This paper deals with a fundamental study on the control problem of distributed parameter and nonlinear flow fields using a high-performance supercomputer as a controller. A state observer is proposed as a basic element of the control system to estimate the complete state from the mathematical model and partial measurements. A SIMPLER-based flow simulation algorithm is used as the mathematical model of the system. The error of the numerical solution evaluated by comparison with the measurement is fed back to the mathematical model to reduce the discrepancy. The adoption of this feedback loop distinguishes the state observer from ordinary flow simulations. With this feedback loop the computational result may converge even for unstable flows with fluctuations. A numerical simulation has been performed for a turbulent flow through a square duct to confirm the validity of the present state observer. A real flow is simulated using a standard numerical solution of a developed turbulent flow. Appropriate choice of the proportional feedback gain accelerates the convergence in the simulation by a factor of 60 and reduces the error in estimating the perturbation velocity component by a factor of 0.5 in comparison with he ordinary flow simulation without a feedback loop.
|ジャーナル||Nippon Kikai Gakkai Ronbunshu, B Hen/Transactions of the Japan Society of Mechanical Engineers, Part B|
|出版ステータス||Published - 1996|
ASJC Scopus subject areas