TED-AJ03-378 MIXING CONTROL BY ARRAY OF JETS WITH NEURAL NETWORK

概要

論文の詳細を見る
Control of scalar transport and local distribution in a turbulent flow is of significance for improvement of fluid devices. For a broad range of mixing control, an array of jets might have high potentiality compared with a single jet, because array of jets could be extended by larger number of operation variables, which is directly attributed to controllability of flow. Characteristics of an array of jets are different from those of a single jet as shown in Fig. 1. The flow structure can be classified into three regions. The first region near the nozzles is called developing region, where reverse flows are observed between the jets. The next region from a point where the reverse flow terminates and jets start to merge is named merging region. Finally, all jets fully coalesce and behave as the same as a single jet. This region is called combined region. In the present study, two types of array of jets were investigated in airflow. One was an array of triple parallel 2-dimensional jets (width 15mm, nozzle spacing 50mm and Reynolds number 4000-8000). The other was a staggered array of seven circular jets (diameter 13mm, nozzle spacing 17mm and Reynolds number 1700-9500). Three hot-wire probes were arrayed in line for sensors to detecting velocity distribution. Characteristics of the velocity field were examined by particle image velocimetry (PIV) and concentration field, measured by image processing of scattering intensity of tracers, was visualized in cross-section of the flow to evaluate shape and orientation of jets diffusion. We applied feedback control of mean velocity distribution in triple parallel jets. The exits velocity of the three jets were determined based on the detected velocity distribution by hot-wire probes in real time. The peak velocity position representing the mean velocity distribution was evaluated. Fluctuation intensity of the peak position, which was non-dimensionalized by the nozzle width, was reduced from 0.017 to 0.007,and the system became robust with the recovering time shortened for cross-flow disturbance as shown in Fig. 2. Concentration profile was controlled in a staggered array of seven circular jets, considering more practical uses. The ratio of the surrounding jets velocity can control the shape of concentration profile so that a directional shape was achieved as shown in Fig. 3. The center jet velocity controlled positioning of such profile formation, because it dominates inclination of the surrounding jet toward the center. The orientation of concentration profile was rotated as a time sequential control. Neural network can provide further complicated shape of velocity or concentration profile as a controller of the nozzle exits velocity. Array of jets can cover global practical uses as a functional fluid device, owing to its controllability of scalar transport.[figure]
一般社団法人日本機械学会の論文

TED-AJ03-378 MIXING CONTROL BY ARRAY OF JETS WITH NEURAL NETWORK

スポンサーリンク

概要

著者

関連論文

スポンサーリンク