https://www.mdu.se/

A unifying and accurate model to predict Critical Heat Flux (CHF) over a wide range of conditions has been elusive since wall boiling research emerged. With the release of the data utilized in the development of the 2006 Groeneveld CHF lookup table (LUT), by far the most extensive public CHF database available to date (nearly 25000 data points), development of data-driven predictions models over a large parameter space in simple geometry (vertical, uniformly heated round tubes) can be performed. Furthermore, the popularization of machine learning techniques to solve regression problems has led to more advanced tools for analyzing large and complex databases. This work compares three machine learning algorithms to predict the entire LUT CHF test database. For each selected regression algorithm (ν-Support vector, Gaussian process, and neural network), an optimized hyperparameter set is fitted. The best-performing algorithm is the neural network, which can achieve a standard deviation of the predicted/measured factor of 12.3%, three times lower than the LUT. In comparison, the Gaussian process regression and the ν-Support vector regression achieve a standard deviation of 17.7%, about two times lower than the LUT. All considered algorithms hence significantly outperform the LUT prediction performance. The neural network model and training methodology are designed to prevent overfitting, which is confirmed by data analysis of the predictions. Finally, a feasibility study of transfer learning is presented and future development directions (including uncertainty quantification) are discussed.