Numerical methods for modelling the viscous effects on the interactions between multiple wave energy converters

Abstract

The vast and rich body of literature covering the numerical modelling of hydrodynamic floating body systems has demonstrated their great power and versatility when applied to offshore marine energy systems. It is possible to model almost any type of physical phenomenon which could be expected within such a system, however, limitations of computing power continue to restrict the usage of the most comprehensive models to very narrow and focused design applications. Despite the continued evolution of parallel computing, one major issue that users of computational tools invariably face is how to simplify their modelled systems in order to achieve practically the necessary computations, whilst capturing enough of the pertinent physics, with great enough ‘resolution’, to give robust results. The challenge is, in particular, to accurately deliver a complete spectrum of results, that account for all of the anticipated sea conditions and allow for the optimisation of different control scenarios. This thesis examines the uncertainty associated with the effects of viscosity and nonlinear behaviour on a small scale model of an oscillating system. There are a wide range of Computational Fluid Dynamics (CFD) methods which capture viscous effects. In general however, the oscillating, six degree-of-freedom floating body problem is best approached using a linear potential flow based Boundary Element Method (BEM), as the time taken to process an equivalent model will differ by several orders of magnitude. For modelling control scenarios and investigating the effects of different sea states, CFD is highly impractical. As potential flows are inviscid by definition, it is therefore important to know how much of an impact viscosity has on the solution, particularly when different scales are of interest during device development. The first aim was to develop verified and validated solutions for a generic type decaying system. The arrangement studied was adapted from an array tank test experiment which was undertaken in 2013 by an external consortium (Stratigaki et al., 2014). Solutions were delivered for various configurations and gave relatively close approximations of the experimental measurements, with the modelling uncertainties attributed to transient nonlinear effects and to dissipative effects. It was not possible however to discern the independent damping processes. A set of CFD models was then developed in order to investigate the above discrepancies, by numerically capturing the nonlinear effects, and the effects of viscosity. The uncontrolled mechanical effects of the experiment could then be deduced by elimination, using known response patterns from the measurements and derived results from the CFD simulations. The numerical uncertainty however posed a significant challenge, with the outcomes supported by verification evidence, and detailed discussions relating to the model configuration. Finally, the impact of viscous and nonlinear effects were examined for two different interacting systems – for two neighbouring devices, and an in-line array of five devices. The importance of interaction behaviour was tested by considering the transfer of radiation forces between the model wave energy converters, due to the widely accepted notion that array effects can impact on energy production yields. As there are only very limited examples of multi-body interaction analysis of wave energy devices using CFD, the results with this work provide important evidence to substantiate the use of CFD for power production evaluations of wave energy arrays. An effective methodology has been outlined in this thesis for delivering specific tests to examine the effects of viscosity and nonlinear processes on a particular shape of floating device. By evaluating both the inviscid and viscous solutions using a nonlinear model, the extraction of systematic mechanical effects from experimental measurements can be achieved. As these uncontrolled frictional effects can be related to the device motion in a relatively straightforward manner, they can be accommodated within efficient potential flow model, even if it transpires that they are nonlinear. The viscous effects are more complex; however, by decomposing into shear and pressure components, it may in some situations be possible to capture partially the dynamics as a further damping term in the efficient time-domain type solver. This is an area of further work.