Two basic features of the CANDU (CANada Deuterium Uranium. Registered trademark.) nuclear power reactor are the use of heavy water as neutron moderator and the use of pressure tubes to contain the reactor fuel and coolant. Contemporary CANDU reactors also use heavy water as primary coolant. As Figure 1 shows, heat is transferred from the fuel to the primary coolant, which transports the heat to steam generators. The Steam Generator secondary side forms part of a conventional steam power cycle.
The low-pressure heavy-water moderator, contained in a vessel called the calandria, provides efficient fuel utilization and permits the use of natural rather than enriched uranium. The moderator is separate from the high-pressure coolant.
A CANDU fuel channel contains 12 or 13 fuel bundles end-to-end. Each fuel bundle comprises an array of fuel elements held together by Zircaloy end-plates. As Figure 2 shows, a fuel element is made from UO2 fuel pellets contained in a thin cylindrical Zircaloy sheath. The coolant, at a pressure of about 10 MPa, is contained in zirconium alloy pressure tubes, each of which is insulated from an outer calandria tube by CO2 gas.
A CANDU reactor employs several hundred horizontal fuel channels, only two being shown in Figure 1: the horizontal orientation facilitates on-power refueling.
A fuel channel in a CANDU reactor is normally supplied with up to 25 kg/s of coolant and produces up to 7 MW of power. The coolant can boil slightly, which increases steam generator temperatures, thereby increasing the plant's thermodynamic efficiency.
Under normal operating conditions, the heat transfer from fuel to coolant is high and the sheath temperature is only a few tens of degrees higher than coolant temperature. However, at abnormally high-channel power or abnormally low-channel flow, the rate of heat transfer to the coolant can deteriorate due to the development of a vapor film at the fuel surface. This condition, called dryout, is avoided with margins in power and flow. The reactor is tripped (power reduced) if measurements show that the dryout condition is being approached. (See Burnout.)
The channel condition at dryout is measured in CANDU-specific full-scale tests using electrically-heated fuel-string simulators (Leung et al., 1985). The tests are done over a range of mass flows and pressures, and for several channel axial power distributions.
Light water, instead of heavy water, is used as coolant in the tests. This requires a conversion of the dryout data, which is done using the fluid-to-fluid modeling techniques of Ahmed (1971). The boiling length/critical quality approach of Bertoletti (1965) is used to correlate the data: dryout is assumed to have occurred if the steam quality in the fuel channel exceeds the dryout steam quality (flow-cross-section averaged). The dryout steam Quality Xc is correlated as a function of the boiling length L (the distance along the channel from the onset of boiling to the location of dryout). This approach suggests the existence of an upstream effect—instead of dryout being a purely local phenomenon—and is necessary in accurately correlating data over a range of axial power shapes.
The correlation has the functional form:
where a and b depend on pressure and mass flux. In the tests, the location of dryout is identified using sliding thermocouples. The location of onset of boiling and Xc are calculated from measurements of power, pressure, coolant inlet temperature and flow.
The above dryout correlation is used in a reactor-system thermohydraulic code. The code is used to simulate a CANDU reactor under a large number of operating conditions, with different three-dimensional neutron flux shapes and power distributions. It generates the channel power at dryout (called the critical channel power) for all channels and all operating conditions. This information, together with detailed information on the three-dimensional neutron flux shapes, is used to establish the trip setpoints for neutron flux as measured by detectors at a number of in-core locations. The detectors are part of a protection system that provides a 99% probability that the reactor will be tripped before dryout in any channel, given an accident leading to overpower.
In applying the thermohydraulic code, an assumption generally used is that the condition in each reactor header stays constant as power is varied in a particular fuel channel. (Figure 1 shows fuel channels connected to headers via feeder pipes.) In particular, the header-to-header pressure drop, ΔP, is conservatively assumed to be constant. However, if the power were to increase, the extra boiling would cause a reduction in flow and an increase in the head from the pumps. The ΔP would increase.
When setting overpower trips in CANDU reactors, dryout is defined and measured as the first small additional rise in fuel-sheath temperature as power is being increased. However, dryout occurs at a level of flow and steam quality such that any further increase of temperature with power is "slow," rather than "fast," as described by Groeneveld and Borodin (1979). Some wetting of the surface still occurs while the temperature is less than 382°C, the minimum film-boiling temperature (Groeneveld and Stewart (1982)). From the full-scale tests, sheath temperatures are typically less than 500°C, even for powers 10% beyond dryout. (See also Postdryout Heat Transfer.)
Ahmed, S. Y. (1971) Fluid-to-fluid Modeling of Critical Heat Flux: A Compensated Distortion Model, AECL Report, AECL-3663.
Bertoletti, S. et al. (1965) Heat Transfer Crisis with Steam-Water Mixtures, Energia Nucleare, 12, 3.
Groeneveld, D. C. and Borodin, A. S. (1979) The Occurrence of Slow Dryout in Forced Convective Flow, Second Multi-Phase Flow and Heat Transfer Symposium Workshop, Miami.
Groeneveld, D. C. and Stewart, J. C. (1982) The Minimum Film Boiling Temperature for Water During Film Boiling Collapse, Proceedings of the Seventh Int. Heat Transfer Conference, Munich.
Leung, A., Merlo, E. E., Gacesa, M., and Groeneveld, D. C. (1985) Critical Channel Power Evaluation Methodologies at Atomic Energy of Canada Limited, Proceedings of the 6th Annual CNS Conference, Ottawa, June.