Page 138 - Dynamics and Control of Nuclear Reactors

134 CHAPTER 11 Nuclear reactor safety

The Doppler coefficient in the fuel is always negative, so fuel temperature is a
stabilizing effect. The power coefficient is the net effect of the negative Doppler
effect and the positive void reactivity effect. Since at low power levels, fuel temper-
ature feedback is smaller than at high power levels, the positive void effect domi-
nates. The RBMK has a very undesirable large positive power coefficient at
low power.
Another important safety-related feature of the RBMK reactor is the design of its
control rods. Insertion drives them into water-filled channels. The positive reactivity
associated with removal of water would be overwhelmed if the control rods con-
tained a strong neutron absorber along its full length. But the bottom section of
the control rods contains graphite, a much weaker neutron absorber than water. Con-
sequently, the bottom of the control rod introduces positive reactivity upon insertion.
The part of the control above the graphite section contains a strong neutron absorber
and it introduces negative reactivity when that portion enters the core.
The accident occurred because of problems initiated by an experiment designed
to evaluate the potential of a way to improve emergency cooling of the reactor. In the
event of a reactor scram and simultaneously losing electrical power, diesel genera-
tors start to provide electricity to power cooling pumps, but the rise to full generator
power is slow. So, the possibility of getting temporary electrical power from the tur-
bine as it coasted down was to be evaluated.
The experiment was to be performed with the reactor at a power level of 700 to
1000 MWth. This would have avoided the high positive power coefficient at lower
power levels. However, stabilizing the power level at the desired value did not occur
as planned. Furthermore, delays caused an operator shift change to operators who
were not as well informed about the test procedure.
Power reduction from full power began and had reached about 50% power when
the dispatcher prohibited further power reduction because of grid power require-
ments. After a delay, permission was granted to continue with power reduction.
In the attempt to reduce the power level to that needed for the experiment, power
was inadvertently reduced to a very low level (around 30 MWth). At this point
the operator started removing control rods to increase power. Because of Xe-135
buildup at low power due to I-135 decay, a number of control rods were withdrawn
to compensate for the reactivity loss due to Xe-135. Power eventually stabilized at
around 200 MWth. It was decided to proceed with the test even though the power
level was far below that prescribed for the test. It was a condition in which the reactor
had a very strong void coefficient. It was a fatal decision.
The steam flow to the turbine was stopped to start the test. Pumping power
decreased as their electrical power from the slowing turbine—generator decreased.
Coolant flow to the reactor decreased, boiling of coolant increased, reactivity
increased, and fission power increased rapidly. The increased power caused further
boiling and further increased reactivity. This void reactivity (along with reactivity
increases caused by burnout of Xe-135) caused strong reactivity feedback and con-
tinuing uncontrolled power increase. Because key personnel were killed and records
were lost, there is confusion about events that occurred at this point. It is possible that
an attempt to stop the power rise by inserting control rods made the problem worse