In principle, criticality accidents such as that of Chernobyl should be impos­sible for an ADSR. However, this is true only as long as one can monitor the effective value of the reactivity. As shown in chapter 7, this monitoring cannot be done solely by relating the beam energy to the reactor output energy: an increase of the reactivity of the subcritical part can be accom­panied by local poisoning of the spallation source in such a way that the output energy does not increase but may, on the contrary, decrease until a critical situation appears. It is thus necessary to devise elaborate ways to monitor the effective reactivity of the subcritical array.

One aspect, which is seldom stressed, of ADSRs is that it requires more technical skill and good maintenance to keep them running than for critical reactors. Indeed, high-intensity accelerators are and will remain rather diffi­cult to operate. Loss of expertise of the staff as well as poor maintenance will, inexorably, decrease the performance of the machine until it finally stops. In contrast, as the recent past shows, critical reactors are apt to run in rather bad shape and do not necessarily need the best team to be operated, with the dangers associated with such a situation. It can be argued, therefore, that ADSRs can offer safety against a societal disorder.

Aside from criticality accidents, ADSRs are subject to risks similar to those of critical reactors, such as solid fuel core melt-down, radioactive leaks into the environment, etc. In addition, the coupling between the accelerator and the subcritical medium may be the origin of weaknesses with respect to safety, such as window breaking or propagation of radio­activity through the accelerator.

For large subcriticality levels of more than a few per cent, the delayed neutron fraction has no influence on the safety of the reactor. This means that it becomes possible to use fuels with large minor actinide concentrations or plutonium without compensating for the small delayed fraction by the presence of 238U. Similarly, the sign of the temperature and void coefficients have a reduced influence. However, they should be limited so that sub­criticality should be guaranteed at all power levels of the reactor. In particu­lar, overly negative coefficients should be avoided to prevent criticality in the case when an accelerator trip leads to a sharp fall of the reactor power.

The high tolerance level of ADSRs with respect to the fuel’s neutronic properties should make them excellent tools to study new reactor concepts by relaxing many safety conditions. For example, the same accelerator could feed different subcritical systems like molten salt, gas or lead cooled reactors. Such prototypes could allow studies of corrosion, radiation defects and fuel evolution in realistic conditions with less stringent criticality- control-related constraints.