On Vapor/Steam Explosion Analysis

THE NUCLEAR NEWS INTERVIEW Fauske and Henry: On vapor/steam explosion analysis

Vapor explosions are an integral part of reactor safety evaluations when accident conditions could lead to molten fuel coming in contact with liquid coolant.

he book Experimental Technical Bases for Evalu- involving these types of explosions must be evaluated ating Vapor/Steam Explosions in Nuclear Reactor in a manner consistent with the available experimen- TSafety was recently published by the American tal technical bases. Their new book, they say, provides a Nuclear Society. The authors, Hans Fauske and Robert common reference that includes the total experimental Henry, say that the possibility of vapor/steam explo- database for vapor/steam explosions, as well as informa- sions occurring at nuclear power plants is an important tion directly related to molten materials and the coolant consideration for safety assessments of events wherein a of interest. high-temperature molten mass could come into contact Early in his career, Fauske joined Argonne National with a liquid coolant. Laboratory, where he focused on reactor safety evalua- The authors stress that potential accident conditions tions for both fast and light-water reactors. Vapor explosions were a key element of the safety evaluations for the Fast Flux Test Fa- cility and the Clinch River Breeder Reactor designs, and steam explosions were an important part of the safety assessments for water-cooled reactors. Fauske left ANL in 1980 to establish, with Robert Henry and Michael Grolmes, Fauske & Associ- ates. In 2011, Fauske stepped away from management responsibilities at the company to spend more time on engineering evaluations and writing. Henry also worked at ANL, under Fauske: “We wrote the book to provide Henry: “Vapor explosions in general a convenient, common reference for are an integral part of reactor safety Fauske’s direction, on both fast reactor experimentalists or analysts.” evaluations.” and LWR safety. In March 1979, the

August 2017 • Nuclear News • 29 Three Mile Island-2 accident occurred, Through the years, Fauske & Associates and during the following summer and has been deeply involved in nuclear and fall, Henry was on loan from ANL to the chemical reactor safety evaluations. The Nuclear Safety Analysis Center, which company developed the Modular Acci- was established by the Electric Power dent Analysis Program (MAAP), a com- Research Institute to study the data ob- puter code that runs faster than real time tained from the accident. By that winter and can evaluate the response of boiling he was part of an ANL group that worked water reactors and pressurized water re- with Commonwealth Edison to conduct actors for a broad spectrum of accident a near-term study on the implications of sequences. Steam explosion modeling is the TMI-2 accident for the Zion reactor, an important component of the MAAP located north of Chicago, Ill. The issue of evaluations. steam explosions was a major component Rick Michal, director of ANS’s Depart- of the Zion study. Henry also authored TMI-2: An Event ment of Scientific Publications and Standards, conducted in Accident Management for Light-Water–Moderated Re- this interview with Fauske and Henry, whose new book actors, published by ANS in 2011. is available through ANS and at Amazon.com.

Why did you write this book? describing the test facilities and pro- in Chapter 2) that are completely consis- Henry: The book is needed by the nucle- viding a summary of the experimental tent with the experimental bases lead to a ar community. Let me begin by explaining results. Equally important, the 19 pages set of requirements for a large-scale vapor that a steam explosion is referred to as a of references in our book tell the readers explosion that could not be satisfied un- vapor explosion when water is the volatile where they can find more details on these der accident conditions. For commercial liquid that is vaporized. Vapor explosions experiments should they have a desire to water-cooled nuclear reactors with oxid- in general are an integral part of reactor read the original documents. ic fuel and zirconium alloy fuel pin clad- safety evaluations if accident conditions Fauske: A second but related reason ding, the experimental bases (discussed could lead to high-temperature molten for the book is that we have observed that in Chapters 3 and 4) that are derived fuel coming into contact with the liquid many of the original experiments don’t from numerous experiments throughout coolant. Both Hans and I have been active seem to be referenced in some of the more the world show only very weak events, for more than four decades in developing recent papers, even when comparable re- and these can be characterized in a sim- the understanding of vapor explosions for sults have been obtained. This can be for ple, bounding manner for severe accident different types of reactor designs. During many reasons, such as that the authors analyses. Chapters 5 and 6 discuss those this time we have had the opportunity to may not be aware of some of the original experiments that are related to the role know and work with many talented, ded- experiments, the original documents have of aluminum in water-cooled reactors icated engineers and scientists who have been difficult to find, or the authors docu- such as were observed in the BORAX-I touched on one or more of the numerous menting some of the more current works and SPERT-I tests, as well as the accident facets that comprise this phenomenon. In didn’t have time to develop an in-depth that occurred in the SL-1 reactor. High- addition, we have had the opportunity to understanding of the original documents. temperature molten aluminum is a very perform experiments, observe other ex- Whatever the reason, this experimental reactive metal in the presence of steam. periments, and examine the relevant data database is too broad and too important These components of the experimental that has been gained from programs con- to reactor accident evaluations and se- databases address the mechanisms and ducted all over the world. vere accident management guidelines to the necessary conditions for a steam ex- It has always been our conviction that be minimized. As a result, we wrote the plosion to transition into a chemical ex- any evaluation of vapor explosions for book to provide a convenient, common plosion. These are particularly relevant to potential reactor accident conditions reference for experimentalists or analysts special-purpose reactors that differ greatly must be consistent with the entire ex- to ensure that they are consistent with the from those just mentioned but neverthe- perimental database that has been de- broad experimental technical bases given less have aluminum structures in a water- veloped. To quote W. Edwards Deming, in the book when they are planning their cooled reactor. Consequently, these need “Without data you are just a person with respective approaches to vapor explosion to be considered in light of the possibility an opinion.” The relevant data associated experiments or models and when they are of a steam explosion alone, as well as the with vapor explosions now spans at least documenting their results. possibility, or lack thereof, for an explosive five decades, and some of the original Henry: A third reason for this book is chemical reaction to be initiated. papers, reports, conference proceedings, that we have found through this growing and other source materials are somewhat experimental basis that the major obser- Is the book intended to address the severe difficult to find, while some old reports vations have consistently demonstrated a accident analysis for all liquid-cooled reac- exist only as copies that are virtually few important behaviors that have differ- tors and not just commercial nuclear power unreadable. Therefore, our first reason ent characters, depending on the reactor reactors? for writing the book was to define the design in question. For the liquid sodium– Henry: Yes. Severe accident analyses technical bases by listing the key exper- cooled fast reactors, the general behavior must consider those conditions that could iments, from our perspective, and then principles for a vapor explosion (explained cause the reactor fuel to become molten,

30 • Nuclear News • August 2017 www.ans.org/nn Interview: Fauske and Henry which generally means very high tempera- plosion to occur, there must be nucle- throughout the world. Can you give some tures, and then come into direct contact ation of liquid to vapor and rapid energy examples of the extent of the experimental with the liquid coolant. This applies to all transfer to generate vapor faster than the database for the key principles mentioned reactors with a liquid coolant in which ac- surrounding medium can respond to the previously? cident conditions could result in very high rapidly increasing steam volume. Of these, Henry: The importance of the interface fuel temperatures and includes water- the method of vapor nucleation from the temperature principle is clearly illustrated cooled commercial power reactors, liquid liquid state is the most important. It differs by experiments performed at Argonne Na- metal–cooled fast reactors, and special- by orders of magnitude from the normal tional Laboratory using simulant materi- purpose reactors. boiling processes that experience nucle- als. These are further supported by the ex- ation at imperfections—for example, crev- periments where high-temperature solids Could you list the conditions or require- ices, scratches, and pits—in the solid sur- dropped into water did not produce a steam ments that could lead to a vapor/steam face for a liquid-solid boundary, such as a explosion. In addition, the experimental explosion? pot of water on a stove or a fuel rod surface database includes numerous in-reactor Fauske: First, there must be a liquid- transferring energy to the water coolant. and out-of-reactor experiments where liquid system in which the high- In addition, the instantaneous vapor- high-temperature molten oxidic materials temperature liquid comes into direct con- ization rate is orders of magnitude greater were poured into sodium with no initiation tact with the colder, more volatile liquid. than the well-known Second, experiments show that the high critical heat flux pro- “The experimental database temperature of the hot liquid must be cess for steady-state greater than a specific temperature for boiling behavior. For includes numerous in-reactor the vapor nucleation rate to be of suffi- a vapor explosion, the and out-of-reactor experiments cient magnitude to generate a significant nucleation sites are shock wave. In addition, experiments formed at a liquid- where high-temperature molten show that the temperature of the hot liq- liquid boundary that uid must be sufficiently high such that the has no such imperfec- oxidic materials were poured contact temperature is equal to, or great- tions, and, as a result, er than, the spontaneous temperature of much higher coolant into sodium with no initiation of the coolant. This spontaneous nucleation temperatures are re- large-scale vapor explosions.” temperature approaches, or is equal to, quired for nucleation the homogeneous nucleation temperature to occur. Once these temperatures are of large-scale vapor explosions because the of the coolant. These three points are dis- reached, however, the density of the nu- contact interface principle was not satisfied cussed in Chapter 2. And last, the ambient cleation sites quickly becomes enormous. due to the thermal properties of sodium. pressure surrounding the liquid-liquid This produces the immense steam genera- This further supports this principle. system must be less than the minimum tion rate that forms a shock wave and the Fauske: Numerous experiments have pressure that can prevent the occurrence explosive behavior. been performed with different liquid- of an explosive interaction, as is discussed liquid mixtures and at different scales in Chapter 3. There are many types of nuclear reactors in to demonstrate the influence of ambient the world. How do these principles relate to pressure on the capability to initiate a va- Could you expand on why there is such various types of reactors? por explosion. Simulant fluid experiments a strong emphasis on the experimental Henry: The basic vapor explosion gen- at ANL first demonstrated this influence, database? eral behavior principle—that the contact and additional confirmatory experiments Fauske: While the phenomenon of va- interface temperature must equal or ex- were performed with molten salt and wa- por explosions is a physical process that ceed the coolant spontaneous nucleation ter (a very explosive liquid pair) in the has been part of severe accident evalua- temperature—would not be satisfied for test facilities at the ISPRA laboratory in tions for decades, there are many different the accident conditions of a liquid metal– northern Italy. Furthermore, additional designs that need to be evaluated with re- cooled reactor. Thus, large-scale vapor small-scale experiments on the growth spect to whether or not such an event could explosions could not be initiated. For rate of vapor bubbles that are initiated by occur, and if it could occur, what the ener- commercial water-cooled reactors with homogeneous nucleation have also clear- gy released by the explosion would be. The oxide fuel and zirconium alloy fuel pin ly shown the influence of pressure. All of answers to these two questions are design cladding, the interface temperature prin- these contributions demonstrate this im- dependent, and each design should have ciple is satisfied, but the experimental re- portant behavior that could be influential a relevant technical basis to illustrate the sults discussed in Chapters 3 and 4, which for some accident conditions in water- influence of the design, as well as the range are from numerous experiments that were cooled power reactors. of possible severe accident conditions. Ex- performed in different laboratories in dif- perimental evidence is essential to confi- ferent countries, show either no explosive As we all know, there have been accidents dently cover the extent of such conditions. interactions or only very weak events if at Fermi-1, Three Mile Island-2, Cher- they could be initiated. Equally important, nobyl-4, and Fukushima Daiichi. Were The central thesis of the book is that any these can be characterized in a bounding any important lessons taken from these representation of vapor/steam explosions manner for severe accident analyses using accidents that confirm/supplement the ex- must be consistent with the body of exper- a simple theoretical representation for the perimental database? imental work reported in the literature. energy transfer. It should be noted that the Fauske: The liquid sodium–cooled, me- What are the key scientific principles that interface temperature principle is also sat- tallic uranium–fueled Fermi-1 reactor ex- are demonstrated by the composite exper- isfied for the molten aluminum-water sys- perienced the melting of parts of three fuel imental works? tem discussed in Chapters 5 and 6. assemblies. Considerable molten material Henry: The key principles are the mix- came in contact with the liquid sodium ing of the high-temperature melt and the Several times both of you have emphasized coolant, but there was nothing in the ac- coolant to a size scale for the melt of about that the important experimental observa- cident record or the defueling process that a centimeter in diameter. For a vapor ex- tions have been documented in laboratories indicated anything like a vapor explosive

August 2017 • Nuclear News • 31 behavior. This is consistent with the ex- droplets. As molten material increases, In another example, light-water– and perimental database. more steam is formed, and multiple ex- heavy-water–cooled commercial nucle- Henry: For Three Mile Island, the pres- perimental programs show that eventual- ar power reactors satisfy this interface sure in the reactor vessel when molten core ly the water can be dispersed by the steam temperature principle, but the database materials drained into water was much flow. When this is reached, no additional shows that the conditions of rapid melt greater than the minimum pressure that molten debris can be mixed into the wa- freezing due to the thermal radiation– experiments (and analyses) have shown ter. With the high melting temperature of dominated film boiling and the thermal would have prevented a steam explosion. molten uranium dioxide, or molten mix- properties of oxide fuels result in a very So, the accident behavior is consistent tures of core materials (corium), the steam low efficiency (conversion ratio) for ther- with the experimental database. generation rate needed to disperse the mal explosions. Combining these, one The Chernobyl RBMK design was a water typically requires only a very small finds that the damage potential for steam water- cooled, graphite-moderated reactor fraction of the core inventory, typically explosions is very small. Moreover, the that had a positive void reactivity coef- less than 1 percent. database also shows that any attempt to ficient. This was a design that was much drive the molten materials together at different from the BWRs and PWRs op- What determines if a steam explosion high velocity causes localized entrain- erated in the United States and elsewhere could transition into a chemical explosion? ment and quenching such that no signif- in the world. This design, combined with Fauske: If a high-temperature reactive icant molten core material was accumu- the sequence of events that preceded the molten metal is poured into water, the lated within the water. initiation of the Chernobyl accident, led metal-water reaction releases hydrogen. A third example is that special-purpose to a core condition in which the core was There are two controlling features: First, reactors can be fabricated with different- rapidly overheating and boiling away the the solubility characteristic of hydrogen ly designed fuel, constructed with other available water inventory. With the pos- in the molten metal as a function of tem- materials, and have substantially differ- itive void reactivity, this induced a rapid perature, and second, the metal tempera- ent operating parameters. Each design increase in the core power. The USSR’s ture. If the metal sol- report on the accident estimated that the ubility increases with “There are three types of core power reached 500 times the peak op- temperature and the erating power over several tens of seconds. metal temperature organizations that need to Consequently, the explosion that occurred is sufficiently high, a in the accident was due to a large pressur- significant quantity of consider the possible role ization of the reactor system that eventu- the hydrogen that was of vapor/steam explosions ally ruptured and discharged radioactive generated can be dis- fission products to the environment. This solved in the metal. If for individual designs: the was not a steam explosion. the local metal-water All three of the Fukushima Daiichi re- contacts escalate in- reactor vendor/designer, the actors that experienced severe core dam- to a steam explosion operating organization, and age were depressurized during the acci- or are forced into dents. In these accidents, steam explosions contact by an exter- the regulatory agencies.” were not prevented by elevated pressures, nal explosive trigger, so potentially they could have occurred. the rapid cooling induced in the molten needs to be evaluated individually. This Most certainly there were hydrogen ex- metal results in supersaturation of hydro- also means, however, that the possibili- plosions in the various reactors. The eval- gen, which then rapidly exits the solution, ty of rapid chemical reactions that could uations of the plant data, however, and the causing fragmentation of a thin layer of cause a steam explosion to transition in- robotic excursions into the containments molten aluminum that is dissolved by the to a chemical explosion would need to be of Units 1 and 2 have shown no indication hydrogen. These fine-scale molten metal considered for water-cooled reactors that of any damage or any other observation fragments rapidly oxidize in the steam, would use highly reactive metals with a that could be attributed to a steam explo- initiating a much more energetic chemical hydrogen solubility characteristic that in- sion event. This was consistent with the explosion. creases with temperature. very weak events that have been observed Fauske: Essentially, there are three in experiments with oxidic core materi- You mentioned that this book summarizes types of organizations that need to con- als. In this regard, it should also be noted the technical bases and provides a common sider the possible role of vapor/steam that even these weak events have been ob- means to discuss and address this issue for explosions for individual designs: the served only when they are triggered by ex- various designs. Could you discuss how reactor vendor/designer, the operating plosive external triggers, such as blasting vendors, utilities/operators, and regulatory organization—for example, a utility, na- caps or exploding wires. agencies could perform their evaluations tional laboratory, industrial laboratory, or for the different types of reactors? university—and the regulatory agencies. Since the strength of a steam explosion Henry: The differences in reactor de- Of these, the reactor vendor needs to ad- would be determined by the amount of signs require that the information from dress the design of interest. The operating molten core material mixed in the water, the experimental technical database be ap- organization may have more than one de- what controls how much molten material plied in a different but consistent manner sign under its responsibility, and the reg- could be mixed with the coolant for water- for each design. For example, the spectrum ulatory agency needs to have a consistent cooled reactors? of accident conditions for liquid metal re- methodology that is applied to all designs Fauske: Experiments show that when actors that are oxide fueled does not sat- under its purview. With varying needs for the molten material drains into water, isfy the interface temperature principle. consistency in the approach, each organi- it breaks up into capillary-size droplets, Furthermore, the metallic uranium and/ zation has somewhat different needs for about 1 cm in diameter, and, simultane- or mixed uranium-plutonium metallic- the experimental database. ously, steam is generated due to film boil- fueled reactors would melt and shut down An illustration of how the experimental ing—very slowly compared to the genera- long before they would be at a sufficient bases need to be understood and used is tion rate during an explosion—around the temperature to satisfy that principle. the SL-1 accident that occurred in January

32 • Nuclear News • August 2017 www.ans.org/nn Interview: Fauske and Henry 1961, early in the life of commercial reactor programs. The accident is discussed in the book, but here it is sufficient to note that the reactor was fueled with metallic uranium-aluminum alloy fuel with aluminum cladding and cooled by water. The accident was due to a fast reactivity ramp rate induced by an undefined mainte- nance error. A steam explosion resulted, which also had a chemical component that caused major damage to the reactor and fatally injured the three-man operating crew. (Chapter 6 provides a discussion and evaluation of the accident, as well as evaluations of the BORAX-I and SPERT-I reactor experiments.) At the time of the Reactor Safety Study (WASH-1400), the sequence of events associated with the SL-1 explosion was used to structure a conservative probabilistic approach to the possible radiological consequences of a commercial reactor accident. The experimental technical bases now define the importance of aluminum in the accident, as well as the equally influential reactivity ramp in terms of melting the fuel within a fraction of a second, in the presence of water, and driving the fuel temperatures to values in excess of 1,400 K. Henry: With this knowledge base, a reactor vendor for a water-cooled commercial power reactor could begin to address the steam explosion issue by stating, “Our design has no aluminum in the reactor core, and it is impossible to melt the oxide fuel in the presence of water.” From this foundation, the vendor’s evaluation could progress to how an accident sequence could melt fuel, how rapidly the molten core materials could be brought in contact with water, how much molten material could be “coarsely mixed” with water, and what the bounding energy release from such an interaction would be. An operating organization would likely use the experimental bases in much the same way for a power reactor, but if the organization were responsible for a special- purpose water-cooled reactor with aluminum in the core, there would be a need to evaluate which accident sequences could melt the fuel and what the maximum temperature of the fuel would be in the presence of water. Would it be essentially the melting temperature of the fuel, or could it be significantly greater? The peak temperatures of the fuel and the extent of rapid oxidation in a steam explosion would need to be assessed for such designs. Regulatory agencies may or may not have all three types—liquid-metal reactors, commercial power reactors, and special-purpose reactors—within their purview. As a result, those individuals responsible for severe accident evaluations within the agency would need to be famil- iar with and use the entire experimental basis in a consistent manner. NN

August 2017 • Nuclear News • 33