* pendulum condition with free hanging load

Transcription

1 OTC Computer Analysis of Heavy Lift Operations by H.J.J. van den Boom, J.N. Dekker, and R.P. Dallinga, Maritime Research Inst. Netherlands Copyright 1988 Offshore Technology Conference This paper was presented at the 20th Annual OTC in Houston, Texas, May 2-5, to copy is restricted to an abstract of not more than 300 words. The material is subject to correction by the author. Permission. - ABSTRACT - construction^ and Offshore platfoms are carried out in exposed areas and at growing Not Only entire sides can be lifted but the large jackets can now be carried out by vessels with... dual cranes. take advantage the large cost-savings related to such operations, the feasibility, workand risk level have to be evaluated accurateb. this end computer simulations of the behaviour of the total heavy lift system due to wind, waves, current and hoistinglde-ballasting are required. A general purpose degrees Of simulation model covering the mechanism of motion has been developed. The computer model is able to provide the motions Of the crane load and barge and the related tensions in hoisting wires, slings and mooring lines as well as the barge-load impact forces. special attention is paid to the sinulation of partly submerged jacket-type structures. he computer program has been validated with results of an extensive model test program. INTRODUCTION. v.... Large S-emi-submersible crane vessels (SSCV's) with lifting capacities of up to 14,000 tonnes are used nowadays for the installation of platforms in exposed areas* A considerable increase in load weights and volumes is expected because of the proven savings that: result from heavy offshore lifts. (see ref. [l.] ). By means of the dual crane option, it is not only possible to lift complete topsides, but also to per- References and'ilustratfons at end of paper. form installation of liftable jackets (Fig. 1). The conventional way of jacket platform installation is the transportation barge launch procedure. In general this launch causes an extreme loading on the structure. Furthermore buoyancy requirements result in large diameter tubulars and extra tanks. For upending ballast systems are needed. All launch related structural provisions increase the wave and current loading. It is therefore evident that the launch procedure has a strong impact on the design, engineering and construction of jacket-type Flatforms. A liftable jacket can be designed for loads exerted to the structure during its operational life time. Hence such jackets are outstanding from the point of view of design and construction work, use of steel and maintenance. It is generally accepted that liftable jackets are 30 to 40 percent more economic than conventionally launched jackets. Lifting operations at exposed locations with rapidly increasing weights, volumes and operational complexity cannot be performed on the basis of experience only. Evaluation of the feasibility and workability of such operations requires a detailed analysis of the system's dynamic behaviour as induced by waves, wind and current in combination with the hoisting action and the (de)ballasting procedures. For topside installation the following phases in the dynamic behaviour of the crane system may be distinguished: crane vessel in mooring system ;k side-by-side mooring of crane vessel and barge * pre-hoist condition with pre-tensioned slings * hoisting/de-ballssting with load/barge impacts * pendulum condition with free hanging load * positioning and lowering on support structures The most important items in the dynamic behaviour during these stages are the motions of the crane tips and load, the hoisting wire and sling tensions and the impact forces on the stabbing piles. In Fig. 2 typical time traces of the hoisting phase for a 8000 tonnes nodule are presented.

2 2 COMWTER ANALYSIS OF HEAVY LIFT OPERATIONS OTC! 5819 For jacket installation by means of a dual crane lift the follawing typical stages can be recognized: - lift-off immersion of jacket upending positioning * mating Besides the sling loads, impact loads on the barge and docking pile loads, for large jackets the relative motions (clearances) between the jacket and the crane booms or the hull can be of importance. Since the dynamic behaviour of the system is basically non-linear and non-stationar~, a Proper lifting analysis can only be performed using extensive time domain Simulation tools* For this Purpose MARIN, in close co-operation with Shell Internationale Petroleum Maatschappij (SIPM), has developed the LIFSIM simulation program, by which the behaviour of three bodies with eighteen degrees of freedom can be computed. The three bodies mentioned are a crane vessel, a load and a transportation barge or platform substructure. The elasticity of the crane systems, the hoisting action and the (de)ballasting of the crane vessel are included in the mathematical model. Any other linear or nonlinear mechanical interaction force between the bodies (for instance tugger lines or dynamic control algorithms) can be formulated by means of a user interface that also enables the user to specify additional output records. Recently the computer package has been extended for the analysis of jacket installation. The mathematical model can be outlined as follows: - the time-domain analysis with 18 degrees of freedam is based on the equations of motion according to Cumminsa with convolution integrals which cope with the frequency dependency of the added mass and damping coefficients of barge and crane vessel. - the environmental excitation forces may be arbi- trarys which means that first order wave forces, slowly varying drift forces, current and wind loading can be included. - since retardation functions and first and second order wave force records can be generated frm the results of frequency domain analysis, on a basis of three-dimensional potential theory, arbitrary hull shapes (ship, semi-submersible, catamaran) are allowed. - additional arbitrary forces may be specified as functions of all state variables [including time) through a special user program interface. - a change of mass and inertia as well as resulting forces and moments due to (de)ballasting of the crane vessel is taken into account. - the hoisting speed is optional to the user. - non-linear and asymmetric mooring characteristics can be dealt with. - the influence of the flexibility of the crane system on the motions is modelled. - impact loads due to strongly non-linear 'spring characteristics' of load/barge or loadljacket interferences can be handled. - fluid and windloading on jackets with arbitrary position and orientation are incorporated. To investigate dynamic effects during lift operations and to validate the LIFSIM-package, extensfve model experiments have been carried out in the MARIFf test facilities. These tests were sponsored by SIPM and HEEREMA [2]. COMPUTER MODEL gguations of motion The description of motions of the three body heavy lift system due to environmental loads, mutual mechanical interactions and the lifting operation is based on the impulse response theory to handle che fluid reactive forces. This description has proved to provide accurate motions and mooring forces for single body moored structures such as a jetty moored tanker (ref. [3]). X j,l.. t [(%j+ %j)~j+l%j(t-~) k(~)d~+c kj j X ]=, t) with k = 1,.., 6...( 1) where: = motion in the j-th mode arbitrarily time varyiog external force in the k-th mode of motion M = inertia matrix m = added inertia matrix R = matrix of retardation functions C = matrix of hydrostatic restoring forces. The left-hand side of eq. (1) contains the rigid body reactions and the linear hydromechanic reaction forces. The right-hand side contains the environmental loads due to wind, waves and current and other forces which may be arbitrary functions of the motions of the structure. When it is assumed that the interactions in fluid reactive forces are negligible and provided that non-linear drag terms, mechanical coupling due to the crane system and impacts and restoring forces due to mooring may be formulated as functions of the motions of the three bodies, eq. (1) for this system yields:... (5) +x~~e~z~>x-~... (2)

3 p -p--- 3 OTC ~ VAIT Dg BOOM, DALLINGA, -- & DEKKER ~ l I l in which: Mi = inertia + added inertia matrix of body No. i = matrix of retardation functions of body - NO. i Ci = matrix of hydrostatic restoring forces of body No. i X = motion vector of body No. i = vector of external forces on body No. i. -i order wave drift forces are computed in the form of frequency domain quadratic transfer functions based on the pressure integration formulation.. In order to be able to compute the first and second order wave loads on a vessel in the time domain for arbitrary wave conditions, use is made of the Volterra series formulation. According to this formulation, the total wave load, including first and second order contributions, follows from: In eq. (2) the indices 1, 2 and 3 stand for the crane vessel, the transportation barge and the load respectively. The coefficients of the matrices contained in the left-hand side of eq. (1) and (2), which describe the hydrodynamic reaction forces, are determined employing a three-dimensional, linear potential theory. In this method, the mean wetted part of the hull is approximated by a distribution bf source singularities. The formulation for the potential of the sources complies with the equation of continuity, the linearized free-surface condition, the bottom condition (if finite water depth is considered) and a radiation condition. The unknown source strengths are obtained by imposing the watertightness condition at the mean position of the centre of each source panel. The hydromechanic reaction terms are primarily obtained from potential theory as frequency-dependent added mass and damping coefficients. Based on these coefficients the added inertia matrix mkj and the matrix of retardation functions Rkj are obtained from:. 2 m...( k j kj R (t) = ; b (w)coswtdw in which ak (U), b.(u) = frequency-dependent added mass and da&ing ck~fficient matrices, respectively; and w* = a constant frequency. The advantage of describing the hydrodynamic reaction forces due to added mass and damping effects using retardation functions lies in the fact that arbitrary motions can be accommodated correctly, irrespective of the nature of the motions. Equations of motion based on the direct application of frequency-dependent added mass and damping coefficients have the disadvantage of not being able to cope with, e.g., transient motions that can arise during load-barge impacts. 3) i~~yhich = wave elevation time record and 4 (T),G'f[T T ~ = ) first and second order impulse response functions relating the wave elevation to the force in the k mode, respectively. The impulse response functions are found from: in which Hkl)(w) and H(Z)(wiw2) are the first order wave force transfer fkunction and the second order drift force transfer function, respectively. The advantage of the Volterra series approach is that simulation computations can be carried out based on measured wave records as well as on synthetic wave records. From the outline given above it follows that the simulation approach takes advantage of efficiently computed frequency domain results. For this reason large changes in draft, trim or heel are not accounted for. Wind and current forces are determined by the following empirical relationship: The right-hand side of eq. (1) contains, besides mooring or restraining forces, forces due to wind, waves, and current. Wave forces may be split into two contributions: first order forces that oscillate with wave frequency; and second order mean and slowly varying wave drift forces. First order wave forces are computed in the frequency domain using three-dimensional diffraction theory. The second in which: p = mass density of the fluid V current/wind speed Ak projected area for force in k mode C = drag coefficient for the force component dk in k mode.

4 I 4 COMPUTER ANALYSIS OF HEAVY LIFT OPERATIONS OTC 5819 Current drag force coefficients are best obtained from tests in large basins that allow relatively unrestricted flows in the case of oblique current angles. The influence of water depth is another factor of importance with respect to current forces. The wind force coefficients may be obtained from wind tunnel tests S empirical data bases or separate computer programs. The effects of wind gusts may be, incorporated by use of dynamic wind ~elociti'+s, derived from wind spectra. Crane szstem ---- The crane system consisting of crane(~), hoisting wires, hooks and slings, forms the mechanical connection between the crane vessel and the load. In order to derive the instantaneous mutual forces between these two bodies, the total crane system is considered as a set of linear springs thus neglecting dynamic behaviour of the crane system itself (Fig. 3). Based on the results of full scale load-deflection tests, the crane is schematized as an infinitely stiff beam rotating in a vertical plane around its heelpin due to a linear support spring. Transverse flexibility of the crane tip is represented by a linear spring. The hoisting wires and the individual slings are also represented by linear springs. From the instantaneous body positions the crane system forces are found. A continuous update of the hoisting wire length accounts for the hoisting action. I in the horizontal modes of motion, however, will be governed by the mooring system. Therefore the static characteristics of the mooring system are incorporated while neglecting dynamic behaviour of individual components of the mooring system. The mooring line tensions are found from the instantaneous position of the fair lead locations, and frm the static load elongation characteristics of lines. The direction of the line tension is taken equal to the direction between the points of attachment of the line. When the load suspended from the crane(s) is a partly submerged jacket-type structure, several additional items have to be accounted for. Large rotations require large angle directional transformations. The fluid forces have to be determined on the basis of the instantaneous posirion and orientation of the jacket. In this way the immersion is also incorporated. The fluid forces are partly of potential origin and partly due to viscous effects. Assuming slender tubular members without hydromechanic interaction, these forces may be approximated by the relative motion concept known as the Morison formulation: Impacts * Four 3-D sets of linear springs are chosen to model the impact forces between the suspended load 1 and the body 3. Since the geometry of the stabbing I points may be rather complex, gap width and depth may be specified for all impact points separately. The inertia and drag force shape coefficient (CI,CI)) for cylindrical elements may be formulated in normal and tangential components. For this reason the fluid forces are computed in a local system of co-ordinates. For several SSCV's the use of dynamic ballast systems is an important aspect of the lifting operation. Dynamic ballasting not only secures the even keel condition but it also may contribute to the actual lift offfon. Due to extreme large ballast volumes the mass properties of the crane vessel may also change considerably. For these reasons both changes in ballast mass alid forces are accounted for in the computer model. The restoring forces due to the mooring system are provided by mooring lines yielding tension forces. Due to its large volume, the dynamic behaviour of the moored structure at wave frequencies will hardly be affected by the mooring. The low frequency motion where : [R] = directional transformation matrix U = velocity vector in local coordinates - U = global velocity vector (G + % - ) V = current velocity = orbital velocity 7w J member no. n - total number of submerged members. D = diameter of member 1 = length of member. The instantaneous orbital velocities and acceleration~ of the water particles due to wave action are computed by means of a Volterra series formulation viz. the linear filtering of the given wave train:

5 I OTC VAN DEN BQO_MM, DALLIWA, & DEKKER 5 The models used during the test represented SSCV BERMOD, transportation barge H108, a rectangular module of 8000 tonnes and a lattice structure with a weight of 8000 tonnes at a displacement of 4000 tonnes. (Fig. 5) where: Hv(@> = V(U>/C~(~> W = wave frequency 3 (t) = input wave train. The transfer functions H (U) are readily found v from linear wave theory taking into account the position of the jacket member and integrating the wave action over the length of this member. The wind loads on the above water part of the jacket is also treated according to equation (9) Additional forces By means of a user's interface, additional forces on each of the three bodies may be specified as arbitrary functions of all state variables and time. In this way automatic control procedures, tugger winches, fenders etc. may be accomodated. Based on the mathematical description of motions and forces as outlined above, the computational procedure of LIFSIM as presented in Fig. 4, has been developed. As already indicated by eq. (2), the computer model describes the motions of the three individual bodies including the mechanical coupling due to the crane system and impacts. Computation of fluid reactive forces by means of retardation functions may be considered as linear filtering of the structure's velocity history. First and second order wave forces are obtained from linear and quadratic filtering of the input wave record. Knowing all forces on the individual bodies, the motions may be solved by numerical integration methods. The Runga Kutta algorithm has proved to provide efficient solutions for many situations. In case of heavy impacts the use of Gear's stiff integration method is required. VALIDATION -. In 1987 an extensive model test program was carried out to study dynamic effects in heavy lift operations and to evaluate the simulation program LIFSIM* The test series were carried out at scale 1:40 in the Seakeeping Basin of MARIN sponsered by Shell Internationale Petroleum Maatschappij and Heerema Engineering Services. The test program comprised single body motion tests, hydromechanic interaction tests, load-barge impact tests and hoisting/deballasting operations in irregular waves. During the test series 62 signals were measured. Motions of SSCV, crane tips, module and barge were measured by optical tracking devices, gyroscopes and polarisation filter techniques. Tensions in support wires, hoisting wires, slings and mooring lines were monitored by means of strain gauges techniques. The impact forces between the module and transportation barge were measured by means of 3-component force transducers integrated in the stabbing piles (Fig. 5). These impact forces were recorded with a sample rate of 15 Hz (full scale). Stationary tests Obviously the dynamic effects during a typical hoist period of 80 seconds strongly depend on the individual waves which pass the SSCV in this short time span. Because of the fact that the operation is carried out in irregular waves the level of impact forces and hoisting tensions are strongly related to the starting time of the operation. (Fig. 6). Although the wave excitation is of random nature, due to the hoisting/ballasting procedure the behaviour of the system is a non-periodic deterministic process. (Fig. 7). Contrary to an ergodic stationary process, a single realisation of a deterministic process does not result in valuable statistical information. In- order to derive statistical data for the hoisting operation two approaches may be followed. The operations may be repeated for many different starting times i.e. at different positions in the wave train. This so-called Monte Carlo approach will result in a distribution function for the required value e.g. the extreme impact force. The alternative approach is based on the assumption that the hoistinglde-ballasting action itself will not contribute to the dynamic behaviour of the system. Hence this deterministic component of the process may be treated quasi-statically. The operation is therefore studied by a number of stationary irregular wave tests. The load during such a. test is kept at a chosen pre-tension level or clearance from the supports. The same irregular wave test is repeated for other pre-tension levels and clearances. Each test can be considered as an ergodic stationary random process and can therefore be subjected to statistical analysis. In the model test program both approaches were investigated.

6 6 COMPUTER ANALYSIS OF HEAVY LIFT OPERATIONS OrC 5819 ( The model test program provided much information both from the points of view of the dynamic behaviour itself and the correlation with LIFSIM simulations. The time traces measured have been subjected to several post-processings and evaluations. Spectra and Response Amplitude Operators (RAO) were derived for the test situations without impacts. In this paper only some selected results are presented. Results for the SSCV only are shown in Fig. 8. These results clearly show the "resonant" response in swell waves with increasing non-linearity. I 1 the severest test of the quality of the simulated results, i.e. a direct deterministic comparison with experimentally obtained results. To this end the wave elevation measured during the model tests was used as input for the computational generation of wave forces. Comparisons of motions of SSCV and module clearly show the accurate modelling of the hydrodynamics with some discrepancies in the low frequency components of surge. Hoisting wire and sling tensions also show a good quantitative agreement. With respect to the jacket impact forces {Fig. II) a satisfactory agreement was found. In the stationary irregular wave tests, the prehoist situation was modelled. This situation with 1000 tonnes pre-tension in the hoisting wires can be considered as a test case for linear theory as long as the modules keeps contact with all impact springs and no slack slings occur. Crane tip motions for this situation for both the barge and jacket are compared with the motions of the SSCV only in Fig. Model test results for a realistic hoistingldeballasting situation are shown in Pig. 2. In this case the SSCV was heading irregular waves (significant height 2 m, mean period 8 seconds). From the pre-hoist condition (1240 tonf. pre-tension), the hoisting speed was m/s and the deballasting time 60 seconds. The presented hoisting wire tensions and support forces show characteristic wave frequency oscillation before lift-off. When the hoisting wire tensions equal the weight of the module, the wave induced motions of SSCV and barge result in large impacts and tension variations. Though tension drops are most significant, peak values are of ten referred to by means of Dynamic Amplification Factors (D.A.F.). As stated before the dynamic effects during a 60 seconds hoist strongly depend on the characteristics of the individual waves which pass the SSCV in this short time span. Because of the fact that the tests were performed in irregular waves repetition of the hoisting test for other hoist/deballast starting times resulted in very different levels of impact forces and hoisting tensions. Stationary irregular wave tests with load at several "pre-tension" levels provided more reliable information on such dynamic forces from the viewpoint of statistics. In Figs. 10 and 11 results of a stationary case with irregular seas are presented. The SSCV keeps the load in the still water lift-off position (pretension kn) just on a fixed jacket structure. The total system is subjected to irregular head waves with mean period of 8 seconds and a significant wave height of 2 metres. The use of impulse response techniques to derive first and second order wave force records enables In recent years the understanding of the behaviour of floating structures and the ability to compute wave forces, motions and governing loading has increased considerably. Extensive model test programs [2], [3] have clearly demonstrated the validity and applicability of the presented chain of hydromechanic theories, numerical techniques and computer implementations. For heavy lift operations, including the installation of jackets and entire top sides, to be performed in the coming years, a thorough evaluation of the dynamic behaviour is of prime importance. The feasibility, workability, planning and risks involved can be investigated by means of computer simulations. The presented LIFSM-package can be used for simulation of both deterministic and stationary situations. The pre-processing allows arbitrary hull shapes such as semi-submersibles, ships and barges while the water depth is accounted for. Geometry and elasticity of the cranes, hoisting wires, slings and stabbing piles may be fully specified. Liftlng, upending and mating of jackets can be simulated as well. REFERENCES 1. Michelsen, F.C. and A. Coppens: "On the upgrading of SSCV Hermod to increase its lifting Capacity and the Dynamics of heavy lift operations." OTC paper No. 5820, Houston Boom H.J.J. van den, A. Coppens, R.P. Dallinga and J.G.L. Pijfers: "Motions and Forces During Heavy Lift Operations Offshore". Workshop on Floating Structures and Offshore Operations, Wageningen Oortmerssen, G. van, J.A. Pinkster and B.J.J. van den Boom: "Computer Simulation of Moored Ship Behaviour". Journal of Waterway, Port, Coastal and Ocean Engineering Vol. 112, No. 2, March 1986.

7 Fig. l-liftable jacket oonflguration. Flg. 2-Force records; hoisting 8,000 tonnes from barge in irregular waves. Tine step CRANE SYSTEM t-czl= Time step differential IMPACT Fig. 3-Schematization of lift mechanlca. Fig. 4-Flow diagram of LIFSIM.

8 Fig. 5-Model test setup and stabbing pile instrumentation.

9 XlmO :F' l, 0.. 0,, 4.' ' S, #d.. -. L.. I,. r.. 8w.o YmOI Starting time 1 Starting time 2 m -, Fig. 6-Hoisting from jacket In irregular waves (simulation). I I RANDOM DETERMINISTIC STATIONARY PERIODIC NON-ERGODIC Fig. 7-Ciasslficatlon of a process. Irregular wave tests Frequency of wave encounter Frequency of wave encounter in radls in rad/s.5 Flg. 8-RA0 of heave and pitch of SSCV.

10 SSCV only SSCV with mdule on jacket (1000 t pre-tension) SSCV with module on barae (1000 t pre-tension) - Fig. 9-RA0 l Frequency of wave encounter in rad/s ot verttcal crane motions. Model tests LIFSIM l l Fig., ;.I m. )m,*,wo m.0 W.. m.. m cl.* m. m e xmm Fig. 10-Yolfona durlng stationary impact teats.