Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
s The Personnel Subsystem It is essential to ensure that suitably trained people are avail- able at all levels of plant maintenance, control, and management. Without adequate training, people cannot perform effectively and efficiently. Research on the personnel subsystem is concerned with the design of jobs and the development of systems to ensure that the people assigned to those jobs are sufficient in number and have the requisite knowledge, skills, abilities, aptitudes, and attitudes to perform them effectively. We discuss these research issues under the topics of training, qualifications, and staffing. Although there has been considerable regulatory activity on nuclear plant staffing and on the training and qualification of personnel, there has been almost no research to measure the ef- fectiveness of these activities and little to give direction to such activities in the future. Research on personnel selection, training, and qualifications of personnel does exist in other industries, most notably in commercial and military aviation. Accordingly, an ac- tive program to follow such research and, in some instances, to support it is recommended. Research topics that the NRC should follow and foster include: (1) the general use of part-task sim- ulators; (2) the effectiveness of various training regimens in skill and knowledge development; (3) the identification of techniques to manage stress in operating personnel; (4) the use of psychophys- iological profiles in screening power plant personnel; (5) factors that affect team performance; and (6) the selection of personnel for problem-solving skills. 58
59 There are also a number of areas related to personnel selec- tion, training, and plant staffing that are specific to the nuclear industry. Such topics include: (1) the effectiveness of part-task simulators as training tools for tasks unique to the nuclear m~us- try; (2) the use of replicate simulators as an operator training too! and the role of fidelity, task selection, and performance mea- surement in establishing the effectiveness of fur-scope simulator training; (3) the relative importance of skill-based training ver- sus knowledge-based training in the development of nuclear plant operators; (4) further development of selection tests for nuclear plant operators and the correlation of the selection criteria for operator trainees with operator performance; (5) determination of the practicality and benefits of licensing nuclear operators as teams, rather than as individuals; (6) the effectiveness of periodic retraining for skill maintenance, especially for rarely performed activities; and (7) qualifications of the regulatory staff and their subcontract personnel engaged in measuring the qualifications of nuclear plant operators. TRAINING Measuring Training Effectiveness Rationale and Background The most fundamental change in nuclear power plant train- ing resulting from the Three Mile Island accident has been the adoption of a systems approach to training. The industry, led by the Institute of Nuclear Power Operation (INPO) and encouraged by the NRC, has adapted the Instructional Systems Development (ISD) process developed by the U.S. military and is implement- ing it through INPO's accreditation program. While the research effort required has been small because of the existing data base available from the military, there is now a need for industry-specific research to develop the technology base for further improvements in existing systems and the development of future training sys- tems. For example, one recurring issue in technical training is the effectiveness of part-task simulators. There is general agreement that there can be effective training on part-task simulators, and a general conclusion has been that a mix of part-task and whole-task simulators, in addition to other
60 training media, are required for training on complex systems. But questions remain: What is the proper mix, i.e., what tasks are best trained by one medium versus another? What is the best way to integrate part-task and whole-task training to facilitate positive transfer and rn~nim~ze cost? Are part-task simulators adequate for licensing, or does competence on individual tasks have to be demonstrated within the context of a complete realistic environ- ment on a full-scale sunulator? These are examples of issues that are task- and environment-dependent and require industry-specific research. Additional areas, such as the measurement of training effec- tiveness, the measurement of trainee performance, feedback to trainees, skill retention and decay with respect to retraining re- quirements, require research in the short term for existing systems. Embedded training is currently receiving considerable attention in the training literature and should be explored in the nuclear power plant context. Basic issues of principle versus procedural training and academic versus skill training also require industry-specific research. Nuclear plants are among the most complex man-made sys- tems, and training is a critical element contributing to system safety and availability. If the industry is intent on excellence, it - should begin now to move to the forefront of training research and development. Research Recommendatiom Research on the measurement of training effectiveness should be one of the first areas of attention. Training research in the nuclear industry is very sparse. At the least, a modest program should be initiated for near-term developments for existing plants. In the long term, training should become a significant aspect of the nuclear industry's human factors program, whether the research is conducted by the NRC or the industry. Some training issues can be addressed by small research teams in an academic environment; others may require training simula- tors or in-plant studies. Field research usually cannot progress beyond a limited point without longitudinal studies. For this reason alone, cooperation with operating utilities is essential. In addition, the economic incentives for training research are proba- bly as great as, and perhaps more readily demonstrated than, the
61 safety incentives. Figures given in DOE EP-0096 1983 show very large econorn~c gains from well-designed training schemes. Independent NRC research In the closely related area of oper- ator licensing examinations (testing) appears to be warranted and appropriate; there should be strong interactions between the two programs. New Training Approathes Rationale and Background Raining currently focuses on whole-scope tasks with high face validity an entire emergency incident or an entire plant startup in an environment with high physical fidelity using a full-scale plant- referenced simulator. Training on component tasks, especially the most difficult or critical ones, is given less emphasis. Although the industry, led by INPO, has adopted an aggres- sive systems approach to training, the work to date has focused entirely on the application of training technology and not on more fundamental questions of the nature of skill acquisition, main- tenance, and decay. Research in these areas has the potential to improve the performance of nuclear power plant staff dramatically. New developments in training technology are increasing the potential for alternative approaches that emphasize concept train- ing (HolIan, Hutchins, and Weitzmann, 1984), exploration train- ing for procedural and controls skills (Woolf et al., 1986), and decision-making training for cognitive skills (e.g., Chipman, Segal, and Glaser, 1985~. Developing and applying this knowledge to the nuclear power plant context provides the opportunity to achieve substantial performance unprovements, especially for rarely per- formed or difficult tasks and maneuvers. These new approaches to training may radically affect the methods employed to test operators for certification or recertification, because training and assessment are intimately connected. Research Recommendations Research is required on at least the following topics: The potential of exploration training, which can provide the basis for more performance-orientec! testing of person- nel;
62 Special training for rare events, including those beyond the design basis; Skill and theoretical training on difficult but infrequent maneuvers; Problem-solving training, especially in the context of man- aging problems involving many people and facilities. This research should be coordinated with current EPR! pro- grams. There Is already a considerable research base in each of these areas, which, if properly used, would provide a cost-effective start to develop new innovations. Significant improvements in operator skills might be obtained by more frequent simulator exercises, but it is not clear that the exercises normally chosen for simulator training are effective in developing operator skills to cope with the unusual. Finally, it is not clear what impact the physical fidelity of a simulator has on training effectiveness. Some plants train operators on simulators built for other plants. These simulators may have different dy- nam~cs and control board arrangements than the plant at which the operators work. In addition to acquiring and maintaining skills for specific -emergency scenarios, plant personnel must master the skills re- quired for dealing with novel, complex, and ambiguous events. Research is required to understand how best to obtain these skills. Similarly,-the social context in which performance takes place must be understood. For example, team training (e.g., NUREG/CR- 4258, 1985b) is an issue that has great potential for safety improve- ments in nuclear power plant operations (see Foushee, 1984, for a description of the effect of team factors on safety in commercial aviation). The key is to develop a research program that deals with the types of skills required in the nuclear power plant context and that feeds into the systems approach to training that has been widely adopted by industry. Classes of independent variables of importance include train- ing equipment (e.g., fulI-scale simulators versus part-task simu- lators), training contexts (e.g., individual versus team), training techniques, and personal characteristics of the trainees (e.g., ed- ucation, experience, stress tolerance). Tests should include both familiar and novel situations. The question of verisimilitude of the test bed with actual plant events is itself a question worthy of
63 research; training on unlikely sequences of events might prove to be superior preparation for emergencies than training on known accident scenarios. A variety of dependent measures must be ex- plored (e.g., "objective" performance, performance ratings) and measured over various time intervab to explore skill decay. This research will have direct implications and provide a base for a variety of issues that face the NRC. For example, ques- tions of academic education versus skill training, the frequency and content of requalification exams, the role of team training and qualifications In the licensing process, and the role of plant personnel in severe accident mitigation would all benefit from the proposed program of research. A training research program should be a central and ongoing element in human factors research. QUALI}lICATIONS Degree Requirements Rationale and Background This section is concerned only with whether control room staff should have engineering degrees because of the immediate relevance of this topic to the NRC. Many other important re- search issues related to personnel qualifications are not discussed. One example involves the proficiency and qualifications of main- tenance personnel; each plant has relatively large numbers of me- chanics, electricians, and instrument-and-control technicians that perform thousands of different maintenance tasks of various levels of difficulty at widely differing intervals of time under stressful environmental conditions. One question for which there Is currently no answer has been the subject of intense discussion and disagreement following the events at Three Mile Island: Should all control room staff, or at least senior control room operators, have engineering degrees? Intuitively it seems obvious that more knowledge must help an operator, particularly when dealing with a situation for which the operating procedures are inadequate or for which there has been no training. There will always be more ways for a system to fait than skills for which there is time to train. In some cases it may be necessary for an operator to draw on general engineering knowledge and knowledge of nuclear physics, in addition to plant
64 systems knowledge, to diagnose the state of the plant and decide on appropriate actions. It has been suggested by some within the NRC that degree qualifications in the relevant disciplines im- prove an operator's ability, and this position has led, on several occasions, to the statement that degree qualifications should be mandatory. It is, in fact, difficult to determine the appropriate degree qual- ifications for a job in a high-technology system. Several countries require higher academic qualifications than does the United States (OCED-NEA, 1985), but other characteristics, such as national character, organizational structure, physical plant differences, and training make it impossible to conclude that this requirement is a valid one for the United States. In France, for example, the safety engineer (equivalent to the shift technical advisor) not only has de- gree qualifications, but also has responsibility for safety decisions during emergencies. In Canada, training to be fully qualified takes seven years and requires higher qualifications than In the United States, but reactor designs are quite different and are more au- tomated than U.S. plants. Thus, international comparisons alone cannot be used to decide the question. A few laboratory studies exist on the effectiveness of theoret- ical knowledge on fault diagnosis (e.g., Duncan, 1981; Shepherd et al., 1977i Surgenor, 1981~. None has found such knowledge to be beneficial. However, this finding is inconclusive because it has not been made clear what is meant by theoretical knowl- edge. The theoretical knowledge needed by operators is not the same as that needed for engineering analysis (Brown, Moran, and Williams, 1982; Roth, Bennett, and Woods, in press). Traditional engineering curricula do not provide instruction on fault diagno- sis. However, it does not necessarily follow that education at the engineering level is of no value or merit for plant operations. There is widespread opposition to the idea of degree require- ments among operators in the United States (see, e.g., Professional Reactor Operator Society, 1987~. It would be easy to dismiss this opposition as the natural objection of qualified, experienced op- erators to being required to go back to school and being faced with a threat to their careers, but that would be less than fair to the operators. Some have taken degrees after being qualified, and a number of them have paid for the degree themselves and were not given paid time off to earn them. In short, many operators with degrees were penalized rather than rewarded for their efforts.
65 Furthermore, several have reported that the courses they took were largely irrelevant to their needs (Professional Reactor Op- erator Society, 1987), a position supported by NUREG/CR-4051 (1985c). Research Recommendations There are two levels to the degree question. First, from the perspective of the human-system interface, what is skill and ex- pertise? Second, from the organizational perspective, where do people come fro~what are their career paths? If the organiza- tional climate changes, it will affect the organizational perspective of what degrees are required. Current degree programs do not couple theoretical knowledge to plan~specific practice and as such are not likely to enhance operational skills (Moray, 1986~. But it still seems plausible to assume that better understanding of a plant, even at a theoretical level, would provide better diagnostic and control skills. Before the decision is made to make a degree mandatory, research is required to establish the necessary course content to ensure that such courses are available and to ensure that people are rewarded, or at least not penalized, for taking them. Such re- search is methodologically difficult, and the design of appropriate courses may be costly. There is also the problem of locating insti- tutions able and willing to provide such courses. This research is urgent and should have higher priority. An injudicious regulation could lead to problems with both morale and recruiting without necessarily improving safety. Testing and Licensing Rationale and Background The NRC licensing and requalification examinations adm~n- istered to reactor operators and senior operators represent an important method for ensuring that these personnel can operate the plant in a manner consistent with public health and safety. Al- though the NRC has taken several steps to ensure that questions on the examination are related to the knowledge an operator must master (e.g., NUREG-1021, 1983e; NUREG-1122,1985f; NUREG- 1123, 1986a), research is still required to improve the examination
66 and the process of its administration. Specifically, questions have been raised by people close to the examination process as to the qualifications of the examiner, the interval between requalification examinations, the feasibility of licensing crews of operators, and the validation of and general improvement in the jo~relatedness of examinations. Research that delineates the questions germane to the ex~nination process and gives them priorities will help the NRC make better decisions as to whom to entrust with the operational safety of nuclear power plants. There is a legal requirement that control room and senior con- trol room operators must be licensed by the NRC before they are allowed to handle the controls in a nuclear power plant control room. The NRC administers a licensing examination comprised of written questions and simulator exercises (if a replicate simula- tor exists) to ensure that candidates have the requisite knowledge and ability. This procedure is based on the assumption that sat- isfactory measures of performance are being used. At present, however, there is no basis on which to effectively determine satin factory measures. Research is therefore needed in this area and is discussed in the section on human performance. Fundamental to any examination are issues of its validity, re- liability, and utility. Although the NRC has moved to a program in which the examination content includes items judged to be important to safety, this should be viewed as only a first step. A research program that evaluates the strengths and weaknesses of the current examination and results in recommended improve- ments is required. There exists extensive examination technology (e.g., Anastasi, 1985) demonstrating that improved tests in terms of validity and utility result in improved performance by operas tars. This technology should be addressed as a first step in research planning. Shikiar, Saari, and Wood (1984), NUREG-1122 (1985f), and NUREG-1123 (1986a) form the basis for a content sampling ap- proach to licensing examinations. In particular, Shikiar, Saari, and Wood (1984), which forms the basis for NUREG-1021 (1983e), de- lineates an approach to be used in constructing written, simulator, and oral questions, as well as guidelines for grading responses to the questions. However, many of these research issues have yet to be addressed in a systematic way.
67 Research Recommendatiom Improvements ~ the validity, reliability, and utility of licens- ing examinations depend on better understanding of the issues dis- cussed above In the sections on training effectiveness, new training approaches, and human performance. Developments in these areas should be applied to develop and test improvements in licensing examinations. The result should be an unproved examination pro- cess and an increase in the confidence of the industry and the public in the Incensing process. STAFFING Shift Technical Advisor A relatively new plant position ~ that of the shift technical advisor (STA). It is clear both from informal and formal reports (NUREG/CR-4280, 19853) that research is needed on how to improve the effectiveness of this position. The contrast between U.S. practice, in which the STA is a purveyor of expertise if and when it is requested, and the French practice, in which a person with those skins is seen as the one responsible for making and implementing decisions when an emergency occurs, is striking. The relation between the control room operator hierarchy and its responsibilities and emergency management has recently been examined by the Atomic Energy Control Board of Canada (Feher et al., 1987a, 1987b). This work should be reviewed as a way of enhancing an understanding of issues in this area. If the knowledge and skills of the STA are essential to safety, then research is needed on how to improve that role and its relation to other members of the control room crew. Screening and Selection A classic topic in this area is the set of dimensions on which to base the selection of personnel. If one is thinking in terms of fault diagnosis and fault management, then selection for problem- solving skills would be the pertinent dimension. If one thinks in terms of the ability to remain alert during prolonged periods of normal operation when little intervention by the operator is re- quired, then different personality characteristics may be relevant. Research is also needed on the selection of personnel other than
68 operators, including both maintenance personnel and managers. Whatever the result, it is essential to realize that selection alone cannot guarantee effectiveness; it must be related to training, to workplace management and morale, to recruiting and retention, to job descriptions, and to other personnel issues and seen as part of a systems approach. A well-established technology is available for selecting people who can perform jobs well or who can be efficiently trained to do so on the basis of their abilities and aptitudes (Dunnette, 1976~. The NRC's plan (NUREG-0985, Rev. 1, 1984g) emphasized the need for selection research; however, for reasons of economy, this emphasis was later abandoned (NUREG-0985, Rev. 2, 1986f). Ac- cording to SECY-85-129 (1985k), the Edison Electric Institute is currently researching selection with nuclear power plant personnel. Another aspect of selection is to identify people with an apti- tude for problem solving under risk and time stress. Some research on this topic has been done in aircraft pilot selection (e.g., Go- pher, 1982~. It is highly desirable to have screening instruments and procedures that identify psychologically high-risk individuab for industries like nuclear power. However, the practicality of se- lection or screening on the basis of personality variables remains highly controversial. Cough (1976) concluded that research tend- - ing to support the construct validity of personality measures would be adequate to justify their use in industrial selection. McCormick and Ilgen (1985, Chapter 10), suggested that a rather rigorous set of requirements should be met before personality measures are used for this purpose. "Standards for Educational and Psycho- logical Testings (American Psychological Association, 1985) also supports a cautious approach to such use. Shift Scheduling and Vigilance Rationale and Background The recent cases of operators found sleeping ~ a control room illustrate the relevance of the issues of shift scheduling and vigi- lance (or sustained attention) to nuclear power plants. Poor work schedules can lead to stress, fatigue, and other undesirable side effects (Hockey, 1983; Eastman Kodak, 1986~. Boring tasks and inappropriate shift structure can produce loss of alertness, which
69 can lead to serious industrial accidents (Comrn~ssion of Inquiry Hinton Train Collision, 1986~. . Large bodies of research exist on each of these topics (e.g., Davies and Parasuraman, 1982; Parasuraman, 1984; Co~quohoun and Rutenfranz, 1980~. Industrial accident data indicate large changes in efficiency as a function of time of day. Poor work schedules can increase the probability of human error. The shift duration, the total amount of work per week, and the pattern of work rotation are all critical factors in determining the amount and quality of human performance under different work sched- ules. A variety of shift schedules are used in the nuclear industry (NUREG/CR-4248, 19851), including some that are known to cause high fatigue and poor performance. For example, two fac- tors that, in general, degrade performance are day-night-afternoon shift rotations and weekly rotation schedules (Eastman Kodak, 1986~. Ever since research was conducted on the behavior of radar and sonar operators in World War IT, it has been known that people are often unable to maintain a vigilant attitude when monitoring systems for long periods during which few significant events occur. While the phenomenon is generally well understood (Hockey, 1984; Warm and Parasuraman, 1987), methods to prevent it are often not easy to devise. The central problem is how to keep operators and other personnel alert. There is a tendency to look for hardware solutions to this problem such as auditory alarms and acknowI- edgment systems. For example, on Japanese long-distance trains, the engineer must respond to an alerting signal every few min- utes or the train is automatically halted. However, experimental evidence suggests responses to signals can become automatic and do not always ensure vigilance (Oswald, 1962~. There are many factors that contribute to a loss of vigilance, including fatigue, in- appropriate shift schedules, and the effects of circadian biological rhythms. Treating causes of losses of vigilance may prove to be more effective than treating the symptoms. There are also other factors that condition how effective an alerting device is or how disruptive fatigue is. If morale is high and a strong desire to excel exists, the operator is likely to carefully monitor the state of the plant even under difficult circumstances. If morale is low, he is likely to do the minimum needed to acknowI- edge an alerting signal without actually checking for changes in
70 plant state. Such an attitude ~ as likely to be due to poor manage- ment style and practice as it is to any inherent lack of motivation in plant personnel. This is an example of why research is needed at all levels shown in Figure 1. Research Recommendatiom Because large amounts of research on both shift scheduling and vigilance exist, applied research in the nuclear industry should consist of synthesizing this knowledge base for application to power plants and developing and testing possible treatments (informed innovation). A model for this process can be found in EPRI's project on heat stress management (EPRT, 1986c, in press).