ML20205M369
| ML20205M369 | |
| Person / Time | |
|---|---|
| Site: | River Bend  |
| Issue date: | 03/26/1987 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML20205M337 | List: |
| References | |
| NUDOCS 8704020366 | |
| Download: ML20205M369 (17) | |
Text
{{#Wiki_filter:r j f . f [s.n tt% 'c,
' UNITED STATES f [*- i NUCLEAR REGULATORY COMMISSION l . ; E WASHINGTON, D. C. 20$55 o S
's, ...../
ENCLOSURE SUPPLEMENTAL SAFETY EVALUATION REPORT REGARDING THE MARK III CONTAINMENT RELATFD ISSUES GULF STATES UTILITIES RIVER BEND STATION. UNIT 1 DOCKET NO. 50-458 6.2 Containment Systems 6.2.1.9 Mark 111-Related issues The details of the staff's review of each of the 66 individual Humphrey ! Concerns (covering 22 major areas) are contained in River Bend SSER No. 2 In that evaluation, the staff concluded that all but two major areas and a small portion of a third area have been satisfactorily resolved for River Bend. By letter dated July 31, 1986, the licensee provided additional information to resolve these residual issues. The detailed staff evaluations are contained in the supplement to Appendix K. All of the remaining issues have been satisfactorily resolved except Humphrey Concern 3.1. In this concern, the licensee has not satisfactorily demonstrated the capability of the RHR system's discharge line to accommodate chugging type lateral loads when the RHR system is operating in the steam condensing mode (SCM). The licensee was informed of our conclusions regarding this issue. In response, GSU has indicated that they do not intend to utilize the RHR system in the SCM for ; future plant operations. Therefore, further staff review on this issue is not l necessary and will not be pursued. l As a result, the staff will continue to require the River Bend License l Condition as indicated in River Bend SSER No. 2; i.e., "GSU shall not use the : residualheatremoval(RHR)systeminthesteamcondersingmode". Accordinoly. Humphrey Concern 3.1 is considered resolved. However, the staff ! has prov'ded an evaluation of this issue in the supplement to Appendix K in order to provide guidance for GSU if further evaluation is desired. Overall, the staff concludes that, subject to continued adherence to the associated license condition, this matter of the Mark !!!-related issues is resolved for River Rend Station Unit 1. 1 0704020066 070326 PDR ADOCK 05000450 P pon
. _ - ~ . _ - . - - . . - - - .- . _ - - - -. .
(SUPPLEMENT TO APPENDIX K) Humphrey Concern 1.3 Additional submerged structure loads may be applied to submerged structures near local encroachments. Evaluation The loads addressed under this item fall into two categories: (1)loadson submergedboundaries(i.e.,drywellwall,basemat,containmentwall),and l (2) loads on submerged structures such as pipes and beams, f A two-dimensional (SOLAV01) simulation was employed to determine the effect of RAS encroachments on these loads (Reference 1.3.1). An increase in the con-tainment wall and drywell wall loads were predicted which are within the design l load specification. The drywell wall pressure is unaffected by this simulation since it is assumed equal to peak drywell pressure. The submerged structure loads are developed using the velocity field derived from the 50LAV01 simulation. The increased loads produced by the encroachments are stated to be bounded by the design basis LOCA bubble loads. l Conclusion l The staff considers this concern to be resolved based on the stated margin ; l betweendesignandtheloadsestimatedwithaconservative(twodimensional) i simulation of the encroachment effect. ! i Reference 1.3.1 Enclosure to GSU Letter No RBG-24,131, dated July 31, 1986 from l l J.E. Booker (GSU) to H.R. Denton (NRC). i l
Humphrev Concern 2.1 The annular regions between the safety reifef valve ifnes and the drywell wall penetration sleeves may produce condensation oscillation (CO) frequencies rear the drywell and containment wall structural resonance frequencies. Evaluation l i As stated above, the concern is that additional and unaccounted for suppression pool boundary leads may be produced by steam condensation at the exit of the l sleeve annulus. However, the scope of this concern is expanded considerably by Humphrey in Reference 2.1.1. It is speculated that, due to resonant coupline between the sleeve annulus C0 and the sleeve annulus acoustics, the pressure loads at higher frequencies could be amplified. The licensee first addressed this concern using the generic approach described
- in Reference 2.1.2, this approach is applicable to the River Bend design. This methodology derives sleeve annulus CO loads by scaling down the main vent CO load data base. The potential for resonant amplification of these loads was not l addressed.
l The use of the main vent C0 data base for devrlopment of sleeve annulus loads I cannot be rigorously defended because of substantial differences in geometry. I However, based on the relative size of the steam / water interface that would exist at the sleeve annulus, it could be judged that any additional non-resonant loads that nay occur will be second order relative to main vent loads. This judgement applies to both C0 and chugging loads. The added sleeve CO loads that were first proposed were substantial. For example, the peak-to-peak pressure amplitude (PPA) was about 20% of that used in the main vent load definition. They were also applied uniformly in the circum-ferential directien, which represents a sizable conservatism. This is because there are roughly twice as many main vents as drywell penetration sleeves. These modifications are clearly more than second order. Thus, provided it could have been demonstrated that resonant amplification does not occur, the CO loads which were specified would have been considered adequate. In an attempt to demonstrate the absence of such a coupling, results from General Electric's 4TCO tests (Reference 2.1.3) were cited under the Clinton Power Station (CPS) docket (Reference 2.1.4). A review of this material indicated that the contrary was the case. That is, the data implied that the type of resonant coupling suagested by Mr. Humphrey was not only possible, but apparently had actually occurred. In fact, it can be inferred from this data that resonance causes about a two fold increase in the basic C0 loads. As a result of this finding, a completely new methodology for the C0 boundary loads was sroposed, also under the cps docket (Reference 0.1.4). This methodo-ology has been developed by General Electric utfitzing the Mark i Full Scale TestFacility(FSTF) database (2.1.5). This new design loading results in a substantial increase in the pressure loading at the hicher end of the frequency spectrum (20 50 Hz). For exampic, in terms of an ampitfied response spectrum, l 1
= . - ---
j 4-the load intensity is about double the one first proposed. However, these new loads are shown to be bounded by other design basis loads. For example, on the drywell wall the sleeve C0 load, when added to the main vent CO, is bounded by the chugging load specification. On the containment wall, the bound is previded with considerable margin by the pool swell bourdary load. The applicability of the FSTF data to the sleeve annulus loads involves considerable uncertainty because of the great disparity in geometries between the two designs. This applies not only for the steam-water interface at the I t respective pipe exits but, more importantly, for the acoustic path throuoh , which the mechanism that drives the C0 phenomenon is transmitted. In fact, it is not completely apparent that for the FSTF case the system has actually achieved a condition of resonant coupling. This is because of the complexity of the FSTF vent system; i.e., the eight downcomers are connected to a vent header, which in turn, is connected to a main vent which then connects to a simulated drywell. Because of this complexity, it is difficult to ascertain the effective relevant vent system natural frequency with sufficient precision. Despite these uncertainties, there are several factors that may be cited that compensate for any possible inadequacy. First, there is the qualitative evi-dence from the 4700 tests (Reference 2.1.3) that, even when resonant coupling clearly occurs, the load amplification is limited to less than a two-fold factor. Also, the generally conservative application of tne available results provides increased confidence in the adequacy of the load determination method. For example, the pressure results cbserved on 24 inch diameter downcomers are taken directly and applied to the much sma11er.14 inch diameter; safety relief valve discharge line (SRVDL) sleeves. Also, in developing the loads on the suppression pool boundaries, the effective source (steam bubble) radius was assumed equal to that of the sleeve without taking into account the actual presence of the SRVDL itself. As shown in a recent submittal under the CPS docket (Reference 2.1.6), this results in a margin of over 30% in the loads that were developed. The staff believes the margin would be even greater if the steam bubble were modeled more realitisteally (Reference 2.1.3). Additional conservatism stems from the use of conventional acoustic analysis techniques for determination of pressure attenuation from the source to the pool boundaries. And neglected dissipative mechanisms that are present in the suppression pool would further reduce the loads. Insofar as the chugging loads are concerned, the staff has not received a description of these loads directly from GSU. However, this information exists (Reference 2.1.7) and has been assessed by the Mark !!! Containment Issues Review Panel (Reference 2.1.8). The findings of this papel were that the prososed loads were only about 6% of the main vent chugging loads and are easily bounded by the design leads. The staff is satisfied that this is the case. Note that resonance effects are not expected to play any role or influerce the chugging phenomenon associated with the sleeve annulus. Conclusions The staff considers this sefety concern to have been satisfactorily resolved : because the conservatively estimated new loads have been demonstrated to be bounded by other design loads. :
i . i }' 1
- l. References l 2.1.1 Humphrey Engineerin 17. 1982; from i J.M. Humphrey (HEI)g, Inc.; Letter(NRC).
to A. Schwencer dated Jure l 2.1.2 IPC Letter No. U-0714; dated May 25, 1984; from D.I. Herborn (IPC) l l toA.Schwencer(NRC). ! 2.1.3 Bird, P.F., et al.; "4T Condensation Oscillation Test Progran f l Final Test Report", General Electric Report NEDE-24811-P; May 1980. { 2.1.4 IPC Letter No. U-600319; dated November 25. 1985; from F.A. Spangenberg 1 (IPC) to W.R. Butler (NRC). 2.1.5 Fitzsimmons, G.W. et al.; " Mark I Containment Program - Full-Scale l Test Program Final Report", General Electric Report NEDE-24539-P;
- April, 1979.
l 2.1.6 IPC Letter No. U-60058E; dated May 30, 1986; from F.A. Spangenberg (IPC) tow.R. Butler (NRC). J 2.1.7 Enercon Letter No. RWE-0G-060; dated May 25, 1983; from R.W. Evans j (Erercon) to B.R. Patel (Creare R&D). 2.1.8 Mark !!! Contanment Issues Review Panel; " Assessment of Humphrey l Concerns"; CREARE R&D, Inc.; Technical Memorandum TM-928; July , i 1984. '{ 2.1.9 GSU Letter Mc. RBG-24,131; dated July 31, 1986; from J.E. Booker I (GSU) to H.R. Denton (NRC). l' I l l 3 i I l , i 1 i i j i
Humphrey Concerns 2.2 and 2.3 2.2 The potential condensation oscillation and chugging loads produced through the annular area between the SRVDL and sleeve may apply unaccounted for loads to the SRVDL. Since the SRVDL is unsupported from the quencher to the inside of the drywell wall, this may result in failure of the line. 2.3 The potential condensation oscillation and chugging loads produced through the annular area between the SRVDL and sleeve may apply unaccounted for loads to the penetration sleeve. The loads may also be at or near the natural frequency of the sleeve. Evaluation The concern is that the steam condensation process (C0 and chugging) at the sleeve annulus exit will give rise to loads on the SRVDL and SRVDL sleeve 1 analogous to the lateral loads experienced by Mark I and Mark II downcomers
- during postulated LOCA blowdowns, and that these structures have not been designed to accommodate them.
f The licensee's specification for chugging loads is given in Reference 2.2.1. I The load has a half-sinusoidal time dependence with a duration of 3 mseconds (.003 seconds) and a peak amplitude of 22 Kips. It is stated that this load derives from the Mark II load methodology of Reference 2.2.2 as modified by the NRC Staff's acceptance criteria (Reference 2.2.3). The load was developed by scaling down the peak amplitude to the outside diameter of the SRVDL sleeve and accounting for the fact that there are fewer chug sources created by flow through the SRVDL sleeve annulus than exist during DBA blowdowns through the Mark II pressure suppression system (i.e.,16 SRVs for the RBS vs. about 100 downcomers in a typical Mark II plant). Scaling down for pipe diameter is accomplished by assuming a 1.7 power dependence of the peak load amplitude on diameter. Load reductions for fewer chug sources utilizes the staff approved statistical representation for these loads (Reference 2.2.3). The region of application of the load was also scaled down using a first power dependence on
- diameter. It is stated that the stresses / loads resulting from application of these loads are either bounded by other design loads, or are less than the pertinent allowables.
i The appifcability of the Mark II results for the present application is somewhat uncertain due to the substantial geometric differences (straight down vs. inclined pipe and circular vs. annular cross section). The use of a 1.7 power dependence of peak amplitude on pipe diameter is somewhat less conservative than we would have preferred since the available data (Reference 2.2.4) exhibit exponent values that range from 0.7 to 1.7. On the other hand, no credit was taken for the presence of the SRVDL in the steam bubble. This provides a substantial
- conservatism that the staff judges more than compensates for any possible non-conservatism in selecting this exponent. Therefore, we find the approach reasonable.
. No C0 lateral loads are specified by the licensee. In Reference 2.2.1, it is argued that since the suppression pool end of the SRVDL sleeves are truncated perpendicularly to the discharge line axis, the potential for asymmetric dynamic pressure loads arising at that end is precluded. The staff believes that this argument has sufficient merit to justify the absence of any expif cit C0 lateral load specification on the SRVDL and SRVDL sleeves during postulated LOCA conditions. Conclusion The staff considers this concern to be resolved because of the specification of a conservative chunging load for the SRVDL and SRVDL sleeve, and because the RBS configuration for these structures does not permit development of any signifi-cant unbalanced pressure loading during the C0 phase of the LOCA blowdown. References 2.2.1 Enclosure to GSU Letter No. RBG-24,131; dated July 31, 1986; from J.E. Booker (GSV) to H.R. Denton (NRC). 2.2.2 Davis, W.M.; " Mark II Main Vent lateral Loads"; GE Report NEDE-23806-P, October, 1978. 2.2.3 Anderson, C.; " Mark II Containment Program Load Evaluation and Acceptance Criteria"; NRC NUREG-0808, 2.2.4 General Electric Letter MFN-050-80; dated February 29, 1980; from R.H. Buckholz (GE) to C.J. Anderson (NRC).
Humphrey Concerns 3.1, 3.3. 3.7 3.1 The design of the STRIDE (Standard Reacter Island Design) did not consider vent clearing, condensation oscillation, and chugging loads which might be produced by the actuation of the RHR heat exchanger relief valves. 3.3 Dischargt from the RHR relief valves may produce bubble discharge or other submerged structure loads on equipment in the suppression pool. 3.7 The concerts related to the RHR heat exchanger relief valve discharge
' lines should also be addressed for all other relief lines that exhaust into the pool.
Evaluation
! The concern is that, besides the main safety / relief valves (MSRVs), there are a number of other valves that discharge fluids into the suppression pool. As a result, they could produce loads analogous to those associated with MSRV discharges and/or LOCA blowdowns through downcomers. These loads have not been accounted for in plant design.
i For the RHR system, flow through the heat exchanger relief valves can occur when it is operating in the steam condensing mode (SCM). During such operation, the heat exchanger is pressurized to about 200 psi by a pressure control valve (pCV). Should the PCV fail, resulting in elevated pressures, the heat exchanger relief valves would actuate at their setpoint (about 500 psi) and vent steam to the suppresson pool via the relief valve discharge lines. Steam discharges would also be possible in the event that the relief valve itself was to fail open, although in this case, the steam flow rates would be much 4 less. Also, under normal operating conditions, small amounts of steam and noncondensibles are continuously bled from the heat exchanger into the relief valve discharge line to maintain heat exchanger efficiency. As stated in Reference 3.1.1, "The flow rate...is such that chugging will occur in the pool
, as long as the RHR system operates in this mode.".
The licensee has supplied a plant unique response for RHR discharge pipe water clearing jet loads and for air bubble loads, including submerged structure loads. l The methods used are conventional and correspond, for the most part, to GESSAR , and/or FSAR methods. Details of these analyses and procedures are given in ' Reference 3.1.2. The ifcensee states that, for the most part, the loads are negligible or bounded by MSRV and/or LOCA loads. Where loads were not bounded, the affected structures were reanalyzed to confirm their capability. For C0 and chugging loads on pool boundaries and for submerged structures, the , licensee utilized the generic methods, employed by all the Mark III owners. l A detailed description cf these methods is given in Reference 3.1.1 and 3.1.3. l Generally, the method derives from conservative application of the Mark II CO and chugging load methodologies. Source terms (i.e., forcing functions) are l developed from conservative chugging and C0 pressure signatures selected from ! the Mark II data base. These source terms are applied without any modification to acccunt for the difference in pipe diameter between the RBS relief valve I 4_
-9 discharge lire (12 inches) and that from the data base (24 inches in the test facility) . This is a significant conservatism since it is well established that the source strengths scale with pipe area. The pressure loads generated using these sources were shown to be bounded by MSRV or other design loads.
The adequacy of the C0 load also needs to be judged in the context of the potential for unstable steam condensation; i.e., elevated pool temperatures. This aspect is discussed under Humphrey Concern 3.6. A detailed description of lateral loads on the RHR discharge line due to chugging is given in Reference 3.1.2. The load is time dependent (triangular) with a peak amplitude of 22.2 Kips and a duration of 3 ms. It is uniformly distributed over an application region extending 0.53 to 2.13 ft from the discharge pipe exit. It is stated that this load specification derives from the Mark II chugging lateral load definition on the containment downcomers. It differs in that it utilizes a triangular impulse rather than the half-sinusoid employed by the original methodology (Reference 3.1.4). Also, it differs in that the peak amplitude and region of application have been scale down to account for the difference in pipe diameter between the RHR discharge line (12 inches) and the standard Mark II downcomer (24 inches). This is accomplished by assuming peak amplitude to scale with the 1.7 power of pipe diamter and the application region with the first power. No rationale for changing the impulse shape is provided despite the fact that the total impulse is reduced about 25%. In Reference 3.1.2 it is stated that application of these loads lead to stresses which are acceptable relative to allowables. Also, pipe supports and penetra-tions were found to be qualified. In the staff's judgement the lateral load specification proposed by GSU has several important deficiencies. We have c1 ready noted one of them above, the use of a triangular impulse function without any justification for this choice. In addition, the peak amplitude used by the licensee is not totally consistent with the stochastic nature of the chugging phenomenon, as exemplified in the NRC approved load method. We refer to the fact that the peak load amplitude depends on the total number of chugs that would be expected during a particular accident scenario and the desired non-exceedance probability. As an example, for the DBA LOCA in a Mark II plant with a population of 100 downcomes, a peak design load of 65 Kips was employed to insure that statistically, no member of that population is likely to experience an exceedance of this load during the 100 or so chugs expected during this LOCA. Since it was this value that was scaled down to derive the 22.2 Kip design amplitude, it follows that this choice is acceptable only for a structure that will not experience more than about 100,000 chugs. The licensee has not given us any way of estimating whether this is adequate or not. A final deficiency of the load specification, as we see it, is the failure to recognize the randomness of chugging lateral loads in terms of their direction. For a straight down pipe this could be accounted for simply by applying the load normal to the pipe axis in the direction which would maximum stresses in any bracing or support structures as may exist. For the RBS discharge pipe, however, since it is equipped with d tee at the end, the proper Way to account for directional randomness is to
apply the load in such a way that both axial and torsional stresses arise in addition to those usually associated with lateral (along the pipe axis) application of the load. According to Reference 3.1.5, such an application may have been done but, to date, this has not been confirmed in any way. In order to technically resolve this issue, the licensee needs to provide some reasonable upper bound estimate of the number of chugs, N, that the RHR discharge pipe is likely to experienence during the life of the plant. For this purpcse a chugging period of one second would be acceptable. Note that this is dcuble the one used by the Containment Issues Owners Group in Reference 3.1.6. With
- N cstablished, a suitable value of the peak load amplitude fer application to the RHR discharge pipe, F, would folicw from the relation F = 1.88 1 n N (a)
This relation derives from Equation (6) of Reference 3.1.7 by scaling down for pipe diameter and assuming that the desired non-exceedance probability, P, is equal to the reciprocal of N. To provide some indication of the order of magnitude of these parameters, when evaluating the value of F needed to insure non-exceedance for the case where chugging occurs continuously throughout a 40 year plant life!! This corresponds to slightly more than one billion chugs (N = 1.3x10'). The corresponding value for F derived from relation (a) is then about 40 Kips. Thus, with less than a two-fold increase in the peak load amplitude, the number of chugs the discharge pipe could experience without exceedance is increased by four orders of magnitude. From this perspective it would not appear that the licensee should have any difficulty resolving this concern in terms of peak amplitude. Loads due to other fluid discharges are considered to be bounded by the RHR heat exchanger relief valve loads. In general, this is a reasonable position given the relatively low flow rates and/or smaller discharge line diameters and fluid state. A list of all such relief lines was provided in Reference 3.1.2 together with data characterizing their size and flow rates. One dis-charge line which has some potential for creating significant loading is the RCIC turbine exhaust line. In this case the flow rate / diameter combination results in steam flux rates that are in the chugging regime. Significant loads due to the reflood and air / water clearing loads could also occur since this pipe has a relatively large diameter. The concern relating to the RCIC discharge is addressed by the licensee by citing Reference 3.1.8. This material is intended to support the position that operating experience has shown that there are no dynamic load problems associated with operation of the RCIC system. Conclusion This issue is considered closed for water clearing, air bubble and C0 loads based on the conservatisms used to develop them and/or the wide margins exhibited relative to other design loads.
- l l
l Fer chugging type lateral loads on the RHR discharge line, the licensee needs to rltrify how the selected design loads have been applied and provide a clearer picture o' the loading that will be experienced by this structure durfrg the life of the plant. Otherwise, specification of a 40 Kip icad applied laterally, axially, and as a torque would suffice to close this issue technically. As a result, GSU will not use the RHR system in the SCH: Therefore, based on this administrative control, the staff finds this issue satisfactorily resolved. As for the loads associated with RCIC operation, the steff judges that the long and favorable operating experience of the RCIC turbine exhaust line demenstrates the adequacy of the system. This system employs a load mitigetor (sparger), whereas the RHR relief valve discharge Ifne dces not. Therefore, the staff considers this issue satisfactorily resolved. References i 3.1.1 Attachment I to MP&L Letter No. AECM-83/0146; dated March 23, 1983; from L.F. Dale (MP&L) to H.R. Denton (NRC). 3.1.2 Enclosure to GSU Letter No. RBG-24,131; dated July 31, 1986; from J.E. Booker (GSU) to H.R. Denton. 3.1.3 Ashley, G.K. and Leong, T.S.; "An Approach to Chugging. Assessment of RHP Steam Discharge C0 in Mark III Containments"; Bechtel Report; March, 1984. 3.1.4 Anderson, C.; " Mark II Containment Program Load Evaluation and Acceptance Criteria"; NRC NUREG-0808. 3.1.5 Communication provided during NRC/BNL/GSU telephone conference; September 9, 1986. 3.1.6 Mark III Centainment Issues Review Panel; " Assessment of Humphrey Concerns"; CREARE R&D, Inc.; Technical Memorandum TM-928; July 1984. 3.1.7 Lehner, J.R. and Sonin A. A.; " Determining a lateral Load Speci#ication During Chugging in a Mark II Containment"; Structural Mechanics in Reactor Technology, Vol. J; August 22-26, 1983; pp. 75-80. 3.1.8 Enercon Services, Inc.; Letter No. JRC-0G-142; from J.R. Corn to L.A. England; " History of Development of the RCIC Exhaust Sparger"; dated April 10, 1985. 1
Humphrey Concern 3.4 The RHR heat exchanger relief valve discharge lines are provided with vacuum breakers to prevent negative pressure in the lines when discharging steam is condensed in the pool. If the valves experience repeated actuation, the vacuum breaker sizing may not be adequate to prevent drawing slugs of water back through the discharge piping. These slugs of water may apply impact loads to the relief valve or be discharged back into the pool at the next relief valve actuation and apply impact loads to submerged structures. Evaluation The concern is that the various steam discharge lines may not have been equipped with properly sized vacuum breakers. This is a credible concern in view of the historical development of the same issue for the MSRVs. Because the potential for subsequent actuations was not fully appreciated in the early stages, the MSRV discharge lines were originally equipped with undersized vacuum breakers. When very high reflood elevations were encountered during tests with subsequent actuation, it became evident that this was so and much larger vacuum breakers were installed (from I inch to as much as 10 inches in diameter or 2 at 6 inches in diameter). In Reference 3.4.1, the licensee indicates that the reflood analysis was carried out using an existing GE model (Reference 3.4.2). Additional detafis used in the analysis are presented in Reference 3.4.3. No indication is given there whether credit is taken for the effect of the bleed flow that keeps the pipe pressurized. In Reference 3.4.4 it is suggested that this is the case. However, the ifcensee has provided informal assurance that this is not (Reference 3.4.5). In References 3.4.1 and 3.4.3, the licensee indicates that the RBS RHR heat exchanger discharge lines are equipped with 0.75 inch vacuum breakers. This is reflected by the results reported by the licensee which indicate that maximum reflood lengths exceeding 50 feet. This is surprisingly high value but, according to the licensee, well below any crucial element of the piping system. Specifically, the SRVs are about 12 feet above peak reflood elevation while the vacuum breakers themselves, although at about the same elevation, are only partially flooded by the rising water leg (Reference 3.4.5). In any case, the piping and piping support dynamic loads were evaluated for a second or subsequent actuation at this peak reflood level using, standard methods that are acceptable to the NRC (Reference 3.4.6). It was stated that the structures were found to have sufficient margin to accommodate these loads. Conclusion The staff considers this issue adequately addressed by the licensee and, therefore, closed. References 3.4.1 Enclosure to GSU Letter No. RBG-19,972; dated January 23, 1985; from J.E. Booker (GSU) to H.R. Denton (NRC).
13 - 3.4.2 Wheeler, A.J. , Dougherty, D. A. ; " Analytical Model for_ Computing Water Rise in Safety. Relief Valve Dischcrge Line Following Valve Closure"; GE Document No. NEDE-23898-P; October,1978. 3.4.3 Enclosure to GSU Letter No. RBG-24,131; dated July 31, 1986; from J.E. Booker (GSU) to H.R. Denton (NRC).
- 3.4.4 Mark III Containment Issues Review Panel; " Assessment of Humphrey ,
Concerns"; CREARE R&D, Inc.; Technical Memorandum TM-928; July, 1984 3.4.5 Communication.provided during NRC/BNL/GSU telephone conference, September 9, 1986. 3.4.6 SWEC Computer Code STEHAM, FSAR Appendix 3A. i 4 l
- - , - ,_--,--r -,-r + r--. e ---r,- ,, mm,- - ,.-, , - - - - ,w
i . r. 14 - Humphrey Concern 3.6 i If the RHR heat exchanger relief valves discharge steam to the upper levels of
- the suppression pool following a design basis accident,'they will significantly 1 aggravate suppression pool temperature stratification.
Evaluation
- As stated above, the concern is that because of the presence of a layer of i 1
heated water near the pool surface, the pressure and temperature response in the containment will be underestimated by any analysis which assumes equality between the wetwell air space and the suppression pool-bulk temperatures. In
! this sense, the concern is similar to that expressed by Humphrey Concern ~4.4.
I i This aspect of the concern has been addressed by the licensee via the generic
- approach first provided under the MP&L docket (Reference 3.6.1). A demonstrably conservative model of thermal deposition, pool mixing and thermal stratification
- was developed and applied using GGNS plant parameters. Based on this model, it was indicated in Reference 3.6.2 that during the first 15 minutes or so of this steam discharge, the difference between the average pool surface temperature and pool bulk temperature increases at less than 1 F per minute. In fact, at 10 t minutes into the blowdown this difference is only 7.5 F. Since the RBS analysis assumes that the surface temperature of the pool is always 5 F greater than the 1 bulk temperature (Reference 3.6.3) the resulting non-conservatism will be minimal and easily bounded by the margins that have been demonstrated to exist in heat exchanger performance (Reference 3.6.4).
! Based on our review of the available information we conclude that the results described above are equally applicable to the RBS. In fact, they are probably conservative since the RBS discharge line is equipped with a tee. This provides improved mixing compared to steam discharge through a straight down pipe as assumed in the analysis. Thus, so long as the blowdown does not proceed for i much more than ten minutes, we would not expect any severe adverse effects to +
result from this postulated accident scenario. The licensee indicates 3 (Reference 3.6.5) that operation in the SCM is operator intensive, so that , detection of a failure and termination of the steam flow could be accomplished } within two minutes. This appears somewhat optimistic, but shutdown within j ten minutes is, in our judgement, a conservative and reasonable estimate.- Despite the above conclusion, this concern is not entirely reso1'ved. As noted ' under Humphrey Concern 3.1, this relates to the potential for unstable steam j condensation due to elevated pool temperature. For this case, the behavior of i local pool temperature rather than average pool surface temperature is of interest. According to the analytical model, the trend of this parameter j depends on the total pool volume that participates in direct mixing with the i steam jet. For the " worst case" estimate (mixing volume equal to 10% of the total), a local temperature rise of about 6*F per minute is predicted. This j trend drops off linearly with assumed mixing volume size (i.e., 3*F per minute : i for 20% mixing volume) so that a-certain amount of uncertainty exists in ! i evaluating this trend. i i l m-------i--y--rv a--Yw--+---ww-
-,qw 4-rep r- ---yr y- w7vgy-r -w-rm- % y -
y' w- vy--w y-wr- yw-wy r-*'"- -"tw--r- w---- --
--r>----yg<c--- y-+-~1r -
-^rw**+-
i Based on our review of the available information, we believe the worst case trend of 6 F/ min. to be overly conservative. For example, the response observed in the Quad City plant during a SRV blowdown thrcugh a ramshead discharge device (Reference 3.6.6) averaged only 1 F/ min. over the first ten minutes of the steam discharge. This was with a steam flow rate double that used in the current analysis.* In part, the disparity between this result and the analysis can be attributed to the difference in mixing geometry represented by the ramshead in the first case and the straight down discharge pipe as assumed in the analysis. The strong influence of steam jet orientation or, temperature response is demonstrated by the results observed with SRV discharges through a quencher device (Reference 3.6.7). In this case, with radially directed steam flow, the temperature increased at an average rate of 8'F/ min. with only a modestly higher flow rate (about 50%) than in the Quad City tests. The Quad City results should be reasonably applicable to the RBS despite the fact that the RBS discharge orientation differs somewhat (by 25 ) from the circumferential discharge used in the Quad City plant (Reference 3.6.8). Note that this difference causes only a 9% reduction in circumferential momentum flux. In any case, qualitatively, we would expect an even slower temperature rise rate in the RBS because of the lower steam discharge rate. However, in the absence of any quantitative information connecting temperature response to steam flow rate, we find that this represents a margin of safety for the present application. This margin is needed to cover other uncertainties that exist such as differences in pool geometry as cited above. In a cover letter to the licensee's most recent submittal (Reference 3.6.9) it is indicated that SCM operation is contemplated with suppression pool local and bulk temperatures "below 130 F." This represents a unique procedural change relative to all other Mark III Ifcensees, in which they have commited to non-use of the SCM under post-LOCA conditions. To address the issue for the RBS, we have had to reexmaine the staff position relative to pool temperature limit for a rams-head type device. To quote from Reference 3.6.10, this position is "...a suppression pool temperature limit has not been adequately established for the ramshead device." However, in the same reference it is further stated that "the limited arrount of plant operational data may be considered as supporting data for some specific zones of mass flow and pool temperature." The limited plant data referred to is summarized in Figure 3.6.1 which has been excerpted from Reference 3.6.10. The notation " demonstrated" signifies thatstablesteamcondensationwasattainedeverywhereuptotheindicatep/sec) line. In particular, at the steam flux rate of interest (about 50 lbs/ft stable steam condensation would be expected up to a temperature of 150*F.
*The parameter that controls mixing is the steam momentum. However, since l the exit flow areas and reservoir conditions are all roughly comparable, the !
mass flow rate is equivalent.
i
- For the reasons' we have already enumerated, we' expect that'any steam discharge 3
through the RBS RHR heat exchanger SRV will result in no more than a 1 F/ min. i local temperature rise. Also, we do not expect this discharge to proceed for longer than ten minutes. Accordingly, if failure of the'PCV valve occurs during SCM. operation at pool temperatures below 130 F, we would not~ anticipate the occurrence of containment loads associated with unstable steam condensation
; due to elevated pool temperatures, i
Conclusion i . . . ! The issue raised by this concern is considered to be satisfactorily resolved for the RBS. In particular, operation in the SCM would be acceptable at' local suppression pool temperatures up to but not exceeding 130 F. ) References i 3.6.1 MP&L Letter No. AECM-82/574; dated December 3, 1983; from-L.F. Dale (MP&L) to H.R. Denton (NRC).
- 3.6.2 Meeting Handout " Response to Question 9.2"; NRC/ Mark III/GE j Meeting; May 19 and 20, 1983.
i 3.6.3 River Bend Station FSAR Section 6.2.1 3.6.4 Attachment to MP&L Letter No. AECM-82/353; dated August 19, 1982; l from L.F. Dale (MP&L) to H.R. Denton (NRC). ) 3.6.5 Enclosure to GSU Letter No. R8G-19,972; dated January, 23, 1985; from J.E. Booker (GSU) to H.R. Denton (NRC). 3.6.6 Dougherty, D.A., " Suppression Pool Temperature Response to a Safety / Relief Valve Discharge Through a Ramshead in Mark I and II Containments"; General Electric Co. Report NEDC-23689-P, j Class III; March,1978.
! 3.6.7 Patterson, B.J.; "Monticello T-Quencher Thermal Mixing Test.
] Final Report"; General Electric Co. Report NEDC-24542-P, Class i III; April, 1979. } } 3.6.8 River Bend sketches (portions of RHR piping); dated O vember 15,
- 1983 provided by the GSU to the NRC in November, 1986.
! 3.6.9 Enclosure to GSU Letter No. RBG-24,131; dated July 31, 1986; from l J.E. Booker (GSU) to H.R. Denton (NRC). i ! 3.6.10 " Mark'II Containment Lead Plant Program Load Evaluation and - Acceptance Criteria"; NUREG-0487; October,1978. I l i 1 k
m A
)
A (T a TN R AA T I SL P D
. E T .
A R T S . N O M _ E D , t
- a
, t a
D
^ ,
l a n F ) i o
- C C t N 0 ,
E a O 6 1 ) S r I - T C 2T e p A ( T D ' F O T N N E R V V M A A H R M O C T S L l
- 1(
L S t E P , n R x a s
' u l l P
'E , F G s S 1
- ' S
- h I
6 M 3 A e _ E r T u _ S g _ C i E F S M 2 ) I _ B L T F . , ( A % _ 0 D _ N J \ 4
- L i- N E
\ ,- s II g
s
)
C ( D - N E
, - - _ . ~
0 0 0 0 W 2 5 1 0 1 5 a- ,W C f2 '-{ _h _ _}}