自然循環ループ内安定・不安定流のカオス解析

－温度時系列データのフラクタル次元－

抄録

An experimental investigation has been made on the flow in a natural convection loop. Time series data of temperature oscillation were obtained for various types of the flow. Based on these data, attractors in high-dimensional space have been reconstructed by obtaining optimum delay time. The correlation dimensions D_c for these attractors were calculated by the method of correlation integration and plotted against heat inputs. The values of D_c could be classified into three regions according to the value of heat flux q, 1) D_c ≥ 13 in the laminar region (q ≤ 4.13kW/m²), 2) 1.1 ≤ D_c ≤ 2.4 in the unstable region (5.51 ≤ q ≤ 22.0), and 3) D_c ≥ 11.5 in the turbulent region (q ≥ 23.4). The values of D_c are non-integer in all regions. These results are consistent with chaotic characteristics. Observed in detail, however, the region 2) contains three types of flows: Turbulent, unstable with and without flow reversals. The last one was considered to be unstable flow in spite of non-reversal, because of its small D_c.