The heat transfer model plays a significant role in improving the steel quality during quench cooling by water jet impingement. However, the available models are still incapable of accurately modeling the heat transfer process. In this work, we perform the multiple impingements of jet on high-temperature steel plate surface experimentally and reveal the effects of surface temperatures, jet velocity, moving speed, and equalization time on the heat transfer behaviors. Then, we propose a time-constrained surface heat flux calculation method to measure the transient heat flux more accurately. Even in the case of lower temperature measurement frequency and larger heat conduction delay, precise calculation results can also be predicted. Importantly, a machine learning method is employed to fit and analyze a large number of experimental data, the machine learning model can predict two-dimensional (2D) transient heat flux in the moving cooling process and the accuracy can be continuously reinforced with the increase of experimental data. Based on the well-trained machine learning model, a heat flux prediction software is developed and open access at GitHub. This work provides an accurate heat transfer model for numerical analysis and shows great application potential in the steel industry, and the results will help deepen the understanding of the heat transfer mechanism.
Read full abstract