformation pattern in each mode must
be computed prior to combination.
The following guidelines are given for determination of the number of modes to include in the
free vibration analysis:
- the sum of the effective modal masses for each translational direction should not be taken
less than 90% of the total corresponding translational masses applied to the model
- the number of mode shapes must cover the peak area of the acceleration response
spectrum (This depend on the roughness of the model, the accuracy of the model, the
goal of the analysis etc., i.e. investigations are necessary in each case).
The number of modes necessary also depend on the response wanted, e.g. higher local
modes can be very important for topside structures located in cantilever part of the deck, and
higher vertical modes can be important for piles. Hence, with these aspects in mind, the
analyst should do some investigations to ensure that sufficient modes are included.
The selection of proper damping values is generally difficult, and depends on the following
aspects:
- structural damping
- hydrodynamic damping
- soil damping
- the modes will mobilize the soil (and the surrounding water) unequally
- different modes will govern different parts of the structure
One specific damping value will then apply to each mode.
The following strategy is suggested: The response spectrum analysis is basically related to
linear elastic problems, and no account should initially be taken for possible plastic behaviour.
On this basis 3% damping (for all modes) is suggested for the structure. Hydrodynamic
damping is explicitly neglected. The structural damping is then combined with the frequency
dependent soil damping computed in the soil-structure interaction analysis. The combined
damping in each mode is calculated based on the deformation work (elastic energy) applied
to the structure and the soil. The basic relation for calculation of modal damping for a
specified mode j is:

∑D • E i, j i, j
Dj = i
(7.2)
∑E i
i, j

where Di,j = damping ratio for element no. i in mode no. j.


Ei,j = deformation work applied to this element in mode no. j.
Both the real part (stiffness) and the imaginary part (damping) of the complex foundation
impedances from the SSI-analysis (see Section 7.3.2) are frequency dependent. Hence, the
selection of spring values to be adopted in the free vibration analysis must be evaluated. one
possible approach Is to select the spring values associated with the mode responsible for the
major contribution to the response wanted. It appears that the frequency dependent stiffness
of the foundation very often is well simulated for the governing modes by a combination of a
static spring and added soil mass. The following relation is then used :
K(ω) = Ks - ω • m
2
(7.3)
Where K(ω) : frequency dependent stiffness
Ks : static stiffness
ω : frequency
m ; added soil mass
Calculation of added soil mass sill thus be performed for each degree of freedom to obtain a
best fit with resulting stiffness in the frequency range of interest.
7.4 Response analysis of the topside main structures
7.4.1 General
In Section 7.3 alternative methods for earthquake response analysis were discussed. In
particular, the response spectrum method was addressed along with the "substructure
method" for solving the soil-structure interaction problem.
The examples in Section 7.3 demonstrate the complexity of an earthquake response analysis
of an offshore platform. Further, with reference to the acceleration response spectrum and the
natural frequencies of the platform (substructure and deck), significant dynamic effects of the
topside main structures can be expected, with major contribution from higher platform
frequency modes.
For the topside main structures (modules), e.g. living quarter, conventional modules, derrick,
and in particular flare boom, the importance of accounting for higher platform modes is
significant because of:
- global module modes correspond to higher platform modes
- cantilever effects
- interaction effects.
Experience from different response spectrum analyses of both jacket platforms and gravity
base concrete platforms, with detailed models of the topside main structures included,
indicate that more than 50 platform modes are normally necessary to include.
Figures 7.10 and 7.11 demonstrate clearly that the important module modes can be (and will
most probably be) located in the peak area of the acceleration floor response spectrum,
indicating significant dynamic response.
For proper estimation of the response of the topside main structures to earthquake excitation,
the platform with the topside main structures must be considered as one dynamic system,
Involving detailed dynamic models of these topside structures in the platform model.
The inclusion of a large number of platform modes is theoretically required for proper
representation of the topside response. As the accuracy of the computed higher modes will
reduce, the reliability of the results from a response spectrum analysis can be questioned.
Studies with systematic comparison with results from direct dynamic solutions, which confirm
the results from a response spectrum analysis, are not available in the literature.
Important aspects and input parameters, e.g. criteria for determining number of modes and
damping, are discussed in Section 7.3. The CQC-method (Complete Quadratic Combination)
for combining modal maxima should be used, particularly for topside main structures. See
Section 7.7.4 regarding the CQC-method.
For code checking and ductility evaluation, see Section 7.6. Use of the behaviour factor q as
described in Section 7.6.4 is recommended, when relevant.

7.4.2 Discussion of alternative analysis methods


Various methods have been adopted in North Sea projects for earthquake response analysis
and code checking of the topside main structures. With basis in a response spectrum analysis
the following approaches have been observed:
1) The mass of each module is concentrated in the support nodes, and the module stiffness
is not included (decoupling). A static analysis with code checking is subsequently carried
out with the detailed module model.
2) Simulation of each module as a "pyramid" in the global platform model with the top
located in the centre of gravity, and the total mass concentrated in this point. The stiffness
is adjusted to fit with the fundamental period of the modules. A static analysis with code
checking is subsequently carried out with the detailed module model using the
accelerations from the response spectrum analysis.
3) Detailed models of each topside main structure are included in the platform model.
Maximum translational accelerations are computed in each module node. A static
analysis with code checking is subsequently carried out with the detailed module model.
4) Detailed models of each main topside structure are included in the platform model.
Maximum translational support reactions are computed, and each component summed
according to a statistically method, e.g. the SRSS-method (Square Root of Sum of
Squares). Equivalent accelerations in the three translational directions are then
established. A static analysis with code checking is subsequently carried out with the
detailed module model.
5) Development of floor response spectra in each module support, and averaging these
spectra to provide excitation criteria for the module analysis. A response spectrum
analysis of the module is subsequently carried out, and element forces/moments from
permanent loads and live loads are combined with the dynamic element forces/moment
prior to code checking.
6) Response spectrum analysis of the complete platform model, with detailed dynamic
models of the modules. Element forces moments are computed directly prior to code
checking.
Due to reasons explained in Section 7.3 only methods 5) and 6) can be recommended.
Methods 1) to 4) are too approximate and will give misleading results. This is further
documented in the examples in Section 7.4.3.

7.4.3 Response spectrum analysis


This Section presents results from detailed response spectrum analyses of a conventional
module. The results relate to the methods 3), 4) and 5) in Section 7.4.2.
The horizontal and vertical floor design spectra for a conventional module are given as an
example in Figure 7.13, along with the corresponding free field spectra. It appears that the
acceleration spectrum ordinates have been significantly amplified due to the filtering process
from free field to the deck level. Further, the 20 first module modes are located in the peak of
the horizontal as well as in the vertical floor spectra. In Table 7.3 the 20 first periods of the
module are listed together with the corresponding modal loadfactors and the effective modal
masses. The total weight of this module is approximately 2500 tons. With respect to the
general dynamic behaviour, not considering the dynamic load, the sum of the modal masses
indicates that 20 modes will be sufficient. Figure 7.13, also confirm that 20 modes will be
sufficient with respect to the dynamic loading.
In Table 7.4 the horizontal and vertical spectrum cordinates are given along with the modal
loadfactors. The product of the modal loadfactor and the corresponding spectrum ordinate is
given, indicating that mode nos. 1, 2, 4, 5, 9, 10, 17, 18 and 19 may contribute significantly.
Note that it is the mode shapes that finally will determine the modal contributions, and that the
modal contribution will differ from element to element.
For the element having the highest utilization ratio from the response spectrum analysis the
modal maximum axial force for the 10 first modes is indicated in Figure 7.14, i.e. the relative
contribution to the axial force from each mode. For this element it appears that the module
modes nos. 1, 2, 4 and 9 are important to include. (Note that other examples may result in
significant contribution of higher modes.)
FIGURE 7.13 – FLOOR ACCELERATION SPECTRA AND FREE FIELD ACCELERATION
SPECTRA (EXAMPLE)

TABLE 7.3 – PERIODS, MODAL LOADFACTORS AND EFFECTIVE MODAL MASSES


(EXAMPLE)
TABLE 7.4 – PERIODS, HORIZONTAL AND VERTICAL INPUT SPECTRUM ORDINATES,
MODAL LOADFACTORS, AND PRODUCT OF MODAL LOADFACTORS AND
SPECTRUM ORDINATES (EXAMPLE)

FIGURE 7.14 – MODAL MAXIMUM AXIAL FORCE CONTRIBUTION FOR THE 10 FIRST MODES
(EXAMPLE, THE ELEMENT WITH HIGHEST UTILIZATION RATIO)

In Figures 7.15 and 7.16 utilization ratios for elements in the-longitudinal frames of the
module are presented for three different methods adopted in practice. The utilizations apply to
the combination of earthquake loads with gravity loads associated with permanent loads and
variable functional loads.
Method A: Response spectrum analysis based on floor spectra, with direct
computation of element forces/moments.This method correspond to
method 5) in Section 7.4.2.
Method B: Static analysis based on nodal translational accelerations from a
response spectrum analysis. This method correspond to method 3) in
Section 7.4.2.
Method C: Static analysis based on equivalent overall accelerations in the three
translational directions, with basis in the support reaction forces from
a response spectrum analysis. This method correspond to method 4)
in Section 7.4.2.
The results related to methods A, B and C are presented in Figures 7.15 and 7.16 for buckling
and yield respectively. The model is identical for all three cases. The response spectrum
analysis is identical with respect to number of modes, damping, modal summation, etc.
Similar post-processing (code check) is applied for all three cases.
By inspection of the results, and having in mind that method A is the "correct" one, method B
and C will provide misleading results. Methods B and C may give both "conservative" and
"non-conservative" results.
The main reason for these differences are related to the fact that the modal summation must
be applied to the response quantity in question (e.g. element forces/moments), which is
documented in Section 7.3.
FIGURE 7.15 – BUCKLING UTILIZATION FOR ELEMENTS IN A MODULE (EXAMPLE),
METHODS A, B, AND C
FIGURE 7.16 – YIELD UTILIZATION FOR ELEMENTS IN A MODULE (EXAMPLE), METHODS A,
B AND C.
7.5 Response analysis of the equipment and the secondary structures
7.5.1 General
In contrast to the procedure often used in determining the static response, a dynamic analysis
calculated in steps can not in general be recommended, calculating one part of the structure
after the other. As discussed in Sections 7.3 and 7.4, the substructure, the deck, and the
topside main structures should be modeled as one dynamic system with accurate
representation of the stiffness and the mass. However, light equipment and secondary
structures can often be decoupled, and then checked afterwards applying the maximum total
acceleration in a quasi-static approach. It is important to be aware of that light equipment may
respond dynamically when interacting with higher platform modes. For equipment which by
failing may cause a major hazard to the safety of personnel or to environment, closer
investigations are recommended. A simplified method, in which interaction effects are
accounted for in case of decoupling, is suggested and described In Section 7.5.3.
Guidelines for equipment classification is given In Section 7.6.2. Class Ia and Ib equipment
relates to the safety production and safety evacuation systems.
A reliable estimation of the response of Class Ia and Class Ib equipment requires a detailed
dynamic platform model. If only global platform results are to be determined, the dynamic
model can generally be restricted to represent few modes of low frequency. In contrast, a
mode of high frequency can significantly influence a local response if the mode shape exhibits
a large value at this point. It Is therefore important to distinguish between global and local
results, and the corresponding design analysis strategies.
For code checking and ductility evaluation, see Section 7.6. Use of the behaviour factor q as
described in Section 7.6.4 is recommended, when relevant.

7.5.2 Quasi-static analysis


The earthquake response of Class Ia and Class Ib equipment and light secondary structures
are normally calculated with use of a quasi-static analysis, based on accelerations computed
in the corresponding supports or in the neighboring nodes.
The response spectrum analysis of the dynamic platform model provides element forces and
accelerations in the substructure, in the deck, and in the topside main structures. Note the
importance of applying the modal summation procedure (CQC) directly on the response
quantity (i.e. element forces, accelerations etc.) in question, as described in Sections 7.3, 7.4,
and 7.7.4.
The accelerations will then provide a basis for design of light equipment. If the equipment item
is located e.g. at or near a module floor beam, proper response of the latter should be
estimated. This may in certain cases involve increased detailedness of the platform dynamic
model. Similarly, the response of components supporting e.g. risers and flares must be
properly estimated.
The required detailedness of the platform dynamic model must be determined based on an
evaluation of the consequences of a failure in the particular equipment items.

7.5.3 Simplified response spectrum method


Class Ia and Class Ib equipment (as described in Section 7.6.2) are normally not incorporated
in the global platform dynamic model. The response of such equipment to earthquake loading
is in practice determined based on computed acceleration values in supports or neighboring
nodes, see Section 7.5.2. This is a simple method which ignore some dynamic effects. For
equipment which by failure may cause a major hazard to personnel or environment, a more
sophisticated method may be applied. In ref. /24/ such a method is presented which efficiently
generates floor spectra including the effect of interaction.
The method provides various statistical measures of the equipment response given directly in
terms of the input ground response spectrum. The effect of Interaction between the
equipment and the structure can be significant if the equipment frequency coincides with one
of the structure frequencies. The interaction effects will increase with increasing equipment
mass.
Further, closely spaced modes will occur in the combined system. The response associated
with such modes will be highly correlated. Therefore, the correlation between closely spaced
frequencies can be important to account for. In ref. /24/ examples are given, demonstrating
remarkable agreement with a direct numerical analysis.
The method given in /24/ give guidance how to develop spectra which can be used directly by
the equipment analyst to design any equipment item having a prescribed mass, frequency
and damping. (Note that complete understanding and use of the method requires use of
ref. /24/ and knowledge in the dynamic theory, the theory of stochastic processes, and the
theory of the response spectrum method.)
The main steps are:
1) Define the dynamic properties of the platform (or module), i.e. stiffness, mass and
damping.
2) Define the equipment dynamic properties, i.e. one-degree of freedom system.
3) Identify the structural natural frequencies and the components of the mode shapes
associated with the attachment degrees of freedom.
4) Compute the spectral moments of the normal coordinates, in terms of the input response
spectrum.
5) Compute the spectral moments of response in terms of spectral moments of individual
normal coordinates.
6) Compute the equipment response quantities, i.e.
- the root mean square response
- the mean peak response
- the standard deviation of the peak response
- The mean peak response curves which include the effect of equipment-structure
interaction, can then be used directly by the equipment analyst.

7.6 Strength and ductility evaluation


7.6.1 General
Response analyses of substructure, deck, topside main structures, and equipment and
secondary structures are described, in Sections 7.3 to 7.5. The following Sections give
guidance how to perform strength and ductility evaluation of structural elements, based on
results from these analyses.
Topside structures/structural elements and equipment are classified according to their
importance in Section 7.6.2. A procedure for code checking structural elements is described
in Section 7.6.5. Emphasis is given to the uncertainties, the assumptions, and the
conservatism associated with the presented methods.
For structures satisfying particular configuration and strength, a simplified procedure for
documentation of both strength and ductility is presented in Section 7.6.4 with use of elastic
analysis and elastic theory.
Dynamic response of light equipment in structures often causes problems in piping, control
systems and electrical systems. Proper functioning of such important equipment on offshore
installations during and after an earthquake is not only necessary in order to minimize the
probability for catastrophic consequences, such as major fires and explosions, but also to
guarantee the adequate functioning of emergency or safety facilities that are necessary for
safe escape after an earthquake.
Topside equipment, piping and other appurtenances are only briefly considered in design
codes. API RP2A /3/ recommend that these should be designed and supported such that their
natural periods of vibration are substantially less than those of the significant platform modes.
It is good design practice to pay attention to designing equipment and piping supports,
welding details, and appurtenances to minimize the probability of damage to the drilling and
production facilities in case of an earthquake. Also, free-standing items such as drill pipes and
casings should be secured to reduce the likelihood of damage during an earthquake.
Main emphasis is given to design guidance for load-bearing structures, see Chapter 1. As
mentioned above and in Chapter 2, other components and equipment can also be of major
importance for safety evaluation considering the abnormal earthquake event. Such "non-
structural" components and equipment are only briefly mentioned in the following text.

7.6.2 Equipment classification


All systems, equipment and equipment supports which by failure may cause a major hazard
to personnel or major pollution of environment should be designed to withstand the abnormal
earthquake event. The selection of specific items or systems to be considered as important
equipment is part of the general risk analysis study of the platform system, topside facilities
etc., see Chapter 2. The equipment classification given below is taken from /5/, and may be
used as guideline for selection of important equipment.
The equipment is classified into two main classes, Class I and Class II, according to
importance, as follows:
Class Ia equipment: Safety production systems which by failing will cause a major hazard
to the safety of personnel or to environment.
Examples on typical items are:
- wellheads and conductors
- flow-lines, risers
- high-energy pressure vessels and piping
- control systems and components related to safe shutdown
- load-bearing structures/structural components directly supporting
such equipment, e.g. clamps for risers, module floor beams, and
main chords and supports of flare
Class Ia equipment should remain intact and under control during
and after the abnormal earthquake event.
Class Ib equipment:
Safety evacuation systems necessary for a safe evacuation of
personnel during emergency situations.
Examples on typical Items are:
- fire fighting systems
- emergency communications systems
- emergency power systems
- living quarters
- escape ways
- launching systems for lifeboats
- load-bearing structures/structural components supporting such
equipment
Class Ib equipment should remain operative after the abnormal
earthquake event.
Class II equipment:
Equipment that by failing/overturning may cause failure of Class I
equipment or cause a major hazard to personnel.
Examples on typical items are:
- modules
- derricks
- work over rigs
- flare boom (see also Class Ia)
Class II equipment should remain under control during the abnormal
earthquake event. The support fastenings should be properly
designed, and the connections should behave ductile.
It is common design practice to design Class Ia equipment limited to elastic response. This to
avoid excessive deformations which in turn may cause failure In interconnected equipment
(e.g. gas flow-lines, etc.).
Allowable deformations caused by structural plastification should be evaluated in each case
for Class Ib equipment such that the evacuation systems maintain its intended use. Hence
such an evaluation may reveal that particular components should be limited to elastic
response.
Class II equipment can normally be designed with use of plastic theory. Note that analyses in
many cases lead to responses within elastic range, and code checking following Section 7.6.5
is in many cases sufficient. See also Section 7.6.4 regarding use of elastic analysis to
document sufficient strength and ductility. For Class II equipment, ultimate strength and
ductility of supports and main load-bearing elements connected to the supports are of main
importance.
Equipment items that do not fall with Class I and Class II need normally not be designed to
resist the abnormal earthquake event.

7.6.3 Equipment substructure interaction


This sections describe the effect topside main structure can have on the dynamic response of
the substructure/deck. Guidelines are given whether topside structures should be considered
as coupled or decoupled.
On the other hand, dynamic response of the topside structures itself is of main importance.
For this reason, topside main structures should be included in the response analysis as
described in Sections 7.3 and 7.4, and the below listed provisions disregarded.
Parameters for determination whether the substructure and the equipment should be
considered as coupled or decoupled systems are the mass, the stiffness and the connectivity
of the equipment in relation to the same parameters for the substructure/deck. Numerical
criteria for decoupling are usually based on the ratios Rm and R f given as:
Equipment mass
Rm =
Mass of substructu re / deck
Fundamental frequency of equipment
Rf =
Frequency of dominant substructure motion
The dominant substructure motion frequencies should include at least 2 modes in each
horizontal direction and the first vertical mode.
The following criteria may be used as a basis in the evaluation if decoupling can be accepted:
a) If Rm < 0.01, decoupling may be accepted for any Rf. However, very light equipment may
respond dynamically when interacting with higher platform modes. If this is a Class I
equipment, closer investigations are recommended
b) If 0.01 ≤ Rm ≤ 0.1, decoupling may be acceptable if Rf ≤ 0.8 or Rf ≥ 1.25. However,
interaction effects with higher platform modes should not be ignored.
c) If Rm > 0.01, a model of the equipment should be included in the platform dynamic model.
d) It is strongly recommended that topside main structures such as modules, flare boom,
drilling structure, living quarter and other slender structures are incorporated in the
platform model.

7.6.4 Use of the behaviour factor q


A simplified approach to account for the energy dissipation capacity through ductile behaviour
is to apply the behaviour parameter q, using q = 2.0, and perform elastic response analysis.
These principles are used in Eurocode No. 8 /18/, ECCS /19/, and API RP2A /3/, see
Sections 7.1.4 and 7.1.5.
Hence the acceleration values given by the response spectrum are divided by 2.0, and the
earthquake responses from this reduced spectrum are combined with the static load effects
such as gravity, buoyancy and hydrostatic loading. Alternatively the load effects from PLS
earthquake excitation and static loads are combined with use of the factors 0.5 and 1.0
respectively. See Section 7.6.5 regarding combination of loads/load effects and code
checking.
This procedure is recommended to be applied for Class Ib and Class II equipment
(see Section 7.6.2) and module support frames (platform decks) to document sufficient
strength and ductility for steel truss structures according to NPD Requirements /1/. If this
procedure is applied for Class Ib equipment, it shall be demonstrated with evaluation of the
structural plastification that the evacuation systems maintain its intended use.
All the following provisions shall be satisfied to apply this procedure:
a) Sufficient elastic strength are documented applying linear elastic response analysis using
half earthquake ground motions (as specified for the NPD PLS check) combined with
static load effects.
b) Satisfy geometrical requirements according to principles given for jacket structures in API
RP2A section 2.3.6d.2 /3/. These requirements concern:
- truss configurations with 4x2 chord elements or more (as jacket type structure with 8
or more legs)
- diagonal bracing configuration ("K"-bracing is not used)
- horizontal/vertical members between chords are provided
- column slenderness of primary diagonal bracings is limited to 80
- cross section compactness, satisfying e.g. NS3472 /17/ cross section classes 1 or 2,
whichever relevant. Alternatively the local buckling of the cross section should be
prevented by meeting the "two times the response" requirement as given for chord
members in c) below
c) Sufficient elastic strength are documented (applying linear elastic response analysts)
for chord members, using two times the earthquake responses (forces, moments)
obtained in a) above combined with static load effects (i.e. earthquake response
spectrum not reduced for particular elements).
d) Horizontal/vertical members between chords are documented to have sufficient
compression capacity to support the redistribution of loads resulting from the buckling
of adjacent diagonal braces (using half earthquake ground motions combined with
static load effects as in a) above).
e) Joints for primary structural members should be sized-for similar strength as the
connecting members, i.e. either the tensile yield strength or the compressive buckling
strength of the members framing into the joint, as appropriate for the ultimate
behaviour of the structure. Bending capacity of the members should also be
considered.
f) Both material and load coefficients should be taken equal to 1.0.
Regarding item b) above, the provision with 4x2 chord elements are related to structural
ductility beyond the elastic range. Sufficient ductility has been demonstrated for an eight-
leg structure with "X"-bracing in the two-legged bents and diagonal bracings in opposite
direction in the four-legged bents (as indicated in Figure 7.18). Thus, a portion of the
shear at each level is carried in diagonal tension members regardless of the direction of
the induced-earthquake loading. The tension brace should not be dependent on the
stability of a companion compression brace. This is also considered satisfied for a
structure with 2x2 chord elements, with "X"-bracing in both two-legged bents.
Similar principle can be applied for e.g. deck structures with upper and lower chords as
the two-legged bents. Then the vertical members between the chords shall have sufficient
compression capacity to support the redistribution of loads resulting from buckling of
vertical "X"-bracing panel between chords.

7.6.5 Code checking


As described in Section 7.3 the response spectrum method is derived from the mode
superposition principle of solving the dynamic equation of equilibrium. Hence this method
applies only to linear elastic FEM computer models. It is important to be aware of the
following effects when code checking (post-processing) results obtained from the response
spectrum method:
- the sign of each stress resultant is lost
- only the maximum value of each stress resultant is computed (maximum values occur at
different times)
- the shape of the bending moment diagram along the element is unknown.
- use of CQC or equivalent procedures will not result in determination of reliable von Mises
stresses (see Section 7.7.4).
Consequently, a code check must assume the worst combination of dynamic element
forces/moments for each element as well as the most conservative combination of the
dynamic and the static element forces/moments. A systematic quantification of the
conservatism associated with this combination and the final code check results for different
type of structures is presently not available in the literature.
Note that the phase and sign information are maintained with use of time history analysis or
frequency domain analysis. The main reasons for the continued use of the response
spectrum method for the structural response analysis are given in Section 7.3.1, hence the
assumptions and limitations listed below are related to the use of this method.
In the following a procedure for code checking structural beam column elements according to
NPD /1/ and NS3472 /17/ is described. In general plastic section properties may be applied
according to principles in NS3472 (for cross section classes 1 and 2 in para 5.2).
The described code check is to be used as an approximate scanning as follows:
i) Elements satisfying the "yield" and "stability" checks are considered to fulfill
NPD/NS3472 requirements
ii) For elements not satisfying the "yield" and/or "stability" checks, further assessments
may be performed to document sufficient safety. Plastic theory may then be applied
according to principles in NS3472 for check of ultimate load bearing capacity.
Regarding ii) above, use of the behaviour factor q = 2.0 as described in Section 7.6.4 shall be
limited to elastic theory (analyses and capacities).
Combination static and dynamic loads:
- Yield check: The axial force from earthquake is given similar sign as the axial
force from static in order to obtain, in combination with bending, the
maximum equivalent stresses.
- Stability check: The axial force from earthquake is given "compression sign"
(normally minus) in order to obtain, in combination with bending,
maximum axial compression or minimum axial tension. Absolute
values of both static and earthquake bending moments are used in
the combination. Note that lateral buckling may be a possible failure
mode for beam-columns with axial tension and significant bending
moment.

Limitations:
- Yield and stability check of members are included. Punching check/check of nodes are
not described.

Non-conservative assumptions:
- Lateral buckling of open beam-column sections shall be prevented as assumed in the
design (code check). If the compression flange is not laterally supported, manual
calculations should be performed including effects of lateral buckling, ref. /17/. This may
be a non-conservative assumption if not properly accounted for.

Conservative assumptions:
- Evaluate if the considered member is part of NPD requirements regarding abnormal
earthquake event. The structure or structural element may in certain cases be allowed to
experience local damage, see /1/ sec. 2.4.
- Elastic FEM analysis is performed. Ultimate behaviour using elasto-plastic material
behaviour may in many cases be applied.
- Code checks based on elastic cross section properties are often applied. Plastic cross
section properties may in many cases be applied, in particular if the cross section satisfy
class 1 or class 2 according to NS3472 /17/.
- The analysis is based on a conventional response spectrum analysis, hence the phase
and sign information is lost. This normally implies:
i) Results from static loadcase with positive and negative signs in three positions
along the beam.
ii) Results from earthquake loadcase given with only positive signs in 3 position
along the beam. The moment distribution along the beam is unknown. The
responses (Fx, Qy, Qz, Mx, My, Mz) may not act similtaneously due to loss of
phase information.
iii) Results from i) and ii) above combined, with most conservative signs from
earthquake applied. Hence the combination is conservative, and in many cases
conservative. (Use of time history analysis exclude this conservatism).
iv) As the actual shape of the moment distribution along the beam-column is
unknown, the maximum bending moments Mmax (both axes) should be used in
the stability check formulas given in NS 3472 Table 5.4.1, /17/. Further manual
evaluations may be performed in each particular case to validate if these bending
moments may be taken less than Mmax. Note that Mmax shall be applied if lateral
buckling is a relevant failure mode.
7.7 Examples and guidelines for structural configuration
7.7.1 Structural configuration
Main principles of structural configuration and structural design for platforms subjected to
NPD abnormal earthquake event are given in Section 7.1.6 and below. The following
principles and examples can be used as guidance to achieve an economic and safe design
against earthquake loading:
a) The structure should be given a design which is documented to ascertain adequate
strength and ductility for dissipation of the earthquake energy. Certain criteria for
geometric and material ductility should be satisfied in order to be able to absorb sufficient
energy, see e.g. Sections 7.7.2 and 7.7.3. A simplified approach to ascertain adequate
strength and ductility is to apply the behaviour factor q as described in Sections 7.1.5 and
7.6.4 using half the response spectrum values.
b) Connections of primary members should have a strength at least to that of the parent
members framing into the joint, see e.g. API RP2A /3/ section 2.3.6e.
c) Pronounced weak sections and abrupt change in strength should be avoided. This is to
reduce local strain peaks which may cause structural failure.
d) Energy absorption in slender members and stiffeners with a non-ductile post-buckling
behaviour should be avoided. I.e. load transfer to bracings in tension should be utilized
when adjacent compression members buckle.
e) The materials in the structure should be ductile with sufficient fracture toughness
properties.
f) The ultimate tensile strength should be higher than the yield strength, such that rupture
does not occur at first yield even if the fracture toughness is high.
Effect of strain hardening in steel material should be considered in the design as
additional load effect in the structure for component adjacent to tensile members.
g) The onset of local buckling should be avoided by e.g. sufficiently small diameter/thickness
and breadth/thickness ratios in member cross sections.
h) High local strains in joints which may result in weak links should be avoided. High local
strain concentration in e.g. shear panels may, however, be advantageous in order to
control the dissipation of the earthquake energy. This principle is often used for design of
tall buildings in seismic areas. Possible development of plastic hinges should occur in
sections preselected by the designer with due consideration to possible redistribution of
force in the redundant structure.
i) For steel-plated structures the ductility may be documented through the post-buckling
behaviour of the structure. Stiffeners should be designed with compact cross sections to
avoid stiffener buckling.

7.7.2 Brace configuration in jackets


API RP2A /3/ guidelines include provision for configuring and proportioning members in
vertical frames in a typical jacket structure. These guidelines may also be applied in general
for e.g. deck structures with diagonal trusses.
The purpose is to provide for redistribution of the horizontal shear loads in the vertical frames
as buckling occurs in diagonal bracings, and to improve the post-buckling behaviour of the
structure.
Figure 7.17 shows examples of vertical frame configuration which do-not meet the guidelines.
Diagonal bracing in the vertical frames should be configured such that shear forces between
horizontal frames or in vertical runs are distributed approximately equally to both tension and
compression diagonal braces (see Section 7.7.1 item d), and that "K" bracings is not used
where the ability of a panel to transmit shear is lost if the compression brace buckles.
Horizontal members between all adjacent legs at horizontal framing levels in vertical frames
should be included. These frames should have sufficient compression capacity to support the
redistribution of loads resulting from the buckling of adjacent diagonal braces. Figure 7.18
show examples of vertical frame configurations meeting the guidelines.
FIGURE 7.17 – VERTICAL FRAME CONFIGURATIONS NOT MEETING GUIDELINES /3/

FIGURE 7.18 – VERTICAL FRAME CONFIGURATIONS MEETING GUIDELINES /3/

7.7.3 Tubular steel members


Tubular steel members are today widely used in offshore structures. Regarding design
against earthquake, information on inelastic behaviour and buckling performance are of
particular interest. /10/ presents a comprehensive state-of-the-art summary of the theory,
behaviour and design of tubular members as used in offshore installations.
Recommendation to slenderness and D/t ratio (diameter/thickness) of structural members
subjected to overload under cyclic loading is in API RP2A /3/ related to the ductility
requirements for the case when analysis is not required, see Section 7.1.4 and
Section 7.6.4 item b). The following is recommended:
- The slenderness ratio of primary diagonal bracing in vertical frames is limited to 80.
- Their ratio of diameter to thickness is limited to 36 (13100/Fy for Fy in MPa)
For members not satisfying these criteria, effects of possible reduced post-yield and post-
buckling behaviour should be considered.
Post-buckling behaviour, D/t-ratio
A fundamental requirement for satisfactory post-elastic behaviour is that the tubular member
should have sufficient rotational capacity before ovalization or local buckling to permit the
formation of plastic hinge. Figure 7.19 shows examples of inelastic behaviour of tubular with
different D/t ratios. Reference is given to /10/ and /11/.

FIGURE 7.19 – INELASTIC BEHAVIOUR OF TUBULAR SECTIONS WITH DIFFERENT D/t


RATIOS /10/

The figure indicates the moment curvature relationship for various D/t ratios. Members with a
D/t ratio of 36 or less were found to qualify as compact sections /11/, in the sense that they
possessed sufficient plastic hinge rotation capacity. Tubes with D/t of up to 48 also
demonstrated considerable ductility, and developed the theoretical limit load prior to failure,
although local buckling initiated earlier.
It should be kept in mind that the energy dissipation capacity of tubular members is
significantly reduced in the presence of sustained axial compression and/or external
hydrostatic pressure /10/, /12/.

Moment-thrust-curvature relations
Empirical moment-thrust-curvature (M-P-φ) curves for tubular beam-columns with small
slenderness are given in /10/ and /11/, see Figure 7.20. Non-linear aspects include early
yielding due to residual stresses, with a gradual transition to the fully plastic moment. With
continued plastic distortion, local buckling eventually leads to a loss of moment capacity, with
the critical curvature being a function of the D/t ratio.
FIGURE 7.20 – MOMENT-THRUST-CURVATURE RELATIONS FOR TUBE /10/

Beam-column buckling
Residual stresses and initial out-of straightness are responsible for realistic column design
curves, being lower than the ideal elasto-plastic. Figure 7.21 shows a typical column design
curve, and Figure 7.22 shows a typical interaction curve for eccentrically loaded columns with
different slendernesses /10/.

FIGURE 7.21 – ULTIMATE STRENGTH COLUMN CURVES


FIGURE 7.22 – INTERACTION CURVES FOR ECCENTRICALLY LOADED COLUMNS /10/
: perfect column
--------------- : with residual stresses and imperfections

Plastic mechanism
Tubular beam-columns (bracings) in offshore platforms are subject to bending moments
caused by lateral loads and frame effects as well as the axial loads of primary interest.
Assuming reasonably compact sections (say D/t ≤ 48) and small bending moments due to
frame effects, the ultimate capacity of such members, Nu, may be represented by the
following ultimate strength interaction formula:

1 M  πN 
= cos  u  (7.4)
N 1.69M y  2Nk 
1− u
PE
where
M = elastic bending moment

ql 2
M =
12
q = lateral load
My = elastic bending moment capacity
Mp
My = for tubular members
1.27
Mp = plastic bending moment capacity

qpl2
Mp =
16
qp = ultimate (mechanism) lateral load in the absence of axial load.
Nk = characteristic buckling strength, fixed ends (k=0.5)
- linear formula for simplicity:
Nk / Ny = 1 – 0.384 λ for λ < 3
- column buckling curves may alternatively be applied
Ny = axial yield capacity (when M = 0)
PE = Euler load with hinges at ends (dua to yielding, k = 1.0)
The above N-M interaction curves are shown in Figure 7.23. Secondary bending moments
due to frame deflection have little influence on the ultimate capacity of the beam0column,
although the corresponding secondary bending moments show up prominently in elastic
analysis and design procedures.

FIGURE 7.23 – ULTIMATE STRENGTH OF FIXED END TUBULAR STRUTS

Maximum load-bearing capacity for uniform loading is when


q = qp and Nu = 0
resulting in
16 ∗ 1.27
M= My = 1.69 M y
12
Hence, M/My in Figure 7.23 include both the following:
- increase in capacity due to bending moment redistribution when assuming plastic
mechanism compared to elastic capacity (factor approximately 16/12 = 1.33)
- Increase in capacity due to effects from plastic section modulus. (factor approximately
1.27 for tubular members)

Degrading post-buckling behaviour


Having defined the ultimate capacity of tubular beam-columns, it now remains to describe
their inelastic behaviour in the post-buckling range /10/, /13/. The degrading post-buckling and
hysteretic behaviour of a fixed tubular member is illustrated in Figure 7.24. The
comprehensive envelope in Figure 7.24 (curves 1 to 3) shows a typical post-buckling
behaviour with the reduced load-bearing capacity.
FIGURE 7.24 – EXAMPLE OF FIXED-ENDED COLUMN BEHAVIOUR UNDER ONE CYCLE OF
LOADING /10/

The principal non-linear behaviour of a cross-braced frame, such as a piled steel jacket
subjected to large horizontal displacements, is inelastic buckling and tensile yielding of the
bracing members followed by inelastic bending of the columns at the level of the damaged
bracing. After both braces at a given level have been stretch, they tend to go slack, resulting a
dog-bone type of hysteresis loop for the overall structure.
The overall structural response is strongly dependent on the non-linear buckling behaviour of
the diagonal braces which in turn is very sensitive to the flexibility of the end supports.
API RP2A /3/ recommend to improve the post-buckling behaviour of the frame by:
- bracing configuration as given in Section 7.7.2
- sufficient strength and ductility of member connections
- redistribution of loads to members in tension
- horizontal members with sufficient capacity to support the redistribution of loads.
Documentation of energy absorption in members with hysteretic behaviour as shown in
Figure 7.24 is an uncertain and complex process, which is recommended to be, avoided.

7.7.4 Response spectrum analysis


In Section 7.3.3 it is emphasized to be aware of the following aspects of the response
spectrum method:
- only the maximum value of each response is estimated
- the summation of the modal maxima must be performed for the actual response quantity
in question, e.g. element forces/moments
This is demonstrated and illustrated with an example in the following. It is convenient to focus
on the general formula for the “true” effective forces given in eq. (7.5). The development of
this formula is related to the mode superposition principle and the formulation of the
earthquake load, see text book, e.g. /16/.
L 
f ( t ) = mφ n ωn • Vn ( t ) (7.5)
Mn 
where f(t) : forces at the time instant t
m : three-dimensional mass matrix
φ : three-dimensional mode shape matrix, containing the important modes
Ln : modal participation factor for mode n
Mn : modal mass for mode n
Ln
: modal load factor
Mn
ωn : natural frequency for mode n
Vn(t) : Duhamel integral for mode n (see e.g. /16/)
For the cantilever beam in Figure 7.25, with the mass concentrated in three nodal points and
with only three translational degree of freedom considered, the effective forces acting on the
structure of the time t is then :

FIGURE 7.25 – CANTILEVER BEAM MODEL

According to the basis for the response spectrum method the following relationship exists:
Sa (ξ1,T1) = ω1V1(t, ξ1)max
Sa (ξ2,T2) = ω2V2(t, ξ2)max (7.7)
Sa (ξ3,T3) = ω3V3(t, ξ3)max
Where
Sa( ) : acceleration spectrum value for the modal period Ti and modal damping ξ i.
When inspecting the formula for the Duhamel integral, Vn( ), it appears that the displacement
spectrum is a “true” spectrum. The response spectrum method, as used in practice, assumes
that the maximum velocity and acceleration can be approximated by the ω-relationship
(harmonic response). Due to this approximation, the velocity and the acceleration spectra are
denoted pseudo-velocity and pseudo-acceleration spectra:
Sv ≈ ω Sd
Sa ≈ ω Sd
2
(7.8)
This approximation is proven to be acceptable for short period structure, i.e. below 1.0
seconds /8/. The validity of this approximation for high period structures, e.g. offshore
platforms for which higher modes may contribute significantly to the response is questioned
and is to date not properly documented.
As the maximum value of each of the three terms in eq. (7.6) will occur at different time
instances, ref. Vn(t), the relationship in eq. (7.7) can not be directly inserted in eq. (7.6). This
will be too conservative. Each of the three terms in eq. (7.6) must be considered separately,
and each term contains the “modal maxima” of the effective forces, i.e:

Note that the sign of the φ-terms and the sign of the participation factors. L are so far
maintained. However, the time and the phase information between the three term are lost. A
statistical method is therefore necessary for combination of the above three terms. The
SRSS-method, Square Root of the Sum of Square, has traditionally been used for two-
dimensional problems, i.e.:

The sign information is lost in this squaring process.


For three-dimensional problems, where mode shapes can be closely spaced and correlated,
the summation method called the “Complete Quadratic Combination” (CQC) should be used.
For typical response component, Uk, this summation procedure will read:

where Ui k : modal maximum of the response component k in mode i


ρi j : cross modal coefficient
Uj k : modal maximum of the response component k in mode j
In general, the cross modal coefficients, ρi j, are functions of the duration and the frequency
content of the loading, the modal frequencies, and the modal damping ratios. If the duration of
the earthquake is long compared to the periods of the structure, and if the earthquake
spectrum is smooth over a wide range of frequencies, the expression for the coefficients can
be approximated by,

where r = ωj / ωi
For constant damping in all modes, eq. (7.12) reduces to:

The CQC-method for combining modal maxima is based on fundamental theories of random
vibration. For development of the method, see /23/.

For the cantilever beam in Figure 7.25 the mode shapes are well separated, and ρi j, (with i
different from j) is therefore very small. As ρ1,1 = 1, ρ2,2 = 1, and ρ3,3 = 1, the formula
degenerates to the SRSS combination method as the six first terms vanish, see eq. (7.10).
For offshore platforms and in particular when analysing the topside main structure, the effect
of the cross modal coefficients must be accounted for.
With reference to eq. (7.5) the effective “true” nodal accelerations are given by:

Similarly to eq. (7.9) the maximum modal effective acceleration are:


Multiplying these accelerations with the corresponding modal masses gives the effective
modal maximum forces as in eq. (7.9), i.e. the sign information is maintained. The maximum
nodal accelerations can be computed according to the SRSS-method, i.e:

Again, the sign information is lost the squaring process, implying that the total inertia force
acting on the structure can not be computed based on these accelerations. This effect is
demonstrated in an example with figures below. See also Section 7.4.3.
As shown above the maximum total inertia force in each node is correctly estimated by
multiplying the nodal mass by the acceleration value from the response spectrum analysis.
However, as the sign information is lost in the squaring process, the shear forces (element
forces) along the beam and the base shear can not be estimated from the total nodal
acceleration value. To properly account for the sign of the modal inertia forces, i.e. the shape
of each mode, the approximate superposition procedure must be applied directly to the
response quantity in question.
The correct formula for the maximum base shear for the cantilever beam will, when summing
the shear forces per mode, read:

The importance of using the actual response quantity is demonstrated with figures below.
Similar notations as used above are applied. See also Figures 7.25 and 7.26.
From eq. (7.9):

From eq. (7.10):

From eq. (7.18):

Note that the maximum shear forces Qmax (or any shear forces) can not be computed by
adding f1, f2 and f3 (resulting in Q = 41.7).

FIGURE 7.26 – CANTILEVER BEAM MODEL, PSUEDO-ACCELERATION SPECTRUM, AND


FIRST THREE MODE SHAPES (EXAMPLE)

7.7.5 Articulated columns


Factors which should be considered in the design of mooring buoys include the long
fundamental period of vibration, and the forces on the universal joint and the cantilever beam
due to the vertical acceleration components.
The design forces on the universal joint from submerged weight, storm waves, wind etc. are
usually balanced to keep the forces on the joint to a minimum. Even relatively low levels of
earthquake ground shaking may therefore exceed the other design forces, see
Figure 7.27, /5/.
FIGURE 7.27 – VERTICAL FORCES ON THE UNIVERSAL JOINT (EXAMPLE)

7.7.6 Tension leg platforms


Tension leg platforms (Figure 7.28) are affected by the vertical component of the ground
shaking and possible also by seaquakes.
Possible "slack" in the tethers is one of the critical conditions for these elements under
earthquake excitation. Further the vertical ground shaking should be considered for design of
the anchor piles.
Due to the rapid rate of loading during earthquakes, the platform will not be able to follow the
vertical motion of the ground, and the tethers will be extended according to the movements of
the ground. Hence, the vertical motion of the sea bottom during earthquakes should be
considered, including effects of. dynamic amplification. This vertical motion may in the North
Sea be some few centimeters.
Linear elastic behaviour may be assumed, and the response spectrum method can normally
be applied.

FIGURE 7.28 – TENSION LEG PLATFORM


7.8 References

/1/ NPD:
"Regulations for structural design of loadbearing structures for exploitation of
petroleum resources", Norwegian Petroleum Directorate, 1984.
/2/ NPD:
"Guidelines for determination of loads and load effects", Norwegian Petroleum
Directorate, 1987.
/3/ API RP2A:
"Recommended Practice for Planning, Design and Constructing Fixed Offshore
th
Platforms", 17 Edition. American Petroleum Institute.
/4/ F. Ringdal et. al.:
"Earthquake Hazard Offshore, Norway. A study for NTNF Safety Offshore
Committee", Norsar Contribution No 302, Oslo, 1982.
/5/ Selnes, P.B., Ringdal, F., Jakobsen, B. Hansteen,
H. Hove, K.,: "Earthquake Risk and Design of Structures Offshore Norway." A study
in the NTNF "Safety Offshore Program", Oslo, Dec. 1983.
/6/ Penzien, J. and Watabe, M.:
"Characteristic 3-dimensional earthquake ground motion", Earthquake Engineering
and Structural Dynamics, Vol. 3, 1975, pp 365-374.
/7/ Kausel, E., Whitman, R.V., Morray, J.P. and Elsabee, F.:
"The Spring Method for Embeded Foundations", Nuclear Engineering and Design 48
(1978), 377-392.
/8/ Wolf, J.P.:
"Dynamic soil-structure interaction", Prentice Hall, 1985.
/9/ Kallaby, J., and Mitchell, W.W.:
"Guidelines for Design of Offshore Structures for Earthquake Environment",
Proceedings of the Second International Conference on Microzonation. San
Francisco. November-December 1978.
/10/ Chen, W.F. and Han, D.J.:
"Tubular Members in Offshore Structures", Pitman Publishing Inc., 1985.
/11/ Sherman, D.R. and Erzurumlu, H.:
"Ultimate Capacity of Tubular Beam-Columns", presented at the Aug. 23-26, 1976,
ASCE National Structural Engineering Conference, held at Madison, Wisc.
/12/ Marshall, P.W. et. al.:
"In-elastic Behaviour of Members and Structures", Combined Preprint for Session 45,
ASCE Annual Convention and Exposition, Committee on Tubular Structures, Preprint
3302, Chicago, Oct. 1978.
/13/ Marshall, P.W., Gates, W.E. and Anagnostopoulos, S.:
"In-elastic Dynamic Analysis of Tubular Offshore Structures", Paper No. OTC 2908,
Proceedings from the Ninth Offshore Technology Conference, Houston, Texas,
pp. 235-246, 1977.
/14/ Dowrick, D.S.:
"Earthquake resistant design. A manual for engineers and architects", John Wiley &
Sons Ltd., 1977.
/15/ Newmark, N.M. and Rosenblueth, E.:
"Fundamentals of Earthquake Engineering", Prentice-Hall, Inc., 1971.
/16/ Clough, R.W. and Penzien, J.:
"Dynamics of Structures", McGraw-Hill Kogakusha, Ltd., 1975.
/17/ NS3472:
"Structural design of steel structures", Norwegian Standards Association, June 1984.
/18/ Eurocode No. 8:
"Common unified Rules for Structures in Seismic Zones", Commission of the
European Countries, Draft report EUR 8850, 1984.
/19/ ECCS:
"Study on Design of Steel Building in Earthquake Zones", European Convention for
Constructional Steelworks, Publication No. 47, 1986.
/20/ Nadim, F. and To, P.:
"Soil-structure interaction of offshore structures; time domain versus frequency
domain analysis", NGI paper
/21/ Zienkiewicz, O.C., Lewis, R.W. and Stagg, K.G.:
"Numerical Methods in Offshore Engineering", John Wiley & Sons, Ltd. 1979.
/22/ Langen, I. and Sigbjornsson, R.:
"Dynamisk analyse av konstruksjoner", TAPIR 1979 (In Norwegian).
/23/ Der Kiureghian, A.:
"On the Response of Structures to Stationary Excitation", report no. UCB/EERC-
79/32. Earthquake Engineering Research Center, University of California, Berkeley,
1979.
/24/ Der Kiureghian, A., Sackman, J.L. and Nour-Omid, B.:
"Dynamic Response of Light Equipment in Structures", report no. UCB/EERC-81/05.
Earthquake Engineering Research Center, University, of California, Berkeley, April
1981
7.9 Illustrative examples
This example illustrates how to use the q-factor method, Section 7.6, on a module frame
exposed to an earthquake. The q-method is an fictitious method which use half earthquake
loads in an elastic analysis to document plastic strength.
The frame used is taken from a drilling module with a movable derrick. The derrick is located
in the most unfavourable position.
This example does not intend to illustrate the method of analysis
Input data:
Horizontal earthquake acceleration ax = 0.2 g
Vertical earthquake acceleration az = 0.3 g
Q-factor q = 2.0
Yield stress fy = 345 N/mm²

Load factor γf = 1.0

Material factor γm = 1.0

Loads :
Piping, 213.50 t
main deck 2.364 t/m, B.o.p. deck 1.576 t/m
Mechanical, 124.00 t
Main deck 0.7626 t/m, B.o.p. deck 0.7626 t/m
top deck 0.6471 t/m
Derrick, 1101.56 t
Piperack, 200.00 t
Top deck 0 to 44.9 m from, left side 4.454 t/m
Electrical, hvac, etc., 185.70 t
Main deck 1.1422 t/m, B.o.p. deck 1.1422 t/m,
Top deck 0.9704 t/m
Steel Weight 641.50 t
Total static load 2466.30 t
The drilling module frame is shown in Figure 7.29 with beam sizes and the derrick in a
extreme position.

FIGURE 7.29 – DRILLING MODULE FRAME USED IN THIS EXAMPLE.


To check the module frame design the provisions in Section 7.6.4 is followed. Combination of
static and dynamic loads is in accordance with Section 7.6.5. For yield check, the axial force
from the earthquake is given similar sign as the axial force from static loads in order to obtain,
in combination with bending, the maximum equivalent stresses. For stability check, the axial
force from earthquake is given "compression sign" in order to obtain, in combination with
bending, maximum axial compression or minimum axial tension. Absolute values of both
static and earthquake bending moments are used in the combination. Elements satisfying the
"yield" and "stability" checks are considered to fulfil NPD/NS3472 requirements.
a) Elastic strength
Sufficient elastic strength are documented using half earthquake ground motions
combined with static load effects. The results is shown in Figures 7.30 and 7.31.

FIGURE 7.30 – USAGE FACTORS FOR BUCKLING ANALYSIS

FIGURE 7.31 – USAGE FACTORS, COMBINED AXIAL AND BENDING STRESS.


USAGE FACTOR = ACTUAL STRESS/YIELD STRESS.

b) Geometrical requirements
- The module is made up of 2x2 chord elements with x-bracings in both 2 legged bents.
This configuration provides that shearforce is carried in diagonal tension members
independent of the direction of the earthquake loading. The tension brace is then not
dependent on the stability of a companion compression brace.
- Vertical members between chords are provided. The vertical chords shall support the
redistribution of loads resulting from buckling of vertical "x"-bracing panel between
chords.
- To achieve that global buckling is prevented before the element reach yield stress the
maximum column slenderness of primary diagonal bracings is 80. For the module
frame maximum column slenderness for primary diagonal bracing is:
Maximum column slenderness = L k / i = 64 < 80
- If the cross section compactness satisfies NS3472 class 1 or 2 the member should
have sufficient capacity before local buckling to permit the forming of plastic hinge.
The module frame in this example satisfies NS3472 class 1:

c) Chord members.
Sufficient elastic strength are documented for the chord members suing two times the
earthquake load used in a), i.e. q = 1.0, see Figure 7.32.

FIGURE 7.32 – USAGE FACTORS, COMBINED AXIAL AND BENDING STRESS.


USAGE FACTOR = ACTUAL STRESS/YIELD STRESS

d) Vertical members between chords.


Vertical members between chords shall have sufficient capacity to support the
redistribution of loads resulting from the buckling of adjacent diagonal braces. Half
earthquake load as used in a) is applied. The result of the elastic analysis is displayed in
Figure 7.33 and 7.34.

FIGURE 7.33 – USAGE FACTOR FOR BUCKLING ANALYSIS


FIGURE 7.34 – USAGE FACTORS, COMBINED AXIAL AND BENDING STRESS
USAGE FACTOR = ACTUAL STRESS/YIELD STRESS

e) Joints.
The philosophy of the design method is that vertical and diagonal members between the
chords shall have sufficient capacity to support the redistribution of loads resulting from
buckling of vertical x-bracing panel between chords. The joints must consequently be
designed to resist such loads.
The joint design criteria will be directly related to the force that gives yield for tension
diagonal, horizontal and vertical member and the buckling force for the bracing in
compression, i.e. the joint should have at least the strength of connecting members.
The design procedure will begin with a collapse mechanism analysis, to determine the
tension and compression members. Then, based on these forces the detailed calculation
of the joint is carried out.
8. DESIGN GUIDANCE FOR RESIDUAL STRENGTH ANALYSIS
8.1 Introduction
8.1.1 General
The purpose of the present section is to give guidance on residual strength analysis of
structures with damages caused by direct action of accidental loads.The damages, estimated
by the methods described in Section 3 - 7 are typically characterized by local dents,
permanent lateral distortion of member axis, over-all buckling, cracks, change of material
properties and loss of entire members.
According to NPD /1/ the structure in damaged condition shall resist environmental loads with
a specified return period. Local nonlinear behaviour is accepted, provided that a global
collapse mechanism is not initiated. Hence, the residual strength assessment is strictly a
matter of nonlinear systems analysis. At the present state of the art, however, models for the
nonlinear behaviour of damaged members are often not available. Accordingly, linear
systems analysis is applied combined with simple formulas for residual strength of damaged
elements. This approach is normally conservative as elasto-plastic strength reserves and load
redistribution effects are not utilized.

FIGURE 8.1 – FORCE DIDAGRAM FOR DENTED TUBE /2/

These considerations are further illustrated in Figure 8.1, which shows the idealized
behaviour of a dented tube, see /2/. The first yield approach (see Section 8.3.1) yield
compressive capacity Ny, whereas the ultimate strength is reached at Nu. Beyond this level,
the capacity drops obeying the plastic interaction curve, possibly with further reduction due to
local buckling.
In spite of the strength degradation, the global strength of an indeterminate structure may
increase due to load shedding to uncollapsed members. In a “true” nonlinear analysis such
effects should be included.

FIGURE 8.2 – IDEAL ELASTIC-RIGID PLASTIC ANALYSIS OF COMPRESSION MEMBERS /3/


The approach illustrated in Figure 8.2 is often useful with respect to residual strength
analysis, especially of beam-columns /3/. The total response is considered to consist of two
phases, the initial elastic deformation of a column with initial inperfection, (curve O’ E) and the
rigid plastic solution corresponding to a yield hinge at midspan (curves CD). The actual
response follows closely these curves except in the vicinity of the ulti-mate load, where the
behaviour is considerably more complex. However, in this region the actual curve may be
sketched approximately by “engineering judge-ment”. Besides, the bifurcation load between
the elastic and the rigid-plastic solution differs generally little from the actual ultimate load in
case where the post-collapse drop in load-carrying capacity is slow or moderately steep
(Curves CD1, CD2). This is typically the case for beam-columns with initial imperfections.
However, slender shell structures and stiffeners with very small imperfections may exhibit a
very steep unloading after ultimate load. They behave in a "brittle" manner and are very
imperfection sensitive so that bifurcation point should not be used as a measure of the
residual strength.
Alternatively, the ultimate load may be estimated by a first yield criterion. This is normally
more conservative and is used in Section 8.3.
For unstiffened tubular (Section 8.3.1) and stiffened plates (Section 8.3.4) guidance is given
also on the elastic and rigid-plastic solution. Apart from determining the bifurcation point, this
information may be used in an extended linear systems analysis by means of the procedure
described in Section 8.4.2.
The guidance below concerns the capacity of damaged members against a single overload. It
should be observed, however, that several damages may also include "micro" defects, such
as material flaws, changed material properties in cold worked region; changed material
properties due to fire, stress concentrations due to local irregular geometries etc., which may
produce rapid crack growth. The corresponding degradation of cross-sectional capacity may
cause the member to fail at a lower load than predicted on the basis of the direct damage
from the accidental load. Crack growth is a matter of fracture mechanics analysis and will not
be discussed herein.

8.1.2 Definitions
Bifurcation load: Load at which the response changes from ideal elastic to ideal plastic
behaviour.

8.1.3 Notations

A Total area of plate stiffener, undamaged shell section


Ae Total area of effective plate-stiffener
A Reduced cross-sectional area of dented tube
E Elastic modulus
I Moment of inertia
M Bending moment
Mp Plastic bending moment
N Axial force
Nd Axial force in dented part of cross-section
NE Euler buckling force
Np Plastic axial force
W Elastic section modulus
Wr Reduced elastic section modulus of dented tube
Z Plastic section modulus
Zr Reduced plastic section modulus of dented tube
b Plate width between stiffeners
be Effective width
d Tube diameter
e Shift of neutral axis in dented section
fE Euler buckling stress
fy Yield stress
l Element length
le Effective length
t Tube thickness
α Half angle of damaged tube section
β Plate slenderness, tube wall slenderness factor
δd Dent depth
δe End shortening
δo Permanent lateral deflection
η Distance to centre point of dented area
Reduced slenderness

φ Stress reduction factor


ρ Factor
σ Stress
σcrit Ante Critical stress

8.2 Design strategy


The following approach is recommended:
- Evaluate possible crack growth in damaged area. If adequate safety against member
failure due to crack growth in the design life or before repair can not be documented, the
damaged element should be considered non effective.
- Perform (linear) global analysis to determine load effects due to environmental loads. The
damage should be modeled as accurately as possible.
- Compare load effects for all platform members with criteria for element strength. Residual
strength for damaged elements are given in Sections 8.3.1 - 8.3.5.
- If strength criteria are violated, repeat global analysis. Assume no or appropriate reduced
strength of damaged members.
- If adequate residual strength cannot be documented by proper use of linear methods,
potential strength reserves should be explored by means of nonlinear methods of
analysis.
8.3 Capacities of damaged elements
8.3.1 Dented unstiffened tubulars
Analytical models for residual strength assessment of unstiffened tubulars with local dents
and permanent lateral deflections cab be found for example in /2/,/4/. Following the approach
adopted in /2/ the cross-sectional shape in the dented region is idealized by subdividing into
an undamaged part and a dented, flat section, see Figure 8.3.

FIGURE 8.3 – IDEAL DENTED TUBE /2/


On the basis of a first yield criterion the capacity can be determined from the expressions

and

where δ0 = permanent lateral deflection


δd = dent depth
d = tube diameter
Nd = load carried by the dented part of the cross section
e = shift of neutral axis due to dent
η = distance to centre point of dented area (see Figure 8.3)
J = moment of inertia of undamaged section
Ar = reduced cross-sectional area (see Figure 8.4)
Wr = reduced sectional modulus (see Figure 8.4)
le = effective buckling depth
The expressions are based upon the asumption that the dent is located in the vicinity of ½.
For off-centre locations the above formulas yield concervative results. More refined calculations may
be performed with the procedure described in /2/.
FIGURE 8.4 – SHIFT OF NEUTRAL AXIS, AREA AND ELASTIC SECTION MODULUD OF
DENTED CROSS-SECTION VERSUS DENT DEPTH.

For thin walled tubes, i.e. d/t > 80, premature local buckling may take place. This can be
accounted for by reducing the yield stress by a factor φ, to be taken as /2/.
δd
for < 0.02,
d
1
φ=
1 + λ4
where

fy
λ=
fE
t
fE = 1.21 ρE
d
d  
− 1
t
ρ = 1 − 0.0077 1 + 259
t d 
 
δd
for > 0.02,
d
E t
φ =1 if 3 > 0.08
fy d
else

11.095β − 30.47β2 β > 0.164


φ = 
1.0 β ≥ 0.164

where

β=3
E t
(1 + sin α )
fy d
The numerical procedure of /2/ accounts also for the strength reserves in the elastic-plastic
region. Theoretical predictions agree excellently with experimental results for a wide range of
tube geometries, loads and boundary conditions.
The bifurcation load (see Section 8.1) may be determined from the following expressions

where the axial force and bending moment should satisfy the plastic interaction function for
the damaged section.

Here is:

= 1
clamped at ends
C
0 rotational ly free ends
δe = end shortening
l = element length

= reduced slenderness ratio for undamaged tube


Np = yield force of undamaged section
Mp = plastic bending moment of undamaged section
Otherwise, the notation above applies. It is emphasized that this procedure does not account
for dent growth in the post-collapse range and possible local buckling at member ends.

8.3.2 Cracked unstiffened tubulars


The behaviour of cracked tubes has been studied for example in /5/, /6/. The plastic cross-
sectional capacities will be reduced for 2 reasons.
- reduced cross-sectional area
- reduced allowable tensible strain at the crack tips due to brittle or ductile fracture
Little information is available on critical crack sizes for realistic geometries, local stress
distributions and fabrication procedures. However, it is noted that the material commonly used
In offshore structures is very ductile.
In lieue of more accurate information Figure 8.5 may be used. It is based on the following
assumptions
- Critical crack tip (tensile) strain is estimated on the basis of the COD criterion /7/ for
different crack sizes
- The cracked part of the cross-section can take compressive forces but no tension
- There is no limitation as to the magnitude of compressive strains
The stress resultants complying with these assumptions are plotted in Figure 8.5 for different
crack sizes. A practical check is to ensure that the stress resultants work within the interaction
surface.
Combined with crack growth analysis, the curves may also be used to predict the time when
the crack size becomes critical.
It should be recognized that Figure 8.5 relates to a single overload. If there are multiple
excursions close to the interaction surface, possible premature failure due to low-cycle fatigue
effects should be considered.

FIGURE 8.5 REDUCED PLASTIC CAPACITIES DUE TO CRACK


8.3.2 Stiffened cylindrical shells
Experimental and theoretical investigations of dented, stiffened cylindrical shells are reported
in /8-10/. The studies conclude that the compressive strength may be assessed by an
effective section approach, i.e. the dented section is assumed ineffective.
This yields

where
Ar = effective cross-sectional area of dented section (see Figure 8.4)
A = area of undamaged section
Zr = reduced section modulus of
σcrit = axial strength of undamaged section
σ = axial strength of damaged section
Alternatively, the axial strength may be taken directly from Figure 8.6.
For slender geometries, i.e. d/t > 1000 the strength should be taken as /10/.
σ* = 0.90 • σ

FIGURE 8.6 – COMPRESSIVE RESIDUAL STRENGTH (REPRODUCED FROM /9/)


8.3.4 Plated members with permanent deflections
Extensive numerical analysis have been carried out on plates with different geo-metries and
various subjected to combinations of the bi-axial compression and lateral loading , see
eg. /11-14/. The residual strength of moderately distorted plated in unaxial compression may
be assessed by an effective width approach /13/.

b = plate width between stiffeners


be = effective width
t = plate thickness
δ = maximum permanent distortion
β = plate slenderness

b fy
β =
t E
For combined in-plane loading various interaction formulas have been proposed /14/.
If the stiffener is subjected to permanent deformation as well, the compressive residual
strength can be assessed by means of a first yield criterion, i.e.

where
N = axial force
Ae = Astiff + bet = total area of effective plate-stiffener
Astiff = Area of stiffener without plate flange
W = elastic section modulus of effective plate-stiffener
The effective permanent distortion should be taken as

where
A = A stiff + bt = total area of plate stiffener
δs = permanent distortion of stiffener
zp = distance from neutral axis to plate midplane.
be = effective width of plate (see above)
The Euler buckling stress of the plate-stiffener is

where
J = moment inertia of effective plate-stiffener
le = effective buckling strength
The effective buckling strength should be taken equal to the distance between the bending
moment inflection points. For multispan deformations

where
l = distance between transverse frames
It is noted that first failure does not necessarily imply collapse due to the difference in capacity
of plate induced and stiffener-induced failure.
A necessary condition for the above analysis is that the cross-sectional shape has not been
distorted during the accidental loading.
The bifurcation load (see Section 8.1) may be obtained from the following expressions

and

where the axial force and bending moment should satisfy the appropriate interaction function,
see Figure 8.7.
The following notation apply

N = axial force
M =bending moment
δe = end shortening
δo = initial lateral deformation
l = element strength

= reduced slenderness ratio


Np = plastic axial force of plate-stiffener
Mp = plastic bending moment of plate-stiffener
For plate induced failure, the plastic capabilities should be based upon the effective width of
the plate flange.
Further information on the residual strength of stiffened plates and plate assemblages cab be
found in /15,/16/.
FIGURE 8.7 – PLASTIC INTERGRATION FUNCTIONS FOR VARIOUS SECTIONS

8.3.5 Built-up members with initial deformations


For built-up members like box-girders and plate girders the effect of permanent deformations
in stiffened and unstiffened flanges can be carried out by means of the methods described for
plates members.
It is anticipated that tension flanges in a static analysis are not influenced by initial
imperfections. However, it should be observed that the load effect may as well be
compressive due to the cyclic nature of the waves forces.
The collapse behaviour of stiffened plate girders have been subjected to extensive
experimental and theoretical investigations /17, 18/. Design formulas are also available /19/.
For transversely stiffened webs the ultimate load is governed by the ability of the tension field
to develop in the web plating. Conceivably, this is not Impaired for moderate interframe plate
deformations, provided that the plate flanges remain undamaged. (Indeed, the elastic
stiffness is reduced). Hence, the residual strength may be assumed equal to the ultimate
strength of undamaged cross-section.
Longitudinal stiffeners tend to increase the ultimate strength of the web. However, beyond
ultimate load there is a rapid decrease to a postfailure plateau corresponding to no
longitudinal stiffeners /18/. Accordingly, if the web distortions include the longitudinal
stiffeners, the residual strength should be taken equal to the ultimate strength of a
transversely stiffened girder, assuming no contribution from the longitudinal stiffeners.
If also the transverse stiffeners and/or the plate flanges have suffered plastic deformations,
the residual strength of the girder is reduced. However, little information is available for
quantitative assessment of such damages.
8.4 Systems analysis
Methods for systems analysis may be categorized into three different groups
- linear methods
- extended linear methods
- non-linear techniques

8.4.1 Linear methods


This category comprises analysis with standard linear FE programs. In damaged condition the
members in question are removed or substituted by forces or equivalent members with
reduced properties so as to represent their residual strength. Collapse is defined as first yield,
or buckling. In effect this is no real progressive collapse (PLS) simulation but rather a check
according to the ultimate limit state criterion (ULS).

8.4.2 Extended linear methods


This technique is based upon application of a linear FE-program, but differs from the previous
approach in the sense that the behaviour after first failure is modeled, although in a simplified
way.
When failure is detected the element is removed and are substituted by external forces Rpc
equal to its post-collapse resistance before reanalyses. Members subjected to tensile yield
are typically replaced by the yield force, buckling members by a specified fraction of the
buckling force. If large plastic strains are Involved the effect of possible strain hardening
should also be considered. The post-buckling resistance may be updated at each step.
The procedure is described schematically in Figure 8.8 for the case of a dented tubular
member /2/. First, the effective buckling length for the dented tube is calculated in a global
response analysis, where the dented section is replaced by an equivalent tubular element.
The dimensions of the equivalent element is determined so as to represent as closely as
possible the plastic interaction curve for the dented section. In addition to the lateral
displacement, the equivalent tube is located eccentrically with respect to the neutral axis.
On the basis of the effective length and the dent geometry, the load-end shortening curve for
the dented tube is assessed either as in /2/ by means of a computer program or by analytic
means, for example by rigid-plastic methods as discussed in Section 8.1.
If the actual compression load is higher than the ultimate load calculated, the effective
buckling length of the damaged member is removed and its load effect is replaced by a force
equal to the ultimate load. Then, a new systems analysis is carried out and the relative
displacements between the nodes of the damaged element is found. The residual strength is
updated on the basis of the load-end shortening curve. The procedure is repeated until the
desired level of accuracy is obtained. During this process other members may buckle or yield
in tension. The procedure described above is then also applied for the buckling member with
no initial damage, while the tension member is replaced by its yield load.
FIGURE 8.8 – EXTENDED LINEAR SYSTEM ANALYSIS WITH A DENTED MEMBER /2/
8.4.3 Nonlinear techniques
The conventional finite element formulation has been widely developed into the nonlinear
range of analysis, and several nonlinear systems are available, see for example /20-21/. In
principle such analyses will yield the most accurate prediction of the residual strength.
However, for analysis of space frames, such as offshore platforms, the use of these systems
is costly and complicated. Several finite elements must be used per structural element. In
addition, numerical integration must be carried out over the cross-section as well as along the
length of the member. Hence, it may be difficult for an inexperienced analyst to select the
proper modeling.
It should be recognized that even if the conventional non-linear programs in principle are
capable of calculating the "true" response history, they can only do this within the inherent
limitations of the element modeling. For example, if beam elements are used, they fail to take
into account local buckling and dent growth in the post-collapse range, fracture etc. These
effects can be of utmost importance with respect to progressive collapse behaviour of actual
structures
This recognition has inspired the development of various simplified procedures, but which use
the framework of general nonlinear finite element programs. Some methods are based on
phenomenological models, where for example the force-axial displacement relationship of a
compression member is established analytically or by experiments, and represented as a one
degree-of-freedom system /22-24/. The load displacement curve is often approximated by
piece wise linearization, as illustrated in Figure 8.9 /24/.

FIGURE 8.9 – SIMPLIFIED STIFFNESS REPRESENTATION OF PLATE PANEL /24/

This has the following advantages:


- the simplified stiffness representation is valid for a wide range of strains
- only those members with changing stiffness need to be modified
- the global stiffness matrix is not rebuilt; the old element stiffness is subtracted and the
new one is added
Extensive reserve- and residual strength analysis of jacket platforms is reported in /25/. The
program used in these studies /23/ offers non-linear beam columns which can develop plastic
hinges as well as truss members, which yield in tension and buckle in compression. The load-
end shortening relationships have been correlated against tests.
Another technique is based on the so-called Idealized Structural Unit Method /26/, originally
developed for analysis of ship structures. In an extension to offshore structures /27/ member
failure occurs either in the form of plastic hinges or by buckling. Because no coupling exists
between axial displacements and lateral rotation, buckling does not emerge through energy
principles, but is introduced artificially by comparing the axial force with design curves.
The approach above has inspired the development of the computer programs in /28, 29/.
Here, the limitation mentioned above has been eliminated by reformulating the method in the
large displacement range by means of the virtual work principles.
The principles behind the programs of /27-29/ are quite similar and are illustrated in
Figure 8.9 for the program of /29/. The basic idea is to use one finite element per structural
element, which allows FE meshes from linear analysis to be used. The elements are
subdivided into an elastic part and a plastic part (yield hinges). In this way tangential stiffness
matrices can be established analytically; no numerical integrations is required. Loads are
incremented analytically and scaled if new plastic hinges or global instability are detected. A
dented tube model is included and the initiation and growth of local buckling in tubes and
plate girder webs is taken into account.

FIGURE 8.10 – BASIC CONCEPTS OF NON-LINEAR SYSTEM ANALYSIS WITH YIELD


HINGES /29/

8.4.4 Cyclic degradation of platform stiffness


The methods described in Sections 8.4.1-8.4.3 are concerned with the resistance against a
single overload. It is recognized that an extreme environmental load may be accompanied by
repetitive close-to-extreme loads. Thus, potential low-cycle fatigue failure should be observed,
if a high plastic utilization is accounted for in the static strength verification.
A review of the post-buckling and cyclic behaviour of beam-columns and frames Is given in
/30/. A typical load-displacement history is traced in Figure 8.11. This cyclic straining will
impose heavy demands on the ductility of the structure, especially if local imperfections
(dents, cracks) are present.
A structure subjected to variable repeated loading may fail at a load that is smaller than the
static collapse load. A mechanism is formed if the magnitude of irrecoverable plastic rotations
increases indefinitely after a number of load repetitions. This process is denoted incremental
collapse. Alternatively, the structure may subjected to alternating plasticity, giving failure due
to accumulated plastic strain. This is denoted low-cycle fatigue.
The damage due to cyclic plastic strains may be assessed by the Manson-Coffin formulae
/31/, which is the counterpart to the Palmgren-Miner rule in the high-cycle fatigue region.
However, an important limitation with plastic hinge based methods is that they often give
strain resultants rather than strains. A strain model has to be developed before the Manson-
Coffin formula can be used.
FIGURE 8.11 – CYCLIC LOAD-DISPLACEMENT RELATIONSHIP FOR BEAM-COLUMN /30/

8.5 References
/1/ Norwegian Petroleum Directorate:
Regulation for Load-Bearing Structures Intended for Exploitation of Petroleum
Resources, Stavanger, 1984.
/2/ Taby, J.:
"Ultimate and Postulitmate Strength of Dented Tubular Members", Dr.ing. thesis,
Division of Marine Structures, with Appendices, NTH, Trondheim, 1986.
/3/ Murray, N.W.:
"Recent Research into the Behaviour of Thin-Walled Steel Structures", Steel
Structures - Recent Research Advances and Their Application to Design, Elsevier
Appl. Science Publ, London, 1986.
/4/ Ellinas, C.P.:
"Ultimate Strength of Damaged Tubular Bracing Members", Journ. of Struct. Engin.,
Vol. 110, No. 2, pp. 245-259, February 1984.
/5/ Yao, T.:
"Elastic Plastic Analysis of Structural System Containing Cracked Members", Div. of
Marine Structures, The Norwegian institute of Technology, March 1986.
/6/ Hessen, G. and Moan, T.:
rd
"Fracture Mechanics Analysis of Stiffened Tubular Members", 3 Int. Symp. on
Practical Design of Ships and Mobile Units, Prads '87, Trondheim, 1987.
/7/ Andersen, T.L., Leggatt, R.H. and Garwood, S.G.:
"The Use of CTOD Methods in Fitness for Purpose Analysis", Workshop on CTOD
Methodology, GKSS Research Centre, Geestacht, Germany, 1985.
/8/ Ronalds, B.F.:
"Mechanics of Dented Orthogonally Stiffened Cylinders", Doctor of Phil. Thesis,
Imperial College, London, 1985.
/9/ Oroufriou, A.:
"Collapse of Damaged Ring Stiffened Cylinders", Doctor of Phil. Thesis, Imperial
College, London 1987.
/10/ Dowling, P.J., Ronalds, B.F., Onoutriou, A. and
rd
Harding, J.E.: "Resistance, of Buoyant Columns to Vessel Impact", 3 Int. Symp. on
Pract. Design of Ships and Offshore Units, Prads '87, Trondheim, 1987.
/11/ Dow, R.S. and Smith, C.S.:
"Effects of Localized Imperfections on Compressive Strength of long Rectangular
Plates", Journ. of Constructional Steel Research, Vol. 4, pp. 51-76, 1984.
/12/ Carlsen, C.A.:
"Simplified Collapse Analysis of Stiffened Plates", Norwegian Maritime Research,
No. 4, 1977.
/13/ Faulkner, D.:
"Design Against Collapse of For Marine Structures", Advances in Marine Technology,
Trondheim, 1979.
/14/ Soares, C.G. and Slreide, T.H.:
"Behaviour and Design of Stiffened Plates Under Predominantly Compressive
Loads", International Shipbuilding Progress, Vol. 30, No. 341, 1983.
/15/ Gerard, G.:
"Compressive Strength of Flat Stiffened Panels", NACA, TN3785, 1957.
/16/ Smith, C.S. and Dow, R.S.:
"Residual Strength of Damaged Steel Ships and Offshore Structures", Journal of
Constructional Steel Research, Vol. 1, No. 4, 1981.
/17/ Tang, K.H. and Evans, H.R.:
"Transverse Stiffeners for Plate Girder Webs - an Experimental Study", Journal of
Constructional Steel Research, 1984.
/18/ Evans, H.R. and Tang, K.H.:
"The Influence of Longitudinal Web Stiffeners Upon the Collapse Behaviour of Plate
Girders", Journal of Constructional Steel Research, 1984.
/19/ Rockey, K.C., Valtinat, G. and Tang, K.H.:
"The Design of Transverse Stiffeners on Webs Loaded in Shear - An Ultimate Load
Approach", Proc. Institution of Civil Engineers, Part 2, pp. 1069-1099, 1981.
/20/ Bergan, P.G. and Arnesen, A.:
th
"FENRIS - A General Purpose Finite Element Program", 4 Intern. Conf. on Finite
element Systems, Southampton, July 1983.
/21/ ABAQUS - General Purpose Nonlinear Finite Element Program, Hibbit, Karlsson,
Sorensen, Providence, RI.
/22/ Mondkar, D.P. and Powell, G.H.:
"ANSR-I General Purpose Program for Analysis of Non-linear Structural Response",
Earthquake Engng Research Center, University of California, Berkeley, Report 75/37,
1975.
/23/ INTRA - PMB Engineering Systems.
/24/ Pettersen, E. and Valsg}rd, S.:
"Collision Resistance of Marine Structures", In Structural Crashworthi-ness, Ed. N.
Jones and T. Wiueerzbichi, Butterworths, London, pp. 338-370, 1983.
/25/ Lloyd, J.R. and Clawson, W.C.:
"Reserve and Residual Strength of Pile-Founded Offshore Platforms, Proc. of Design
- Inspection - Redundancy Symposium, Williamsburg, Virginia, 1983.
/26/ Ueda, Y. and Rashed, S.M.H.:
"An Ultimate Transverse Strength Analysis of Ship Structures", Journ. of the Society
of Naval Architects of Japan, Vol. 136, 1974.
/27/ Ueda, Y, Rashed, S.M.H., and Nakacho, K.:
"New Efficient and Accurate Method of Nonlinear Analysis of Offshore Tubular Frame
(the Idealized Structural Unit Method)", OMAE, 1984, pp. 260-268.
/28/ Nedergaard, H. and Pedersen, P.T.:
"Analysis Procedure for Space Frames vith Material and Geometrical Non-linearities",
Europe-US Symposium on Finite Element Methods Nonlinear Problems, The
Norwegian Institute of Technology, August 12-16, 1985.
/29/ Soreide, T.H. and Amdahl, J.:
"A Computer Program for Ultimate Strength Analysis of Frame Offshore Structures -
Theory Manual", SINTEF Report No. STF71 A86049, Trondheim, 1986.
/30/ Chen, W.F. and Sugimoto, F.:
"Fabricated Tubular Colums Used in Offshore Structures", Shell Structures - Stability
and Strength, R. Narayaman (Ed.), Elsevier Appl. Publ., New York, 1985.

8.6 ILLUSTRATIVE EXAMPLE

A brace is damaged from a minor ship collision. Calculate the residual strength and the load-
end displacement relationship for the brace.
(Estimate the residual strength for both pinned and clamped ends, and on the basis of a first
yield as well as a bifurcation load criteria.)
FIRST YIELD CRITERION:
Residual strength for the brace based on the first yield criterion is 6674 kN with pinned ends
and 7215 kN with clamped ends.
THE BIFURCATION LOAD CRITERIA
The plastic part will remain the same as for pinned ends, while the elastic part
becomes:

The graphical solution of the equations are shown on the figures below.
Residual strength of the two cases estimated by the two methods:

First yield criteria Bifurcation load criteria


Pinned ends 6674 kN 7125 kN
Clamped ends 7215 kN 7545 kN

You might also like