基于机器学习的LNG气化器排水自由跌水换热入口水流方向优化研究

Optimization of Water Flow Direction for Free-Falling Water Heat Exchange in LNG Vaporizer Drainage via Machine Learning

摘要: 本文基于计算流体动力学对基于自由跌水的江水源LNG气化器排水升温过程进行了数值模拟，获得了在不同入口流速下及入口水流方向角对排水升温性能的影响规律。水流与渠道底部接触时最大速度随着入口水流方向角的增大而减小。当水流方向角在0˚到60˚之间时，出口温度随水流方向角的增大而增大；当水流方向角超过60˚后，出口温度随水流方向角的增大而减小。基于机器学习构建了入口水流方向角、入口流速与出口温度之间的预测模型。进一步基于梯度下降算法，以最大化出口水温为目标，获得了在入口流速为3 m/s时，最优的入口水流方向角(59.798˚)，此时水流的出口温度为280.5 K，相较于入口水流方向角为0˚时，出口温度提升了3 K。

Abstract: Based on computational fluid dynamics, this study conducted a numerical simulation of the heating process of LNG vaporizer drainage using free-falling river water. The influence of different water flow direction angles on the heating performance under varying inlet flow velocities was investigated. The maximum velocity when the water flow contacts the channel bottom decreases as the inlet flow direction angle increases. The outlet temperature increases with the increase of the flow direction angle when the angle is between 0˚ and 60˚. However, when the angle exceeds 60˚, the outlet temperature decreases as the angle further increases. Furthermore, a machine learning-based prediction model was developed to establish the relationship between the inlet flow direction angle, flow velocity, and outlet temperature. Using the gradient descent algorithm with the objective of maximizing the outlet water temperature, the optimal inlet flow direction angle was determined to be 59.798˚ at an inlet flow velocity of 3 m/s. At this angle, the outlet water temperature reaches 280.5 K, representing an increase of 3 K compared to the case with a 0˚ inlet flow direction angle.