Chapter 4. TMI Event Precursors

The information in this section is taken from the Rogovin Report. (Nope, I didn’t make any of this up).

Dopchie Letters

On April 27, 1971 a concern came in a letter to the AEC from H. Dopchie who was doing work for the Belgium government on reactor issues. The crux of Dopchie’s issue was that he noted for a leak in the vapor (steam) space of a Westinghouse Pressurizer (2-loop PWR at that time) that the Pressurizer water level would rise (not lower); hence neither the Low Level (Pressurizer) Signal nor the associated Safety Injection (SI) Signal would be actuated. Thus for this SBLOCA High Pressure Injection would not be actuated. In these early Westinghouse designs the Safety Injection Signal relied on what is referred to as “coincidence logic”, such that SI actuation was dependent on both low RCS pressure and low Pressurizer level occurring at the same time (in coincidence).  With Dopchie noting in this event the Pressurizer level would rise not fall, he correctly concluded that High Pressure Injection would not be automatically initiated within the time frame assumed in the Safety Analysis. Dopchie basically asked if the AEC had “investigated the consequences” of this event.

On June 25, 1971 Dopchie subsequently sent a letter to AEC noting with his subsequent evaluation he determined with a leak size of 2 inches or smaller, the Operator had 30 minutes to manually actuate High Pressure Injection and for leaks larger than 6 inches the Containment pressure signal would automatically initiate High Pressure Injection. The Rogovin Report contains a lot of subsequent discussion between Dopchie, the AEC, and Westinghouse on this issue.

When I cut through all the crap in the Rogovin Report on the Dopchie Letter, here is what I see. Notwithstanding the obvious 10 CFR Part 21 issue involved on Westinghouse’s part, and including the September 13, 1971 AEC response letter to Dopchie concluding the High Pressure Injection actuation system was OK; I see a significant design failure of the High Pressure Injection actuation system, including all the then operating Westinghouse two loop USA PWRs. The result of this would be a SBLOCA in the steam space of the Pressurizer, resulting in, for discussion at 10 minutes into the event, no High Pressure Injection flow. Sound familiar? Both Davis Besse and TMI and all two loop Westinghouse PWRs would have no High Pressure Injection flow in that time frame. At DB and TMI the Operators turned it off in response to their training and procedures, at two loop Westinghouse PWRs the design failure prevented it from actuating. With respect to the coolant being lost out the SBLOCA does it make a difference? Not really. Further Dopchie noted the Operator had 30 minutes to manually actuate High Pressure Injection based on Containment pressure. Sound familiar? At DB, in response to Containment pressure I isolated the leak at 20 minutes, followed by manual High Pressure Injection initiation within about the 30 minute time frame.

My point is “what is basically different about what I did and what all these people decided was acceptable”, rather than modify the design. Yet I made an “Operator Error”, while the word “error” is never used in connection with all the hands that touched the failure on this issue at that time.

The Beznau Incident

Well folks would you believe the Dopchie Letter postulated incident actually happened on August 20, 1974? It was at the two loop Westinghouse NOK 1 plant in Beznau Switzerland. Short summary:

“The particular incident in question began with the reactor operating at 100% power. A trip of one of the two turbine generators occurred. As a result, the reactor coolant system temperature and pressure increased rapidly and both PORVs opened. One PORV failed to close and a subsequent depressurization of the reactor coolant system occurred. The reactor tripped on low pressure as a result of this depressurization. As pressure continued to decrease, steam formed in the reactor coolant system hot leg and Pressurizer level began to rise. It eventually increased past the 100% point and remained off scale for 3 to 5 minutes. The operators were able to identify that the PORV was open in approximately 2 to 3 minutes and shut the isolation valve (there is no indication of what caused the operators to realize in such a short period of time that the PORV was open). NOTE: Other places in the Report mention the PORVs had a mechanical direct indicating position indication, not an implied indication like DB and TMI. Well…duh! END NOTE After the PORV was shut, the Pressurizer level fell rapidly as the steam bubbles in the reactor coolant system collapsed. Finally, approximately 12 minutes into the incident the Pressurizer level reached the 5% point and high pressure injection was initiated. In this particular design, a coincident initiation was required for high pressure injection actuation. This initiation required both a low reactor coolant system pressure and a low Pressurizer level. Therefore, because the Pressurizer level went off scale high due to void formation in the reactor coolant system, the Pressurizer level did not decrease initially and did not cause high pressure injection to begin until 12 minutes into the incident.

The incident was analyzed by a team from Westinghouse's Brussels, Belgium office and a report prepared. This report was distributed to various individuals in the Westinghouse domestic reactor offices in Pittsburgh, Pa. The analysis indicated that all existing protection systems had performed properly.”  NOTE: Whew, if they hadn’t performed properly that would have required a 10 CRF Part 21 report. I bet the Rogovin Investigation was sure glad they didn’t have to deal with that (wink, wink).

You can read the rest of this section in the Rogovin Report and find a few other stunning revelations. Also see if you can find who made an “error.”

The Michelson Report

It appears this report first appeared as a hand-written draft on September 1, 1977 (~3 weeks before the DB event, 18 months before TMI). This report is well documented and discussed in Rogovin. Carl Michelson was a TVA employee, working on the review of the B&W 205 Fuel Assembly design, and also a consultant to the Advisory Committee on Reactor Safety (ACRS). The crux of the report was that Michelson realized that a break in the Pressurizer steam space of the B&W design would cause the Pressurizer level to go up (not down) and in subsequent versions of the report and questions he posed to B&W, he even stated this would likely cause Operators to (mistakenly) terminate High Pressure Injection flow based on the response of the Pressurizer level. Sound familiar?

The bloody details are in the Report. But I will add that I counted the hands this report touched, just as mentioned in the Rogovin Report. I didn’t count anybody twice. My total including organizations directly exposed is the ACRS, the Pebble Springs License Application, the Bellefonte plant review, 2 NRC personnel, and 6 B&W personnel, all before the TMI accident. Would anyone care to venture a guess just how many times the word “error” is used in the Rogovin Report in describing this known ERROR in the B&W plant Operator training program and procedure guidance?