The reactor is supposed to be lowered in power to 700 - 1000 MWt. The power is being lowered slowly to decrease the build-up of Xenon in the reactor. As a reactor decreases in power a large amount of Xenon will start to build-up, especially if the reactor has a rapid decrease in power. Xenon is known as a "poison", a neutron absorber that removes neutrons from the fission chain reaction in a reactor, thereby decreasing the reactor's reactivity. The reactor power reaches a level of 1600 MWt at 03:47:00 where it is halted.
The range of 700 - 1000 MWt is chosen by test designers as the minimum allowable power of the reactor. At lower power levels there are thermal and neutronic instabilities. Although, Soviet engineers have stated that the test could be run at zero power, which would simulate a scram of the reactor control rods.
Switching off the Emergency Core Cooling System (ECCS) is a violation of regulations. The test required that the ECCS be switched off to ensure that the ECCS would not activate itself during the testing procedures. However, the disconnecting of the ECCS does not contribute to the accident. It is important to note that if the ECCS would still have been connected it may have reduced the impact of the accident.
At this time the reactor is held at a constant power. The power is scheduled to keep decreasing. The power hold at 1600 MWt is done at the request of the Kiev dispatcher. The power is held in check for almost nine hours. The Kiev dispatcher needs more power than is expected at the time of test.
Two things happened because of the delay of the test. First, the hold in reactor power leads to a further reduction of Xenon buildup. This is a side benefit from having the testing delayed. The second side effect is the timing of the actual experiment. The test is now going to be held during the 'night' shift instead of the 'day' shift that had been planned.
The Kiev dispatcher permits the continuation of the descent in power. The power descent should continuously decrease until it reaches 700 MWt - 1000 MWt. The time span of the power at 1600 MWt is 19 hours and 23 minutes.
At 24:00 there is the normal switch of shifts. After about 25 minutes of being at the reactor controls, the power level is reduced below 700 MWt. This time is linked to the actual start of the accident. The reduction below 700 MWt is in direct violation with the normal and safety test procedures. This is caused by the operator switching from Local Automatic Control to Global Power, or full control. The operator failed to enter hold power at required level, which will stabilize the reactor at a level power. Subsequently, power is reduced to approximately 30 MWt.
The sharp reduction in power causes an intense increase in Xenon buildup. The build-up of Xenon causes negative reactivity. The reactor can become sub-critical and the test could be lost. The operators have a decision to make, whether or not to continue with the test or scrap the test and try again at a later date. After approximately 4 minutes, the operators prepare to resume the test. As a result of those extra 4 minutes, the Xenon build-up has increased even more. In order for the operators to resume the test it would be necessary to increase the reactor power to a level above 700 MWt, the normal and safe operating level. As the Xenon keeps increasing it becomes more difficult to increase power without removing more control rods than is allowed.
At approximately 00:40, the reactor operators start to withdraw control rods in an attempt to increase the power of the reactor. To do this it is necessary to withdraw more control rods than is allowed by operational safety. At this point the Operating Reactivity Margin (ORM), keeps decreasing. It is necessary to keep this margin above 30 rods, as is seen in the next time step, the Chernobyl reactor is well below its level of operation.
The reactor power is stabilized at 200 MWt. There is difficulty acquiring and holding this power level. The difficulty in increasing the power is due to the extremely low Operating Reactivity Margin (ORM). ORM is a calculated reactivity margin based on power and control rod distribution and positions, given in terms of number of control rods, where 30 is the minimum allowable. In certain circumstances ORM may be lowered to 15, but as is noted the ORM of Chernobyl is well below this level of acceptance. The reactor currently had an ORM of 7.2 control rods. It is important to note that to operate below 30 rods it is necessary to get the Station Manager's approval. This approval is not granted, thereby causing another violation of procedures.
At this point the Xenon build-up has increased to levels above what is anticipated for the testing procedures. Also, the power level of 200 MWt is below 700 MWt, which is a direct violation of the test safety procedures. The power is only increased to 200 MWt because there is no excess reactivity available to increase the power to necessary levels. This power level is continuously fluctuating.
At 1:03, it is possible to better stabilize the reactor at a power level of 200 MWt. This is done by connecting the fourth main cooling pump to the left loop of the system. This is part of the normal testing procedure; however, the pump is not supposed to be connected at such a low power level. The connecting of the cooling pump increased the flow of the coolant throughout the system. At this point the coolant inlet-to-core temperature is approaching closer to saturation temperature. There is little to no boiling due to low power and increased flow rate of the coolant. This leads to an increase of negative reactivity, which will lower the power of the core. Lowering the power causes the eventual need to withdraw even more control rods to compensate for the lack of power.
The fourth Main Cooling Pump is connected to the right side of the cooling loop. This is as is prescribed in the original testing procedures. The total flow of the system increases to 57,000 m3/hour in excess of the maximum flow permitted at such a low power. The operation of the additional pumps leads to overcooling of the core. The low water temperature leads to a reduction of steam pressure to the turbine.
At this point the core contains a very low void coefficient (a rate of change in the reactivity of a water reactor system resulting from the formation of steam bubbles as the power level and temperature level increase), low ORM, slow power measurement response, and pumps near cavitation (Vapor bubbles appearing in liquid caused by the liquid pressure decreasing below the vapor pressure. Cavitation can rapidly erode metallic surfaces and eventually destroy them.). The operational mode is now very unstable.
In order to keep the reactor operating at such a low steam pressure and low water level it becomes necessary for the operators to shut down the Emergency Protection Signals. This will allow the test to continue without the reactor accidentally scramming. This is an extremely serious violation of normal operating procedures.
In recognition of a very low level of water in the steam separators, the operators increase the amount of feedwater (water being injected into the core) flow. The new cold water reaching the reactor core causes a sharp drop in the steam fraction of the coolant and a corresponding power decrease. The overall feedwater flow is increased to three times the amount of balanced flow for such a low power level. This new core subcooling increases the poisoning of the core and a greater void collapse. Particularly at low power levels, a reduction of the void produces a negative reactivity insertion which acts to lower the power of the core. For the operators to continue with the experiment it is necessary to remove even more control rods to keep the reactor at it's stationary power.
At this point more of the manual control rods are removed from the core. There are probably only 8 manual control rods currently in the core. A minimum of 15 manual control rods must be inserted in the reactor at all times. To account for the 7 missing manual control rods, there are a number of automatic control rods that are in place to cause a total of at least 15 control rods in the reactor. Even though there is a total of 15 control rods in the core, it is still a violation of procedures to have less than 15 manual control rods.
The operators initiate a feedwater flow reduction for a target value of two-thirds of the balanced flow. The operators realize that the feedwater flow is too high and the water level has stabilized in the steam drums at 1:22:10. The operator is unable to stop the feedwater flow rate at the desired level because of the coarseness of the system at such an operating regime. At this point, control rods are inserted to compensate for the added reactivity associated with the increased voiding.
The operators make a print out of the distribution of power density and the position of every control rod in the reactor. This is done in accordance with the testing procedures to establish the flux distributions and reactivity margin before the beginning the test. The lengthy computer calculation time actually provides a print out of the reactor core a few minutes later. There is conformation that the ORM is half of the minimum allowed. At this point the operator should initiate immediate shutdown of the reactor. This warning is ignored and the test is initiated.
After the feedwater flow is stabilized, the operators begin the test by closing the TG #8 turbine (emergency) stop-valve. This valve is closed to allow for the repeating of the experiment as necessary. This is the last safety system to be disconnected to avoid an inadvertent scram of the reactor. The only emergency scram systems left are high power trip and low period trip, which are linearly related (a low period is equivalent to high reactor power). The test is now initiated.
The closing of this inlet valve to the turbine will cause the turbine to slow down because there is no steam being supplied to the turbine. The flow rate begins to fall in the four main cooling pumps. The steam pressure increases due to the removal of the relief of steam associated with the turbine and the reduction of the feedwater flow. These factors combine to increase the void coefficient of the coolant and at the same time the increase the power of the reactor.
Ten seconds after the initial start of the test, at 1:23:04, there is a group of automatic control rods that start driving out of the core. This is caused by a decrease in the void associated with the increasing system pressure. This also leads to there being less than 15 total control rods in the core. Eleven seconds later, two groups of automated control rods start to drive back into the reactor core. This is the result of the continuously reducing coolant flow and the approach of warmer water to the core which increases core power. Both the losses of coolant and the power increase lead to added reactivity to the core.
The net reactivity of the core is now continuously increasing with the accelerating power increase. At 1:23:31, the insertion of control rods is unable to balance the increased reactivity. At this point the power is increasing, yet not enough to melt the fuel.
The operator notices that there is movement in the automatic control rods, the power is increasing, and that the test is effectively over. At 1:23:40, the Unit-4 shift foreman gives the order to scram the reactor. Unfortunately, at this point there is insufficient negative reactivity in the rods that are in the core. The rods that are on top of the core can not be lowered fast enough to counteract the power increase that is now occurring in the reactor. At this point the reactor is at approximately 320 MWt and increasing rapidly.
The insertion of all of the rods at the top of the core centralized all of the reactivity in the lower region of the core. The estimated reactivity addition to the core is approximately $1. The reactor is now prompt critical. This leads to a run away nuclear reaction.
At 1:23:43, the power of the reactor core is approximately 530 MWt. The high power alarm is triggered and the short period alarm is also triggered. Note, that these are the only two alarms that are basically in operation. The emergency protection system is not sufficient enough to prevent the runaway power. In this same second, calculations by soviets indicate that the power is approximately 3800 MWt, approximately 120% of full power.
At 1:23:44, the reactor core attains it's first peak in power. It is calculated that this power reached 325 GWt, approximately 109 times nominal full power. This large and rapid power increase leads to fuel fragmentation, which causes large steam spikes. This reverses the flow of the coolant and closes the main reactor pump check valves. As a result, the void produces another neutron pulse and a second power surge.
Operators hear banging noises and see were rods stopped before they reached the bottom of the reactor. This is probably caused by the distortion to the reactor after the first power surge. The first power explosion could be the rise and fall of the 2500 metric ton Upper Biological Shield. The first power surge is probably stopped by the Doppler effect and partial rod insertion. The Doppler effect is a shift with temperature of the interaction rate between neutrons and reactor materials, such as fuel rods, structural materials, and fertile materials. The shift can appreciably affect the neutron density and hence the reactivity of reactors.
At 1:23:45, the reactor attains the second spike in power caused by the reversal of the feedwater flow into the core. The steam drum pressures starts to exceed "accident level" and pressure relief valves stay open. They reach accident levels at 1:23:46.
At 1:23:48, the first "thermal explosion" blows the top off the reactor. This destroys the reactor hall. The second explosion is caused by the expansion of the fuel vapor. From this point on there is no more information from the reactor control room. The two explosions cause hot fragments and spark emission. These cause about thirty fires on roofs, etc. A reactor fire results with the reaction of air to graphite in the reactor.
Alexander R. Sich, Ph.D. Dissertation, Department of Nuclear Engineering, M.I.T., 1994.
McPherson, Donald G., A Chronology of the Chernobyl-4 Accident, ANS Topical Meeting on Radiological Accidents, date????, pg. 29 to 37.
Kress, T. S., Jankowski, M. W., Joosten, J. K., & Powers, D. A., The Chernobyl Accident. Nuclear Safety, Special Section: Chernobyl, Vol. 28, No. 1, January-March 1987, pg.7 to 9.
http://www.uilondon.org/cherntime.html, The accident at Chernobyl Unit 4. Uranium Institute 1996.
Nuclear Terms A Brief Glossary, 2nd edition. U.S. Atomic Energy Commission, Division of Technical information, January 1969.