Experimental study of thermal limits with one-side heated smooth channel for plasma facing component safety

Abstract

In order to stably operate the equipment inside the tokamak, which is loaded with a heat flux of several MW/m2 under the one-side heating condition, it is necessary to thoroughly prepare for various thermal engineering limits that may occur under the high heat flux load condition. In this study, we have experimentally explored critical heat flux (CHF) and onset of flow instability (OFI), which are considered potential threats in a DEMO fusion power plant. Specifically, the effect of system parameters on CHF was investigated. The results indicate that with an increase in subcooling and mass flux, the CHF increased, as it induced a faster bubble condensation near the CHF. As the system pressure increased, the CHF also increased. This is because the bubble size reduction effect was dominant in the pressure range of 1–10 bar. Most of the existing CHF correlations could evaluate the CHF with reasonable accuracy of within 25%; especially, the Boscary CHF correlation yielded the highest accuracy with an average error of 12%. Similar to CHF, OFI, which is a measure of the sudden fluctuations in the system pressure caused by a large amount of vapor generated due to the high heat flux, tended to increase as the subcooling, mass flow rate, and system pressure increased. Most of the existing OFI correlations yielded large error rates (more than 135%) as these correlations were primarily developed for micro-channels. Therefore, in this study, a new OFI correlation was developed using a Python code, in combination with an artificial intelligence (AI) regression method. The developed correlation can be used in the cooling system design of tokamaks, which involve a high-heat load condition on one-side of the reactor.

Study on onset of nucleate boiling with screw tube for fusion reactor application

“…To address this issue, the RANdom SAMple consensus (RANSAC) algorithm was used in our regression analysis. RANSAC has the characteristic of completely ignoring data above a certain threshold, allowing for the mitigation of the influence of outliers and the derivation of an optimal model supported by the bulk of the data [53,54]. The RANSAC algorithm operates under the following assumptions: (1) that the dataset contains outliers; (2) that the dataset can be expressed using mathematical model factors; (3) that inlier data can be incorrectly determined to be outliers as a result of extreme values, incorrect measurement of data, or incorrect assumptions about data interpretation; and (4) that there is an algorithm that can predict the model parameters or fit the data in a given set of inliers.…”

Predicting the onset of nucleate boiling heat flux of one-side heated flat heat sink in sub-cooled flow conditions for fusion divertor cooling

“…To overcome this limitation, the random sample consensus (RANSAC) algorithm has been added to the linear regression. Because this algorithm adopts a model supported by a large amount of data as the final product, it is possible to minimize the influence of outliers [54,55]. Finally, the Python code is designed to go through an iteration process so that the code can learn through machine learning through the linear regression analysis process evaluated in the previous step.…”