Numerical investigation of rod bundle thermal–hydraulic behavior during reflood transients using COBRA-TF

Abstract

Accurate prediction of the reflood transients during LOCA has long been a challenging task. One reason for this is the substantial cost incurred to perform large-scale reflood tests, which are formidable in experimental expenditure and intensive in technology as well as program management. Another reason preventing a comprehensive understanding of reflood transients is the high-level complexity involved in the two-phase flow mass and heat transfer processes, which makes the measurement extremely difficult and inefficient. The situation has been significantly improved with the design and operation of the Nuclear Regulatory Commission (NRC)/Pennsylvania State University (PSU) Rod Bundle Heat Transfer (RBHT) test facility. A variety of advanced instrumentations were developed and used at this facility and very high-resolution data has been obtained, especially for the liquid droplet field. In the current study taking advantage of the unique NRC/PSU RBHT data, an extensive and comprehensive code evaluation and validation is carried out using the thermal–hydraulic sub-channel analysis code COBRA-TF. The system parametric effects investigated include: the system pressure, inlet liquid subcooling temperature, inlet flooding rate and rod bundle power. A variety of thermal–hydraulic quantities predicted by the numerical code are evaluated including: the quench front propagation, cladding, spacer grid and vapor temperature variations, two-phase pressure drop, liquid droplet velocity and coolant void fraction. In addition, the prediction errors are presented for each of the quantities in great detail in order to have a comprehensive understanding of the code performance. In general, COBRA-TF agrees relatively well with the experimental data in terms of the overall quench front propagation, only predicting slightly earlier quench. However, under low inlet subcooling and high flooding rate conditions, significant discrepancies are observed. The comparison with rod bundle thermal–hydraulic parameters indicates that COBRA-TF is able to predict the cladding temperature before quench well within a 15% range. The spacer grid and vapor temperature predictions are within 20% error. While the overall droplet velocity prediction is found to be within 30% error, COBRA-TF is able to capture the decreasing trend for droplet velocity during reflood. In addition, the prediction of the dispersed flow film boiling void fraction shows that the code always under-predicts the void fraction. While the most prediction discrepancy is found to be the two-phase flow pressure drop after quench, which involves more than 50% under-prediction for the bulk liquid boiling regime. The present study clarifies the effect of different system parameters on the various two-phase flow quantities during reflood. The results obtained provide answers for many existing code modeling issues and thus can be instructive and useful for future model development and code upgrading.