2000 Volume 26 Issue 4 Pages 596-603
The progress of a mixing pattern is calculated with the knowledge of displacement of fluid particles obtained by integrating the velocity vectors with respect to time. In time-periodic flow, especially the mixing pattern at a discrete time whose interval is equal to multiple periods is able to be estimated with the knowledge of the fluid motion during only one period. In this study, the new discrete model is proposed, and this model enables expression the mixing process in the time-periodic laminar flow. Both the independent variables, time and space, and the dependent variable, concentration, are digitized there. As a result, continuous fluid motion is discretely treated, and the change in mixing pattern is expressed as rotation of a vector representing the mixing pattern. A new algorithm is developed in order to reproduce precisely the fluid motion during one period. In this algorithm, the combinational optimization method based on the neural network theory is used together with the domain reduction method. By use of the discrete model, the progress of mixing pattern in laminar flow is able to be calculated fast and precisely.