Part 6: Prompt Critical

Image Credit: chidori@emptytriangle.com

 

With the reactor outwardly stable, the operators were ready to begin the rundown test. The procedure was that the steam shut-off valves would be closed, isolating the turbine from the reactor. Four of the eight MCPs would continue operating as normal, powered from the grid. The other four MCPs would receive their power from the still-spinning turbine, and the voltage measurements would be used to evaluate the test results. Simultaneously with closure of the steam valves, electrical division personnel were supposed to engage a temporary system (triggered by the 'MPA' button) meant to simulate the signals of a maximum design-basis accident. Meanwhile, the oscillograph would record the necessary data.

 

Razim Davletbaev recalled the beginning of the test:

 

Unit 4 shift supervisor Alexander Akimov (who died of acute radiation sickness in May 1986) approached each operator, also briefly instructing senior turbine control engineer Igor Kirshenbaum that when commencing the experiment he needed to cut off steam to turbine No. 8. Then Akimov asked the operators if they were ready, and "Dontekhenergo" test representative Metlenko commanded: "Attention, engage oscillograph."

Yuri Tregub relayed much the same thing:

As the experiment managers, Dyatlov and Akimov were placed in the center, to monitor the panel, and periodically went to and fro. Then Metlenko sat not far from Akimov's work station, picked up a phone. They arranged that Metlenko would engage the oscillograph, to record the test results, Kirshenbaum would shut the steam stop-valves. The rundown experiment began. They shut off the steam to the turbine and meanwhile watched how long the rundown will last. And then the command was given, by Akimov. Kirshenbaum—I stood next to him—closed the stop-valve, Metlenko gave some command into his phone...

 

The steam valves were closed, which revealed one final mistake in the test procedure. According to the prevailing studies and accident modelling (as cited in INSAG-7), it only delayed the explosion, but many maintain that it made all the difference. Ordinarily when the turbine was disconnected, an automated AZ-5 signal would shut down the reactor. However, this precise signal had already been blocked forty minutes previously, in order to carry out the turbine vibration test. Therefore the reactor kept running, and the control rods' graphite displacers did not yet begin pushing away the water filling the bottoms of the channels.

Controversy: It is often alleged that the operators deliberately left the reactor running, in order to repeat the experiment if needed. One prominent RBMK engineer has pointed out that a repeat attempt would have been infeasible, requiring perhaps an hour of electrical work, in which time the reactor would smother itself in xenon. Motivation aside, the delayed scram may nonetheless have been deliberate. The previous rundown test in 1985 had been carried out not before but following a planned shutdown, and the reactor remained running for several minutes after the end of the test. Therefore the step of shutting down the reactor was not included in the program. Still, the likely answer is to be found in the transcript of a politburo meeting: Akimov wrote in the hospital that there had been a simple miscommunication, and he continued waiting for confirmation from Kirshenbaum or a command from Dyatlov.

 

The steam valves having been closed, Gennadii Metlenko describes another minor hiccup related to the 'MPA' button:

 

They were supposed to press the buttons simultaneously. But Lisyuk was late with his button by 3-4 seconds, or actually 6 seconds later. As Lisyuk explained, he misunderstood the command. Therefore, the initial phase of the test began without the rundown block, and only after pressing the MPA button did the regulator transfer the generator's electrical parameters to the proper regime. Engaging the rundown block created an impulse in the electrical circuits for a fraction of a second, leading to a short-term drawdown in the power supply sections. Then the regime stabilized. The drawdown was around 5%. 

 

Myths/Misinformation: In other words, the turbine spent several (in reality exactly 6.5) seconds coasting down without providing any voltage to the four MCPs. This had no apparent effect on the reactor, but because there was a 6.5 second disconnect between closing the steam valves and beginning the test proper, many Soviet and post-Soviet investigators were able to 'forget' the intermediary period. This meant that the raw data could be conveniently misinterpreted to show that the reactor exploded and the pumps failed 6.5 seconds earlier, before the operators attempted to shut it down. Failure of the pumps due to cavitation remains the favored theory of the Kurchatov Institute and NIKIET, since it shifts blame from the reactor itself. But prior to and during the test, the MCPs gained a sizable safety margin to cavitation and flow rates dropped back into compliance. The data unambiguously show that flow rates did not fall until 5 seconds after the power surge began.

 

The rundown test proceeded for 36.2 seconds, during which time four main circulating pumps (MCPs) continued operating as normal, and the remaining four were powered from the ever-decreasing voltage provided by the turbine. The result was a steady reduction in coolant supply of 10-15%, as ably described by Dyatlov:

 

...the revolutions and flow rate of the circulating pumps declined... The flow rate of the other four rose slightly, but the overall coolant flow rate fell 10-15% over 40 seconds. This introduced positive reactivity into the reactor, the AR steadily maintained reactor power, compensating for this reactivity. Until 1:23:40 no change in parameters was observed. The rundown proceeded calmly. Quiet in the control room, no conversations.

 

Controversy: Virtually the exact same assessment can be found on page 65 of INSAG-7, based on raw data recovered from the reactor, rather than eyewitness accounts. However, there are some discrepancies and nuances. The INSAG-1 report from 1986 depicts a general movement of all three groups of automatic control rods (AR-1, AR-2, AR-3) into the core, one after the other. Prominent investigators such as Karpan describe the same phenomenon. INSAG-7, however, draws on diagnostic signals to suggest that only AR-1 could have theoretically approached the bottom of the core, getting no closer than 1.4 meters (This is due to a "1 overcompensation upwards" signal registering until 1:23:30). Regardless, it is clear that the automatic regulators were successfully dampening the positive reactivity caused by the rundown test. 

 

Reactor power according to the ionization chambers remained at 200 MW, while the PPDDCS recorded a power increase between 15 and 40 MW, some of which preceded the test itself. Other developments during the test included a minor increase in drum separators' water level and pressure, which had a modest contribution to the boiling and positive reactivity in the core. On a certain level it is somewhat surprising that reactor power did not begin rising faster and sooner, as some accident modelling has suggested it should have (the Kurchatov Institute predicted an increase to over 500 MW by 1:23:40). But facts remain facts.

 

It is a great irony that if the data presented by the Soviets to the IAEA in 1986 is correct, the downward movement of the automatic control rods had likely increased ORM back above 15. Combined with the declining flow rates in the MCPs, this meant that all operating parameters of the reactor were now within normal limits, just seconds before its destruction. No warning signs or emergency signals were being produced.

 

Technical Explanations: Two more instances of intense irony deserve mention. When the AZ-5 button was pressed at 1:23:40, the diesel generators had reached nearly full load. Data recovered from the oscilloscope would later show that the rundown experiment had been a success. Sufficient voltage had been provided to validate the concept as an additional safety measure. But the reactor itself was gone. Secondly, during the upcoming planned maintenance period, the bottom-insert UPS rods were supposed to be retrofitted to move into the core during an AZ-5 signal. Nikolai Karpan's calculations have shown that on April 26th this improvement would have prevented the disaster. Therefore Unit 4 exploded during a successful safety test, just hours before receiving the safety upgrade that would have saved it from its greatest flaw.

A graph from the INSAG-1 report to the IAEA in 1986. Filled with accurate data and a few carefully-placed bald-faced lies.

 

Let us step back and quote the eyewitnesses, starting with Yuri Tregub: 

 

We didn't know how the equipment would respond to the rundown, so in the first seconds I perceived... some sort of troubling sound started up. I thought it was the sound of the braking turbine. I remember all this somewhat vaguely... the sound itself I don't remember, but I remember how I described it in those first days: as if a Volga at full speed hit the brakes and spun out.

 

Du-du-du-du... that kind of sound, turning into a rumble. The building started vibrating. Yes, I thought, this isn't good. But that is probably just because of the rundown.

 

The control room shuddered. But not like during an earthquake. If you count to ten, there was a roar, the frequency of the vibrations fell. But their strength increased. Then came the blow. Because I was closer to the turbine, I assumed that it had thrown a blade. But that is just subjective, because I didn't see anything like that.

 

Kirshenbaum cried: "Water hammer in the deaerators!" The blow was nothing much. Compared to what came after. Although it was a strong blow. The control room shook. And when the SIUT shouted, I noticed that the emergency steam blow-out valve signals had lit up. The thought flickered in my mind: "Eight valves... wide open!" I rushed away, and there followed the second blow. Now that was a very strong blow. Plaster rained down, the whole building reeled... the lights went out, then emergency power came on. I rushed away from the spot where I stood, because I couldn't see anything there. I saw only that the main emergency blow-off valves were open. One valve open meant an accident, and eight valves... that is already something... something supernatural. Our only hope was that the signal was false, due to a water hammer.

If you string together the above quote with Tregub's other paragraph at the top of the page, you will see that the timing is far from clear. How soon after the start of the experiment did the vibrations start, and when was AZ-5 pressed? Other eyewitnesses such as Razim Davletbaev provide that information:

 

When the revolutions of the turbogenerator slowed to the level called for by the program... Akimov made a command to shut down the reactor, which was done by the control panel operator. After some period of time (how many seconds passed, I don't remember) a humming noise started.

 

While working at nuclear plants in various roles, I have experienced numerous emergency situations, including those accompanied by strong noises. But this hum was of a completely unfamiliar nature, with a very low tone, like a person moaning. Similar effects have been described by witnesses of earthquakes and volcanic eruptions.

 

The walls and floors staggered sharply, dust and small fragments fell from the ceiling, the lights winked out, leaving semi-darkness, only the emergency lights were on, then immediately came the second deafening blow, accompanied by thunderous reverberations. The lights came back on again, everyone in the control room stood still, the operators shouted to each other over the noise, trying to discern what had happened, what had occurred.

 

Dyatlov, standing at this point between the shift supervisor's table and the safety systems panel, loudly commanded: "Cool the reactor with emergency speed!"

 

Now we see the version of events that is indicated by INSAG-7, based in the diagnostic data and quantitative evidence. But Grigorii Lisyuk throws a spanner in the works:

 

Then there was a calm conversation, about how it was time to shut down the reactor. Then the SIUR announced that reactor power was changing with emergency speed. Then Akimov gave the sharp command: "AZ-5!" He ripped the paper covering from some button or other and someone pressed it, either him or Toptunov. After that came the explosion. When the howl of the explosion began to recede (when 1-3 seconds had passed), I saw Dyatlov approaching from the right towards the center of the control room. He told everyone to move to the reserve control room. But no one left. "Diesels!" Akimov exclaimed, and began to turn on the emergency pumps.

 
There were reports of fires in the turbine hall, on the feedwater pump platform, etc. Akimov tried to call the fire brigade, but the phones were out. And also, the dosimetrist did not let us into ABK-2, and there were 40-50 of us. In response to the question: "What's the situation?" he said, "Up to 40,000 beta particles."

Judge: The SIUR's shout that the reactor power was changing with emergency speed, was that before the AZ-5 button was pressed?
 

Lisyuk: Yes.

 

Lisyuk was the only eyewitness who described a power surge predating AZ-5, so naturally the Soviet prosecutor wanted to boost his version of events. The Soviet report to the IAEA also implied this sequence without outright saying so, probably because the data simply did not support it. Meanwhile many investigators chose to believe that the reactor had exploded before AZ-5 was even pressed, and the eyewitnesses be damned.

 

Alexander Akimov wrote the following note to Nikolai Karpan on April 26th:

 

The explosion occurred after pressing AZ-5. All actions were made in accordance with A. S. Dyatlov's instructions...

 

Meanwhile Leonid Toptunov's written testimony in court included the following statement: 

At the moment of the blast (or immediately afterwards) the control rods stopped moving...

 

In Lisyuk's written testimony (no doubt assisted by the KGB), he stated outright that Akimov pressed AZ-5. In his oral testimony above, he clarified that he had not in fact seen who pressed the button, nor was he able to identify what button was pressed. This is not surprising, given his vantage point on the opposite side of the control room. Lisyuk was also an electrician, likely unfamiliar with the SIUR panel and distracted by the electrical instruments which were his purview. In a legitimate trial with a bona fide defense attorney, his testimony would have been picked apart in cross examination. The simplest explanation is that Lisyuk simply did not notice Toptunov press AZ-5, as his attention was elsewhere. The button he saw being pressed was possibly the one that disconnected the control rods servos so that the rods would fall under their own weight. He may also have seen the second press of AZ-5 that was recorded on teletype. 

 

Teletype printout of reactor signals. At 1:23:39: "Emergency protection 5". At 1:23:41: "Emergency p..." The double AZ-5 signal recorded may be a glitch in the systems or the result of vibrations, but some have theorized that Toptunov physically pressed the button twice. He could have been attempting to move every control rod downwards a few centimeters, to maintain the current power level. The rods would stop moving when the button was released (this was changed after the accident, and the button replaced with a switch). Image Credit: Nikolai Karpan

 

Various eyewitnesses have given a wide range of perceived durations between the start of the test and the explosion or accident-related noises, so the non-temporal description by Genadii Metlenko is very helpful:

 

The turbine revolutions steadily decreased, frequency and voltage decreased accordingly. When the turbine revolutions decreased to 2100, and frequency to 35 Hz, voltage 0.7 of nominal, I heard a rolling thunder, as can happen during water hammers. The sound came from the direction of the turbine hall. The building began vibrating strongly. Rubbish dropped from the ceiling. I had the impression that the control room was collapsing. The command "shut down the reactor" was given by Akimov when the turbine fell to 2500 revolutions/minute. I remember this well, since I was observing the turbine revolutions. The command was given in a calm tone of voice.

 

Anatoliy Dyatlov:

 

Having heard some kind of conversation, I turned and saw that reactor engineer L. Toptunov was speaking with A. Akimov. I was ten meters away from them and did not hear what Toptunov said. Sasha Akimov ordered to shut down the reactor and gestured with his finger: hit the button. Myself I turned back to the safety panel, which I had been observing.

 

There was nothing nervous in their behavior; it was a calm conversation and calm command. This is confirmed by G. P. Metlenko and A. Kukhar of the electrical division, who had just entered the control room.

 

Why Akimov delayed the command to shut down the reactor will never be known. In the first days following the accident we still had discussions, before we were put in separate hospital rooms, and I could have asked, but I did not view it as significant. The explosion could have occurred 36 seconds earlier, that is the only difference.

At 1:23:40 the emergency protection button to shut down the reactor was pressed, upon completion of the test. This button is used in both emergency and in ordinary situations. 187 control rods moved into the active zone and according to all the canon should have halted the chain reaction.

But at 1:23:43 signals appeared announcing excess reactor power and reduction in power multiplication period (high rate of power increase). These signals alone would insert control rods into the zone, but they were already moving due to the AZ-5 button. Other emergency warnings and signals appeared: power increase, pressure increase in the primary circuit...

At 1:23:47: explosion, shaking the entire building, and 1-2 seconds later, by my subjective impression, another more powerful explosion. The emergency protection rods stopped before reaching the halfway mark. That's all.

In such a businesslike everyday situation the RBMK-1000 reactor of ChNPP Unit 4 was blown up by the emergency protection button (!?!?).

 

Dyatlov's indignation is understandable. But it would be months before anyone in the control room learned exactly what had happened, assuming they lived to see it at all. Soviet investigators would arrive at a tolerably complete picture of the accident's causes within just a few weeks. The rest of the world would wait years.

  

 Continue to Part 7: Bitter Wormwood

 

The 'golden' deaerator corridor by which Lisyuk and other test spectators evacuated to the eastern half of the plant, soon followed by the first accident victims such as Shashenok and Degtaryenko, who had suffered steam burns. Other operators would head in the other direction, to attempt to save the burning reactor.

Sources Cited:

• Nikolai Karpan, Revenge of the Peaceful Atom.
• Yuri Scherbak, Chernobyl.
• Razim Davletbaev, The Last Shift. 
• Anatoliy Dyatlov, Chernobyl: How It Was. 
• INSAG-7.