In recent decades, CFD has been used extensively both in the industry and in the research field, but it’s still facing many difficulties. Perhaps one of the most notable difficulty is the automation of the numerical mesh generation. Current meshing algorithms only considers the topology of the geometry, therefore engineers have to take into account the flow characteristics themselves and modify the mesh accordingly. There have been several publications in the nineties and the 2000s where machine learning has been applied to numerical mesh generation or mesh optimization. These approaches have few downsides. They often require some sort of preprocessing, and most of them use meshes as training data. Using a diverse training data set with good quality is crucial for training a machine learning model, since its performance is highly dependent on is. With the usage of a physics informed neural network the need for meshes to generate training data could be prevented. The model can predict the solution of a wide variety of differential equations, thus can be used to approximate a flow field. The approximate then can be combined with a meshing algorithm to produce a graded mesh which considers physical properties alongside the domain’s properties In this work the variational tetrahedral meshing algorithm by Aliez et al. was combined with the physics informed neural network to produce mesh grading. The algorithm can adjust the mesh density utilizing a weight function. The idea is to compute a weight value for each node in the function of the predicted flow field, the larger the weight the denser the mesh be around that node. The paper describes a geometry-optimal function, but it is also possible to implement other functions.Read
Floating sensor buoy devices (FSBDs) are used mainly to gather data from underwater and water surface to respond to naval as well as maritime industry requests. Due to their complex access on the sea, the FSBDs are self powered. Among the renewable energy options, wave energy converters (WECs) have many advantages. One particular method to extract the consistently dense, sea wave energy, is a via heaving buoy, equipped with an internal, inertial power take-off (PTO), thereby acting as a point absorber. A noted unresolved problem for this type of WEC, is large amplitude pitch/roll motion. This has a two-fold negative effect on the FSBD system: (1) Reduction of electrical energy from the WEC system (2) Deteriorated wireless sensor network capability, due to misalignment of vertical antenna. Parametric resonance can be the cause of very large amplitude, unstable, pitch/roll motions in floating bodies. However, this nonlinear phenomenon, has not previously been implemented in the literature for this type of WEC-FSBD. The objective of this study is therefore to include the effect of parametric resonance, when ensuring the vertical stability of a WEC-FSBD device. Read
The oceans are one of the largest resources of mankind, covering more than 70% of the Earth’s surface and maintaining huge industries such as fishing, shipping, tourism, offshore oil, gas and renewable energies. Numerical wave tanks (NWTs) can be a vital tool for understanding, designing and optimizing systems in the ocean. A key element of the NWT is the wavemaker, which is responsible for generating NWT waves. For the wavemaker to generate desired water waves for modelling of nonlinear free surface waves, hydrodynamic forces and floating body motions, fine-tuning and iterative calibration are required, which is computationally intensive and thereby increasing the number of CFD simulations to run and the user’s work. This study reports on a numerical wavetank implemented in OpenFOAM and improving the wave generation capabilities of impulse source wavemaker method by employing data-driven system identification and machine learning techniques to identify models mapping a desired wave surface-elevation time- trace at a probe to the wavemaker input required to create it. A comparative study of performances of various model structures is presented for both forward and inverse NWT problems based on their modal accuracy, complexity and computational cost. It was demonstrated that Autoregressive with exogeneous inputs emerged as the best method to model the dynamic behaviour of NWT satisfactorily when trained for a single training example of medium sized amplitude. ARX was able to validate well over the datasets with amplitude variations of almost 10 times the amplitude of the actual training dataset.Read
Nowadays, with the development of technology, we try to avoid wire usage and reduce the size of devices and their power consumption. As power consumption is decreasing due to the development of new devices, this allows us numerous opportunities for monitoring low-powered wireless sensors. Energy harvesting is one of the most eective methods of contributing the power of these sensors. This applied technique allows us to convert lost energy from the process into new energy for a new process. This will reduce the maintenance cost of the sensors and will contribute to the long-term functioning of them. Sensors will be able to operate continuously with no need to replace batteries or use extremely complicated wired systems. Wireless energy harvesting devices also open to us new opportunities from geographical aspects, as it is now easier to provide sensors for pumps that operate on deep-water platforms or marine machines. The author will discuss vibration energy harvesting devices and will show how dierent types of vibration harvesting devices can be used to capture the mechanical energy parasitically lost to the environment by the operating pumps. Mostly concentrating on piezoelectric energy harvesting devices of various designs and power outputs author will also try to develop a model to obtain an optimal design with optimal power output for sensors. Read
The popularity of electrical power generation from renewable energy resources has been increasing a lot in the last decades. Ocean energy in addition to other traditional renewable resources has been sparking a greater interest in researchers. This is because it appeared as an important and promising renewable energy resource due to its enormous reserves and wide distribution. It might be a practical long-term solution for satisfying the electricity demand for countries with coasts facing the ocean. Ocean energy includes many types, the most important one is wave energy. The potential of this source of energy is great, and exploiting it efficiently can be a great step towards meeting renewable energy targets. Oscillating water columns (OWC), are a type of wave energy converters which operate by converting wave energy into pneumatic energy, whereby wave oscillations change water levels inside of a chamber to force entrapped air through a turbine. One potential drawback of this concept is the noise generated by the turbine. In order to investigate and reduce the noise created by OWC turbines, an acoustic model of the turbine operation must be developed. However, to replicate the turbine operation in the acoustic model, it is first necessary to model the rest of the OWC system to provide information on the operating conditions. This project will involve the development of turbine model, using the Open-Source CFD software OpenFOAM, which is to be coupled to a Numerical Wave Tank (NWT).Read
Coastal erosion is an ongoing natural phenomenon, which can reduce coastlines. Nowadays, nourishment and coastal barrier protection methods are used to tackle this issue. Recently, an alternative solution is developed using wave transformation tech- niques. The main idea is to use a variable bathymetry to shift the phase of the part of the wave front to be 180 degrees out of phase with the other parts of the wave front so that they cancel each other out. In this thesis, a Numerical Wave Tank (NWT) is created to create a simulation of the alternative solution. The NWT provides a reliable numerical tool for bathymetry anal- ysis, enabling cost-effective test bed for analysing multiple geometries. To create a model with high accuracy and low computational cost, numerical result is validated using similar bathymetry case and mesh convergence study is conducted as well. Volume of fluid model and shallow water equations model are used. The first is able to provide a phase difference and an accurate wave characteristic. From the phase differ- ence results, the most efficient shoal height, which shows nearly 180 o phase difference, can be found from several cases presented. In addition, wave breaking phenomenon when the shoal has certain height is also observed. However, at large computational cost, restrict- ing its use to 2D simulations and it also cause a limitation in using a lot of geometries. The later, The shallow water equations are much faster and could be utilised for 3D, however in the test cases performed poor accuracy was delivered. Being able to make this model work will be beneficial for further study.Read
Wave energy converters (WEC) are part of the renewable energy sources that harness the energy of the oceans. The development of WECs pioneered the creation of Numerical Wave Tanks (NWT) where wave solid interaction (WSI) is modelled. The NWT has been under development for a number of years for the purpose of testing and optimizing WEC design. To date, the OpenFOAM NWT has only considered single floating bodies, however some classes of WECs comprise of multiple floating bodies. Multiple floating bodies were were investigated in the NWT in two version of Open- FOAM numerical solver, Foundation version 6 and ESI version 18.12 was used. The investigation was focused on meshing strategies for NWTs and floating bodies and two different body motion solvers rigidBodyMotion and sixDoFRigidBodyMotion compared for better implementation of floating bodies. The results of the simulations presented in this work shows that OpenFoam capable modelling multiple floating bodies in NWTs. Read
This thesis will involve development of a Numerical Wave Tank (NWT) using ANSYS Fluent. The NWT has already been under development using the Open-Source CFD software OpenFOAM, for the purpose of testing and optimising marine renewable energy devices, such as wave energy converters. This thesis have analysed experimental data of regular and focused wave kinematics from a test campaign conducted in July 2018 in the COAST laboratory in Plymouth University, UK, Then establish a two-dimensional numerical wave tank according to the entity of the wave flume in the experiment, and realize the verification of the NWT experiment results by comparing the numerical result with the experimental data. In the experiment, the resistance wave probe was used to measure the free surface elevation and the particle image velocimetry (PIV) to measure the particle velocity. The first set of simulation is performed on regular waves, and it is expected that the simulation settings will be verified by comparing the results with the expected Stokes theory. And also made a verification of wave maker. It is found that NWT can produce the desired wave shape very well by using dynamic grid technology to simulate the motion of the wave maker. The second set of simulation was performed on focused waves. Compared the free surface elevations at different locations between the numerical results and experimental data. It can be found that the numerical results don’t have a good agreement with the experimental data along the wave tank to reproduce the process of the wave focusing. Because there is a large error between the amplitude of the free surface elevation of numerical wave tank and experimental results. So the NWT was not able to accurately reproduce the focused waves tested experimentally in the wave flume. Read
In this document, experimental results from previous work were analyzed then used in order to model a waveflume in OpenFoam. The theories are firstly studied to extract all the essential parameters that are necessary to justify the good use of the numeric model. A specific case was used in the intermediate water waves field. It was found that, despite the use of the laminar type of simulation, the extrapolation method and the Wheeler stretching method are not describing well the results obtained by the simulation. Further work on different type of simulations must be made to confirm or invalidate the previous assessment. Read