INSTALLATION OF LARGE SUBSEA PACKAGE

Transcription

1 INSTALLATION OF LARGE SUBSEA PACKAGE François Pétrié (Océanide), Benjamin Rousse (Océanide), Christophe Ricbourg (Stat Marine), Bernard Molin (Ecole Centrale Marseille), Guillaume de Hauteclocque (Bureau Veritas) ABSTRACT Deepwater field developments are usually associated with large submarine package to be installed such as suction anchors, manifolds and other subsea equipment. Key data for the lowering operation engineering are the slamming, added mass and drag coefficients of the package. These hydrodynamic values are not well known. They are usually considered as a constant whereas they are depending on many parameters such as crane motion amplitude and period, package porosity... This lack of data may lead to over-design the lifting equipment and/or to over-restrain the operating environmental conditions. Consequently, this could have a significant impact on the cost & planning of the operation. This paper presents the main results of the recent work that has been performed for the joint industry project Offshore installation of heavy package, in 28 and early 29. The objective of this program is to establish more refined hydrodynamic coefficients and, if needed, to propose a detailed methodology for the engineering of such operation. In particular, model tests results are compared to DNV recommended practice. 1. INTRODUCTION The aim of this document is to present the results from the French joint industry project Installation of Large Subsea Package. This research project is leaded by Océanide, in partnership with Ecole Centrale de Marseille, Stat Marine and Bureau Veritas. It is sponsored by Total, Technip, Doris Engineering and Saipem. 1/19

2 After a brief analysis of the large subsea package installation challenges, some of the project results are presented. First, the selection of the studied cases is briefly introduced. Then, devices and results from the model tests performed in the offshore tank basin BGO FIRST located at La Seyne sur mer, France are detailed. Finally, theoretical model for hydrodynamic coefficients calculation and design recommendations for heavy lifting are proposed. 2. Subsea package installation challenges During the installation of offshore fields, very large subsea packages are laid down on the seabed. Some of these packages are over 1 ton. Those operations are completed from specific vessels equipped with heavy lifting devices. With the growth of the packages and the very deep water depth of new fields, the engineering offices are often at the limit for lay down design. This comes from the lack of data on hydrodynamic coefficients used for the design. Those hydrodynamics data are the added mass, the radiation and quadratic damping and the slamming load in the splash zone. This lack of data leads to over-conservatism and so to increase the lifting device capacities or to reduce drastically the operating environmental conditions. The consequence is an important operating cost impact and a planning drift. Packages shapes Large packages can generally be classed into two main shapes: Cylindrical shape: mainly suction anchors. It is a cylinder shape opened at bottom and partly opened at top with events. The size and the number of the events impact the hydrodynamic characteristics of the package. Parallelepiped shape: number of subsea equipment can be mentioned (manifold, flet, sled ). They are usually supported by a mudmat for geotechnical stability. Due to its shape and its size, the mudmat is often the preponderant component of the overall hydrodynamic loads. It is usually perforated to let the water escape during the seabed laying. 2/19

3 Installation phases Water entry The water entry is a critical phase for object laying. The package is suspended to the lifting vessel crane. During the lowering through the free surface and the wave zone, the package buoyancy and the slamming loads can destabilize the equilibrium. During this phase, cable slack events have to be avoided. When the package is immerged, hydrodynamic damping is influenced by the free surface proximity. Lowering object in infinite water The lowering phase is continuous. There is no brutal equilibrium change like for water entry or seabed laying. In very deep water, the design of this phase shall take into account the dynamic answer of the system cable+package and check that the natural periods of the system are away from the wave periods on site. Lowering close to seabed The problematic is the same as for the infinite water lowering. The only change is that the seabed proximity influences the added mass and the hydrodynamic damping of the system. 3/19

4 3. Studied cases Context Installation design of large structures requires knowing the hydrodynamic parameters like the added mass, the added damping but also the slamming loads. Depending on the installation phase (see above), these parameters may vary with the motion of the structure, with the waves parameters, with the distance from the free surface or from the seabed, but also with other parameters like the porosity of the structure and its inclination. As several studies were conducted in the past to characterize the hydrodynamic behaviour of suction piles, it was decided at the beginning of the present research project, to focus on the large parallelepiped shaped structures and especially on the perforated mudmats with a skirt, as it is usually one of the main components of the hydrodynamic loading. L = 1.5 x l l h = l/8 Skirt The parameters of the tests were defined as described hereafter. Wave conditions: Three sea-states were considered: Hs = 1m, Tp = 4s Hs = 1.5m, Tp = 6s Hs = 1.5m, Tp = 8s where Hs is the significant wave height and Tp the peak period. On installation site, the peak period may be larger, in particular in West Africa. But, for long wave lengths, motions of the installation vessel are more or less in phase with the swell. These long wave conditions are generally not critical regarding the crane tip motions and thus were disregarded. 4/19

5 Lowering velocity: Typical lowering velocity recommended by DNV is.5 m/s (crane hook velocity). Practically, the crane operator adjusts this velocity depending on the observed situation for each phase. At the splash zone crossing, the lowering may be stopped when the mudmat is just above a wave crest, and then the crossing is started with the proper velocity, avoiding to have the mudmat crossing again the free surface but also to have snap loads. Hence, the lowering velocity may be different from.5 m/s, and this value is deemed to be an upper bound velocity. During the tests of splash zone crossing, two lowering velocities were used:.5m/s and a lower velocity. Mudmat motions: The motions of the mudmats are derived from the wave conditions and from RAOs of typical installation vessels (see figure below). Crane tip RAO of vertical displacement RAO amplitude (m/m) Wave period (s) Vessel 1 amplitude Vessel 2 amplitude Envelop (21deg heading) Vessel 1 phase Vessel 2 phase RAO phase (deg) The motions of the mudmat may depend on the depth, since the stiffness of the system is linked to the paid-out length of cable. Consequently, the period of the motion may be in the vicinity of the natural period of the system (cable, mudmat and hydrodynamics) or not. 5/19

6 Impact velocity: For the slamming, the impact velocity was estimated through the DNV formula. Two formula were used and the results analyzed (oldest one provided in the Rules for Planning and Executions of Marine operations dated 1996, and the recent Recommended Practice DNV- RP-H13). The highest value obtained is 3 m/s which accounts for several conservatisms and is consequently considered as an upper bound. Finally, four impact velocities where selected:.5, 1, 2 and 3 m/s. Porosity: Most of the mudmats made of perforated plates (and not truss) have a rate holes/plate (called porosity) designed for soil stability purpose. Nevertheless the porosity hydrodynamic impact is investigated. Three porosity rates are studied: plain mudmat (no porosity), low porosity (named porosity 1), high porosity (named porosity 2). 6/19

7 4. Model tests Model tests setup Model tests have been conducted in Froude s similitude. The model scale is 1/16 th. The model is composed of: a mudmat a 1D motion generator Each mudmat tested is composed of: a 5mm thickness steel plate perforated or not (model scale) 4 vertical skirts Mudmat 1 Mudmat 2 Mudmat 3 Figure 1: mudmat model The 1D motion to be imposed to the model is heave translation. The motion generator is then composed of: a vertical electric jack a vertical beam mounted on a trolley and guide rail equipped with one load sensor 4 load sensors between the mudmat model and the vertical beam an inclination assembly to allow the mudmat model inclination 7/19

8 Figure 2: overall assembly 15 5 Figure 3: inclination assembly Measurement The wave elevation is measured from a wave probe located at the same distance from the wave maker than the model. The model displacement is measured from the optical system without contact Krypton and its acceleration by an accelerometer. The load is measured in two areas: directly on the mudmat model and between the trolley and the jack. 8/19

9 5. Slamming Loads Slamming on still water For slamming on still water tests, the mudmat impacts the still water surface with a constant velocity. The results for these tests are summarized on the following figures for three inclinations (, 5, 15 ): Porosity deg 5deg 15deg Cs*S (m 2 ) impact velocity (m/s) Figure 4: slamming on still water porosity 1 Porosity 2 Cs*S (m 2 ) deg 5deg 15deg impact velocity (m/s) Figure 5: slamming on still water porosity 2 Those results lead to the following conclusions: Cs decreases with the porosity growth Cs decreases for inclined mudmat 9/19

10 Slamming on irregular waves With the aim to be more realistic, some tests have been performed on regular and irregular waves. Only the irregular wave tests are presented here. For irregular wave tests, the approach is different than for still water. The slamming load is treated as a statistical value. The equivalent of 1 hour tests (scale 1) has been performed. Around 4 slamming impacts have been measured for each test. The mudmat imposed motion is a harmonic oscillation at wave Tp, around the free surface mean elevation. For each water entry, the slamming load is measured. The measures of Cs for each impact during one test are presented on Figure 6 below: Porosity1 - Irregular Waves Tests Cs*S (m 2 ) impact velocity (m/s) Figure 6: Cs repartition deg inclination The Figure 6 gives the repartition of the Cs with the impact velocity. With these results, the Cs is characterized and its variation with the impact velocity can be taken into account for the design. 1/19

11 Design impact velocity In order to compute the design loads, a design velocity has to be associated with the slamming coefficient. Figure 7 sums up the velocities to take into account in the impact velocity calculation. Crane tip vertical velocity Lowering velocity Wave velocity Figure 7: Velocities v L : Lowering velocity v ct (t) : Crane tip velocity v ct : significant value v w (t) : Wave particle vertical velocity v w : significant value v (t) : Relative motion between the crane tip and the surface v : significant value The temporal impact velocity is: v ( t) = v ( t) v ( t) + v = v ( t) + v impact w cr L L In a design approach, an impact design velocity should be available from the significant value. An impact design value is thus defined as: Where (α ) v impact = v + C(α ) v C is function of the risk α to be taken. C(α ) Rayleigh law has been shown to be a conservative estimate). L is defined by the distribution of v (a The significant ative velocity v is often taken as the quadratic sum of v ct and v w : v ² = vct ² + vw² This implies that the wave velocity and the crane velocity are independent; moreover, this does not take into account the diffraction/radiation of the waves by the vessel. The link between v (t) and v ct (t) leads to overestimate the velocity for high period (the free surface and the crane tip are moving together). The lack of diffraction significantly overestimates the velocity in shielded area. (See Figure 8) 11/19

12 Figure 8: v around a supply for different heading and Tp (Pierson-Moskowitz spectra, long crested) As the impact forces are proportional to the square of the velocity, a better estimation of v can lead to significant differences in the hydrodynamic loadings. Figure 9 shows the evolution of v in the lowering area shielded by the vessel (3 ative to the wave). Vre (m/s) Exact Quadratic sum Tp (s) Figure 9: v 12/19

13 6. Added mass and damping Oscillation tests To measure the mudmat added mass and damping, oscillation tests have been performed with the following variable parameters: Amplitude and period of the harmonic motion Sea surface or seabed distance Porosity of the mudmat A total of 72 tests have been performed. The mudmat trajectory can be written as Z = A.sin(ω.t) (1) The hydrodynamic load is 2 m33 2 F( t) = ρaω sin( ωt) ρaω ρ b33 cos( ωt) ρω with ρ = water density A = harmonic motion amplitude ω = motion pulsation m 33 = added mass term b 33 = damping term. Drag, radiation and pressure loss through perforations (2) From the measured load during the harmonic oscillation tests, the added mass and damping terms are identified thank to a sliding Fourier analyse. This sliding Fourier analyse allows identifying the part of the load in phase with the velocity and the part of the load in phase with the acceleration (cosines and sinus term in equation Erreur! Source du renvoi introuvable.)). Those results (m 33 and b 33 ) are compared to theoretical ones in the next section. 13/19

14 Theoretical calculation Model description Figure 1 below gives the reference geometry: a perforated horizontal plate with vertical skirts. The problem is solved for axisymmetric (3D) or 2D geometries. The axisymmetric geometry is applicable for suction anchors or nearly square mudmats whereas the 2D geometry is applicable for elongated mudmats. Figure 1: geometry The structure has a harmonic imposed motion with ω the motion pulsation and A the amplitude. The problem is solved within the scope of linearized potential flow theory. The velocity potential Φ ( x, z, t) or Φ ( R, z, t) is written: Φ,, = R, Φ,, = R, (3) i i ( ) ( ) e t t x z t ω ω ϕ x z ou ( R z t) ϕ ( R z) e The fluid domain is split into three subdomains: Subdomain 1: outside, from R (or x) = a to infinity Subdomain 2: between the seabed and the perforated plate Subdomain 3: above the structure In each subdomain the velocity potential is written as an Eigen-function expansion. The pressure loss condition at the porous plate is written as a quadratic expression of the ative vertical velocity through the perforated plate: 1 τ p2 p3 = ρ ( Φ cos ) cos 2 z Aω ωt Φ z Aω ωt. (4) 2 µτ Here τ is the porosity, or open-area ratio, and µ a discharge coefficient, close to 1. The velocity potentials and normal velocities are then matched as the common boundaries. For details see Molin & Nielsen (24) [1] or Molin et al. (27) [2]. A no-flow condition is 14/19

15 written on the skirt. The pressure loss equation (4) is satisfied through an iterative procedure initiated from the solid case solution. Two porosity cases are studied at several motion amplitudes. We shall note that the calculations with the two porosities are redundant since the hydrodynamic coefficients only depend on the «porous Keulegan Carpenter» number defined as ~ 1 τ A c = (5) 2µτ a K 2 Calculations are first performed at mid-water, far from the sea surface. The hydrodynamic coefficients are then nearly insensitive to the oscillation frequency. In the figures, the added mass is divided by the fluid density and the damping by the fluid density times the frequency to be comparable to the added mass. If the forced motion is Z = Asin ωt, the hydrodynamic load is: 2 m33 2 b33 F( t) = ρaω sinωt ρaω cosωt (6) ρ ρω M33/ρ B33/ρω 5 4 M33 porosity 1 M33 porosity 2 B33 porosity 1 B33 porosity A (m) Figure 11: porous mudmat at mid-water 15/19

16 Axisymmetric (3D) versus 2D model The Figure 12 below shows the comparison between the axisymmetric and the 2D model. For the axisymmetric model, the diameter used is calculated to have the same area than the rectangular model. 7 6 M33/ρ B33/ρω M33 2D M33 3D B33 2D B33 3D A (m) Figure 12: Axisymmetric (3D) versus 2D comparison for porosity 2 mid-water These models are used to evaluate the m 33 and b 33 damping for several configurations. The main variable parameters are the porosity and the distance to the sea surface or to the seabed. The results from the calculation are compared to the model tests results. Implementation of the edge drag The model described above takes into account the flow separation through the perforations, idealized as porosity, but not the flow separation at the outer edges or at the skirt base. In Molin et al. (27) [2], an empirical correction of added mass and damping calculated is proposed, based on a drag term ated to the averaged flow velocity through the plate. This correction has been implemented in the axisymmetric and 2D models even if the pertinence of this method for structure equipped with skirts can be questioned. Based on works of Graham (198) [3] on plates, followed by Sandvik et al. (26) [4], the drag coefficient is ated to the Keulegan-Carpenter number by the ation: C D = (7) 1 3 α K / C 16/19

17 Comparison with model tests -Added Mass Ma/rho D 3D Model test Porosity 1 Model test Porosity KC porous The fact that Figure 13 : Added mass in infinite water K ~ c is the main parameters of interest is thus confirmed by the experiment. The calculation slightly over-estimates the results, the axisymetric model being closer to the experiment than the 2D model. Compare the added mass of the opaque mudmat without skirts; the added mass of a circular disk is 8% higher (45% for the 2D plate). Those differences are however not big enough to explain the difference between the model tests and the calculations. The difference is more likely due to the drag on the edge of the mudmat, which is not taken into account in the calculation on Figure 13. Figure 14 presents the results obtained taking into account an empirical correction for the drag on the edge. The agreement is slightly closer to the experiment (Note: K ~ c is not the only parameter anymore). 17/19

18 6 5 4 Ma/rho D porosity 1 alpha=4 3D porosity 2 alpha=4 3D without edge effect Model test Porosity 1 Model test Porosity KC porous Figure 14 : Edge effect on the added mass -Damping D calculation Model test, porosity 1 Model test, porosity 2 b33/rho KC porous Figure 15: Damping in infinite water Regarding the damping coefficient, the agreement with the experiment is not as good as for the added mass. K ~ c does not seem to be the only governing parameter. This is due to fact that the damping is more sensitive to the drag on the edge of the object. The results with the correction for the edge effects are presented on Figure /19

19 D porosity 1 alpha=4 3D porosity 2 alpha=4 3D without edge effect Model test, porosity 1 Model test, porosity 2 b33/rho KC porous Figure 16 : Edge effect on the damping Taking into account the edge effect thus allows a much more accurate calculation of the damping coefficient. However, the parameter alpha used for the correction is expected to be sensitive to the thickness of a plate (or skirt length). 7. Conclusions This research project has lead to a significant progress in the knowledge of hydrodynamic coefficients used for heavy lifting operations. This better knowledge will allow optimising heavy lifting operations and then to reduce their cost. We want to thank the sponsors of this project: Total, Technip, Saipem and Doris Engineering. A second phase of this project is under discussion for studying more package shapes. 8. References [1] B. MOLIN & F.G. NIELSEN 24 Heave added mass and damping of a perforated disk below the free surface, Proc. 19th Int. Workshop Water Waves & Floating Bodies, Cortona ( [2] B. MOLIN, F. REMY & T. RIPPOL 27 Experimental study of the heave added mass and damping of solid and perforated disks close to the free surface, Proc. IMAM Conf., Varna. [3] GRAHAM J.M.R. 198 The forces on sharp-edged cylinders in oscillatory flow at low Keulegan-Carpenter numbers, J. Fluid Mech., [4] SANDVIK P.C., SOLAAS F. & NIELSEN F.G. 26 Hydrodynamic forces on ventilated structures, Proc. 16th International Offshore & Polar Eng. Conf., San Francisco. 19/19