ML20217D723

Discusses thermal-hydraulic five-yr Research

ML20217D723
Person / Time
Issue date:	09/06/1996
From:	Taylor J NRC OFFICE OF THE EXECUTIVE DIRECTOR FOR OPERATIONS (EDO)
To:	Shirley Ann Jackson, Rogers K, The Chairman NRC COMMISSION (OCM)
Shared Package
ML20217D728	List: ML20217D723 ML20217G453
References
NUDOCS 9701090198
Download: ML20217D723 (31)

Text

_ _ . . ___ _ _ ._ _ __ ____ _-_

g

["*g* UNITED STATES g 4 ,

. g V

j2 NUCLEAR REGULATORY COMMISSION WASHINoToN, D.C. 20h0001

• ...+! September 6, 1996 MEMORANDUM T0: Chairman Jackson Comissioner Rogers Comissioner Dieus Comissioner Diaz Comissioner McGaffigan FROM: James M. Taylor '

ExecutiveDirytrfor rations

SUBJECT:

THERMAL-HYD LIC FIVE-YEAR RESEARCH PLAN Thermal-hydraulics has been a major area of nuclear safety research since regulatory research began at NRC. Traditionally, thermal-hydraulic research has had two principal aspects: developing system-level computer codes and conducting of both large- and small-scale experiments. The staff's use of NRC-developed thermal-hydraulic codes has been an integral part of the licensing process, and these codes have provided the NRC with the ability to l perform independent analyses as mandated by the Energy Reorganization Act of 1974 (P.L.93-438) and as expected by the public. Over the past two decades,

! the NRC has spent, and is continuing to spend, substantial resources to demonstrate that its thermal-hydraulic codes are valid to analyze complex transients, accidents, and other off-normal conditions. In addition to improving and validating these codes, the NRC has performed, and most recently, is performing, in support of the AP600 certification, sufficient thermal-hydraulic experiments of various sizes and scales to gain an understancing of thermal-hydraulic phenomena. These tests have aided in improving NRC's analytical methods and in assessing the design, testing, and analysis of vendors and licensees.

The safety issues that drive the need for thermal-hydraulic research include:

Operational events and operational concerns continue to be of safety importance both domestically and internationally. These events require analysis by the NRC to understand their potential safety and generic implic .tions. These events and conditions include: BWR oscillations, steam generator tube rupture, PWR RCP seal failure, cooldown by natural circulation, station blackout, BWR vessel thermal stratification, boron .aixing during ATWS, pressurized thermal shock, inter-loop blowdown, and performance of safety features.

International thermal-hydraulic research continues, using integral test facilities such as ROSA (Japan), BETHSY (France), and PIPER (Italy). There is a potential for new data from these test facilities to alter our present understanding of accident scenarios.

This could lead to the reevaluation of design margins or procedures.

CONTACTS: M. Wayne Hodges, 415-5728 Thomas L. King, 415-5790 Farouk Eltawila, 415-5741 Joseph M. Kelly, 415-6295 Pre-decisional - For Internal Use Only O.1O3 M '

. 2

• Requests from the nuclear industry to modify operating licenses ,

(e.g., power upgrades), technical specifications, and emergency I procedures will continue to be received and will require a capability for independent analysis.

1 Risk-inforned regulation is becoming more important in the agency's l decision-making process. For the agency to understand the risk significance of various sequences, those sequences must be properly analyzed with a code that is robust and fast running.

• Applications for the certification of new designs may be received in the future. An advanced analysis capability that is flexible enough to be readily modified to account for new design features is needed to facilitate the agency's review.

The objectives of this paper are to inform the Commission of the staff's plan to improve and maintain its capability in thermal-hydraulics, including:

1. The major goals (near-term and long-term) of the thermal-hydraulic research program (THRP);

2. The specific activities associated with maintaining core competency in the areas of thermal-hydraulics, reactor physics, and plant-transient analysis codes, including international leadership and l cooperation; l

! 3. The staff's plan to develop a state-of-the-art plant transient code to replace current codes;

4. The experimental programs to obtain. fundamental data and information to support the development of advanced thermal-hydraulic models.

BACKGROUND:

After the NRC promulgated the revised ECCS rule, both the funding and prominence of thermal-hydraulic research at NRC declined. The recent advent of advanced light water reactor designs emphasized the importance of maintaining a viable, world-c ass thermal-hydraulic research program, including developing and assessing the thermal-hydraulic computer codes (the codes), and has resulted in our rebuilding our thermal-hydraulic capability.

The basis for the existing thermal-hydraulic codes was developed 20 to 30 years ago to analyze large break loss-of-coolant accidents using coding architecture and numerical methods that are now obsolete. Operating experience and ap,lications for passive reactor designs have demonstrated the need for a wider range of capability. This, in turn, led the code developers to modify the codes in an ad hoc fashion, which resulted in difficulty in preparing or modifying plant input decks, difficulty in interpreting the results, and frequent user intervention during the simulation of a transient because the codes are slow running Given our extensive data base and knowledge of numerical modeling techniques, combined with a new generation of high speed parallel computers, these codes no longer provide the best tool for

4

. 3 the job. Moreover, maintaining and running the existing codes is excessively 4 labor intensive, requiring many skilled people at different national laboratories. NRC can restructure the existing codes into a single code that is faster, more robust, and more user-friendly, which provides substantial saving in staff and contractor time to use and maintain the code.

DISCUSSION:

A large amount of thermal-hydraulic research results now exist, and this plan is intended as a needed step to the critical and focused examination of the NRC's future needs. These needs, including the analytical tools, experimental data, and staff and contractor capability, are presented below.

i The NRC is recognized as a leader in nuclear safety and is called upon to i address issues that are facing the industry domestically and internationally.

To sustain and enhance these capabilities, a stable, challenging research environment must be maintained. This includes providing stable long-term funding and interesting work to retain talented researchers. This section

contains an analysis of the issues associated with developing of long-term research to maintain technical expertise, including maintaining, updating, and restructuring the codes and maintaining certain experimental facilities. This 4

plan reflects the comitment made in a memorandum dated June 30, 1994, to the Commissioners.

Goals of the Thermal-Hydraulics Research Procram:

The near-term goals of the thermal-hydraulic research program are to:

I

1. Develop and maintain in-house capabilities:

a) Perform plant transient analyses using the current thermal-

! hydraulic codes and modify and assess the codes as necessary.

b) Participate in the development and evaluation of experiments needed for code development, assessment, and improvement.

c) Provide the technological bases for regulatory decisions involving thermal-hydraulics.

2. Maintain the exirng NRC plant transient analysis codes (RELAP5, TRAC-P, TRAC-B,RAMONA). In this context, maintenance not only-3 includes correcting errors identified by users but also includes i needed development, improvement,.and assessment.

The long-term goals are:

3. Maintain some experimental capabilities to address phenomena relevant ,

to nuclear safety and to provide validation data to cover plant parameter ranges of interest. This includes continued support of the domestic experimental programs at Oregon State University (OSU),

Purdue University (PUMA), and the University of Maryland (UMD). We will continue to interact with other international programs, e.g., i 1

t

- .w-

. 4

-ROSA (Japan), and BETHSY (France), to share information and participate in coo)erative programs of interest to the NRC. We also plan to maintain t.1e Code Applications and Maintenance Program (CAMP).

4. Combine the different modeling attributes embodied in the RELAP5, TRAC-P, TRAC-B, and RAMONA codes into a single state-of-the-art computer code. This will be accomplished by capitalizing on lessons learned from previous code development )rograms and exploiting new technology that has been developed or t1at is evolving, e.g., .

parallel computing environment, advances in modeling, and computation of two-phase fluid dynamics. In addition, the staff will remain cognizant of developments in computational fluid dynamics technology.

, The most successful of these technologies would then be considered l for incorporation in the plant transient code or as a stand-alone tool to-address specific issues that require such complex l computational tec1nology.

i

5. Incorporate user experience into an expert system. Such a system can advise the user on the type of nodalization to use, the uncertainty in the results, and can provide ready visualization of the processes that occur in a transient.

The staff's current estimate is that the near-term program will continue to meet NRC's existing needs but in less than an optimum manner. The long-term goals will continue to provide valuable information for the foreseeable future, and substantial improvements in code performance will be achieved in five years.

Maintainino Core comoetency:

To maintain competency, key researchers and the capability for plant analysis must be maintained. However, as a practical matter, this will happen only if researchers can be assured of a stable program and are involved in interesting and creative work, not just " maintenance of code." Cooperative research and agreements designed to spur collaboration with international organizations can also play a role. The intent is to maintain a cadre of researchers in-h: .se and at different universities and contractors' site. These experts wn1 be available to res >ond to technical questions as they arise and to develop acdels that can >e incorporated i..to NRC codes.

The research plan, discussed in detail in the Appendix, spans four different areas that are strongly related to each other: 1) reactor safety code development, 2) two-phase flow modeling, 3) thermal-hydraulic experiments, and

4) in-house capability. The following is a sumary of the plan.

Reactor Safety Code Develcoment Procram and Two-Phase Flow Modelina:

In the area of code development, the NRC plant transient codes embody much of the staff's knowledge about thennal-hydraulics and reactor physics; and they are essential to maintaining a strong and effective regulatory program.

Future uses of the cedes include support for increased power ratings, risk-

- __ -_~ - . . - - - . -. - ~ _ . -- - . -

, a 3

informed regulation, analyses of ope.ating events, and addressing issues that are facing the industry domestically and internationally. Currently, the NRC supports the develoament and maintenance of the RELAP5, TRAC-P, TRAC-B, and RAMONA codes at INEi., LANL, Penn State University, and BNL, respectively.

Since each code has its own mission and is maintained at a separate institution, there is little opportunity to consolidate the available talent base and costs. This plan will consolidate the thermal-hydraulic activities 4

and thus promote greater sharing of expertise and. experience.

As we reflect on our operational experiences with these codes, it has become increasingly evident t1at we need to consider modifying our overall thern.al-hydraulic code strategy and realign the objectives for each code so they better match today's and future needs. Furthermore, with the increased demand to reduce the budget, we have questioned our adherence to maintaining several codes that embody the same characteristics and diverge only on few models designed to address specific safety issues. It is our conclusion. (hat the current advances in software engineering, data distribution, expert systems and graphical user interfaces, machine intelligence, and knowledge of thermal-hydraulic phenomena will enable us to consolidate the NRC transient analysis capabilities into a single code without adversely impacting the existog capabilities.

Therefore, in FY 1997-1998, we will combine the different modeling attributes embodied in the TRAC-P, TRAC-8, and RAMONA codes into a single TRAC code as a

, first step in consolidation. In addition, over the next 3-5 years, we will

support, as our ultimate goal, the development of a state-of-the-art code and j data base to embody the capabilities in the combined TRAC code with those now existing in the RELAP5 code. This is summarized further below and in more detail in the Appendix.

Development of a State-of-the-Art Plant Transient Code:

As stated earlier, the underlying basis for the current codes was developed 20-30 years ago using coding-architecture that is now obsolete. Using modern 4

computer techniques (e.g., parallel processing) and more efficient user interfaces, naintenance and modification costs can be reduced, execution can be improved, and portab"ity can be enhanced. In the long term this will be cost saving.

I The consolidated code will be organized along functional lines, with a staff member responsible for each subject area. It is expected that, for some disciplines, the staff will initially need the assistance of consultants who specialize in a given area, but if that discipline is considered to be integral to developing expertise within the NRC, it is also expected that the staff will develop the expertise during the course of the project. The functional areas are:

• Physical Model Development - to upgrade or develop the constitutive

models necessary to enable the simulation of irrportant phenomena with good fidelity.

I

. 6

• Numerical Methods - to explore solution strategies for the two-fluid equation set and advanced matrix solution techniques.

• Data Base Structure and Code Architecture - the consolidated code will be modular and will have component-based input / output and physical models. The data structure will be chosen so as to not overly restrict development and modification activities and it will be easily amenable  !

to parallelization. '

• Neutronics - a 3-D coupled thermal-hydraulics, neutronics capability will be included in the consolidated code.

• Graphical User's Interface (GUI) - to make the code easier to use, from the perspective of both input and output. We would employ the services of a professional software development company. A staff member thoroughly familiar with reactor plant systems, and experimental facilities, and the input to a thermal-hydraulic code would work with the GUI developer to make certain that the product will meet our needs.

• Compatibility with other codes - such as the SCDAP code for severe l accident analysis or the CONTAIN code for coupled reactor system / containment analysis, in addition, a translator must be provided to preserve investments in the preparation of plant input decks for the current versions of the codes.

• Developmental Assessment - the consolidated code will be assessed against a wide variety of conditions to establish confidence in the results. Some of this activity can be performed within the agency.

Additional assessment, maintenance, user support, and archiving activities of the consolidated code will be the responsibility of a contractor. In preparing the detailed procedures for the assessment process, we will use the guidance provided in the agency Code Scaling, Assessment, and Uncertainty methodology to establish and measure acceptable limits for code accuracy and uncertainty.

• Documentation - documentation will be maintained contemporaneously with the de"elopment and mai*tenance of the code. User's manuals and other documentation will_ be in Hypertext Markup Language (HTML) to allow cro.

reference to documents that can be physically located in the same computer or another computer connected to a network.

To accomplish the above, a group of experts has been convened to identify approaches and to comment on a staff-developed plan to develop the code. This plan, which specifies the functional requirements for a consolidated code, will also be discussed at an OECD/CSHI workshop, hosted by the NRC in Annapolis, Maryland on November 5-8,-1996. We will discuss this plan and all facets of its implementation with the ACRS in early 1997, including the experimental programs discussed below. In addition, the implementation of this plan will include -the full participation of the NRC internal thermal-

s 7

1 j hydraulic users from NRR, AE00, RES, in the detailed design, development,

modification, and maintenance of both the existing suite of ccdes, as well as the next generation code.

4 Eggerimental Procrams:

In the area of experimental programs, we are planning to maintain a boiling-

water reactor test facility at Purdue University (PUMA), a pressurized-water reactor test facility at Oregon State University (APEX) and a B&W once through

! steam generator test facility at the University of Maryland (THECA) to:

• Perform independent confirmatory tests of an applicant's design to ensure that potential problems are fully explored i

i e Provide additional independent data in areas of particular importance for existing plants

• Provide a data base for thermal-hydraulic code validation In addition, where practical advanced instrumentation will be used to obtain reliable multi-phase mass flow measurements, void fractions, two-phase

density, and other needed information to improve basic modeling of the two-phase processes. It should be noted that modification to the OSU and PUMA

facilities may be needed to preclude any infringement on Westinghouse and GE proprietary design information that is incorporated in the design of these 2 facilities. An OECD/CSNI Specialists Meeting on Advanced Instrumentation and Measurement Techniques, hosted by NRC, will be held from March 17-20, 1997, in

Santa Bhrbara, California, which will identify state-of-the-art

' instrumentations as well as promising concepts.

In-house Capabilities:

In regard to in-house capabilities, it should be noted that the capability to analyze potential plant transients and accidents is necessary for carrying out

- the NRC mission. The need to perform plant transient analysis; e.g., design-basis accidents as well as non design-basis events, such as multiple system or component failures, common-mode failures, or operator errors, will not j

diminish with the completion of the certification of AP600.

Until recent years, NRC h;. maintained little or no in-house capability to e independently assess safety issues for advanced reactors or operating plants of either domestic or foreign design. As our budget is reduced, there will be more reliance on the staff to fill the gap created by reduced contracter support. Because of the complexity of the different thermal-hydraulic-and reactor physics issues, replacing the contractors' capabilities developed over-the past 20 years of research by in-house capabilities will require a commitment to maintain staff. In the last four months, RES has recruited four

engineers with experience in thermal-hydraulic phenomena, numerical methods, j and code development. In the fall of 1996, a graduate fellow will rejoin the staff after finishing her Ph.D. at MIT. Finally, a new graduate fellow will join the staff in the fall of 1996. NRC now has a nucleus for a good thermal-l hydraulic team and will continue to recruit and hire individuals with skills

/

! g we need, and to train young engineers -to replace those leaving the staff.

Establishing in-house capabilities supplemented by potential sources of-outside expertise will enable NRC to respond effectively to emerging needs.

Finally, we will keep the codes and the staff at a state-of-the-art level through participation in CAMP, OECD/CSNI, and other international programs; using the codes in response to specific requests; and checking them against new experimental data developed by the NRC and others.

RES0URCE COMMITMENT:

i l

The FTE and dollars shown in the FY97 and FY98 columns of Table 1 are included i in the budget request to OMB (FY98 Blue Book).. In preparing this research plan, the staff assumed these funding levels will be sustained through FY 2001. The following table identifies the staff and contractor resource estimates by functional area. Additional computer resources, not specifically '

addressed in this plan, will be needed to upgrade the NRC infrastructure that will be necessary to support this plan, in order to leverage resources, the plan seeks cooperative research on-thermal-hydraulics and on the state-of-the-art plant transient code with OECD member countries, the Commission of European Communities (CEC), JAERI and NUPEC in Japan, and CEA in France. Further leveraging of our resources will be sought by offering experimental research, such as that being conducted at Purdue University, Oregon State University, and the University of Maryland, as

. a quid pro quo to obtain relevant data from international experimental programs such as ROSA in Japan and BETHSY in France.

RELATIONSHIP TO COMMISSION'S STRATEGIC ASSESSMENT The Thermal-Hydraulic Research Plan will bi modified to take into account the Comission's final views on NRC's research program as set forth in the Strategic Assessment Issue Paper on Research (DSI 22).

COORDINATION A draft version of this plan was reviewed by selected members of the There 1-Hydraulic Expert Consultants (See memorandum dated January 9,1995, from James Taylor to the Commissioners). The consultants comments are reflected in the 2nclosed plan. In particular, we eceived comments from Professors Todreas (MIT), Wallis (Dartmouth), Ishii (Purdue), Banerjee (UCSB), and Reyes (Oregon State University)

Finally, the staff will meet with and brief both the ACRS Subcommittee on Thermal-Hydraulic Phenomena and the full ACRS on this plan. These meetings are tentatively scheduled for September and October 1996. We will incorporate the ACRS coments as appropriate and finalize the plan.

Attachments:

1. Table 1

2. Appendix cc: SECY OGC OCA OPA

- APPENDIX STRATEGIC RESEARCH PLAN FOR MAINTAINING CORE COMPETENCY IN THERMAL-HYDRAULICS It is essential for the NRC to maintain a high level of research expertise in thermal-hydraulics and reactor safety and to continuously improve our capability to analyze plant transients. The goal of this plan is to ensure that the NRC is able of providing thermal-hydraulic support for regulatory decision making. To meet this goal, the staff must have expertise in four areas that are strongly related to each other:

1. Reactor Safety Code Development

2. Two-Phase Flow Modeling

3. Thermal-Hydraulic Experiments

4. In-house Capability Each of these four areas is discussed below. For each area, an introduction is followed by sections on the significance of the problem, the identified needs, and a strategic plan for that area.

1.- Reactor Safety Code Development To audit vendor or licensee analyses of new or existing designs, to establish and revise regulatory requirements, to study operating events, and to anticipate problems of potential significance requires thermal-hydraulic analysis capabilities that are unique to the NRC. This is because the appropriate tools do not exist outside of the nuclear industry and entities within the industry have inherent conflicts of interest with the NRC.

Therefore, the NRC must have a capability for independent analysis, including both the tools and a cadre of experts capable of using them. The NRC currently relies on four different thermal-hydraulic system analysis codes.

Consolidating the modeling attributes embodied in these four codes into a single state-of-the-art code is the goal of the research effort detailed here.

This consolidation woul ' exploit new technology in the areas of parallel computing, two-phase t19,# modeling, and computational methods.

1.1 Code Development: Backgro"nd The NRC currently maintains four thermal-hydraulic computer codes of similar, but not identical, capability. For pressurized water reactors, the RELAP5 code provides a primarily one-dimensional representation of the flow field (some limited capability to model transverse flows is also available through the use of " cross-flow" junctions) and includes both point and one-dimensional reactor kinetics models. RELAPS is used primarily for small-break LOCA and plant transient analyses but lacks models needed for the analysis of large-break LOCA transients. Analyses requiring the modeling of multidimensional flows, and in particular large-break LOCAs, use the TRAC-P code. In principle, RELAP5 was supposed to be a fast-running " simple" code for long-

. 2 term transients, while TRAC-P would provide a more detailed description of the flow field and be suitable for faster (i.e., shorter) transients and also for benchmarking RELAP5. Over the years, this distinction has been blurred, and today many of these two codes' capabilities overlap, yet the two codes often 3 use different constitutive models for the same phenomena.

for analyzing boiling water reactors, the situation is somewhat similar. The RAMONA code provides a very simple one-dimensional representation of the flow field but contains a three-dimensional reactor kinetics package. For a more detailed representation of the flow field, the TRAC-B code was developed from the TRAC-P code. In addition to adding EWR-specific models (e.g., jet pumps),

the TRAC-B code implemented a different constitutive package and numerical scheme from its namesake; and since their separation, each of the two TRAC codes has followed its own independen3 path of development, it should be

' recognized that all four of these code: were initially developed for large main frame computers and have been modif f ed in a piecemeal fashion for um on workstations.

l 1.2 Code Development: Significance of the Problem As briefly outlined above, the NRC currently supports four different bermal-hydraulic analysis codes. The cost of this support is prohibitive, in terms of both budget and impact on our effort to rebuild and maintain a core competency in the area of thermal-hydraulics.

Part of the problem is the dilution of resources by supporting four codes; but of equal or perhaps greater importance, is the diminishing return on future research investment when it is invested to "fix up" old computer codes that are mired in obsolete technologies. The problems caused by having four codes are discussed below in three general categories: direct costs, impact on staff capability, and thermal-hydraulics code capabilities. The last subject area, thermal-hydraulics code capabilities, contains more detail and is further subdivided into four areas: maintainability, code accuracy, code speed and robustness, and user friendliness.

1.2.1 Direct Costs Direct costs are the support needed to maintain four code development and maintenance teams at three DOE laboratories and one university. Other direct costs accrue because of the nature of doing things in quadruplicate. For example, as part of the ALWR program, NRC funded both INEL and LANL to develop AP600 input decks for the RELAP5 and TRAC-P codes respectively. Concurrently, NRC Research also funded BNL to develop SBWR input decks for the RELAP5 and RAMONA codes. NRR often faces the same duplication of costs in needing plant decks for both RELAP5 and TRAC-P. These costs are not limited to just the initial input deck development, but they continue as the input decks have to be " maintained

• as the code input description changes with more recent code versions. Also, part of maintaining a thermal-hydraulics code is ensuring its simulation fidelity through the process of developmental assessment. Again, this effort must be duplicated as each code must be assessed for the complete range of phenomena over which it will be applied and not just for those that

3 are unique to its particular application. Finally, as will be discussed below under thermal-hydraulic code capability, the archaic nature of the architecture used in these codes makes their maintenance, finding and correcting user-identified errors, much more time-consuming (by a factor of 2 to 4) than for a well-engineered software product.

1.2.2 Staff Capability One of the goals is to upgrade staff capability to a world-class level of expertise on thermal-hydraulics. Splitting up our efforts into four parts is not an efficient-way of achieving this. First, the staff must be trained not only to use four different codes but also to understand their numerical and-physical models (and their underlying assumptions and limitations). Second, the existence of four codes means that there are (at least) four different contracts that need to be managed with all the associated paperwork and contractor interaction that siphons off some of the available staff resources.

l Finally, the upgrade of staff capabilities is-further impeded by the fact that l

all four of these codes are very difficult to use, both in the sense of input deck preparation and the interpretation of results. Although building a graphical user's interface is a priority item, a good interface will include

, code specific features such as automatic user guidelines and on-line help.

This will be expensive and cannot be done four times, 1.2.3 Thermal-Hydraulic Code Capabilities The subject of thermal-hydraulic code capabilities is divided into four parts:

maintainability, code accuracy, code speed and robustness, and user friendliness.

1.2.3.1 Maintainability As regards maintainability, one overriding factor, that also affects the other three areas, is simply the age of the codes. These codes were developed in the 70s, long before the revolution caused by the introduction of high-performance workstations and memory that is cheap, fast, and abundant.

Consequently, these codes were developed with an architecture aimed at optimizing performance on obsolete machines that were severely limited in memory. To overcome these memory limitations and allow dynamic memory allocation, the code developers were forced to employ elegant programming-styles (such as ' container erays" and ' bit packing") that have severely compromised readability, maintainability, and portability (e.g., separate code versions with machine-dependent options for different types workstations).

Engineers then spend their time not resolving fundamental deficiencies in the numerical and physical models but rather trying to decipher cryptic coding and work around the limitations inherent in the data base structure. One result of this is that current efforts to update models in these codes for advanced light water reactor analyses are costing several times more than necessary in terms of both time and money.

/

4 1.2.3.2 Code Accuracy Code accuracy cercsrns the simulation fidelity of important phenomena for both reactor systems and experimental facilities over the full range of parameters for which these >henomena are expected to occur. The majority of the constitutive paccages are different for all four codes, even though most of .

the phenomena are the same. As noted above, the code architecture makes it very difficult to modify the models in these codes (without introducing a large number of ' bugs"). Furthermore, the code models have been hardwired into a package of correlations, using smoothing functions to minimize discontinuities between correlations and explicit ramming functions to solve difficulties caused by the interaction of the physical models with the numerical scheme. Consequently, the numerical solution algorithms and the physical models are not separable, so that improving the physical behavior of one model can degrade the overall performance in unanticipated ways.

Consequently, there is inherent difficulty to modifying the code to upgrade the physical models to keep them state of the art.

Code accuracy is further adversely impacted because the majority of the physical models to be found in the literature were not formulated to be compatible with the framework of a two-fluid code. Also, the models are i developed to be applicable to one regime and not as part of a consistent

! package, leading to a patchwork quilt of correlations stitched together out of l expediency. Therefore, the current sets of constitutive relations do not take full advantage of the current data base, which leads to a larger degree of uncertainty in our calculations and to potentially erroneous calculations.

1.2.3.3 Code Speed and Robustness The real time required to simulate a transient is a product of both the code's speed and robustness, both of which greatly impact the efficiency of the analyst using the code. Current codes are often poorly structured, because new features were often added in a quick "fix up' mode, so that the resulting coding is very inefficient. Also, complex data structures, resulting from optimization for machines with small memories, impede the ability to apply new and potentially more efficient matrix solution algorithms, as the programming effort (and the probability of introducing errors) is enormously increased These same factors also limit the potential of present codes to take adnntage of one method to speed up codes by parallel processing.

Time-step size is the other factor affecting code speed, and it is primarily governed by stability and convergence considerations. A systematic effort will be needed to trace the source of these limits so that the efficiencies of more implicit schemes can be realized.

Robustness concerns the ability of a code to calculate a given transient through to completion without user intervention. When the code fails frequently and is restarted by the user with a different time-step strategy, it takes longer to reach the end of the transient. Code speed and robustness affect all users but are of particular significance to those conducting PRA studies, as they must run large sensitivity matrices of calculations.

5 1.2.3.4 User Friendliness User friendliness concerns the degree of difficulty one encounters in using a code, for both the laborious task of input decs preparation and the equally daunting task of interpreting the output. Overlaying both of these issues is the so-called ' user effect," that is, the likelihood of different code users getting significantly different results for the same transient even though using the same code.

The current codes require monumental efforts to prepare the input decks and l often put a large burden on the user, in the name of providing flexibility, by giving the user too many input options and no on-line guidance. Further, the current codes have demonstrated a distressing tendency to produce results that are time-step dependent, and the time-step control is largely left in the user's hands. Finally, interpreting the results has become somewhat easier because of the development of back-end interfaces, XMGR5 for the .. LAP 5 code and XTV for the TRAC-P code, but much remains to be done to bring this to the current state of the art in the software industry. Again, dividing resources between multiple efforts not only dilutes the effort but also makes the user's task more difficult as multiple code interfaces, each with its own philosophy, must be learned.

In summary, our current suite of thermal-hydraulic analysis codes suffers significant deficiencies with respect to the current state of the art in terms of: programming style, numerical techniques, the two-phase flow model, the reactor kinetics model, the constitutive relations, and user interfaces.

( Correcting these deficiencies is greatly en: umbered, if not prevented, by both I

the multiplicity of codes and by the difficulty of modifying these codes because of their antiquated programming styles.

1.3 Code Development: Identified Needs As regards the tools that will be needed to provide the NRC with the necessary analysis capability, the primary need is for a system thermal-hydraulic code applicable to current generation PWRs and BWRs for both large- and small-break LOCAs and for operating transients. This basic capability must be modular in nature to allow for fut :e enhancements that might be needed to accommodate other designs, such as advanced passive LWRs or university research reactors.

This basic system thermal-hydraulic analysis capability also ncads to be con.patible with other codes to ;.erform coupled reactor system / containment, coupled thermal-hydraulics /neutronics calculations, and coupled thermal-hydraulic / severe accident calculations. Finally, though our present analysis tools meet some of these needs, significant deficiencies exist and upgrading is needed in several areas.

. Accuracy: Present calculational uncertainties are larger than our data base warrants, possibly leading to overly conservative calculations.

. Speed and Robustness: Determining uncertainties requires the simulation of a large number of transients. These runs need to execute quickly and without frequent " crashes" that require user intervention.

. 6

• User Friendliness: Both pre- and post-processing tools need to be l upgraded to make input deck preparation and modification simpler and less sensitive to ' user effects,' as well as to make the code results easier to comprehend.

-Along with the above computational tools, a core competency in thermal-hydraulic code development and reactor safety analysis needs to be rebuilt and subsequently maintained. Experience over the last several years has shown that trying to rebuild this capability in a ' crash program

• to meet a specific need, e.g., the analysis of advanced passive reactor designs, is very costly in terms of both time and money.

Tc effectively and efficiently meet the future thermal-hydraulic analysis nesds of the NRC, a sustained long-term effort is needed. The effort envisioned by the staff would entail the development of a next-generation reactor safety thermal-hydraulic analysis code with a viable experimenta, program and the development of a world-class thermal-hydraulic research team wit l11n the NRC. The proposed research is outlined below showing how these three elements can be woven into one cohesive program.

1.4 Code Development: Strategic Plan As briefly outlined above, the NRC currently supports four different thermal-hydraulic analysis codes involving three different DOE laboratories and one university. The continuance of this situation indefinitely - in a future of declining budgets - is clearly untenable as ever-higher fractions of the available research resources would be needed to maintain outmoded technologies with little or no advancement to kee) u) with the state of the art. To effectively address this situation, )ota a short-term and a long-term strategy are needed.

The short-term strategy is to combine the capabilities of the codes so that they can be used to meet NRC's current analysis needs in a more efficient manner. To meet the objectives of the short-term strategy, the two versions of the TRAC code will be merged and a 3-D neutronics package will added during the next 2 to 3 years. The resulting single code will replace the RAMONA, TRAC-8, and MAC-P codes for 1 rge-break LOCAs, ATWS, and reactivity accident calculations, leading to a reduction in maintenance costs. Though some modernization of the TRAC architecture will have been accomplished during this effort, the combined code will retairi the ' procedural' structure of its progenitors as opposed to a thoroughly modern ' object oriented' architecture.

After the consolidation period is complete, the combineo TRAC code will be put in a maintenance mode until its replacement by the next generation code.

Concurrent with the above consolidation effort, a general purpose thermal-hydraulics graphical user's interface will be developed by a contractor.

Also, the RELAPS code will be maintained as the tool for analyzing small-break LOCAs, operational transients, and passive ALWRs. This maintenance period for RELAP5 will be for five years, during which time the development of the next generation code (see below) will have progressed to the point that the capabilities of RELAP5 and the consolidated TRAC code will have been recovered v

1 7

and they can be replaced. After two years of tr. ting and assessment, maintenance of the TRAC and RELAP5 codes will be discontinued, j The long-term strategy will result in a thermal-hydraulic systems code employing the following elements:

. Start with the capabilities of the current four codes,

. Implement modern code architecture, l

. Upgrade the numerical solution scheme, l

• Improve the two-phase flow model,

. Improve the constitutive modols and correlations, if needed, and

• Improve the user's interface and reduce the magnitude of the user effect.

In this way, the different modeling capabilities embodied in the RELAPS, TRAC-P TRAC-B, and RAMONA codes will be combined into a single state-of-the-art computer code, exploiting new technology in the areas of parallel processing, two-phase flow modeling, and computational methods, thereby enabling us to ca)italire on lessons learned from existing code development without suastantially changing models that are shown to be acceptable. A brief description of these elements, the rationale for their inclusion in a new thermal-hydraulic analysis code, and their relationship to other elements of the long-term strategy is given below.

1.4.1 Moders. Code Architecture in the future, the outmoded programming techniques used for the current codes would continue to hinder our efforts to maintain a state-of-the-art analysis capability. Implementing a modern code's architecture is essential, one that would have the following attributes.

• Adapts easily to a parallel processing environment (increased speed),

. Is highly readable and has a data base that is easy to modify (minimized maintenance / develop......t costs) .

• Maximizes portability and minimizes machine and compiler dependency, e Keeps the numerical scheme and constitutive models separate and employs a component-based structure for the physical models (easy to upgrade numerics and models).

Implementing a modern architecture consists not only of using a modern programming language, either C++ or Fortran 90, but also of using a modern data base and modular structure. To make the next generation code truly modular, that is, modular by both component and function, the modern software development paradigm of an object oriented programming will be adopted. To

1 8

accomplish this, it will be more efficient (and produce a higher quality product) to 're-engineer

• the models from the existing codes into a new architecture, than to try to retrofit a new architecture onto an existing code.

In addition, substantial im)rovements could be made in significant areas: the numerical scheme, the two-plase flow model, the physical models and correlations, and the user's interface, in some cases, the technology needed to make these improvements is readily available and requires an implementation effort as opposed to a research effort. These upgrades will be implemented directly into the systems code. In other cases, a significant component of the technology remains to be developed and small exploratory research efforts will be launched to develop this technology. In the description of the activities given below the distinction between evolutionary improvements ,

(onesthatrequireimplementationonly)andmorerevolutionaryefforts(that  ;

require development beyond the current state of the art) will be indicated.  ;

1.4.2 Numerical solution Scheme l I To improve code speed and robustness, it will be necessary to upgrade the numerical solution scheme. At present, long-term analyses using the RELAp5 code are hindered by the explicit nature of the numerical scheme as the time step is limited by the Courant condition. Often, the resulting time step is on the order of 0.005 seconds. A more implicit scheme could employ time steps on the order of seconds or even tens of seconds, thereby greatly reducing the time needed to complete a calculation. A more implicit scheme has been successfully implemented in the TRAC-P code, however, its performance can be degraded by ' problem numerical stability," leading to time-step reduction, code failures, or numerically driven oscillations. Improvements in the numerical scheme would lead to reduced run times and less frequent code crashes requiring user interventior,.

The SETS (Stability Enhancing Two-Step) method from the TRAC-P code will be used as the base numerical solution algorithm. Also, a systematic effort will be conducted to uncover and eliminate the root causes of ' numerical events.'

In particular, efforts will be devoted to the handling of phase disappearance, the appearance of non-condensable gases, ' water packing,' and intelligent time-step control. These efforts are evolutionary in nature and will include investigating the use of methods such as the stiffly stable' schemes, higher order differencing schemes (for tuermal stratification and two-phase level tracking), and multidimensional solution schemes that have a higher level of-implicitness.

Although the thermal-hydraulic codes that are discussed in this paper are essential to understanding fluid system performance, in certain situations it -

is important to ur.derstand the phenomena that occur within particular components (such as steam generators) themselves. For this purpose, our system-level codes are not well suited, and we need to have a tool available that has different capabilities. These codes are known as computational fluid dynamics (CFD) codes, and they are used in many commercial applications, such as chemical plants, combustion systems, and aerodynamics, to provide detailed information about fluid behavior,

i .

! To support the agency need in this area, a series of pilot projects will be initiated to determine how this technology can best fit into the agency's toolbox. These projects may be carried out in collaboration with l

universities, outside corporations, international organizations, and in-house staff, and they will consider various CFD products that are available from both U.S. Government and commercial sources. We will evaluate the ability of )

different codes to track complex fluid interfaces in viscous multi-phase flow l geometries, as well as their ability to model mixtures of vapor and liquid that contain small bubbles or drops. The evaluation is expected to improve our insights in fluid dynamics and might eventually lead to adopting of similar methods for interface tracking and adaptive meshing, particle and lattice gas methods, and sub-grid scale modeling in the system codes. Even if

, these techniques are not incorporated into the system codes, we expect _that the pilot programs will identify the appropriate CFD technology that the agency should use for component-level analysis of fluid-dynamics nroblems, such as in steam generators.

1.4.3 Two-Phase Flow Model In concert with improving the efficiency of the computational tools, it is necessary to ir.; prove the fidelity of their simulations as well, which will require improving the degree of sophistication in the representation of two-phase flow, immediate gains can be made by adding a droplet field to the I current two-fluid model (as was done in the COBRA / TRAC code, developed l initially by the NRC and now used by Westinghouse). The addition of a droplet field allows for a much improved representation of the two-phase flow field for regimes in which the liquid phase has two characteristic velocities, such as the annular / mist flow regime. Upgrading the two-phase model from the two-fluid to the three-field formulation is an evolutionary effort and will be incorporated in the systems code.

Of equal significance would be the replacement of flow regime maps used to characterize the nature of the two-phase interface with a dynamic flow regime model. Here, the traditional flow regime map would be replaced by introducir.; interfacial area transport equations whose source / sink terms represent the processes that govern the creation or destruction of interfacial area (e.g., bubbles en escence or break up). Thus, the empiricism inherent in the modeling of two-phase flow would be moved to a more fundamental level.

Ttis technology is far from being fully developed and must be considered revolutionary in nature, especially for two-phase flow in complex geometries such as reactor coolant systems. However, the instrumentation has now matured (see the discussion under two-phase flow modeling in Section 2) and an experimental program, going hand-in-hand with the effort to improve the computational model, would greatly enhance our predictive caoability.

1.4.4 Models and Correlations Even with a more fundamental model for two-phase flow as described above, a systems thermal-hydraulic analysis code will retain a set of models and correlations that includes hundreds of empirical relations. At present, the models and correlations employed in these four codes are inconsistent (i.e.,

different models are employed for the same phenomena in different codes),

/

10 often employ ad hoc formulations or undocumented smoothing functions, and do not reflect the knowledge embedded in the existing experimental data base.

Together with improving the description of two-phase flow (see above), some of the greatest gains can be realized through a comprehensive upgrading of the models and correlations. i This effort will include the estabitshment of an electronic data base that contains the supporting empirical evidence for each of the podels or correlations for phenomena judged to be of high importance. Then, a quantitative review of the applicability of the models/ correlations in the current codes will be conducted. For these high-ranked phenomena, if the accuracy of the present model is found to be insufficient, either a new model will be developed from the existing data base or separate effects tests will be conducted to generate the needed data base as necessary. In this approach, there are two features that have not generally been present in the past:-(l) the needed models will be developed within the framework of a two-fluid code and (2) the associated data base will become part of the code documentation ,

and electronic archive such that it will be readily available for assessing l future model upgrades. l This effort to upgrade the models and correlations is evolutionary, if the ,

research into modeling two-phase flow through the use of interfacial area '

transport equations has promising results, an experimental program will be needed to develop the necessary constitutive relations as part of the exploratory research effort.

~

One of the key processes in assessing the system code capability for transient analyses of nuclear reactors is establishing of the code scale-up capability to plant conditions. Although it would be most desirable to verify code performance against actual plant transients and accidents, this is usually impractical. Instead, the codes are verified primarily by comparing their predictive results against the measured results of scaled experimental test facilities. In order to establish that the code behavior at small scale is applicable to analyses of the full-scale reactor systems, three important activities need to be performed:

1. The code assessment team n t first assess the scaling bue for the various experimental facilities to ensure that the test equipment does not distort the phenomena of interest in a significant way.

2. The assessment team must then establish that the application of the code at the reduced scale of the test facilities does not violate any limits of applicability of any internal code models.

3. The assessment team must then establish that the code performance in predicting the behavior of the experimental test facility can be scaled up to the' full scale of the operating reactor.

De results of currently exuting phenomena identincation and ranking tables (PIRTs) mill be used to help establah priorities in upgrading the constitutive relations.

Ki

. 11 These three activities, when taken tcgether, are used to demonstrate that the code models and constitutive relations within the code, and the code as a whole, can be applied to analyses of tbn full-scale plant. As part of this thermal-hydraulic research plan, we will review scaling philosophies and programs used in the past and will develop a unified philosophy for addressing scaling effects as a part of overall code assessment.

1.4.5 Improved User's Interface The term " users' interface" essentially relates to the degree of difficulty encountered in using a code, for both the laborious task of input deck preparation and the equally daunting task of interpreting the out3ut.

Overlaying both of these issues is the so-called ' user effect,' t1at is, the likelihood of different code users getting significantly different results for the same transient even though using the same code. Current codes require monumental efforts to prepare the input decks and often put a large burden on the user, in the name of providing flexibility, by giving the user too many input options and no on-line guidance. Clearly.-the area of user interfaces-is one in which a large effort is needed to:

. Make input more ' hardware

• oriented instead of ' code' oriented. For example, a user would enter the pipe schedule and diameter instead of individual volumes and flow areas for computational volumes.

• Build user guidelines into the user interface so that default noding -

schemes are automatically generated.

. Provide greater guidance on the objectives and limitations of the user input option and provide more default settings,

. Implement more ' intelligent' time-step control algorithms decreasing the -

sensitivity of the results.to time step size, with an option for " hands-off" use.

. Make the post-processing tool more flexible and easier to use so that the analyst has more help when trying to interpret the code results.

The activity to improve the user's interface was started in FY-96.

1.5 Summiary of Code DeveWment Plan In summary, the proposed long-term strategy is to develop a single state-of-the-art code, ta(ing the best of all the available codes, using modern code development practices, and incorporating advances in modeling, numerical methods, and graphical interfaces from other disciplines. As discussed below, some research effort would be executed in-house, drawing on outside expertise of consultants, so that the resulting knowledge base would be developed and reside in-the staff.

2. Two-Phase Flow Modeling

i 12 The success and the quality of the future plant transient code largely depends on the availability of a significantly improved two-phase flow formulation and l constitutive relations supported by detailed experimental data. Therefore, this Research Plan calls for significant research effort in the-areas of two-phase flow modeling, instrumentation, and separate effect experiments that should be pursued systematically and with clearly defined objectives.

Phenomena identification and Ranking Tables will be used to determine the various characteristics and properties of models and processes that should be formulated clearly, on a rational basis, and supported by experimental data, for this purpose, s)ecially designed instrumentation and experiments are required that must )e used in conjunction with and in support of analytical investigations.

2.1 Significance of the Problem The weakest link in the two-phase flow formulation is the constitutive equations for the interfacial interaction terms. The difficulties arise from the complicated motion and geometry of interfaces in a general two-phase flow.

Furthermore, these constitutive equations should be expressed by the macroscopic variables based on proper averaging.

The interfacial transport terms are strongly related to the interfacial area l concentration and to the local transpott mechanisms such as the degree of )

turbulence near interfaces. The driving forces for the interfacial transport depend on the local turbulence, transport properties, driving potentials, and some length scale at the interfaces. This length scale may be related to a 4 transient time such as the particle residence time or to the interfacial area 4 concentration and void fraction.

One of the_ major difficulties in developing a reliable two-fluid formulation is modeling of the constitutive relations for the interfacial transfer of momentum and energy, which does not have a counterpart in a single-phase flow analysis. To mechanistically model the constitutive relations for the interfacial transfer and turbulent transfer in two-phase flow requires detailed local measurements of the interfacial area, interfacial velocity, phase velocities, and turbulence, which were not available until quite recently. In the last five years, there have been excellent advances in laal instrumentation technology for two-phase flow. These developments were due to advances in electronics, local multi-sensor techniques, and optical methods.

Now the local interfacial area concentration, void fraction, interface velocity, Sauter mean diameter, phase velocities, and turbulence in two-phase flow can be measured. These parameters give great insight into the interfacial transfer and turbulent transfer mechanisms. Many of the three-dimensional transfer phenomena can now be measured and quantified such that modeling of the constitutive relations for the interfacial and turbulent transfers becomes realistic.

The new approach for modeling of the interfacial structure that replaces the conventional flow regime maps and criteria should be one of the focal points of the research. The introduction of the interfacial area transport equation or multi-field approach is now possible. The modeling of-the interfacial structure is directly related to the foundation of the new two-fluid model, t

. 13 l 2.2 Identified Needs 1

! The conceptual models that describe the steady state and dynamic characteristics of structured multi-phase media should be formulated in terms of the appropriate field equations and closure relations. However, the derivation of such equations for the flow of structured media is considerably more complicated than for single-phase flow. In multi-phase or multi-com)onent flows, the presence of interfaces introduces great difficulty in the mat 1ematical and physical formulations of the problem. From the point of view of physics, the difficulties that are encountered in deriving the field and closure equations appropriate to multi-phase flow systems stem from the presence of the deformable interface and the fact that both the steady and dynamic characteristics of multi-phase flows depend upon the structure of the flow.

From the standpoint of analysis, there is a need for improved methods of accounting for-the structure and local phenomena in two-phase systems. From the standpoint of experimentation, there is a need for new and improved measurements for local phenomena to support constitutive equation development. ,

7 The two must proceed in concert for success in producing new reliable  !

computational methods. Also, development of advanced instrumentation development for two-phase flow systems is a necessary component of thermal-hydraulic research. The instrumentation is the basic tool for the fundamental experimental research focused on the important phenomena in two-phase flow.

2.3 Strategic Plan The strategic plan for the advancement of the state of the art in two-phase flow modeling contain. three complementary activities: 4

1. Use of advanced two-phase flow instrumentation

2. Performance of fundamental two-phase flow experiments

3. Development. of improved phenomenological models for each of these three activities, a list of proposed research efforts is given below.

2.3.1 Use of Advanced Two-Phase Instrumentation Some advanced instrumentation development will be included in the program as listed below:

• Multi-sensor conductivity probes for the measurement of local interfacial area, void fraction, particle size, and interfacial velocity, particularly for a boiling water system

. Measures of entrainment rate, deposition rate, and droplet size, in high

_ velocity two-phase flow.

1 .

14

• Measures of mass flux and vapor quality

.

• Measures of critical flow

• Flow visuklization and characterization of interfacial geometry

• Measures of liquid flow rate using modified magnetic flow meters or other methods l . Global void sensors The use of such advanced instrumentation will enable us to obtain data needed for model development and code assessment. l 3.3.2 Fundamental Two-Phase Experiments Using state-of-the-art instrumentation,_ fundamental experiments focused on the ,

important problems and phenomena can be studied and a database for a model development effort can be established. The following art some of the recommended experiments that will be part of the overall experimental program described in section 3.

• Interfacial area measurements focused on developing a data base for the coalescence sink _ term and disintegration source term in the area l transport equation. This should be performed for both vertical and horizontal flow at several hydraulic diameters.

• Dynamics and instability experiments for single phase and two-phase natural circulation.

• Flashing phenomena in stagnated fluid or in a natural circulation system

. Annular flow experiment focused on the entrainment rate, deposition rate, droplet size, film thickness, and interfacial shear 2.3.3 Development of Phenomenolog.cv. Models The proposed model development activities are listed below. Note that some of these efforts are dependent on the experimental activities regarding instrumentation.

2.3.3.1 Interfacial Area Modeling for predicting the thermal-hydraulic behavior of two-phase flows, the interfacial structure is one of the most important factors. Traditionally, the effects of the interfacial structure have been analyzed using the two-phase flow regimes and regime transition criteria. However, this traditional approach has a number of shortcomings. First, the flow regime

15 transition criteria are algebraic relations that do not fully reflect the true dynamic nature of changes in the interfacial structure. Hence the effects of the entrance or boundary cannot be taken into account correctly, nor can the gradual transition between regimes. Secondiv the method based on the flow regimetransitioncriteriaisatwo-stepmetho,dthatrequirestheregime-dependent closure relations for the interfacial area effects. Normally, the effects of these are imbedded in the correlations implicitly; therefore, the compound errors from this approach can be significant.

RES will develop an interfacial area transport equation for the first-order characterization of interfacial structures. For goou mechanistic modeling, it is necessary to study bubble coalescence and break-up criteria to get information on the maximum bubble size and bubble size distribution. These are important in the formation of a link between the flow-pattern transition and the characteristics of the interfacial structure, such as interfacial area concentration and void fraction distributions.

Bubble coalescence and break up arocesses are considered explicitly to develop a more mechanistic model. For t11s purpose, the use of an interfacial area transport equation for two-phase flow appears to be most suitable. The concept of the interfacial area transport equation was suggested by Ishii in 1975 and subsequently applied for annular mist flow by Kataoka and Ishii to predict the entrainment and deposition processes. The mechanism of the transition from bubbly to slug flow can be considered as the elimination of the dispersed phase by the coalescence mechanism, whereas in the annular to annular-mist flow transition the dispersed phase is created by the droplet entrainment process. The processes are almost completely in the opposite direction. Hence it can be concluded that once the rate processes of the coalescence and bubble breakup are modeled, the gradual transition from the bubbly to slug flow can be predicted through-the interfacial area transport equation.

2.3.3.2 Pilot Code Development Using Interfacial Area Transport Equation The effect of the interfacial area transport equation on the overall two-fluid model formulation and numerical solution method should be studied through a simple one-dimensional pilot code. This will give insight to the dynamic effects of the transport equation, stability of the differential equation system, accuracy of the constitutive relations, and efficiency of the numerical method.

2.3.3.3 Two-phase Flow Instability at Low Pressure At low pressure, two-phase flow systems tend to be quite unstable because of several mechanisms, in particular, natural circulation two-phase flow at a low pressure is highly unstable because manometric, density wave, chugging, and flashing-induced instabilities. This is because of the flow and void generation are closely coupled in a natural circulation system. Furthermore, because of the large density ratio between liquid and vapor, small fluctuations in heat transfer result in significant void fluctuations.

However, two-phase natural circulation is a key in most of the advanced LWR designs that use the automatic depressurization systems and depend on the

/

16 gravity-induced flow. Most of the existing studies have been performed for a forced flow system at relatively high pressure, hence it is necessary to carry out some basic research to understand these instabilities.

2.3.3.4 Constitutive Relation Development Constitutive relations and correlations are used in the two-fluid model to close the two sets of conservation equations of mass, momentum, and energy.

In particulur, the interfacial transfer terms couple the mass, momentum, and energy of phases. There are several areas in which improved constitutive relations can make a large difference in the accuracy and reliability of code ,

predictions based on the two-fluid model formulation, as follows.

+ Interfacial Heat Transfer at low Pressure: The current algebraic heat transfer model for the interfacial energy transfer is too tensitive to the instantaneous changes in the system pressure through the use of the saturation temperature of the interface, particularly at low pressure.

This leads to considerable fluctuating energy transfer between the liquid and vapor and leads to oscillatory void fraction predictions.

The actual physical process involves the transient thermal boundary layer development, which should exhibit some effects of time delay.

Either a time lag model that leads to a difference differential equation or an exponential relaxation model may be used to fix this problem. ,

. Interfacial Momentum Transfer: The constitutive relations for interfacial drag and shear for certain regimes require further study.

These are (1) inverted flow regimes in the post-dry-out region, (2) annular flow at high pressure, and (3) developing flow where void distribution changes rapidly.

• Thermal Non-equilibrium Model: Significant thermal non-equilibrium occurs during flashing, direct contact condensation, and post-CHF heat transfer. Among these, flashing and direct contact condensation are particularly important for advanced LWRs. A mechanistic model of flashing based on the nucleation site density is in the early stage of the development; however, it has the potential to eliminate the large uncertainty in the existing empirical correlation and the shortcomi.-,s of the thermal equilibrium model. The condensation of large voicies of steam with noncondensable gas by injected subcooled water is another important problem, yet there are no reliable models or data. Similarly, the condenration of steam with noncondensable gases in a heat exchanger or in a pooi of subcooled water is not well understood.

3. Experiments Experiments simulating reactor plant designs and their components are necessary in order to:

• Perform independent confirmatory tests of an applicant's design to ensure that potential problems are fully explored

17 l

• Provide additional independent data in areas of pi.rticular importance for existing plants

• Provide additional data for thermal-hydraulic code validation.

i These activities require that testing be conducted in scaled integral test facilities. When properly instrumented. these same integral facilities may be operated in a separate effect mode to provide more specific code assessment data and to help establish a data base for model development. It is necessary to continue staff involvement with integral facility experimental programs so that technical skills are not lost.

In 1976, the NRC created a Reactor Safety Data Bank to provide a central, readily accessible repository of qualified test results of tests performed in experimental facilities and reactors. These data were produced by experiments that took place over a period of several decades, in test facilities such as LOFT that cannot be replicated. It is therefore vital that the agency not lose either the data from the experiments or the information needed to accurately model the test facilities for code validation purposes. The staff is in the process of transferring the data bank, from INEL to an internal NRC computer system, and will ensure that both the data from the tests and the experimental facility configuration information are maintained for the use of agency thermal-hydraulic code developers and other code users.

3.1 Significance of the Problem large thermal-hydraulic experimental facilities are costly to maintain. There are several available around the world (e.g., in ilapan, Switzerland, France) that could probably be used if the need arose for integral experimeits.

However, smaller scale, university-run facilities provide a more economical alternative with the added advantages of maintaining expertise in nuclear technology in universities and a stream of trained graduates. Gaps in the knowledge of two-phase flow need to be filled in order to conduct regulatory analyses; this can best be accomplished with small-scale facilities in a research environment such as exists at universities.

There are three facilit 4s (OSU, Purdue, ar.d VMD) that have experimental equipment as well as a team of thermal-hydraulic experts. In addition to providing sup) ort so that these facilities and on-site teams can be maintained, t1e NRC should provide an environment in which research teams from other institutions will have access both to these experimental facilities and to the necessary facility sup) ort staff. In this way, these three experimental facilities will se shared between on-site and visiting research teams.

3.2 Identified Needs Code validation and a greater understanding of thermal-hydraulic phenomena depend on properly scaled, designed, instrumented, and conducted experiments.

The data base used to develop and assess the existing thermal-hydraulic codes ,

was developed in the 1970s. Because of their intrusive nature, and the long time delay, the instruments that were used were inadequate to provide

- - - - = .

. 18 sufficient data to develop models to represent the complex thermal-h draulic phenomena. New,lessintrusiveinstrumentshavebeenusedsuccessfuflyin other fields. Advanced instruments-can be used to obtain reliable multi-phase mass flow measurements, void fri.ctions, two-phase density, and other needed information to identify phase configurations and interfacial areas to improve basic modeling of the two-phase processes.

To develop a data base that is adequate for code validation and for developing l

of state-of-the-art models, the NRC must maintain the existing experimental

! facilities and u) grade their instrumentation as described above. These facilities can tien be used to obtain separate effects test data both through university research and international collaboration.

3.3 Strategic Plan for Experiments l

L The strategic plan for experiments provices for maintaining the three existing integral test facilities (OSU, Purdue, and UMD). In addition, to-enlarge the data base for code validation and model development, these facilities would be used to conduct separate effects tests.

For each of the three facilities under consideration, we have developed a preliminary list of experiments that could provide us with experimental information that is currently needed for future code development. We will continue to review this list as the development and maintenance efforts proceed, to ascertain whether new or different experiments are needed or whether the information is not needed or is available from other sources.

3.3.1 Separate Effect Tests: Purdue University's PUMA Facility The PUHA facility at Purdue was originally designed for the confirmatory integral test for the GE SBWR design. This facility has a large number of instruments and includes the capability for local void measurements and flow visualization. Each of the components displays some fundamental characteristics of various two-phase flow systems, it is quite possible to=

run the PUMA facility for various separate effect tests to obtain fundamental data focused on particular phenomena. The separate effect tests that can be performed without-any major ge metrical modifications are listed here.

3.3.1.1 Reactor Pressure Vessel (RPV) and Automatic Depressurization System (ADS)

. Single Phase Natural Circulation Benchmark Test:

Focused on the natural circulation rate, two- and three-dimensional energy distribution, and flows instability.

. Two-phase Quasi-steady Natural Circulation Test: Focused on void distribution, relative velocity, two-phase level, natural circulation rate, void generation by flashing, and various flow instabilities.

. Rapid Depressurization and Flashing Test: Focused on flashing phenomena and void generation, void distribution, relative velocity, transient

. 19 behavior of void fraction, flow, temperature and two-phase level, and flow instabilities.

• Critical Flow at Low Pressure Test: Focused on break flow and its  ;

measurement for both large (HSL, DPV, and SRV break) and small breaks.

. Downcomer Mixing Test: Focused on cold water injection into the RPV through GDCS, IC, or FWL nozzles, mixing of subcooled water with saturated water and the two-phase mixture, void collapse, condensation, reestablishment of natural circulation, and transition between single phase and two-phase flow.

. Boiler-Condenser Mode Operation Test: Using the RPV and ICS, the steady boiler-condenser mode of core cooling is studied. With limited l modifications, reflux ondenaar ma e operation is also possible.

3.3.1.2 Drywell Phenomena The major focus is steam mixing with noncondensable gas in the dry well. The inertia transition and plume regimes are studied separately. Another focus is the effect of the vacuum breaker operation on the noncondensable gas I distribution.

3.3.2 Separate Effects Tests: Oregon State University APEX Facility The APEX facility was specifically designed to obtain integral system thermal-hydraulic data for the proposed AP600 design. However, with improved instrumentation, separate effects data can be generated that would pertain to not only the AP600 design but to generic PWRs as well.

3.3.2.1 Flow Stability and Heat Transfer in Forced Flow and Gravity Driven System e Steam Generator Heat Transfer: Steady state and transient tests to improve understanding and modeling of heat transfer processes from primary to secondary.

. Two-Phase Natural Circulation: Perform natural circulation tests for the primary loop and eteam generator with reduced system inventory to identify the conditions for the onset of instability.

. Onset of Tube Voiding: Perform natural circulation tests with reduced primary pressure to study the onset of tube voiding and breaking of the natural circulation loop. The prediction of this phenomenon is important to the potential occurrence of thermal stratification in the cold legs for the A_P600 and for PTS in existing PWRs.

3.3.2.2 Critical Flow in Valves and Orifices Perform critical flow tests under multiple chocked flow conditions (resonance effects) in spargers and valves and under single choked flow condition such as breaks.

20 3.3.2.3 Thermal Stratification Construct a thermal fluid mixing map which describes the primary loop conditions under which cold leg thermal stratification can occur.

3.3.2.4 Two-phase Fluid Flow Pattern and Flow Pattern Transition in Complex Reactor Components

. Counter-Current Flooding Limit (CCFL): Identify the conditions at which the pressurizer cannot drain because of the CCFL at the surge line during operation of safety relief valves or the ADS systems.

Complement the system tests with air / water bench tests to permit flow visualization and characterization of flow patterns.

• Level Swell: Perform pressurizer blowdown tests to determine level swell and phase separation during flashing conditions.

3.3.2.5 Phase Separation in Tees Perform flow visualization and phase separation tests suitable for assessment or development of off-take model for geometries typical of hot leg / surge line and hot leg / ADS-4 conditions.

3.3.2.6 Multi-dimensional Turbulent Mixing Induced by Tube Bundle Boiling

• Determine the heat transfer characteristics, both in-tube and pool-side, for a heat exchanger submerged in a pool.

. Investigate thermal plume behavior in and around the submerged bundle and develop data on thermal stratification at the pool surface.

3.3.3 THECA program at University of Maryland Facility One of the characteristics of the thermal-hydraulic experiments for code assessments (THECA) program is the flexibility of the test facility, resulting in low operating costs that would allow performing extensive sequences of repeat tests. In addition, because of the proximity of the VMD to the NRc headquarters, we will be able to use the staff in executing the experiiuents, thus providing the staff with hands-on experience. The following are some of 4

the tests to be investigated under the THECA program.

. Liquid thermal stratification under vapor space--conditions for stable existence and for an onset of rapid condensation

. Single Loop Interruption / Resumption Mode--associated with natural circulation behavior.

. Single Loop Condensation Controlled Mode--originating from condensation of two-phase flow entrained over the candy-cane (B&W hot leg).

21

• Cold Leg Downcomer Flow Distribution--related to multidimensional effects in nearly stagnated system important for pressurized thermal shock (PTS).

4. In-house capability l It is essential for the NRC to sustain the highest level of research expertise l in thermal-hydraulics and reactor safety and to continuously improve our capability to analyze plant transient. To this end, a new direction has been set, one in which challenging research activities will be conducted in-house or in close collaboration with a contractor. Not only is this better for the agency's long-term interests, it is necessitated by declining capabilities at national laboratories and the declining budget for research, future thermal-hydraulic research activities will be focused primarily in three areas:

1. Reactor Safety Code Development: The next generation thermal-hydraulic code (see Section 1 of this appendix) will have strong involvement of the NRC staff, and in addition to providing the computational tool for the future, will provide the opportunity for our junior staff to become tomorrow's experts.

2. Reactor Analysis: The staff will use the current (and future) code to perform analyses of both plant transients and integral facility experiments, requiring the current staff training program to continue.

3. Thermal-Hydraulic Experiments: Although the experiments will not be conducted here, the staff will actively participate with university researchers to develop a data base sufficient for future model development (the fundamental tests described in Section 11) and model assessment (the separate effect tests of Section 111).

A stable long-term research budget is needed to accomplish the above, which will result not only in the development of computational tools and expanded data bases, but also in a research staff capable of meeting the agency's needs in the fut"re.

4.1 Plan One of the primary goals of this research >1an is the development of a world class thermal-hydraulics research team witiin the NRC. To do this, core competency in thermal-hydraulic code development and reactor safety analysis needs to be rebuilt and subsequently retained. The expertise required is above and beyond that resulting from a university nuclear engineering curriculum and can only be developed through performing research with the aid and supervision of a suitable mentor. To this end, two new branch members have been recruited; one with experience in numerical methods to supplement the two-)hase model development experience of a current senior staff member, and anotler with experience in reactor plant analysis. These three individuals will form the nucleus of code development and analysis teams.

Given the above goals and budgetary constraints, development of the next generation thermal-hydraulic code would be undertaken by a small, well-

/

22 organized team with less expenditure of resources and would have the crucial benefit of allowing the staff to develop ex>ertise for the future. This team would be organized along functional lines t1at could be pursued independently for example, physical model development, numerical solution, neutronics, data management and parallel processing, and development of a graphical user interface.

In functional ereas in which the staff is expert, the lead role would be l undertaken by the appropriate staff member. In other areas, when the expert I individual is from industry or academia, an NRC staff member would work closely with consultants, not as a project manager, but in an " apprentice" role. Such apprenticeships are designed to ensure a " technology transfer" between the outside consultant and the staff, so that expertise in each critical functional area would be developed in-house.

As for plant transient analyses .the current staff retraining program will continue and will be expedited by the addition of the new staff member. The program in two-phase fundamental experiments will provide an opportunity for the staff to collaborate with university researchers and develop expertise in the area of two-phase flow physics.

4.2 Near Term Plan

1. Continue training the staff to run and interpret our computer codes.

2. Recruit one more code developer to supplement the existing one, and recruit a staff member with analysis experience. This part of the plan is complete.

3. Investigate the use of a commercial contractor to maintain RELAPS and service the CAMP users (in lieu-of a national laboratory) for improved cost and performance.

4. Move RAMONA maintenance in-house.

5. Continue to analyze the systems tests from ROSA, SPES, and OSU in-house.

6. Continue international interactions on codes and data, domestic and foreign. Organize OECD/CSNI Workshop on the requirement for transient thermal-hydraulic and neutronic codes (to be held in Annapolis, November 5-8,1996).

7. Sponsor in-house courses and seminars and international workshops to hone and maintain skills.

4.3 Long Term Plan To achieve a state-of-the-art plant transient code and the associated expertise within the NRC requires a commitment to a modest long-term program that would involve:

• 23

• Assignment of five to six branch members to the code development team on a full-time basis

• Placement of consulting contracts for about five outside experts for at least five years.

• -Contractor support to perform tasks such as code configuration control, code maintenance, and user support.

Similarly, the programs in two-phase flow fundamentals and separate effects testing would require the staff to work closely with university professors to

formulate and conduct experimental programs to obtain information on some I

phenomena or process or on some integral response. Management will ensure interaction on specifications for tests at an early stage between the staff and contractors-responsible for model developments and thnse responsible for experiments. This interaction is to be coordinate how the facility is nodalized and how it is instrumented, as well as to ensure that measurements are sufficient for model development needs. In addition, we plan to:

1. Train staff to run and interpret the new thermal-hydraulic code.

2. Cont.inue courses, seminars, and workshops to maintain expertise.

3. After evolving to one code, rely on NRC staff to develop additional models for the code. Use a contractor to maintain the code and to support code users.

4. Remain current on international experimental programs through cooperative efforts.