## Importance of Fluid-Structure Interaction on Dropped Lifeboats - A parametric study used to explore the importance of hydroelasticity on complex low rigidity structures using decision factors

##### Master thesis

##### Permanent lenke

http://hdl.handle.net/11250/2350706##### Utgivelsesdato

2015##### Metadata

Vis full innførsel##### Samlinger

- Institutt for marin teknikk [2923]

##### Sammendrag

Experimental methods have previously been used in the design of lifeboats; this approach is both time-consuming and expensive. The experimental approach is suitable to find the loads and accelerations on an existing design; however, it is not very useful for the development of a new design. In order to continue the innovation and development of lifeboat design there has been a shift towards the application of numerical tools. Today, Computational Fluid Dynamics (CFD), Finite Element Analysis (FEA), model and full-scale experiments are used in research, investigation and development of lifeboats. The use of CFD and FEA is increasing and there has been demonstrated a general good agreement between numerical results and full-scale tests. However, the elastic deformations on the hull during impact are normally ignored and the validity of this simplification is questioned since large deformations have been observed. A possible solution to this problem is to perform Fluid-Structure Interaction (FSI) analysis, where CFD and FEA are coupled. Today, the most common procedure is to use the pressure from the CFD analysis as boundary conditions for the FEA analysis. With this approach the deformations during impact are ignored. A more robust, but expensive, solution is to perform a two-way interaction analysis where information is exchanged between the CFD-domain and FEM-domain during the analysis. In this way the deformation during impact is addressed. For lifeboats, this method is most applicable during slamming or more specifically when hydroelasticity occurs. Hydroelasticity is a phenomenon which is propagated in the structure as fast vibrations, which will alter the dynamic loading and boundary condition on the structure.
The overall purpose of this thesis is to explore the importance of fluid-structure interaction, or hydroelasticity, on lifeboats. Specifically, the goal is to explore when it is important to account for hydroelasticity for complex low rigidity structures using decision factors. By exploring decision factors, one can more easily understand if it is necessary to account for hydroelasticity in numerical simulations. The main body of work has been divided into two main tasks: 1) A verification study and 2) a parametric study of different impact velocities, deadrise angles and material properties, in order to explore the importance of hydroelasticity using decision factors.
A brief introduction to the theory behind CFD and FEA is presented along with fluid-structure interaction and different coupling algorithms. Hydroelasticity is described and multiple decision factors are presented.
In this thesis, Star-CCM+, developed by CD-Adapco, is used in the CFD calculation while Abaqus, developed by Simula, is used in the FEA calculations. A co-simulation engine developed by Simula couples the two softwares together in order to exchange information. Star-CCM+ transfers the pressure and wall shear stress to Abaqus while Abaqus transfers the nodal displacements to Star-CCM+.
In order to verify that these two softwares were able to recreate a hydroelastic event, a verification study of a full-scale test was performed. A full 3D FEA model and a full-scale test report were supplied by the manufacturer. A 3 meter free-fall drop test of a conventional lifeboat was performed, where the acceleration on multiple places on the boat was measured. The model was reduced to 2D and altered with respect to geometric stiffness using springs and additional mass. Convergence studies for both FEA and CFD were performed. The mass was tuned in such a way that the maximum acceleration in 2D would be equal to the maximum acceleration in 3D, this is necessary due to the pressure distribution in 3D. The acceleration from the co-simulation was compared to the acceleration at the same location in the full-scale test. The results are in good agreement with the test results and they yield approximately the same maximum acceleration and oscillation period even though the mass have been tuned. From the spectrum of the pressure time series we can observe that a hydroelastic event was recreated as we can observe several distinctive peaks. We can also see from the pressure time series the pressure oscillations which are associated with hydroelasticity. The results are considered satisfactory and it is believed that the method can be trusted.
A total of 120 simulations were performed with a range of six different deadrise angles, five impact velocities and four different material properties in the parametric study. In addition, 16 models with the verification study model are simulated where only the impact velocity is altered. The maximum strain in a node in a platefield was evaluated. The results demonstrates a clear trend that was consistent with previously published articles. As the deadrise angles decreases, the impact velocity increases, and the stiffness decreases, the importance of hydroelasticity will increase. The decision factor R_B defined in Bereznitski2001 predicts that the response should be hydroelastic, while the factor R_F defined in Faltinsen1999 predicts that the response should almost be quasi-static. The quasi-static strains is found using Euler-Bernoulli beam theory. According to the results, all the simulations are dominated by hydroelasticity; hence $R_B$ is the best suited decision factor for this problem. It is also observed that the when the importance of hydroelasticity increases, the maximum strain in the plate decreases in relationship to the quasi-static strain.
Error sources are discussed and it was discovered that the simulations are unstable, due to the fact that a repetition study yields a different maximum strain, while the profile of the strain time history is very similar. The uncertainty of the results was studied and it was discovered that the uncertainty for the mean of the nine simulations are 10.8 \% within 95 \% confidence interval. The main reasons are assumed to be the interaction with the eigenmode introduced by the springs, the presence of bifurcation points, and the capabilities of the implicit dynamic solver in Abaqus. Although instabilities and an uncertainty are observed, the results are regarded as representative and the model is able to describe hydroelastic events. This conclusion is supported by \cite{Faltinsen2005} as he observed some variance for the strain when an experiment was performed with a given impact velocity.
The influence on the point of flow separation is investigated by applying different viscous models.It is concluded that the point of flow separation is not affected by different viscous models as the flow follows the hull and only separates at the edge. The effect from different viscous regimes cannot be clearly observed on the translation, velocity or acceleration of the hull. This can be explained due to the fact that the pressure time history is fairly similar and only small differences in the magnitude of the pressure exist, this is not significant as it is the impulse of the slamming load which is important for the response. It is concluded that the difference from a laminar flow to a turbulence model is negligible.
Further work is also discussed.