Talk:Three Mile Island accident

TMI accident remediation

I'm new to the Wiki community, at least as a contributor. I wanted to add what I believe is useful commentary about some fixes that were made to help prevent future nuclear accidents. Specifically, I wanted to add right after the following sentence in the existing article:

"Because of the lack of a dedicated instrument to measure the level of water in the core, operators judged the level of water in the core solely by the level in the pressurizer. Since it was high, they assumed that the core was properly covered with coolant, unaware that because of the voids forming in the reactor vessel, the indicator provided false readings."

...a comment something like this:

"Shortly after the TMI accident, the nuclear steam supply system vendor Combustion Engineering developed just such a dedicated level sensor. The sensor used a device called a heated-junction thermocouple, and took advantage of the fact that water is a better conductor of heat than steam. This level sensor was inserted into an in-core instrumentation thimble inside the reactor core. At 5 different points along the approximate 12 foot length of this sensor, a pair of thermocouples were placed - one heated and one unheated. The temperature difference between the heated and unheated thermocouples would indicate whether or not water covered that point on the sensor. If there was water, the temperature difference was a few degrees Centigrade; no water - the difference would jump to several hundred degrees Centigrade. The temperature difference for each of the five thermocouple pairs was transmitted to a water level indicator in the power plant control room. Each of the five signals was reduced to a green or red light - green for water, red for steam. If all five thermocouple pairs indicated water, the plant operator would see a vertical stack of five green lights - good times. If an accident occurred, and water in the core started to flash to steam, the lights would change from green to red, beginning at the top of the stack. This is one of the remediation steps taken to reduce the likelihood of human error at nuclear power plants. This flashing of water to steam occurs isentropically - at a constant pressure - underscoring why traditional pressure sensors were useless. The first such HJTC in-core water level sensor was installed at the Waterford III Nuclear Generating Station in 1984 by Combustion Engineering engineers."

Interested in feedback on how to incorporate this, where to incorporate it, whether it's even relevant to the core story... ...I think it is because this article, very well-written as it is, leaves the reader to wonder what has been done to redress this particular problem.

Johnjmccauley (talk) 16:32, 6 March 2010 (UTC)

You bring up a great point. We could actually create a new section in this article with the title being something like: "How TMI affected the nuclear industry". I know Shearon Harris Nuclear Power Plant installed a Reactor Vessel Level Indication System (RVLIS) due to the accident at TMI. However, its indications are different from what you described, it has an entire console dedicated to the output of RVLIS and has a 1980's style black and green screen showing an outline of the RV and shows water level using two arrows (must be using two channels). There are many other examples. Such as the bypass valves for the Condensate Polisher Demineralizers (CPD) now have auxiliary tanks to store pressurized air in order to operate the bypass valves even if there's a complete loss of Instrument Air in the power plant. TMI lost their main feed water supply due to the bypass valves to the CPD not opening. Another example is the installation of the Hydrogen removal system in the reactor containment building. High Hydrogen concentrations became a concern days after the TMI accident, therefore, the industry came up with two methods to remove Hydrogen from the containment atmosphere. There are also rigorous processes (engineered and proceduralized) that reduce moisture in the Instrument Air system of the power plant. Water in the IA system prevented valves from operating correctly. But for the most part, it really wasn't an engineering mistake that caused the accident, but mainly human error. So a lot of the fixes made in the industry have to do with operator training and the implementation of procedures. Also, training for how the Main Control Room reacts to events have changed based on the accident at TMI. As one operator from TMI at the time of the accident said that if the operators in the MCR did nothing and allowed the computer to remain in control, the accident would not have happened. Gilawson (talk) 00:17, 19 May 2010 (UTC)

I think that these improvements would currently best fit in the "Lessons Learned" section. How the TMI2 meltdown effected the nuclear industry already has a section that talks about general trends rather than specific changes. I suspect that there were many improvements made to reactor designs and instrumentation, and I wonder if there should be a separate listing of them, perhaps incorporating major lessons learned from several nuclear reactor accidents on the separate page. Or there could be a section on these added to the List_of_nuclear_accidents. Drcarasco (talk) 08:47, 30 November 2010 (UTC)

" But for the most part, it really wasn't an engineering mistake that caused the accident, but mainly human error." I totally disagree with this statement. You have it backwards. Also: "As one operator from TMI at the time of the accident said that if the operators in the MCR did nothing and allowed the computer to remain in control, the accident would not have happened." This statement sounds like nonsense. Can you source it? TMI was primarily caused by engineering failures, from the PORV failure to the system response. 172.130.45.164 (talk) 04:28, 14 April 2011 (UTC) BG A few years before the TMI accident the Beznau Switzerland reactor had a similar open PORV valve accident, and the Integrated Control System automatically stopped water injection into the system because of pressurizer high water level indication. The Beznau staff quickly identified the problem, but the TMI staff initially followed procedure as they were trained to do and they did what the Integrated Control System would. 172.129.29.244 (talk) 19:49, 14 April 2011 (UTC) BG

There were a lot of people problems associated with TMI:

Reactor operators were not trained to deal with accident conditions, and the NRC had not established effective communication with utilities. Moreover, once the accident occurred, the lines of authority proved to be ill defined. The public received conflicting reports that caused needless panic and evacuations. It was these systemic weaknesses in the regulatory system that allowed gifted people to make the mistakes they did. -- Thomas Wellock, (22 September 2005). "Three Mile Island: A Nuclear Crisis in Historical Perspective (Book review)" The Historian, Vol. 67.

-- Johnfos (talk) 06:33, 14 April 2011 (UTC)

Definitely. Reactor operators were not trained to deal with accident conditions. That says a lot. 172.163.37.92 (talk) 22:37, 16 April 2011 (UTC) BG

Suggestions for additions to the article

An excellant and refreshing article. Thank you for noting the PORV valve had previously failed on 11 occasions. Here are some suggestions for additions to the article:

(A) A few years before the TMI accident the Beznau Switzerland reactor had a similar open PORV valve accident, and the Integrated Control System automatically stopped water injection into the system because of pressurizer high water level indication. The Beznau staff quickly identified the problem, but the TMI staff initially went by procedure as they were trained to do and they did what the Integrated Control System would.

(B) The PORV was probably guaranteed to fail under normal conditions for 2 reasons: (2) Due to the B&W design using a smaller volume pressurizer to save money, the PORV valve would frequently open under routine shutdowns - much more frequently than the original Westinghouse PWR designs. (2) More importantly, due to the pressurizer spray turning on prior to PORV opening, the PORV would have to handle water droplets in the saturated steam. This water droplet/steam mixture is what causes the sound of a freight train, and this mixture will destroy anything, including the convoluted path in a PORV valve. To my knowledge this is still not addressed for PORV valves. A simple steam dryer box prior to the PORV would be a solution.

(C) If steam forms in the reactor vessel, and the pressurizer vents, water is ejected from the pressure vessel because the pressurrizer connection (at the coolant loops) is effectively connected about 1/3 of the way down from the top of the pressure vessel. (Your mom’s pressure cooker had a vent at the top, not 1/3 of the way down on the side.) This is why after the TMI accident vents were connected at the top of the pressure vessel to vent steam or hydrogen in this kind of emergency, so the core can always be covered with water. IF there is enough fresh water, the option even exists to let it boil - venting some radioactive steam into a steam dryer/condensate tank outside the building can save the core. There's no substitute for a reliable 100,000+ gallon water source . The reactor building should have vents that open on over-pressure, and the building sometimes should be vented in an emergency. The most important thing by far is core integrity, as the poor Japanese are about to find out. 172.164.188.98 (talk) 16:38, 13 March 2011 (UTC)Bernie Goetz

A Wiki article on the Japan nuclear accident(s) already started. They can still benefit from the TMI Wiki article although they are boiling water reactors. The piping described above doesn't directly apply because the Japanese reactors are BWRs, but like TMI they still had an unnecessary situation of cores not covered with water. There should have been a convection heat exchanger for decay heat. I can't believe they are flooding the core with seawater, but its an old facility anyway. It might be a blessing in disguise, the Japanese can build modern facilities much better and maybe show others a better way. Steam bubbles, hydrogen, and other things must be handled better. Reactors should not be built in areas with a history of tsunami flooding, like the east coast of Japan. 172.162.139.33 (talk) 21:27, 13 March 2011 (UTC) 172.162.82.165 (talk) 23:54, 13 March 2011 (UTC)BG

Due to the loss of heat removal from the primary loop and the failure of the auxiliary system to activate, the primary side pressure began to increase, triggering the pilot-operated relief valve (PORV) at the top of the pressurizer to open automatically. The PORV should have closed again when the excess pressure had been released and electric power to the solenoid of the pilot was automatically cut, but instead the main relief valve stuck open due to a mechanical fault. The open valve permitted coolant water to escape from the primary system, and was the principal mechanical cause of the crisis that followed.

In this paragraph I miss the note that cutting the elecric power to the solenoid was intended to shut the valve again. —Preceding unsigned comment added by 81.183.131.97 (talk) 13:58, 14 March 2011 (UTC)

Lessons Not Learned...

Because of the production of hydrogen at the Three Mile Island incident, vent was added to the design to vent hydrogen. The venting of hydrogen was responsible for large explosions in the 2011 Japanese nuclear incident. It seems sad that there was no provision made to flare (burn) the hydrogen into much less explosive water, which could have have been condensed and contained. —Preceding unsigned comment added by 203.114.136.217 (talk) 03:00, 15 March 2011 (UTC)

A steam bubble formed in the top of the pressure vessel first. I'm not sure, but I think the added vents were to vent steam first (so the core doesn't become uncovered), and then hydrogen if the core becomes uncovered. I think the main pressure vessel vent should go to steam dryer and then to a vented condensate tank located outside the containment building.172.129.181.52 (talk) 15:42, 9 April 2011 (UTC) BG

Biggest lesson not learned: The operators were fall guys. The official reports stated the fundamental cause of the accident was operator error, not design errors. And possibly PORV failure was not adequately addressed. A PORV opening on a Westinghouse PWR would have been considered a minor accident, but on a B&W reactor it was routine. 172.129.127.212 (talk) 14:17, 15 March 2011 (UTC) BG

Bernie, if you're interested (I read your words as you're connected to the industry), you might want to read Dyatlov's recall of Chernobyl events as well as K-19 crew's (automated translation is bitter but getting plain misinformation is poisonous, and there was deliberate misinformation, "east and west", regarding both of these events). In short, at least several men among those operating the reactors weren't merely "trained" but knew the theory and practice to the degree that only knowing them better than designers (or knowing the particular construction negligences) could have saved them. --Gvy (talk) 12:07, 19 March 2011 (UTC)

Yes, afterwards the operators were somehow expected to understand the plant design better than the TMI designers. The Japanese of course will do a reevaluation of reactor designs after their terrible meltdowns, and they can be expected to do a more complete analysis than was done after TMI. 172.162.121.239 (talk) 14:54, 2 April 2011 (UTC) BG

Human Factors

In the second paragraph of the Human factors – confusion over valve status section, this sentence:

"The design of the PORV indicator light was fundamentally flawed, because it implied that the PORV was shut when it went dark."

Usually, an indicator lamp on a control panel has some sort of annotation to tell observers what the lamp means. Usually this is some sort of text or image that attaches meaning to the lamp -- the indicators on your car's instrument cluster for example. For completeness, this paragraph should include the PORV lamp's annotation. Anyone know what that annotation is (was)? -- Trappist the monk (talk) 18:00, 20 March 2011 (UTC)

You have a point. The article more or less explained the PORV indicator light was simply in parallel with the PORV coil, so the PORV indicator light only indicated power to the PORV coil. It did not directly measure the PORV position. If the PORV stuck in one position, the operators wouldn't know that by looking at the PORV indicator light. I'd have to look it up, but the PORV indicator light basically showed something like PORV CLOSED or PORV OPEN. So the operators looking at it had unreliable information. More can also be said about the unreliable water level indicators. 172.163.121.246 (talk) 04:13, 21 March 2011 (UTC)BG

Ah, the metaphorical light glimmers a bit. Your comment about the indicator lamp being in parallel with the solenoid coil is not part of the article. Similarly unclear is the purpose of the solenoid. I would guess that it is supposed to open the PORV. The WP PORV article makes no mention of the solenoid's function. So that leads me to another question: What is the function of the solenoid on the PORV in the TMI design? -- Trappist the monk (talk) 12:57, 21 March 2011 (UTC)

Yes, the solenoid would open the PORV. This is a decent article but it could be a great article with additions like the one you suggest.

The article should also go more into the steam-bubble/water mixture. It was at TMI that that this mixture and its dramatic consequences were discovered: erroneous water level readings, water ejection, pump cavitation, and a steam bubble in the pressure vessel. The lack of knowledge of this mixture was one of the basic causes of the accident. 172.130.57.210 (talk) 14:56, 21 March 2011 (UTC) BG

I thought that the sentence was clear enough, and did not understand the "clarification needed (see talk)" annotation. But, yes, the addition of the label on the indicator, if it can be discovered, would be a contribution. We assume it would say "valve position", whereas it should have said "value energized", or some such. ( Martin | talk • contribs 17:09, 25 March 2011 (UTC))

For what it says, yeah, the sentence is clear enough. But it left me wondering what the designers did to make the indicator lamp ambiguous enough that the operators could misinterpret its true meaning. These people are, after all, highly trained, intelligent folk. I don't think we can make any assumptions about the lamp's annotation in the same way that we can't make assumptions about the lamp's color - yet another unknown. Nor should we speculate. -- Trappist the monk (talk) 14:23, 26 March 2011 (UTC)

Back then the meaning of a lit bulb next to a control switch on a panel or the meaning of an annunciator window was not always clear. For example, a light may have indicated that a solenoid was energized to open a valve like a PORV, but that doesn't necessarily mean the valve, in fact, opened (it might have been stuck closed). Conversely, the light going out (indicating the solenoid was no longer energized) does not necessarily mean the valve closed if it was stuck open. For additional confidence in the actual state of a valve you would rely on position switches that actuate based on the position of the valve stem (which are still not guarantees since it is possible the valve stem may have separated from the disk). In addition, the control switches on the panels were not necessarily arranged in an intuitive, human-factored way that allowed an operator to tell at a glance whether the necessary valves had repositioned and pumps had started to provide the safety function. Improving human factors in control rooms was yet another outcome of TMI. -- (UTC)Blubbaloo (talk) 18:56, 27 March 2011

Moved Blubbaloo signature to what I hope was the correct place. Before I did this, the Human Factors section had been disappeared. -- Trappist the monk (talk) 19:35, 27 March 2011 (UTC)

Trappist's question DOES bear some merit, as some may not understand the nature of the lamp, due to the valve sticking. The indicator in question only is lit when the solenoid coil is energized, which should open the valve. The solenoid de-energizing would then have a mechanism, typically a spring, that closes the valve. The light is off, but IS the valve truly closed? Under normal conditions, yes. That day, no, the valve failed to close for unknown reasons (at least, I've found no indications of the cause of the failure.) Think of a car with a low oil level indicator light. If the light is on, your engine needs oil. If it's off, NORMALLY, the oil level is acceptable, but if the bulb or sender failed, you'd get no indication and have a seized engine. That is one of the reasons of the lamp test when you start your car, so that you can notice if an indicator bulb failed. In the case of TMI-2, were there flow sensors at the output of the PORV valve, a simple OR detection could be applied or even better, NAND test for a secondary valve failure indicator. But then, one can have sensors to monitor sensors ad-absurdium...Wzrd1 (talk) 17:56, 15 September 2011 (UTC)

Xe-131 is stable

"Noble gases such as xenon 131 (a byproduct of the decay of radioactive iodine) made up the bulk of the release of radioactive materials from TMI-2,.."

The problem with this statement is that while Xe-131 is indeed the result of I-131 decay, Xe-131 is a stable isotope. I believe that radioactive noble gases were released at TMI, but they consisted primarily of Kr-85. The Wikipedia article on Kr-85 says that TMI-2 released about 50,000 Ci of Kr-85, so I'm pretty sure (but not certain) that the text should be changed to say this. Karn (talk) 13:11, 26 March 2011 (UTC)

Correct. The Rogovin report says the release was primarily "'noble' (chemically inert) gases, xenon and krypton" that emit "both beta and gamma rays." Clearly this is not Xenon 131. It also says, "During the course of the accident, approximately 2.5 million curies of radioactive noble gases and 15 curies of radioiodines were released." I'll change the article to quote Rogovin.--RichardMathews (talk) 16:42, 21 April 2011 (UTC)

Missing timeline

• From "sound an alarm at 4:11 a.m." on, we have a quite precise timeline of the events. But when did SCRAM occur in the first place ? At 3:59 ? 4:00 ? 4:01 ? 4:02 ? 4:05 ? Teofilo talk 12:04, 31 March 2011 (UTC)