The prediction of the moderator temperature distribution in a CANDU reactor is important in establishing its ability to act as an emergency heat sink for certain beyond design basis accidents. This analysis typically relies on computational models which are benchmarked against experimental data from small-scale test facilities. These small-scale models prioritize the matching of the Archimedes number (Ar) of the full-scale reactor, which represents the ratio of buoyancy forces to inertial forces. Concerns regarding similarity between the reactor and small-scale facilities may exist due to a large difference in scale, as well as geometric simplifications made due to practical limitations.
This study examined the behaviour within an approximately 1/16 scale facility representative of the Bruce A calandria vessel, which features a unique inlet and outlet configuration. Experimental measurements were obtained for a range of power and flow conditions. Unsteady RANS simulations of the small-scale facility were also performed using the realizable k-epsilon model. Goals of the study included the assessment of the unique moderator inlet on the flow patterns inside the calandria vessel and how well existing CFD modelling approaches replicated these features.
The observed flow and temperature distributions in the scale facility did not appear highly sensitive to changes in Ar. For all tested conditions, a large front-to-back recirculation pattern resulted from the asymmetric inlet arrangement. Peak temperatures consistently occurred toward the front of the vessel where inertial flows were assisted by buoyancy induced flows.
Under steady-state conditions, unsteady and three-dimensional behaviour was observed within the vessel. Temperature fluctuations near the upper rear end of the vessel arose from the unstable interaction between cool downward flow from the inlets and upward buoyant flow from the tube bank. In the peak temperature regions, flow direction was relatively consistent in the upward direction.
The simulations tended to overpredict the peak temperatures within the vessel by approximately 0.5 – 3.8 ºC. This behaviour was attributed to the model tending to underpredict the upward velocities entering the base of the tube bank in the peak temperature regions. As Ar increased and buoyancy effects became more significant in determining the local velocities, agreement between the predicted and measured velocities was improved.
The similarity between the small-scale model and the full-size reactor was also assessed through comparisons to existing simulations of the full-size calandria. There was qualitative similarity between the two geometries, albeit at lower Ar for the small-scale facility. This suggested that buoyancy effects were more significant in the small-scale facility compared to the full-size calandria. This was attributed to the use of surface heating (as opposed to volumetric heating in the reactor), and relatively high surface heat fluxes caused by a reduced number of tube bank elements. / Thesis / Doctor of Philosophy (PhD)
| Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/23863 |
| Date | January 2019 |
| Creators | Strack, James Michael Vincent |
| Contributors | Novog, David, Engineering Physics |
| Source Sets | McMaster University |
| Language | English |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0023 seconds