The Fukushima Syndrome, 2

This is the tenth article in the Where We Came From series.

Last week: Attempting to learn more about the ongoing crisis at the Fukushima Daiichi reactor, I study a complex simulation program that was released on the Atari 400/800 computer in 1981, SCRAM by Chris Crawford. I’ve read the manual and bought the T-shirt. Now it’s time to show this reactor what I’m made of.

Atoms, apparently.

It’s probably just me, but I like to visualise the process of nuclear fission as atomic sex. All it takes is just one neutron to penetrate the heart of an atom, to fertilise it. The fertilised atom divides into two and, in this orgasm of reproduction, fires more neutrons into the atomic void. These neutrons find other atoms to fertilise… and the reactor core becomes a vast, self-sustaining sex orgy.

But what happens if the orgy gets out of control? How do you stop nuclear reproduction? Obvious. Nuclear condoms.

Enrico Fermi was well aware of the problem when he and his associates were building the first nuclear reactor on a University of Chicago squash court in 1942. They used neutron-absorbing control rods to manage the reaction. Withdrawing the rods from the core would allow a self-sustaining nuclear reaction to take place and re-inserting them back in would subdue it. As this was the first reactor, the test of theory, control was their primary concern.

One story goes that the rods were held up by a rope and a codeword to order the rope be cut was “SCRAM” which either stood for “Safety Control Rod Ax Man” or “Start Cutting Right Away, Man”. Another story is that the emergency shutdown button was marked “SCRAM” because the team would have to scram out of there if the button was actually needed. Whilst the origin of the term “scram” in the nuclear context is unclear, its use is ubiquitous.

At Fukushima Daiichi, the reactors were scrammed as soon as the earthquake hit. This is standard procedure. An hour later, a tsunami battered the plant and disabled the backup power generators – sending the reactors hurtling towards meltdown.

Why?

A Reactor Operator’s Diary

As explained last week, SCRAM is a hardcore simulation with limited visual feedback. It’s a game that takes its subject matter seriously and is impossible to figure out without the manual. Rather than try to explain in text how SCRAM functions, I put together a video instead (featuring original music composed by regular commentmeister Sid Menon, a.k.a. BeamSplashX).

What follows is the diary of my progress through the risk levels. I only considered a risk level “completed” if I hit the corresponding energy production target in the manual and achieved cold shutdown.

Risk 1: 1000 MWH

Meltdown! First quake, temperature shoots up in the primary loop so I assume it must be one of the primary loop pumps. Fixed. Second quake, saw the same thing, made the same conclusion and wrong-o. Waste most of my workers fixing the perfectly good components. Turns out to be a broken pump in the tertiary loop. Third quake leads me on a similar wild goose chase. After my fifth quake I have only five workers remaining, and the temperature is rising. Meltdown occurs with around 814 MWH generated.

Successful on the second attempt, so at least I’m learning.

Risk 2: 900 MWH

Game aspect becomes more prominent but am I learning anything about Fukushima? All the new reactor knowledge in my head was acquired via the manual and not the game itself.

But I’ve learnt another lesson. In our CSI times, we’ve become entrenched in a make-believe of computer systems that tell us answers and spell out the truth. Reality is not quite like that. I like the idea of having to work out where the problem is myself.

Not sure of the purpose of the auxiliary feedwater tank attached to the middle loop, so unlikely I am playing the way Crawford intended.

Seems like games on risk levels 1 and 2 take around 20-25 minutes and it gets dull waiting for something to happen. But I reach the production target of 900MWH and achieve cold shutdown first time.

Risk 3: 800 MWH

Getting better at this. A drop in temperature at a loop inflow is a key signal for pump failure – as coolant isn’t flowing as fast, heat backs up at the top of the loop and, just for a moment, drops at the bottom. But composite problems are a nightmare. The signals are confused when you have multiple component problems, gets frustrating when you can’t work out what’s going on without guesswork.

After reaching target, I bring the control rods in only to get hit by a cluster of quakes. Think I’m going to lose… but cold shutdown is achieved just in the nick of time.

Risk 4: 700 MWH

The tedium between quakes now becomes tension. I worry whether I can fix a broken component in time before the next quake hits.

I reach 292 MWH and have 45 workers left. When I reach 418 MWH, failure to find which valve is broken leads to wasted workers, 25 left. Have a bout of “steam voiding” so to cool the core I push the control rods in; the power output falls. I literally despite a fall in power output because the longer you play, the more quakes you are exposed to.

I eventually run out of workers around 550 MWH and then lots of quakes follow. It’s clear I’m not going to make it and I wonder how much of this is luck. I restart.

I take action to make SCRAM an easier ride and begin to game the simulation.

First: I use quake-free mode Risk 0 to pump the pressure up a bit and ensure the turbine is generating 999 MW before I increase the risk level. Although the energy score is reset when you change level, the current production rate of the reactor is preserved.

Second: I decide not to fix every component problem until I need to, because quakes don’t wait for me to track them down. Until now I’ve been doing experiments to find out which component was broken and this tinkering destabilises the reactor state – this means if a quake hits, it’s difficult to see what new dynamic has arrived.

On my second attempt, as I approach 700MWH I get hit with multiple pump problems. 645MWH, 20 Workers left. The shutdown phase is really tricky, not just because it is so slow but also as the system is in temperature freefall, reading pump problems is virtually impossible.

But cold shutdown is achieved, 726 MWH and 15 workers left.

Risk 5: 600 MWH

Success! 628 MWH, 25 workers left. When temperatures start rising I try to fix within seconds. A drop in power output is more fatal to your long-term game health than anything else, as it substantially elongates your game, ratcheting up the quake risk. And I ram the control condoms into the reactor as soon as I get near the target. I become efficient at my job.

Risk 6: 500 MWH

Takes two attempts to succeed. 522 MWH, 25 workers left.

Risk 7: 400 MWH

Here is where I skewer any pretence of simulation. You can turn the auxiliary pumps on which will have no effect on the reactor, provided you leave the auxiliary valves closed. Once pumps are activated, they become potential targets for each quake so I’m using the game’s strange logic of “one quake, one break” to deflect damage away from components that are vital for success. This is truly a strategy that would work in real life.

Cold shutdown achieved with 414 MWH produced, 35 workers left.

Risk 8: 300 MWH

I make 321 MWH but cannot survive the onslaught of quakes that befall me during reactor scram. Unable to identify what’s broken during the shutdown, I run out of workers with random guesses.

I realise I’ve had enough and decide to end my experiment.

Fail Safe

The game itself didn’t teach me anything – it was the manual that did all the work. The game often overwhelms the player with problems and, on the higher risk levels, I never got the sense I was resolving them with anything less than luck. It’s not particularly fun, it doesn’t educate through its mechanics and it feels unfair. I would’ve liked more emphasis on black box problem-solving in a real-world context and less on anxiety.

Crawford came to similar conclusions in his book, Chris Crawford on Game Design:

“All in all, Scram was a stupid game devoid of entertainment value. … I could hide behind the fact that software in those days was mostly bad. … If I had it all to do over again, I would start my design process by asking myself, “What is fun and interesting about nuclear power plants?” The answer, of course, would be “Not much,” and I would walk away from the idea of building such a game.”

So is SCRAM just an oddity from an era when game design was crawling out of the swamp? Is it the kind of experimental work we wouldn’t see now? Hmm…

• Oakflat Nuclear Power Plant Simulator (video)
• Nuclear Power Plant
• BWR Nuclear Simulator

If you’re looking for evidence that the 80s were the only decade where true gaming experimentation occurred, you are not going to find it here.

Twilight’s Last Gleaming

So SCRAM never told me why Fukushima Daiichi experienced core meltdown. Why was it important for the generators in Fukushima to keep running if the control rods were inserted? Even in the game, if your pumps break, the control rods are not sufficient to shut down the plant. I had to fall back on the internet for the answer.

A reactor core also produces what is known as decay heat because it is radioactive. A concentration of decay heat alone will be enough to bring about a meltdown and so cooling is essential. You can’t just shut it off because the nature of the very material itself is unstable. So a key to safe reactor design is being able to keep the coolant circulating in even the most extreme scenarios.

If Fukushima Daiichi had better defences against the tsunami, we wouldn’t be talking about a 20km exclusion zone right now. But it’s important to note there’s another Fukushima nuclear plant, Fukushima Daini, which was also overwhelmed by the tsunami. It achieved cold shutdown four days after the earthquake and is not in danger of meltdown.

It’s true that the Japanese nuclear reactor industry has been bizarrely complacent about safety – there have been many concerns and incidents over the years. But there’s also something to be said about the Swiss cheese model of system accidents. In a complex system, it is difficult to identify exactly how disaster could strike. Hindsight and experience are much better at suggesting improvements to a system than thought experimenting our way to the perfect design. We learn through error.

But with nuclear power, accidents have the potential for large-scale environmental damage. Charles Perrow, who devised “normal accident theory” even went so far as to say “some technologies, such as nuclear power, should simply be abandoned because they are not worth the risk.”

Anti-nuclear hysteria has inculcated a nuclear industry that is hyper-defensive and dwells in a state of minimal transparency. After the first explosion at Fukushima Daiichi the media talked up the danger of explosions for days, suggesting that Japan was teetering on the edge of a new Chernobyl. This hype obscured the real story of a tsunami-devastated coast, of death beyond comprehension. Nuclear power comes in for a lot of criticism but in the wake of a media frenzy like we have just seen, it is not out of place to ask whether nuclear power gets a raw deal.

But the nuclear industry does itself no favours. The latest bad news came from Reuters:

The workers who stayed on to try to stabilise the plant in the darkest hours after March 11 were lauded as the “Fukushima 50” for their selflessness. But behind the heroism is a legacy of Japanese nuclear workers facing hazards with little oversight, according to interviews with more than two dozen current and former nuclear workers, doctors and others.

Since the start of the nuclear boom in the 1970s, Japan’s utilities have relied on temporary workers for maintenance and plant repair jobs, the experts said. They were often paid in cash with little training and no follow-up health screening.

I just wish the trials and tribulations of nuclear power were as simple those an Atari gamer faced in the compact, sterile microcosm of SCRAM: A Nuclear Power Plant Simulation.