Table 1 lists more information structure from the physical model test. Two cases modelled in the present paper Figure 2 shows the two structure configurations modelled in the present paper, named config A and config C. The difference between the two Config Calm water Wave height, m Wave period, s configurations is that the C config includes a large horizontal plated depth, m solid deck, risers on both front columns and a small grated screen Case 1 C 0. Although the regular braces were neglected based on a CFD sensitivity study proving that waves did not break during the physical test program, some minor the braces do not notably impact the hydrodynamics.
The horizontal spilling at the wave crest was observed. The actual paddle motions from the physical model study were assigned by CFD modelers.
This time-series of In the VOF method, two phases are represented with a single fluid paddle motions is given in Fig. More information about the whose density and viscosity is a function of! Time-series of the horizontal location of paddles generating the Boundary Conditions irregular long-crested waves for Case 2.
Two laws of physics relevant to this problem are the Because of the symmetry of waves and configurations, shown in Fig. OpenFOAM just enough to prevent water to reach the top boundary plane. For Case 1, the length of the tank along the direction of dynamics with a powerful body-fitting meshing utility. The center of the structure is thus VOF is an interface capturing method.
To track the paddles. The total size of the tank along the X axis for Case 2 is The size of the tank along the Y axis was! In the present study! This equation is then only applicable to the interface region. The third term The mesh cells are cubical equilateral away from the structure and in the left-hand side of the above equation is not needed for the hexahedral and split hexahedral to conform to the structure as one theoretical representation of the interface and is only added for the approaches the structure.
BlockMesh and SnappyHexMesh utilities of numerical stability and the conservation of a sharp interface. The pressures, global forces, and the main overturning moment were cubical cell size for Case 2 is 0. Figure 5 shows locations and names approximately 12 cells. Mesh is not refined at the water-air interface.
Case 1 was run on 50 parallel processors and each wave period took approximately 1. Case 2 was run on 90 processors and Computational time step size changes automatically during both simulation cases to satisfy accuracy and convergence conditions. For both cases, typical order of magnitude of time step size is s. Based on the wave theory chart by Le Mehaute , Stokes IV is the most suitable wave theory for this case. We first investigated how waves propagate and evolve with time and location in the numerical tank in the absence of the structure.
We attempted to generate waves based on three different wave theories b separately: Stokes V, 5th order stream function theory, and cnoidal theories provided by IHFOAM.
For the 5th order stream function wave generation a calculation, based on Fenton, , was required prior to Fig. Locations and names of sensors.
When the wave condition structure and flow speeds, b water surface elevations for this case, given in Table 1, is input to the CFD model, generated waves become unrealistically unstable and break after a short distance. The present CFD model correctly predicts the X component of the flow This instability is regardless of the wave theory used.
Instability of at three different locations, Fig. Note that all velocity probes are interface for steep waves Afshar, Wave steepness wave on the symmetry plane perpendicular to Y axis. Modelled waves satisfy target trough and crest elevations for Case 1. After successfully reproducing the highly nonlinear regular waves in the numerical tank, Fig. To avoid the interaction of undesirably evolved waves as they propagate in the domain, as c discussed above, the center of the structure was positioned Fig.
Modelled speeds are consistent with measurements. The X approximately 3. In component of velocity at three different locations shown in Fig. Results were very consistent with measurements, Fig.
Physically generated waves have slightly smaller period than the target wave period. This is the main reason why the time values associated with the extremums of most of the modelled quantities do not match those of the measured.
Modelled run-ups are generally consistent with measurements. Peak values of modelled local pressures on columns tend to be generally slightly larger than the measured values. This is true for both the underwater sensor, AB53, and other pressure sensors located above the still waterline.
In the physical model study, the underside of the top deck had been instrumented with several local pressure sensors. Modelled pressures for some of those sensors were compared with b measurements; however, the present model results were not consistent with measurements. Impacts on the underside of the deck are extremely complex and stochastic, and even the physical model had issues of repeatability of impact events during duplicated tests. Whether increasing the output frequency of the CFD model improves the estimation of the local c pressures under the deck was not investigated.
Modelled air gaps are consistent with measurements. Note that waves for the present case are high enough to interact with the large horizontal plated solid deck whose lower surface is approximately 0. Generation of the numerical waves was based on the actual paddle motions, Fig. Unlike Case 1 where waves were generated at a stationary boundary, d for Case 2 waves were generated by the horizontal motion of the boundary similar to the physical test program.
In this subsection we have briefly studied the applicability of OLAFOAM for the reproduction of the physical test results for the irregular wave condition. It is known that the mesh cell size impacts the numerical diffusion, an e unwanted behavior of most of mesh-based CFD models. This diffusion Fig.
Modelled local pressures are higher than measurements for some could lead to the damping of waves propagating in a numerical tank. In of wave cycles except for the underwater sensor AB53 where modelled the context of the mesh-based CFD modeling of waves, it is generally peak values are slightly larger for all wave cycles.
For the present Case 2 involving Global X and the dynamic Z components of the force, and the Y irregular waves with heights ranging approximately from 0. The X component, in-line with the propagation because computational time will be unpractically large. The dynamic Z component of the force shows two local peaks, associated with the We studied the applicability of OLAFOAM for irregular wave impact of run-ups with the top deck; the existence of these peaks are generation in the basin in the absence of the structure and also the correctly predicted by the present CFD model although the peak values effect of cell size on modelled waves.
Two cell sizes were considered: a are generally slightly underpredicted, Fig. The Y component of the overturning moment is very large computational time. Nevertheless, the Modelled waves at three different locations were compared with modelled overturning moment is consistent with measurements, measurements from the physical model study.
Modelled waves have particularly for the peak values, Fig. Particularly, the largest wave is very well reproduced. Results of the coarse 3D model do not significantly differ from those of the fine 2D model.
This confirms that results are not considerably mesh dependent from the course mesh to the fine mesh considered here. Whether further mesh refinements improve model results was not investigated.
First, by considering the time-varying control delays and earthquakes, a delayed dynamic model of the marine structure subject to earthquake is established. The simulation results show that the vibration of marine structures caused by earthquakes can be effectively reduced through the designed control scheme.
Skip to main content. This service is more advanced with JavaScript available. Advertisement Hide. Conference paper First Online: 19 October This is a preview of subscription content, log in to check access. Si, Y. Zhang, B. Springer, Singapore Wang, S. China Ocean Eng. Som, A. Basic Appl. Wu, B. Control Health Monit. Kazemy, A. ISA Trans. Moharrami, M. Neurocomputing , — CrossRef Google Scholar. Ma, H.
0コメント