An active design method for high-precision heat transfer correlation

Accurate heat transfer correlations are necessary to ensure the efficiency and safety of the cycle system. The selection of correlation factors, dimensionless forms, and correlation forms are critical factors to affect precision of the heat transfer correlations. However, existing techniques usually couple these selection processes, without independently optimizing each, which limits universality and accuracy of the correlations. To overcome this problem, this study proposes an active design method for deriving high-precision heat transfer correlations. Based on numerical model, active design of the heat transfer correlation is achieved by decoupling the correlation factor, dimensionless form and correlation form selection process. In order to verify the effectiveness of the proposed method, the prediction performance of the heat transfer correlations obtained by the active design method and the traditional method is compared based on experiments and a one-dimensional dynamic model, taking the construction of the heat transfer correlation of CO2 in the new variable cross-section printed circuit heat exchanger as an example. The results show that the non-linear functional relationship between density and Nusselt number (Nu) is better characterised by a quadratic polynomial function than by a power function. The maximum error Nu obtained through active design method is 14.99 % while the maximum error Nu obtained through traditional design method is 26.32 %. The active design method has the prediction accuracy of 99.40 % for outlet temperature. Near the near-critical region of sudden thermophysical change, the maximum error of the active-design method is 3.20 %, while that of the traditional design method is 5.93 %.