Effect of helium injection on diffusion dominated air ingress accidents in pebble bed reactors

Abstract

The primary objective of this thesis was to validate the sustained counter air diffusion (SCAD) method at preventing natural circulation onset in diffusion dominated air ingress accidents. The analysis presented in this thesis starts with a vertically oriented rupture of a coaxial pipe. Air enters into the reactor cavity at a rate dictated by diffusion, until the buoyancy force is strong enough to initiate natural convective flow through the reactor. The SCAD method, developed by Yan et al. reduces the buoyancy force in a high temperature gas reactor (HTGR), during the lengthy diffusion phase, by injecting minute amounts of helium into the top of the reactor to set up a counter helium-air diffusion circuit. By delaying the onset of natural circulation, air enters the reactor only at diffusion transport rates, instead of much higher natural convection transport rates. Thus, the air ingress rate is reduced by several orders of magnitude. Without the continuous convective driven supply of "fresh" air the threat of oxidizing graphite components is significantly reduced. To validate SCAD a small scale simulated Pebble Bed Reactor (PBR) was constructed and a series of air ingress experiments with and without helium injection were conducted. In addition, Computational Fluid Dynamic (CFD) simulations were performed using FLUENT @ to model the experiment and gain further insight into the behavior of the flow field leading up to the onset of natural circulation. In order to have the CFD predicted natural circulation onset time better match the experimentally determined onset time, the initial helium fraction in the numerical model had to be reduced by 15%. This reduction is within the uncertainty of the experimental set-up. This change helped display an important feature of the behavior of air ingress accidents. With the initial helium fraction in the simulated reactor at 100% the first half of the transient is a very slow completely diffusion dominated transport phase. The second half of the transient had an air transport rate that had an increasing natural convective transport contribution leading up to the onset of natural circulation and complete natural convective transport. Reducing the initial helium fraction by only 15% caused that initial very slow, pure diffusion transport phase to be bypassed and the natural circulation onset time was dictated by the combined effects of free convection and diffusion transport, not simply diffusion. A full scale PBR experiencing a similar accident will have the core entirely filled with helium. Thus, for a vertically oriented double ended guillotine (DEG) large-break loss of coolant accident (LB-LOCA) the subsequent air ingress rate will be dictated by the slow diffusion of air into the reactor cavity, for most of the transient. For the helium injection tests, even at the at the lowest tested injection rate, both the experiment and the CFD simulation showed that natural circulation was prevented over a time period twice as long as the time to onset. The tests showed that without helium injection, natural circulation started after about 117 minutes on average. With helium injection, natural circulation did not start after 240 minutes when the experiment was terminated. Additional injection tests were run where after 240 minutes the helium injection was terminated, but data continued to be taken. In these tests natural circulation was initiated in approximately 120 minutes after termination of helium injection confirming the helium injection flow was preventing natural circulation from starting. The lowest tested helium injection rate corresponded to 0.01% of the test assembly's total volume per minute, demonstrating how small of a flow rate is needed for the SCAD method to work. Minimal helium injection is not intended to be an emergency core cooling system but rather a system to prevent or delay natural circulation which would result in a large amount of air ingress. The system response was formulated non-dimensionally to quantify the impact SCAD has on the driving parameters that impact the onset of natural circulation, namely the buoyancy force, mass flow rate, and density ratio between the hot and cold leg. The results showed that SCAD suppresses the buoyancy force and forces a mass flow (transport) rate that causes any changes in the hot leg density to be counter-acted by density changes in the cold leg. The transport rate that is established is orders of magnitude less than the natural circulation transport rate. Using the driving nondimensional parameters, a methodology was also developed in order to formulate a correlation to estimate the minimum injection rate (MIR) of helium to prevent the onset of natural circulation. In order to properly derive a correlation for the MIR, further experiments and/or simulations are required over different geometrical configurations. The non-dimensional analysis showed that Yan's MIR estimate was conservative for the experimental configuration, and would be conservative for a full scale PBR. Therefore, Yan's MIR calculation was used to provide an order of magnitude estimate for the helium injection rate in a full scale PBR. The resulting MIR of helium for a full scale PBR was 5.36 g/hr, which corresponds to storing only 11.6 kg of helium on-site to prevent the onset of natural circulation for three full months. The experiment and CFD simulations were performed using an inverted U-tube which simulates a vertically oriented pipe configuration. If the pipe break occurs in a horizontal configuration, the air ingress phenomena could be substantially different depending on the break size and orientation. Thus, this thesis concludes that the method is capable of preventing natural circulation onset as long as air ingress occurs at transport rates comparable to diffusion after the break occurs.