| dc.description.abstract | This masters thesis presents a possible solution for deepwater model testing of floating production units (FPU) in ocean basins, and investigates some aspects of the method. At present, the existing ocean basins have a maximum water depth of around 10 meters. Since typical model scales are between 1 : 30 and 1 : 100, the maximum depth using conventional model scales is around 1000 meters. Today, deepwater model testing in ocean basins are limited to ultra small scale modelling, and truncated modelling.
In ultra small scale testing, both the FPU, mooring lines and risers are scaled to fit the available basin depth. However, ultra small scales tends to increase the uncertainties regarding the results, and it becomes increasingly difficult to accurately make a small model as the tolerances are reduced.
In a truncated model test, a smaller scale factor is used. In practice, this means that the scaled system will go deeper than the available ocean basin depth, and the lines are thus truncated at the ocean basin bottom. The experimental results are then used to calibrate a numerical model of the system which is then used to simulate the total scaled system. Since the final results rely on a numerical model, there is a risk that higher order phenomena like green water are not properly simulated. At present, this method is preferred by SINTEF MARINTEK.
The method presented in this thesis is a hybrid method that combines a scaled physical model of the FPU and a scaled numerical model of the mooring lines and risers in a real time experiment. The numerical model will receive real time information about the motions and velocities of the FPU model, and calculate the corresponding mooring line and riser forces. The resultant force from the mooring lines and risers are then passed to a controller that actuates the forces on to the FPU model using a set of lines. In order for this method to work, it is important that both the numerical code and the actuator can work in real time at a frequency resolution that can properly replicate the mooring lines and risers. The method can be called a semi-coupled method as it takes into account the forces from the mooring lines without iterating the forces and responses to dynamic equilibrium in each step.
There are mainly two advantages of using a numerical model of the mooring lines. First and foremost, the depth of the ocean basin is no longer a limitation for the choice of model scale. This means that larger models can be used, and thus decreasing the uncertainties related to small scales. Secondly, the viscous forces on the numerical model can be correctly modelled as the drag coefficients can be manually set.
In order to assess the active hybrid model testing (AHMT) method, the method have been divided into two parts:
1. A simulator for the vessel will send displacement data to RIFLEX that simulates the lines. The forces obtained in RIFLEX will then be passed back to the simulator and taken into account.
2. Make a controller that can actuate a reference force on an object that moves. The controller will measure the actual forces applied, and correct itself onto the reference force.
The last part is treated in MSc. Student Kristian Dahl's masters thesis, and the first part is assessed in this thesis. In order to assess the system, two cases of a barge moored at 1000 meters water depth have been studied in scale 1 : 40.
The first case is a one degree of freedom (DOF) system of a barge moored free to translate in surge with two mooring lines subjected to a sea state with a significant wave height of 13 meters and a peak period of 13 seconds. The model of the barge is made in JAVA, and the mooring lines is modelled in RIFLEX. The JAVA model and RIFLEX model exchanges data in real time over the network using the high level architecture (HLA) simulation standard, and have been tested at working frequencies at 100 Hz, 67 Hz and 50 Hz.
The second case is a six DOF system of a barge moored with four mooring lines subjected to the same sea state as the one DOF system, now directed 45 degrees off from the barge longitudinal direction. The barge is modelled in JAVA, the mooring lines are modelled in RIFLEX, the data exchange is obtained using HLA and the working frequencies are the same as for the one DOF case.
The results from both cases where compared to a RIFLEX run with the simulated response as a forced top node motion. The results from this RIFLEX run would be in dynamic equilibrium, and the difference between this force history and the force histories obtained from the HLA run was used to assess the results. From the results, it was seen that the real time calculated forces from RIFLEX delivered close to correct forces regarding the amplitudes of the forces, but had a phase lag of four to five time steps. It was also seen that the JAVA code and RIFLEX could communicate in real time at frequencies higher than 100 Hz, and it was concluded that the real time capabilities of RIFLEX with HLA is feasible for the cases investigated in this thesis. | nb_NO |
