Model Tests and Computer Simulations for Njord FPU Gas Module Installation

Transcription

1 Marine Operations Specialty Symposium (MOSS2008), pp.1 20 c 2008 CORE, National University of Singapore, Singapore Model Tests and Computer Simulations for Njord FPU Gas Module Installation Remmelt van der Wal 1, Hans Cozijn 1 and Chris Dunlop 2 1 Maritime Research Institute Netherlands (MARIN), Wageningen, the Netherlands 2 Saipem UK Limited, London, United Kingdom Recently, two gas modules have been installed on the Njord FPU, on the Norwegian Continental Shelf, to enable gas exports from the Njord Field facilities. The 100 and 200 tonnes modules were placed onto the FPU deck by Saipem s dynamically positioned S7000 SSCV. Detailed model tests and computer simulations were carried out by MARIN and Saipem UK. The purpose of this analysis was to investigate the limits of the offshore installation s operability. Experiences gained from the actual installation in the field proved to be valuable feedback for the performed analysis. Fig. 1. Model Tests in MARIN s Offshore Basin. The model tests were carried out in MARIN s Offshore Basin. Various sea states and wave directions, relevant to the Njord field, were modelled. Detailed scale models of the S7000, the Njord FPU and both modules were constructed at a scale of 1:50. The S7000 was equipped with a DP system, including the controller, Kalman filters and azimuthing thrusters. The installation sequences were modelled in detail by using an automatic control of the crane. Measured signals included motions of the floaters and gas modules, loads in the hoisting arrangement and loads on the supports and guides. The FPU mooring loads and S7000 thruster performance were monitored as well. Time domain computer simulations were carried out using MARIN s simulation program LIFSIM. The simulation model included both floaters, the crane and hoisting arrangement, the gas modules and the guiding system on the FPU deck. First, model test results were used to optimise the performance and accuracy of the simulation model. The tuning of the LIFSIM program included the calibration of damping levels and the detailed mathematical modelling 1

2 2 R. van der Wal, H. Cozijn and C. Dunlop of the guides and supports. A good agreement between the model test and simulation results was obtained. Secondly, a large number of simulations were performed to investigate the different environmental conditions, as well as different approach and lowering scenarios. Finally, observations and experiences for the actual installation of the modules, which took place earlier in the summer of 2007, are presented in this paper. Introduction Two gas modules have been installed on the Njord FPU recently, to enable gas export from the Njord Field facilities. The modules have been placed on the Njord FPU by Saipem s semisubmersible crane vessel S7000. The weight of the modules was 100 and 200 tonnes. Prior to the installation a study was carried out to optimise the lifting operation and to investigate the limits of the environmental conditions in which the modules could be installed safely. The study included a combination of hydrodynamic scale model tests and time-domain computer simulations. Furthermore, observations made during the actual offshore installation can be used as feedback to improve the quality of similar studies in the future. Figure 2 below shows in a schematical manner the interaction between scale model tests, time-domain computer simulations and full scale measurements. Each has its own advantages and these can be exploited by using the tests, simulations and full scale measurements in combination. The hydrodynamic scale model tests deliver detailed information on the behaviour of the considered system in a set of controlled environmental conditions. The properties of the models, such as weight distributions and stiffness values, can be accurately calibrated. Furthermore, all relevant physical aspects are automatically taken into account. Besides confirming the overall feasibility of the operation, the model test results include data that can be used to calibrate the numerical models of the system, such as hydrodynamic (viscous) damping levels and the structural response of the models on impacts. The main advantage of time-domain computer simulations is that it is relatively easy to investigate a large number of environmental conditions and installation scenarios. Compared to scale model tests the costs are relatively low. The accuracy of the simulation model can be optimised by comparing and tuning it against model test results. The final proof of the installation s feasibility could of course only be found at full scale, when the real life S7000 crane vessel placed the two gas modules on the deck of the Njord FPU. Full scale measurements of motions and loads can provide valuable feedback on the Model Tests Computer Simulations Full Scale Measurements Fig. 2. Interaction between Model Tests, Simulations and Full Scale Measurements.

3 Model Tests and Computer Simulations for Njord Fpu Gas Module Installation 3 quality of the model tests and computer simulations (1, 2 and 3). It should be noted, however, that on board measurements can face a range of practical problems and should not interfere with the installation process itself. Furthermore, it is of course not possible to control the environmental conditions on-site. And finally, not all parameters, such as weight distributions and line pretensions, can be determined with the same accuracy as they can in model tests and computer simulations. An additional possibility, which was not considered in the present case, is the training of the crew by modelling the offshore installation procedures, including all platforms, on a training simulator. This approach enables the crew to practice the operation without any risk of damage, even in difficult circumstances, while it is also possible to test and improve the installation procedures. The present paper first discusses first the model tests that were carried out in MARIN s Offshore Basin. A description of the models is given, as well as a description of the modelled environmental conditions and test scope. Second, the time-domain computer simulations are discussed. A description of the applied mathematical models is given. Furthermore, the tuning of the simulation program is discussed, as well as the simulation scope. Third, observations made during the actual installation of the two modules on the Njord FPU are presented. Nomenclature ABS B D DEH DP FDM FPU GM T H L L oa L pp L pontoon M SSCV T TEG Model Tests Acrylonitrile Butadiene Styrene (a plastic) Width, [m] Vessel displacement mass, [tonnes] Column diameter, [m] Dehydration module Dynamic Positioning Fused Deposition Modelling Floating Production Unit Transverse metacentric height, [m] Height, [m] Length, [m] Vessel overall length, [m] Vessel length between perpendiculars, [m] Pontoon length, [m] Mass, [tonnes] Semi-submersible crane vessel Vessel draft, [m] Regeneration Module An extensive series of model tests was carried out in MARIN s Offshore Basin, at a scale of 1:50. The Offshore Basin has been in operation since the year 2000 and offers state-of-the-art capabilities for the generation of current, waves and wind in offshore model tests (4 and 5). A large number of different wave conditions was modelled, varying in wave height, period and direction. Wind and current were not modelled. The environmental conditions considered in the model tests formed a systematic set in order to optimise the tuning process of the mathematical models.

4 4 R. van der Wal, H. Cozijn and C. Dunlop Scale models The model tests included the moored Njord FPU, the dynamically positioned semisubmersible crane vessel S7000, a remotely operated crane, the two gas modules and a detailed modelling of the guide and support system on the FPU deck. Semi-submersible crane vessel S7000 The crane vessel was represented by an existing model of the S7000, which was provided to MARIN by saipem UK. all measuring devices and other equipment necessary for the model tests was built into the model by MARIN and the required weight distribution was calibrated. The main particulars of the S7000 SSCV, as calibrated in the preparations of the model tests and used in the time-domain computer simulations, are presented in Table 1 below. The photograph in Figure 3 below shows the instrumented S7000 model directly after completion in the workshop. It is noted that no top-sides were included on the S7000 model, because during the present model tests no wind was generated in the basin. Table 1. S7000 Main Particulars. Designation Unit Magnitude L oa m L pp m B m 87.0 T m 27.5 tonnes 165,540 GM T m 18.1 Fig. 3. Model of the S7000 Crane Vessel.

5 Model Tests and Computer Simulations for Njord Fpu Gas Module Installation 5 Crane, hoisting arrangement and tugger lines Only the port side crane was installed on the deck of the S7000. The crane and its boom were positioned as specified by Saipem UK and fixed. The modules could now be brought to the correct position above the FPU deck by changing the x- and y-position of the dynamically positioned SSCV. For both top-side modules the complete hoisting arrangement consisted of the crane main hoisting wire with the crane hook, two upper slings connected between the hook and a spreader bar and four lower slings connected between the bar and the four corners of the lifted module. The weight of the main hook and spreader bar were modelled to scale, as well as the axial stiffness of all lines using linear springs. A photograph of the model crane and a schematic representation of the hoisting arrangement are shown in Figure 4 below. In the model tests the lowering of the module was achieved using a remote control winch with a steering unit capable of performing prescribed lowering sequences. This approach ensured the repeatability of the process during the model tests, as well as the exact modelling of lowering speed and vertical position of the lifted module. Tugger wires were positioned between the lifted module and the S7000 crane vessel to avoid undesired yaw motions of the modules during the lifting operation. The tugger wires were used during the hovering tests in the Offshore Basin, but were removed for the lowering tests. Fig. 4. S7000 Crane Model and Hoisting Arrangement.

6 6 R. van der Wal, H. Cozijn and C. Dunlop DP system the S7000 model was equipped with a fully functional real time DP system, specifically designed for use in model tests. This model DP system contains all components included in a full scale DP system on board, such as a controller, a kalman filter, a thruster allocation algorithm and azimuthing thrusters. Detailed modelling of the DP system ensures a realistic response of the dynamically positioned crane vessel during the model tests. The S7000 model was equipped with four azimuthing thrusters, of which the azimuth angles and thruster RPMs are controlled by the DP system. In the model tests these four model thrusters represented the 12 controllable pitch thrusters that are present on the actual vessel. In Figure 5 below one of the model azimuthing thrusters is shown. The S7000 position measured in the basin is used as input for the model DP system. The Kalman filter included in the model DP system separates the low frequency vessel motions from the measured position error. Subsequently, a PID controller determines the total required thrust forces. The control coefficients used during the model tests were specified by Saipem UK and were based on the settings applied in the actual S7000 vessel. Finally, an allocation algorithm is used to efficiently distribute the total required forces over the available azimuthing thrusters. It is noted that the response times for the thruster azimuth angles and RPMs were also correctly modelled. Njord FPU model The hull of the Njord FPU model was constructed of PVC pipe and plate material. A deck frame was modelled of thin-walled steel pipes and a wooden deck was added. Similar to the S7000 model, the correct weight distribution was calibrated after all necessary equipment had been built into the model. The main particulars of the Njord FPU model are presented in Table 2 below. Figure 6 below shows the instrumented Njord FPU model directly after completion in the workshop. No top-sides were included on the Njord FPU model, except for one of the deck cranes, which was located near the positions of the new top-side modules. Fig. 5. Azimuthing Thruster on the S7000.

7 Model Tests and Computer Simulations for Njord Fpu Gas Module Installation 7 Table 2. Main Particulars of the Njord FPU. Designation Unit Magnitude L pontoon m B m 81.5 D m 16.0 T m 23.0 tonnes 45,630 GM T m 4.0 Fig. 6. Model of the Njord FPU. Mooring system and risers The actual Njord FPU platform is moored to the sea bed by means of four bundles of three mooring lines. In the model tests a simplified mooring system consisting of four mooring chains was used to represent the actual mooring system. The model test mooring system was designed such, that load-displacement characteristics close to those of the actual mooring system were achieved. The model mooring lines were constructed of segments of chain and steel wire. A calibrated axial linear spring was included in each of the lines to achieve the correct total axial stiffness of the lines. Using a similar approach as for the mooring lines, the 20 risers of the actual Njord FPU platform were represented in the model tests by a set of five equivalent risers. These were designed such that the horizontal pretension, stiffness and drag loads were representative of the situation in the field. The model risers were constructed of silicon hoses with a weighted steel wire inside. The purpose of the hose was to achieve the required outside diameter of the riser, while the wire was calibrated to create the correct underwater weight. Guiding system The most complex items to construct in the model tests were the guides and supports on the Njord FPU. The purpose of the system of primary and secondary guides is to direct the topside modules on to their supports, during the final stages of lowering. The presence of the guides ensures that the modules can be placed safely, without a risk of damage, e.g. due to

8 8 R. van der Wal, H. Cozijn and C. Dunlop Fig. 7. Guiding system and instrumentation. accidental collision between the lifted top-side module and existing equipment on the FPU deck. Figure 7 below shows the guiding system, supports and instrumentation for the DEH module. In order to accurately measure any (impact) loads between the lifted top-side modules and the guiding system, a detailed modelling of the primary and secondary guides and supports was required. An accurate representation of the geometry, as well as the correct modelling of the stiffness and deformations of the guiding system, is essential to measure the impact loads as good as possible. This was achieved by designing and constructing the system of guides inside a very stiff housing, so that all components could be calibrated in the dedicated instrumentation workshop, after which it could be built into the Njord FPU model as a single unit. Installed top-side modules Two different top-side modules were installed on the deck, being the TEG (regeneration) module and the DEH (dehydration) module. The main structure of the models representing the top-side modules was constructed using a technique called FDM (fused deposition modelling). This novel technique could be best described as 3D printing, as the models are built up layer by layer in tiny droplets, based directly on their 3D AutoCAD drawings. The end result is a geometrically accurate model made of ABS (a synthetic material), which

9 Model Tests and Computer Simulations for Njord Fpu Gas Module Installation 9 Fig. 8. Models of the Top-side Modules. is light, strong and stiff. After construction of the module frames the necessary instrumentation (e.g. for the motion measurements) were included, as well as calibrated ballast weights, to achieve the correct total weight distributions. Both modules are shown on the photograph in Figure 8 below. The bumpers, which are shown in the photograph below, were accurately modelled using brass material. The weights and dimensions of the top-side modules are presented in Table 3 below. Instrumentation The signals measured during the model tests included motions of the floaters and the lifted modules, mooring loads, lifting loads and loads on the guiding system, as well as measurements of the S7000 thruster RPMs and azimuth angles. The motions of the Njord FPU, the S7000 crane vessel and the module suspended from the crane were measured using an optical motion measuring system. The system measures the positions of three infrared LEDs, placed on each of the models, and derives from these the motions in six degrees of freedom. In this manner the motions could be measured with an accuracy of better than 0.5 mm/0.1 deg (model scale values). The line tensions in the mooring lines and hoisting wires were measured using ring shaped strain gauge transducers. The (impact) loads on the guiding system and module supports were measured using strain gauge transducers built into a purpose- built frame. This enabled the measurement of the loads on the individual guides and supports, while still achieving a sufficiently stiff model with an accurate geometrical and mechanical representation of the actual guides and supports. Table 3. Size and Weight of the Top-side Modules. Designation Unit Magnitude TEG Magnitude DEH L m B m H m tonnes

10 10 R. van der Wal, H. Cozijn and C. Dunlop Fig. 9. Schematic view of set-up. Table 4. Tested combinations Hs and Tp. Significant wave Height Hs 1.0 m 1.5 m 2 m 2.5 m 6 s x x x x Wave peak 8 s x x x x period Tp 10 s x x x x The RPMs and azimuth angles of the thrusters on the S7000 model were recorded using digital transponders (pulse counters). A schematical test set-up can be found in Figure 9. Shown are the SSCV, crane and hoisting arrangement, the FPU and one of the lifted modules. The tugger lines are shown as well. Environmental conditions Wave conditions were modelled using Torsethaugen irregular waves and in regular waves. Wind and current were not modelled. For the present work, waves were modelled from three directions in the Offshore Basin. By rotating the set-up, any wave direction can be modelled. The modelled relative wave directions are shown in Figure 10 below. The significant wave height varied from 1.0 m to 2.5 m. The corresponding wave periods were 6, 8, and 10 s. The combinations can be found in Table Fig. 10. Modelled wave directions.

11 Test scope Model Tests and Computer Simulations for Njord Fpu Gas Module Installation 11 During the installation a number of different phases can be distinguished from the moment the module is lifted and suspended in the hoisting arrangement and the moment the module is placed on the deck: (1) Hovering phase: the module is lifted and in position at about 2 4 m from the guide. No contact between the module and the primary guides is found. Main focus was on the module. (2) First impact phase: the module is brought closer to the guides (clearance between 0.5 and 1.0 m) by positioning the SSCV closer to the FPU. Impact loads are found more frequently. Main focus was on the module motions and the magnitude of the impacts on the primary guides. (3) Pretension ( overboom ) phase: module is pushed against the guide constantly by positioning the SSCV further towards the FPU. Main focus was on the module motions and the magnitude of impacts on the primary guides. (4) Lowering phase: the module is lowered against the guides with 0.1 m/s until the module is fixed on the FPU deck and the hoist wire is slack. Main focus was on the module motions and the magnitude of impacts on the primary and secondary guides and the permanent supports. Prior to the actual tests decay tests were carried out to derive the natural periods and damping levels, which is required input for tuning the numerical tool. The initial DP control coefficients were obtained from SAIPEM UK. DP optimisation tests were done to improve and obtain the required DP performance of the SSCV by fine-tuning the control coefficients. Model test results The different phases of the installation were tested in separate tests. The results below show the typical behaviour of the DEH module for stages I III. Focus is laid on the module motions and the loads on the guides and supports. Figure 11 shows the surge motions of the SSCV (red line) and the DEH module during the hovering phase (I), the first impact phase (II) and the pretension phase (III). The different positions during the three stages can be clearly identified. As a result of the more forward position of the SSCV, the module is closer to the primary guides and impact loads start to occur. The motion behaviour of the module changes as well. In the hovering phase, when there is no contact between FPU and module, the DEH module shows low frequency (LF) motions which are caused by the SSCV motions and motion frequencies that match the pendulum natural frequency. During stage III, when the module is against the guide, mainly LF motions are observed, caused by the LF motions of the FPU (natural surge period FPU = rad/s). The higher frequencies are no longer present. This shift can be observed in Figure 12 below. As a result of the more forward position of the SSCV, the module is closer to the primary guides. The LF and WF motions of the FPU and SSCV and the motions of the lifted module result in contact between the guides and the module and thus impact loads. This is shown in Figure 13. In Stage I no impacts are found. For Stage II the module contacts the primary guides frequently. For Stage III the impacts are more frequent and show the highest peaks.

12 12 R. van der Wal, H. Cozijn and C. Dunlop X SSCV X DEH II III I Fig. 11. Surge motions SSCV and DEH during hovering. Fig. 12. spectral density DEH motions for Stages I III. Once the module is positioned against the primary guides by moving the SSCV sufficiently forward, a pretension situation occurred. From this moment on, the lowering of the module was started. The lowering procedure stopped when the module was positioned on the four permanent supports (BASE 1-4) and the hoisting wire was slack. What can be observed in Figure 14 is that the weight of the module is non-uniformly distributed along the four permanent supports. A construction that is supported by four points is statically undetermined. The particular distribution of the weight over each corner for the relative stiff module is a result of the position of the centre of gravity and it is very sensitive to small deviations in the support heights. This will not be the case in reality because the stiffness of the real module and the deck will be smaller and therefore small deformations are allowed. Furthermore, the lowering speed of 0.1 m/s seems not sufficient to let the module follow the vertical motions of the FPU. A detail of the first impact loads on BASE 2 is shown in

13 Model Tests and Computer Simulations for Njord Fpu Gas Module Installation 13 Fig. 13. Impact loads on primary guides DEH module. Fig. 14. Impact loads on permanent guides DEH module. Figure 15. After the first impact the module is lifted again due to the relative motions of the crane tip and the FPU deck. Based on the measured impact loads for the installation in various sea states and wave directions and the allowable limits for impact loads on the permanent supports and horizontal guides, the workability can be determined.

14 14 R. van der Wal, H. Cozijn and C. Dunlop Fig. 15. Detail impact load BASE 2. Time Domain Computer Simulations The time-domain simulations in the present study were carried out using the LIFSIM program, which can model the behaviour of up to three (floating) bodies, including all hydrodynamic and mechanical interactions between them. Using a time-domain program, the actual lowering procedure can be modelled as well as the approach manoeuvres itself. The program was originally developed as part of a Joint Industry Project in the 1980s (1 and 2). Developments on the mathematical models within the program continue until the present day, for example concerning the hydrodynamic interactions between the floaters (6). The simulation tool is used on a regular basis, for simulation of lift operations (3), but due to its flexibility also for other types of simulation studies involving multiple floating bodies (6). Diffraction-radiation calculations Prior to the time-domain simulations a frequency domain linear diffraction analysis was carried out for the S7000 crane vessel and the Njord FPU. In the diffraction calculations both vessels are placed at their correct relative positions. The results of the analysis include hydrodynamic reaction forces (added mass and damping coefficients), as well as first order (linear) wave forces and second order (quadratic) wave drift forces. All hydrodynamic interactions between the two floaters are taken into account. Figure 16 below shows the panel distributions of the S7000 crane vessel and the Njord FPU as used in the diffraction calculations. Modelling of the bumpers, guides and supports The guides and bumpers were different for both modules. For the DEH module the two primary guides and four secondary guides were defined with a horizontal stiffness depending on the vertical position of the impact. The modelled stiffness was a linearisation of the actual stiffness of the guides. The bumpers on the modules are defined as a series of adjoining points. Figure 17 shows a visual representation of the points that define the guides and bumpers of the DEH module in LIFSIM. As the secondary guides are limited in height, the loads between bumper and secondary guides are only calculated if the bumper at a certain time step is below the top of the secondary guide. The supports on the FPU were modelled as impact points. Coordinates of the supports were identical to those used in the model tests. The applied vertical stiffness was derived from the model tests. Some damping was added to account for some energy dissipation.

15 Model Tests and Computer Simulations for Njord Fpu Gas Module Installation 15 Fig. 16. Diffraction Analysis Panel Distribution. DEH guides and bumper modelling y-coordinate 2.0 bumpers vertical guides x-coordinate Fig. 17. Visual presentation of modelling guides and bumpers. Modelling of the hoisting arrangement and tugger wires The lifting arrangement was modelled as four slings with zero mass between crane hook and the module pad eyes. The spreader bar was not modelled as a separate body. The tugger wires were modelled as weightless, constant tension lines with a linear stiffness, placed between the connections points on the hovering top-side module and the S7000 crane. Two points are defined on the crane boom and two points on the module pad eyes. The tugger wires were crossed from PS crane boom to SB pad eye on the module and vice versa. Lifsim Simulation Tool After all parts of the installation were modelled, the user-defined input was stored in records that are used in LIFSIM. The LIFSIM program reads the frequency domain results from the diffraction analysis and transforms these into time domain data that are used in the actual

16 16 R. van der Wal, H. Cozijn and C. Dunlop simulations. The frequency dependent added mass and damping coefficients are transformed into a set of frequency independent added mass coefficients and retardation functions, which can be used in the time domain. In addition, linear and quadratic damping coefficients are included to represent e.g. viscous roll damping. Furthermore, the vessel low frequency damping, which is always of viscous origin, is modelled using an empirical mathematical model. Further modelling aspects: A time record of the wave elevation that was calibrated for the model tests was used in the simulations. This enables a direct comparison between the simulations and model tests. Based on the two-body diffraction analysis results first and second order wave force time series on the SSCV and the FPU were generated. In the time domain simulations the DP system of the S7000 and the Njord FPU mooring system and risers were not actually modelled. Instead, these were represented by restoring force matrices. As a result the mooring and DP system stiffness were linearised. The crane on the S7000 was modelled by defining the exact position of the crane tip as a fixed point on the vessel. Tuning of the simulation model Tuning time domain simulations to model test results requires a lot of effort and experience for complex systems. The resulting motion behaviour of each body depends on the applied damping viscous coefficient, but also on potential damping (due to wave radiation) and mutual interaction between the bodies. To asses the overall quality of the tuning process, a comparison is made of natural periods, overall damping levels between simulations and model tests for the decay tests and the comparison between the simulations and model tests for the hovering and lowering tests. The following steps can be distinguished in the tuning process that was done: The measured wave elevation as calibrated in the model test program was used to generate 1st and 2nd order wave forces on the FPU and SSCV. This ensures that computed 1st and 2nd order motions of the vessels are in phase with the measured motions of the vessels. Calm water decay tests were assessed and simulated to find reliable damping coefficients. This was done for all degrees of freedom of SSCV and FPU and for both modules, with and without tugger wires. Next, time domain simulations were performed and the simulation results were compared to the model test results. By performing the simulations in order of increasing complexity, the numerical model could be refined step by step and final results of the simulations were tuned quite closely to the model tests. For the hovering (stationary) simulations a comparison was made between the behaviour of the three bodies by comparing the time traces as well the standard deviation for each degree of freedom. In Figures 18 and 19 a comparison is made between the FPU and SSCV surge motions in Hs=1.5 m seastate. A good overall agreement between model tests and simulations was found. What can be observed is that after the first phase the transition to the second phase shows an overshoot for the simulations. This is caused by the fact that in the simulations the step further towards the FPU was obtained by applying a higher constant force acting instantly on the SSCV, causing an overshoot. During the model tests the set-point of the DP system

17 Model Tests and Computer Simulations for Njord Fpu Gas Module Installation 17 nd. simulation model test Fig. 18. SSCV x-motions. simulation model test Fig. 19. FPU x-motions. was moved forward slowly, resulting in a slow forward movement of the SSCV and no overshoot was present. For the modules not only the motions, but also the reaction forces between the modules and the FPU guides were considered. Initially, significant differences were found for the motions and thus the loads on the guides between the model tests and the simulations using the damping coefficients based on decay tests on the modules. Therefore, damping coefficients of the modules and on the primary guides were derived using a trial and error process of changing the damping coefficients until a satisfactory agreement between the model test results and the simulations An example of the impact loads on the primary guides can be found in Figure 20. A similar approach was followed for the dynamic lowering simulations. The dynamic behaviour of both FPU and SSCV was well predicted using the original damping coefficients, derived from the decay tests. Further tuning of the module damping coefficients was required, as there was frequent contact between the modules and the guides as well as contact between the modules and the fenders and permanent supports. The lowering sequence was repeated ten times per test, shown in Figures The same sequence was modelled numerically and these results are also shown in the same Figures. Figure 21 shows the vertical position of the module. The ten lowering events can be identified clearly. The results of the simulations match the model test results quite well. By lowering the DEH module, the FPU pitch angle changes two degrees. In the model tests the FPU was given a pre-tilt angle of two degrees while this was not done in the simulations. This explains the two-degrees difference in the FPU mean pitch angle shown in Figure 21. Figures show the impact loads on one of the permanent supports, BASE 2 and the loads in the hoist

18 18 R. van der Wal, H. Cozijn and C. Dunlop simulation model test Fig. 20. DEH x-motions and primary guide loads. simulation model test Fig. 21. DEH heave motions and FPU pitch motions during lowering. wire. The behaviour of the loads is comparable. The behaviour of the impact loads is comparable but the magnitude of the peaks is different. This is a result of the modelling of the supports in the model tests (a system with four pick-up points is statically undetermined). The measured impact loads were compared with the allowable limits and the effect of the wave height and peak period on the magnitude of the impacts was investigated. Although the impact loads during lowering were the prime interest of these tests, the results in Figure 23 also show a second impact when the module was lifted from the FPU deck for the next lowering sequence. These impact load values are not further discussed since they are not relevant for the present study.

19 Model Tests and Computer Simulations for Njord Fpu Gas Module Installation 19 simulation model test Fig. 22. Loads in hoist wire during lowering. Fig. 23. Loads on permanent supports during lowering. Offshore Installation of the Modules The actual installation of the modules was carried out in the summer of The sea state during the operation was ideal; a flat calm seastate was reported. All through the operation no unusual module motions were observed. The modules were lifted from the S7000 deck and slewed over the bow. Subsequently, the S7000 moved into position on DP. Final minor adjustments for the installation positions were made by luffing and slewing the crane. The tugger lines worked well: unwanted movement of the modules was not observed. Before the set-down of the modules, the FPU was trimmed over. After the module was set-down, which was observed for the TEG module in particular, the FPU had returned to a level condition. Conclusions Based on the results of the model tests and computer simulations, as well as the observations made on board during the actual installation, the following can be concluded. The combination of physical scale model tests, time-domain computer simulations and full scale proved to be a powerful approach for the analysis of the Njord FPU top-side installation.

20 20 R. van der Wal, H. Cozijn and C. Dunlop In the model tests an accurate modelling of the physical models, control systems and environmental conditions was achieved. Special attention was paid to the DP system of the S7000, the modelling of the guides and permanent supports on the FPU, the topside gas modules and the approach and lowering sequence. The combination of all these separate aspects proved to be particularly complex. From the results of the model tests the parameters were derived which are essential for the tuning and validation of the LIFSIM model. These parameters included damping levels, modelling of the guides and support and modelling of the impact loads. After tuning and validation, the LIFSIM model was capable of accurately reproducing the model test results, for both motions and loads. The tuned model was used by Saipem UK for further analysis and optimisation of the installation operation. For example, the model was used to determine the operational limits. Although no actual measurements were carried out during the actual offshore installation, it was observed that the motions of the floaters and top-side modules resembled the behaviour in the model tests and computer simulations. Acknowledgements The authors would like to thank Saipem UK and StatoilHydro for their permission to publish this technical paper. The information in this paper is based on the results of the model tests carried out at MARIN s Offshore Basin and the LIFSIM computer simulation study performed by MARIN and Saipem UK. References 1. Computer Analysis of Heavy Lift Operations, H.J.J. van den Boom, J.N. Dekker and R.P. Dallinga (MARIN), Offshore Technology Conference, Houston, 1988, OTC Monitoring Offshore Lift Dynamics, R. Wouts, T. Coppens (Heeremac) and H.J.J. van den Boom (MARIN), Offshore Technology Conference, Houston, 1992, OTC A Dynamic Model for Lifting Heavy Modules Between Two Floating Offshore Structures, Radboud R.T. van Dijk (MARIN), Alex A. Hendriks (Heerema Marine Contractors Nederland B.V.) and Lars Friisk (Statoil ASA), EuroDyn Conference, Paris, Features of the State-of-the-Art Deepwater Offshore Basin, B. Bucher, J.E.W. Wichers and J.J. de Wilde (MARIN), Offshore Technology Conference, Houston, 1999, OTC Important Environmental Modelling Aspects for Ultra Deep Water Model Tests, B. Buchner, J.L. Cozijn, R.R.T. van Dijk and J.E.W. Wichers (MARIN), Deep Offshore Technology Conference, Rio de Janeiro, Numerical Multiple-body Simulations of Side-by-side Mooring to an FPSO, B. Buchner, J.J. de Wilde and A.W. van Dijk, ISOPE Conference, Stavanger, ISOPE2001-JSC-285.