Low-dimensional representation and sensor based estimation of bluff body wakes and vortex induced vibrations

点击翻译按钮获取中文摘要

This proposal is focused on the development of low-dimensional models of turbulent wakes and vortex induced vibrations which are amenable to sensor-based training/estimation for monitoring applications. Major applications of this research include: (i) maintaining the integrity of civil structures and mechanical engineering devices which may undergo significant unsteady cross-flow loading, (ii) monitoring and control of unsteady flows in industrial applications, (iii) the harvesting of unsteady fluid kinetic energy wind and water flows through exploiting flow-induced vibrations. These are of emerging importance in engineering and science due to the need for safe infrastructure and sustainable energy sources. The long-term objective of this research program (10+ years) is to develop a generalizable methodology which enables multi-sensor systems to be trained for real-time predictions and robust deployments. The research program can be divided into two phases. Phase I research is focused on advanced multi-scale decomposition techniques including the time-domain Spectral Proper Orthogonal Decomposition (SPOD) and multiscale Proper Orthogonal Decomposition (mPOD) for turbulent wakes and vortex induced vibrations. The essential hypothesis in these techniques is that it is possible and desirable to control the spectral nature of the data reduction such that dynamically relevant multi-scale phenomena may be represented in distinct modes, enabling an improved interpretation of the physical processes occurring within the flow. Moreover, these approaches may improve the separation of coherent structures from incoherent fluctuations. The goal of Phase I is to experimentally test the aforementioned hypothesis through relevant and practical engineering problems. Phase II research involves sensor-based estimation of turbulent wakes and vortex induced vibrations. The low-dimensional subspace established through SPOD and mPOD within Phase I will serve as the estimation basis. Several techniques including Linear and Quadratic Stochastic Estimation, and more modern approaches using Artificial Neural Networks will be utilized in developing mapping functions between the sensor signals and the desired modes to be estimated. The significant experimental component of the research program will be carried out in the applicants operational research water tunnel facilities which are equipped with state-of-the-art flow diagnostic equipment. The research program will contribute fundamental advancement in physical interpretation and modeling of turbulent wakes and vortex induced vibrations in practical engineering applications and support the development of sensor-based estimation techniques for these flows. The demonstrated generalizability of the developed flow modeling and sensor-based estimation techniques will be a catalyst for industry partners in Canada to implement these techniques for multi-sensor system monitoring (safety, integrity) and control.