No abstract

No abstract

No abstract

The leakage problem of supercritical carbon dioxide (SCO2) axial-inflow turbine brings great challenges to the efficiency and security of the power system. Labyrinth seals are usually utilized to improve the leakage characteristics of the blade tip. In this paper, a 1.5-stage SCO2 axial-inflow turbine is established and labyrinth seals are arranged on the top of the first stage stator and rotor blades. The effects of seal clearance, groove on seal cavity surface and circle groove shape on flow characteristics and aerodynamic performance under different pressure ratio are investigated. Increasing seal clearance can significantly weaken the turbine performance. Arranging rectangle, circle and V-shaped grooves on the seal cavity surface near the outlet of the seal gap can enhance the energy dissipation, reduce the relative leakage and improve the power and efficiency. Increasing the groove width can improve the aerodynamic performance while the effect of the groove depth is weak. The configuration where the circle groove width is 50% of the pitch of seal tooth achieves the best performance with the relative leakage of stator1 and rotor, power and efficiency of 6.04 × 10−3, 8.09 × 10−3, 3.467 MW and 86.86% respectively. With an increase in pressure ratio, the relative leakage increases firstly and then remains almost constant. The power increases while the efficiency increases firstly and then decreases, reaching the peak value under the design condition.

Aiming at the problems of a long design period and imperfect surrogate modeling in the field of airfoil design optimization, a convolutional neural network framework for airfoil design and performance prediction (DPCNN) is established based on the deep learning method. The airfoil profile parameterization, physical field prediction, and performance prediction are achieved. The results show that the DPCNN framework can generate substantial perfect airfoil profiles with only three geometric parameters. It has significant advantages such as good robustness, great convergence, fast computation speed, and high prediction accuracy compared with the conventional machine learning method. When the train size is 0.1, the predicted results can be obtained within 5 ms. The prediction absolute errors of physical field of most sample points are lower than 0.002, and the relative errors of aerodynamic performance parameters are lower than 2.5%. Finally, the optimization attempt of operating parameters is completed by gradient descent method, which shows good stability and convergence. Overall, the DPCNN framework in this paper has outstanding advantages in time cost and prediction accuracy.