Abstract

A monitored nuclear reactor parameter, such as a startup range neutron monitoring signal, is filtered at a filter having a preselected time constant and amplified with the output being passed to one channel of a two channel comparator. The level of the amplification and the filter time constant are chosen to discriminate between normally increasing startup neutron density including the prompt-jump effect and exponentially increasing neutron density due to the occurrence of reactor casualty in view of the design parameters of the reactor. This same startup range neutron monitoring signal is passed unprocessed to the other channel of the two channel comparator. Comparison of the two signals is made with the binary output of the comparator flagging excess of the unprocessed signal over the filtered and amplified signal. Presence of the appropriate reactor flag is typically used to cause rod withdrawal blockage and/or SCRAM. The time constant of the filter causes the monitored neutron density signal to be delayed in its tracking of reactor startup. The level of amplification of the monitored neutron density signal establishes the reference level in the comparator which when exceeded sets the comparator flag. Consequently, comparator response time varies inversely with the severity of increase in neutron density. Preferably a first amplification level is used for determining rod withdrawal block and a second and higher level of amplification is used for plant SCRAM comparators. A dynamic display to the operator of comparison between monitored neutron density and the dynamically tailored trip levels is provided.