ESDEP WG 15A
STRUCTURAL SYSTEMS: OFFSHORE
To present the analysis procedures for offshore structures relating to fatigue, abnormal and accident conditions, load-out and transportation, installation and local design.
Lecture 15A.1: Offshore Structures: General Introduction
Lecture 15A.2: Loads I: Introduction and Environmental Loads
Lecture 15A.3: Loads II: Other Loads
Lecture 15A.4: Analysis I
Methods of fatigue analysis are described including the fatigue model (structural, hydrodynamic loading, and joint stress models) and the methods of fatigue damage assessment.
Abnormal and accidental conditions are considered relating to earthquake, impact and progressive collapse.
Analyses required for load-out and transportation and for installation are outlined. Local analysis for specific parts of the structure which are better treated by dedicated models outside of the global analysis are identified.
A fatigue analysis is performed for those structures sensitive to the action of cyclic loadings such as:
The in-place model is used for the fatigue analysis.
Quasi-static analysis is often chosen; it permits all local stresses to be comprehensively represented. The dynamic effects are accounted for by factoring the loads by the relevant DAF.
Modal analysis may be used instead; it offers computational efficiency, but may also overlook important local response modes, particularly near the waterline where direct wave action causes high out-of-plane bending (see Section 5.2). The mode - acceleration method may overcome this problem.
A very large number of computer runs may be necessary to evaluate the stress range at the joints. The wave is repeatedly generated for:
Nominal joint stresses are calculated for eight points around the circumference of the brace. The maximum local (hot spot) stress is obtained by multiplying the former by a stress concentration factor (SCF) given by parametric formulae which are functions of the joint geometry and the load pattern (balanced/unbalanced).
The fatigue failure of joints in offshore structures primarily depends on the stress ranges and their number of occurrences, formulated by S-N curves:
log Ni = log a + mlog Dsi
The number of cycles to failure Ni corresponds to a stress range. The effect of the constant stresses, mainly welding residual stresses, is implicitly accounted for in this formulation.
The cumulative damage caused by ni cycles of stress Dsi, over the operational life of the platform (30 to 50 years) is obtained by the Palmgren-Miner rule:
D =
The limit of this ratio depends on the position of the joint with respect to the splash zone (typically +/-4m on either side of the mean sea level). The ratio should normally not exceed:
The damage may alternatively be expressed in closed form:
D = 
where
a
, m are coefficients of the selected S-N curve.Ds
is the stress range exceeded once in N cycles.k is a long-term distribution parameter, depending on the position of the joint in the structure.
N is the total number of cycles.
This analysis consists of time-domain analysis of the structure. The main advantage of this representation is that non-linear effects (drag, high order wave theories) are handled explicitly.
A minimum of four regular waves described in terms of height and associated period are considered for each heading angle.
Waves of a given height are not characterised by a unique frequency, but rather by a range of frequencies. If this range corresponds to a peak in the structural response, the fatigue life predicted by the deterministic method can be seriously distorted.
This problem is overcome by using a scatter diagram, in which the joint occurrence of wave height and period is quantified. Wave directionality may also be accounted for. Eventually the most thorough representation of a sea state consists of:
This approach requires that the physical process be approximately linear (or properly linearised) and stationary. Transfer functions TF are determined from time-domain analyses involving various wave heights, each with different period and incidence:
The response has normally a narrow-banded spectrum and can be described by a Rayleigh distribution.
The zero-upcrossing frequency of stress cycles is then approximated by:
Tz = 
where mn is the nth order moment of the response.
The significant stress range is readily obtained for each sea state as:
ssig = 
where S(w,q) is the directional wave energy spectrum.
The fatigue damage caused by the fluctuating part of wind (gusts) on slender structures like flare booms and bridges is usually predicted by spectral methods.
The main feature of such analysis is the introduction of coherence functions accounting for the spanwise correlation of forces.
Vortex induced failure occurs for tubes subjected to a uniform or oscillating flow of fluid.
Within a specific range of fluid velocities, eddies are shed at a frequency close to the resonant frequency of the member.
This phenomenon involves forced displacements, which can be determined by models such as those suggested in [1].
This type of analysis addresses conditions which may considerably affect the integrity of the structure, but only have a limited risk of occurrence.
Typically all events with a probability level less than the 10-4 threshold are disregarded.
Particular attention shall be paid to:
The seismic forces in a structure are highly dependent on its dynamic characteristics. Design recommendations are given by API to determine an efficient geometry. The recommendations call for:
Earthquake analyses can be carried out according to the general methods presented in Lecture 15A.4.
However their distinctive feature is that they represent essentially a base motion problem and that the seismic loads are therefore dependent on the dynamic characteristics of the structure.
Modal spectral response analysis is normally used. It consists of a superposition of maximum mode response and forms a response spectrum curve characteristic of the input motion. This spectrum is the result of time-histories of a SDOF system for different natural periods of vibration and damping.
Direct time integration can be used instead for specific accelerograms adapted to the site.
The analysis of impact loads on structures is carried out locally using simple plastic models [2].
Should a more sophisticated analysis be required, it can be accomplished using time-domain techniques presented in Section 6 of Lecture 15A.4.
The whole energy must be absorbed within acceptable deformations.
When a wellhead protection cover is hit by a drill collar, or a tube (jacket leg, fender) is crushed by a supply boat, two load/deformation mechanisms occur simultaneously:
Owing to the current lack of definitive guidance regarding explosions and fire, the behaviour of structures in such events has so far been only predicted by simple models based on:
In the aftermath of recent mishaps however, more accurate analyses may become mandatory, based on a better understanding of the pressure-time histories and the effective resistance and response of structures to explosions and fire.
Some elements of the structure (legs, bracings, bulkheads) may partially or completely loose their strength as a result of accidental damage.
The purpose of such analysis is to ensure that the spare resistance of the remaining structure is sufficient to allow the loads to redistribute.
Since such a configuration is only temporary (mobilisation period prior to repairs) and that operations will also be restricted around the damaged area, reduced live and environmental loads are generally accepted.
In this analysis, the damaged elements are removed from the model. Their residual strength may be represented by forces applied at the boundary nodes with the intact structure.
The load-out procedure consists in moving the jacket or module from its construction site to the transportation barge by skidding, or by using trailers underneath it.
The barge may be floating and is continuously deballasted as the package progresses onto it, or grounded on the bottom of the harbour.
The most severe configuration during skidding occurs when the part of the structure is cantilevering out:
The analysis should also investigate the possibility of high local reactions being the result of settlement of the skidway or errors in the ballasting procedure.
As the reaction on each trailer can be kept constant, analysis of load-out by trailers only requires a single step to determine the optimal distribution of trailers.
The model consists of the rigid-body assembly of the barge and the structure.
Barges are in general characterised by a low length/beam ratio and a high beam/draught ratio, as well as sharp corners which introduce heavy viscous damping.
For jacket transport, particular care shall be taken in the representation of overhanging parts (legs, buoyancy tanks) which contribute significantly to the righting moment.
Dry-transported decks and modules may be simply represented by their mass and moments of inertia.
This analysis shall provide the linear and angular accelerations and displacements of the structure to be entered in the structural model as inertia forces, and also the partition and intensity of buoyancy and slamming forces.
The jacket model is a simplified version of the in-place model, from which eccentricities and local reinforcements may be omitted.
The barge is modelled as a plane grid, with members having the equivalent properties of the longitudinal and transversal bulkheads.
As the barge passes over a wave trough or a crest, a portion only of the barge is supported by buoyancy (long barges may be spanning over a whole trough or be half-cantilevered).
The model therefore represents the jacket and the barge as two structures coupled together by the seafastening members.
A three dimensional analysis is carried out to evaluate the global forces acting on the jacket at various time steps during the launch sequence.
At each time step, the jacket/barge rigid body system is repositioned to equilibrate the internal and external forces produced by:
The maximum reaction on the rocker arm is normally obtained when the jacket just starts rotating about the rocker hinge.
The structural model is in all aspects identical to the one used for the transportation analysis, with possibly a finer representation of the launch legs.
The rocker arm is also represented as a vertical beam hinged approximately at midspan. Interface loads obtained by the rigid body analysis are input at boundary conditions on the launch legs. All interface members must remain in compression, otherwise they are inactivated and the analysis restarted for that step.
Once the tilting phase has begun, the jacket is analysed at least for each main leg node being at the vertical of the rocker arm pivot.
No dedicated structural analysis is required for this phase, which is essentially a naval architecture problem.
A local analysis of the lugs is performed for crane-assisted upendings.
Docking of a jacket onto a pre-installed template requires guides to be analysed for local impact. The same requirement applied for bumpers to aid the installation of modules.
The condition where the jacket may for a while stand unpiled on the seafloor is analysed for the design installation wave.
The stability of the jacket as a whole (overturning tendency) is investigated, together with the resistance of the mudmats against soil pressure.
The piles are checked during driving for the dynamic stresses caused by the impact wave of the hammer blow. The maximum cantilevered (stick-up) length of pile must be established for the self-weight of the pile and hammer combined, accounting for first and second order moments arising from the pile batter. Hydrodynamic actions are added for underwater driving.
Elements in the vicinity of the piles (guides, sleeves) shall also be checked, see Section 5.1.
The model used for the lift analysis of a structure consists of the in-place model plus the representation of the rigging arrangement (slings, spreader frames).
For single lifts the slings converge towards the hook joint, which is the sole vertical support in the model and shall be located exactly on the vertical through the centre of gracity (CoG) of the model.
For heavier dual-crane lifts, the CoG shall be contained in the vertical plane defined by the two hook joints.
The mathematical instability of the model with respect to horizontal forces is avoided by using soft horizontal springs at the padeyes. The force and elongation in these springs should always remain small.
Different factors are applied to the basic sling forces to account for specific effects during lifting operations.
This factor represents the effect of fabrication tolerances and lack-of-fit of the slings on the load repartition in a statically undetermined rigging arrangement (4 slings or more). Skew factors may either be directly computed by applying to a pair of opposite slings a temperature difference such that their elongation/shortening corresponds to the mismatch, or determined arbitrarily (typically 1/3 - 2/3 repartition).
This factor accounts for global dynamic effects normally experienced during lifting operations. DnV [24] recommends minimum values as follows:
|
Lifted Weight W (tonnes) |
up to 100 t |
100 t to 1000t |
1000 t to 2500t |
more than 2500 t |
|
DAF offshore |
1,30 |
1,20 |
1,15 |
1,10 |
|
DAF inshore |
1,15 |
1,10 |
1,05 |
1,05 |
This factor accounts for additional sling loading caused by the rotation of the lifted object about a horizontal axis and by the longitudinal deviation of the hooks from their theoretical position in the case of a multi-hook lift. It shall normally be based on 5° and 3° tilt respectively depending on whether cranes are on different vessels or not.
This factor accounts for the rotation of the lifted object about a vertical axis (equal to 1,05 typically).
Forces in elements checked under lift conditions are multiplied by a factor reflecting the consequence a failure of that specific element would have on the integrity of the overall structure:
Local analyses address specific parts of the structure which are better treated by dedicated models outside the global analysis.
The list of analyses below is not exhaustive and more information can be found in [1-24] which provide a complete design procedure in each particular case.
Underwater pile/sleeve connection is usually achieved by grouting the annulus between the outside of the pile and the inner sleeve.
The main verifications address:
Horizontal members (conductor guide frames in particular) located within the splash zone (+/-5m on either side of the mean-sea-level approximately) shall be analysed for fatigue caused by repeated wave slamming.
A slamming coefficient Cs=3,5 is often selected.
Typical straightened nodes (ring-stiffened nodes, bottle legs nodes with diaphragms) are analysed by finite-elements models, from which parametric envelope formulae are drawn and applied to all nodes representative of the same class.
Static In-Place and Fatigue
Risers, caissons and J-tubes are verified either by structural or piping programs for the action of environmental forces, internal pressure and temperature. Particular attention is paid to the bends not always satisfactorily represented by structural programs and the location of the touch-down point now known a-priori.
A fatigue analysis is also performed to assess the fatigue damage to the clamps and the attachments to the jacket.
Pull-In
J-tubes are empty ducts continuously guiding a post-installed riser pulled inside. They are verified by empirical plastic models against the forces generated during pull-in by the friction of the cable and the deformation of the pull head, see [22].
Conductors are analysed in-place as beam columns on discrete simple supports, these being provided by the horizontal framing of the jacket (typically 20 to 25 m span).
The installation sequence of the different casings must be considered to assess the distribution of stresses in the different tubes forming the overall composite section.
Also the portion of compression force in the conductor caused by the hanging casings is regarded as an internal force (similar to prestressing) which therefore does not induce any buckling tendency, see [23].
The helideck is normally designed to resist an impact load equal to 2,5 times the take-off weight of the heaviest helicopter factored by a DAF of 1,30.
Plastic theories are applicable for designing the plate and stiffeners, while the main framing is analysed elastically.
Analyses of flare booms particularly consider:
× check the interfaces between the different analyses and ensure the consistency of the input/output.
× verify the validity of the data resulting from a complex analysis against a simplified model, which can also be used to assess the influence of a particular parameter.
× make full use of "good engineering judgement" to criticise the unexpected results of an analysis.
[1] Skop R.A. & Griffin O.M., An Heuristic Model for Determining Flow-Induced Vibrations of Offshore Structures/OTC paper 1843, May 1973.
[2] De Oliveira J.G., The Behaviour of Steel Offshore Structures under Accidental Collisions/OTC paper 4136, May 1981.
[3] API-RP2A, Recommended Practice for Planning, Designing and Constructing Fixed Offshore Platforms/18th edition, September1989.
[4] DnV, Rules for the Classification of Fixed Offshore Structures, September 1989.
[5] DnV, Standard for Insurance Warranty Surveys in Marine Operations, June 1985.
[6] NPD, Regulation for Structural Design of Loadbearing Structures Intended for Exploitation of Petroleum Resources, October1984 and Veiledning om Utforming, Beregning og Dimensjonering av Stalkonstruksjoner i Petroleumsvirksomheten, December1989.
[7] DoE, Offshore Installations: Guidance on Design and Construction/London, April 1984.
[8] McClelland B. & Reifel M.D., Planning and Design of Fixed Offshore Platforms/Van Nostrand Reinhold, 1986.
[9] UEG, Node Flexibility and its Effect on Jacket Structures/CIRIA Report UR22, 1984.
[10] Hallam M.G., Heaf N.J. & Wootton L.R., Dynamics of Marine Structures/ CIRIA Report UR8 (2nd edition), October 1978.
[11] Wilson J.F., Dynamics of Offshore Structures/Wiley Interscience, 1984.
[12] Clough R.W. & Penzien J., Dynamics of Structures/McGraw-Hill, New York, 1975.
[13] Newland D.E., Random Vibrations and Spectral Analysis/Longman Scientific (2nd edition), 1984.
[14] Zienkiewicz O.C., Lewis R.W. & Stagg K.G., Numerical Methods in Offshore Engineering/Wiley Interscience, 1978.
[15] Davenport A.G., The Response of Slender Line-Like Structures to a Gusty Wind/ICE Vol.23, 1962.
[16] Williams A.K. & Rhinne J.E., Fatigue Analysis of Steel Offshore Structures/ICE Vol.60, November 1976.
[17] Anagnostopoulos S.A., Wave and Earthquake Response of Offshore Structures: Evaluation of Modal Solutions/ASCE J. of the Structural Div., vol. 108, No ST10, October 1982.
[18] Chianis J.W. & Mangiavacchi A., A Critical Review of Transportation Analysis Procedures/OTC paper 4617, May1983.
[19] Kaplan P. Jiang C.W. & Bentson J, Hydrodynamic Analysis of Barge-Platform Systems in Waves/Royal Inst. of Naval Architects, London, April 1982.
[20] Hambro L., Jacket Launching Simulation by Differentiation of Constraints/ Applied Ocean Research, Vol.4 No.3, 1982.
[21] Bunce J.W. & Wyatt T.A., Development of Unified Design Criteria for Heavy Lift Operations Offshore/OTC paper 4192, May 1982.
[22] Walker A.C. & Davies P., A Design Basis for the J-Tube Method of Riser Installation/J. of Energy Resources Technology, pp. 263-270, September 1983.
[23] Stahl B. & Baur M.P., Design Methodology for Offshore Platform Conductors/J. of Petroleum Technology, November 1983.
[24] DnV - Rules for the Classification of Steel Ships, January 1989.