ML20077F416

Review of Combustion Engineering,Sys 80+ H-Mitigating Igniter Sys,

ML20077F416
Person / Time
Site:	05200002
Issue date:	05/31/1994
From:	Ginsberg T, Pratt W, Yang J BROOKHAVEN NATIONAL LABORATORY
To:	Office of Nuclear Reactor Regulation
Shared Package
ML20076E998	List: ML20076E995 ML20076E999 ML20077F411 ML20077F415 ML20077F416 ML20077F425
References
CON-FIN-J-2022 NUDOCS 9412140070
Download: ML20077F416 (20)

Text

_ _ _ _ - _ _

-J-2022 2-15-94 1 Rev.1 5-11 94 i

A REVIEW OF TIIE t COMBUSTION ENGINEERING SYSTEM 80+

HYDROGEN-MITIGATING IGNITER SYSTEM l

T. Ginsberg, J. W. Yang, and W.T. Pratt J. Shepherd and D. Stamps l

May 1994 l SAFETY AND RISK EVALUATION DIVISION 9412140070 941209 PDR ADOCK 05200002 A PDR Enclsoure 2

I A REVIEW OF TIIE COMilUSTION ENGINEERING SYSTEh! 80+

fTDROGEN-MITIG ATING IGNITER SYSTEM T. Ginsberg, J. W. Yang and W.T. Pratt Brookhaven National Laboratory Department of Advanced Tectmology Upton, NY 11973 J. Shepherd California Institute of Technology Department of Aeronautics Aeronautics 105-50 Pasadena, CA 91125 D. Stamps Sandia National Laboratory Containment Modeling Department P.O. Box $800 Albuquerque, NM 87185 May 1994 Prepared for Containment Systems & Severe Accident Branch Division of Systems Safety and Analysis Office of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Washington, DC 20555 Contract No. DE-AC02-76CH00016 FIN 3-2022

l 1

AllSTRACT This report presents a summary of an assessment of the CE System 80+ hydrogen mitigation system by  ;

individuals with combustion and accident analysis expertise. The CE 80+ igniter system, consisting of 80 igniters  !

distributed across the containment building, was analyzed based upon a set of igniter placement principles which {

were developed during the review based upon the available experim^ tal data for combustion of mixtures of hydrogen, air, and steam and the experience of the reviewers, i

The individuals participating in the review recognized the lack of prototypic experimental data to support the rE80+ igniter placement scheme, and the unavailability of verified analytical tools for multicompartment ambustion calculations to support the igniter system design. As a result, the assessment of the CE design was cased upon engineering judgment, guided by the experimental evidence and the combustion experience of the reviewers. The assessment was based, to a large extent, on the review of the results of the large-scale NTS experimental program, the 1/4-scale IICOG experimental program, and on the extensive experimental efforts at many laboratories around the world which focussed on the deflagration and diffusion flame combustion characteristics of mixtures of hydrogen, air, and steam.

Despite the deficiencies in multicompartment combustion analytical capability and the lack of prototypic experimental data, the engineering judgment of the reviewers is that it is likely that the hydrogen releases during .

the CE 80+ accident progressions would be ignited and burned in the presence of active igniters in a benign manner. On the basis of the available experimental evidence, the reviewers' judgment is that the igniter concept is a reasonable one for mitigating the buildup of hydrogen in containment. It is believed that the hydrogen concentration through contaimnent would be maintained at levels less than 10 percent, except possibly for the IRWST volume, for small regions around the hydrogen release location, and for small pockets elsewhere in containment. As long as the igniters are powered-on prior to hydrogen release to containment, no significant pressure threat to containment is expected due to the combustion processes.

iii

l' TABLI: OF CONTENTS ABSTRACT .

iii ACKNOWLEDGEMENTS . . . . . vi I. INTRODUCTION ,. .

I

2. IGNITER SYSTEM FUNCTIONAL REQUIREMENT . . . . 2

3. IIYDROGEN RELEASE ACCIDENT SEQUENCES AND ASSUMPTIONS . . 2

4. PREVIOUS IIYDROGEN COMBUSTION RESEARCII AND IMPLICATIONS . . . . 4 4.1 Experimental Research . . .

4 4.2 Analytical Capabilities .

6 4.3 Assessment ofimplications of Previous Research . . . . 6 4.3.1 Basic Arguments . . .

. 6 4.3.2 Application to Loss-of-Coolant Accidents . . 7 4.3.3 Application to Plant Transient Sequences . . 8 5.

BNL CONTAIN CALCULATIONS TO SUPPORT IGNITER PLACEMENT JUDGEMENTS 8 5.1 Calculation Results 8 5.1.1 No Igniters . 8 5.1.2 Igniters On .

. 9 5.2 Implications . .

. 10

6. IGNITER PLACEMENT CRITERIA . .

10 6.1 Overall Approach to Specification of Igniter Configuration. . . 10 6.2 Summary of Generic Igniter System implementation Criteria 11 6.3 Ig. niter Spacing . .

I1

7. CRITIQUE OF THE CE80+ IGNITER PLACEMENT SCliEME . 12 7.1 Introductory Remarks . .

12 7.2 Comments on CE80+ Igniter Configuration . . . . 12 7.2.1 Effectiveness for Transient Sequences: In-Vessel liydrogen Release 12 7.2.2 Effectiveness for Loss-ot-Coolant Accidents: In-Vesselliydrogen Release . 13 7.2.3 Effectiveness for Ex-Vessel Combustible Gas Releases .

13 7.2.4 Comments on Igniter System Functionality. 14

8. CONCLUSION .. ,

. . . 14

9. REFERENCES . .

15 v

o I -

ACKNOWLEDGEMENTS The authors wish to acknowledge the guidance and advice of Mr. Michael Snodderly of the Containment Systems and Severe Accident Branch of the U.S. Nuclear Regulatory Commission's Office of Nuclear Reactor Regulation in the course of the review (" the CE System 80+ hydrogen mitigation system. We also appreciate the efforts of Ms. J. Frejka, who prepared this manuscript for publication.

vi

1. INTRODUCTION A severe accident in the Combustion Engineering (CEJ System 80+ which would result in hydrogen production fro 100 percent clad metal. water reaction, would result in an average hydrogen concentration in containment of 13.6 percent on a dry-air basis [1]. In order to comply with the Code of Federal Regulations regarding permissible hydrogen concentration during a degraded core accident, CE has designed a Ilydrogen Mitigation System the use of glow-plug igniters and has defined the purpose of the system as follows:

" ..While it is highly unlikely that the hydrogen generated will be sufficient to fail the containment, a llydrogen Mitigation System (HMS) has been it.corporated into the System 80+ design to provide added assurance that hydrogen concentrations will be maintained at nondetonable levels even during the most limiting severe accident. To this end, the HMS is designed to accomodate the hydrogen production from 100 percent fuel clad metal-water reaction and maintain the average containment hydrogen concentration below 10 percent in accordance with 10CFR 50.34(f) for a degraded core accident." " ..The HMS consists of 42 igniters (subsequently modified to 80) which are divided into redundant groups, Group A and Group B [2 A number of technical experts in the fields of combustion and accident analysis were requested by NRC to review, and provide an assessment of, the CE igniter system approach to hydrogen control during severe accidents, and to suggest potential improvements, if judged necessary. The individuals participating in this review are the authots of this report and are referred to as the " technical experts" or " reviewers" throughout the body of the report. The re was of limited scope, and was conducted only for the CE System 80+ plant. The conclusions reached here a to this system.

The technical experts reviewed previous experimental and analytical research work related to hydrogen mixing containment, hydrogen combustion in the presence of igniters and igniter system effectiveness. The experts conclude that there exists documented experience with igniter systems which suggests that they may be used to minimize the buildup of hydrogen in containment during severe accidene The available methodolories and experimental data, though incomplete, provides a level of understanding of the b isic phenomenology to enable identification of a set of criteria for placement of igniters in the CE80+ containment l'.silding with the objectiv; of minimizing the buildup of hydrogen. The CE System 80+ design was analyzed based tgon a set of criteria dncloped during the review and a judgment was made about the effectiveness of the system. lecommendations v cre made for improvement of the design.

A meeting was held to discuss the issues related to the CE System 80+ igniter system design'. This report provides a summary of the assessments made at this meeting of the CE System 80+ igniter system design by the experts.

Additional information provided subsequent to the meeting is also included in the report, along with any resulting potential impacts on conclusions. Section 2 defines the functional requirements of the igniter system which was used to make judgements about the proposed design. Section 3 presents the accident sequences leading to hydrogen release which were considered in the analysis and discusses assumptions about the operation of the system. BNL CONTAIN calculations which were carried out to support the assessment of the CE proposed igniter system are summarized in Section 4. These calculations were carried out after the meeting of August 10-11, 1993. Previous research and implications are discussed in Section 5, and Section 6 presents basic criteria wnich were used to assess the propo CE design. Section 7 provides comments on the CE igniter placement strategy and offers recommendations to improve the design. Uncertainties and recommendations for confirmatory research related to optimal igniter placement are made in Section 8.

' Technical Review Meeting to Assess the Combustion Engineering System 80+ Hydrogen Mitigating System, Bethesda, MD, August 10 11, 1994. Attending the meeting were: T. Ginsberg, W. T. Pratt, J. W. Yang, Brookhaven National Laboratory: R. Barrett, M. Franovich, C. Iloxie, J. Kudrick, A. Malliakos, M. Snodderly, and C. Tinkler, U.S. Nuclear Regulatory Commission: D. Stamps, Sandia National Laboratory-and J. Shepherd, California Institute of Technology.

I

I

2. IGNITER SYSTEM FUNCTIONAL REQUIREMENT i l

The basic definition of functional requirement for the icmter system relates to the requirement of 10CFR50.34 which specifies that the system should [31:

I

" Provide a system for hydrogen control that can safely accc tadate hydrogen generated by the equivalent of i a 100 percent fuel-clad metal water reaction " "The hydrogen control system and associated systems shall '

provide, with reasonable assurance, that: Uniformly distributed hydrogen concentrations in the containment do not exceed 10 percent during and following an accident.. " and "Combastible concentrations of hydrogen will not collect in areas where unintended combustion or detonation could cause loss of containment integrity or loss of appropriate mitigating features."

In the assessment of the CE80+ igniter system, the reviewers focussed on development of a response to the following question:

Could the technical experts be satisfied, with reasonable confidence, that with the igniter system in place, no region of the CE 80+ containment would have a hydrogen concentration greater than 10 percent with a volume large enough to cause containment failure if detonated ?

Two basic assumptions were made in this assessment of the CE80+ system: (i) the igniters are powered on ca *y in the accident sequence, before hydrogen release from the primary system to the containment, and (ii) both Groups A and B are functional during the entire accident sequence with a high reliability. The first assumption precludes the possibility that igniters will be powered on at a time that a sensitive hydrogen mixture already exists in containment.

3, HYDROGEN RELEASE ACCIDENT SEQUENCES AND ASSUMPTIONS The CESSA Design Certification Report [2l has identified several severe accident sequences which have larger frequencies and may involve the release of a large quantity of hydrogen into the containment. The accidents identified involve station blackout, large and small break LOCA, total loss of feedwater, and steam generator tube rupture sewarios. The amount of hydrogen production, release rate and release compartment in the containment depend on the specific accident sequences. A brief description of these sequences are given below.

The station blackout (SBO) sequence consists of a loss of all ac power. Station batteries may or may not be available during the accident. With the failure of all safety injections, the scenario leads to core uncovery, fuel heat-up and clad oxidation, which in turn produces hydrogen. The scenario is characterized by high pressures in the primary system, slower core heat-up and relatively lower hydrogen generation in the reactor vessel. With the activation of the Prirrary Safety Valve (PSV), hydrogen as well as steam are released to the IRWST. According to MELCOR analyses, the hydrogen and steam release rates from the primary system are on the order of i and 100 kg/s, respectively. The System 80+ IRWST contains 545,800 gallons of water and a freeboard space of 57,100 ft). The steam released is likely to be condensed in the water pool and the hydrogen will enter the freeboard space. According to the latest CE design [1], there are four vent outlets located inside the wing walls in the steam generator 2

compartment. The total flow area of the vents are 200 ft . The size and the location of the vent outlets are important design parameters which affect the flow circulation between the IRWST freeboard space and the steam generator or/and the lower compartments. The vent size must be sufficient to allow the escape of of combustion products and introduction of oxygen into the IRWST when igniters are activated to control the hydrogen burning.

The pipe-break LOCA results in a rapid depressurization in the primary system. The assumption of failure of all safety injection system leads to core uncovery followed by rapid fuel heat-up and a relatively large hydrogen production in

'he reactor vessel. The release rate and release compartment deperd on the size and location of the break. If the break is located at the cold. leg or hot leg, hydrogen as well as steam will be released into the lower section of the steam generator compartment. Since steam is not removed by immediate condensation, the atmosphere in the source compartment consists of a large fraction of steam. In many cases, the atmosphere is steam inerted and ignitions are prevented during the release period. The System 80+ steam generators are located within four wing walls which do not form a complete enclosure. There are multiple flow paths connecting adjacent compartments, which would 2

enhance the hydrogen distribution in the containment. IIence, concentration of hydrogen in the source compartment is expected to be much lower than that in the SBO scenario.

For the case of a break at the surge line, hydrogen and steam are released to the pressurizer room. The System 80+

pressurizer room is a long enclosure with limited flow passages at the bottom and upper section of the room, liydrogen distribution from the pressurizer to the rest of the containment by natural circulation could be delayed by the limited flow passage.

He sequence involving the total loss of feedwater would lead to steam generator dryout and primary system heat-up, if the failure of all engineered safety feature systems are assumed. The primary system response in this scenario is similar to that in the station blackout. Hydrogen and steam produced in the core are released through the cyclic openings of the PSV into the IRWST.

De sequence of steam generator tube rupture is different than the other sequences discussed above. In this sequence, the hydrogen produced in the core is released through the tube break to the secondary system and eventually released outside the containment. Ilowever, most of the analyses show that clad oxidation in the core is not more than 40 percent for this scenario. This implies that about 60 percent of clad zircaloy are still available for the ex-vessel hydrogen generation in the containment (i.e., by corium-roncrete interaction). Ilydrogen control would also be needed to ensure combustion of hydrogen released from core-concrete interactions for this sequence. >

The in-vessel hydrogen generation is controlled by several factors, such as the metallic zircaloy surface area and the amount of steam available for the oxidation reaction. All the computer analyses for severe accidents show that the quantity of hydrogen produced in the reactor vessel is less than that equivalent to 100 percent oxidation of the active clad as specified in 10 CFR 50.34(f). The System 80+ active clad contains about 26536 kg (58500 lb) of zircaloy.

A 100 percent oxidation of the clad will produce 1164 kg of hydrogen. The average hydrogen release rate from the primary system to the IRWST computed by most analyses is in the order of 0.2 kg/s. At this average rate, the release period is 5820 seconds (1.6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br />) for the case of 100 percent metal-water reaction (MWR). Since hydrogen is released by the cyclic opening of the safety relieve valves, the actual release rate could be higher and the release period longer than these averaged values. The igniters designed to control hydrogen burning must be able to handle the high release rate and be functional for a long period of time. It should be pointed out that there is a large uncertainty in the estimated release rate and release period. While 10CFR50.34 specifies consideration of 100 percent MWR, currently available mechanistic models, based upon experimental data, predict metal-water reactions of significantly lower yields.

Without restoting the safety injection systems, continued cote uncovery will eventually lead to breach of the reactor vessel. At the time of reactor vessel breach, hydrogen and corium are released into the cavity region. If the cavity flooding system is not activated, the process of corium-concrete interactions (CCI) will lead to the generation of a large quantity of combustible gases (CO and 112). The System 80+ cavity is directly connected to the Maintenance Access, Ventilation and Equipment Chase (MAVEC) which has a door and a damper open to the lower compartment.

These two openings will allow the release of hot gases generated by the CCI from the cavity to the lower compartment. Gases in the cavity can also escape through the nozzle region around the reactor vessel to the refuel pool area. Since gases in the cavity (the lowest compartment in the containment) are at elevated temperatures, the buoyancy force will enhance the natural circulation and redistribution of hydrogen in the containment.

Containment sprays play an important role in hydrogen concentration and combustion in the containment. The System 80+ containment has two independent spray system trains ivhich use the IRWST as a water source The flow rate of each pump is $000 gpm. The spray system can be powered by the on-site, off site, and the emergency power sources.

Thus, containment spray can be available for both transient sequences and the pipe. break LOCA events. The spray system is designed to maintain the containment pressure and temperature below the containment design specifications (53 psig and 290 F). The spray is automatically activated by the Safety Injection Actuation Signal (SIAS) and the Containment Spray Actuation Signal (CSAS). For a transient sequence, the in-vessel released steam and energy are absorbed by the IRWST water pool. The relatively low containment pressure and temperature during the hydrogen release period would most likely he unab!c to activate the containment spray. However, sprays would likely not be necessary here, since the bulk of the steam released would be condensed in the IRWST, and the dry atmosphere would tw able to initiate hydrogen ignition without initiation of sprays through operator action. For a pipe-break event, steam 3

l~

and its associated energy are directly released to the containment. The release could increase the containment pressure and temperature to the level at which the sprays are automatically activated. Steam removal by sprays will allow hydrogen ignition if the igniters are available. The impact of hydrogen ignition as a result of sprays depends on the amount of hydrogen already accumulated in the containment, the steam condensation rate, and the flow mixing enhanced by turbulence created by the sprays. It is noted that the System 80+ spray heads are located about 91 ft above the operation floor. Containment sprays will directly affect the steam condensation, hydroger mixing and ignition in the dome area, where the atmosphere is likely to be rich in oxygen and hydrogen. Atmospheres in other compartments, particularly these in the lower section of the containment, are affected by sprays through the inter-compartment flow circulation patterns.

In order to determine whether the CE80+ igniter system satisfies the functional requirements discussed in Section 2 above, it must be established, by a combination of experiments, analysis and engineering judgement, that the hydrogen concentration during the hydrogen release sequences described in this section is less than 10 percent on a containment. wide basis with igniters operating, except for perhaps small regions of containtnent which have sufficiently small volumes so as not to cause a containment threat if detonated. ' The next section discusses the experimental database which is available to support the analysis of the effectiveness of the igniter system and, in addition, the status of appropriate analytical models which could be used to compute the hydrogen concentraJan

' distribution in containment during the hydrogen release events discussed above.

4.

PREVIOUS IIYDROGEN CO5fBUSTION RESEARCil AND DIPLICATIONS The reviewers surveyed the experimental and analytical work which had been performed related to hydrogen mixing, transport and combustion in the presence of operating igniters, in containment buildings. The following summarizes the conclusions of the survey:

4.1 Experimental Research The technical experts are unaware of any systematic study of the effect of igniter placement on hydrocen combustion behavior in systems which are typical of multicompartment containment geometries. Additionally, there are no hydrogen combustion data available from experimental systems which are prototypic of a PWR or, in patticular, of the CE80+ containment system which would be useful for assessment of igniter system performance.

While prototypic data are not available, important research which is relevant to the question of igniter system performance, was performed during the 1980s on the phenomena of ignition and combustion efficiency of mixtures of hydrogen, steam and air, igniter effectiveness during spray operation, igniter effectiveness in condensing steam atmospheres, and igniter effectiveness as a function of temperature for glow-plug igniters (e.g.,4-8]. The NTS experiments [4] provided data on the performance of igniters during the hydrogen release phase of the accident sequences discussed in the previous section. The Mark.ll!IICOG [5] experiments provided data on the combustion behavior of hydrogen released through spargers from a pool of water, similar in concept to the releases in the CE80+

IRWST pool The reviewers believe that the available data suggests that igniters would be effective in mitigating against the accumulation of high concentrations of hydrogen in containment and concluded that the data can be used to support development of criteria for igniter placement to promote combustion of hydrogen at low concentratien.

The National Research Council (NatRC) report 19] on hydrogen control cited the following findings based upon a review of a number of experimental programs at several laboratories:

"1.

The igniters will initiate combustion for hydrogen concentrations of 5 to 8 percent, but all the hydrogen will not be consumed.

2.

Operation of sprays has little effect on the ability of shielded thermal igniters to initiate combustions. At low hydrogen concentrations of 5 to 8 percent, water spray promotes more complete hydrogen combustion because of induced turbulence.

4

L

3. Steam concentrations of up to 40 percent by volume do not effect the ability of the igniter to initiate combustion, nor does the steam dramatically suppress peak pressures generated by a burn.

4. For hydrogen concentrations of 10 to 12 percent in the presence of sufficient oxygen, all hydrogen present in the atmosphere will burn.

5. The igniter can initiate hydrogen burning under transient conditions of continuous injection of hydrogen and steam unless the mixture is outside the flammability limits."

The NatRC report concludes that " Tests indicate that shielded igniters will operate reliably under conditions anticipated in nuclear reactor accidems except for total loss of power." The reviewers felt that these conclusions, published in 1987, have not been invalidated by any work performed since that time and that the conclusions are still generally valid. The assessment of the CE80+ igniter system reported here did not address the situation of total loss of power.

The igniters were assumed powered on early in the accident sequence, prior to generation of significant quantities of hydrogen.

Two experimental programs have provHed data on the effectiveness of igniters in terms of their ability to mitigate against accumulation of hydrogen in the presence of hydrogen sources. The EPRI Nevada Test Site (NTS) [4] and the Mark-l!I IICOG [5] experimental programs, described briefly below:

The NTS experiments were carried out in a 2ft diameter,74,000 ft' volume, spherical vessel. One phase of the program was carried out with the intent of studying the concept of deliberate ignition of hydrogen as a control measure. These experiments, involving combustion during continuous hydrogen injection from a location near the bottom of the vessel, demonstrated that shortly after initiation of hydrogen injection, a diffusion flame generally (although not always) became anchored at the source and remained there until nearly the end of the injection period.

Experiments with an igniter placed well above the hydrogen source led to early ignition with resulting flame flashback to the source, forming an anchored diffusion flame. In experiments with the igniter beneath the source of hydrogen, ignition was delayed until the light hydrogen filled from the top downward to the position of the igniter, allowing more hydrogen to accumulate until lightoff occured. In one of these experiments (C-9/P-8) with the igniter beneath the source, the extent of hydrogen accumulation at the top of the vessel became of concern. The source was turned off, fans were turned to mix the gases in the vessel. A mild deflagration resulted. This points to the wisdom of igniter placement above potential sources of hydrogen. No significant pressure rises (less than 17 psi) were measured in any of these continuous injection experiments. The authors consider that these NTS experiments demonstrated that: (i) igniters, when turned on early, provide an effective means of burning hydrogen to preclude buildup to large concentrations in a large open volume, (ii) the dominant mode of combustion was as a diffusion flame, anchored at the source, with no evidence of a deflagration, (iii) no evidence was presented which demonstrated adverse effects of the use of igniters, e.g., large pressure spikes or accelerated flames and (iv) for effective operation, the igniters should be placed in regions above the source of hydrogen. Dry-air hydrogen concentrations were measured at limited locations, and the continuous injection data presented in the NTS report are limited. Concentration gradients were observed in this vessel, due to the presence of a source of hydrogen in a limited-volume vessel and, in addition, due to the strong influence of bouyancy on the flow and mixing processes. The data suggest the existence of a region of high concentration in the vicinity of the source k) cation, and development of a stratified hydrogen distribution resulting from bouyancy-induced flow of hydrogen towards the top of the vessel and its subsequent redistribution in the downward direction [10).

The Mark-l!I IICOG experimental program was carried out in a 1/4-scale facility, the central feature of which was an annular pool representing the suppression pool. Eleven spargers (for simulation of Perry and Grand Gulf), spaced circumferentially around the pool were used in the experiments along with a total of 56 igniters placed within the experimental apparatus. (The plant installation incorporates 20 igniters around the suppression pool, with a total of

% in the containment). Twelve glow-plugs were spaced around the annular pool region at a level around 5 ft above th.: water level. These carefully-designed and scaled experiments demonstrated that hydrogen released from spargers submerged in a water pool would, in the presence of igniters above the pool, burn as diffusion flames which are anchored at the pool surface. Additionally, the experiments demonstrated that the ignition system maintained the background concentration of hydrogen to below 4-5 percent by volume. The annular geometry of the Mark-Il!

suppression pool is similar to that of the IRWST pool in the CE80+. The authors concluded that the HCOG 5

-experiments suggested that hydrogen released from spargers which are submerged in the IRWST pool would burn as diffusion flames if a suitable number of igniters were located in the vapor space of the IRWST tank. Under these -

conditions, development of detonable mixtures inthe vapor space would be difficult.

4.2 Analytical Capabilities In order to rigorously demonstrate that the requirements of 10CFR50.34 can be satisfied by an igniter system of a given design, an analytical tool is required which can predict the hydrogen concentration distribution in containment with igniters on while hydrogen flows into the containment building from the primary system as a result of a LOCA or a transient, or as hydrogen is generated as a result of core-concrete interactions. Sprays and fans may be on or off. The containment atmosphere may be temperature-stratified at the time of hydrogen injection. The analysis-requires models for core degradation, which would provide the hydrogen-steam source to the contairuuent, a core-concrete interactions model, a hydrogen transport model to predict local species concentrations and flow between subcompartments and, finally, a combustion model which suitably characterizes the combustion processes.

The reviewers considered available analytical tools le.g.,11-13] in the context of the international standard problem exercise to predict the German HDR containment mixing and transport experiments El1.2 and El1.4 [141. These experiments are rather complex and will not be discussed here in detail. Experiment El1.2 involved injection of hydrogen-helium mixture from a release point just below the containment operating floor, while El1.4 injected the mixture from an elevation well-below the containment dome elevation. The available analytical tools did a reascnable job of predicting the (relatively uniform) species concentvtions in Ell.4. The experimental sequence for Experiment Ell.2 involved four distinct stages: a steam injection phase, which established the initial temperature and steam concentration distribution in the containment, a phase of hydrogen-helium injection, a second period of steam injection and, finally, a phase of natural cooldown followed by delivery of water spray to the outside of the containment shell.

For all phases beginning with the hydrogen-helium injection, the experimental results showed strong gas concentration gradients in the containment dome region and between the dome and the region of containment beneath the location of injection. The analytical methods did not capture the strong differences in gas concentration. A number of hypotheses have been proposed to explain the discrepancies between the experimental results and calculational models, including the inability to predict bouyant gas concentrations in the presence of strong pre-existing temperature gradients in containment, the basic failing of lumped-parameter codes to resolve gradients in either temperature, velocity or species concentration within a subcompartment. The authors generally felt that available analytical codes have not been adequately assessed against either . mixing data or combustion data in systems simulating multicompartment containments. The failure to capture the most important features of El1.2 was taken as demonstration of the basic inability of current analytical methods to predict concentration distributions in containments under conditions involving steam injection, heat transfer to boundaries accompanied by condensation, hydrogen injection, mixing and multicomparment transport. More generally, computer codes designed to predict hydrogen combustion ' phenomena in containments have not be assessed against experimental data from prototypic multicompartment combustion experiments.

4.3 Assessment of implications of Previous Research 4.3.1 Basic Arguments The reviewers recognized that systematic studies of igniter systems have not been performed and that data prototypic of the CE80+ system are also not available. Moreover, it recognized that a reliable, verified mathematical model for prediction of hydrogen concentration during hydrogen injection and combustion applicable to the CE80+

containment is not available. Despite these deficiencies, the reviewers concluded that there is a substantial base of experience with igniters and igniter performance to suggest that igniter systems can successfully and safely burn hydrogen at low concentration and thereby mitigate potential threats to containment integrity which would result if hydrogen were left to accumulate. If designed and implemented based upon engineering judgement derived from principles which are suggested by the available experimental evidence, the experts believe that igniters can succeed in benignly burning hydrogen and thereby maintaining the hydrogen concentration at the levels required by 10CFR50.34. It is assumed that the igniters are powered on early in the severe accident sequence, prior to generation of significant quantities of hydrogen. The basis for the above judgement is presented below:

6

- _~

The review by the National Research Council which led to the findings presented in Section 4.1 above are felt to be valid. Agreement with these findings is taken to imply that igniters can effectively initiate the deflagration mode of combustion in preexisting mixtures of hydrogen, air and steam at hydrogen concentrations in the range 5-8 percent,

' and that combustion of mixtures of 10 percent hydrogen or greater would be nearly complete if adequate oxygen is available. These conclusions apply as long as the steam content is low enough so that the mixture is not steam inerted.

The NTS experiments discussed above are interpreted to suggest that, during accidents which are initiated by pipe breaks and transients which occur in regions of containment which have adequate supply of oxygen, igniters in those regions may lead to ignition of diffusion flames which would be anchored to the source location. The conditions under which stable diffusion flames would exist at the source location, however, are not well defined. Detailed models or criteria for diffusion flame stability (flame lift-off, blowoff and relocation) are not available, and were recommended by the NatRC report for further study. If conditions are favorable for continuous diffusion flames, combustion will continue for the hydrogen source duration period, with little bulk accumulation of hydrogen in the containment. If insufficient oxygen is available in the break compartment, then the hydrogen will begin to accumulate there, and will then flow to adjacent compartments in which sufficient oxygen will be available to support combustion. Igniters, if appropriately placed in these downstream compartments, may initiate combustion of the hydrogen entering these compartments as a diffusion flame or, if not, will initiate mixture deflagrations when the mixture composition approaches the 5-8 percent range. It is expected that the igniters will maintain the hydrogen in the compartments downstream of the break compartment below the required 10 percent level The 1/4-scale llCOG experiments discussed in Section 4.1 are interpreted to imply that, for accidents which are initiated by transients, in which hydrogen-s'eam mixtures are released to the IRWST in the CE80+ system.

appropriately spaced igniters in the IRWST vapor space region would initiate combustion of hydrogen above the water pool if sufficient oxygen exists and if the region is not steam-inerted (steam condensation is effective). The hydrogen, if released from spargers similar in design to those of the HCOG cxperiments, would burn as diffusion flames '

i anchored to the pool surface. If conditions in the IRWST are such that combustion cannot be sustained, then hydrogen mixtures would be transported to the steam generator (SG) compartment.. Igniters placed in the steam generator compartment would either lead to ignition of the mixtures as they pass into the SG compartment as low velocity jets, or would burn the mixture within the SG room if not steam inerted or oxygen depleted. If the mixtures do not burn in the SG room, then they would eventually reach the dome region, where an array of igniters would initiate burns in that region if, as is likely for these sequences, conditions are not steam inerted.

The above discussion has made reference to an " adequate

• number of igniters distributed through the CE80+

containment. The question about numbers of igniters will be addressed in Sections 6 and 7 below.

4.3.2 Application to Loss-of Coolant Accidents For severe accidents initiated by loss-of coolant accidents, high-velocity jets of high-temperature hydrogen and steam will enter the containment atmosphere from break locations in the regions of comainment identified in Section 3 of this report. The jets are expected to burn upon ignition for molar ratios of steam-to-hydrogen of less than 9:1 and if adequate oxygen is available [15]. The NTS experiments suggest that hydrogen released from the break will, under a wide range of conditions, continue to burn as a diffusion flame anchored at the break location if adequate oxygen is available and the steam content of the jet permits. Despite the many uncertainties, the authors believe that this scenario is a reasonable expectation, given the presence of an adequate coverage of :gniters. The question of coverage is discus 3cd further in Sections 6 and 7. If conditions for diffusion flame combustion at the site of release are not satisfied, then the hydrogen will eventually be transported to the containment dome where adequate oxygen will be available to support combustion. The gases flowing between compartments may burn off as diffusion flames. If they do not, then it is expected, based upon the previous deflagration research, identified in the National Research Council report [9], that igniters in the downstream compartments will ignite the mixtures in their vicinity and deflagrations

]

would maintain hydrogen concentrations below the required 10 percent level.

If the mixture in any region of containment is steam-inerted, then two cases must be considered: (i) sprays are assumed off and stay off and (ii) sprays are assumed initially off and are turned on at some later time. If the sprays are off and stay off, then the steam will condense verv slowly by natural circulation, leading to slow reduction of the local steam concentrations. When these steam concentrations fall below the inerting limit, approximately 55 percent 7

l

l steam, then the igniters will effectively initiate hydrogen combustion and the hydrogen will likely burn off. If the sprays are turned on after accumulating hydrogen, then the possibility arises that if steam condensation is sufficiently rapid, then igniters may irutiate a burn which could begin to rapidly propagate through a suddenly deinerted, sensitive mixture of hydrogen-air, perhaps leading to significant combustion pressure rise. An assessment of the potential for rapid combustion events, with the generation of pressures larger than for constant volume combustion would have to be considered. This scenario could, conceptually at least, occur in the containment dome region. Small-scale experiments have been performed [16) in which mixtures of hydrogen, air and steam have been suddenly exposed to rapid condensation, with igmters turned on. Over the range of initial mixture compositions and condensation rates considered, the mixtures always burned at their flammability limit, based on accepted flammability maps. Based upon this evidence, it is expected that condensation will not lead to pres. ares above those which wordd be expected based upon constant volume combustion calculations, llowever, supporting 'alculations are needed and NRC/RES is supporting a series of experiments at SNL to provide further confirmatory evidence related to this issue (17].

4.3.3 Application to Plant Transient Sequences During CE80+ plant transient initiated sequences, prt.nry system hydrogen and steam would flow into the IRWST water pool prior to entering the containment atmospre. The hydrogen-steam mixtures would be distributed and sparged through the pool at severa. 'screte locations, and the steam would condense. A major characteristic of these l sequences is that the steam content in containment is expected to be significant below the inerting steam concentration.

Igniters are placed at several locations within the IRWST above the water pool, close to, but not directly abo.: the

' spargers, with the intent of burning the hydrogen as it is released from the pool. The igniters should not be deliberately placed in a location where it is likely to be exposed to a flame for an extended period.

If adequate oxygen is available to support combustion, then it is expected, based upon the llCOG experiments, that the igniters will initiate combustion within the IRWST vapor space when the local concentrations reach 5-8 percent, and the hydrogen will burn at the surface of the pool as diffusion flames anchored to the pool surface. This assumption will be assessed in more detail at a later date, after more detailed consideration of sparger design, igniter location and other geometric and flow considerations. If the conditions within the IRWST cannot support combustion.

then hydrogen will llow through vent areas in the roof of the IRWST into the steam generator room. Igniters placed near these vent areas and within the steam generator room will initiate combustion in their vicinity and, it is believed, will ignite the mixture as it enters this room and the mixture will burn as a diffusion flame. If the steam generator room cannot sustain mixture combustion due to oxygen depletion, then the combustible gases will flow to the containment dome region where additional igniters are placed to burn the gaser within that region. It is expected that the existence of several layers of igniters, at the source location in the IRWST, in the steam generator room and.

finally, in the dome, would provide assurance that the hydrogen would burn benignly upon release during transient accident progressions, and w;"Id preclude accumulation of hydrogen to detonable concentrations.

5.

IlNL CONTAIN CALCULATIONS TO SUPPORT IGNITER PLACEMENT JUDGEMENTS CONTAIN analyses were performed for the CE System 80+ containment using a 15-cell nodalization [181. The development of the nodalization was based on considerations of potential hydrogen sources, containment interior structures and on the CE proposed placement of igniters. Two studies were performed to simulate severe accidents based on th. in-vessel releases of hydrogen and steam computed by the MEI.COR code. Section 5.1 summarizes the calculation results.

5.1 Calculation Results 5.1.1 No Igniters a 'e fW ries of calculations considers the cases of no actuation of igniters. This study was performed in order to

,he maximum concentration of hydrogen in the containment and identify the potential regions where ole mixtures could be accumulated.

8

-l l

The results show that, for a station blackout sequence, hydrogen concentrations greater than 10 percent are predicted l

for the IRWST, cavity, nozzle compartment, lower section of the lower compartment and the refuel pool area. In l these regions, the steam concentrations are less than 24 percent and the oxygen concentrations are sufficient to form a highly combustible or detonable mixture It is noted that the maximum hydrogen concentration is predicted as high as 60 percent in the IRWST These calculations indicate that a hydrogen mitigation system is needed to ensure that hydrogen concentrations are kept below 10 percent during transient sequences in which steam is endensed in the IRWST.

For a small break LOCA sequence with no sprays initially turned on, hydrogen and steam are directly released from the reactor vessel to the steam generator compartment. Due to the addition of large quantities of steam in the containment, hydrogen concentrations greater than about 5 percent were only predicted for the cavity region into which combustible gases are released by the corium-concrete interactions during the ex vessel release phase of the sequence. Under these conditions, the steam concentrations in containment are very high, i.e., in excess of 40 percent, and the oxygen content in the release compartment is likely to be low, due to the purging effect of the steam.

As a result, the hydrogen may not burn in the source compartment, despite the presence of igniters. Hydrogen in the

' calculations performed in this study builds up in containment to global concentrations of 5 percent in the presence of a large inventory of steam. The hydrogen concentiation would subsequently increase upon condensation of the steam, either by sprays, or as a result of condensation heat transfer to the containment boundaries. A hydrogen mitigation system is required to ensure that, upon condensation, the hydrogen burns as soon as minimum combustion conditions are achieved, and thereby avoid reaching detonable levels.

5.1.2 Igmters On The second series of calculations considers the impact of controlled burning on containment performance. Calculations were performed for a station blackout sequence and a small break LOCA with a 30-minute spray.

The CONTAIN results indicate that, for the station blackout sequence, multiple deflagration or diffusion flame burns occur in the IRWST, cavity, the nozzle compartment, the lower section of the lower compartment, the steam generator room and the refuel pool area. All these regions are equipped with igniters, except the nozz!c compartment in which the burn is induced by propagation from the cavity area. The total burn time is about 111 minutes (from 9305 to 15990 s), and a total of 657 kg of hydrogen is consumed (about 68 percent of the hydrogen in the containmen0. The burn period covers the periods of both in-vessel release from 5400 to 13700 seconds and the ex-vessel release due to corium/ concrete interaction starting from 10997 seconds. During the burn period, each burn is terminated either by the depletion of oxygen in the compartment or by the ccmbustion completeness model used in the code. The CONTAIN code uses empirical correlations of combustion completeness for deflagration. The combustion is complete if the hydrogen concentration is equal or greater than 8 percent; otherwise the combustion is incomplete. As a result of the CONTAIN deflagration model and the assumed presence of igniters in each of the compartments, the hydrogen concentration will not be predicted to be greater than 8 percent in any compartment which is not oxygen depleted or steam inerted. For a diffusion flame burn, the fraction of the incoming combustible gas burned is assumed to bc 1.

According to the analyses, the hydrogen concentrations vary between 2-8 percent during the hydrogen injection period.

The hydrogen concentration is predicted to be between 1.6 percent to 2.4 percent in the containment at the end of the burning period.

For the small break LOCA sequence, combustion occurs before sprays are actuated. CONTAIN predicted that one hydrogen combustion event occurred in the steam generator compartment during the break release time period as a result of the igniters in the source compartment. Following this initial burn, which involved less than 100 kg of i hyd ogen, combustion conditions were not again reached in the steam generator compartment. Sprays are assumed activated for a period of 30 minutes during the time period of hydrogen release from the primary system. Combustion occurs once again after vessel fadure and core-concrete interactions have initiated. Burns are predicted for the lower compartment and steam generator region and a large amount of CO is involved in the burning. The total burn time is about 201 minutes ( from 690 to 12780 s) and a total of 302 kg of hydrogen is consumed (about 42 percent of the hydrogen in the containment). Similar to the station blackout scenario, each burn is terminated either due to the depletion of oxygen in the compartment or by the CONTAIN combustion completeness model. At the end of the burn period, hydrogen concentrations ate about 2.5 percent to 3 percent in the containment. The calculation result indicates i that the controlled burn maintains the btogen concentrations at low levels. This result is, of course, a consequence 9 l l

l l' l of the combustion model used in the calculation. Because the code has not been appropriately assessed against multicompartment combustion experiments, the result cannot be used to conclude that there would be no regions of I containment which would hase concentrations above 10 percent. The CONTAIN analysis also indicates that controlled

{

burns have no appreciable effect on containment pressure, but the repeated burns yield high temperatures in some '

regions in the containment.

I 5.2 Implications The calculations performed assuming no igniters on con 0rm the CE conclusion that a hydrogen mitigation system is required since, in their abseace, detonable mixtures may develop in the litWST and other regions in the event of a severe accident initiated by a transient event in which steam is condensed in the pool. The calculations which were performed for the LOCA initiators demonstrate that under some conditions the steam concentrations will be very high, and will preclude immediate combustion at the site of release or, perhaps, even elsewhere in containment, especially if sprays are not on or are not effective. Under these circumstances, combustion will be initated later in time, upon activation of sprays. The nature of combustion under conditions of sprays on and igniters on must, therefore, be addressed for these sequences.

6. IGNITElt PLACEMENT CIIITEltlA After consideration of previous experimental and analytical research described in Section 4 and consideration of the BNL CONTAIN calculations discussed in Section 5, a set of igniter placement principles were developed which could be used to assess the proposed CE80+ igniter system design. These criteria are largely based upon engineering judgement, guided by the available experimental database, insights gained from the CONTAIN calculations, and the combustion phenomenology experience of the individuals involved in the review. These criteris ve discussed below.

6.1 Overall Approach to Specincation of igniter Configuration Igniter placement specification begins with identi0 cation of all credible severe accident sequences which can lead to release of hydrogen to the containment building, and the associated release locations. The general principle which is adopted is to ensure that any compartment of containment which could contain hydrogen, either a compartment containing the release location or a compartment which receives hydrogen from a neighboring compartment, would contain at least one pair of igniters (one each from Group A and Group B). More than a single pair enuld be specified, depending upon the volume of the compartment. The igniters would be positioned so as initiate combustion at the lowest hydrogen concentration and as early as possible, as close to the source as possible and to promote as complete combustion as possible upon development of a combustible mixture.

The igniters should be positioned above potential source locations. Otherwise, bouyancy will carry hydrogen upwards, and hydrogen will generally arrive at igniter locations beneath the source after some time delay and, therefore, after some potentially dangerous accumulation above the source, as it did in NTS Test C-9 [4). While the igniters should be placed above potential source k> cations, they should not be placed too high in a compartment. They should be l

placed in a manner that takes advantage of upward flame propagation, which would be initiated at lower hydrogen concentration than downward propagation.

' Each potential source compartment is analyzed separately, and its neighbors are identified as receivers. Igniters are placed in these receiver compartments, as close to the entry location of flow into that compartment as possible, in order to promote initiation of a diffusion llame burn as the flow passes from the source into the receiver compartment.

l De receiver compartments will also have igniters positioned in their interior, so as to promote combustion within us volume should a combustible mixture develop. Since hydrogen has a strong potential to stratify in any bounded volume, igniters should be also be placed high in any compartment which has no outward flow nath at its highest elevations, so as to promote combustion in these stratified regions. As above, igmters should be placed in a manner that takes advantage of upward flame propagation, and the igniters should not be placed just adjacent tc the ceiling of a compartment.

10 t

l 6.2 Summary of Generic Igniter System Implementation Criteria The above discussion is summarized in the following basic criteria:

(1) Examine all credible release sequences and identify hydrogen source compartments and release locations.

(2) Locate igniters in all potential source compartments to promote combustion as a diffusion flame f anchored at the source location. Place these igniters above the potential source locations, including above water pool through which hydrogen is sparged from submerged source locations (spargers).

l (3) Follow flows from each potential source compartment to adjacent receiver compartments, and locate igniters to promote combustion as a diffusion flame upon entry of the mixture to the receiver compartments.

(4) locate igniters within all compartments which could contain hydrogen to promote early deflagration mode of combustion within the compartment volume as mixture composition becomes combustible.

Avoid placement too high in compartment. Take advantage of upward flame propagation.

(5) Compartments which are " dead-ended" at the top should contain igniters above potential source locations or flow entry points. The containment dome should have igniters sufficiently high so as to promote combustion of stratified mixtures.

(6) Any compartments which are judged to be questionable as far as buildup of combustible mixtures, should either contain igniters to promote early combustion at low concentration, or it should be required to demonstrate that a detonation would not threaten containment integrity or survival of critical equipment.

6.3 Igniter Spacing i

The above discussion has thus far avoided the issue of igniter spacing or, to put it another way, the question of how '

many igniters to be positioned in any given compartment. Although this issue is probably best addressed by prototypic experiments, previous experience was used to provide some guidance.

The Segouyah ice condenser plant contains a total of 68 igniters distributed in its containment volume. The Grand Gulf Mark-lli plant contains 90 igniters, distributed through its drywell volume.

The NTS experiment provided limited demonstration that in a closed spherical volume of diameter 52 ft, that ignition and combustion processes were relatively benign in the presence of hydrogen release and active igniters. If we take 52 ft as a typical spacing, then the associated volume is about 75,000 ft). We will take this as rough guidance for the minimum number of igniters to be placed in any large compartment (where volume is greater than 75,000 ft').

If we take the entire CE80+ containment, with a volume 3.8 x 10* ft' as a whole, then we find that this rule +f-thumb would give about 50 igniter locations distributed through the building. CE has specified a total of 80 igniters in two systems (Groups A and B) operated by independent power supplies. Based upon the discussion in Reference [19],

it has been assumed in this analysis that both Groups A and B are functional, and that the igniters withio each of the two systems are operational upon demand with high reliability. While a rigorously defensible igniter spacing criterion is not available at this time, the largest scale combustion experiment involving a continuous hydrogen source in the presence of igniters u hich has been performed to date using a large open volume demonstrated effective ignition with one or more igniters in a volume of 75,000 ft'. His volume per igniter is taken as a rough measure of igniter spacing l

adequacy.

igniter basis. Based upon this measure of acceptance, the number of igniters is adequate on an average volume per 1

l l

11

l =

7. CRITIQUE OF TIIE CE80+ IGNITER Pl.ACEMENT SCIIEME 7.1 . Introductory Remarks Three versions of the CE80+ igniter configuration scheme were examined. The first was presented in the CESSAR submittal [2), the second was reviewed at the Technical Review

• feeting of August 10-11, 1993, and the last is the October 1993 CE submittal of Appendix 14.llK (1J.

At the meeting of August 1993, the CE configuration included a total of 66 glow-plug igniters distributed in containment and divided into two " redundant." groups, supplied by independent power supplies. The CE submittal as of the August 1993 was reviewed at the meeting against the criteria of Section 6 above. As a result of this examination, the NRC recommended that CE add additional igniters in five specific locations: (i) the pressurizer compartment, (ii) the MAVEC area downstream of the reactor cavity, (iii) the IRWST area, (iv) the Letdown Hen Exchanger Room, and (v) the Regenerative !! cat Exchanger Room. An additional 14 igniters were recommended with  ;

a minimum of 4 to be added to the IRWST area. The CE specification at the time of this meeting was that the IRWST -

area was vented to the lower compartment through a vent area of 1000 ft2 , with the vents located inside the " wing wall" This vent area was felt to be important from the standpoint of ensuring good circulation of containment oxygn through the IRWST area in order to promote efficient combustion above the IRWST pool.

The authors examined the CE submittal of October 1993, and found that, with 80 glow-plug igniters, CE has gennally met the criteria outlined in Section 6 above with their modified design. The IRWST vent area, however, has been significantly reduced and the locations of the vents have been changed since the meeting of August 1993 The IRWST performance is discussed further in Section 7.2.1, 7.2 Comments on CE 80+ Igniter Configuration The CE igniter configuration is presented in Table 6.1-1 of their Appendix 19.llK submittal. The following is an assessment of the effectiveness of the design, based upon engineering judgement. It is assumed that the igniters have been turned on prior to release of hydrogen to the containment. In vessel releases are discussed in Sections 7.2.1 and 7.2.2, while ex-vessel releases are discussed in Section 7.2.3. A total of 80 glow-plug igniters are distributed through comainment.

7.2.1 Effectiveness for Transient Sequences: In-Vessel flydrogen Release Four igniters have been specified by CE to be placed above the IRWST pool to promote combustion of hydrogen released from the two submerged spargers locations during transient events. In the event of loss of either of the two igniter system trains, one igniter would remain active in the region adjacent to each sparger release point. In order to promote efficient combustion in this region, the flow paths entering and exiting the IRWST volume should be of sufficiently large flow area so as to ensure adequate supply of oxygen from the regions above this volume. Initially, the CE design specified four vent areas to the steam generator compartment with a total area of about 1000 ft 2. These vents were initially located outside the steam generator " wing walls." The open vent area has now been specified as 2

200 ft [1j. The hydrogen combustion hehavior in the IRWST depends on the rate of hydrogen release into the region, the rate of supply of air from containment to the tank volume, the extent of mixing, and the effectiveness of the ignition source. llydrogen could burn as a fuel. air mixture within the IRWST volume, it could be as a diffusion flame anchored to the IRWST pool surface, or it could burn as a diffusion flame exiting the IRWST vents. The assumption of hydrogen combustion as a diffusion flame anchored at the pool surface, in analogy with the llCOG experiments, may not be valid if the supply of air to the region is limited as a result of the small vent area ilydrogen accumulation within the IRWST would be more likely if the hydrogen does not burn off at the pool surface.

One implication of the vent area design change by CE since the August 1993 meeting is that there is more of a chance nc,w that hydrogen will emerge f rom the IRWST volume unburned. One level of defense against hydrogen buildup may not be effective and, as a result, more reliance has to be placed on the igniters positioned in compartments downstream of the IRWST voLme to burn the hydrogen released during transient sequences.

12

l in the October 1993 version of the CE igniter locations, the IRWST vents were placed on the outer side of the steam generator wing walls, and igniters were positioned on these walls adjacent to the vent areas in order to promote ignition of any hydrogen which would emerge from the IRWST volume unburned. If adequate oxygen is available.

then the jets of hydrogen emerging from the IRWST volume will likely burn upon entry to the steam generator compartment. Igniters are available in the steam generator compartment at the 126-ft level to promote combustion in the steam generator room as a deflagration if necessary. If hydrogen should emerge from the steam generator compartment into the dome region, igniters are available at the 164-ft,200-ft and, finally, at the 237-ft, levels to promote combustion at those locations. Two levels of ioniters are located in the annulus between the crane wall and the containment shell to promote combustion of any hydrogen v hich is transported from the steam generator compartment radially outward to this annular region.

7.2.2 Effectiveness for Loss-of-Coolant Accidents: In Vessel flydrocen Release 4

CE has positioned igniters at ne 126-ft level to promote cornbustion as diffusion flames of hydrogen released from pipe breaks in the steam generar compartment located beneath these igniters. The igniters placed at the 164 ft, 200-ft and 237-ft levels are designed to promote combustion of hydrogen which is not burned as diffusion flames within the lower and steam generator compartn":nts. Two levels of igniters are located in the annulus between the crane wall and the containment shell to promote combustion of any hydrogen which is transported from the steam generator compartment radially outward to this annular region.

A characteristic feature of LOCA sequcnces is the large quantity of steam released into containment along with the hydrogen. Two conditions must be considered: (i) sprays are activated late, at some time during release of hydrogen to the containment, and (ii) sprays are activated prior to release of hydrogen.

d.} Sprays activated late. after some hydrocen accumulation if sprays are not available prior to hydrogen release, then the steam concentration in containment will be relatively large at the time of hydrogen release. It is expected that in some cases, oxygen would be depleted from the steam generator compartment as a result of displacement by steam re! cased at an earher time in the sequence. Additionally, the steam content of the steam generator compartment may be so high that a diffusion llame at the break location cannot be supported. Information is lacking here on the effect of atmospheric steam content on conditions for stable diffusion llame combustion. Similarly, the steam concentration in the containment dome and elsen here could be so large and hydrogen concentration so low, that the igniters would be ineffective in igniting the mixtures in the various regions of containment. In the worst instance, hydrogen will accumulate in containment until the sprays are turned on. Upon activation of the containment sprays, it is expected that the steam condensation will lead to ignition and combustion of the hydrogen mixtures as the mixtures reach their ignition limits at lower steam concentrations, as discussed in Section 4.3.2. No containment-threatening pressure events are expected.

Gi) Sorays on ririor to hydrocen relenq With sprays on prior to hydrogen release, steam concentrations would tend to be low, and the igniters would be more effective in burning hydrogen as it is released from the primary system, either as a diffusion flame at the site of the break, or as deflagrations in either the steam generator compartment or the containment dome, thereby limiting the accumulation of hydrogen in containment. It is believed that there are an adequate number of igniters placed in the steam generator room and in the lower and upper dome regions to effectively ignite and bum hydrogen released from breaks either in the piping in the steam generator compartment or in the surge line.

7.2.3 Effectiveness for Ex-Vessel Combustible Gas Releases igniters are placed in the reactor :avity and in the MAVEC area to promote ignition and combustion of hydrogen and carbon monoxide gases released during core. concrete interactions (CCI). These igniters would attempt to burn the gases before they discharge to the lower compartment. If the gases enter the lower compartment unburned, there are additonal igniters on the wing-walls available in that region to ignite them, if these igniters fail to ignite, then the igniters in the containment dome region are available. Additionally, igniters are positioned around the refueling cavity 13

,l to ignite gases released from CCI which flow inside the between the reactor vessel and the biological shield towards the containment dome.

7.2.4 Comments on Igniter System Functionality The authors consider that the CE80+ igniter system provides adequate coverage of the containmen' building with bot igniter Groups A and B functional with high reliability. This conclusion would not apply if either of the two Groups were to be unavailable, or if a "substantional" number u igniters failed to perform upon demand. The reviewers believe that igniter reliability is an important issue which should be addressed in the reliability assurance program.

L

8. CONCI,USION The largest scale experiments which simulate continuous hydrogen-steam release in the presence of active igniters l the NTS and 1/4-Scale llCOG experiments, suggest that igniters effectively maintained low hydrogen concentrations in the experimental test vessels. While detailed hydrogen concentration mappings are not available from these experiments, the background (away from the sources) hydrogen concentrations which are reported are less 5-6 perce in the case of the llCOG cxperiments. In the case of the NTS experiments, detailed data are not presented for local hydrogen concentration, but for those data which are shown, it appears that hydrogen concentrations are less thart 7-8 percent during the combustion period (prior to oxygen depiction.) Ignition, in these experiments, was typally observed to result in a " light-off" burn of hydrogen injected into the experimental vessels up to the time of ignition. I with burn of subsequently released hydrogen as a diffusion name anchored near the site of release. Neither of these experiments provided evidence of any adverse effects of the presence of igniters during the time scale of release of hydrogen into the experimental vessels. Additionally, previous investigations demonstrate the effectiveness of igniters in initiating combustion in the range of 5-8 percent hydrogen, and achieving nearly complete combustion of uniform i

mixtures of hydrogen, air and steam for hydrogen concentrations of 10 percent or greater and steam concentrations t

less than about 30 percent [20]. On the basis of the available experimental evidence, the judgement of the authors is that the igniter concept is a reasonable one for mitigating the buildup of hydrogen in containment.

The reviewers recognized the lack of prototypic experimental data to support the CE80+ igniter placement scheme, and the unavailability of verified analytical tools for multicompartment combustion calculations to support the design.

As a result, the assessment of the CE design for implementation of the igniter system concept was based upon engineering judgement, guided by the available experimental evidence and the combustion experience of the members.

The igniter system implementation criteria were developed to provide a basis for assessment of the CE igniter system.

The latest CE80 Appendix 19.11K igniter system design submittal of October 1993 differs in two respects from the version the experts reviewed on August 10-11,1993: (i) fourteen more igniters were added based upon the experts' recommendations and (ii) the IRWST region vent area and location were altered. The additional igniters impact favorably on the assessment, as would be expected. Ilowever, the assessment is that acceptable arguments have not been presented by CE to confirm that the vent area associated with the IRWST is adequate for supply of air for efficient combustion in the IRWST volume. liigh concentrations of hydrogen in the IRWST region cannot be precluded if, however, the vent area is small, adequate air would likely not be available to support combustion.

Additionally, the expectation of the efficacy of the igniters in the IRWST region is based upon an analogy with the 1/4-Scale llCOG cxperiments. There are, however, differences between the two systems in terms of the number and type of spargers, number of igniters, scale and flow paths to the region above the ' vater pool, and uncertainties remain with respect to the potential for combustion of hydrogen as diffusion flames above the IRWST pool.

Despite the deficiencies in multicompartrnent combustion analytical capability, and the lack of prototypic experimental data, the engineering judgement of the members is that it is likely that the hydrogen released during the CE 80+

accident progressions discussed in Section 3 would be ignited and bumed in a benign manner. The reviewers believe that the system will succeed in satisfying its functional requirement as defined in Section 2 of this report.The reviewers believe that the hydrogen concentration through containment would be maintained at levels less than 10 percent, except possibly for the IRWST volume, discussed above, for small regions around the hydrogen release ,

location, and for small pockets elsewhere in containment. As long as the igniters are powered on prior to hydrogen  ;

release, no significant pressure threat to containment is expected due to the combustion processes in the presence of '

la l

l

'f i i

l the igniters. Containment temperatures, however, would remain elevated for some period of time. These conclusions apply to the CE System 80+ system with the 80 igmters in Groups A and B functional and available upon demand with a high reliability.

9. REFERENCES 1.

CESSAR Design Certification, Appendix 19.llK, " Assessment of the System 80+ liydrogen Mitigation System for Application in a Severe Accident Environment," ABB-Combustion, Combustion Engineering, Inc. (October 1993),

2.

CESSAR Design Certification, Amendment N, ABB Combustion Combustion Engineering, Inc., Section

19. I1.3.4 (April 1993).

3.

Code of Federal Regulations,10CFR50.34, U.S. Government Printing Office, Washington, D.C. (January 1991).

4.

Thompson, R.T., "Large-Scale liydrogen Combustion Experiments " EPRI Report, NP-3878, Vol. I and Vol. 2 (1988).

5.

Tamanini, F., et al., "llydrogen Combustion Experiments in a 1/4-Scale Model of a Nuclear Power Plant' Containment", in Twenty-Second Syraposium, The Combustion Institute, 1715-1722 (1989).

6.

Tamm,11., et al., " Effectiveness of Thermal Ignition Devices in lean flydrogen-Air-Steam Mixtures," EPRI Report NP-2956 (March 1985).

7.

Marshall, B.W., "flydrogen: Air: Steam Flammability Limits and Combustion Characteristics in the FITS Vessel," NUREG/CR 3468, SAND 0383 (December 1986).

8.

Lowry, W.E., et al., " Final Results of the 1-lydrogen Igniter Experimental Program," NUREG/CR-2486.

UCRL-53036 (1982).

9.

National Research Council (NatRC) Report " Technical Aspects of flydrogen Control and Combustion in Severe Light-Water Reactor Accidents," National Academy Press, Washington, DC (1987).

10 Shepherd, J.E., " Analysis of Diffusion Flame Tests," NUREG/CR-4534, SAND 86-0419 (August 1987).

. I1. Murata, K. K., et al., " User's Manual for CONTAIN 1.1: A Computer Code for Severe Nuclear Reactor Accident Containment Analysis," NUREG/CR 5026, Sandia National Laboratories (November 1989).

12.

Summers, R. M., et al., "MELCOR 1.8.0: A Computer Code for Nuclear Reactor Severe Accident Source Term and Risk Assessment Analysis," NUREG/CR-5531 Sandia National Laboratories (January 1991).

13. MAAP 3.0B User's Manual, FAI(March 1990).

14.

International Standard Problem 29, " Distribution of flydrogen Within the llDR Containment Under Severe Accident Conditions," Committee on the Safety of Nuclear installations. OECD Nuclear Energy Agency (February 1993).

15.

Shepherd, J.E., "Ilydrogen-Steam and Jet Flame Facility and Experiments," NUREGICR-3638, SAND 84-0060 (October 1984).

16.

Tamm,11., et al., " Effectiveness of Thermal Ignition Devices in Rich Ilydrogen-Air-Steam Mixtures." EPRI Report NP-5254 (1987).

15

s b

l l

1

17. Stamps, D., Personal Communication (February 1994). I 18.

Yang, J., "CONTAIN Analysis ofliydrogen Distribution and combustion for CE System 80+ Containment."

Draft Report for NRC/NRR (February 1994L l l

19. Snodderly, M., Personal Communication with W.T. Pratt (January 1994).

20.

Wong, C. E., "IIECTR Analyses of the Nevada Test Site (NTS) Premixed Combustion Experiments,"

NUREG/CR-4916, SAND 87-0956 (November 1988).

l l

l l

1 I

l I

l l

l l

l l

l I

i 16 l

_ _ _ - _ _ _ _ _ _