Three Myths of the Three Mile Island Accident

Nuclear engineer Arnie Gundersen gives a talk on his calculations of the amount of radiation released during the accident at Three Mile Island. Mr. Gundersen's calculations differ from those of the NRC's and official industry estimates. Arnie Gundersen was a senior executive in the nuclear industry with over twenty years experience. Mr. Gundersen holds Bachelors and Masters degrees in Engineering from Rensselaer Polytechnic Institute. In 1990 he came forward as a whistleblower and was fired that same year. Over the next several years, his case got a great deal of attention, and he testified before Congress during hearings on ways to protect whistleblowers. Mr. Gundersen is now a prominent nuclear safety expert witness.

Associated Materials

Epidemiologist Steve Wing Discusses the Increases in Cancer Rates After Three Mile Island Accident (Part 1)

Epidemiologist Steve Wing Discusses the Increases in Cancer Rates After Three Mile Island Accident (Part 2)




Arnie Gundersen: I'm Arnie Gundersen. I was on the industry side of this argument until about 1992. I had people reporting to me at Three Mile Island during the recovery. I had a TMI shirt that said "I survived Three Mile Island" and I was on television saying that...I think I said "the Titanic hit the iceberg and the iceberg sunk". So, I was clearly of the opinion that this was a non-event until about 1992. I started looking into it, and was asked to be an expert at the trial in 1994 and really had a chance to dig into it then, and my opinions have essentially gone 180 degrees. It is a significant event that we need to learn from if we are to have a new generation of nuclear plants which is sort of the direction I am taking this. And then the other thing is, that from a health effect we better understand what happened here and with Dr. Wing's data if we can learn from it as well.

So there are two issues: What happened in the past to the people who may have been exposed here in Harrisburg and Middletown and then also, looking forward, what have we learned from TMI and that is what I will be talking about.

One last thing, Dr. Wing and I have never met until today. It is interesting, I had these Curie release numbers for years and as it turns out, they were dramatically different than what the Nuclear Regulatory Commission says. So I have been talking and talking and finally I realized that Dr. Wing's epidemiological data, which this release data supports... But we have never met and I would not have known him until today.

The title is Three Myths. There are more than three myths, but these are the three I decided to cover in my 20 minutes:

1. Should an evacuation have been ordered? 2. Did the containment leak? 3. How much radiation was really released?

Like I said, there are more myths, but given 20 minutes, this will be a rush.

So, "Should an evacuation have been ordered?" is the first thing, and I break that into 3 segments. I break it into before 7 o'clock, what information was known to the people of Three Mile Island before 7, 7:30, 8 in the morning, and then around 10 and then around 2 on the first day. I will talk to that.

So, around 7 o'clock in the morning, an engineer and his supervisor using an approved procedure calculated that the exposure in Goldsboro might be as high as 10 (R) an hour. Now it was an approved procedure and people had worked on it for years. And it was actually a TMI Unit 1 procedure. So this is not a new procedure. By the procedure, an evacuation was required. There is no doubt that by the written process that people not in a crisis situation had available to them, by 7-7:30 in the morning, an evacuation was required. At 7:30, TMI called the state and told them that they had 10 (R) an hour but they said that it seemed too conservative. One of the things they said is that the pressure inside the containment was not as high as they had expected. But the state was aware that the calculation showed 10 (R) an hour. But TMI's position was that it seemed too conservative. They said that the pressure was not high enough. Well, within the calculation, there was no pressure dependency. So basically they went outside the realm in a crisis situation, as opposed to letting the procedure govern how you should be working your way through.

What they did not tell the state in that 7:30 phone call is that employees working outside had already begun to receive exposures. There is at least one case of an exposure of 20 millirem to an employee who was out on the grounds before 7:30 in the morning.

They did not tell the state that already almost every radiation detector in the plant was off scale. They did say that a helicopter flew to Goldsboro at about 7:30 in the morning and found no radiation. Now there are two problems with that. Actually there are 3 problems with that. It was a very calm day. They will admit that the helicopter actually got to Goldsboro before the plume would have gotten to Goldsboro. So had there been radiation coming, it would not have gotten there and so based on that, they said, well, there was no radiation in Goldsboro. That is problem #1.

Problem # 2 was that the plume, the center line of the plume was right there. In Goldsboro, if they were off by 6 degrees, in other words, if they were not right on the center line of the plume, they could have been off by a factor of 10,000 in the dose they recorded. It is something called a Chi/Q, it is a dispersion coefficient and the plume would have been narrowly concentrated and a 6 degree position error where they put the instrument would have resulted in as much as a 10,000-fold difference in radiation that they measured. So that is problem #2. It is very hard to chase a plume. And problem number 3 is that the helicopter actually arrived on site at 8:30. There was no helicopter at 7:30. So my belief is that at 7:30, procedures told them they should have evacuated and in a situation like this, you do not try to change the rules on the fly.

The next time I suggest would have been a good time to have evacuated, is around 10 o'clock in the morning, between 10 and 11. By then, they knew that core thermal couples, that is a device to measure temperature inside the core, were measuring 2,100 degrees Fahrenheit. Well, normally they would measure about 500 degrees and 2,100 degrees indicates that the fuel rods are entering something called a zirc-water reaction. Fuel rods are made of zirconium and they scavenge oxygen out of H2O, so the oxygen gets pulled out of the H20 releasing hydrogen. So by 10 o'clock in the morning, they knew that there was hydrogen being generated. That should have surprised no one. In the hot leg from the reactor, that is the leg that would normally carry the hot water out of the reactor, the thermal couples were reading in excess of 700 degrees. Now, given the pressure that the reactor is capable of withstanding, it could not have been water at 700 degrees because the relief valve would have either opened or the vessel would crack.

So that tells me that there was air, hot air, running through the hot leg at 700 degrees being already heated by the core, another indication that there was not enough cooling, or no cooling, going through the core.

Also by 10 o'clock, their bad reactor cooling pumps were massive, 3-4,000 horsepower pumps. And the amperage for those pumps was very low. And that is an indication that they are not pumping water, it is an indication that they are pumping steam or air. And the next thing is that in a pressurized water reactor, they have neutron monitors in the core, but they also have neutron monitors outside of the nuclear reactor. And the neutron monitors outside of the nuclear reactor were reading very high levels of neutrons. Well, what that means is that there was no water to moderate the neutrons. Even if the reaction was shut down, there was still more neutrons than they had ever experienced outside the core. And that is an indication that the core had lost it's water and was uncovered.

Again around 10 o'clock, the radiation monitors in the dome of the containment were at lethal levels, thousands of (R) an hour. Again an indication that fuel is breaking down. So, you had indications of zirc-water reaction from the temperature, you had indications of fuel breaking down from these radiation levels. Someone took a reactor cooling sample and noticed that when they down to a line and they opened a little spigot, filled a vial and normally those vials are very non-radioactive. This was reading 200 R an hour. That is lethal in 2 hours. That is an incredible amount, another indication of fuel failure. Around 10 o'clock in the morning, Health Physics asked the plant management to evacuate the auxiliary building. So all these things were happening and yet the state was not told that things were really out of control.

The Plant Manager at the time was a guy named Miller and here is what he had to say over the next couple of years about what was going on in that time frame. "They were hot enough that they scared you." And he was talking about the in-core temperature. Well, if you are scared, one would think that an evacuation might be in order. "Pretty early we were scared. Radiation was all over the place. Everything was off scale." Another indication; if you are scared, it is about time to at least tell the civilians that is is time to move out. I should note that everything I have got up here is substantiated with footnotes in a report which will be up on TMIA's website. Literally, every one of these quotes is referenced back to a reference document and the same with the temperatures earlier.

This was another interesting quote: "We don't know where the hell the plant was going." Now Miller said that in a phone call to Parsippany, Parsippany was the headquarters office, at 7:30 in the morning. He had the sense to tape the call and I had a chance to read the transcript. And it was pretty clear in my mind that Miller was suggesting we should go to a general emergency. And the people in Parsippany talked him down to a site emergency. And this was one of the quotes from his call to Parsippany. So Parsippany backed Miller down at 7:30. But I think until 7:30, his heart was in the right place, and I think he was suggesting that it is time to order an evacuation. After that, he changed his tune. But at 7:30 in the morning, I think his heart was in the right place. And finally, "We were not in our minds convinced that the core was totally covered." That is another indication that it is time to let the civilians know to head for the hills. But it did not happen.

Last time where I think anybody of conscience would have ordered an evacuation, is before 2 o'clock in the afternoon. Based on the core temperature, we have been over this, clearly hydrogen was being generated. It could not have not been happening. There is a good double negative.

This is an important piece, the next two. At 12:20, the NRC asked TMI, "What is the temperature in the core?" TMI got back to them shortly thereafter and they said, "We don't know. The computer is printing question marks." And then they said, "That means that the computer is messed up." In fact, question marks meant that the temperature in the core was over 700 degrees. They did not know how high, but they knew that it was high, and it was another indication of a meltdown in progress. So the other piece of this is that two temperature indications were not in question marks. They were 599 degrees. And they never told the NRC that, hey, we got two that are still on scale that are reading 599, in fact they said that the computer is messed up. Well, a couple minutes before 2, there was an hydrogen explosion. Now the industry will call it a hydrogen burn, but it was a hydrogen explosion.

In lay terms, there was a big change in pressure and a loud noise that shook the building. To me that is an explosion. In industry jargon, it is a hydrogen burn. But it was an explosion.

The NRC was informed 2 days later. Plant Manager Miller was in the control room at the time, based on affidavits from 4 reactor operators. They all said Miller knew about it. And the control room shook. Now when your building starts shaking, I think that is about the last indication you need that you really should let the civilians know to head for the hills. After that it was unconscionable that an evacuation was not ordered on the first day. I would say 7:30, but even if you give them the benefit of the doubt at 2, an evacuation should have been ordered on the first day.

Did the containment leak? What is wrong with that picture? This is the audience participation section. See the spike. This is not post grad engineering, there is a spike there. This is the containment pressure. This recording was available to the operators in the control room. And a little before 2 o'clock, there was a huge spike in pressure. The peak of that is about 28 pounds. It is not clear that that really is the peak, the needle was moving so fast it might have left the page. But the industry's position is that the peak containment pressure was designed for 40 or 50, so therefore the containment did not leak, the peak was below it. There are a couple of problems with that. Now what I have done on the next slide is I have cut out from 2 o'clock to about 4 o'clock, but this is the slide I will essentially be referring to in the discussion.

To me, what is important here is that before the spike the containment was pressurized. It was at roughly 2-3 pounds of pressure. That means it is containing, because the core is generating a lot of heat, and just like in a pressure cooker, the pressure is above what it is outside. After the accident spike, the containment goes down to zero. It sits at outside pressure after that. The rapid drop was due to, of course, containment sprays going on and mechanisms to drop the containment pressure quickly and on top of that, with a hydrogen explosion, it would not stay high anyway. After about 5 minutes, any of those pressure mitigating and energy removal systems were gone. If the containment really had maintained it's integrity, this line should be about 3 pounds higher than where it is. The same spike, I just shortened it a little bit. So the other piece of this is that is the containment pressure for the entire containment. There is something called sub-compartment pressurization. This explosion did not occur in the whole containment, it occurred in a sub-compartment. Some photographs 5 or 6 years after the fact showed doors being blown off their hinges in a sub-compartment. So it is very likely that a sub-compartment could have exceeded 100 pounds per square inch. So what I believe happened, based on sub-compartment pressure, is that a leak occurred in a portion of the containment wall, perhaps not all through the containment, but a portion of that containment wall got a crack and started to leak. And I base that on a couple of things.

Before 3 pounds: after zero: never returning. Zero is atmospheric in this system. Now during the trial, the plaintiffs hired Dr. Reytblatt from the University at Bridgeport. Dr. Reytblatt is a structural engineering professor at Bridgeport and this is what he said, "A plausible release of up to 8 - 10% of the volatiles may have occurred due to the unavailability of the containment system at the time of the accident."

So, Reytblatt concluded that about 10% of the radiation within the containment leaked out as a result of that pressure spike. I will get to what that means as far as total radiation in a minute. But if you do not want to believe Arnie Gundersen, this guy is a professor of structural engineering at the University of Bridgeport and his expert report is part of the trial transcript.

To my mind, that was not enough. So, what I did is, I went back into old plant data, conveniently provided by John Daniel, the industry expert against me. So this is industry data I went back through. And I found 3 radiation detectors that went off scale within an hour after the explosion. Now remember, most of the radiation detectors had already gone off scale. I found 3 that were on scale that suddenly then went off scale immediately after the explosion. The first one recorded a 5-fold increase, and here are the numbers here. You can trace it back. The second one recorded a 10-fold increase and then went off scale. This one I think is the most interesting. This one doubled and it was protected by 4" of lead. Well, 4" of lead will eliminate everything except the most powerful gamma rays. So in addition to a doubling of incredibly powerful gamma rays, what this also shows is that there had to be low level gammas and a lot of alphas and a lot of beta that were also released that this instrument never picked up.

So for those 3 reasons, the shape of the curve, Dr. Reytblatt's analysis and this forensic evidence, I believe I can show that the containment leaked.

Question from the audience: And where were those monitors located? On site?

Arnie Gundersen: You would have to go back in the transcript. They were very near the containment. In the annular gap around the containment and in areas right next to it and in the auxiliary building.

"How much radiation escaped?" is the last question. I should preface this: I do not believe that there is a UFO in area 51, I believe that Kennedy was shot by one guy and I believe that two airplanes crashed into the World Trade Center and knocked it down and there is no conspiracy on any of that. However, I believe that the data for Three Mile Island on releases is significantly larger than is reported by the federal government.

OK, this is how Dilbert would explain it: Panel 1: "A reporter wants to see you." Panel 2: "He claims we have been delivering all of our garbage to the local park for twenty years. Panel 3: "How is that even possible?" "The secret is in the spreading."


The secret is in the spreading. We will talk about that. You have to remember that there is no measurement of how much radiation was released. Every monitor was broken, it had failed high, it had burned out like turning a camera toward the sun, or something like that. Every radiation monitor in the plant had burned out on the releases. So what happens is that assumptions had to be made based on off-site exposures. For instance, if I have got 7 packs of cigarettes (I don't smoke cigarettes) and we were in a large auditorium and not a small room, and I started to smoke, but you couldn't see me smoking, but you could smell the smoke and you are not downward from me, you are off to the side, and I asked you at the end of my smoking how many cigarettes did I smoke? Well, you could make assumptions about the concentration of smoke and back calculate how many cigarettes. And there is a lot of assumptions that have to be made, but what has happened in Three Mile Island is that a lot assumptions were made and all of them were non-conservative. They were all low ball assumptions. And I will talk about that. So, the key is the assumptions.

Another Dilbert: Panel 1: I can do this feasibility analysis in two minutes. Panel 2: It's the worst idea in the world. Numbers don't lie. Panel 3: Our CEO loves the idea. Luckily, assumptions do lie. (laughter) And you can see Dilbert is my favorite comic strip. There are no more Dilberts. O.K.

John Collins was a guy pretty high up in the NRC structure and this is what he had to say about off-site monitoring. Collins said: "My problem . . . the concern I have about aerial monitoring was that for the first three days we were pretty much into a very static air condition. There was very little dispersion. When you were flying your helicopter and taking your aerial measurements, you were actually reading erroneous readings... I really doubt some of the measurements that were made." You know if you watch a helicopter do a rescue, what happens? It is taking clean air from up above and blowing it down and pushing the water out. Well the same thing was happening with the aerial surveys around TMI. The helicopter was taking clean air and blowing it down on the radiation detectors that hung below.

So I agree with Collins, whatever data came off the helicopters is erroneous.

Second is that the wind was very light. And this is important in a river valley site. And I think rolls into what Dr. Wing will show. In a river valley site in the morning, you get very, very static air and the plume was meandering but it was not traveling very fast and that is a concern. More on Collins: "... not only should we have good monitors but also people who understand how to use them. That was a problem since day one. They get data and no one sits down and evaluates the data to try and understand what it means." This is the NRC talking about the data which was recorded off site after the accident.

And the last, and probably the most important is, they had to chase the plume in a car. As I was explaining, in a plume variation, from the center of the plume, 6 degrees off, if you miss the plume by 600 feet, you would be measuring 1,000 or 10,000 times less radiation than was on the center line. So when you hear of a person being exposed (the metallic taste, hair loss issues) and perhaps the neighbor wasn't, the reason was that the dispersion of the plume was very narrow. You could easily have a factor of 10,000 according to a Dr. Vergeiner, who was the meteorologist on the job, when you look just 600 feet off at a mile, it would be about 1200 feet off at 2 miles. But, again the further out you go, you just have to move a couple hundred feet off the center of the plume to have a dramatic difference in the amount of radiation.

The NRC estimates that about 10 million Curies of radiation were released; it is on their website. A Curie is 37 billion disintegrations per second. So, just for the heck of it, I multiplied 37 billion times 10 million and there is an awful lot of zeros there. These are disintegrations per second and in static air, that radiation stays behind and just keeps disintegrating at that rate every second until it can get blown out.

I think the NRC's estimate is wrong and I have got a couple of different ways of proving it. But it is important, we are starting at a number 10,000,000, we won't worry about curies, but disintegration, is what the NRC said was released.

This number was created by an NRC manager named Lake Barrett. That is not a place, that is a person's name. And it is actually a NUREG-0637, Appendix C, his analysis. Barrett used time average plume dispersion as opposed to hour-to-hour plume dispersion, and that has a tendency of flattening the curve, so it reduces the exposures. Barrett assumed that the center of the plume hit the detector. And I have already shown that if you were off by 600 feet, you have got a factor of 10,000 difference. Barrett then averaged 7 days or 8 days or 10 days of data and wound up with a number lower than any of the other numbers in his calculations. It is kind of interesting. This is from Barrett. Barrett says on the first day of the accident, 14 million Curies were released. Well the NRC's website, of which he was a member, says there is 10 total. If you add up all of Barrett's numbers, he comes up with 36 million Curies. So this is the NRC's estimate, but the website shows 10. And on top of that, the NRC's estimate is, the time averaging of the dispersion can cause a 10-fold error. Being on the center line of the plume versus being off the plume by just a little bit can cause a thousand, some of the data says a 10,000 fold error. And averaging the data changes it by about a factor of 3. The net effect is that the NRC's 10 million could be wrong a thousand fold. The NRC's could be low, based on those assumptions by a thousand fold.

Another way of coming at the data. The industry hired a guy named John Daniel to go up against me in court. This was during the trial. And Daniel magically calculated 10 million curies. When he listed his assumptions, I went back through the assumptions and using his assumptions correctly, I came up with 150 million. Well then Daniel came back and the judge let him re-do his expert report, and the industry's number is now 17 million Curies released. Well this puts the NRC in an interesting position, because the guardian of public health and safety has the lowest estimate on the totem pole for what was released from the accident. Even the industry is almost twice as high. Their [NRC] own experts are 3 times as high and in fact, if you look at the data, and all of the non-conservative assumptions, it can easily be on the order of a thousand times higher than the NRC's estimate, which puts you at around a billion Curies. There is another way of getting at that also. Dr. Reytblatt, the structural engineer on the containment, said that about 10 percent of the material inside the containment leaked. Dr. Akers, which is an industry expert, said there was 10 billion Curies inside the containment. 10 percent of that is around a billion. So Aker's number, this is an industry guy, this is not me, says 10 billion, a tenth of it got out by Reytblatt, puts you at around a billion Curies, not the 10 million the NRC is advertising on their website.

Recently released records from Hershey Chocolate were provided in a book by Dr. Helen Caldicott, "Nuclear Power is Not the Answer". She is quoting Hershey data in her book that shows that iodine-131 was measured in cow's milk 150 miles away from Middletown. Well, 10 million curies at the NRC website does not get you to being able to detect iodine 150 miles out. I have not seen the hard copy of her report, but she is going on 4, 5, or 6 Hershey internal documents. We all know, and it was publicized, that Hershey froze milk. And that was a good prudent business decision. But Hershey had data apparently by Dr. Caldicott, Hershey had data that would have been helpful to civilians in the area, to let them know that, in fact, the plume was out at 150 miles. That is pretty significant because this time of year, grass does not grow very fast, so the cows were on silage, which meant that they were probably getting the iodine as an inhalation dose, as opposed to eating it out of the grass.

Anyway, this stuff rolls into the future of nuclear power, which I wanted to talk about on this slide. If you believe that only 10 million Curies got out, the NRC has made that sort of gospel, it is something called the alternate source term. It has allowed power plants to reduce the amount of radiation which they claim to release, which then turns around and then they can increase the power. So, a lot of plants have gone through power upgrades as a result of Three Mile Island because the NRC is allowing them to say, well, a lot less radiation got out than we thought. So, they have lowered the source term which has allowed them to crank up the power so they can get back to where they were. But in fact, if you do not believe the NRC's 10 million Curies, then the alternate source term and all these power increases is, in fact, wrong.

There is also, less robust containments are planned for the next generation reactors on the basis of how well TMI survived it. But I think the data does not show that TMI survived it. And finally, there is a lot of consideration of collapsing evacuation zones, or even eliminating evacuation zones, based on the success of TMI and I really question that.

So, my conclusion is that I think that the numbers on the NRC's website are wrong and they are wrong by between a factor of a hundred and a factor of a thousand. If everything, if all these non-conservatives build up, it could be off by 10,000, but some of these have to go the other way. So, between a hundredfold and a thousandfold increase from where the NRC's website is. And, I also conclude that the containment failed after the hydrogen detonation.