ML19308C318
| ML19308C318 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 03/28/1979 |
| From: | Seminara J LOCKHEED MISSILES & SPACE CO., INC. |
| To: | Chipman G NRC - NRC THREE MILE ISLAND TASK FORCE |
| References | |
| TASK-TF, TASK-TMR NUDOCS 8001220711 | |
| Download: ML19308C318 (8) | |
Text
.
J. L. S; min:ra CO*MENTS ON THE PEELIMINAJY DRAFT OF THE SIG STAFF REPORT ON HUMAN FACTORS EVALUATIO!! RELATED TO THE YHREE MIIE ISLAND ACCIDENT Many of the comments made and submitteo rith regard to the ESSEX first draft apply to this SIG draft and will not be repeated here. Most of my remarks deal with the Findings / Conclusions & Recommendations Section.
First a word about organization. It seems to me that a one page crisp summary or abstract is needed as the first page of the report. This sunnary would state the purpose of the investigation, the four or five major findings and conclusions and the major mcommendations. If you had this overview up front, then you could structure the report in a more traditional manner with Section 2 at the end of the report instead of in the beginning.
Some of the major findings in my view include o The TMI-2 accident can be attributed to an insufficient mgard for the human factors aspects of control room design, procedures, training, and other factors that inpact the quality of the man-machine interface.
O The IMI-2 control room was designed in accordance with prevailing HF standards and guidelines in the nuclear industry but fell far short of the human factors state-of-the-art as applied to military and aerospace systems where the consequences of degraded operator performance cannot be tolerated.
o The human f actors problems in control room design, training, selection, procedures, etc. found at IMI-2 are not unique to this plant but can be also observed in varying degrees at other operational or near-operational nuclear power plants.
o There has been a systematic disregard for human factors engineering methods, data, design principles and evaluation criteria by the NRC and the various components of the nuclear power industry.
o Since the control room and operator training programs are not designed to minimize the potential for operator error, we cannot attribute the TMI-2 accident to f aulty performance on the part of the control room operators or their supervisors.
8 o 01220 7/ l
RECOt'MENDATIONS 1.
The NRC should develop a human factors engineering capability comparable to that available in military / aerospace system developments. This Human Factors organization should be placed at a sufficiently high organizational level to ref]e et the importance of human factors in the light of TMI-2 events. This orgahization should be charged with the msponsibility for raising the awareness level.for human factors aspects of control room operations first within the NRC and secondly within the industry at large. The NRC Human Factors organization should be initiated by recruiting a nationally recognized human factors expert from the military / space community. (Let's avoid the natural temptation to select one of the old gang and call him the human factors expert.) It has to be someone from the outside.
The functions of this HF organization would include:
o Development of a program of research to develop a human factors data base that would be applied in resolving basic human factors issues und9r the purvue of the NRC. This research would be conducted by both in-house human factors experta with help from outside human factors organizations as needed.
o Development of an applied HF pmgram based on available human factors research data. The Applied HF effort would include the development of evaluation criteria for control rooms, training programs, selection pmgrams, procedures, organizational interfaces, etc. Such criteria would be applied in the licensing process, relicensing, operator licensing, inspections, LER reviews, etc. The existance of such criteria would promote industry standards that utilities and vendors would follow as prerequisites to licensing.
o Maintain contact with and act in cooperation with on-going research efforts conducted by the EPRI, INPO, IEE, vendor research efforts, etc.
o Interact with other NRC organizations to ensum that NRC guidelines and practices take full cognizance of human factors concerns, e.g.,
in developing separation criteria have all human factors issues been considered in the trade-off process?
Promote the development of human factors technology in the utilities o
and the vendors that serve the industry.
o Extend the present comoern for power plant operations to the maintain-enance area, quality assurance, health ph,tsics, rad waste handling and other areas of plant operations where human factors considerations can degrade plant safety, efficiency and public saftey
2.
Under the leadership of the NRC Human Factors Organization, a program of control room enhancement should be initiated. This program should have near-term and long-term goals. Much can be done to upgrade control rooms without interrupting power generation. Ionger term measures might be accomplished during periodically scheduled extended outages.
Enhancement approaches should also be developed for the ninety or so control rooms that are not operational and in varying stages of developmalt.
More extensive upgrading opportunities may be available with these control rooms that are still in the formative stages.
3 Human factors guidelines need to be developed for future control rooms.
The question of standardization has been a controversial one over the years. In the wake of TMI, this issue should be reexamined. Standardiz-ation should be considered for both the control room and the plant design as a whole. In military and aerospace system developments, complex equip-ment systems are developed a generation at a time with many copies of one standard design. For example there are X number of submarines of the 616 class. As soon as the design of one generation of subnarines is completed, design work commences on the next generation.
The design process can be more systematic and reflect the systems engineering approach when it covers the design of many copies of a system as opposed to a separate design effort for each individual submarine. The standard nuclear plant would also shorten the span of time between concept and operational implementation.
h.
There are fundamental problems with training programs for power plant pe rsonnel. First there is no good definition of the operator's role. No formal job analyses have been conducted to establish the skills, knowledge, personality characteristics and other attributes which operator candidates should have. Consequently, trainee selection techniques may be imprecise or invalid. Similary, training programs cannot be designed to meet specific
)
performance criteria.
The NRC licensing examination process is not viewed by the overwhelming majority of operators as a measure of future operational i
success on the job. A majority of trainers concur in this view. Training programs do not measure up to the training state-of-the-art in other contexts where the operator's role can jeopardize public safety.
1 5.
While insufficient attention has been paid to the human factors aspects of operator training, selection and control room design, the situation is even worse for non-operational aspects of the plant. There has been j
almost a total disregard for human factors aspects of plant maintenance activities, health physics protection, quality assurance, rad waste j
handling, etc. The concern for human factors aspects of plant operations advocated above should also be extended to other facets of power plant activities.
6.
Additional research is required to determine the most effective structure of the operator's job. The operator's job can be described as one.which is largely filled with boredom with occasional moments of sheer terror.
Many operator's seek other forms of' work to avoid the stress of shift work. The operator is confined to the control room for long stretchs
with no opportunity' to feel the pulse of the plant. Many belong to unions i
and an adversary relationship with management is sometimes encountered.
There is a concern that operators will strike and leave the plant operations
.L to licensed managencnt personnel who are relatively inexperienced. There are many indications that the operator's role and status in the plant could bear further scrutiny and possible reformulation.
i
- )
I l
l i
t
)
k
(,
r L
DETAILED C0t?'ENTS: SIG PRELIMINAR DRAFT PAGE PARA COMMENT h-3
..., approximately 2,000 represented by the Human Factors Society 7
2 Can all selection criteria be expressed quantitatively?
12 1
Few control rooms reveal an organized effort to integrate the needs of the man in the machine.
12 2
I don't consider CC an example of a "worka ble" design process... not if a " workable" design process results in mirror-imaging, back panels, annunciator systems that overload the operator, etc.
15 1
Viewed in the light of the industry-wide development
- process, 15 2
It may not be necessary for training personnel to be involved with procedures. It may be better to get operators in the loop.
16 2
We might say a word about establishing crew size on the basis of analyses of workload associated with normal and off-normal situations. If we, as experts, say that the manning appeared OK then we can be accused of being as non-rigorous as those we are judging.
16 3
This paragraph supports the view that a thorough job analysis is needed to define required operator skills and knowledge.
I find it difficult to believe thelast sentence.
18 1
Most training harps on emergency conditions, at least in simulators. Why presume the premise made? I have never heard anyone voice this sentiment. I think the statement is unfair and should be backed up with evidence. I think that most training deficiencies that exist are based on a lack of sufficient research in this area and a general ignorance of the state-of-the art in training.
18 2
Training at TMI-2 was not especially deficient....
I know of no operator training program which provides measures of performance beyond very subjective observations and comments by instructors. EPRI is sponsoring a study to develop a performance measure-ment system for operator performance.
18 3
Add: This is common practice in the industry 18 h
Self-study is also not unusual in the industry.
PAGE PARA.
COMPENT 20 2
The last sentence is far too strong. TMI-2 was more than a training disaster. The CR didn't help much.
If serious accidents are bound to happen based on the training observed, we can expect same all over the country.
25 2
I guess I have a philosophical disagreement with the writer here. I see no excuse for poorly designed panels whether for the systems manager role or the maintenance role. Costly errors could and are made in the Maintenance function. Granted, the panels must definitely be well-human engineered for systems management in a crisie. However, let's not let the designer off the hook for other panels. If nothing else, well human engineered panels facilitate training.
26 1
We should add a couple of sentences describing the nature of the fixes that will correct the situation.
Sometimes backfits, if not well thought out, will introduce new operational problems.
27 1
You might have had the decency to spell out the references to the EPRI studies a little better so that the reader might obtain copies and increase my royalties.
Why don't you reference Fred Finlayson's work? Why piss off two members of the peer review group?
27 2
There is also a need for a definitive study /ies to establish the infonnational needs of operators based on functions and task analyses.
27 2.2 2.
statement far too weakJ Military and aerospace experience has shown that human engineering aspects of a system get reasonable attention when a systems engineering approach is applied and all participants in the design, procurement, and evaluation process understand human factors goals, methods, standards and evaluation procedures.
l 28 h
Establishment.... The first sentence is misleading.
l
[
While the establishment.of a specific group in the NRC to review and evaluate human factors engineer-ing is essential, it is not the entire answer to the problem. The NRC must promote the acceptance of human factors engineering within the industry. -
Human factors engineering must...
30 1
Omit last sentence. I have not seen designs that are far superior.
l In this summary we should hit the big stuff e l
30 h.
informationoverloadfromtoomanyannuncia$ ors.g.,
rather than panel bulbs difficult to change.
7
\\
PAGE PARA.
CO? TENT 30 B.
1.
Don't buy this statement. It isn't substantiated and probably can't be to anyone's satisfaction 31 2.
Is it necessary to mention which plants we visited for comparative purposes?
The quality of CC and Oconee HF is not high enough for me to single them out as paragons of virtue.
If you really looked at Oconee in depth you might decide TMI-2 is better.
31 5.
Same comment as for 2.
32 h
The mirrored scales at CC don't do you much good when the 75 me6ers are placed above proper limits.
The operator has to haul out a fancy ladder to avoid parallax. Who wants to go to that trouble? They don't.
Again, it looks silly to point out how good CC is compared to TMI-2 when we see that 75 meters deviate from anthropometric limits.
33 1
Should each utility do a task analysis to identify i
"saftey" instruments? Should one task analysis per major system vendor be done? Are there any shortcuts short of seventy analyses?
Maybe the NRC should sponsor the analyses?
What is a state-of-the-art supervisor's console?
Perhaps we shouldn't be specific here. How about:
t One the greatest shotcomings in the design of present generation control rooms appears to be a lack of appropriate diagnostic aids to allow the operator to assess the nature of disturbances and proceed unerringly to the correct conclusions and associated course of corrective action. Research is needed to determine the best method /s for augmenting oper-ational control rooms to provide additional diag-nostic capabilities.
3h h.
L Say something about negative transfer of training when training on a simulator that differs from the operational control room design.
I
_