--To make things more clear, if a better moderator is available --
graphite in RBMK and heavy water in CANDU -- ordinary water serves as a
poison.
************************************************************************
Its not always that simple.
To quote from a CNS conference paper last year,
<begin quote>
Coolant-void reactivity is determined by the net result due to the loss of absorption (a positive
reactivity change) and the loss of moderation (a negative reactivity change). Spatial and spectral
changes of the neutron flux in the lattice cell due to voiding of the coolant, as well as the nuclide
composition in the fuel, also affect the coolant-void reactivity to a lesser extent.
......
The neutron spectrum "hardens" when the coolant is lost and the dips at low epithermal neutron
energies become more pronounced. [there is] a general increase in the fast and epithermal
neutron flux in the moderator region as a result of increased escape of fast neutrons from the
voided channel, and an associated decrease in the thermal neutron flux due to the lack of moderation.
[.....there is a] large resonance absorption cross section of 238 U between 1 eV and 1 keV, [and a] prominent broad resonance of 239 Pu fission cross section centred at 0.3 eV. The behaviours of these cross sections have a major influence on the coolant-void reactivity in the ..... lattice.
.....
Voiding of the H2O coolant removes its moderation effect, which renders the lattice
even more under-moderated. The shift in neutron spectrum upon coolant voiding increases the
absorption in 238 U in the resonance region. There is also a significant reduction of the fission-yield
in 239 Pu due to the decrease in neutron flux near 0.3 eV. These negative reactivity effects
reduce the overall coolant-void reactivity in the lattice cell that would be expected from the loss
of substantial absorption in the H2O coolant.
<end quote>
Hope this helps.
Jaro