L. De Amicis, G. Mahoney - Saipem
F. Grealish, A. Connaire - MCS
As the offshore industry develops deep and ultra deepwater locations, steel catenary risers (SCRs) or lazy wave SCRs often are the risers of choice. The requirement for more accurate SCR modeling is driven by the criticality of the SCR strength and fatigue characteristics. Increasing costs, complex metocean environments, demanding external turret-moored vessel motions, and corrosive internal fluids drive the performance requirements of hang-off and touchdown-point riser pipe and make for careful design procedures and more demanding modeling requirements.
For most South Atlantic ocean projects, the riser system is part of a SURF engineering, procurement, construction, and installation (EPCI) contract. The design used for the risers must be compatible with the schedule of the EPCI contract and should facilitate long-lead procurement items, qualification programs, and key interfaces with third-parties at an early contract stage.
To cope with the schedule and technical-economic constraints, Saipem has developed a rationalized and phased riser design approach that targets early definition of key responses and management of uncertainties, with focused design methods/tools to confirm key riser responses once input data is sufficient.
The design of SCRs connected to FPSOs by means of taper, flexible jumper, or other articulation often requires the use of advanced finite element (FE) techniques to accurately represent the complex interaction of the SCR with the environment and support structures.
The ability to develop and to apply advanced key riser designs plays a significant role in optimum project execution. MCS’ specialized analytical support and software development have made important contributions to the overall definition of advanced modeling techniques for riser design.
SCR design challenges
For SCR conceptual studies and projects to date, there are two critical design challenges: characterizing behavior at the vessel hang-off interface, normally by flexible joint articulation, and modeling the SCR interaction with the seabed. The response characteristics of the SCR at the vessel hang-off and at the seabed are complex, nonlinear phenomena, so it is important not to restrict the modeling to linear approximations.
A significant amount of work (still to be fully benchmarked against field-proven performance) has been done to develop advanced models at these critical interfaces, although most analyses apply linearized solutions to these nonlinear problems.
Riser design has two main phases. The preliminary design phase assesses extreme, survival, and fatigue conditions and identifies the critical and nonlinear responses at the vessel hang-off and at touchdown, where the behavior is affected by the pipe/soil interaction and articulation stiffness.
Both responses strongly depend on vertical motion of the vessel at the hang-off. Dedicated software can screen complex swell and local sea wave data, FPSO RAO response, as well as quasi-static effects to identify the critical loads characterized by wave frequencies, directionality, and vessel drafts. Consequently, the feasibility of the proposed SCR configuration can be confirmed by verifying the peak riser stress/strain levels. Moreover, the calculated extreme, survival, and fatigue loads, plus adequate margins to take into account the known uncertainties, determined at this stage are used as input to early interfaces such as procurement of flexible joint and linepipe, welding qualifications, and FPSO interfaces.
In particular, the preliminary phase requires that flexible joint design agree with the manufacturer on minimum and maximum allowable quasi-static stiffness and permissible dynamic factors over the full range of dynamic angles/frequencies. These requirements, together with adequate margins and contingencies, form the basis of the preliminary phase data. These requirements are confirmed later in the project execution by FE and factory acceptance testing.
Preliminary design generally considers unburied pipe conditions with upper-bound linear soil stiffness for maximum trench depth. For extreme environments, the dynamic bending moment is amplified to account for overburden and suction. Soil data interpretation is done in the preliminary phase so the modeling assumptions can be confirmed during the detail design.
The detail verification phase follows the preliminary phase and aims to confirm the design margins and interface data by analyzing all loadcases on the basis of the final data. Attention is focused on the key issues of the SCR behavior highlighted in the previous phase. The verification phase also incorporates feedback from the material and welding qualifications trials.
Based on previous project experience in deepwater South Atlantic environments, the critical loadcases with respect to pipe stress levels, welding performance requirements, and connector performance are such that advanced FEM techniques are required to confirm the design assumptions in the preliminary phase.
Where more flexible joint modeling accuracy is needed, a nonlinear dynamic rotation/bending moment response using MCS’ advanced nonlinear software can analyze the critical loadcases. With a similar approach, the effect of trenching and dynamic embedment in extreme and fatigue environments is considered for selected cases using non-linear springs/overburden forces to model the pipe soil interaction.
Advanced joint modeling
A flexible joint incorporates alternating laminations of spherical rubber and steel components within a steel support structure that includes an extension for welding to the SCR. The rotational stiffness of the flexible joint is nonlinear with different behavior under quasi-static and dynamic loading, and typically is described by a rotation-moment curve.
Industry standard time-domain FE software tools can model a nonlinear quasi-static stiffness but are not adapted to model the combined quasi-static and dynamic stiffness interaction of the flexible joint. Therefore the current conservative modeling practice for predicting SCR response is to model linear stiffness expressed as a factored quasi-static value corresponding to 1X standard deviation of the rotation (ϕST) from a three-hour irregular sea analysis.
The more advanced method developed by MCS uses histograms of joint rotations to determine an equivalent linear stiffness that gives the same fatigue damage as the fully nonlinear model. This entails determining a factor that can be used to select the flexible joint linear stiffness, as follows:
- Determine the equivalent flexible joint rotation (ϕLS) using the nonlinear flexible joint rotation-moment curve and compare this to the standard deviation of rotation (ϕST), to give the following selection factor: Fϕ = ϕLS / ϕST
- From this, determine the equivalent linear flexjoint stiffness that gives the same damage in a fully nonlinear mode. Those results can be used to select the most appropriate linear flexible joint stiffness across a broad range of seastate simulations.
A comparison of traditional versus advanced methods to predict SCR response and fatigue life in the critical region of the flexible joint shows that the more accurate advanced method predicts 20% higher fatigue life. However, demonstrating the feasibility of the advanced modeling requires effort and, therefore, the application of the methodology to the critical loadcases only, as identified as part of a phased design process, is the optimum trade-off with respect to the progress of detail design.
Advanced soil modeling
When lifted from the trench, an SCR experiences loads from the weight of typical clay soil backfill, suction, and adhesion.
A piecewise linear model shows how the suction force is a function of the remolded soil shear strength, plus pipe diameter, bearing width, and depth. The piecewise linear model uses three linear components to describe suction mobilization; maximum sustained uplift force during a plateau phase, and suction release resulting in breakout of the SCR. These components are defined by the maximum breakout force and break-out displacement.
Adhesion results from contact with the trench sides or during vertical passage through the flowing soil, and may exist over the entire depth of the trench.
The preliminary design phase considers a flat seabed with linear stiffness to model the pipe/soil interaction in both fatigue and extreme conditions. For extreme conditions, a factor on the dynamic bending moment (typically 1.25) is applied to account for suction effects. A factor closer to unity is used for fatigue conditions. Validation/optimization of these assumptions is reached with advanced nonlinear pipe soil interaction modeling of key loadcases, and it is part of a typical detail design phase.
An advanced time-domain FE formulation, with the capability to model an arbitrary seabed profile, elastic seabed stiffness, and implicit nonlinear suction with and without hysteretic-type load path, is used to model the SCR pipe/soil interaction for key cases in detail design phase.
The nonlinear suction force can be modeled using springs following a force deflection curve. To define the pipe/soil interaction when the pipe moves down into the trench, a feature in the MCS Flexcom software applies the suction force until the displacement of the node is equal to the breakout displacement. Following breakout, the spring is not active until the node again contacts the seabed, in order to represent the hysteretic type load path. The accuracy of the modeling is increased step by step, moving from the simple approach to the hysteretic-type load path.
For fully developed trench profiles, typically up to five times the pipe outer diameter, irregular time domain analysis shows that the standard deviation of bending moment increases with severity of seastate and depth of entrenchment, i.e. level of pipe soil interaction. The results indicate that a factor of 1.25 is consistent with accurate modeling of severe trench conditions and also suggests that the traditional approach of ignoring complex soil/pipe interaction in the assessment of fatigue is justified with a typical deep trench, considering the overall level of uncertainty in environment modeling, FPSO motions, and fatigue performance.
Hysteretic-type load path models show a trend that suggests the method with implicit definition of the springs considering the hysteretic effect generally yields lower moment amplitudes. High levels of suction and overburden increase the bending moment. Hence, the increase in standard deviation of bending moment at the departure point of the trench may exceed the case with flat seabed and no suction consideration.