Part 5: The Test Begins

Unit 3 MCPs (Photo Credit: Martin Malik)

To review, around 1:00 reactor power had reached 200 MW. The turbine vibration test was proceeding smoothly, to everyone's relief. SIUB Boris Stolyarchuk was still working hard to get the water level in the left-side drum separators above -600mm, but no one was losing any sleep over that. Operating reactivity margin (ORM) was within normal limits as far as the personnel could tell, although the reported value (between 17 and 19 rods) may have already been outdated, from the calculation produced by the Skala at 0:39.

 

Now it was time to continue with the rundown test program. The first order of business was to enable the 7th and 8th main circulating pumps (MCPs), two more than were usually in operation. This was done at 1:03 and 1:07. With two additional MCPs engaged, the amount of coolant flowing through the reactor increased. This dampened steam production (and hence reactivity) in the active zone, which required the SIUR to raise control rods to compensate. ORM dropped from its previous value. In addition, the reduction in steam caused water levels to fall again in the drum separators.

Myths/Misinformation: In what readers may recognize as a recurring theme, the Soviets claimed that operating all eight MCPs simultaneously was a violation of the regulations. It was not, although the rules were changed after the accident to forbid this. In general the Soviets wrote new rules after the accident and then tried to accuse the operators of breaking the regulations that did not exist yet.

Boris Stolyarchuk in 2004

Now that the reactor was no longer in danger of smothering itself in xenon, Stolyarchuk no doubt felt free to finally tackle the drum separators' water level problem. He let loose with a sudden influx of cold feedwater, sending flow rates rising off the charts as shown below. Combined with the influence of the additional MCPs, this large addition of colder water further throttled reactivity. As boiling reached a minimum, the signal "1 overcompensation upwards" was recorded, meaning that the automatic control rods were rising to prevent a reduction in reactor power. Toptunov had no choice but to aid them by withdrawing manual control rods as well, including many of the remaining bottom-insert USP rods.

Feedwater refers to excess water added to the drum separators to make up for coolant that turns to steam. At 1:06 and 1:19 Stolyarchuk can be seen struggling to top up the drum separators. Graph copied from accidont.ru (Author Viktor Dmitriev)

Finally, at this point the operating reactivity margin (ORM) was clearly below the limit of 15 rods. According to calculations performed by Nikolai Karpan, the additional coolant flow reduced ORM by 2.5 rods, while the influx of cold feedwater reduced ORM by 5.5 rods. Another 1-2 rods of ORM was likely lost due to xenon poisoning during the lead up to the test. The end result was that when the Skala's DREG program scanned the control rod positions at 1:22:30, the ORM value had fallen to 7.5 rods. 

 

Myths/Misinformation: It is often claimed that the operators received a computer printout alerting them to the low ORM value and suggested that they shut down the reactor. In fact the Skala computer never calculated ORM based on the data received at 1:22:30, because the explosion happened first. The Skala operators had been scheduled to deliver their next ORM report only at 2:00. The accident investigators needed to recover the raw data and perform the ORM calculations on the Skala computer at the Kurskaya NPP. A less accurate value of only 2 rods ORM was produced along the way, which the Soviets trotted out to make the operators look worse. Of course, even if the operators HAD learned of the low ORM value and decided to abort the test, the reactor likely still would have exploded due to the tip effect of the control rods.

 

As visualized in the diagram below, between 0:39 and 1:22:30 the operators had gradually removed control rods in response to various developments. The total linear length of inserted absorber rods had decreased from 102 meters to 48 meters (bearing in mind that ORM does not have a direct, simple relationship to the number of rods inserted).

Control rod insertion depths (centimeters from top of core) as they changed between 0:39 and 1:22:30. Some small rod movements omitted. Graph copied from Karpan, Revenge of the Peaceful Atom.

 

As is immediately evident, numerous rods had been removed from the core. However, by no means was there a complete lack of absorbers, and some (primarily automatic control rods) had been inserted farther. The bottom-insert USP rods had preferentially been removed, leading to a paucity of neutron absorption and control in the very section of core that was most exposed to the tip effect and positive void coefficient. The automatic control rods did not have graphite displacers, so the tip effect applied only to the manual control rods which were completely withdrawn from the core. Between 0:39 and 1:22:30 the number of rods which could produce a reactivity spike upon insertion grew from 151 to 164. Calculations by Nikolai Karpan indicate that AZ-5 would have safely shut down the reactor at 0:39, but would have destroyed it at 1:22:30.

 

Human Factors: Did the operators realize that ORM was too low? It is equally plausible that they were too busy to make an extra calculation request, or that they had an intuition of the low value, but took no action. Speculation aside, we do have a pretty good idea of what their attitude towards ORM would have been. Namely, that ORM was a non-safety-related parameter of secondary importance. An ORM of less than 15 meant that the reactor could not be restarted for a long time following a shutdown. Since extended maintenance was planned anyways, this factor was irrelevant to them. They were also aware that a low ORM would limit their ability to change the shape of the neutron flux fields (i.e., a rod that is withdrawn cannot be withdrawn farther, so low-reactivity areas of core will be 'stuck' at that level). Since they only intended to work for another twenty minutes, this too was no reason for concern. Moreover, for the first several years of its operation, the RBMK had lacked an ORM limit entirely. That meant that reactor was simply designed to operate with very few rods inserted, and this was even favored as the most fuel-efficient regime. Current regulations still required that the reactor be shut down at 15 rods, but the operators likely viewed this requirement as not applying to transitional regimes at low power, since tracking ORM in real time was impossible. They were accustomed to waiting for the periodic Skala reports before taking action.

 

If Unit 4 had never exploded, but only resulted in a minor inconvenience, a disciplinary investigation of the shift likely would have only bothered mentioning a single violation: the excessive flow rates on two or three of the main circulating pumps (MCPs). With eight pumps engaged and feedwater pouring into the drum separators, Stolyarchuk failed to balance the workload of the MCPs and several began operating above their capacity limits (7000 cubic meters per hour). 

 

Technical Explanations: Unlike the night's other transgressions, this parameter had very direct and obvious safety ramifications: with excessive flow rates the pumps could start cavitating and fail completely. In addition, such a high flow rate meant that coolant exiting the reactor had little time to drop in temperature before arriving back at the core inlet. This was a factor in the accident because the coolant was close to the boiling point, and able to rapidly transition from liquid to steam in the problematic bottom section of the reactor. However, the MCP flow rate violation had no real influence on the accident, since the total flow rate of coolant was not limited by any regulations. In other words, had Stolyarchuk done his work perfectly and exceeded no limits, overall coolant flow rate could still have been the same or even higher, and the accident would have occurred in the same way.

The ORM value (7-8 rods) measured at 1:22:30 was likely the minimum value that night. Shortly after, Stolyarchuk abruptly reduced the feedwater flow rate to its initial value (although MCP flow rates remained high). The result was a slight increase to overall coolant temperature, hence steam production and reactivity. One of the two reactor power instruments (the PPDDCS but not the ionization chambers) recorded a slight increase in reactor power at this point, and according to the data presented by the Soviets at Vienna, some automatic control rods descended partway. In other words, between 1:22:30 and 1:23:40, ORM slowly increased. 

 

Various instances of confusion, errors and rule-breaking have been detailed so far, but the inadvertent deficit of ORM was the only violation that had a direct relationship to the accident. Combined with other factors such as the low steam production and shape of the neutron flux fields, the low ORM meant that the reactor was poised to surge out of control if there was an increase in steam production for any reason. But this was invisible to the operators, who supposed that the fast power coefficient of reactivity would be negative, that the void effect of reactivity was negative, that the void coefficient of reactivity was much smaller, and that the AZ-5 button would safely shut down the reactor in any eventuality.

 

Continue Part 6: Prompt Critical

 

Sources Cited: 

• Nikolai Karpan, Revenge of the Peaceful Atom.
• INSAG-7.