A dynamic model for lifting heavy modules between two floating offshore structures

Transcription

1 A dynamic model for lifting heavy modules between two floating offshore structures Radboud R.T. van Dijk Maritime Research Institute Netherlands (MARIN), Netherlands Alex A. Hendriks Heerema Marine Contractors Nederland B.V., Netherlands Lars Friisk Statoil ASA, Norway ABSTRACT: The installation of new topside modules on existing floaters in deep water poses new cha l- lenges, as both the crane vessel and floater move independently. Installing even a relatively light module on a moving target with existing topsides needs careful consideration. This paper describes the development of a numerical dynamic model to simulate an offshore installation of two modules on a Floating Production Unit (FPU) from a dynamically positioned crane vessel. The main reason for the development of this model was to prevent damage to both the existing equipment and the new modules and to optimise the installation procedure. The numerical model is based on an existing time domain simulation program extended with a dedicated user interface. The approach of this project was to tune and validate the numerical model using the results of an extensive model test program. The methodology of tuning will be presented in this paper along with a comparison between the results of the tuned numerical simulations and the model tests. As a next step in this approach the validated numerical model was used to do Monte Carlo simulations to derive and verify design input. Subsequently operational limits could be determined for all possible sea states and wave directions. Finally, this paper highlights some optimised aspects of the installation procedure. 1 INTRODUCTION Prior to every offshore installation the contractor performs a thorough analysis, often using numerical simulations. In most cases these simulations are done using the time domain simulation program LIFSIM, developed by Shell, HMC and MARIN in the 1980s, or any comparable computer code. The validation of this program has been done using results of model tests during the development of the program, full-scale measurements and extensive operational experience (Wouts et al. 1992). In this case the installation involved installation of two relatively light modules ( 500 t) on an existing floating production unit (FPU) using a semisubmersible crane vessel (SSCV). Although the weight of the modules was not an issue, the motions of the modules relative to the FPU would be higher than for a fixed platform installation. Therefore these motions needed to be assessed carefully as the modules have to be installed between existing equipment on the floater. Moreover, as a result of this, impact loads became an issue. Any damage to existing or new equipment would lead to downtime of the FPU. To derive design loads for the guiding system on the FPU, to determine operational limits of the installation and to optimise the installation procedure, a numerical model was developed and linked to LIFSIM. The approach of this project was to validate and tune the numerical model with a series of model tests and to base design input on statistical results of Monte Carlo simulations for all environmental conditions. 2 NUMERICAL MODEL There are several reasons to use a time domain simulation program to assess the lowering procedure. First of all, in time domain simulations both the wave frequency motions and the low frequency motions can be taken into account. Secondly, it allows evaluation of impact loads. Finally, it is possible to evaluate full dynamic simulations of approach manoeuvres as well as actual lowering of the modules. With the presently available computing power it is possible to run large numbers of time domain simulations in batch in a relatively short time. 2.1 Diffraction analysis As a first step, a two-body diffraction analysis was performed of the SSCV in close proximity to the FPU. The results of this diffraction analysis were used as input for LIFSIM and contain all hydrodynamic interactions between the two bodies, such as effect on added mass, wave shielding, etc. Figure 1

2 shows the panel distribution used in the diffraction analysis. Figure 1: panel distribution for two-body diffraction analysis 2.2 Numerical Model of Guides The next step was to make a detailed numerical model of the guiding system on the FPU and the bumpers on the modules. Figure 2 shows the guides and the two modules as used in the model test program. The guides consist of a set of primary guides, a set of secondary guides, four rubber fenders to cushion the landing and four permanent supports. Figure 3: Schematic guides and bumpers in simulation model 2.3 Tugger wires In order to control the yaw motion of the load during installation tugger wires are often used. A change in line geometry due to the yaw motions of the load results in a yaw restoring moment. These tugger wires were included in the numerical model between the base of the crane and the top corners of the module. The wire tension was kept constant during each simulation. Figure 2: Guides and modules in model tests Figure 3 shows schematically how the guides and bumpers were modelled in the time domain simulation program. To correctly calculate the forces of the modules on the guides, the guide stiffness was defined as function of height of bumper impact above the main FPU deck. For the rubber fenders a non-linear spring was used in the numerical model, which closely resembled the actual compression load characteristic. To account for the damping due to fender hysteresis, different spring values were used for the compression and rebound spring. The permanent supports were modelled as stiff linear springs. 2.4 LIFSIM This user-defined part of the program was then linked to the main LIFSIM program. The mooring of the FPU was simulated by springs. The DP system on the SSCV was simulated using appropriate springs and dampers. For both floaters all relevant coupling modes (such as surgepitch, etc) were taken into account. Figure 4 shows a schematic overview of the numerical model as used in the simulations. Figure 4: Schematic dynamic model overview 2.5 Numerical Stability During approach and lowering of the module high impact loads can occur. In the prototype situation these impacts are dealt with by adding dampers, rubber fenders and allowing some plastic deformation of the guides and bumpers. It is difficult to

3 quantify and accurately model this kind of energy dissipation. In addition, when high stiffness values are used in a numerical model high and unrealistic impact loads can occur. In some cases this may even lead to instability in the simulations. To avoid these kind of problems it is therefore important to 1) use a sufficient small time step for the simulations, 2) add a small amount of damping to model the expected full-scale energy dissipation and 3) use stiffness values that are realistic, taking into account the expected plastic deformation of the members under consideration. As it is difficult to model impact loads on complex structures in model tests, full scale measurements and Finite Element Method (FEM) analysis can provide valuable information. 3 TUNING WITH MODEL TESTS Although the program LIFSIM is thoroughly validated, the extension of the program to simulate the interaction between modules and guides required additional validation and tuning. By validating the numerical results to model test results, any defects in the numerical model could be identified and corrected. Tuning the numerical results to model test results ensured that realistic values were obtained from the simulations. All model tests were simulated for validation and tuning. Different stages of the installation procedure were considered with both one hour stationary model tests and tuning simulations: - Approach of the module (no contact) - Approach of the module (first contact) - Module in the guide (no pretension) - Module in the guide (with pretension) The lowering of the modules on the fenders and permanent supports was considered dynamically in both tests and tuning process. 3.1 Tuning parameters The first step in the tuning process was to determine the viscous damping for the FPU and SSCV, as the diffraction analysis results contain only potential damping. When no dedicated model tests are available, results of previously done tests on similar vessels can be used to make an estimate of the viscous damping coefficients. In this case dedicated model tests were available and the motion decay test results were used to derive viscous damping coefficients and thus obtain realistic motion response of both floaters. Next the motions of the modules in calm water decay tests were simulated and compared to the model tests, in order to derive appropriate damping coefficients for the modules. When the agreement between model tests and numerical simulations was found to be satisfactory for the calm water decay tests, a comparison was made between calculated motions of the three bodies in the numerical simulations and the measured motions in the model tests for the stationary condition in waves. For a proper comparison the first order and second order wave forces in the numerical simulations were based on the measured time traces of the calibrated waves in the model basin. For impact loads on the guides and permanent supports it was not possible to make a comparison of statistics. Therefore a comparison of the Weibull distribution and the most probable maximum (MPM) of the impact loads was made. Finally, for the loads on the rubber fenders and the permanent supports also a comparison of the actual time traces between simulations and model tests was made. 3.2 Tuning Results Below a brief overview of the results of the tuning process is presented. Figure 5 shows a comparison of standard deviation of numerical simulations and model tests. Figure 5: comparison of simulated and measured motions Figure 6 shows a comparison between simulation and model test of the pitch response of the FPU due to lowering of the module on the FPU deck. Figure 6: comparison of simulated and measured FPU response Figure 7 shows a typical example of the comparison of Weibull plots based on simulated impact

4 loads and measured impact loads on the primary guides. Figure 7: comparison of simulated and measured impact loads Figure 8 shows a comparison between simulation and model test of load on the fender during installation. Figure 8: comparison of simulated and measured fender loads 4 ANALYSIS OF INSTALLATION For this installation a design wave height and peak period were defined. The design of guides and permanent supports should correspond to this design condition and the resulting impact loads. With the tuned numerical model a large series of simulations were performed to validate and derive design loads and relative impact velocities. The next step was to perform an operability assessment for all loading conditions, based on the capacity of the design. The simulations also proved to be suitable for evaluation and optimisation of the installation procedure. For educational purposes the simulations have been visualised with three-dimensional graphics. 4.1 Loading conditions For the derivation of design loads and relative velocities, simulations were performed for wave heights and peak periods up to design condition and for eight wave headings with a 45 degrees interval. The simulations were divided into two stages, which, for correspondence with reality, were both considered dynamically: - Approach of the module from hovering cond i- tion to full pretension against the guides - Lowering of the module on the FPU deck For the sea states considered for installation a Torsethaugen wave spectrum was used, as this was considered to be the most appropriate wave spectrum formulation to describe the combined wind seas and swell at the location of the FPU. For this type of wave spectrum the spectral shape changes not only as function of wave period, but also as function of wave height. As a result the motion response of the floaters varies non-linear with wave height. Therefore time domain simulations were performed for all relevant wave periods and wave heights. 4.2 Design loads and relative velocities To obtain a reliable population for statistical postprocessing, for every load case a Monte Carlo simulation has been performed by randomly varying the wave seed. Whereas during the model tests only a limited number of simulations could be performed for a limited number of loading conditions, with LIFSIM it was possible to perform in the order of 50,000 simulations within a realistic time-frame, covering all conditions. The characteristic relative velocities and impact loads were determined by applying a Weibull fit on the simulation results. 4.3 Design considerations The design of the guides and permanent supports was based on the characteristic values found for the worst case loading condition. The horizontal impact loads found in the LIFSIM analysis and model tests were much higher than the traditional, proven-industry, rules of thumb for guide and bumper loads for fixed platform installations. This is because for this project the modules are installed on a floater which is moving significantly relative to the SSCV. The main issue for re-design was the inconsistency between stiffness and strength. To decrease horizontal impact loads a more flexible guide was needed, but this would violate the strength requirements. To resolve the issue, a damped guide concept was developed, with dampers installed on sliding braces. To investigate forces and accelerations in more detail inside the modules, a state-of-the-art FEM program was used. With LIFSIM motion results as input and high-detailed models of the modules, including all local mechanical properties, an assessment of forces and accelerations could be made for the final design.

5 4.4 Operability assessment The characteristic impact loads and relative velocities, used for the design of guides and permanent supports, were used as a unity check criterion for the operability for all the other peak periods and wave headings. Simulations were performed, with significant wave heights exceeding design wave height, yielding a set of limiting wave heights for all installation conditions. against the guides by the pretension. The number of impact is still high, but the magnitude of the impact loads has dropped significantly. 4.5 Installation procedure A number of parameters have been investigated and varied to optimise the installation procedure. A few examples of installation optimisation are given below. During the approach stage tugger wires are used to control the yaw motions of the modules. Figure 9 shows the positive effect of tugger wires on module yaw. Figure 9: Effect of tugger wires on yaw of the module On top of wave frequency motions both the FPU and SSCV exhibit low frequency motions. As a result, there will be a significant low frequency relative motion between the module and the FPU during the approach manoeuvre. The crane operator will have to compensate for these low frequency motions during the approach manoeuvre to avoid contact between the module and existing equipment on the FPU. It was found that a sufficient level of pretension of the module against the guides is important. This phenomenon is illustrated by Figure 10. In the first stage of approach ('hovering') the module has a relatively large clearance to the guides and impact loads hardly occur. In the second stage ('1 st impact') the impacts are more frequent and the loads are higher. In the third stage ('no pretension') the module regularly looses contact with the guides due to the relative motions between SSCV and FPU. The module swings back and forth, picking up speed, and has increased angular motions due to the eccentric point of contact with the guides. This results in significant loads at repeated impacts. In the fourth stage ('pretension') the module is pressed Figure 10: Effect of pretension on guide loads Related to this issue, it was found that the approach speed should not be too low. A relative low approach speed lengthens the period of time between first contact of module and guides and the stage where the module is sufficiently pretensioned against the guides. This increases the chance on high repeated impacts with the guides. A too high approach speed increases the chance on a high first impact between modules and guides. Another phenomenon was identified for the lowering of the modules on the permanent supports. Figure 11 demonstrates that, although the lowering speed is set at the maximum of the hoist capacity, the module cannot always follow the motion response of the FPU deck as a result of first impact. This results in a second impact load on the permanent supports which is significantly higher than the load at first impact. Figure 11: High second impact on permanent support

6 4.6 3D Visualisation Despite the relative low weight of the modules to be installed, this installation is not a routine job. The safety of the operation is enhanced through intensive communication between analysts and the offshore crew. A computer program was developed which can do three-dimensional (3D) visualisation. 3D visualisation of the time-domain simulations proved to be a powerful means to help preparing the crew of both the FPU and the crane vessel for the operation. Moreover, 3D visualisation was a great help in understanding the physics of this operation better. The program links LIFSIM output, consisting of motion time traces of the bodies involved, to 3D graphical models from Microstation. The exact motions and interactions (e.g. collisions) that are simulated with the analyses are visualised in three dimensions. Figure 12 gives a screen shot of a 3D visualisation of the installation of a module. Analysis of the effect of the lowering speed revealed that the highest possible lowering speed of the crane vessel resulted in the lowest vertical loads on the permanent supports and module. 3D visualisation of the simulation can be used effectively to support the offshore crews of the FPU and SSCV in their preparation for a safe operation. In general, the approach of this project proved to be highly valuable for all parties involved. It is concluded that for similar installations, concerning two floaters and impact sensitive equipment, the preferred methodology consists of the following steps: - Validation and tuning of the numerical model with model tests - Performing Monte Carlo analysis with the tuned numerical model to obtain statistically reliable results - Use the analysis results for the design condition as an input for design optimisation - Establish installation limits for all loading cond i- tions - Optimise installation procedure In the project described in this paper the hydrodynamic time domain simulations and FEM analysis were done separately. Motions calculated in LIFSIM were used as input for the FEM analysis to optimise the guides. A future development may to be couple LIFSIM (or similar hydrodynamic computer code) directly to an FE-program. This will allow solving the motions of the floaters (which are dominated by hydrodynamic loads) and the modules (which are dominated by contact loads) more accurately. REFERENCES Figure 12: Screen-shot of 3D visualisation Wouts, R., Coppens, T. & Boom, H.J.J. van den 1992, "Monitoring Offshore Lift Dynamics", OTC CONCLUSIONS AND LESSONS LEARNT Based on the results of this study it could be concluded that the initial design of the guides could be optimised to limit the introduction of high loads between modules and guides. The optimised design of guides ensures a safe operation and avoids damage to existing equipment or modules. Simulations were performed with the optimised design to verify the reduction in loads. By performing simulations for a number of loading conditions, a set of limiting wave heights was determined for all peak periods and wave headings. Using the tuned numerical model the appropriate pretension levels were determined to keep the module steady in the guides during lowering. For the approach manoeuvres the best setting for tugger wire tension was established.