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.
3 Elements of a Future Agenda for Nuclear Safety Research In this chapter the committee identifies some important ques- tions of reactor safety that should be researched. The principles introduced in the preceding chapter, where we looked at nuclear safety research in terms of several broad categories, are applied to each of these topics. The perspective here is narrower; the focus is on specific research topics that the committee believes should constitute elements of the nation's future agenda for nuclear safety research. The list of specific issues that the committee has assembled is presented in Table 3. The table presents in alphabetical order a selection of important topics requiring research. The committee was asked to review a draft report, "NRC Safety Research Pro- gram," by the staff of the NRG, which outlines a small number of research topics they wish to pursueâcomponent aging; thermal hydraulic transients; severe accidents; plant operations; seismic analysis; and waste disposalâand a seventh (advanced reactors) by implication. All of those topics, except regulatory waste man- agement research, are discussed below, along with others that the committee finds important. It is appropriate to reiterate that the focus here is exclusively on reactor safety research and, within that sphere, almost entirely on research in support of current reactors. Other aspects of the nuclear fuel cycle, in particular radioactive waste management, and other important areas of research, such as radiobiology and the health effects of ionizing radiation, are beyond the scope of this report. 25
26 TABLE 3 Elements of a Future Agenda of Nuclear Safety Research Current Plant Designs Behavior of Materials in Nuclear Power Plant Environments D ecommissioning Extended Fuel Cycle Human Factors, Instrumentation and Control, and Operations Nondestructive Testing and Examination Plant Aging (License Extension) Policy Research Quality Assurance/Quality Control Reevaluating Existing Regulations Reliability of Plant Components and Plant Systems Safety Analysis Methodology and Application Seismology, Soil Mechanics, and Structural Response to Seismic Events Severe Accidents Thermal Hydraulics Future Plant Designs The discussion below is intended to point out that, whether or not new reactors are built, outstanding issues remain to be considered and significant research remains to be done if only to provide adequate support for the safety of current reactors. In some areas not enough research is being performed. Such areas include the behavior of materials in nuclear power plant environ- ments, decommissioning, human factors, nondestructive testing, plant aging, policy, and reevaluation of existing regulations. In other areas one or more of the principal sponsors has a sizable program of research under way, and the primary need is to re- focus the work. That category includes research on component and system reliability, nondestructive examination, quality assur- ance/quality control, safety analysis methodology and application, and severe accidents. CURRENT PLANT DESIGNS Behavior of Materials in Nuclear Power Plant Environments Many of the research topics discussed elsewhere in this chap- ter necessitate basic, exploratory, long-term research on materials used in the construction of nuclear power plant structures, sys- tems, and components. Topics such as plant aging, component and system reliability, nondestructive testing and examination,
27 quality assurance/quality control, decommissioning, and research to reevaluate existing regulations, as well as the broad field of advanced reactor R&D, all require basic materials research. Ma- terials science is a highly complex, enormously important field incorporating a number of scientific disciplines. It can and should be developed as a basis for future investigations of such issues as new fabrication processes for use during plant construction; cor- rosion and cracking of piping during normal operations; effects of pressure and temperature gradients and radiation extremes on piping and equipment; inspection frequency and inspection tech- niques; repair and maintenance procedures; occupational health and protection against radiation; and many other topics of rele- vance to plant safety, reliability, and maintainability. Basic and applied research on materials properties and the performance of materials in the different environments found in nuclear power plants are essential in providing a means of confirming the ad- equacy of existing operating parameters, inspection techniques, and operating and maintenance procedures. This is research that industry, the NRC, and DOE all need, whether for shared or different purposes. Not surprisingly, the work is now performed largely on a project-by-project basis. Some basic materials science and engineering research should be under- taken as part of a cooperatively funded program involving indus- try, DOE, and the NRC. Decommissioning Although only a few of the smaller research and demonstra- tion reactors face immediate decommissioning, all the larger com- mercial reactors will eventually be candidates. Decommissioning will require the development of a data base and a methodology, including advanced measurement techniques for determining the quantity and activity of induced radioactive isotopes in remote locations (e.g., concrete and shielding), for treating contaminated materials, for ultimate disposal (including packaging and ship- ping), and for analyzing the potential reuse of certain materials such as stainless steel. A continuing research program is needed to anticipate the much larger needs for decommissioning that will inevitably arise in the years ahead. The NRC has part of the responsibility be- cause it needs the research to set decommissioning regulations.
28 The NRC currently has a small program to transfer lessons learned from DOE's decommissioning of the Shippingport reactor to the NRC, but in general decommissioning research receives low prior- ity within the agency, and no substantial research is being con- ducted. Industry has a responsibility to do research that will establish ways of satisfying the NRC's regulations. DOE shares part of the responsibility for decommissioning research not only because of its mission to conduct basic energy research but also because it has reactors that will soon require decommissioning. It currently is conducting a sizable program of research, the principal elements of which are the decommissioning of the demonstration reactor at Shippingport and the removal and analysis of the core of the damaged Three Mile Island reactor. The results of DOE's current program should be transferable both to the NRC and to industry. Extended Fuel Cycle The economics of nuclear power are strongly determined by the availability of nuclear plants for electricity generation. The frequency and duration of refueling outages are among the factors that influence plant availability. If nuclear fuel were present in reactors for a longer period of time without reducing operating efficiency, nuclear plants might achieve greater availability and spend less for fuel fabrication, spent fuel storage, and waste dis- posal. However, extending the period between refueling outages could in some cases also extend the time between preventive main- tenance and component and system testing. So extended fuel cycle research is aimed not only at demonstrating the safety of new fuels but also at demonstrating the safety of extending the period of time between scheduled outages. Both industry (in this case the utilities and the nuclear fuel suppliers) and DOE have research programs on extending the use- ful life of nuclear fuel. To the extent that this research is motivated by economic considerations, industry should be primarily respon- sible for funding and setting the agenda for the research. DOE began research in this field at a time when uranium resources were thought to be at risk. The purpose of the research was to stretch those resources as far into the future as possible. With the cur- rent worldwide glut of uranium, this justification for the program is
29 reduced. Nevertheless, extended burnup fuels raise potentially sig- nificant issues for the management of the nation's high-level waste repository program, and DOE would be justified in conducting research to assess the potential impacts of high burnup fuel on waste packaging, handling, and repository size requirements. At some point, the NRC may need to conduct confirmatory research in order to permit extended-life fuels to be licensed, but because the lead time for this research is likely to be shorter than the expected lead time to develop the technology, this research need not yet be done. Furthermore, when the time does come to engage in this research, the NRC should ensure that it is done but should not pay for it. This is a case where the NRC should direct industry to pay for the needed research. Human Factors, Instrumentation and Control, and Operations One of the most significant lessons of the accidents at Cher- nobyl and Three Mile Island is that the people who design, operate, maintain, and manage nuclear power plants make up a system ev- ery bit as important to safety as the major components in a nuclear plant. At any time of the day or night, plant operators must be ready and able to diagnose disturbances in plant operations and prevent them from leading to a major accident. We now have enough operating experience to know that human errors are a significant contributor to the class of reportable events that oc- cur at nuclear power plants, and we also know from probabilistic risk analyses that human errors are a significant contributor to plant risk. Operating experience indicates that some errors can be triggered by the failure of instrumentation and control systems to operate reliably and to assist operators in preventing events from occurring that challenge plant safety systems. Still others can be related to faulty or improperly executed operating and maintenance procedures. Instrumentation and control system research should aim to provide (1) diagnostic aids and instrumentation for nonsafety sys- tems so as to reduce human errors and improve early diagnosis of incipient failures, and (2) new techniques and instruments, in- cluding sensors and microprocessors, for on-line calibration and testing.
30 All these areasâhuman behavior, instrumentation and con- trol, and operationsâare in fact related to one another and belong within a single category of human factors research. Human factors research encompasses two broad areas and the following topics: For normal operations: ⢠control room management ⢠maintenance management ⢠training in plant operations and maintenance ⢠management of the interface between control room opera- tion and maintenance ⢠optimization of the division of labor between human oper- ators and plant systems, including development of expert systems For accident situations: ⢠development of sensors, displays, and redundant instru- mentation ⢠development of real-time simulators for training ⢠development of improved accident diagnostics ⢠analysis of operating crew behavior ⢠emergency operating procedures, including development of expert systems ⢠integration of human factors engineering into the design of safeguards and other engineered safety features Both the federal government and industry have responsibil- ities in human factors, instrumentation and control, and opera- tions research, but the government has very little activity of this kind focused on existing reactors. For its part, industry should be funding an integrated program in these areas. But because industry views the NRC licensing process as an impediment to in- corporating advanced technology into current plants, there is little likelihood that an integrated effort will be forthcoming. For this reason the federal government must establish a base program that encompasses these various areas and seek to play a leading role in encouraging the transfer of advanced technology for improving human reliability to the nuclear industry. At present, the NRC has almost no program whatsoever, and few national laboratories have much human factors expertise. The bulk of the expertise is in industry outside the nuclear field.
31 The NRC must sponsor this kind of research because without a major program of its own it will be slow to respond to the improvements currently being developed in some of the rapidly changing technologies in this field, such as process control. Many of the technical fields properly included within human factors research are so new that the NRC will need its own independent research program just to keep track of developments. Perhaps the crucial point to be made is that the NRC needs to establish a regulatory climate that encourages technological advances in the area of human factors, rather than one that is merely neutral with respect to it, as is now the case. DOE should fund and direct its contractors who operate DOE reactors to perform research in all of these areas, because the research is needed for the continued safety of DOE reactors, because it is largely developmental, and because new technology is not required for commercial reactors by existing regulations. DOE's advanced reactor R«kD programs can be expected to produce results applicable to existing plants. The agenda for this research should be set cooperatively by industry, the NRC and DOE. Nondestructive Testing and Examination Nondestructive testing (NDT) and nondestructive examina- tion (NDE) are important technologies for in-place examination of nuclear power plant equipment and systems so that early indica- tions of degradation can be obtained. In principle, NDT and NDE can provide crucial data necessary for determining when to make repairs and modifications to plant components. Further research is needed to develop NDT capabilities not now available, to refine the sensitivity and accuracy of the techniques now in use, and to develop and validate the reliability of advanced methods such as acoustic emission technology. Research is also needed to develop methods for extending NDT to reactor internals. Both the industry and the NRC have relatively large pro- grams in NDE. EPRI sponsors an NDE center in North Carolina, and the reactor vendors and other contractors have established service businesses supplying NDE and NDT techniques to the utilities. The NRC is sponsoring research in this field in order to be able to confirm the applicability and reliability of NDT and NDE techniques used by licensees. Certain types of nondestruc- tive examination and testing are relatively new and could provide
32 new technical approaches to monitoring plant component and sys- tem degradation. Because of this promise, the NRC and industry programs in this area should dedicate a portion of their funding to basic and exploratory research. Plant Aging (License Extension) Current operating plants were licensed to operate for either 30 or 40 years, depending upon when they were built. Some of these have licenses that begin to expire during the 1990s. It is common practice in the United States to extend the lifetime of large industrial facilities beyond what was anticipated when they were built, especially if they are capital cost-intensive. The nuclear industry believes that this will be especially appropriate for those nuclear power plants where no major licensing issues are outstanding and where the capital cost of the facility has been fully paid off. Because the cost of capital is the major component of the cost of nuclear-generated electricity, electricity produced by such plants is relatively inexpensive. These economic considerations, coupled with the current reluctance of utilities to consider building new capital plants of any kind, mean that there will be increasingly intense pressures to keep existing nuclear power plants running as long as is safely possible. In order to extend the life of plants now in operation, the utilities will need to be able to have their operating licenses extended. Consequently, the NRC needs research to enable it to define whether safe conditions for license extension can be found. This will be a difficult and challenging problem, because each request for a license extension will depend upon the operating history, design characteristics, and anticipated mode of future operation of each plant. Although there is work on plant aging in progress, both within the NRC and among the utilities and vendors, the committee has seen no evidence that the results are being used in a systematic way to develop an integrated approach toward life extension research and regulation. The NRC needs and should fund research that will allow it to set new design margins and to evaluate the adequacy of existing ones under extended-life conditions. The utilities need and should fund research to prove that license conditions set by the NRC can be met. Because the industry has enormous incentives to conduct research in this area, it should play a major role in
S3 funding plant aging research. However, DOE also has a role to play because it needs the research to ensure the continued safety of DOE's production reactors, because it can provide industry with useful spin-offs from development work in other areas such as nondestructive testing and examination, and because it can provide industry with insights from efforts to design advanced reactors intended to operate for a long time. Since this research is of mutual benefit to the NRC, the DOE, and the utilities, the agenda for plant aging research should be set cooperatively. Indeed, an integrated, well-thought-out program of research should be undertaken. The research should focus on long- term chemical damage to pipes, valves, and other components ex- posed to gases and liquids, on long-term radiation damage to core, structural, electrical, and instrument components, and on long- term effects of operational cycles on mechanical integrity (fatigue and wear). The program should analyze the ability of compo- nents and systems to function beyond their design life. Examples of specific research topics in this field include in situ weld-repair techniques; structural integrity of plant systems, including the long-term integrity of radiation-embrittled materials; on-line diag- nostics to measure degradation, including nondestructive testing; and the effectiveness of in-place annealing on the brittle fracture behavior of reactor pressure vessels. Policy Research One of the primary purposes of nuclear safety research, es- pecially that funded or required by the NRC, is to inform reg- ulatory decisions. However, little systematic research has been conducted on the use of scientific and engineering knowledge in nuclear regulationâa context in which legal requirements, interest group politics, and the exercise of policy discretion by government decisionrnakers are all at least as important as knowledge. Nor has research been conducted into what sort of knowledge is most useful or in what form it is most usefully presented to decisionmakers. Neither the government nor the industry currently funds pol- icy research focused on these issues. Their importance for the effective use of research results justifies a modest effort by both the NRC and the utility industry. One example of the role of such research is in seeking to reconcile the differences in perspective
34 between geophysicists and engineers who design nuclear reactor components, structures, and systems (see below). Quality Assurance/Quality Control Good quality assurance practices are important elements in the design, construction and modification of nuclear power plants and have a direct bearing on safety. The construction of a nuclear power plant must proceed in an orderly manner, with a high de- gree of management control and quality review. Experience has shown that standard construction practices do not consistently ensure that these goals are achieved. Deficiencies in quality assur- ance/quality control (QA-QC) have led to costly delays in plant start-up and, in some cases, have contributed to cancellation of the plant. Research is needed to develop techniques to evaluate the as-built condition of nuclear power plants, and to track these conditions over the life of the plants to ensure that repairs and modifications that were made to correct faulty conditions do not lead to future problems. Topics in this field include the reliabil- ity and efficiency of QA-QC practices, human factors in QA-QC, construction-induced anomalies and their repair, and QA-QC data base management. The aim of this research should be to establish the QA-QC systems for achieving safety and full compliance with codes and standards. The market for improved light water reactors and other ad- vanced reactors is likely to be contingent upon, among other things, the availability of QA-QC practices that afford greater assurance of quality in construction than is now typically being achieved. Research is needed to provide these improved QA-QC practices, including development of more effective systems of QA- QC documentation with advanced approaches to scheduling, con- figuration control, document control, records management, ma- terials control, materials storage, inventory control, and testing. The goal of this research should be to develop superior alternatives to the current paper-intensive approach to quality assurance. Industry has the responsibility to fund research in QA-QC; the NRC should not be funding research in this field. If the industry is unwilling to do the research, it would be more appropriate for the NRC to direct the industry to do it than to fund it itself. The agenda for QA-QC research should be set by industry, with
35 the exception of research in direct support of the NRC's ability to regulate where the agenda should be set cooperatively. The agenda for any basic and exploratory work that may be required should be cooperatively funded and cooperatively set by DOE and industry. Reevaluating Existing Regulations Over the last 25 years the federal government has amassed a large body of codes, standards, criteria, regulatory guides, and rules with which to regulate the various aspects of the nuclear fuel cycle. These were incorporated, one by one, into the framework of existing regulations as they were issued. By now a number of these technical codes and standards are thought to be out of date, if for no other reason than because research in nuclear safety and in other fields has continually revised and augmented the store of knowledge in science and engineering. Furthermore, the industry has produced a base of operating experience that can be used to reevaluate the existing safety margins contained in the regula- tions. A systematic research program is now needed to evaluate the overall adequacy of the existing regulations, integrating mod- ern scientific and engineering understanding and the accumulated lessons of plant operating experience into a more coherent system of regulations. As a consequence of this work, some currently out- standing safety issues may be brought to regulatory closure, and areas where further research is needed may be identified. But the real purpose of a program to reevaluate the existing regulations is to rewrite the regulations so that they are more consistent and more efficient in ensuring public safety. Reliability of Plant Components and Plant Systems Nuclear power plants in the United States average many more reactor shutdowns per year than their Japanese and French coun- terparts. The failure of plant components and systems has been a significant contributing factor in a number of these shutdowns. In many cases the components and systems that have failed (valves, valve operators, pumps, small turbines, control equipment) were not specifically designed for the nuclear industry; they are con- ventional equipment designed for many different industrial appli- cations. Component failures can degrade plant protection systems
36 and challenge the capabilities of the operating staff. The relatively large number of significant failures of components and systems in the last year illustrates the need for increased component and system reliability. Additional research is needed to extend real-time monitoring to other components and systems, to develop better methods of monitoring component adjustments and calibration, to establish better data acquisition and analysis of component reliability, and to optimize programs of preventive maintenance. One goal of this research should be to achieve better understanding of the effects of different system parameters, such as pressure and temperature gradients and fluid flow, on the performance of plant components during transients. Developmental work should be undertaken to determine whether simplifications or other changes in the design of plant systems can be made to increase plant reliability, safety, and economy. Because this is an area of research in which economic and safety objectives are virtually inseparable, it is a prime candidate for cooperative NRC-industry research. The NRC needs research in this field to confirm that the components in the plants it regu- lates are sufficiently reliable to meet overall safety objectives. The industry needs the same assurances, and it also needs improved plant availability. The Institute of Nuclear Power Operations in Atlanta is developing a component reliability data base that holds great promise for future research; both the utilities and the NRC can use this data base to further their research programs. Safety Analysis Methodology and Application Probabilistic risk assessment (PRA) is an analytical technique for evaluating plants and plant systems. Although there are sig- nificant uncertainties in the risk estimates derived from PRA, its techniques have been used to assign priorities to programs of research, design, and plant operations. When properly applied, these methods can assist in interpreting operating experience, in analyzing data on the reliability of components and plant systems, and in identifying potential contributors to severe accidents. The development of PRA methodology necessarily resulted largely from generic studies of the elements of risk in nuclear plants and applications of the methodology to illustrative plants. The generic approach was, and to some extent still is, essential
37 to the development and refinement of the methodology. Early generic risk assessments, however, were widely claimed to provide important insight into the extent of risk associated with nuclear plants in general. But a reading of the PRAs that have been conducted suggests that many, if not most, of the contributors to risk are plant-specific rather than generic in nature. Experience suggests, therefore, that future PRA activities should be directed primarily at plant-specific studies. The pri- mary value of the PRA is not the results it provides to the NRC but rather the knowledge, insight, and decisionmaking capabil- ity it can provide to the utility. The successful application of PRA techniques to specific plants requires the participation of the utility's operating staff, as well as physical examination and anal- ysis ("walk-througha") of the entire plant, comparing the plant as built against the plant's final design. Nevertheless, the NRC needs to be actively involved in developing methods to ensure that utility-sponsored PRAs are adequate and to ensure that it can independently evaluate their strengths and weaknesses and their implications for plant safety. Further improvement of PRA methodology will result from its widespread application to specific plants. However, certain improvements are likely to require generic research. For exam- ple, the analytical treatment of human factors, dependent failures, and external accident initiators in PRAs are difficult areas that can benefit from additional research. The incorporation of recent severe accident research results into PRAs is another area where methodological advances should be sought. Some generic research may be needed to ensure that PRA results drawn from studies of different plants are comparable. Finally, additional research is needed to develop improved methods of identifying reactor acci- dent sequences, plant-specific contributors to risk, and possible means of mitigating potential accidents. Both industry and the NRC should fund research on PRA methodology, and the agenda should be set cooperatively. Seismology, Soil Mechanics, and Structural Response to Seismic Events Current estimates of the risk posed by earthquakes contain large uncertainties and suffer from the difficulty of modeling seis- mic phenomena. Research in this field is needed on a broad range
38 of issues, including responses of plant structures to dynamic loads, systems response of reactors to earthquakes (i.e., the impact of earthquakes on the extent of defense-in-depth), and soil-structure interaction. The further research needed to evaluate and reduce these un- certainties will require the formation and use of a sizable data base on earthquakes, data on structural failures of industrial plants, and data from tests of component fragility under varying loads. It will also require better techniques for modeling complex soil-structure interactions and the response of structural, mechanical, and elec- trical systems to seismic motions. Both the NRC and the industry have significant research pro- grams in the earth sciences largely dedicated to data gathering. The collection of data, although helpful, will not eliminate funda- mental differences of opinion about the likelihood of seismic events of various severities against which plants might be designed. In fact, there is no universally accepted methodology for interpret- ing seismic data, for extrapolating them into regimes for which no actual data exists, or for drawing conclusions from the data and applying those to regulatory questions. It is not surprising, therefore, that in this field the NRC and industry have a history of disagreement. Among the underlying causes of disagreement are important differences in perspective between earth scientists and engineers. Geophysicists tend to have a retrospective focus and to be more concerned with understanding a problem by eliminating uncertainty. They have played a leading role in the government's earth sciences research program and as advisors to the NRC and intervenors on earth sciences issues arising in regulatory proceed- ings. Engineers have a prospective focus and tend to be more concerned with solving problems and designing around uncertain- ties. Civil engineers have tended to play a leading role as earth sciences advisors to industry. This basic difference in perspective between the two groups has led to and probably will continue to result in an inability of technical experts relied upon by the NRC, utilities, and intervenors to reach consensus on earth sciences reg- ulatory issues. The NRC and industry would be well-advised to seek ways to reconcile the difference. Involving both geophysi- cists and engineers in cooperative research or initiating a series of conferences to explore their disagreements might help.
39 The federal government has a substantial program of research in the earth sciences, carried out principally by the U.S. Geolog- ical Survey, much of it directly relevant to the safety of nuclear facilities. The NRC could benefit from better coordination and collaboration with the USGS program. In addition, the NRC needs to focus greater attention on the difficult problem of system and plant response to beyond-design basis earthquake-induced ac- cidents. Severe Accidents One outgrowth of the accident at Three Mile Island was an increased emphasis on research to understand the phenomenol- ogy of severe reactor accidents and how to mitigate them. This research has led to significant advances in the state of our knowl- edge about the complex physical and chemical processes of severe accidents, but uncertainties remain concerning the adequacy of the computational methodology used to analyze severe accidents and the reliability of the experimental data base that supports it. Long before the Soviet nuclear accident at Chernobyl, it was clear from severe accident research that the major source of risk to the public stems from accident sequences that threaten the integrity of reactor containment structures. One of the principal goals of future severe accident research should be to establish con- tainment performance, and this will require further basic research on the physical and chemical processes relevant to severe accidents that might breach containment. Both the NRC and the industry currently have large research programs on severe accidents. Nearly a quarter of the NRC pro- gram, in fact, is devoted either to research on severe accident phenomenology or risk assessments related to it. Severe accident research can be aimed either at preventing or mitigating severe accidents or both, although it is more often aimed at mitigating them. The committee believes that more research on accident preventionâincluding research to analyze the adequacy of alter- native means of core cooling (such as "feed-and-bleed"); to analyze the effects of unusual transients that go beyond the design basis of the plant; and to improve plant-specific PRAsâis required now that the mitigation issues are being resolved. This is an area in which both the industry and the NRC need data and where both need to be able to trust the findings.
40 Experimental work on severe accidents should be organized co- operatively by the industry and the NRC. Funding and agenda setting for methodology development, however, should proceed in- dependently because the NRC should not be solely reliant upon industry in the evaluation of severe accident research results. Thermal Hydraulics Thermal hydraulics has been the dominant area of nuclear safety research since regulatory research began. Prior to the ac- cident at Three Mile Island, nuclear safety research in thermal hydraulics was focused on reactor steady state conditions, re- actor transients, and large-break, loss-of-coolant accidents. The focal piece of the NRC's early research was the Loss-of-Fluid Test (LOFT) program and the LOFT reactor at the Idaho National Engineering Laboratory. In many ways the legacy of LOFTâ the perception of it as an enormously expensive, largely failed and mismanaged project, particularly before a management re- organization in 1977âstill continues to dog the NRC program as a whole. Since Three Mile Island, however, the entire direc- tion of safety-related thermal hydraulics research has changed; the emphasis now is on small-break, loss-of-coolant accidents and off- normal thermal hydraulic plant behavior (so-called "transients"). The principal reason for this focus is the recognition, based on operating experience, that complex transients are a more credible source of accidents than originally thought. Research in this field has two principal aspects: large-scale computer codes and both large- and small-scale experiments. Fu- ture research should further analyze the response of existing plants to thermal hydraulic transients and develop more effective compu- tational tools and numerical methods for modeling and simulating thermal hydraulic phenomena. Research should be designed to provide fast running versions of the existing suite of computer codes for use in nuclear plant analyzers and reactor simulators. These tools could provide analysts with better means of evaluat- ing reactor response to upset conditions and provide reactor oper- ators with better ways of testing emergency operating procedures. As computerized systems become faster, attaining both real-time capability and more realistic simulation, they can be expected to provide invaluable assistance in the training of nuclear power plant
41 operators. Many of the existing codes contain bounding-type as- sumptions rather than realistic models. For advanced computer systems to be completely effective, therefore, substantial research will be needed to upgrade and validate the existing codes and to develop and validate new ones. The current codes are integral parts of the reactor licensing process, and they must be demonstra- bly valid and readily available to all who need them. In addition to complex transients, phenomena such as flow-induced vibrations, water hammer, and the off-normal behavior of steam generators are candidates for early further thermal hydraulics research. Code validation and greater understanding of thermal hy- draulic phenomena both depend upon the results of properly de- signed and conducted experiments. The United States has very few remaining governmental facilities for experimental thermal hydraulics research, and some of these are to be dismantled in a few years. New facilities are being planned or built abroad, in- cluding ROSA IV (Japan), BETHSY (France), and SPES (Italy), and U.S. researchers participate in work at some of them. These facilities have been designed specifically to study small-break, loss-of-coolant accidents and thermal hydraulic transients. Con- struction and operation of facilities for experimental research are time-consuming and expensive, but there is a continuing need for experimentation. The United States must plan on being an active participant in these international programs for the long term. Al- though the utilities have in the past participated in international work of this kind, the international character of the work indicates that the NRC has the principal responsibility for funding and helping to set the agenda for international experimental thermal hydraulics research. Nevertheless, EPRI can and should play an equivalent role. The NRC needs thermal hydraulics research to regulate ex- isting reactors. For example, the thermal hydraulic behavior of some reactors during small-break, loss-of-coolant accidents un- der station blackout conditions (e.g., pump seal loss-of-coolant accidents) are not well understood. The NRC needs to conduct sufficient research in this field to know whether industry has done the right work, correctly and competently, and what the research results mean. For these reasons, some research should be funded by NRC.
42 FUTURE PLANT DESIGNS Because the work of this committee was sponsored by the NRC and because of the limited time available to do the study, the committee tended to focus primarily on the research program of the NRC. The NRC's mission is to regulate the safety of current reactors and to review the safety of reactor designs that industry wishes to build. As a result of the prevailing conditions in the industry and the constraints on the NRC's research budget, the agency is currently sponsoring no research on advanced reactor safety (except for a relatively insignificant amount of "technical assistance"). Essentially all such research is sponsored by DOE and by industry. As a result, the committee did not examine safety research on future plant designs in any depth. Nonetheless, there is one comment related to public policy that the committee must make here. The NRC should now consider what research must be conducted (by the NRC and by others) relative to the safety of new reactor designs and, in particular, what standard of safety such reactors must meet. The urgency of having the NRC address these issues derives from the fact that a consideration of licensing requirements is a necessary part of plant design. Designing an advanced reactor without considering the NRC's safety concerns and regulatory requirements makes no sense, but designers will have no alternative unless the NRC provides adequate guidance. It is poor public policy to have DOE and industry funding advanced reactor research and development while the NRC has failed to lay the regulatory basis for such reactors. The Commission's recent policy statement on advanced reactors fails to provide the program of regulatory research on advanced reactors with the detailed guidance that it requires.
