Report Of The President's Commission On
The Accident At Three Mile Island           > TMI-2 > Kemeny

Quick Links:

Report Index

The Commission

Senior Staff







The Accident


Commission Findings:


 1.  A purpose of the Energy Reorganization Act of 1974 was to divorce the newly created NRC from promotion of nuclear power. According to one of the present NRC commissioners, "I still think it [the NRC] is fundamentally geared to trying to nurture a growing industry." We find that the NRC is so preoccupied with the licensing of plants that it has not given primary consideration to overall safety issues.

2.  NRC labels safety problems that apply to a number of plants as "generic." Once a problem is labeled "generic," the licensing of an individual plant can be completed without resolving the problem. NRC has a history of leaving generic safety problems unresolved for periods of many years -- for example, the problem of anticipated transients without scram. In 1976 during the TMI-2 operating license (OL) review, the Advisory Committee on Reactor Safeguards recommended, as they did in at least one other OL review, that prior to commercial operation further evaluations be done of various possible accidents, including low-probability accidents. NRC staff designated this as a "generic issue." TMI-2 received its OL 2 years later without the resolution of the issue.

3.  Although NRC accumulates an enormous amount of information on the operating experience of plants, there was no systematic method of evaluating these experiences and looking for danger signals of possible generic safety problems. In 1978, the General Accounting Office criticized NRC for this failure, but no corrective action had been taken as of the TMI-2 accident.

4.  The NRC commissioners have largely isolated themselves from the licensing process. Although the commissioners have adopted unnecessarily stringent ex parte rules to preserve their adjudicative impartiality, they have also delegated most of their adjudicative duties to the Atomic Safety and Licensing Appeal Board and actually adjudicate approximately 25 percent of all licensing decisions. That figure is misleadingly high, in part because a number of the decisions do not represent significant adjudicatory events and include decisions on exports. The commissioners have also isolated themselves from the overall management of the NRC.

One of the present NRC commissioners, testifying before Congress,  j said, "There has, I think, been too little Commission involvement in j the setting of safety policy in this agency and too little      J Commission guidance on safety matters to the staff and to the    | board."                               |

5.  The major offices within the NRC operate independently   ' with little evidence of exchange of information or experience. For . example, the fact that operators could be confused due to reliance  : on pressurizer level had been raised at various levels within the  I NRC organization. Yet, the matter "fell between the cracks" and   s never worked its way out of the system prior to the TMI-2 accident, j

6. Licensing of a nuclear plant is a two-step process. First, the company must obtain a construction permit (CP) and several years later must obtain an operating license (OL). The CP stage does not require complete design plans, and therefore the full safety review does not occur until the OL stage. By then, hundreds of millions of dollars have been spent or committed in the construction process. Therefore, the ultimate safety review may be influenced by economic considerations that can lead to a reluctance to order major changes at the OL stage.                           '

7.  The Advisory Committee on Reactor Safeguards (ACRS)    ' reviews all applications for licenses and poses whatever questions it deems appropriate. The ACRS is the only body independent of the NRC staff which regularly reviews safety questions. However, it has established no firm guidelines or procedures, and generally has only monthly meetings of limited duration. ACRS members are part-time and have a very small staff, thus they must rely heavily on the NRC staff for follow-up of their concerns. ACRS members tend to concentrate on their own particular areas of expertise, thereby resulting in a fragmented licensing review.

8.  There are serious inadequacies in the NRC licensing process.

a.  Applicants for licenses are only required to analyze "single-failure" accidents; they are not required to analyze what happens when two systems or components fail independently of each other. The accident at TMI-2 was a multiple-failure accident.

b.  NRC's design safety review places primary emphasis on those items labeled "safety-related." This designation is crucial since items not labeled "safety-related" need not be reviewed in the licensing process, are not required to meet NRC design criteria, need not be testable, do not require redundancy, and are ordinarily not subject to NRC inspection. There are no precise criteria as to which components and systems are to be labeled "safety-related;" the utility makes the initial determination subject to NRC approval. For example, at TMI-2, the PORV was not a "safety-related" item because it had a block valve behind it. On the other hand, the block valve was not "safety-related" because it had a PORV in front of it.

 c.  NRC's reliance upon artificial categories of "safety-related" items has caused it to miss important safety issues and has led the nuclear industry to merely comply with NRC regulations and to equate that compliance with operational safety. Thus, over-emphasis by the NRC process on specific categories of items labeled "safety-related" appears to interfere with the development, throughout the nuclear industry, of a comprehensive safety consciousness, that is, a dynamic day-to-day process for operating safely.

d. There is no identifiable office within NRC responsibile for systems engineering examination of overall plant design and performance, including interactions between major systems.

e.  There is no office within NRC that specifically examines the interface between machines and human beings. There seems to be a persistent assumption that plant safety is assured by engineered equipment, and a concommitant neglect of the human beings who could defeat it if they do not have adequate training, operating procedures, information about plant conditions, and manageable monitors and controls. For example, despite recognition within NRC and various industrial groups that outdated technology in the control room could seriously handicap operators during an accident, NRC continues to license new plants with similarly deficient control rooms. As noted before, problems with the control room contributed to the confusion during the TMI accident. (See also finding A.8.)

f.  The requirement of additional instrumentation to aid in accident diagnosis and control was considered by NRC as early as 1975, but its implementation was delayed by industry opposition as expressed by the Atomic Industrial Forum (AIF). AIF opposition was based on, among other things, the belief that the instrumentation required would cover "Class 9" accidents, and therefore, would extend beyond design-basis requirements. The lack of instrumentation to display in the control room the full range of temperatures from the core thermocouples contributed to the confusion involved in the attempt to rapidly depressurize the primary system on March 28.

g.  It is common to issue operating licenses to plants when there are still "open safety items." When a plant is licensed with many open items, the Division of Operating Reactors, which has the technical expertise to supervise operating plants, may refuse to accept jurisdiction from the Division of Project Management. In effect, the plant then ends up in a regulatory "limbo," receiving insufficient attention from either division. TMI-2 was in this "limbo" at the time of the accident, 13 months after its OL had been issued.

h.  When NRC issues new guidelines concerning safety, there is usually no systematic review, on a plant-by-plant basis, of operating plants and plants under construction for possible "backfitting." For example, Chairman Hendrie explained to a Congressional committee that stricter requirements for on- and off-site emergency plans had not been imposed on any already operating plants because of the need to balance costs against safety. The committee, however, found no significant cost burden in requiring utilities to upgrade and implement emergency plans. Similarly, NRC determined not to backfit the 1975 Standard Review Plan (SRP) to those plants, such as TMI-2, that received construction permits prior to September 1, 1975. According to Roger Mattson, director of the Division of Systems Safety, if individual SRP requirements had been reviewed for possible backfitting, the SRP requirement of diverse containment isolation actuation would probably have been backfitted to plants such as TMI-2. Instead, TMI-2 containment was isolated only when the pressure in the building exceeded 4 pounds per square inch. Thus, containment isolation did not occur until several hours after the start of the accident. However, this delay had little effect on the actual small releases of radioactive material during the accident.

i.  Although decisions of significant public health impact are considered in the licensing process, NRC has no specific mechanism for interactions with public health agencies in the licensing process, other than the U.S. Environmental Protection Agency (EPA) which does review Environmental Impact Statements filed by applicants for CPs and OLs.

9.  The Office of Inspection and Enforcement (I&E) is charged with determining whether licensees are complying with NRC regulations, rules, and licensing conditions. Some serious deficiences in this office are:

a.  A 1978 General Accounting Office report found that I&E inspectors did little independent testing of construction work, relied heavily on the utility's self-evaluation, spent little time observing ongoing construction work, and did not communicate routinely with people who did the actual construction work. Similar problems exist in I&E inspections of operating plants. For example, the principal I&E inspector for TMI-2 completed an inspection shortly before the accident by examining utility records and interviewing plant personnel, but without physically examining any equipment.

b.  A 1978 survey of I&E commissioned by the NRC determined that the majority of inspectors felt their procedures were unclear and lacking in sufficient technical guidance.

c.  Of crucial significance to I&E's system of inspection and enforcement are the Licensee Event Reports (LER) in which utili- ties report and evaluate important incidents. However, both licensees and vendors often have a strong financial disincentive to evaluate and report safety problems that may result in more stringent regulations, at least in part because it is uncertain which entity will ultimately bear the cost of increased safety. I&E makes little effort to systematically review the LERs, has no formal review mechanism for them, and hence, must rely on individuals to remember events and to identify generic concerns.

 d.  I&E inspectors at various times have had difficulties having safety issues that they have raised seriously considered within the office. For example, in 1978 one I&E inspector raised the issue of operator termination of HPI during the September 1977 incident at Davis-Besse. For some 5 months, none of his efforts produced any action. He then took advantage of the "open-door policy" of NRC and went directly to two of the commissioners. These commissioners considered his complaint serious enough to merit further exploration. Unfortunately, this meeting with the commissioners did not take place until one week before the TMI-2 accident.

e.  Early this year, the General Accounting Office concluded that NRC had not made effective use of its authority to assess monetary penalties for significant violations. The report cited cases where I&E consolidated continuing violations into one violation, took too long to impose penalties, and sometimes reduced the penalties to avoid financial hardship for the licensee.

f.  In its investigative report on the TMI-2 accident (NUREG 0600), I&E came to the unequivocal conclusion that if the operators had followed their procedures for loss-of-coolant accidents, there would have been no accident. However, for more than 2 hours on March 28, the operators at TMI did not recognize that they had a loss-of-coolant accident and did not consider the LOCA procedure relevant. In any event, the TMI-2 procedures were inconsistent and misleading in this regard.

10. There is an absence throughout the NRC of any overall system to measure and improve the quality of safety regulations. There are inadequate management and internal quality assurance systems, an inadequate research program, and the absence of any systematic effort to obtain and use the public health-related research of such federal agencies as HEW and EPA.

11. The information and direction issued by NRC to licensees based on operating experience was, at times, fragmented and misleading. For example:

a.  An NRC publication describing the September 1977 Davis-Besse incident made no mention of the fact that operators interrupted HPI. The incident appeared under the heading of "valve malfunction" not "operator error."

b.  In the weeks following the accident, NRC apparently was confused as to what emergency procedures plant operators should follow. Thus, within a short span of time, NRC issued and then either modified or contradicted its post-TMI emergency instructions.

(i) Immediately after the TMI accident, NRC directed operators not to override automatic engineered safety features under any circumstances and to operate high pressure injection without regard for reactor vessel pressure/temperature limits. NRC modified this directive within a short time.

(ii) On April 5, NRC required all licensees operating B&W-designed reactors to revise their procedures so that in the event of HPI initiation with reactor coolant pumps (RCP) operating, at least two RCPs would remain operating. On July 26, NRC took the opposite position and directed licensees to shut down its pumps when HPI initiated. I&E, in its August 1979 report on the TMI accident, stated that the failure of the TMI operators to shut down the RCPs sooner than they did was a potential item of noncompliance.

12. With its present organization, staff, and attitudes, the NRC is unable to fullfill its responsibility for providing an acceptable level of safety for nuclear power plants.