Thermal dynamics aspect identification of loop heat pipe with capillary tube wick using nonlinear autoregressive exogenous neural network

• Published in: Nuclear engineering and technology Vol. 56; no. 12; pp. 5145 - 5153
• Main Authors: Setyo Pambudi, Yoyok Dwi, Giarno, Hatmoko, Sumantri, Antariksawan, Anhar Riza, Kusuma, Mukhsinun Hadi
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01-12-2024; Elsevier; 한국원자력학회
• Subjects: Loop heat pipe; Neural network; Nonlinear autoregressive exogenous; Passive cooling system; Thermal dynamics; 원자력공학; Thermal dynamics; Neural network; Passive cooling system; Nonlinear autoregressive exogenous; Loop heat pipe
• DOI: 10.1016/j.net.2024.07.022
• EISSN: 2234-358X
• ISSN: 1738-5733

The loop heat pipe (LHP) has the potential to be used as a passive cooling system in small modular reactors. The research objective is to study the thermal dynamics of LHP with capillary tube wick using a non-linear autoregressive exogenous (NARX) based on a neural network. The neural network identification of LHP with capillary tube wick was carried out on the MATLAB platform. The experiment data obtained is used to identify the neural network of LHP with capillary tube wick. The temperature of the water as an evaporator heat source was varied at 60, 70, 80, and 90 °C. The LHP was charged with demineralized water with a filling ratio of 100 %. The air as a coolant in condenser section was blown at velocity of 2.5 m/s. The LHP was vacuumed with an initial pressure of 2690 Pa. The result confirmed that NARX based on the neural network model can predict the temperature of the condenser section with a given input set under the steady-state and transient conditions. The coefficient of determination is higher than 0.998 and Mean Square Error (MSE) is below 0.0082. The result obtained shows that the NARX neural network model can predict thermal dynamics phenomena in LHP quickly and precisely.