You are on page 1of 10

OTC 16588 Guidelines on Coupled Analyses of Deepwater Floating Systems

Vigleik L. Hansen, Lihua Wang, Nils Sodahl/ DNV and E. G. Ward/ OTRC/Texas A & M

Copyright 2004, Offshore Technology Conference This paper was prepared for presentation at the Offshore Technology Conference held in Houston, Texas, U.S.A., 36 May 2004. This paper was selected for presentation by an OTC Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Offshore Technology Conference or its officers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.

Abstract Within Deepstar (DS), a Guideline (ref. /1/) representing an industry consensus on deepwater design analysis methods and model test verification has been prepared. As the offshore industry move into deeper water, the capacity to analyze and model test deepwater floating systems are challenged. The designers need efficient tools and procedures on how to determine dimensioning responses. For some of the floating systems installed/planned, coupled analyses is considered a must, both with respect to safety and cost. The paper focuses on how to perform coupled analyses efficiently for deepwater floater and riser/mooring systems. 1. Introduction During the last years, the offshore industry has gained increased experience and knowledge with respect to how integrated, coupled analyses should be performed on floater types like FPSO, TLP, Spar, Semi, with associated riser and mooring systems. The introduction of new floater solutions/combinations, new materials and the fact that the riser/mooring system introduce more non-linearities in deeper water emphasise the need of a unified approach for coupled analyses. Model testing alone can not be regarded as solely suited for final verification/validation of complex floating systems in deep water when e.g. truncations have to be included in the tests. Some years ago, DS invited the offshore industry to perform analyses on some selected floating systems, with FPSO, TLP and Spar as the 3 Theme Structures. The relative large scatter in the calculated results led to the execution of model testing these 3 floating systems (ref. /2/). Following these model tests, a smaller group of vendors (3 basins and 4 designers) with DNV and OTRC as the coordinators were established. New analyses were conducted with the purpose to explain and reduce the scatter in results (refs. /3/, /4/ & /5/). This exercise typically involved modeling the model used in the model tests and necessary calibrations (model test results vs simulations). Further details

on these calibrations and correlations are given in ref. /6/. In parallel with these analyses and calibrations, a Guideline on how to perform coupled analyses of deepwater floating systems was prepared. The main objectives of the Guideline (ref. /1/) are to provide guidance on: when it is deemed necessary to perform coupled analyses. how to efficiently perform coupled analyses. riser and mooring line analyses. relevant load models for floater and slender structures. important coupling effects from slender structures to the floater. the interface between model tests and coupled analyses. which input parameters are of key importance in simulations. obtaining confidence in critical design and validation cycles. The 10 sections of the Guideline deal with the following issues, respectively: 1. Presents objectives, scope and background. 2. Gives an outline description of different floating concepts, riser and mooring systems and an evaluation of coupling effects. 3. Outlines a typical design process for deepwater developments. 4. Provides guidance on how to identify design conditions. 5. Provides guidance on how to perform de-coupled analysis, covering load models, static analysis, frequency domain and time domain analyses. 6. Global slender structure analysis with focus on slender structure load models and FE analyses. 7. Provides guidance on the methodology and procedures of coupled analysis. 8. Model testing and which effects and parameters are of importance. 9. Presents results of sensitivity studies. 10. Statistical post processing of results from the coupled analyses and/or the slender structures. These sections will be, respectively, summarized in the following sections. It is to be noted that Section 8, covering model testing of deep & ultra deep water floating systems is presented in a separate paper, ref. /7/.

OTC 16588

2. Deepwater Floating Systems A common feature of all types of floaters is that they utilize excess buoyancy to support deck payload and provide slender (mooring, tendon, riser) structure tensions. Another common characteristic of all floater types in deep water is that they are soft in the horizontal plane, implying that their surge, sway and yaw eigen periods are generally in the range 100 s 400s. The fundamental differences among the floaters are related to their motions in the vertical plane, i.e., heave, roll and pitch. The floater motions in the vertical plane are decisive for the choice of riser and mooring system. Fig. 1 shows comparisons of typical heave RAOs for different floater types. 3. Design Process for Deepwater Developments It is argued that the design process for deep water floating systems is not significantly different than that for shallower water systems. The difference will depend on the actual water depth, floater concept, mooring system and most importantly, the riser system. Designing floating systems for deeper water implies that the cost of the mooring and riser system will increase relative to the floater compared with shallow water systems. This implies that there is a greater incentive to come up with a cost optimal design for the mooring and riser system. Applying efficient coupled analyses from the conceptual design stage and into the detail design phase would assist the designers in avoiding too conservative margins during the early design stages. If the coupling effects are known to be minor, traditional decoupled analyses may be used to validate the final (or conceptual) design with appropriate model testing. This approach may be valid for water depth up to 3000 5000 feet depending on floater/system and excitation level (e.g. benign areas). For larger water depths and concepts requiring simplifications (e.g. truncations) in the model testing, efficient coupled analyses and subsequent slender structure analyses should be used in design. The term efficient in this context implies that the computation time is less, or equal to real time and no more than 2 3 times real time for more complicated systems with e.g. consideration of riser/hull friction (hysteresis effects). If a Finite Element (FE) approach is applied, typically 1000 1500 elements would suffice with focus on single floater motions and top tensions of the slender structures. 4. Identification of Design Responses The characteristic load effect to be applied in the design checks should reflect the most probable extreme response over a specified design time period. The most probable extreme response during D years is commonly denoted the D-year return period value, which corresponds to an annual exceedance probability of 1/D. Design can be based on either environmental statistics, or response statistics. 4.1 Design Based on Environmental Statistics In this approach, the design extreme responses are obtained by exposing the system to a number of stationary design environmental conditions. Each design condition is a combination of wind, wave and current with typical return periods up to 100 years. Higher return periods (e.g. 1000 years) are also frequently applied to check the robustness of

the floater and slender structures. Each design condition is 3 6 hours long and described in terms of a number of environmental parameters such as significant wave height, peak wave period, wind speed, current speed, etc.. In general, the extreme design responses can be established using either a design wave approach, or a design storm approach. In the former, an extreme regular wave is used as the design wave. This approach is usually not relevant for coupled analyses. In the latter, an extreme irregular wave (short term seastate) is used as the design wave/storm. Time domain analyses is normally applied in both cases. The main challenge related to design criteria based on environmental statistics is that the return period for the characteristic load effect is unknown due to the non-linear behavior of e.g. riser systems. This will in general lead to an inconsistent safety level for different design concepts and failure modes. Acceptable results can, however, be expected for quasi-static systems with moderate non-linearities. A careful evaluation of the design criteria should be carried out for: New concepts Systems with significant non-linear response characteristics Dynamically sensitive systems 4.2 Design Based on Response Statistics Design based on response statistics is in general the recommended procedure and should be considered whenever practically possible for consistent assessment of characteristic load effects (especially when shortcomings in the traditional approach based on environmental statistics have been identified). This is of particular importance for extreme load and accidental load conditions, which are normally associated with the most pronounced nonlinear response characteristics. Consistent assessment of the D-year load effect will in general require a probabilistic response description reflecting the longterm environmental action e.g. on the riser system. The load effect with a return period of D-years, denoted xD, can be found from the long-term load effect distribution as: FX ( x D ) = 1 1 ND

N D - total number of load effect maxima during D years.


FX ( xD ) - long-term peak distribution of the load effect

5. Coupling Effects In this paper, the terms coupled and de-coupled analyses are frequently used. Both approaches are time-domain analyses, and with the use of the Finite Element (FE) approach for the coupled analyses. In case of multiple floaters, both de-coupled and coupled analyses may be performed. 5.1 Definition Coupling effects can be defined as the influence on floater mean position and dynamic response from slender structure restoring, damping and inertia forces, specified as follows:

OTC 16588

5.2 Restoring Force 1) Static restoring force from the station keeping system and the riser system as a function of floater offset. 2) Current loading and its effects on the restoring force of the mooring and riser system. 3) Seafloor friction (if mooring lines and/or risers have seabed contact) 5.3 Damping 4) Damping from mooring and riser system due to dynamics, current etc. 5) Damping due to hull/riser contact 5.4 Inertia 6) Additional inertia forces due to mooring and riser system In a traditional de-coupled analysis, item 1) can be accurately accounted for, items 2), 4) and 6) may be approximated. and items 3) and 5) generally cannot be accounted for. A coupled analysis can yield consistent treatment of all these effects. Coupling effects in terms of damping and current loading due to slender structures in general depend on floater type, water depth, mooring and riser system as well as the excitation level. Most pronounced coupling effects have been experienced for deep water systems in harsh environment. Some few examples are given in the following. 5.5 LF Wind and Wave, Surge/Sway Motions - FPSO The main findings from ref. /8/ for a coupled analysis case of a taut moored (3x3) FPSO in 4200 ft of water were that the damping contributions from the slender structures were around 40 % relative critical in extreme seastates (North Sea & Gulf of Mexico). For more operational seastates the damping from slender structures was around 25 % relative critical. This system included only 3 long wave risers, hence the main damping effect was from the mooring lines (90 % for extreme seastates). The mean drag loads on the slender structures was around 45 % of the mean environmental forces acting on the vessel. Risers contributed most due to larger exposed drag area than the taut mooring lines. It was also noted that the damping level increased due to the presence of the current. The main conclusion from this study was that coupled analyses represent a significant improvement when compared to traditional de-coupled analyses. The case study also revealed that with proper analysis strategy, the computation time for the applied software was acceptable for use in practical design (computation time = real time). The computational efficiency is essential, as it is possible, early in design, to come up with very good estimates of mooring and riser responses and how these responses are transferred into the turret and further into the hull structure. 5.6 LF Wind and Wave Heel Motions Spar De-coupled analyses were mainly applied in the design and verification of the first Classic Spar units in addition to model testing. Resulting heel motion of a Spar is of interest for topside, hull, mooring and risers, hence it is of importance to determine the resulting heel accurately. For these first

Classic Spar designs, typical calculated heel angles were around 9 10 degrees adding all static and dynamic components in a GoM 100 year hurricane environment. Later performed coupled analyses with FE representation of the slender structures have revealed reduced responses, giving resulting heel angles in the order 5 7 degrees. See also refs. /2/, /4/ and /6/. The main reason for this is the reduction in the LF response component. Damping contributions from the slender structures in the FE model is the main reason for this. These damping contributions could have been included in a de-coupled analysis based e.g. on input from model testing. It is, however, considered more efficient and consistent to obtain the results directly from a coupled analysis. 5.7 Heave Responses Spar The heave mode is of special interest for Spar units. This applies to traditional air-can supported riser systems as well as more advanced gimbaled, or tensioner supported systems. Heave excitation for Classic Spar units is minor due to the large draft (typically 650 feet), with the main heave excitation at the keel level. For the Truss Spar, the heave excitation area (hard tank) is closer to the waterline with much higher heave excitation forces. Heave eigenperiods for Spar units may vary from 16 30 seconds depending on global configuration, selected riser supporting system and area of operation. The heave eigenperiod is usually attempted kept outside the resonance area, however the level of heave damping will still be an important parameter to determine such that most accurate heave responses can be predicted. The main heave damping contributions for Spar units (Classic, Truss, Cell) comprise: Hydrodynamic potential damping (small) Viscous hull damping (strakes, trusses, damping plates) Viscous damping from mooring lines Viscous damping from risers Friction damping forces (hull/riser & tensioners) All these damping effects can be simulated in a coupled analysis. Fig. 2 is an attempt to visualize the different damping contributions and how these will influence the heave responses. This figure shows that the heave responses for a tensioner supported Spar unit is not very sensitive to increased heave damping once beyond a certain damping level. This is crucial information e.g. in relation to discussion of damping parameters like drag coefficients for mooring lines and risers, friction coefficients (keel/riser & tensioners), drag coefficients for damping plates and drag coefficients for trusses. The frictional damping simulation challenges the coupled analyses due to hysteresis effects (i.e. stick-slip behaviour) which has to be properly simulated with static and dynamic coefficients of friction in the algoritm. This software implementation usually implies a larger FE model and usually a requirement of smaller time steps in order to avoid numerical problems. Inclusion of friction simulation will typically increase the computation time.

OTC 16588

5.8 Tendon Responses TLP The responses of a TLP are generally separated into three frequency ranges, HF, WF and LF. The HF response will in this context imply springing responses which are important for the fatigue design of the tendon system. For the WF part there is usually good correlation with model test results. The main challenges with respect to TLPs and deeper water will be the LF and the HF responses. Studies performed within the DEEPER JiP (ref. /9/) for a 3-legged TLP in 1500, 3000 and 6000 feet of water concluded that the coupling effects are shown to be significant both for LF and HF motions for this TLP. The work focused on irregular sea (extreme and fatigue) and with correlation to model test results for the 3000 feet water depth. No extreme loop current was simulated. Both coupled and de-coupled analyses were performed. The following conclusions from this work were made: Coupling effects are significant for HF and LF TLP motions High resolution in FEM model of tendons needed Coupled analysis versus de-coupled was shown to introduce significant damping in LF surge and HF pitch. This trend increased with water depth. Coupled analyses increase HF tendon tension for fatigue waves and gives a more broad banded response. Coupled analyses decrease HF tendon tension for extreme waves. Motions and tendon tension dynamics from coupled analysis are qualitatively closer to model tests than decoupled. Current seem to have small influence on the importance of coupling. The HF tendon tension is sensitive to tendon material damping applied in the coupled analyses. The above and other data suggests that coupled analysis is the preferred analysis method when going to deeper water with TLPs. 5.9 SCR Responses Semi The use of steel catenary risers are increasing and can be accommodated on many types of floaters (Semi, Spar, TLP) and geographical areas. The Semi is among the most challenging in relation to use of SCRs. This is mainly linked to the level of motions, especially in the vertical plane (heave, roll, pitch). For SCRs, fatigue is usually the critical design mode with most exposure in the touch down area and the hang-off at the floater. Vortex Induced Vibrations (VIV) is a central issue for fatigue design of SCRs, but not dealt with herein. As already noted, WF motion is insignificantly altered when applying coupled analyses. Any benefit from using coupled analyses will then be linked to the level of LF motions (all 6 DOF). Small Semis will generally have minor slowly varying motions, so the challenge is with larger production Semis where the experience show LF roll/pitch motions at the same level as the WF roll/pitch motions.

The use of coupled analyses for design of SCRs should be considered attractive as one will be able to have both the LF and the WF time domain motions and SCR responses consistently calculated early in the design process. Also to be noted that the LF motions/responses will contain a gust wind induced component at the same importance level as the wave drift component. 5.10 Slender Structure Nonlinearities Despite the differences in design, function and application area for the slender structures (top tensioned riser, compliant risers, fluid transfer lines, mooring lines, tendons) physical behaviour and governing parameters for the response characteristics are quite similar. Such structures are commonly also termed as tensioned structures to reflect that the effective tension is the overall governing parameter for the global configuration (i.e. geometry) and transverse stiffness. An important consequence is that a common overall analysis framework can be applied in load effect analyses of slender structures. Mooring lines and cable/chain systems are not influenced by bending stiffness at all. The other systems have a physical bending stiffness that should be considered in the load effect analyses. A basic understanding of important non-linearities of slender structures is of vital importance for system modelling as well as for selection of adequate global analysis approach (e.g. FD versus TD analysis). Non-linearities will also be decisive for the statistical response characteristics for systems exposed to irregular wave loading. An essential issue is how non-linear properties of the slender structure and hydrodynamic loading mechanisms transform the wave frequency Gaussian excitation (i.e. waves and 1st order floater motion) into nonGaussian responses. Important non-linearities that always should be carefully considered can be summarised as: 1) Geometric stiffness (i.e. contribution from effective tension to transverse stiffness). Tension variation is hence a non-linear effect for slender structures. 2) Hydrodynamic loading. Non-linearities are introduced by the quadratic drag term in the Morison equation expressed by the relative structure-fluid velocity and by integration of hydrodynamic loading to actual surface elevation. 3) Large rotations in 3D space. This is relevant for systems with bending stiffness undergoing two-axial bending. 4) Material- and component non-linearities. 5) Contact problems in terms of seabed contact and hull/riser contact (varying location of contact point and friction forces). The relative importance of these non-linearities is strongly system and excitation dependent. Non-linearities due to item 1) and 2) will, at least to some extent, always be present. Item 3) is relevant for systems with bending stiffness undergoing two-axial bending due to in-plane as well as out of plane excitation, while 4) and 5) are more system specific non-linear effects. Material non-linearities are of special importance for flexible risers and umbilicals (e.g. hysteretic bending moment/ curvature relation due to interlayer stick/slip behaviour) and

OTC 16588

synthetic mooring lines (axial force/elongation hysteresis). Component non-linearities are experienced for several riser system components such as flex-joint, tensioner, bend stiffener etc. It should be noted that external hydrostatic pressure is not considered to be a non-linear effect as hydrostatic pressures normally will be handled by the effective tension/effective weight concept in computer programs tailor made for slender structure analysis. 6. De-Coupled Floater Motion Analysis Floater motions can be split into LF, WF and HF motion components. For the different floater types, certain hydrodynamic effects from various sources are important and must be taken into account in the analysis and design. Many of these hydrodynamic effects (e.g finite wave amplitude, moonpool, VIV, air gap, slamming, green water, viscous roll damping, ringing, springing) are of higher order and a decision has to be made on whether to use frequency domain (FD), or time domain (TD) method in calculating the global floater responses. Many of these effects will usually not influence the floater motions and can be disregarded in coupled analyses. In comparison with FD analysis, the advantage of TD analysis is that it can capture higher order effects without linearizing the problem. In addition, it estimates the maximum responses without making assumptions regarding the response distribution. 6.1 Load Models Load models will typically include hydrostatic effects, wave loads (WF, LF, HF), wind loads, current loads, hull VIV effects, moonpool effects and Mathieu Instability. Further details on these effects may be found in the Guideline, ref. /1/. 6.2 Wave Load Simulation Usually a FD motion analysis (radiation/diffraction analysis and e.g. Morison loading) is the basis for generating transfer functions for frequency dependent wave excitation forces (1. and 2. order), added mass and damping (potential & viscous). In a frequency domain analysis the equations of motions are solved for each of the incoming regular wave components for a wave frequency analysis, and for each of the sum- or difference- frequency combinations for a 2. order analysis (HF, and/or LF response). In relation to determination of wave drift surge/sway excitation for FPSOs, some caution has to be shown when using traditional radiation/diffraction analyses. In ref. /10/ it is documented a relatively large underestimation of the surge drift force on a FPSO exposed to extreme weather. This correlation was based on using results, empirically found from model tests. This observed deviation was assumed to be due to bow effects due to large pitch motions. Also to be noted that the correlation (empirical/predicted) was better for operational conditions. Another example emphasizing the importance of the loading side for coupled analysis is the use of Newmans approximation versus the full Quadratic Transfer Function (QTF) is visualized in Fig. 3 and Fig. 4. This example is with a minimum facilities type surface piercing circular cylinder, connected to a circular, larger diameter shallow base.

Depending on the LF damping levels to be used in these two sets of analyses, differences in wave induced LF heel motions in the order of 50 100% may be obtained. 6.3 Slender Structure Representation In a de-coupled analysis the restoring from mooring lines and risers is normally represented in terms of tabulated quasistatic restoring force as function of offset. This information is used as look-up tables for restoring forces for a given floater position in the de-coupled analyses. It is important to note that slender structure dynamics are not included in de-coupled analyses. This implies that the damping of LF floater motions due to the slender structure dynamics are not included in de-coupled analyses. This effect is important and must be included for deep water concepts. Furthermore, current loading on slender structures is usually not included in the restoring characteristics. The total force on the floater from current loading on the slender structures can be substantial for deep water systems. Seabed/slender structure frictional forces can not be represented by use of restoring force characteristics since they depend on the displacement history. 6.4 Coupling Effects in De-Coupled Analyses Separate assessments of slender structure/floater coupling effects are required due to simplified representation of slender structures in de-coupled analyses. The force on the floater from current loading on the slender structures can be assessed by static analyses using a standard slender analysis computer tool, or by coupled model of the total system. The contribution from current loading on the slender structures can be applied directly as an additional force on the floater in the de-coupled analyses. LF slender structure damping can be assessed based on model testing, or coupled analyses. It is emphasized that the LF damping from the slender structures for some systems are sensitive to the environmental excitation level, implying that different damping level applies e.g. for extreme and fatigue seastates. 7. Global Slender Structure Analysis This section gives an introduction to the slender structure analysis using floater motions predicted by coupled, or decoupled analyses as forced boundary conditions. The slender structure analysis methodology is also a cornerstone in the coupled analysis approach as will be discussed later. The main difference between mooring lines and risers is that the latter has bending stiffness while the former has not. Risers and mooring lines show close similarities with regard to staticand dynamic global behavior. The same overall analysis strategy is therefore applicable to mooring lines and risers. The purpose of global riser system analyses is to estimate the overall static and dynamic structural behaviour by exposing the system to a stationary environmental condition. A global cross sectional description in terms of resulting force/displacement relations (axial force versus axial elongation, bending moment versus curvature, and torsional

OTC 16588

moment versus twist angle) is applied in such analyses. Relevant global response quantities can be grouped as follows: Resulting cross-sectional forces (effective tension, bending moments, torsional moment) Global deflections (curvature, elongation, angular orientation) Global position (coordinates, translations, distances to other structures, position of touch down point on seabed, etc) Support forces at termination to rigid structures (resulting force and moments). These response quantities are given directly as output from global riser analyses. Subsequent detailed cross sectional analysis to determine local stresses and strains can be performed using resulting cross sectional forces from the global analysis as boundary conditions and including possible external/internal pressure loading. 7.1 Finite Element (FE) Approach A FE approach is normally considered for global riser analysis. The most important features required for adequate modeling and analysis of deepwater systems can be summarized as: 3D formulation Consistent treatment of finite rotations in 3D space (relevant for beam elements used for modeling of riser cross sections) Conventional small strain slender beam and bar elements including material and geometric stiffness allowing for nonlinear material properties. Seabed/riser/mooring line contact formulation Hull/riser contact formulations Structural damping formulations Hydrodynamic loading according to the Morison equation expressed by the relative water/structure velocity and acceleration. Regular and irregular loading due to waves and floater motions Current modeling. Special features allowing for efficient modeling of components such as swivels, hinges, buoyancy modules, clump weights, flex-joints, etc. Nonlinear static analyses Eigenvalue analysis Nonlinear time domain dynamic analysis 7.2 Dynamic FE Analyses The treatment of non-linearities is the distinguishing feature among available analysis techniques. Due to the nonlinearities, the response characteristics of riser systems are in general non-Gaussian. Consequently, time domain analyses are the primary method of analysis, especially for the purpose of prediction of extreme responses. 7.2.1 Nonlinear Time Domain Analysis Non-linear time domain analysis is based on a step by step numerical integration of the incremental dynamic equilibrium equations. A Newton-Raphson type of equilibrium iteration is applied at each time step. The non-linear approach will give an adequate description of all non-linear effects and will

consequently give a good representation of a non-Gaussian response. Non-linear simulations will typically be needed for systems undergoing large displacements, rotations, and tension variations, or in situations where description of variable touchdown location, or material non-linearities is important. 7.2.2 Linearized Time Domain Analysis Linearized time domain analysis is based on linearization of the dynamic equilibrium equations with regard to stiffness, damping and inertia forces at static equilibrium position (i.e. structural linearization). This means that the system stiffness, damping and mass matrices are kept constant throughout the analysis and that the system displacement vector can be found by a simple back substitution at each time step. Nonlinear hydrodynamic loading according to the Morison equation is, however, still included. The linearized approach is far more efficient than nonlinear analysis and is hence an attractive alternative when hydrodynamic loading is the major nonlinear contributor. A typical application is analysis of tensioned risers with moderate transverse excursions. 7.2.3 Frequency Domain Analysis Frequency domain analysis is based on linearization of stiffness-, damping-, inertia- and external forces at static equilibrium positions (i.e. structural and load linearization). Stochastic linearization for combined wave/current loading is required for irregular analysis. Frequency domain analysis will always give a Gaussian response and is therefore not recommended for extreme response prediction. The main application area is fatigue calculations and long-term response statistics to identify design conditions to be applied in time domain analyses. The computation time is short when compared to time domain analyses. A more appropriate alternative could be the non-linear frequency domain approach. See ref. /3/ for more details. 8. Coupled Analyses The floater, risers and mooring system comprise an integrated dynamic system responding to environmental loading due to wind, waves and current in a complex way. The response components of the floater will also be present in the slender structure responses. The overall purpose of global performance analyses of offshore floating systems is accurate prediction of floater motion, as well as mooring and riser system responses with due regard of floater/slender structure coupling effects. A coupled dynamic model of a floating system can in principle be obtained by introduction of the rigid body floater model in a FE model of the complete mooring and riser system. Such models can be quite complex, and a masterslave approach is an efficient technique for connecting relevant mooring lines/tethers/risers to the floater. The solution of this coupled system of equations in TD is obtained using a non-linear integration scheme. This will ensure consistent treatment of floater/slender structure coupling effects, i.e. these coupling effects will be automatically included in the solution. Normally the floater model for coupled analysis is identical to the floater model for the de-coupled analysis approach.

OTC 16588

8.1 Coupled System Analyses Floater responses as well as detailed mooring and riser responses can be computed by coupled analysis using a detailed model of the total system. This approach is usually termed Coupled System Analysis reflecting that all relevant system responses are computed directly by the complete coupled response model. This approach is mostly suitable for simple systems where adequate mooring line and riser responses can be predicted by fairly simple FE models. For more complex systems, with large number of finite elements, computational efficiency may be achieved by Selected Modelling. In this approach, only identified critical slender structures are modelled in detail. For other non-critical ones a rather coarse model is applied to reduce the total number of elements and thus the simulation time. This model will behave as a coupled system analysis as described above for the selected critical slender structures, as well as the floater motion.

Traditional assumptions can alternatively be applied considering WF floater motion as dynamic excitation while LF floater motions are accounted for by a representative offset (branch b of Fig. 5). The slender structure is consequently assumed to respond quasi-statically to LF floater motions. In case of TLPs, the flowchart in Fig. 5 could have been extended to include the HF springing components.

8.2 Coupled Floater Motion Analyses Several strategies can be proposed to achieve computational efficiency. All strategies have in common that the floater motion analysis and slender structure analysis are carried out separately (with the exception of selected modelling stated earlier). The first step is always a floater motion analysis. Computed floater motions are then applied as loading in terms of forced boundary displacements in subsequent slender structure analysis. Critically loaded risers and/or mooring lines can then be analysed one by one in the slender structure analyses. This scheme contributes to computational flexibility as well as a significant reduction in computation time. A flow chart of this approach is shown in Fig. 5. By proper modelling, this approach is capable of predicting floater motions and detailed slender structure response with the same precision as the Coupled System Analysis. The main objective of Coupled Floater Motion Analysis is to give a good estimation of floater motions. Detailed slender structure response is secondary. It can therefore be proposed to apply a rather crude slender structure FE model (e.g. crude mesh, no bending stiffness in risers, etc.) in the coupled analysis still catching the main coupling effects (restoring, damping and mass). A crude mesh for a FPSO mooring system in e.g. 1300m water depth could imply mooring line element length of up to 430m (= 1/5 of total length), still capturing the floater motions. Assessments of the maximum and minimum mooring line tensions, however, would require that segment length need to be less than 45 m would be required for this system (see ref. /8/). Different alternatives for interfacing coupled motion analysis with subsequent slender structure analysis are shown in Fig.5. The most direct way is to apply time series of floater motions (combined WF and LF motions) computed by the coupled floater motion analysis as boundary conditions in the slender structure analyses (branch a). This approach will also capture possible LF slender structure dynamics as well as the influence from LF response (possibly quasi-static) on the WF response. Such effects may be of importance for some deepwater mooring line and riser systems.

8.3 Modelling Considerations The primary requirement to the slender structure models is to give an adequate representation of the coupling effects. It is also often desirable to establish some key results for the mooring and riser system directly as output from the coupled floater motion analyses. Such information can be used to identify critically loaded slender structures for subsequent detailed analyses. In most cases, it is convenient to include all mooring lines, tethers and risers in the FE slender structure model. The FE model of each slender structure is simplified by using a rough mesh and omission of bending stiffness for most parts of the riser system. This will allow for output of key slender structure responses (mooring line tensions at fairlead, riser top tensions, tensioner stroke, etc.) directly from the coupled floater motion analysis. More detailed riser responses using a refined FE model is carried out separately in dedicated riser analysis to save computation time and increase the analysis flexibility. Examples being modelling of special components such as taper joints as well as refined mesh for adequate calculation of moment, shear and curvature in critical areas (e.g. touch-down and hang-off areas for SCRs). Further simplifications are possible if the primary objective of the coupled analyses is the prediction of floater motion. This may involve the use of equivalent models for a group of mooring lines, tethers and risers. This will provide less information on slender structure responses, but still be adequate to capture coupling effects.

9. Influence of Analysis Parameters The Guideline (ref. /1/) provides a number of examples on how variations on input parameters will influence the key responses for the 3 Theme Structures. This will not be repeated in this paper. Instead some general comments are provided with respect to which parameters may be considered in relation to potential sensitivity studies. In general the range of variability to be checked out with sensitivity studies will vary from typically 20% and up to 50% and maybe even higher (e.g. in relation to friction coefficients). Some judgement has to be made based on previous industry practice/knowledge, model testing/validation and if possible, full scale correlations. If more refinement is needed in evaluation of which parameters are most critical/important for the actual design, a structural reliability analysis (SRA) may be proposed.

OTC 16588

Excitation Forces
Wave headings Wave height/-period Wave spectra/parameters Wind spectra shape/parameters Combination of wind generated wave & swell Finite wave effects (above/below mean water line) Newmanss approximation versus full QTF (slow drift responses) Slamming effects * Green water * Drag coefficients for hull, moorings, tendons, risers, damping plates, strakes. Drag coefficients when VIV is present Disturbed fluid kinematics (inclusive moonpool kinematics) * Added mass for hull, slender structures, damping plates.

Stiffness/Mass/Buoyancy
Metacentric height (e.g. free surface effects) Mass and buoyancy data/variable load conditions Tension systems/hysteresis effects Mooring stiffness (e.g. non-linear stiffnesses for fiber ropes) Open/closed moonpool (hydrostatic stiffness)

Damping
Drag coefficients (see Excitation Forces above) Viscous hull damping Friction coefficients (inclusive hysteresis) Aerodynamic damping *

General
Time domain versus frequency domain analyses Number/type of risers installed Simulation time, time step & seeds Combination of responses (HF, WF, LF).

10.2 Long Term Response Statistics The long term load effect distribution is a result of the combined wind, wave and current loading on the coupled floater/slender structure system, i.e. a probabilistic description of the response from the long term environmental loading. This long term environmental loading process can be divided into a number of short term stationary conditions. It is further assumed that each short term condition can be completely described by a limited number of environmental parameters. The main challenge with this approach is to establish the short term load effect distribution as non-linear irregular time domain analysis will be required to give an adequate description of the response process. The response surface approach is a direct numerical approximation to the long term distribution of load effect peaks. The computational efforts involved in such techniques are to establish the short term distribution of load effect peaks. Interpolation and extrapolation techniques can thereby be applied to establish the short term distribution for all relevant environmental conditions required for assessment of the long term load effect distribution. The response surface approach will hence enable the computation of the long term load effect distribution considering a possible non-Gaussian short term load effect characteristics. For practical application, it is however crucial that acceptable precision can be obtained by use of relatively few basic seastates (e.g. 5 or less). 11. Conclusions The move of the offshore industry into deeper water and the introduction of new floating concepts and systems, challenge the capability to simulate the behaviour of these integrated systems. This paper focus on the analysis part and describes how efficient coupled and de-coupled, time domain analyses can be performed. Coupled analysis is argued to be the ultimate method, with the possibility to capture all significant coupling effects. De-coupled, time domain analyses with input from e.g. coupled analyses, model testing, or full-scale data is considered as an alternative if minor coupling effects are known to be present. 12. References
1. V. L. Hansen, N. Sodahl, L. Wang, E. G. Ward, J. Wichers, C. T. Stansberg, Guideline on Coupled Analyses and Model Testing of Deepwater Floating Systems, Deepstar Ctr 5401B. J. Wichers, P. Devlin, The Benchmark Model Tests on the Deepstar Theme Structures FPSO, Spar, TLP, OTC 16582 Y. Luo, S. Baudic, P. Poranski, Prediction of FPSO Responses Model Tests vs Analysis, OTC 16585 A. Steen, M-H. Kim, M.B. Irani, Prediction of Spar Responses - Model Tests vs. Analysis, OTC 16583 J. Zou, C. T. Stansberg, H. Ormberg, Prediction of TLP Responses Model Tests vs. Analysis, OTC 16584 E.G. Ward, V. L. Hansen, M. H. Kim, L. Wang, Model-theModel: Validating Analysis Models for Deepwater Structures with Model Tests, OTC 16586 C. T. Stansberg, S. I. Karlsen, E. G. Ward, J. Wichers, M. B. Irani, Model Testing for Ultradeep Waters, OTC 16587

* Usually not important for coupled analyses

10. Statistical Post-Processing The move into deeper water and new floater concepts challenge the assumptions of linearity, stationarity, Gaussianness and narrow-bandedness of the responses. The results from the coupled analyses will typically contain time domain responses in 2 3 frequency bands (HF, WF, LF). Guidance on how to process the results/responses from coupled analyses is needed.

10.1 Short Term Response Statistics In a design storm approach, the load effect can be estimated as the expected- or most probable largest response peak for the specified duration of the design condition. The output from irregular time domain analyses is time traces describing one realization of the load effect. The probabilistic distribution of the load effect is in general nonGaussian. Extrapolation is often involved in estimation of the extreme load effect. The process of estimating extreme load effect from time domain analyses will typically involve the following: Envelope statistics Estimation of extreme values from non-Gaussian load effect time series Estimation of simulation length required to obtain extreme load effect estimates with sufficient statistical confidence. Further details on these issues can be found in the Guideline, ref. /1/.

2. 3. 4. 5. 6.

7.

OTC 16588

8.

Ormberg, H., Sodahl, N., Steinkjer. O., Efficient Analysis of Mooring Systems using De-Coupled and Coupled Analysis, Proceedings 17th OMAE, Paper 1460, Lisbon, Portugal, 1998. 9. DNV, MARINTEK: DEEPER Joint Industry Project. 10. C. T. Stansberg, H. Ormberg, O. Oritsland; Challenges in Deep Water Experiments Hybrid Approach, OMAE01/OFT-1352

QTF Newmans appr.

0.8

0.6

Heave RAOs
2,0
Spar TLP SEM I FPSO _head

0.4

0.2

0 25

1,5 heave ampl./wave ampl.

FPSO _beam

20

25
15

20
10

15

10
5

1,0

Fig. 3 LF Heel Excitation Newmans Approximation


0,5

0,0 0 10 20 30 Wave Period (sec)


1

Full QTF

Fig. 1 Typical Heave RAOs for Different Floaters

0.8

0.6

0.4

0.2

Typical Heave Damping/Response Curve


0 25
20
15
10
15
10

25
20

Tj

Ti

Heave Response

De-Coupled

Coupled wo/friction Coupled w/friction

Fig. 4 LF Heel Excitation Full QTF

Damping Relative Critical

Fig. 2 Typical Heave Damping/ Response Curve Spar

10

OTC 16588

Advanced vessel model Simplified slender structure model

Vessel Motion Analysis

LF & WF vessel motions

(b)

Select vessel motion representation

(a)

Establish representative offset (mean & LF)

WF &LF vessel motions

Vessel WF motion RAO

Advanced slender structure model of each riser & mooring

Slender structure analysis

Slender structure analysis

WF slender structure responses

WF & LF slender structure responses

Fig. 5 Flowchart for Coupled Floater Motion and Slender Structure Analysis

You might also like