Uncertainty quantification methods applied to wind energy systems

考虑到能源需求和资源可用性的数量级，风能是最有前途的可再生能源之一。风力资源在全球广泛分布，而天气预报与需求响应和存储技术正在实现大规模的电网集成。风能发电机可以大规模生产并逐步安装。传统的“丹麦概念”3叶片涡轮机设计的技术进步使10兆瓦以上的机器在商业上可行。GW规模的底部安装的海上风力涡轮机阵列现在是工业实践。一个漂浮的海上风力涡轮机阵列刚刚首次商业化部署，在全球更深的沃茨开辟了额外的风力资源。还正在开发新生的空中风能系统（AWES），其在系绳上放飞风筝/滑翔机以捕获风能，以逐步减少风能转换器的材料要求。大规模AWES如果成功，将理所当然地部署在海上。 所有这些系统的核心是风本身，风本身就是湍流，对结构施加随机载荷，并导致可变的功率输出。在近海，波浪力是一种附加的随机力。快速准确地分析随机载荷的能力是探索新设计概念和提高性能的关键。拟议的研究计划将建立在PI研究小组的大量工作基础上，这些工作涉及应用于能源系统的一系列随机方法，包括将具有不同保真度水平的不同模拟器（空气动力学，结构）融合在一起，以影响系统参数和配置的鲁棒优化。将要开发的随机方法将能够分析风能系统，并直接计算其性能和寿命的不确定性措施，而不是依赖于稳态性能的简化近似。拟议的研究将特别关注开发和应用随机方法来量化浮动常规和AWES的不确定性，然后实现替代概念的设计和优化研究。第一个应用是风力发电场设计，需要快速建模的疲劳负载的涡轮机阵列的基础上，当地的非定常流动。第二个应用程序是在一个单一的机器设计优化循环。风力涡轮机操作从来不是稳定的，因此稳定模型的优化不是最优的或有效的。准确、高效和直接计算对不稳定风和波浪输入的随机响应的能力开辟了使用更有效的优化方法来探索和优化广泛的风力发电系统设计的可能性。

Wind energy is one of the most promising renewable energy sources given order of magnitude energy requirements and resource availability. Wind resources are widely distributed globally, while weather forecasting with demand response and storage technologies are enabling grid integration at-scale. Wind energy generators can be mass produced and installed incrementally. Technology progression in conventional `Danish concept' 3-bladed turbine design has enabled 10+ MW machines to be commercially viable. GW-scale arrays of bottom-mounted offshore wind turbines are now industrial practice. A floating offshore wind turbine array has just been commercially deployed for the first time, opening up additional wind resources globally in deeper waters. Nascent airborne wind energy systems (AWES) that fly kites/gliders on tethers to capture wind energy are also being developed, to pursue a step reduction in material requirements for wind energy converters. Large-scale AWES, if successful, will logically be deployed offshore. At the heart of all these systems is the wind itself, inherently turbulent and imposing stochastic loading on the structures and causing variable power output. Offshore, wave forcing is an additional stochastic forcing. The ability to quickly and accurately analyze stochastic loading is key to exploring new design concepts and improving performance. The proposed research program will build on the PI's research group's large body of work on a range of stochastic methods applied to energy systems, including fusing together different simulators (aerodynamics, structures) with heterogeneous levels of fidelity to affect robust optimization of the system parameters and configurations. The stochastic approaches to be developed will be capable of analyzing wind energy systems and directly computing uncertain measures of their performance and life time, rather than relying on simplified approximations of steady state performance. The proposed research will focus specifically on developing and applying stochastic methods for uncertainty quantification of floating conventional and AWES, to then enable design and optimization studies of alternative concepts. The first application is to wind farm design, requiring fast modeling of fatigue loading for turbine arrays based on local unsteady flows. The second application is inside a single machine design optimization loop. Wind turbine operation is never steady, hence optimization of a steady model is not optimal or efficient. The ability to accurately, efficiently and directly compute the stochastic response to unsteady wind and wave inputs opens up the possibility of using more efficient optimization methods to exploring and optimize a wide range of wind generation system designs.