Estimation of numerical uncertainty in computational fluid dynamics simulations of a passively controlled wave energy converter
Estimation of numerical uncertainty in computational fluid dynamics simulations of a passively controlled wave energy converter
复制标题
DOI：
10.1177/1475090217726884
复制
发表时间：
2018-02
期刊：
影响因子：
--
通讯作者：
Weizhi Wang;Minghao Wu;Johannes Palm;C. Eskilsson
中科院分区：
文献类型：
--
作者：
Weizhi Wang;Minghao Wu;Johannes Palm;C. Eskilsson
传统上，浮式波能转换器的波浪载荷和由此产生的运动是用线性辐射绕射法计算的。然而，对于某些情况，如生存条件、相位控制和工作在共振区的波能转换器，更多的是更完整的数学模型，如计算流体力学，在过去的5年里，计算流体力学在波能场中的应用越来越频繁。然而，在波能领域，对与计算流体力学模拟相关的数值误差、收敛速度和不确定性的严格估计在很大程度上被忽视了。在本文中，我们将形式验证和确认技术应用于被动控制点吸振器的计算流体力学模拟。即使对于近乎线性的入射波，相位控制也会导致运动响应高度非线性。首先，我们证明了计算流体力学模拟与实验数据具有可接受的一致性。然后，我们提出了一个验证和验证研究，重点是解决方案验证，包括空间和时间离散化，迭代和域建模误差。结果表明，空间离散化是误差的主要来源，但时间误差和迭代误差是不可忽略的。采用低纵横比的六面体单元，每波高30个单元，对没有相位控制的浮标，运动响应(除涌浪外)和约束力的不确定度小于5%。由相位控制引起的放大的非线性响应导致数值不确定性的大幅增加，这说明对于高度非线性响应很难获得可靠的解，并且这种情况需要更密集的网格。
The wave loads and the resulting motions of floating wave energy converters are traditionally computed using linear radiation–diffraction methods. Yet for certain cases such as survival conditions, phase control and wave energy converters operating in the resonance region, more complete mathematical models such as computational fluid dynamics are preferred and over the last 5 years, computational fluid dynamics has become more frequently used in the wave energy field. However, rigorous estimation of numerical errors, convergence rates and uncertainties associated with computational fluid dynamics simulations have largely been overlooked in the wave energy sector. In this article, we apply formal verification and validation techniques to computational fluid dynamics simulations of a passively controlled point absorber. The phase control causes the motion response to be highly nonlinear even for almost linear incident waves. First, we show that the computational fluid dynamics simulations have acceptable agreement to experimental data. We then present a verification and validation study focusing on the solution verification covering spatial and temporal discretization, iterative and domain modelling errors. It is shown that the dominating source of errors is, as expected, the spatial discretization, but temporal and iterative errors cannot be neglected. Using hexahedral cells with low aspect ratio and 30 cells per wave height, we obtain results with less than 5% uncertainty in motion response (except for surge) and restraining forces for the buoy without phase control. The amplified nonlinear response due to phase control caused a large increase in numerical uncertainty, illustrating the difficulty to obtain reliable solutions for highly nonlinear responses, and that much denser meshes are required for such cases.