Research and Development Response to the TMI-2 Accident

In response to the accident at Three Mile Island, the NRC re-examined the adequacy of its safety requirements and, consequently, imposed new regulations to correct deficiencies. The regulations defined new requirements for operator training, testing and licensing, and for shift scheduling and overtime. The NRC also began to emphasize the importance of "human factors" in plant performance in an effort to avoid a repeat of the operator errors that had exacerbated the accident. In cooperation with industry groups, it promoted the increased use of reactor simulators and the careful assessment of control rooms and instrumentation.

Most regulations could not be written without some research effort to first provide insight into the issue. Hence, the NRC expanded its research programs to address problems that TMI had highlighted, including human factors, small-break LOCA, fuel damage, fission-product release, and hydrogen generation and control. The primary products resulting from these research efforts were computer codes that could capture the knowledge base for future application.

Human Factors

In a complex industrial facility such as a nuclear power plant, the majority of the tasks are performed by machines. But the human operator is, of course, involved to a great extent in their design, testing, maintenance and operation. The performance of a person working within a complex mechanical system depends on that person's capabilities, limitations and attitudes, as well as on the quality of instructions and training provided. The interface between a machine and its operators in any industrial project is usually known as the human factor.

The Kemeny Commission identified several human factors areas that contributed to the accident. Principal issues include training, procedures, control room design, reporting, and equipment maintenance. The use of plant simulators has been an important tool for improving human factors in all these areas. Through the use of simulators nuclear plant personnel can exercise their skills while human factor scientists can evaluate their success, identify and report common failures, and test machine designs that improve the effectiveness of plant personal.

The nuclear industry owes much of its design, operation and maintenance philosophy to the fossil fuel industry. Unfortunately, an event like TMI-2 reminds everyone that a nuclear power plant is a much more complex machine than any other power plant. Early human factors work following the TMI-2 accident focused on ways to manage this complexity. For the issues identified by the Kemeny Commission, these changes included:

  • Training now accompanies all regular and emergency preparedness tasks. Through the use of computer simulators, training is made to be as realistic as possible.

  • Procedures for both normal and emergency operations were updated to be more technically accurate, better defined, and entirely comprehensible. For tasks performed infrequently and for emergencies, procedures were expanded in detail and clarity.

  • Control room design and layout modifications have been carried out in existing plants to reduce the probability of design-induced error. Such improvements can lead to the prevention of accidents or better management of accidents if they occur, although care must also be taken to avoid causing human errors by changing layouts or designs with which the operator is familiar.

  • Reports are now required for every event which occurs out of the ordinary, a form is completed describing the event, its probable cause and other pertinent information. This information is compiled and evaluated statistically for assessing the likelihood of accidents via methods such as Probabilistic Risk Assessment. If human error information is accurately reported, such statistics are useful for future human factors improvement.

  • Equipment design, maintenance and testing now focuses on improving the identification of equipment and access to it, providing better technical manuals and written procedures, and by designing ergonomically better tools and instruments. Maintenance errors can be reduced further by improving the work environment -- for example, by avoiding extreme temperatures, noise and inadequate lighting.

    Most of these human factors issues are interrelated. As mentioned, training accompanies all regular and emergency preparedness task; however, it also must be used to validate the effectiveness of changes made in the name of human factors. For example, human error during training activities has also been used to explicitly determine how a system or component may fail and what safeguards can be incorporated by the designer to prevent or mitigate such failures.

    Control Room Software Assistants

    One of the major human factors contributions that has been made since the TMI-2 accident has been the introduction of the Safety Parameter Display System (SPDS). As required by the USNRC, every nuclear power plant control room must include a SPDS. NUREG-700 specifically provides information on SPDS requirements based on lessons learned from the TMI-2 accident, recommendations for general emergency preparedness, and human factors criteria, respectively. Following these regulations, all major nuclear vendors in the United States began development of software applications that meet the USNRC's SPDS requirements. This action by the USNRC was quite precedent setting. By introducing computer software applications into the control room of operating plants to assist operators, a plethora of proposed software applications emerged from vendors, academia, government, consulting companies and utilities.

    During the 1980s significant research efforts went into development of "Software Assistants" for reactor operators. The capabilities of these software assistants typically fall into three categories: data interpretation, database management, and analysis and diagnosis tools. Most of these applications fall into the category of data interpretation.

  • Data interpretation software monitors and determines qualitative and quantitative information about the state of various plant systems and conveys that information in a "clearly understandable" format for an operator. These applications are the natural extension to the SPDS software.

  • Database management applications often use the same data acquisition facilities as the SPDS and maintain operational status of redundant systems.

  • Reactor core analysis and diagnosis software are typically the most computational intensity and represent the best utilization of computing power of existing software applications for the nuclear plant operator. Most all major U.S. and International nuclear vendors offer sophisticated core/plant analysis and diagnosis applications.

  • Despite the extensive amount of research and development that has gone into software assistants for operators very few applications have actually entered the control room. Demand, functionality, and limited financial or other resources have narrowed the number of software applications actually making it into the control room of an operating nuclear plant. This small list of software applications is predominately products produced by nuclear vendors.

    LOCA Research Refocused

    The early 1980s was a crossroads in LOCA research. A lot of data had been recovered from PWR and BWR thermal-hydraulic test facilities during the later part of the 1970s and the USNRC was working to understand the lessons arising from this work. At this time many test facilities were being completed, actively operating, or being planned and constructed. The table below compiles a small subset of these facilities and the focus of their research - links provide some detail on their safety contribution. A more complete, although not exhaustive, list is available from this Test Facility Summary Link.


    Model Size Linear Scale


    Integral System

    Separate Effects Test


    Blowdown Heat Transfer


    Reflood Heat Transfer


    ECCS Bypass
    Bldn & Refill


    Pump and Steam Generator


    Upper Plenum Deentrainment


    Nearly Full Scale




    GE TLTA-1

    PKL-340 rod bundle

    1/4 & 1/5 Scale









    CE LOFT Pump












    One of the unexpected circumstances surrounding the TMI-2 accident was the realization that a small break LOCA could result in serious core damage. Granted much of the consequences of the accident were caused when the operators overrided the ECCS system. Nonetheless, the traditional approach to ECCS design was to prepare for the worst case and then everything else should take care of itself. The TMI-2 accident contributed uncertainty to this idea.

    In response, the USNRC refocused its ECCS research on to the Small Break LOCA phenomena. The LOFT program was extended to include some small break tests. Results from these tests recognized that the small break LOCA isn't necessary as benign as original believed; however, existing ECCS designs where adequate to ensure that the small break is no worse than a large break LOCA. The USNRC revised the licensing rules for the small break LOCA based on lessons learn from TMI and LOFT.

    The U.S. reaction to the TMI-2 accident reverberated around the world. Internationally, there was a movement for expanded safety research and internationally cooperation. Through organizations such as the Organisation for Economic Co-operation and Development (OECD), engineers and scientists from around the world could observe and contribute to research in foreign locales.

    Two major international projects included the OECD LOFT project and the 2D/3D Program. As the USNRC's test program was closing out, representatives from OECD countries recognized the uniqueness of the LOFT facility in which significant safety experiments could be carried out. Through Nuclear Energy Agency of the OECD, the OECD LOFT program was organized to focus on more severe transients in which fuel disruption and release of fission products would occur. From the winter of 1983 until the summer of 1985, 8 experiment were conducted - including two that resulted in damage to the fuel and release of fission products.

    The 2D/3D Program was a multinational (Germany, Japan, and the United States), experimental, and analytical nuclear reactor safety research program for investigated multidimensional thermal-hydraulic behavior during the refill and reflood phases of a LOCA in PWRs. Organization for this program began in the spring of 1976; however, it would take a few years until facilities would be completed. The German contribution to the program was the Upper Plenum Test Facility (UPTF), a full-scale facility with vessel, four loops, and a steam-water core simulator. The Japanese operated two large-scale test facilities: Cylindrical Core Test Facility (CCTF) and the Slab Core Test Facility (SCTF). Both facilities are scaled on a power-to-volume basis, preserving full-scale elevations. The United States contribution was to proved advanced two-phase flow instrumentation and analytical support through development of the TRAC best-estimate thermal-hydraulic system simulation computer code.

    By the end of the 1980s, the body of ECCS research was quite extensive. Unfortunately, documentation reflected the fact that the research had been carried out by many different teams of researchers. The NRC decided to take on the task of compiling the significant results in to the "Compendium of ECCS Research for Realistic LOCA Analysis" (NUREG-1230). While the "Compendium" did not take into account the lessons learned from all experiments performed world wide, it provides a very good source for understanding the lessons learned from dozens of tests and hundreds of millions of dollars of investment.

    Best-Estimate LOCA

    As mentioned in the previous section the experimental program of the USNRC and of other countries contributed extensively to the development of Best-Estimate Thermal-Hydraulic computer codes for safety analysis of Light Water Reactor nuclear power plants. Many of the early improvements in these codes are traceable to deficiencies uncovered through modeling of the LOFT and Semiscale facilities. The knowledge and understanding of accident behavior gained from careful analysis of these facilities was instrumental in pointing the direction for further improvements. Alternatively, the experimental programs used code calculations to better understand the behavior of their facilities in terms of their limitations and how they might be improved. The close association of code development and experimental programs produced a synergistic effect in which both benefited.

    During this time of active experimental research, the USNRC sponsored (at various levels of support) the development of a number of computer codes to predict LOCA phenomenon. These included RELAP2, RELAP3, RELAP3B (BNL), RELAP4, RELAP5, TRAC-PF1, TRAC-PD1, TRAC-BF1, RAMONA-3B, THOR, RAMONA-3B, RAMONA-4B, HIPA-PWR and HIPA-BWR. The Los Alamos National Laboratory held responsibility for the TRAC codes (with the exception of TRAC-BF1), the Idaho National Engineering Laboratory and Energy Incorporated (an Idaho-based consulting company) were responsible for RELAP codes, and Brookhaven National Laboratory was responsible for THOR, RAMONA and HIPA codes. In general, the USNRC did not specify a particular objective for the computer code development they were supporting; however, as the computer codes evolved, the USNRC did recognized that some codes had unique strengths and weakness. TRAC, RELAP5 and RAMONA were recognized as the advanced simulation computer codes for PWR and BWR plants safety analysis. THOR's architecture would simplify numerics and constitutive models so that results could be generated quickly. RELAP4 would be developed for licensing calculations based on Appendix K of 10 CFR 50. RELAP4 became the basis for the Water Reactor Evaluation Model (WREM) and TRAC was used to audit safety analysis from vendors and utilities. Before the ACRS, the USNRC did eventually recognize TRAC as the Large Break LOCA code and RELAP5 for Small Break LOCA.

    The fact that there are three USNRC developed best-estimate safety codes - TRAC, RELAP5, and RAMONA - was never an objective of the best-estimate LOCA program. By the mid to late seventies, the USNRC had pretty much defined TRAC as the premier best-estimate, thermal-hydraulic code. Other computer code projects began to see reduced funding or were dropped altogether. RAMONA would survive at a lower funding level by focusing on unique BWR plant phenomena. Meanwhile, RELAP5 would emerge through the "backdoor." At the Idaho National Engineering Laboratory a small team of engineers presented their results from a spin-off project from the experimental work being done at Semiscale. Their task was to investigate the characteristics of various numerical advancement procedures as applied to the difficult problem of modeling two-phase (steam/water) flow. This effort was organized by Dr. Victor Ransom, Dr. John Trapp, and Mr. Richard Wagner, and their results came from the PILOT code which they had work on for nearly 2 years. The PILOT code was no-where-near the sophistication of TRAC; however, based on the success of the PILOT code, NRC would continue to support its incorporation into a system oriented code to supersede the RELAP4 code through the LOFT program.

    The developers goals for this code was to create a fast-running, "best-estimate" code primarily for application to pressurized water reactors. The "best-estimate" descriptor meant that physical phenomena would be mathematically modeled in as realistic a manner as best as possible, rather than using so-called conservative models. Another important goal was ease-of-use, or what we today call "user friendly". This led to the use of free format input, extensive input data checking and diagnostics, and a built-in plotting feature. Finally, the developers wanted a code that was flexible, in the sense that virtually any thermal-hydraulic system could be modeled. Thus they avoided rigidly defined component models and gave the user the freedom to define the system as he/she sees fit.

    During most of the 1980s a "beauty contest" mentality emerged from a perceived competition between RELAP5 and TRAC. The USNRC encouraged this perception believing that the competition would be healthy for the advancement of the Best-Estimate codes. They were right. Both teams worked intensely to show that their code was better. However, by the end of the decade, RELAP5 would emerge as the industry favorite primarily because of its relatively "user-friendly" features and execution speed.

    By the end of the 1980s, the USNRC was fairly satisfied with the ability of computer codes to simulate nuclear plant response under adverse conditions. In 1987 they published NUREG-1230, "Compendium of ECCS Research for Realistic LOCA Analysis." This was followed by a Rule change published in Regulatory Guide 157 presenting a best-estimate approach to licensing nuclear power plants. This approach removed much of the conservatism of 10 CFR 50 Appendix K; however, it requires application of statistics to assure system integrity during an accident.

    Research Fallout from Chernobyl

    The Chernobyl accident has been written about and analyzed from just about every point of view imaginable. The immediate cause of the Chernobyl accident was a mismanaged electrical-engineering experiment. Ultimately, to conduct this experiment, reactor staff had to bypass multiple layers of safety systems that would prevent the reactor instability that the Chernobyl reactor would eventually experience. While this unfavorable safety culture would be the primary cause of the accident, a number of other factors would contribute to the extensiveness of the fallout.

    The design of the Chernobyl reactor presents a number of problems that can only exasperate a severe accident. In contrast to western-styled Light Water Reactors, Chernobyl's RBMK reactor design contains these recognized design flaws:

  • Inherently unstable under conditions of loss-of-coolant.
  • Inherently unstable under conditions of increased temperature.
  • Lacks sufficient containment.

  • The absence of a containment structure was especially important. Post-accident analyses indicated that if there had been a U.S.-style containment, none of the radioactivity would have escaped, and there would have been no injuries or deaths.

    The social fallout from Chernobyl put pressure on the nuclear industry to take a harder look at safety in general and the safety mechanisms that prevent severe accidents. In response, governments and private industry sponsored research in severe accident research, advanced reactors designs, new engineered safety systems, and refined regulation following the Chernobyl accident. However, much of this work just extended the path of research begun with the accident at Three Mile Island.

    Ultimately, attention focused on raising the safety standards of the nuclear industry internationally. The International Nuclear Safety Program was established for this purpose. The INSP offers training and technology transfer to under-developed nuclear countries. They have also been actively involved in offering direct services to address specific needs of operating reactors world wide.

    Containment Research at the SURTSEY Facility

    As Three Mile Island and Chernobyl demonstrated, severe accidents - those accidents that deform or melt fuel - can happen. For western-designed reactors which are designed with a containment, it was generally believed that the containment integrity would be maintain for most of the severest accidents imaginable. This accidents would result in a Direct Containment Heating (DCH) situation in which molted core material breaches the reactor pressure vessel and enters the containment. However, no experimental program had been designed for this issue.

    The SURTSEY Direct Heating Test Facility was built to answer questions about DCH for severe accidents. The SURTSEY facility was sponsored by the USNRC and built at Sandia National Laboratory in Albuquerque, New Mexico. Previous experiments and analysis had suggested that the ejection of core debris into the reactor cavity might result in the molten core debris being lofted into the containment atmosphere. Preliminary calculation performed at SURTSEY suggested that if the energy transfer processes were efficient, only a relatively small fraction of the total core mass would be sufficient to threaten the integrity of some containment structures.

    The SURTSEY facility consisted of a pressure vessel oriented vertically with the lower head flange approximately two meters above the ground. A 1:10 linear scale model of the Zion cavity was placed in the vessel. The cavity exit was located on the vertical centerline of the vessel. The concrete lined cavity was modified by the addition of a 0.36-m square by 0.9-m tall steel "chute" attached to the exit of the cavity. The purpose of the chute was to direct the dispersed debris vertically upward to avoid ablation of the SURTSEY steel shell. The chute terminated approximately two meters above the flow of the cavity. The molten material was produced in the melt generator attached to the cavity at the scaled height of the reactor pressure vessel.

    SURTSEY was designed and constructed to perform experiments where molten debris was ejected into a well-defined and contained atmosphere. The size of the facility allowed realistic behavior of DCH phenomena. Instrumentation included various pressure and temperature measurements to track the dispersal of debris from the "pressure vessel." Additionally, the SURTSEY facility allowed for the debris and aerosol material to be sampled and recovered.

    Ultimately, much of the results from the SURTSEY were captured in new phenomenological models that could be incorporated in to computer codes. The USNRC sponsored the develop of the CONTAIN computer code at Sandia and much of the results at SURTSEY were incorporated into CONTAIN.

    The Advanced Light Reactor Program

    In the mid-1980s the U.S. Nuclear industry, through the Electric Power Research Institute, and the U.S. Department of Energy (DOE) coordinated an effort to initiate the next generation of nuclear power plants. Their objective was to assure the future availability of improved and simplified light water reactor plants to meet new or replacement capacity requirements. EPRI's first goal was to document the industry's requirements for advanced light water reactor (ALWR) plants. An integral part of the industry's requirements was their demand for real licensing reform to avoid the disaster of the 1970s. In general, EPRI called for:

  • Greatly simplified designs
  • Easier construction, operation and maintenance
  • Fully proven technology
  • Greater margin of safety in operating transients and accidents
  • Lower life-cycle costs than current generation nuclear plants
  • Reduced licensing uncertainty

  • EPRI began working with the major nuclear vendors, architect engineering firms, and utilities. Their task was translate the needs of the industry into new ALWR plant designs. Nuclear vendors, such as GE and Westinghouse, took EPRI's design requirements and initiated design teams to begin this process. EPRI and the nuclear vendors were also supported by DOE. While EPRI was focused on ensuring that the next generation of nuclear plants incorporated the lessons learned from the first generation, DOE was concerned with closing the design loop. Specifically, defining requirements of design verification.

    The DOE pledged to sponsor the USNRC Final Design Approval and Certification of two of the more developed design concepts: GE's Advanced Boiling Water Reactor and Combustion Engineering's System 80+. Additionally, they would handle the coordination of a standard safety analysis report and EPRI requirements document to confirm agreement or resolve any differences. Where EPRI and DOE overlapped was with the nuclear vendor designs.

    Probably the most studied part of the new design features of the ALWRs are the safety systems. Designers have taken two different approaches to address safety concerns: 1) redesign of conventional systems for improved reliability and safety and 2) develop passive safety systems. Suggestions for the conventional system redesign included:

  • Increasing the number of ECCS trains from two to four.
  • Maintain a common ECCS and containment spray water source to eliminate the need to switch from an external source and provide a semi-closed system.
  • Inject coolant directly into the reactor pressure vessel
  • Rapid system depressurization to bring pressure below the shut-off head of the ECCS pumps.
  • These choices were implemented in evolutionary designs developed by GE (ABWR) and Combustion Engineering (System 80+). Suggestions for passive safety systems included:

  • Gravity-driven emergency coolant systems
  • System designs that facilitate natural circulation
  • Automatic depressurization systems to bring pressure below ECCS coolant source pressures.
  • Integrated containment cooling systems to mitigate severe accidents.
  • These choices were included in advanced passive designs developed by GE (SBWR) and Westinghouse (AP600).

    The Advanced Liquid Metal Reactor

    While the US DOE was supporting the ALWR program, they were also extremely active in supporting the Advanced Liquid Metal Reactor. This was primary through the activities of the Argonne National Laboratory - West and General Electric. Design concepts from Argonne West's Integral Fast Reactor (IFR) and GE's PRISM (Power Reactor Innovative Small Module) breeder reactor designs were investigated with much early success. It was planned that eventually ANLW's Experimental Breeder Reactor-II would be used to demonstrate the IFR concept. EBR-II and ANLW's Fuel Conditioning Facility is shown in the photo below

    The IFR concept involves a pool type breeder reactor, metal fuel alloy, and pyroprocess fuel cycle. Two features of the IFR make it different from all other liquid-metal reactors around the world: fuel type and fuel cycle technology. The fuel is a metallic alloy of uranium, plutonium, and zirconium. Metal fuel provides important new safety characteristics and allows the possibility of a radically simplified fuel cycle based on pyrometallurgical processes.

    The passive safety characteristics of a metal-fueled ALMR were demonstrated in a landmark series of tests at the EBR-II in 1986. Two classic anticipated transient without scram (ATWS) events were simulated. In separate tests on the same day, actual station blackout and loss-of-heat-sink conditions were created from full power, with scram circuits temporarily bypassed, and with no operator intervention. In both cases, EBR-II simply shut itself down to a low-power condition where the heat was rejected by natural circulation. No damage was done to either the fuel or the reactor system.

    ALMRs require a closed fuel cycle. In the IFR fuel cycle, a relatively high temperature, metal -based process is used - the pyroprocess. This is a simple batch process that involves 1) electrorefining, a single step in which fuel is dissolved and heavy metals separated from fission products by electro transport to a collector cathode; 2) cathode processing, in which heavy metals from the cathodes are purified by retorting, and cast into refined metal ingots; and 3) injection casting, in which new metal fuel rods are made.

    The ALMR is designed as a safe, reliable, and economically competitive liquid-sodium-cooled fast spectrum breeder reactor power plant, with the following key features:

  • Compact reactor modules sized to enable factory fabrication and shipment to either inland or water-side sites, and to permit affordable, full-scale prototype testing to confirm predicted safety and performance characteristics.
  • Passive reactivity reduction during undercooling and over-power transients with failure to scram, to achieve a safe, stable state, with abundant time for shutdown to cold conditions by operator action.
  • Passive decay heat removal for loss-of-heat-sink accidents that is invulnerable to operator errors and equipment failures.
  • Protection against severe accidents by simple and passive safety features so that radioactive releases are small, and formal public evacuation planning and exercises are unnecessary.
  • Capability for breeding more fuel than consumed.
  • Optional capability to use-as fissile material for startup-either plutonium or actinide wastes from light-water reactor spent fuel or excess plutonium available due to nuclear disarmament.
  • Flexibility of core design to use either the reference metal fuel or, alternatively, an oxide fuel cycle.
  • Possibly the most positive advantage of the ALMR designs is its capability to recycle high-level, long half-life actinides back into a reactor core that fissions actinides, due to the ALMR's fast neutron energy spectrum. Fissioning of actinidies precludes their accumulation and need for subsequent disposal. This capability can reduce the management of nuclear wastes from one of tens of thousands of years to a little over 100 years.


    1) L. S. Tong and J. Weisman, Thermal Analysis of Pressurized Water Reactors, American Nuclear Society, La Grange Park, Illinois, 1996.

    2) "The New Reactors," Nuclear News, A Publication of the American Nuclear Society, Vol. 35, No. 12, September 1992.

    3) "Compendium of ECCS Research for Realistic LOCA Analysis," NUREG-1230, March 1988.

    4) L. S. Tong, "USNRC LOCA Research Program," Proceedings from the Eighth Water Reactor Safety Meeting, NUREG/CP-0023, March 1982

    5) F. A. Ross and W. R. Sugnet, "Advanced Light Water Reactors for the Nineties," Proceedings from the Fourteenth Water Reactor Safety Meeting, NUREG/CP-0023, March 1982

    6) W. W. Tarbell, J. E. Brockmann, and M. Pilch, "DCH-1: The First Direct Containment Heating Experiment in the SURTSEY Test Facility," Proceedings from the Fourteenth Water Reactor Safety Meeting, NUREG/CP-0023, March 1982

    7) "Reactor Safety Issues Resolved by the 2D/3D Program," International Agreement Report, NUREG/IA-0127, July 1993.

    8) "The OECD/LOFT Project: Achievements and Significant Results," The Nuclear Energy Agency, OECD, 1990.

    9) "Anticipated and Abnormal Plant Transients," Proceedings from an American Nuclear Society Topical Meeting, Jackson, Wyoming, September 1983.

    10) R. Rhodes, Nuclear Renewal, Penguin Books, 1993.