ML20203H353: Difference between revisions
StriderTol (talk | contribs) (StriderTol Bot insert) |
StriderTol (talk | contribs) (StriderTol Bot change) |
||
Line 17: | Line 17: | ||
=Text= | =Text= | ||
{{#Wiki_filter:}} | {{#Wiki_filter:Docket No. 50-423 B17049 Attachment 2 Millstone Nuclear Power Station, Unit No. 3 Transport of a Small Air Pocket Creare incorporated TM-1876A February 1998 February 1998 | ||
. . __ _.. =_ _ | |||
?M *?*deo M*dtaal: | |||
G PDR | |||
l 1 | |||
TitANSI'OllT OF A SMALL Allt POCKET TM-1876A Febiuary 1998 | |||
s NON Q.A. | |||
TRANSPORT OF A SMALL AIR POCKET Prepared for: | |||
Northeast Utilities Services Company Prepared by: | |||
Ubm N 9V Paul H. Rothe /** Date Vice President IihNn P. hV | |||
'\9illiam E. Nutt | |||
"' 2/M/98 | |||
' Da'te , | |||
Engineer i | |||
Creare incorporated TM 1876A PO Box 71 Project 8084 Ilanover,NH 03755 February 1998 | |||
O TM 1876A TAllLE OF CONTENTS | |||
: 1. S U M M A R Y m........................... .. .... .... ... .............. ...........................................................! | |||
: 2. It A CKG ROUND AN D INTRODUCTION .. ... . . . ..... .. ...... . .......... . .. ..... .. .m..... .1 2.1 I s s Ul! D u 1N IT10N .... .... .. .. .. . .. ........... ..... . . . . .. . . .m.. ...... . ...................................m...........................I 2.2 SiONII1CANCE OF GAS POCKl.T SIZE....... ...... .............................................................. ;.2 2.3 SlONII1CANCl! OP lilE FORM OF lilE O As ..... . .. .................. ......m .m..............m . | |||
.................2 | |||
: 3. TEST D EF l N I TI ON ...... .. ... ... . ...... . . .....m ........... ............. ... ... m. . .... . . .. . ........ 3 3.1 OEOMETkY .... ...m..... ....... ..............m.... .. 3 3.2 Finw SEQUENCE .......... ................ .m. .... ............... =3 3.3 EX AMIN ATION OP TEST UNCERT AINTIES .... ...... ..... ...... ....m......... .................. ....... ............4 3.4 REPUCATION... ..... ..........................................................................................................4 | |||
. 3.5 SCAUNO.... ....... ...... ............. . =5 3.6 OTilER UNCERTAINTIES ; ........ ...................- =5 | |||
: 4. TEST FI N D I N G S . ..... . . . . . . .... .. ... ............ .... . .. . ..... .~....... ..... ....... .. 5 | |||
: 5. INTE R P R ETATl ON .......-.m. .... ~.m.. ....... ..... ..............~.. ...m ...... . . .7 | |||
: 6. R E F E R EN C ES... ... . . . .~........ . --.... .. .~.. ... .. ~ - .~ -.. . .... 9 A PPENDIX A FA Cl LITY .. . ... .. . .. ....-. . . . ........ .... ~ ... .. .. --. A . I A PPENDIX 11. TEST PROCEDURE..... . mm .... .. .~ . .- .....-.-~.. ....~.. ~ .~ .. ..11 1 APPENDIX C . SUPPLEMENTAL CALCULATIONS. . ... ... .. .. ........ . . ... . C .1 A PPENDIX D . SCALING STUDY.. . . .... . . . .. .~ .~ -~.. - ~ .~....D.1 i | |||
. _ . . -= -. . - - - - . _ - - _. _. - - | |||
TM 1876A . | |||
l | |||
: 1. | |||
==SUMMARY== | |||
Tests show that a small air pocket tends to distribute as it is transported through a tortuous array of piping. There are three key findings: | |||
1, The initial air pocket is very small; only five pipe diameters long, | |||
: 2. The form of the air is benign, tiny bubbles, and | |||
: 3. The void fraction was approximately 1% at a pump simulator. | |||
The facility is a transparent model of a portion of the A train of the long term cooling system at Millstone Point 3. The model piping extends frorn the RSS expansion seal upstream of MV8837A to the safety injection pump location following MV8807A. The testing was intended to illustrate the governing multiphase Dow phenomena with confidence and to quantify the rate and form of air transport to pump locations. | |||
: 2. HACKGROUND AND INTRODUCTION 2.1 Issue Dennition During various loss-of coolant accident scenarios for Millstone Point 3 (MP3), a small air - | |||
pocket is trapped in a theimal expansion loop. The air pocket is trapped inherently upon Olling by water of an ordinarily air-filled pipe line to containment spray headers. Later, that air pocket , | |||
is probably ingested by the safety injection or charging pumps, with potential for damage to the pumps. | |||
4 Tests at Creare and analysis by NU establish the size of the air pocket as 1.27 cu ft (best estimate) in 8 inch diameter pipe. Such an air pocket is approximately 5 pipe diameters in length. It is also approximately 80% of the internal volume of the safety injection or charging pumps found by NU to have a volume of approximately 1.6 cu ft. Such an air pocket is a small fraction of piping that extends over 150 pipe diameters from MV8837 to the safety injection pump. Creare's Technical Memorandum TM-1869A is incorporated by attachment and addresses this portion of the work to define the size of the trapped air pocket. | |||
The behavior of pumps when ingesting a water gas now was a subject of extensive research during the 1970s by Creare and others. The findings are described for example by Swift and Block (1982), Rothe and Runstadler (1978), and Patel and Runstadler (1978). A technology summary was prepared by Swift (1982) to assess the state of the art. The cited research encompr.ssed head degradation and resultant unstable operation of pumps. Related technology of cavitation and stall (loss of prime) of pumps is also well established. | |||
A number of reportable events in nuclear power plants during the late 1980s and through the 1990s involved the discovery of unvented high points in the piping to charging pumps. | |||
Information notice IN88-23 of the U.S. NRC summarizes the event at Farley which found a trapped hydrogen pocket of approximately 60 cu ft volume. Subsequent analysis by Creare of this trapped gas pocket was reported at a public meeting attended by Westinghouse and provided by Westinghouse to NU. Swift and Rothe (1990) found analytically: | |||
) | |||
. _ __ _ - _ . _ . _ _ . _ _ _ _ _ . _ _ . - _ - . ~ __. -. . _ | |||
b TM 1876A | |||
: 1. A trapped hydrogen pocket of 60 cu ft in a 6 inch diameter line, | |||
: 2. Subjected to a 6700 gpm flow transient from the RilR pumps, | |||
: 3. Would travel in six seconds through a short run of pipe, | |||
: 4. Arriving at the charging pump as a long and intact column of gas, | |||
: 5. Woyld fully stall the pump, driving residual water to the last few stages, | |||
: 6. Would subject the pump to a massive clearing transient and unbalanced loads, | |||
: 7. With calculated excessive shaft deflection and other mechanical events, and | |||
: 8. leading probably to prompt seizure of the pump. | |||
U.S. NRC concern for such an event is understandable, and indeed, there have been four supplements to IN88 23 as well as unrelated reports of findings of large quantliles of unvented gas and reports of various pump behaviors. | |||
The present report addresses the likely insignificance of a much smaller air pocket, transported to a pump at rnuch lower velocity, in a piping geometry unique to Millstone Point 3. | |||
We find that the air pocket distributes to a long chain of fine bubbles. Our opinion is that this specific flow is benign. | |||
2.2 Signincance of Gas Pocket Size On inspection, a dominant difference between the subject scenario at MP3 and the reported Farley event is the size of the trapped gas pocket, approximately 60 cu ft at Farley and approximately I to 2 cu ft at MP3. Moreover, the larger diameter piping at MP3 has nearly twice the cross-sectional area as that at Farley, the piping is much longer at MP3, and the pumps have twice the volume. | |||
Gas pocket size may be considered first in relation to the piping through which it must be transported. A !arge column of gas near the pump, such as that at Farley, has no possibilhy of appreciable collection in the various air traps presented by the piping. In contrast, a small air pocket such as that found at MP3, will have difficulty proceeding toward the pump through a circuitous pipe network. 'ine piping as built is a nest of dead legs and components all prone to trap air. Pipes slightly sloped from the horizontal will on average tend to trap gas. | |||
Gas pocket size may also be considered separately in relation to the pump. A massive gas column, such as that found at Farley, will surely evacuate, stall, and unbalance the pump. A small air pocket such as at MP3 (even if it were transported compactly in its entirety to the pump location) tends to be a brief and one time transient for the pump. | |||
2.3 Significance of the Form of the Gas Gas in water can take many forms, known as flow regimes. The hydrogen gas at Farley began as a long column filling approximately 300 feet of piping. The operating scenario subjected this long column to the full, transient head of one pump (the RHR) while restricted by another (the charging pump). The initial, transient advance of the water into the gas column was 6700 gpm (approaching 80 ft/sec in water) which effectively pushed and compressed the gas column. The gas moved intact through the piping over the short travel distance to the pump. | |||
2 | |||
b TM 1876A At MP3 the water velocities are much lower, of the order 10 ft/sec or less, initially the water is stagnant, allowing the air to drift and spread. Then the water proceeds at velocities of approxhrately four feet per second, which is well below the condition to clear the gas as a column. M : ' sie gas simply rises as bubbles or layers to fill the tops of pipes as it is dragged | |||
, along slowly. by a now of water underneath it, importantly, the piping at Mp3 includes downcomers, where layers of gas are prone to break up into small bubbles as they are carried downward contrary to the force of gravity. | |||
[ These effects related to flow all tend to spread out the gas and promote a bubbly flow. | |||
They are enhanced by the many traps and obstructions in the piping as the ps must navigate past dead legs, through valves (including part open check valves) and the like. Thus, not only does j realistic piping tend to trap some of the air. It also tends to dilute it with water and render it in a relatively benign, bubbly form. Contrast this form with the description of the Farley situation, which was calculated to involve a long column of gas and slugs of water repeatedly passing through the pump. | |||
: 3. TEST DEFINITION 3.1 Geometry The subject piping at MP3 extends through some 30 independent pipe segments, with numerous valves and tees, and with suitable modeling of dead legs where there is no Dow. The facility diagram, component description, and basis in NU drawings are in Appendix A. The facility is one-quarter scale. Facility piping is primarily 2 inch diameter; comparable piping at MP3 is primarily 8 inch diameter. NU identified Train A as having greater safety significance than Train B. All tests at Creare modeled the Train A geometry. | |||
3.2 Flow Sequence The safety analysis branch at NU has evaluated numerous potential LOCA scenarios on a screening basis using the RELAP computer code and also based on engineering judgment, as reported by NU. An exemplar scenario was specified to Creare that involved three key steps: | |||
: 1. Open MV8837, no Dow, air pocket migration by gravity for 100 seconds, | |||
: 2. Open simulators for MV8804/CV36, begin Gow at 560 gpm plant equivalent for 50 seconds, and | |||
: 3. Open simulators for MV8807, step-up flow to 1230 gpm plant equivalent, until the gas trantportation process ends. | |||
This scenario was identified by the Safety Analysis Branch of NU as the most safety significant now sequence. | |||
We examined the flow sequence in two ways. First, only steps 1 and 2 above were performed, and air was collected at the inlet to valve VS (simulator of MV8807) and in the bypass line P25 leading in MP3 to the charging pumps. That is, we sensed the arrival of air (if any) at tee T5 where the flow splits between the Si and charging pump inlet pining, and we 3 | |||
1 TM 1876A measured the quantity of air collected at the location during the period of time significant to the sequence, in this way, we could determine definitively if there were any adverse buildup of air at valve V5 (MV8807) before it opened. (There was not.) We could also confirm the NU analytical finding that air preferentially passes to valve VS (MV8807) and little carries under to the charging pumps. (We concur.) lly stopping the test at this point in the flow sequence, we also took the opportunity to measure air standing at all locations and to perform an integral check on air volume throughout the facility. | |||
I Secondly, we performed the entire three step flow sequence above. The procedure is detailed in Appendix II. This involved additional air collection stations at other locations, and more stringent verification of their experimental uncertainty, again as described in Appendix II. | |||
That is, the pump simulators in the facility are air collection stations. They are made of segments that expand in diameter and slope upward (to prevent air entrapment) then proceed to a small | |||
* diameter riser for the purpose of observing and timing the rate of air collection. | |||
NU identified a second now sequence of some (but considerably lesser safety significance), it involves two steps: | |||
: 1. Open valve MV8837, no How, air migration by gravity for 100 seconds, and | |||
: 2. Open valves MVE804/CV36, flow at 900 gpm. l That is, by comparison with the primary flow sequence above, the initial flow (900 gpm) is higher (than 560 gpm), but it does not step up to 1230 gpm This test was performed only in the first form above. That is, valve V5 was not opened, air collected at its inlet was measured (very little air was collected). Continued flow at 900 gpm was felt to be less challenging than the step-up to 1230 ppm that had already been tested. | |||
3.3 Examination of Test Uncertainties All tests or analyses have modeling uncertainty. Therefore, an important part of the test definition involved devising means to i.'luminate and quantify these uncertainties. The identified uncertainties were of four kinds: | |||
- 1. Replication and sensitivity, 2, Scaling, | |||
: 3. tvmponent modeling, and | |||
: 4. Instrumentation. i | |||
- NU derived additional confidence from the collaboration of their RELAP calculations with Creare's tests liere we consider the uncertainty of the tests as if they stood alone. | |||
3.4 Replication The tests as performed were intended primarily to illustrate the governing physics of some complicated and unsteady phenomena. That is, these were not meant to be precise | |||
-- production experiments, as would be carried out for example to obtain the flow characteristic of a 4 | |||
O TM 1876A valve. Still, we were determined to assess the sensitivity of the results to acasoned variations of parameters, in part, we did so simply by repeating the key tests a number of times and with varied observers, in part, we adjusted the slope of cedain pipes, predictably with measurable effect. Lastly, we altered the size of the initial air pocket introduced into the facility, in this way, we examined the sensitivity of this test to that input assumption, and we segregated the phenomenon of air trapping from the phenomenon of air transpodation. These uncertainty studies are described in Appendix 13 and are the sole cause of the data scatter reported in Table i below. | |||
3.5 Scaling cude scaling is a well established principle for examination of the stratified multiphase How :egime and its transition to other regimes. At the one quarter scale tested, flow velocity is reduced by a factor of two when Froude number is preserved. Reynolds and Weber numbers are distorted proportionally. | |||
Although these are minor distortions by comparison with the degree of scaling employed routinely in the past by the research community concerned with nuclear reactor safety, it was straightforward to perform a scaling study that provided added confidence in the results. One of - | |||
the pipe segments was identified both on inspection and -in subsequent tests as phenomenologically significant. It involved flow in a riser to a mn to a downcomer, then a nm, and served as a location both to hold air and to disperse it as a chain of small bubbles. We tested this segment of the geometry at two scales; one-quarter scale and one half scale. | |||
As might be expected, the scaling study confinned onu again the suitability of Froude number to scale a stratified now. Further, it confirmed that the bubble production behavior was only slightly modified by scale, and in a manner consistent with expectation. Thin aspect of the work has some original scientific interest to illustrate the behavior of interacting pipe segments. | |||
From an engineering point of view, and for NU's safety determination, it merely confirms the low uncertainty introduced by scale modeling. The scaling study is reported in Appendix D. | |||
3.6 Other Uncertainties The details of the component modeling are described in Appendix A in the context of the geometry detail presented there. The details of the instruments and their uncertainties are described in Appendix B in the context of the detailed test procedures described there. Various-support calculations are described in Appendix C. | |||
: 4. TEST FINDINGS The test findings are the subject of a brief illustrative video tape. This video has been supplied to NU and was played at the enforcement hearing held at Millstone Station on January 13,1998. | |||
The initial small air pocket breaks up to a chain of tiny bubbles, and arrives at the safety injection simulator over_ a period of approximately a minute. Littic or none of the air is 5 | |||
b TM 1876A l transported past the tee and toward the charging pump, a result of centrifugal force at a pipe tee flowing bubbly air water along its top. A significant fraction of the initial air pocket is trapped at various locations throughout the long piping network, The time scale of the scaled facility is twice real time. That is, a minute of facility time is equivalent to two minutes of plant time, in two minutes, a flow of 1230 gpm passes 2460 gallons of water. The initial trapped gas pocket is 1.27 cubic feet or 9.5 gallons. Thus, the air is diluted by water, on average, by a ratio of nearly 300. | |||
At the location of the safety injection pump, air arrives in a manner that is distributed . | |||
over a minute in time but is initially at a rate somewhat higher than average. Table I below summarizes our findings in comparison with the RELAP calculations. Various appendices elaborate. | |||
Table 1. Millstone Unit 3 Pump Inlet Vold Fraction Results Scenario' MV8837 Flow Computer Laboratory Laboratory 3 | |||
Flows Duration RELAP Results Results' (gpm) (sec) Results IX Air. 2X Air (NU) (Creare) (Creare) | |||
: 1. RSS A to 560 150 0% 0% 0% | |||
CliG A 3 | |||
: 2. RSS A to 560-1230 until air 2-4% 0.8 to 1.7% 3 to 7% | |||
Sill /CIIO A fully transported | |||
: 3. RSS A to 900 150 <l% <l% Not Run CilO A&B Notes: | |||
8 See NU Analysis Report M3.EV.98-0018 Rev 0 for description of scenarios. | |||
8 RELAP calculates a detailed transient with an average of 2 to 4 void fraction for 10 to 15 sec. 'Ihe range is that of sensitivity calculations. Instantaneous peaks of 5% and 9% respectively are calculated. | |||
8 Laboratory results inherently average air collection over a period of time. However, tests show that the collection rate is relatively constant during the significant period (unlike the RELAP calculation of a peak). Accordingly we compare the averrges by both approaches. | |||
* 2X air is simply twice as much air as the best estimate volume of the initial air pocket. Though not prototypical, the test shows the sensitivity to various uncertainties. | |||
6 | |||
b TM 1876A | |||
: 5. INTERPRETATION I The mechanical response of a pump is amenable to analysis, test, or to manufacturer representation based on knowledge of its design and experience with like equipment. Pump manufacturers design their equipment to tolerate imbalance ordinarily expected based on manufacturing tolerances or adverse loading. Considerable experience is also gained from long operation of many pumps and from specialized tests. In this respect, the mechanical response of | |||
'; a pump, its susceptibility to mechanical vibration, to excess shaft deflection, bearing wear. life degradation, or even seizure, lie within the ordinary design envelope of ti e equipment. Still, it remains that centrifugal pumps generally and the specific charging and safety injecting pumps at Mp3 are not produced nor qualified for the specific loads due to gas ingestion. Accordingly, NU - | |||
must rely primarily on representation by the pump manufacturer The paragraphs below supplement such representation by logic and by review of the known draa. | |||
Pump head degradation is a process of accumulation of air in a pump, over time, ; | |||
sufficient to replace the water within the pump and prevent the achievement of full centrifugal head and the flow distributloa on which pump effectiveness miles. Such accumulation of air results from a balance between body forces tending to trap air in the pump (centrifugal and l coriolis forces) and the drag of water on the air bubbles and '- yers. Both the fraction of arriving | |||
; air and the duration of air ingestion affect pump performance. | |||
There are no known tests of pump performance when subjected to an air ingestion transient of such brief duration (and air volume) as is of present interest. We suspect that when the entire air pocket is smaller than the intemal volume of the pump, as is the case here, then the odds and severity of appreciable performance degradation or mechanical damage are significantly mitigated. If further, the air is highly distributed to individual bubbles, as also is the case here, | |||
- then the event is further mitigated substantially. These are qualitative observations. It remains that the subject pumps have not been tested to quantify their mechanical tolerance to brief air ingestion at any volume or fraction. No such tests have found damage, nor have they assured continued performance without pump life degradation. For perspective, pumps are routinely designed for much more severe duty of this kind. (For example, certain fuel pumps accept | |||
, sudden intake of liquid in a previously gas space at high liquid flow rates.) Thus, the question is not whether centrifugal pumps can be made in general to withstand such duty, nor whether the subject pumps are mgged. They can and are. Rather, the tolerance of the particular subject pumps has not been tested for this narrow and specific duty. Fortunately, the present event appears to be so mild as to obviate the need for such completeness of evaluation. | |||
4 Extensive research [ Swift,1982] demonstrates no appreciable (reliably measured) head L degradation for inlet void fractions below 2% even when air has been supplied to the pump for an extended period. This is a very stringent criterion involving the ability to detect any effect whatsoever, in research experiments, with air supplied at a given rate forever. | |||
The onset of barely measurable pump performance degradation in research experiments is itself a relatively mild situation. Indeed, researchers can begin to detect the accumulation of air in the pump and corresponding reduction of head when as little as 2 to 5% void fraction is 7 | |||
Nb TM 1876A supt. lied at pump inlet. The pump does not pump as well, though it continues to pump even when supplied a void fraction of up to 15% air for an extended period, in the present situation with a highly distributed flow of a small air pocket, the possibility of pump stall is remote. | |||
Creare finds air arriving at its pump simulator inlet in the range 0.8 to 1.7 % by volume for a period of about 15 seconds. Thereafter air continues to arrive at substantially lower void fraction for approximately one minute. This is a very low total volume of air amounting to a fraction of the initial air pocket, which in turn is smaller than the internal volume of the pump. | |||
Moreover, the air arrives at pump inlet well diluted by water and at fractions shown by research to be unmeasurable in terms of degradation of pump head. Reason would suggest that the unbalance loads produced thereby will have minimal effect on pump mechanical performance, as V/cstinghouse states is the opinion of the manufacturer. | |||
i The onset of now vibrations and surge results from an interaction of the pump's performance with the flow system, particularly its flow inertia, resistance and compliance [Rothe, 1978). Absent a specific system analysis or test,little can be said of the susceptibility of a given system to surge, llowever, in an event of brief duration, there is little prospect of surge, lacking time to develop flow oscillations. Thus, either the void fraction at pump inlet is too low for there l to be any appreciable head degradation and surge, or the duration of head degradation is too brief I for the system to respond and surge. Either way, surge is unlikely, i | |||
in summary, the small initial air pocket at MV8837 is expected to be transported to and ingested by the safety injection pump. The considerations specific to the subject scenario are: | |||
: 1. An initial air pocket of 1.27 cu ft volume, | |||
: 2. A large pump, larger than the air pocket, | |||
: 3. A tortuous run of piping that is 50 times this volume and contains numerous air traps, | |||
: 4. An overhead horizontal run to a downcomer where air is distributed gradually as ; mall bubbles, and | |||
: 5. A flow rate sequence particular to Mp3, specifically a; Stagnation for 100 seconds, b) 560 gpm Cow for 50 seconds, and c) 1230 gpm Dow thereafter. | |||
For the scenario involving all of these specifics, we believe that the outcome would have been benign. Air arriving at the charging pump would be of brief dur .on (of the order of a minute), low void fraction (of the order 1%), and low total volume. Based on reason and the literature of the field, we expect none of stall, surge (now oscillations), appreciable imbalance loads, pump seizure, nor pump life degradation. In the absence of specinc mechanical test data, we rely on the pump manufacturer's assurance of the suggedness of the unit and its design with a factor of three margin for imbalances of various kinds. | |||
8 i | |||
b TM 1876A | |||
: 6. REFERENCES Swift, W.L; Block, J.A.; Model Pumo Performance Program Data Report; EPRI NP-2379, Research Project 3471, Interim P.cport, Electrical Power Research Institute, Palo Aho, CA, May 1982. | |||
Swift, W.L; Kamath, P.S.; Runstadler, P.W. Jr.; Block, J.A.; Two Phase Performance of Scale Models of a Primary Coolant Pump; EPRI NP 2578, Research Project 3471, Final-Report, Electrical Power Research Institute, Palo Alto, CA, Sept.1982. | |||
Rothe, P.ll.; Runstadler, P.W., Jr.; Dolan, F.X.; Pump Surge Due to 7ko Phase Flow; Polvohase Flow in Turbomachinerv, C. Brennen, P. Cooper and P.W. Runstadicr Jr., eds., | |||
Publication 11000123 of the American Society of Mechanical Engineers,1978,121 137. | |||
Putel, B.R.; Runstadler, P.W. Jr.; investigations into the Two Phase Flow flehavior of Centr (fugal Pumps; Pohphase Flow itLTurbonachinery; C. Brennen, P. Cmper and P.W. | |||
Runstadler, Jr., eds., Publication H000123 of the American Society of Mechanical Engineers, 1978,79 100. | |||
Swift, W.L; Kamath, P.S.; Tantillo, T.J.; An Assessment of Residual lleat Removal and Contalmnent Sprav Pumo Performance Under Air and Debris Innestinn Conditions. | |||
NUREG/ CR 2792, Creare TM 825, United States Nuclear Regulatory Commission, Sept,1982. | |||
Swiit, W.L; Rothe, P.iI.; Evaluation of Operability of Charging Pumps; Creare MTG-90 7-302 presented to the Nuclear Regulatory Commission, Rockville MD, July 14,1990, 9 | |||
bb TM 1876A i | |||
Al'I'ENDIX A FACILITY j A-1 | |||
) | |||
b TM 1876A Baseline Facillty at One Ouarter Scale ! | |||
The schematic of the one-quarter scale facility is given here as Figure A1. The diagram serves to number and orient the piping, which includes: | |||
^ ' | |||
4 e llorizontal runs, | |||
* Downcomers (vertical pipes where the flow proceeds downward), and e Risers (vertical pipes where the flow proceeds upward). | |||
Table Al lists the pipes that match those at MP3 and in the NU/RELAP model of MP3, and compares the as-built dimensions with the plant dimensions. The plant was closely matched at one-quarter scale. | |||
Dimensions of the test facility are from as built measurements and can also be found on , | |||
Figure Al, The dimensions of the Plant were taken from a variety of drawings, sketches, and conversations with Northest Utilities Company. The drawings were from the Stress Reconciliation Piping Location isometric drawings 25212-20394 Sil.8,9, and 26; 25212-20404 Sil10; and 12179-C.I. RSS12, Sl! 3. Sketches were titles RSS IlX A Piping. Additional information came fiom Figures 4,5, and 6 of some drawings of the RELAP model titled MP3 RECIRC. PHASE AIR BUBBLE STUDY TRAIN 'B' FAILURE. | |||
Several additional pipes are on Figure A1, for facility purposes such as air collection. All | |||
, dimensions are shown. As described in the test procedures, the facility was refined in minor ways (for example addition I air collection stations were used) fev. some of the supplemental tests. | |||
Three pipes are dead legs either closed off or to a downcomer. Of these one tends to | |||
^ | |||
collect air, while two do not. There are slight differences between the facility and the plant (the facility lengths of dead leg are shorter in the two cases that don't collect air anyway). We believe that these differences are insignificant and also conservative. | |||
A-2 | |||
2.g,,,a,_-.,._.,d,,,44ML'8"'M-' | |||
@eare - pp Tu is76A E | |||
a - | |||
3k | |||
& Wi u ., , | |||
e S | |||
@'i[.4 e% | |||
e li. | |||
l 4,,' JD s | |||
O' Il* , | |||
hn4 t u3 i ! -e N 1 j- ---l g | |||
~ | |||
s e' ' | |||
s | |||
! rI.., .,.'M*-e D.. - | |||
, N. T ' 'x a | |||
l - | |||
N e 4 e | |||
/ 9 6'K# "E | |||
,l e, | |||
y A | |||
e | |||
,g%n f[31ibl,E3dy,e . | |||
44HH M4 4 4444k- --- ' | |||
g | |||
,1 | |||
TM 1876A TAllLE A1. ONE QUARTER SCALE PIPING Test Facility Dia. Length RELAP Plant Plant Comments Piping # In. Ud Number Dia. 1/d PI 2.0 7.3 120 8 6.4 P2 2.0 11.7 130 8 11.6 P3 2.0 4.0 140 8 3.7 P4 2.0 6.5 150 8 6.0 P5 2.0 13.7 157 8 13.5 9 feet to downcomer. | |||
P6 2.0 10.3 160 8 9.6 P7 2.0 26.3 165 8 26.5 P7A 2.5 2.4 170 10 2.0 P8 2.5 3.0 NA 10 Unknown, doesn't collect air. | |||
19 2.5 3.6 , 180 10 3.7 P10 2.5 16.4 190 10 16.2 Pt1 2.5 13.2 200 10 13.2 P12 2.5 16 211,213, 10 37.8 Approx. 31.5 ft to down. | |||
215,217 comer. Doesn't collect air. | |||
P13A 2.5 1.2 210 10 1.8 P13 & 14 2.0 20.7 220 8 17.0 P15 2.0 6.5 230 8 6.5 P16 2.0 4.0 240 8 3.8 P17 2.0 6.3 250 8 5.9 P18 1.5 4.7 260 6 4.4 P19 1.5 8.3 270 6 20.0 Conservatively short. | |||
P20 1.5 16,0 300 6 15.6 P21 1.5 9.3 330 6 9.4 P26 1.5 9.3 320 6 9.4 P27 1.5 14.7 340 & 350 6 13.1 A-4 | |||
S TM 1876A Table A2 provides a chan of facility components. Elbows and tecs are similar to those in the plant. Valve detail data were not available to Creare during construction and are partially available at this writing. Swing check valv' 'he plant are expected to be partially open (about half open) at the subject flows. This degre spening will present a " smile" opening at the top of the pipe (where the air is), thereby tendinh .o trap and distribute air. Create used a partial port (80% open diameter) ball valve at check valve location CV36, a geometry that tends to trap some air while passing the rest in an unhindered manner. Creare did not model another check valve in the plant at all. Motor operated plant valves are generally gate valves, with a stroke over some time period of at least 10 seconds, and an opening that may be full or partial (unknown to us at present). Gradual opening will tend to ramp How, though not to appreciable degree at these now rates and port sizes. Creare used a conservative abrupt opening. Partially open gate valves will tend to trap a!r, as did Creare's panial port ball valve. On balance we view these differences as minor, given two overriding geometric attributes: | |||
* Entrapment and distribution of air due to gravity in horizontal pipe runs, and e Breakup to fine bubbles in downcomers That is, the effects of gravity appear to override component details. | |||
A-5 | |||
O TM-l876A Table A2, One Ouarter Scale Facility Comisonents Piping clear acr3 11c (0.25 inch walls). | |||
1.1.5 inch ID PI8 P21, P26, P27 | |||
: 2. 2.0 inch ID. Pl P7,, P7A, Pl3A, Pl3.Pl7, P22, P25, P29 3, 2.5 inch ID P8 Pl2 1 | |||
Valves I, 2.0 inch partial port ball valves. VI V4, V6, V8, V12 2.1.5 inch gate valves. V5, V9, VIO. | |||
Valves 2,3,4,5,9 and 10 represent valves in the plant. All of the other valves in the test facility are control valves necessary to run the experiments, in the plant, valve MV8837A is a motorized gate valve which opens over a period of 10 seconds; wiercas in the test facility, it is a manually operated partial port ball valve. The details of the geometry and the degree of opening are unknown to use. We believe the use of the ball valve is either accurate or conservative because it does not trap or hold up appreciable gas. | |||
Valves 4,9 and 10 in the test facility and the plant are all gate valves. All though we have no detailed geometry of the plant valves, we believe that the choice is conservative, because we have observed no signincant gas trapped in or tefore the test valves. | |||
Elbows . Standard PVC nttings 1.1.5 inch,90 degree. E8-Ell 2, 2.0 inch,90 degree. El, B2, ES, E12 El3, E14 | |||
: 3. 2.5 inch,90 degree. E3, E4 | |||
: 4. 2.0 inch,45 degree. E6, E7 | |||
: 5. 2.0 inch 22.5 degtce. A1. A2 Elbows E2. Ell and E23 model cibows in the plant. llorizontal cibows were not modeled. All of the other cibows in the facility serve te support the plumbing necessary to run the experiments. The elbows were chosen primarily based on what was immediately available and generally have a radius of curvature of I 5 times the Jiameter. | |||
Teen . Standard PVC fittings | |||
: 1. 2.0 inch. T2 | |||
: 2. 2.5 inch. T3, T4 | |||
: 3. 2.0 x 1.5 x 2.0 ineh. T5 | |||
: 4. l.5 inch T7,T8 1 acking detailed information on the plant elbows wo used standard off-the-shelf right angle Tee's for the facility. Tee's 2 5,7 and 8 represent Tee's in the plant. | |||
110w Metert | |||
: 1. Flow meter #1. King Instrument Co. Rotometer,7205-0201 W 5-40 gpm | |||
: 2. Flow meter # 2. Omega Eng. Inc. Rotometer, PL-75F,4-40 gpm | |||
%e stated accuracy of the Omega meter is 2.5% full scale. The King instrument unit should be similar. | |||
We have noted no discrepancies between the two meters. The King instrument was also crudely checked by performing an actual measurement of its flow rate. The now was set at 10 gpm and fed into a known volume. The fill time was measured and the now rate calculated. %ey were found to match. | |||
A-6 | |||
i S TM 1876A ! | |||
l t | |||
Half Scale Facility B | |||
A segment of the pipe network (four pipes) was modeled at hath one quarter and one half ! | |||
scale. Figure A2 shows the half scale pipe segment, which consists (in flow direction) of: j e Riser P4 - | |||
* Run PS | |||
* Downcomer P6 i e Run P7. | |||
l There are corresponding pipes in the one quarter scale facility. i t | |||
Table A3 compares the one-quaner scale and one-half scale facilities with MP3 plant pipe i dimensions. , | |||
I F | |||
k t | |||
e t | |||
I | |||
-t M | |||
4 A-7 | |||
#,r y co--- , . . . . , , , , , , ,+g_,, , , , ,, , mm,, , , _ _ , . ..m,.-_.#.___. ,p ,_ -, ncm_..,. f.-e__y. . ,yr, , y , . , , , ,_.e- | |||
. . .,-m..-.~.~.__.. _.~~ . - _ _ ____. . - . - - _ , - . _ . . . ..._--._.__m_..m__.m.._ m- _ _ . _ _ _ _ _ _ . . _ . . . . _ . - | |||
t TM-l876A i, | |||
t 4 L i | |||
J l- . | |||
PS g- , | |||
y | |||
@~ s78 u | |||
l Y? ' | |||
44.6 R3 , | |||
4 fW | |||
* P3 1 P) | |||
WCTIR | |||
, ,. a a | |||
[1 vt Rt h<- Peg . .AD g - | |||
6-A9 PI (5 (6 to to 30 t i. | |||
g4 PuuP IAm jaw 36 se e3 , | |||
s _ | |||
u , | |||
WAf(R | |||
.r PUMP ' | |||
\ | |||
i Figure AL. Scliematic of One Ilalf Scale Pipe Segment A -8 t | |||
TM-1876A i | |||
Table A3. Pipe Segmends Scaling Verification Facilities ! | |||
Pipe Section PLANT RELAP 1/4 SCALE 1/2 SCALE Numbers ; | |||
length diameter I/d length diameter I,d Pips length diameter 1/d (inches) (inches) (inches) (inches) Number (inches) (inches) | |||
Riser 37 10 3.7 180 ~ | |||
9 2.5 3.6 P9 28 4 7 ! | |||
First Horizontal 162 10 16.2 190 41 2.5 16.4 P10 78 4 19.5 Run | |||
! Downcomer 132 10 13.2 200 33 2.5 13.2 PII 41.5 4 10.4 Second 18 10 1.8 210 3 2.5 1.2 P13A 78 4 19.5 Horizontal Run Continuation of 136 8 17.0 220 41.5 2.0 20.7 P13&l4 - - - | |||
Second Horizontal Run - | |||
[ | |||
i A-9 e -- - | |||
ENf6 TM-1876A APPENDIX B TEST PROCEDURE l | |||
l B-I | |||
b - TM-1876A Test Procedure - Ouarter Scale Baseline Tests Tests were run in the facility diagrammed in Figure Al. The primary test procedure was as follows: | |||
: 1. Initial Valve Settings a) VI Open b) . .V2 - Open c) . V3 Open | |||
. d) V4 Open c)_ V5 Closed f) V6 Open g) V7 Closed h) V8 Open i) V9 Open j) VIO Open k) Vll Closed 1)_ V12 Open m) BV1 Closed j n) BV2 Closed o) BV3 Closed | |||
: 2. Purge Air From System a) Turn on Pump - | |||
b) Cyc!c V6 to puige air from valve and close it - | |||
c) Cycle V1 until air is purged from valve and leave open. | |||
I d) Cycle V2 until air is purged from valve and leave open. , | |||
e) Deflect and wiggle P5 and cycle V2 until air is purged from P5. | |||
f) Cycle V3 untilit's purged of air, g) Tap on Pl1 until E4 is clear of air h) Generally P12 is clear of air, but if necessary bleed through BV2. | |||
i) Cycle V4 until purged of air j) Cycle V8 until it's purged of air and leave closed - | |||
k) Open V5 and cycle until cleared of air. Leave open | |||
: 1) Cycle V9 and V10 until they are cleared of air m) Cycle V12 until it too is cleared of air | |||
: 3. Adjust Air Measurement Scales a) Partially close V12 b) Open V7 until P25 is approximate! f 3/4 full of water. The top quarter will be air c) Close V7 d) Open VI1 until P29 is approximately 7/8 full of water. | |||
e) Close VII f) Open V8 g) Open V12 B-2 I | |||
., __) | |||
$ TM-1876A | |||
: 4. Set test flow rate, a) Adjust Vi to obtain a flow rate of aoproximately 40 gpm through Flow Meter 1 b) Adjust V8 to obtain a flow of 20.8 gpm through the Si section, Flow Meter 2. | |||
c) Adjust V1 to obtain a flow of 38.3 gpm through Flow Meter 1. | |||
d) Iterate between adjusting VI and V8 to reach flows of 38.3 and 20.8 gpm. | |||
e) Close V12 f) Open V6 until a flow of 17.4 gpm is obtained through Flow Meter i g) Close V2 h) Close V3 i) Turn off pump | |||
: 5. Record the air-water interface level in P25- | |||
: 6. Record the air-water interface level in P29 - | |||
: 7. Introduce starting air volume l- a) Open BV1 until the appropriate amount of air is drawn into Pl. Most tests were run with approximately 34.6 in' air, b) Close BV1 8.- Turn on pump | |||
: 9. At t=0 seconds open V2 | |||
: 10. At t=50 seconds open V3 l 11. At t=75 seconds close V6 and open V12 simultaneously. | |||
: 12. Observe and time as desired the air collection rate in P29. | |||
: 13. Once P25 and P29 have stopped receiving air, a) Close V3 b) Turn off pump c) Slowly close V12 d)_ Slowly close V8 | |||
: 14. . Record the air-water interface level in P29 | |||
: 15. Record the air-water interface level in P25 | |||
: 16. Test is complete. I B-3 o | |||
b TM-1876A : | |||
L Additional Tests . | |||
The previous procedure was used 16 measure the amount of air going to the charging and - | |||
SI pumps and to measure the void fraction going to the Si pump. We also ran a number of tests to determine the amount of air in the SI pump line and the charging pump line at a plant time of 150 seconds (75 sec. In the test facility) with plant flows of 560 and 900 gpm. The test procedure for these tests is similar to the above procedure with the following differences: | |||
2J Cycle V8 until it is purged of air, and leave open 2k. Delete- | |||
: 21. -Iklete. | |||
2m. Delete 3a -- Partially close V8 | |||
.3d Open V5 until P20 is approximately 3/4 full of water. | |||
3e.- Close V5 3 g.- Delete 4a - Adjust VI to obtai$ the desired flow.17.4 gpm for a plant flow of 560 gpm and 28 gpm for a plant flow of.900 gpm. | |||
4b. _ Turn off pump. | |||
4c-l' ~ Delete 6, - Mark the air-water interface level in P20. | |||
I1; At t=75 seconds close V4 - | |||
i | |||
: 12. Turn off pump. ; | |||
: 13. - Slowly close V8 | |||
: 14. Record the air-water interface level in P29. | |||
1 B-4 | |||
b TM.1876A Test Directory 12/12/97 A. Void clearing of 2.25 inch short test loop | |||
: 1. Increased flow stepwise until void cleared. | |||
: 2. Increased flow abruptly to 13.4 gpm, observed clearing | |||
. 3. Increased flow abruptly to 13.4 gpm, observed clearing H. Checked flow rate. | |||
12/15/97 A. Void clearing of 2.25 inch short test loop | |||
: 1. Took videos of voiding at 13.4 gpm. | |||
12/23/97 A. Void clearing of long 2.0 inch loop,1/4 scale (17.4/560 gpm). | |||
Initial void is 151/d's. | |||
: 1. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 -- | |||
: 2. Air volume measurements at t=75 (T=150 in plant)in P20 and P25 | |||
: 3. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 | |||
: 4. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 | |||
: 5. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 12/29/97 A. Void clearing of long 2.0 inch loop,1/4 scale (28.0/900 gpm). | |||
Initial void 151/d's | |||
: 1. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 | |||
: 2. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 | |||
: 3. Air volume measurements at t=75 (T=150 in plant)in P20 and P25 | |||
: 4. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 12/31/97 A. Void clearing of long 2.0 inch loop,1/4 scale (28.0/900 gpm). Initial void = | |||
151/d's | |||
: 1. Air volume measurements at t=75 (T=150 in plant) in P20 and Video B. Void clearing of long 2.0 inch loop,1/4 scale (17.4/560 gpm) Initial void = | |||
151/d's. | |||
: 1. Video 1/5/98 A. Clearing of 1/4 scale loop with SI sect. (17.4,20.8,38.3/560,670,1230 gpm) | |||
Initial void volume is 151/d's. | |||
: 1. Void fraction measurements to SI pump, P29 (Horizontal void trap) | |||
B-5 | |||
@ TM-1876A 1/6/98 A. Clearing of 1/4 scale loop with Si sectior. :.;:d horizontal void trap. (17.4,20.8, 38.3/560,670,1230 gpm) | |||
: 1. 5.461/d's initial void (34.6 in* =1.27 ft' in plant) 2.121/d's initial void (74 in') | |||
: 3. 5.461/d's initial void (34.6 in3 =1.27 ft' in plant) 1/7/98 A. Clearing of 1/4 scale loop with SI section and horizontal void trap. (17.4,20.8, 38.3/560,670,1230 gpm) | |||
: 1. 5.461/d's initial void (34.6 in3 =1.27 ft3 in plant) | |||
B.1/4 scale clearing with 28 gpm to chuging pump only (900 gpm in piant) 1-4. Let run for approximately 5 min. Measured air in P20 & P25. | |||
C.* 1/4 scale clearing with 28 gpm to charging pump only (900 gpm in plant) | |||
I'& 2. Let run for 75 s. Measured air in P20 & P25. | |||
1/8/98 A. Numerous Froude number tests with 2.5 and 4 inch test sections 1/9/98 ; | |||
' A. Measured volume of air at top of elbow E16 B. Complete tests with angled entry to pipe trap P29. Flow of 17.4 / 20.8 / 38.2 gpm in test facility (560 / 670 /1230 gpm in plant),5.461/d's. | |||
: 1. Air volume in P25 | |||
: 2. Air volume in P25 and P29 , | |||
: 3. & 4. Air volume in P25 and P29, and air flow : ate to P29 (SI pump) - | |||
5.& 6. Air flow rate to P29 (SI pump). | |||
1/12/98 A. Complete tests with angled entry to pipe trap P29. Flow of 17.4 / 20.8 / 38.2 , | |||
gpm in test facility (560 / 670 /1230 gpm in plant). Twice the initial void volume,121/d's. | |||
: 1. & 2. Volume in P25 and P29, and air flow rue to P29 (SI pump) | |||
: 3. Air flow rate to P29 (SI pump) | |||
B-6 | |||
b TM-1876A Haseline Data (Plant Sequence 560-1230 gpm) | |||
The initial bubble size in P2 was approximately 34.6 in'. | |||
Test # Air Volume Air Volume Air Flow Rate P25 in' P29 in' 3 P29 in /s 1/9/98 B 1. 2.55 (0.41 1/d) - -- -- | |||
B2. 2.75 (0.441/d) 25.1 (4.01/d) -- | |||
B3. 2.95 (0.471/d) 25.7 (4.1 1/d) 0.70 (over 27 sec) | |||
B4. 2.75 (0.441/d) 23.4 (3.71/d) 1.32 (over 7.13 sec) | |||
B5. 1.22 (over 7.76 sec) | |||
B6. 1.56 (over 4.02 sec) | |||
Tests were also mn with an initial bubble size of approximately 75 in3 , including initial | |||
. data on 1/6/98 and the data below. | |||
Test # Air Volume Air Volume Air Flow Rate 3 3 3 P25 in P29 in P29 in /s 1/12/98 A1. 9.42 (l.51/d) 69.9 (l1.11/d) 5.40 (10.02 sec) | |||
A2. 0.39 7 69.5 (11.11/d) 4.96 (10.76sec) | |||
A3. -- -- | |||
6.59 (4.77 sec) | |||
B-7 | |||
x 0 TM-1876A | |||
-In aildition to the tests reported above which involved different sampling intervals, a number of verifications were performed. Piping slope was adjusted at various points of the pipe network,'with slight differences in result, Sampling interval was_ varied, to assess the degree of peaking in the rate of air collection. (This effect is encompassed in _the results above). Two? | |||
different air collection geometries were employed. Initially, horizontal pipe P27 expended into a horizontal run of twice the diameter, which was then collected by a vertical pipe at top. This geometry tended to trap an air bubble at the expansion. Subsequently, the expansion pipe was . . | |||
sloped,- te_nding'to trap air at the high point of the" slope. The latter geometry was_ preferred: | |||
because little air in fact made it to the high point (it was removed into the vertical collector). | |||
= Moreover, the air at the high point was measured (post test) and used to provide a high bound on | |||
, - the total air collection. Downcomer pipe P30 was observed during tests, no air was carried under in that pipe. | |||
Lastly, air is held up for a time in the expansion pipe P28 to the collector pipe P29,- | |||
thereby altering the apparent rate of collection. Accordingly, we placed two simultaneous observers at pipes P27 and P29 respectively, each timing their observation of the air transported or collected. Collection in the vertical column was analyzed as above, an accurate measure but- , | |||
subject to air holdup in pipe P28. The observations of P27 are not influenced by holdup in pipe . | |||
- P27, but are more difficult to make, The observer et P27 found that the air flow rate was at first - | |||
relatively constant and highly visible, then abruptly tapered off. He measured the first time | |||
-interval (of appreciable air f,'ow), however attributed the entire collected air volume to that interval.- In this way, the result of the second observer is an upper bound in two respects; First it is insensitive to holdup in pipe P28. Second, it overestimates the amount of air delivered in the | |||
: measured interval. The' P27 approach was indeed higher than the P29 approach, but by only' 20E Accordingly, the P29 result has been reported and its uncertainty is indicated by this independent verification at P27. | |||
B-8 | |||
_ _ _a | |||
(OQfe TM 1876^ | |||
APPENDIX C SUPPLEMENTAL CALCULATIONS l | |||
C-1 | |||
NO TM-1876A Froude Number Froude Number and related calculations were performed to determine the appropriate Dows in scaled test facilities. To maintain a similar How regime in our scale facility, it must operate at the same Froude Number as the plant. The Froude Number is defined as: , | |||
F = v/(gd)" (1) where v is the velocity in the pipe, t si the acceleration of gravity, and d is the pipe diameter. To determine the flow in the scaled facility, one calculates the Froude Number in the plant for the - | |||
conditions of interest. This combined with g and d of the scaled facility allow one to determine the velocity of the flow rate in the scale model, and hence, the flow rate. For example, the initial flow in the plant to the charging pump is 560 gpm. From the flow rate and the cross sectional area of the 8 inch pipe, it can be readily calculated that the flow velocity in the 8 inch sections is 3.574 ft/s. Using this velocity and the above equation yields a Froude Number of 0.774. | |||
Plugging this back into the above equation for the corresponding 2 inch diameter pipe in the one-quarter scale facility gives a velocity of 1.79 ft/s and a flow rate of 17.4 gpm. Similar calculations were done for plant flows of 670,900, and 1230 gpm to obtain their scaled flow rates of 20.8,28.0 and 38.3 gpm in 2 inch piping. | |||
Air Volume Collected ' | |||
The air volume to reach the line running to the SI pumps and the line leading to the charging puinps was trapped and measured at a given point in time; typically 75 seconds after test initiates (equivalent to 150 seconds of plant time)._ These tests use the modified test procedure given at the end of the Test Procedure in Appendix B. The only calculation performed ; | |||
with this data was a straight forward air volume calculation based on the measured column height of air and the known column diameter. This volume can also be expressed as a fraction of the | |||
-initial air void. For example, one of our tests gave a column height difference of 1/16 and 3/16 inches in P25 and P20 respectively. This height combined with the cross sectional areas of the two columns (The diameters for P25 and P20 are 2.0 and 1.5 inches respectively.) yields air volumes of 0.2 and 0.3 in' in P25 and P20. A similar calculation was done for P29 for some of the tests. | |||
Void Frution During a test, we first determined rate of air flow into the SI and charging pump lines by measuring the rate of air entrapment in our two air traps P25 and P29. Dividing this by the water L flow rate yields the peak void fraction in the flow. Calculations were only perfo; -d for the SI pump in that the air flow to the charging pump was extremely low. In one test (w...ch had an initial air pocket twice that of the best estimate), the air level in P29 was found to drop 18 inches in 10 seconds. Eighteen inches of a two inch diameter column yields a total of 56.5 in3 or a rate 3 | |||
of 5.66 in /sec. The water flow rate through piping leading to the SI pump in the scale model is 3 | |||
20.8 gpm or 80.1 in /sec. Dividing the air flow rate by the water flow rate yields a void fraction of 7.1 % | |||
C-2 | |||
, _=. = | |||
s | |||
TM 1876A-Initial Air Pocket Volume To support our testing, an initial calculation was'done to determine the appropriate quantity of air for the start of the tests.: We received verbal instruction that the initial air volume in the plant is 1.27 ft' in an 8 inch pipe. This represents 2,195 in' at full scale and 34.3 in' at one-quarter scale. For simplicity, it is also a length of pipe that is 5.6 times its diameter, in the horizontal pipe P2, this was just slightly less than half full of air. The section of 2 inch diameter pipe is 23.5 inches long. Its ratio of length to diameter is 11.75. Italf of this is 5.8 We filled pipe P2 of the one-quarter scale facility _with air equivalent to just under one half of the pipe, matching the ratio 5.6 above, I | |||
i | |||
) | |||
C-3 | |||
h[@[Q TM-1876A APPENDIX D SCALING STUDY D-1 | |||
N EO TM-1876A On inspection we expected overhead pipe P10 in the one-quarter-scale facility to tend to trap air for a time in the fonn of a stratified How. Moreover, downeomer Pl1 tends to develop a bubbly flow, but has difficulty drawing the air from the run P10. The multiphase flow behavior of a run feeding a downcomer is not well documented in the literature, although we have seen it before (including some 1996 tests for NU for other purpose). | |||
We modeled this pipe segment at two scales, as described in Appendix A. This gave us an opportunity to confirm and illustrate the appropriateness of Froude scaling. And it also allowed us to examine the effect of Weber number on the rate and completeness of bubble formation. | |||
Figure D1 shows the degree of air entrapment as a function of Froude number with steady flow. The data are identical in the tested range, within experimental uncertainty of these quick single-sample tests. We would expect if anything that the rate and degree of air removal might increase slightly at the larger scale, a result of increased flow velocity (and Weber number) at fixed Froude number. This effect, if any, is indistinguishable in the present data. | |||
The actual situation is an unsteady flow. We tested with the overhead run (P10 at one-quarter scale) initially about half full and the water flow stagnant. Then water flow was stepped up and held steady. Air is removed in a spectrum of bubble sizes. Fine bubbles are quickly transported down the downcomer at velocities near that of the water. However, such fines constitute a small fraction of the total (initial) air volume. | |||
We were able to observe " neutral" bubbles that were dragged downward into the downcomer, then tended to stay relatively in place for long periods. At times such bubbles would move downward briefly, then they would rise again. Over time, statistically, such neutral bubbles would tend to progress downward and toward the bottom of the downcomer pipe. But then they would often agglomerate to a larger bubble and rise against the flow. As a result of these multiple and competing mechanisms, air tended to remain for long periods in the downcomer, and the process of air removal from the overhead run above was slow. | |||
Recall that in Figure Dl, very little more air was removed at half-scale than at one-quarter scale, even after waiting for a long period. Qualitatively, bubbles were slightly larger at the larger scale, perhaps by 10% of diameter. That is, bubble diameter was nearly the same in size when the pipe was nearly doubled in diameter. Similarly, the bubble entrainment process was slower at the larger scale, a result of Froude scaling. However, when time was properly normalized, the transients were similar within our degree to define them. | |||
Such details are of some scientific interest. For engineering purposes or the present NU evaluation, summary findings are: | |||
: 1. Froude scaling is appropriate, | |||
: 2. A long chain of small bubbles is observed at both one quarter and one-half scale, | |||
: 3. Projection to full scale tatroduces minimal uncertainty. | |||
D-2 i | |||
3 ] | |||
I re ru.1876A Froude Number Scaling 0.6 ....i....i.... .... | |||
....i.... | |||
e 2.5 inch pipe : | |||
E 4.0 inch pipe _ | |||
0.5 - | |||
e . . | |||
e . | |||
5 0.4 - - | |||
u - | |||
e . | |||
.9- e . | |||
@; 0.3 - - | |||
n . | |||
.9 - | |||
e c . | |||
u . | |||
5, 0.2 - | |||
m - | |||
. m . | |||
0.1 - E - | |||
~ | |||
. e . | |||
0.0 ''''''''''''''''''''''O'8'''' ' | |||
O.2 0.3 0.4 0.5 O.6 0.7 0.8 Froude Number Figure DI. Air Entrapment Declines As Flow Rate Increases D.3}} |
Latest revision as of 16:42, 31 December 2020
ML20203H353 | |
Person / Time | |
---|---|
Site: | Millstone ![]() |
Issue date: | 02/24/1998 |
From: | Nutt W, Rothe P CREARE, INC. |
To: | |
Shared Package | |
ML20203H345 | List: |
References | |
TM-1876A, NUDOCS 9803030203 | |
Download: ML20203H353 (36) | |
Text
Docket No. 50-423 B17049 Attachment 2 Millstone Nuclear Power Station, Unit No. 3 Transport of a Small Air Pocket Creare incorporated TM-1876A February 1998 February 1998
. . __ _.. =_ _
?M *?*deo M*dtaal:
G PDR
l 1
TitANSI'OllT OF A SMALL Allt POCKET TM-1876A Febiuary 1998
s NON Q.A.
TRANSPORT OF A SMALL AIR POCKET Prepared for:
Northeast Utilities Services Company Prepared by:
Ubm N 9V Paul H. Rothe /** Date Vice President IihNn P. hV
'\9illiam E. Nutt
"' 2/M/98
' Da'te ,
Engineer i
Creare incorporated TM 1876A PO Box 71 Project 8084 Ilanover,NH 03755 February 1998
O TM 1876A TAllLE OF CONTENTS
- 1. S U M M A R Y m........................... .. .... .... ... .............. ...........................................................!
- 2. It A CKG ROUND AN D INTRODUCTION .. ... . . . ..... .. ...... . .......... . .. ..... .. .m..... .1 2.1 I s s Ul! D u 1N IT10N .... .... .. .. .. . .. ........... ..... . . . . .. . . .m.. ...... . ...................................m...........................I 2.2 SiONII1CANCE OF GAS POCKl.T SIZE....... ...... .............................................................. ;.2 2.3 SlONII1CANCl! OP lilE FORM OF lilE O As ..... . .. .................. ......m .m..............m .
.................2
- 3. TEST D EF l N I TI ON ...... .. ... ... . ...... . . .....m ........... ............. ... ... m. . .... . . .. . ........ 3 3.1 OEOMETkY .... ...m..... ....... ..............m.... .. 3 3.2 Finw SEQUENCE .......... ................ .m. .... ............... =3 3.3 EX AMIN ATION OP TEST UNCERT AINTIES .... ...... ..... ...... ....m......... .................. ....... ............4 3.4 REPUCATION... ..... ..........................................................................................................4
. 3.5 SCAUNO.... ....... ...... ............. . =5 3.6 OTilER UNCERTAINTIES ; ........ ...................- =5
- 4. TEST FI N D I N G S . ..... . . . . . . .... .. ... ............ .... . .. . ..... .~....... ..... ....... .. 5
- 5. INTE R P R ETATl ON .......-.m. .... ~.m.. ....... ..... ..............~.. ...m ...... . . .7
- 6. R E F E R EN C ES... ... . . . .~........ . --.... .. .~.. ... .. ~ - .~ -.. . .... 9 A PPENDIX A FA Cl LITY .. . ... .. . .. ....-. . . . ........ .... ~ ... .. .. --. A . I A PPENDIX 11. TEST PROCEDURE..... . mm .... .. .~ . .- .....-.-~.. ....~.. ~ .~ .. ..11 1 APPENDIX C . SUPPLEMENTAL CALCULATIONS. . ... ... .. .. ........ . . ... . C .1 A PPENDIX D . SCALING STUDY.. . . .... . . . .. .~ .~ -~.. - ~ .~....D.1 i
. _ . . -= -. . - - - - . _ - - _. _. - -
TM 1876A .
l
- 1.
SUMMARY
Tests show that a small air pocket tends to distribute as it is transported through a tortuous array of piping. There are three key findings:
1, The initial air pocket is very small; only five pipe diameters long,
- 2. The form of the air is benign, tiny bubbles, and
- 3. The void fraction was approximately 1% at a pump simulator.
The facility is a transparent model of a portion of the A train of the long term cooling system at Millstone Point 3. The model piping extends frorn the RSS expansion seal upstream of MV8837A to the safety injection pump location following MV8807A. The testing was intended to illustrate the governing multiphase Dow phenomena with confidence and to quantify the rate and form of air transport to pump locations.
- 2. HACKGROUND AND INTRODUCTION 2.1 Issue Dennition During various loss-of coolant accident scenarios for Millstone Point 3 (MP3), a small air -
pocket is trapped in a theimal expansion loop. The air pocket is trapped inherently upon Olling by water of an ordinarily air-filled pipe line to containment spray headers. Later, that air pocket ,
is probably ingested by the safety injection or charging pumps, with potential for damage to the pumps.
4 Tests at Creare and analysis by NU establish the size of the air pocket as 1.27 cu ft (best estimate) in 8 inch diameter pipe. Such an air pocket is approximately 5 pipe diameters in length. It is also approximately 80% of the internal volume of the safety injection or charging pumps found by NU to have a volume of approximately 1.6 cu ft. Such an air pocket is a small fraction of piping that extends over 150 pipe diameters from MV8837 to the safety injection pump. Creare's Technical Memorandum TM-1869A is incorporated by attachment and addresses this portion of the work to define the size of the trapped air pocket.
The behavior of pumps when ingesting a water gas now was a subject of extensive research during the 1970s by Creare and others. The findings are described for example by Swift and Block (1982), Rothe and Runstadler (1978), and Patel and Runstadler (1978). A technology summary was prepared by Swift (1982) to assess the state of the art. The cited research encompr.ssed head degradation and resultant unstable operation of pumps. Related technology of cavitation and stall (loss of prime) of pumps is also well established.
A number of reportable events in nuclear power plants during the late 1980s and through the 1990s involved the discovery of unvented high points in the piping to charging pumps.
Information notice IN88-23 of the U.S. NRC summarizes the event at Farley which found a trapped hydrogen pocket of approximately 60 cu ft volume. Subsequent analysis by Creare of this trapped gas pocket was reported at a public meeting attended by Westinghouse and provided by Westinghouse to NU. Swift and Rothe (1990) found analytically:
)
. _ __ _ - _ . _ . _ _ . _ _ _ _ _ . _ _ . - _ - . ~ __. -. . _
b TM 1876A
- 1. A trapped hydrogen pocket of 60 cu ft in a 6 inch diameter line,
- 2. Subjected to a 6700 gpm flow transient from the RilR pumps,
- 3. Would travel in six seconds through a short run of pipe,
- 4. Arriving at the charging pump as a long and intact column of gas,
- 5. Woyld fully stall the pump, driving residual water to the last few stages,
- 6. Would subject the pump to a massive clearing transient and unbalanced loads,
- 7. With calculated excessive shaft deflection and other mechanical events, and
- 8. leading probably to prompt seizure of the pump.
U.S. NRC concern for such an event is understandable, and indeed, there have been four supplements to IN88 23 as well as unrelated reports of findings of large quantliles of unvented gas and reports of various pump behaviors.
The present report addresses the likely insignificance of a much smaller air pocket, transported to a pump at rnuch lower velocity, in a piping geometry unique to Millstone Point 3.
We find that the air pocket distributes to a long chain of fine bubbles. Our opinion is that this specific flow is benign.
2.2 Signincance of Gas Pocket Size On inspection, a dominant difference between the subject scenario at MP3 and the reported Farley event is the size of the trapped gas pocket, approximately 60 cu ft at Farley and approximately I to 2 cu ft at MP3. Moreover, the larger diameter piping at MP3 has nearly twice the cross-sectional area as that at Farley, the piping is much longer at MP3, and the pumps have twice the volume.
Gas pocket size may be considered first in relation to the piping through which it must be transported. A !arge column of gas near the pump, such as that at Farley, has no possibilhy of appreciable collection in the various air traps presented by the piping. In contrast, a small air pocket such as that found at MP3, will have difficulty proceeding toward the pump through a circuitous pipe network. 'ine piping as built is a nest of dead legs and components all prone to trap air. Pipes slightly sloped from the horizontal will on average tend to trap gas.
Gas pocket size may also be considered separately in relation to the pump. A massive gas column, such as that found at Farley, will surely evacuate, stall, and unbalance the pump. A small air pocket such as at MP3 (even if it were transported compactly in its entirety to the pump location) tends to be a brief and one time transient for the pump.
2.3 Significance of the Form of the Gas Gas in water can take many forms, known as flow regimes. The hydrogen gas at Farley began as a long column filling approximately 300 feet of piping. The operating scenario subjected this long column to the full, transient head of one pump (the RHR) while restricted by another (the charging pump). The initial, transient advance of the water into the gas column was 6700 gpm (approaching 80 ft/sec in water) which effectively pushed and compressed the gas column. The gas moved intact through the piping over the short travel distance to the pump.
2
b TM 1876A At MP3 the water velocities are much lower, of the order 10 ft/sec or less, initially the water is stagnant, allowing the air to drift and spread. Then the water proceeds at velocities of approxhrately four feet per second, which is well below the condition to clear the gas as a column. M : ' sie gas simply rises as bubbles or layers to fill the tops of pipes as it is dragged
, along slowly. by a now of water underneath it, importantly, the piping at Mp3 includes downcomers, where layers of gas are prone to break up into small bubbles as they are carried downward contrary to the force of gravity.
[ These effects related to flow all tend to spread out the gas and promote a bubbly flow.
They are enhanced by the many traps and obstructions in the piping as the ps must navigate past dead legs, through valves (including part open check valves) and the like. Thus, not only does j realistic piping tend to trap some of the air. It also tends to dilute it with water and render it in a relatively benign, bubbly form. Contrast this form with the description of the Farley situation, which was calculated to involve a long column of gas and slugs of water repeatedly passing through the pump.
- 3. TEST DEFINITION 3.1 Geometry The subject piping at MP3 extends through some 30 independent pipe segments, with numerous valves and tees, and with suitable modeling of dead legs where there is no Dow. The facility diagram, component description, and basis in NU drawings are in Appendix A. The facility is one-quarter scale. Facility piping is primarily 2 inch diameter; comparable piping at MP3 is primarily 8 inch diameter. NU identified Train A as having greater safety significance than Train B. All tests at Creare modeled the Train A geometry.
3.2 Flow Sequence The safety analysis branch at NU has evaluated numerous potential LOCA scenarios on a screening basis using the RELAP computer code and also based on engineering judgment, as reported by NU. An exemplar scenario was specified to Creare that involved three key steps:
- 1. Open MV8837, no Dow, air pocket migration by gravity for 100 seconds,
- 2. Open simulators for MV8804/CV36, begin Gow at 560 gpm plant equivalent for 50 seconds, and
- 3. Open simulators for MV8807, step-up flow to 1230 gpm plant equivalent, until the gas trantportation process ends.
This scenario was identified by the Safety Analysis Branch of NU as the most safety significant now sequence.
We examined the flow sequence in two ways. First, only steps 1 and 2 above were performed, and air was collected at the inlet to valve VS (simulator of MV8807) and in the bypass line P25 leading in MP3 to the charging pumps. That is, we sensed the arrival of air (if any) at tee T5 where the flow splits between the Si and charging pump inlet pining, and we 3
1 TM 1876A measured the quantity of air collected at the location during the period of time significant to the sequence, in this way, we could determine definitively if there were any adverse buildup of air at valve V5 (MV8807) before it opened. (There was not.) We could also confirm the NU analytical finding that air preferentially passes to valve VS (MV8807) and little carries under to the charging pumps. (We concur.) lly stopping the test at this point in the flow sequence, we also took the opportunity to measure air standing at all locations and to perform an integral check on air volume throughout the facility.
I Secondly, we performed the entire three step flow sequence above. The procedure is detailed in Appendix II. This involved additional air collection stations at other locations, and more stringent verification of their experimental uncertainty, again as described in Appendix II.
That is, the pump simulators in the facility are air collection stations. They are made of segments that expand in diameter and slope upward (to prevent air entrapment) then proceed to a small
- diameter riser for the purpose of observing and timing the rate of air collection.
NU identified a second now sequence of some (but considerably lesser safety significance), it involves two steps:
- 1. Open valve MV8837, no How, air migration by gravity for 100 seconds, and
- 2. Open valves MVE804/CV36, flow at 900 gpm. l That is, by comparison with the primary flow sequence above, the initial flow (900 gpm) is higher (than 560 gpm), but it does not step up to 1230 gpm This test was performed only in the first form above. That is, valve V5 was not opened, air collected at its inlet was measured (very little air was collected). Continued flow at 900 gpm was felt to be less challenging than the step-up to 1230 ppm that had already been tested.
3.3 Examination of Test Uncertainties All tests or analyses have modeling uncertainty. Therefore, an important part of the test definition involved devising means to i.'luminate and quantify these uncertainties. The identified uncertainties were of four kinds:
- 1. Replication and sensitivity, 2, Scaling,
- 3. tvmponent modeling, and
- 4. Instrumentation. i
- NU derived additional confidence from the collaboration of their RELAP calculations with Creare's tests liere we consider the uncertainty of the tests as if they stood alone.
3.4 Replication The tests as performed were intended primarily to illustrate the governing physics of some complicated and unsteady phenomena. That is, these were not meant to be precise
-- production experiments, as would be carried out for example to obtain the flow characteristic of a 4
O TM 1876A valve. Still, we were determined to assess the sensitivity of the results to acasoned variations of parameters, in part, we did so simply by repeating the key tests a number of times and with varied observers, in part, we adjusted the slope of cedain pipes, predictably with measurable effect. Lastly, we altered the size of the initial air pocket introduced into the facility, in this way, we examined the sensitivity of this test to that input assumption, and we segregated the phenomenon of air trapping from the phenomenon of air transpodation. These uncertainty studies are described in Appendix 13 and are the sole cause of the data scatter reported in Table i below.
3.5 Scaling cude scaling is a well established principle for examination of the stratified multiphase How :egime and its transition to other regimes. At the one quarter scale tested, flow velocity is reduced by a factor of two when Froude number is preserved. Reynolds and Weber numbers are distorted proportionally.
Although these are minor distortions by comparison with the degree of scaling employed routinely in the past by the research community concerned with nuclear reactor safety, it was straightforward to perform a scaling study that provided added confidence in the results. One of -
the pipe segments was identified both on inspection and -in subsequent tests as phenomenologically significant. It involved flow in a riser to a mn to a downcomer, then a nm, and served as a location both to hold air and to disperse it as a chain of small bubbles. We tested this segment of the geometry at two scales; one-quarter scale and one half scale.
As might be expected, the scaling study confinned onu again the suitability of Froude number to scale a stratified now. Further, it confirmed that the bubble production behavior was only slightly modified by scale, and in a manner consistent with expectation. Thin aspect of the work has some original scientific interest to illustrate the behavior of interacting pipe segments.
From an engineering point of view, and for NU's safety determination, it merely confirms the low uncertainty introduced by scale modeling. The scaling study is reported in Appendix D.
3.6 Other Uncertainties The details of the component modeling are described in Appendix A in the context of the geometry detail presented there. The details of the instruments and their uncertainties are described in Appendix B in the context of the detailed test procedures described there. Various-support calculations are described in Appendix C.
- 4. TEST FINDINGS The test findings are the subject of a brief illustrative video tape. This video has been supplied to NU and was played at the enforcement hearing held at Millstone Station on January 13,1998.
The initial small air pocket breaks up to a chain of tiny bubbles, and arrives at the safety injection simulator over_ a period of approximately a minute. Littic or none of the air is 5
b TM 1876A l transported past the tee and toward the charging pump, a result of centrifugal force at a pipe tee flowing bubbly air water along its top. A significant fraction of the initial air pocket is trapped at various locations throughout the long piping network, The time scale of the scaled facility is twice real time. That is, a minute of facility time is equivalent to two minutes of plant time, in two minutes, a flow of 1230 gpm passes 2460 gallons of water. The initial trapped gas pocket is 1.27 cubic feet or 9.5 gallons. Thus, the air is diluted by water, on average, by a ratio of nearly 300.
At the location of the safety injection pump, air arrives in a manner that is distributed .
over a minute in time but is initially at a rate somewhat higher than average. Table I below summarizes our findings in comparison with the RELAP calculations. Various appendices elaborate.
Table 1. Millstone Unit 3 Pump Inlet Vold Fraction Results Scenario' MV8837 Flow Computer Laboratory Laboratory 3
Flows Duration RELAP Results Results' (gpm) (sec) Results IX Air. 2X Air (NU) (Creare) (Creare)
- 1. RSS A to 560 150 0% 0% 0%
CliG A 3
- 2. RSS A to 560-1230 until air 2-4% 0.8 to 1.7% 3 to 7%
Sill /CIIO A fully transported
- 3. RSS A to 900 150 <l% <l% Not Run CilO A&B Notes:
8 See NU Analysis Report M3.EV.98-0018 Rev 0 for description of scenarios.
8 RELAP calculates a detailed transient with an average of 2 to 4 void fraction for 10 to 15 sec. 'Ihe range is that of sensitivity calculations. Instantaneous peaks of 5% and 9% respectively are calculated.
8 Laboratory results inherently average air collection over a period of time. However, tests show that the collection rate is relatively constant during the significant period (unlike the RELAP calculation of a peak). Accordingly we compare the averrges by both approaches.
- 2X air is simply twice as much air as the best estimate volume of the initial air pocket. Though not prototypical, the test shows the sensitivity to various uncertainties.
6
b TM 1876A
- 5. INTERPRETATION I The mechanical response of a pump is amenable to analysis, test, or to manufacturer representation based on knowledge of its design and experience with like equipment. Pump manufacturers design their equipment to tolerate imbalance ordinarily expected based on manufacturing tolerances or adverse loading. Considerable experience is also gained from long operation of many pumps and from specialized tests. In this respect, the mechanical response of
'; a pump, its susceptibility to mechanical vibration, to excess shaft deflection, bearing wear. life degradation, or even seizure, lie within the ordinary design envelope of ti e equipment. Still, it remains that centrifugal pumps generally and the specific charging and safety injecting pumps at Mp3 are not produced nor qualified for the specific loads due to gas ingestion. Accordingly, NU -
must rely primarily on representation by the pump manufacturer The paragraphs below supplement such representation by logic and by review of the known draa.
Pump head degradation is a process of accumulation of air in a pump, over time, ;
sufficient to replace the water within the pump and prevent the achievement of full centrifugal head and the flow distributloa on which pump effectiveness miles. Such accumulation of air results from a balance between body forces tending to trap air in the pump (centrifugal and l coriolis forces) and the drag of water on the air bubbles and '- yers. Both the fraction of arriving
- air and the duration of air ingestion affect pump performance.
There are no known tests of pump performance when subjected to an air ingestion transient of such brief duration (and air volume) as is of present interest. We suspect that when the entire air pocket is smaller than the intemal volume of the pump, as is the case here, then the odds and severity of appreciable performance degradation or mechanical damage are significantly mitigated. If further, the air is highly distributed to individual bubbles, as also is the case here,
- then the event is further mitigated substantially. These are qualitative observations. It remains that the subject pumps have not been tested to quantify their mechanical tolerance to brief air ingestion at any volume or fraction. No such tests have found damage, nor have they assured continued performance without pump life degradation. For perspective, pumps are routinely designed for much more severe duty of this kind. (For example, certain fuel pumps accept
, sudden intake of liquid in a previously gas space at high liquid flow rates.) Thus, the question is not whether centrifugal pumps can be made in general to withstand such duty, nor whether the subject pumps are mgged. They can and are. Rather, the tolerance of the particular subject pumps has not been tested for this narrow and specific duty. Fortunately, the present event appears to be so mild as to obviate the need for such completeness of evaluation.
4 Extensive research [ Swift,1982] demonstrates no appreciable (reliably measured) head L degradation for inlet void fractions below 2% even when air has been supplied to the pump for an extended period. This is a very stringent criterion involving the ability to detect any effect whatsoever, in research experiments, with air supplied at a given rate forever.
The onset of barely measurable pump performance degradation in research experiments is itself a relatively mild situation. Indeed, researchers can begin to detect the accumulation of air in the pump and corresponding reduction of head when as little as 2 to 5% void fraction is 7
Nb TM 1876A supt. lied at pump inlet. The pump does not pump as well, though it continues to pump even when supplied a void fraction of up to 15% air for an extended period, in the present situation with a highly distributed flow of a small air pocket, the possibility of pump stall is remote.
Creare finds air arriving at its pump simulator inlet in the range 0.8 to 1.7 % by volume for a period of about 15 seconds. Thereafter air continues to arrive at substantially lower void fraction for approximately one minute. This is a very low total volume of air amounting to a fraction of the initial air pocket, which in turn is smaller than the internal volume of the pump.
Moreover, the air arrives at pump inlet well diluted by water and at fractions shown by research to be unmeasurable in terms of degradation of pump head. Reason would suggest that the unbalance loads produced thereby will have minimal effect on pump mechanical performance, as V/cstinghouse states is the opinion of the manufacturer.
i The onset of now vibrations and surge results from an interaction of the pump's performance with the flow system, particularly its flow inertia, resistance and compliance [Rothe, 1978). Absent a specific system analysis or test,little can be said of the susceptibility of a given system to surge, llowever, in an event of brief duration, there is little prospect of surge, lacking time to develop flow oscillations. Thus, either the void fraction at pump inlet is too low for there l to be any appreciable head degradation and surge, or the duration of head degradation is too brief I for the system to respond and surge. Either way, surge is unlikely, i
in summary, the small initial air pocket at MV8837 is expected to be transported to and ingested by the safety injection pump. The considerations specific to the subject scenario are:
- 1. An initial air pocket of 1.27 cu ft volume,
- 2. A large pump, larger than the air pocket,
- 3. A tortuous run of piping that is 50 times this volume and contains numerous air traps,
- 4. An overhead horizontal run to a downcomer where air is distributed gradually as ; mall bubbles, and
- 5. A flow rate sequence particular to Mp3, specifically a; Stagnation for 100 seconds, b) 560 gpm Cow for 50 seconds, and c) 1230 gpm Dow thereafter.
For the scenario involving all of these specifics, we believe that the outcome would have been benign. Air arriving at the charging pump would be of brief dur .on (of the order of a minute), low void fraction (of the order 1%), and low total volume. Based on reason and the literature of the field, we expect none of stall, surge (now oscillations), appreciable imbalance loads, pump seizure, nor pump life degradation. In the absence of specinc mechanical test data, we rely on the pump manufacturer's assurance of the suggedness of the unit and its design with a factor of three margin for imbalances of various kinds.
8 i
b TM 1876A
- 6. REFERENCES Swift, W.L; Block, J.A.; Model Pumo Performance Program Data Report; EPRI NP-2379, Research Project 3471, Interim P.cport, Electrical Power Research Institute, Palo Aho, CA, May 1982.
Swift, W.L; Kamath, P.S.; Runstadler, P.W. Jr.; Block, J.A.; Two Phase Performance of Scale Models of a Primary Coolant Pump; EPRI NP 2578, Research Project 3471, Final-Report, Electrical Power Research Institute, Palo Alto, CA, Sept.1982.
Rothe, P.ll.; Runstadler, P.W., Jr.; Dolan, F.X.; Pump Surge Due to 7ko Phase Flow; Polvohase Flow in Turbomachinerv, C. Brennen, P. Cooper and P.W. Runstadicr Jr., eds.,
Publication 11000123 of the American Society of Mechanical Engineers,1978,121 137.
Putel, B.R.; Runstadler, P.W. Jr.; investigations into the Two Phase Flow flehavior of Centr (fugal Pumps; Pohphase Flow itLTurbonachinery; C. Brennen, P. Cmper and P.W.
Runstadler, Jr., eds., Publication H000123 of the American Society of Mechanical Engineers, 1978,79 100.
Swift, W.L; Kamath, P.S.; Tantillo, T.J.; An Assessment of Residual lleat Removal and Contalmnent Sprav Pumo Performance Under Air and Debris Innestinn Conditions.
NUREG/ CR 2792, Creare TM 825, United States Nuclear Regulatory Commission, Sept,1982.
Swiit, W.L; Rothe, P.iI.; Evaluation of Operability of Charging Pumps; Creare MTG-90 7-302 presented to the Nuclear Regulatory Commission, Rockville MD, July 14,1990, 9
bb TM 1876A i
Al'I'ENDIX A FACILITY j A-1
)
b TM 1876A Baseline Facillty at One Ouarter Scale !
The schematic of the one-quarter scale facility is given here as Figure A1. The diagram serves to number and orient the piping, which includes:
^ '
4 e llorizontal runs,
- Downcomers (vertical pipes where the flow proceeds downward), and e Risers (vertical pipes where the flow proceeds upward).
Table Al lists the pipes that match those at MP3 and in the NU/RELAP model of MP3, and compares the as-built dimensions with the plant dimensions. The plant was closely matched at one-quarter scale.
Dimensions of the test facility are from as built measurements and can also be found on ,
Figure Al, The dimensions of the Plant were taken from a variety of drawings, sketches, and conversations with Northest Utilities Company. The drawings were from the Stress Reconciliation Piping Location isometric drawings 25212-20394 Sil.8,9, and 26; 25212-20404 Sil10; and 12179-C.I. RSS12, Sl! 3. Sketches were titles RSS IlX A Piping. Additional information came fiom Figures 4,5, and 6 of some drawings of the RELAP model titled MP3 RECIRC. PHASE AIR BUBBLE STUDY TRAIN 'B' FAILURE.
Several additional pipes are on Figure A1, for facility purposes such as air collection. All
, dimensions are shown. As described in the test procedures, the facility was refined in minor ways (for example addition I air collection stations were used) fev. some of the supplemental tests.
Three pipes are dead legs either closed off or to a downcomer. Of these one tends to
^
collect air, while two do not. There are slight differences between the facility and the plant (the facility lengths of dead leg are shorter in the two cases that don't collect air anyway). We believe that these differences are insignificant and also conservative.
A-2
2.g,,,a,_-.,._.,d,,,44ML'8"'M-'
@eare - pp Tu is76A E
a -
3k
& Wi u ., ,
e S
@'i[.4 e%
e li.
l 4,,' JD s
O' Il* ,
hn4 t u3 i ! -e N 1 j- ---l g
~
s e' '
s
! rI.., .,.'M*-e D.. -
, N. T ' 'x a
l -
N e 4 e
/ 9 6'K# "E
,l e,
y A
e
,g%n f[31ibl,E3dy,e .
44HH M4 4 4444k- --- '
g
,1
TM 1876A TAllLE A1. ONE QUARTER SCALE PIPING Test Facility Dia. Length RELAP Plant Plant Comments Piping # In. Ud Number Dia. 1/d PI 2.0 7.3 120 8 6.4 P2 2.0 11.7 130 8 11.6 P3 2.0 4.0 140 8 3.7 P4 2.0 6.5 150 8 6.0 P5 2.0 13.7 157 8 13.5 9 feet to downcomer.
P6 2.0 10.3 160 8 9.6 P7 2.0 26.3 165 8 26.5 P7A 2.5 2.4 170 10 2.0 P8 2.5 3.0 NA 10 Unknown, doesn't collect air.
19 2.5 3.6 , 180 10 3.7 P10 2.5 16.4 190 10 16.2 Pt1 2.5 13.2 200 10 13.2 P12 2.5 16 211,213, 10 37.8 Approx. 31.5 ft to down.
215,217 comer. Doesn't collect air.
P13A 2.5 1.2 210 10 1.8 P13 & 14 2.0 20.7 220 8 17.0 P15 2.0 6.5 230 8 6.5 P16 2.0 4.0 240 8 3.8 P17 2.0 6.3 250 8 5.9 P18 1.5 4.7 260 6 4.4 P19 1.5 8.3 270 6 20.0 Conservatively short.
P20 1.5 16,0 300 6 15.6 P21 1.5 9.3 330 6 9.4 P26 1.5 9.3 320 6 9.4 P27 1.5 14.7 340 & 350 6 13.1 A-4
S TM 1876A Table A2 provides a chan of facility components. Elbows and tecs are similar to those in the plant. Valve detail data were not available to Creare during construction and are partially available at this writing. Swing check valv' 'he plant are expected to be partially open (about half open) at the subject flows. This degre spening will present a " smile" opening at the top of the pipe (where the air is), thereby tendinh .o trap and distribute air. Create used a partial port (80% open diameter) ball valve at check valve location CV36, a geometry that tends to trap some air while passing the rest in an unhindered manner. Creare did not model another check valve in the plant at all. Motor operated plant valves are generally gate valves, with a stroke over some time period of at least 10 seconds, and an opening that may be full or partial (unknown to us at present). Gradual opening will tend to ramp How, though not to appreciable degree at these now rates and port sizes. Creare used a conservative abrupt opening. Partially open gate valves will tend to trap a!r, as did Creare's panial port ball valve. On balance we view these differences as minor, given two overriding geometric attributes:
- Entrapment and distribution of air due to gravity in horizontal pipe runs, and e Breakup to fine bubbles in downcomers That is, the effects of gravity appear to override component details.
A-5
O TM-l876A Table A2, One Ouarter Scale Facility Comisonents Piping clear acr3 11c (0.25 inch walls).
1.1.5 inch ID PI8 P21, P26, P27
- 2. 2.0 inch ID. Pl P7,, P7A, Pl3A, Pl3.Pl7, P22, P25, P29 3, 2.5 inch ID P8 Pl2 1
Valves I, 2.0 inch partial port ball valves. VI V4, V6, V8, V12 2.1.5 inch gate valves. V5, V9, VIO.
Valves 2,3,4,5,9 and 10 represent valves in the plant. All of the other valves in the test facility are control valves necessary to run the experiments, in the plant, valve MV8837A is a motorized gate valve which opens over a period of 10 seconds; wiercas in the test facility, it is a manually operated partial port ball valve. The details of the geometry and the degree of opening are unknown to use. We believe the use of the ball valve is either accurate or conservative because it does not trap or hold up appreciable gas.
Valves 4,9 and 10 in the test facility and the plant are all gate valves. All though we have no detailed geometry of the plant valves, we believe that the choice is conservative, because we have observed no signincant gas trapped in or tefore the test valves.
Elbows . Standard PVC nttings 1.1.5 inch,90 degree. E8-Ell 2, 2.0 inch,90 degree. El, B2, ES, E12 El3, E14
- 3. 2.5 inch,90 degree. E3, E4
- 4. 2.0 inch,45 degree. E6, E7
- 5. 2.0 inch 22.5 degtce. A1. A2 Elbows E2. Ell and E23 model cibows in the plant. llorizontal cibows were not modeled. All of the other cibows in the facility serve te support the plumbing necessary to run the experiments. The elbows were chosen primarily based on what was immediately available and generally have a radius of curvature of I 5 times the Jiameter.
Teen . Standard PVC fittings
- 1. 2.0 inch. T2
- 2. 2.5 inch. T3, T4
- 3. 2.0 x 1.5 x 2.0 ineh. T5
- 4. l.5 inch T7,T8 1 acking detailed information on the plant elbows wo used standard off-the-shelf right angle Tee's for the facility. Tee's 2 5,7 and 8 represent Tee's in the plant.
110w Metert
- 1. Flow meter #1. King Instrument Co. Rotometer,7205-0201 W 5-40 gpm
- 2. Flow meter # 2. Omega Eng. Inc. Rotometer, PL-75F,4-40 gpm
%e stated accuracy of the Omega meter is 2.5% full scale. The King instrument unit should be similar.
We have noted no discrepancies between the two meters. The King instrument was also crudely checked by performing an actual measurement of its flow rate. The now was set at 10 gpm and fed into a known volume. The fill time was measured and the now rate calculated. %ey were found to match.
A-6
i S TM 1876A !
l t
Half Scale Facility B
A segment of the pipe network (four pipes) was modeled at hath one quarter and one half !
scale. Figure A2 shows the half scale pipe segment, which consists (in flow direction) of: j e Riser P4 -
- Run PS
- Downcomer P6 i e Run P7.
l There are corresponding pipes in the one quarter scale facility. i t
Table A3 compares the one-quaner scale and one-half scale facilities with MP3 plant pipe i dimensions. ,
I F
k t
e t
I
-t M
4 A-7
- ,r y co--- , . . . . , , , , , , ,+g_,, , , , ,, , mm,, , , _ _ , . ..m,.-_.#.___. ,p ,_ -, ncm_..,. f.-e__y. . ,yr, , y , . , , , ,_.e-
. . .,-m..-.~.~.__.. _.~~ . - _ _ ____. . - . - - _ , - . _ . . . ..._--._.__m_..m__.m.._ m- _ _ . _ _ _ _ _ _ . . _ . . . . _ . -
t TM-l876A i,
t 4 L i
J l- .
PS g- ,
y
@~ s78 u
l Y? '
44.6 R3 ,
4 fW
- P3 1 P)
WCTIR
, ,. a a
[1 vt Rt h<- Peg . .AD g -
6-A9 PI (5 (6 to to 30 t i.
g4 PuuP IAm jaw 36 se e3 ,
s _
u ,
WAf(R
.r PUMP '
\
i Figure AL. Scliematic of One Ilalf Scale Pipe Segment A -8 t
TM-1876A i
Table A3. Pipe Segmends Scaling Verification Facilities !
Pipe Section PLANT RELAP 1/4 SCALE 1/2 SCALE Numbers ;
length diameter I/d length diameter I,d Pips length diameter 1/d (inches) (inches) (inches) (inches) Number (inches) (inches)
Riser 37 10 3.7 180 ~
9 2.5 3.6 P9 28 4 7 !
First Horizontal 162 10 16.2 190 41 2.5 16.4 P10 78 4 19.5 Run
! Downcomer 132 10 13.2 200 33 2.5 13.2 PII 41.5 4 10.4 Second 18 10 1.8 210 3 2.5 1.2 P13A 78 4 19.5 Horizontal Run Continuation of 136 8 17.0 220 41.5 2.0 20.7 P13&l4 - - -
Second Horizontal Run -
[
i A-9 e -- -
ENf6 TM-1876A APPENDIX B TEST PROCEDURE l
l B-I
b - TM-1876A Test Procedure - Ouarter Scale Baseline Tests Tests were run in the facility diagrammed in Figure Al. The primary test procedure was as follows:
- 1. Initial Valve Settings a) VI Open b) . .V2 - Open c) . V3 Open
. d) V4 Open c)_ V5 Closed f) V6 Open g) V7 Closed h) V8 Open i) V9 Open j) VIO Open k) Vll Closed 1)_ V12 Open m) BV1 Closed j n) BV2 Closed o) BV3 Closed
- 2. Purge Air From System a) Turn on Pump -
b) Cyc!c V6 to puige air from valve and close it -
c) Cycle V1 until air is purged from valve and leave open.
I d) Cycle V2 until air is purged from valve and leave open. ,
e) Deflect and wiggle P5 and cycle V2 until air is purged from P5.
f) Cycle V3 untilit's purged of air, g) Tap on Pl1 until E4 is clear of air h) Generally P12 is clear of air, but if necessary bleed through BV2.
i) Cycle V4 until purged of air j) Cycle V8 until it's purged of air and leave closed -
k) Open V5 and cycle until cleared of air. Leave open
- 1) Cycle V9 and V10 until they are cleared of air m) Cycle V12 until it too is cleared of air
- 3. Adjust Air Measurement Scales a) Partially close V12 b) Open V7 until P25 is approximate! f 3/4 full of water. The top quarter will be air c) Close V7 d) Open VI1 until P29 is approximately 7/8 full of water.
e) Close VII f) Open V8 g) Open V12 B-2 I
., __)
$ TM-1876A
- 4. Set test flow rate, a) Adjust Vi to obtain a flow rate of aoproximately 40 gpm through Flow Meter 1 b) Adjust V8 to obtain a flow of 20.8 gpm through the Si section, Flow Meter 2.
c) Adjust V1 to obtain a flow of 38.3 gpm through Flow Meter 1.
d) Iterate between adjusting VI and V8 to reach flows of 38.3 and 20.8 gpm.
e) Close V12 f) Open V6 until a flow of 17.4 gpm is obtained through Flow Meter i g) Close V2 h) Close V3 i) Turn off pump
- 5. Record the air-water interface level in P25-
- 6. Record the air-water interface level in P29 -
- 7. Introduce starting air volume l- a) Open BV1 until the appropriate amount of air is drawn into Pl. Most tests were run with approximately 34.6 in' air, b) Close BV1 8.- Turn on pump
- 9. At t=0 seconds open V2
- 10. At t=50 seconds open V3 l 11. At t=75 seconds close V6 and open V12 simultaneously.
- 12. Observe and time as desired the air collection rate in P29.
- 13. Once P25 and P29 have stopped receiving air, a) Close V3 b) Turn off pump c) Slowly close V12 d)_ Slowly close V8
- 14. . Record the air-water interface level in P29
- 15. Record the air-water interface level in P25
- 16. Test is complete. I B-3 o
b TM-1876A :
L Additional Tests .
The previous procedure was used 16 measure the amount of air going to the charging and -
SI pumps and to measure the void fraction going to the Si pump. We also ran a number of tests to determine the amount of air in the SI pump line and the charging pump line at a plant time of 150 seconds (75 sec. In the test facility) with plant flows of 560 and 900 gpm. The test procedure for these tests is similar to the above procedure with the following differences:
2J Cycle V8 until it is purged of air, and leave open 2k. Delete-
- 21. -Iklete.
2m. Delete 3a -- Partially close V8
.3d Open V5 until P20 is approximately 3/4 full of water.
3e.- Close V5 3 g.- Delete 4a - Adjust VI to obtai$ the desired flow.17.4 gpm for a plant flow of 560 gpm and 28 gpm for a plant flow of.900 gpm.
4b. _ Turn off pump.
4c-l' ~ Delete 6, - Mark the air-water interface level in P20.
I1; At t=75 seconds close V4 -
i
- 12. Turn off pump. ;
- 13. - Slowly close V8
- 14. Record the air-water interface level in P29.
1 B-4
b TM.1876A Test Directory 12/12/97 A. Void clearing of 2.25 inch short test loop
- 1. Increased flow stepwise until void cleared.
- 2. Increased flow abruptly to 13.4 gpm, observed clearing
. 3. Increased flow abruptly to 13.4 gpm, observed clearing H. Checked flow rate.
12/15/97 A. Void clearing of 2.25 inch short test loop
- 1. Took videos of voiding at 13.4 gpm.
12/23/97 A. Void clearing of long 2.0 inch loop,1/4 scale (17.4/560 gpm).
Initial void is 151/d's.
- 1. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 --
- 2. Air volume measurements at t=75 (T=150 in plant)in P20 and P25
- 3. Air volume measurements at t=75 (T=150 in plant) in P20 and P25
- 4. Air volume measurements at t=75 (T=150 in plant) in P20 and P25
- 5. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 12/29/97 A. Void clearing of long 2.0 inch loop,1/4 scale (28.0/900 gpm).
Initial void 151/d's
- 1. Air volume measurements at t=75 (T=150 in plant) in P20 and P25
- 2. Air volume measurements at t=75 (T=150 in plant) in P20 and P25
- 3. Air volume measurements at t=75 (T=150 in plant)in P20 and P25
- 4. Air volume measurements at t=75 (T=150 in plant) in P20 and P25 12/31/97 A. Void clearing of long 2.0 inch loop,1/4 scale (28.0/900 gpm). Initial void =
151/d's
- 1. Air volume measurements at t=75 (T=150 in plant) in P20 and Video B. Void clearing of long 2.0 inch loop,1/4 scale (17.4/560 gpm) Initial void =
151/d's.
- 1. Video 1/5/98 A. Clearing of 1/4 scale loop with SI sect. (17.4,20.8,38.3/560,670,1230 gpm)
Initial void volume is 151/d's.
- 1. Void fraction measurements to SI pump, P29 (Horizontal void trap)
B-5
@ TM-1876A 1/6/98 A. Clearing of 1/4 scale loop with Si sectior. :.;:d horizontal void trap. (17.4,20.8, 38.3/560,670,1230 gpm)
- 1. 5.461/d's initial void (34.6 in* =1.27 ft' in plant) 2.121/d's initial void (74 in')
- 3. 5.461/d's initial void (34.6 in3 =1.27 ft' in plant) 1/7/98 A. Clearing of 1/4 scale loop with SI section and horizontal void trap. (17.4,20.8, 38.3/560,670,1230 gpm)
- 1. 5.461/d's initial void (34.6 in3 =1.27 ft3 in plant)
B.1/4 scale clearing with 28 gpm to chuging pump only (900 gpm in piant) 1-4. Let run for approximately 5 min. Measured air in P20 & P25.
C.* 1/4 scale clearing with 28 gpm to charging pump only (900 gpm in plant)
I'& 2. Let run for 75 s. Measured air in P20 & P25.
1/8/98 A. Numerous Froude number tests with 2.5 and 4 inch test sections 1/9/98 ;
' A. Measured volume of air at top of elbow E16 B. Complete tests with angled entry to pipe trap P29. Flow of 17.4 / 20.8 / 38.2 gpm in test facility (560 / 670 /1230 gpm in plant),5.461/d's.
- 1. Air volume in P25
- 2. Air volume in P25 and P29 ,
- 3. & 4. Air volume in P25 and P29, and air flow : ate to P29 (SI pump) -
5.& 6. Air flow rate to P29 (SI pump).
1/12/98 A. Complete tests with angled entry to pipe trap P29. Flow of 17.4 / 20.8 / 38.2 ,
gpm in test facility (560 / 670 /1230 gpm in plant). Twice the initial void volume,121/d's.
- 1. & 2. Volume in P25 and P29, and air flow rue to P29 (SI pump)
- 3. Air flow rate to P29 (SI pump)
B-6
b TM-1876A Haseline Data (Plant Sequence 560-1230 gpm)
The initial bubble size in P2 was approximately 34.6 in'.
Test # Air Volume Air Volume Air Flow Rate P25 in' P29 in' 3 P29 in /s 1/9/98 B 1. 2.55 (0.41 1/d) - -- --
B2. 2.75 (0.441/d) 25.1 (4.01/d) --
B3. 2.95 (0.471/d) 25.7 (4.1 1/d) 0.70 (over 27 sec)
B4. 2.75 (0.441/d) 23.4 (3.71/d) 1.32 (over 7.13 sec)
B5. 1.22 (over 7.76 sec)
B6. 1.56 (over 4.02 sec)
Tests were also mn with an initial bubble size of approximately 75 in3 , including initial
. data on 1/6/98 and the data below.
Test # Air Volume Air Volume Air Flow Rate 3 3 3 P25 in P29 in P29 in /s 1/12/98 A1. 9.42 (l.51/d) 69.9 (l1.11/d) 5.40 (10.02 sec)
A2. 0.39 7 69.5 (11.11/d) 4.96 (10.76sec)
A3. -- --
6.59 (4.77 sec)
B-7
x 0 TM-1876A
-In aildition to the tests reported above which involved different sampling intervals, a number of verifications were performed. Piping slope was adjusted at various points of the pipe network,'with slight differences in result, Sampling interval was_ varied, to assess the degree of peaking in the rate of air collection. (This effect is encompassed in _the results above). Two?
different air collection geometries were employed. Initially, horizontal pipe P27 expended into a horizontal run of twice the diameter, which was then collected by a vertical pipe at top. This geometry tended to trap an air bubble at the expansion. Subsequently, the expansion pipe was . .
sloped,- te_nding'to trap air at the high point of the" slope. The latter geometry was_ preferred:
because little air in fact made it to the high point (it was removed into the vertical collector).
= Moreover, the air at the high point was measured (post test) and used to provide a high bound on
, - the total air collection. Downcomer pipe P30 was observed during tests, no air was carried under in that pipe.
Lastly, air is held up for a time in the expansion pipe P28 to the collector pipe P29,-
thereby altering the apparent rate of collection. Accordingly, we placed two simultaneous observers at pipes P27 and P29 respectively, each timing their observation of the air transported or collected. Collection in the vertical column was analyzed as above, an accurate measure but- ,
subject to air holdup in pipe P28. The observations of P27 are not influenced by holdup in pipe .
- P27, but are more difficult to make, The observer et P27 found that the air flow rate was at first -
relatively constant and highly visible, then abruptly tapered off. He measured the first time
-interval (of appreciable air f,'ow), however attributed the entire collected air volume to that interval.- In this way, the result of the second observer is an upper bound in two respects; First it is insensitive to holdup in pipe P28. Second, it overestimates the amount of air delivered in the
- measured interval. The' P27 approach was indeed higher than the P29 approach, but by only' 20E Accordingly, the P29 result has been reported and its uncertainty is indicated by this independent verification at P27.
B-8
_ _ _a
(OQfe TM 1876^
APPENDIX C SUPPLEMENTAL CALCULATIONS l
C-1
NO TM-1876A Froude Number Froude Number and related calculations were performed to determine the appropriate Dows in scaled test facilities. To maintain a similar How regime in our scale facility, it must operate at the same Froude Number as the plant. The Froude Number is defined as: ,
F = v/(gd)" (1) where v is the velocity in the pipe, t si the acceleration of gravity, and d is the pipe diameter. To determine the flow in the scaled facility, one calculates the Froude Number in the plant for the -
conditions of interest. This combined with g and d of the scaled facility allow one to determine the velocity of the flow rate in the scale model, and hence, the flow rate. For example, the initial flow in the plant to the charging pump is 560 gpm. From the flow rate and the cross sectional area of the 8 inch pipe, it can be readily calculated that the flow velocity in the 8 inch sections is 3.574 ft/s. Using this velocity and the above equation yields a Froude Number of 0.774.
Plugging this back into the above equation for the corresponding 2 inch diameter pipe in the one-quarter scale facility gives a velocity of 1.79 ft/s and a flow rate of 17.4 gpm. Similar calculations were done for plant flows of 670,900, and 1230 gpm to obtain their scaled flow rates of 20.8,28.0 and 38.3 gpm in 2 inch piping.
Air Volume Collected '
The air volume to reach the line running to the SI pumps and the line leading to the charging puinps was trapped and measured at a given point in time; typically 75 seconds after test initiates (equivalent to 150 seconds of plant time)._ These tests use the modified test procedure given at the end of the Test Procedure in Appendix B. The only calculation performed ;
with this data was a straight forward air volume calculation based on the measured column height of air and the known column diameter. This volume can also be expressed as a fraction of the
-initial air void. For example, one of our tests gave a column height difference of 1/16 and 3/16 inches in P25 and P20 respectively. This height combined with the cross sectional areas of the two columns (The diameters for P25 and P20 are 2.0 and 1.5 inches respectively.) yields air volumes of 0.2 and 0.3 in' in P25 and P20. A similar calculation was done for P29 for some of the tests.
Void Frution During a test, we first determined rate of air flow into the SI and charging pump lines by measuring the rate of air entrapment in our two air traps P25 and P29. Dividing this by the water L flow rate yields the peak void fraction in the flow. Calculations were only perfo; -d for the SI pump in that the air flow to the charging pump was extremely low. In one test (w...ch had an initial air pocket twice that of the best estimate), the air level in P29 was found to drop 18 inches in 10 seconds. Eighteen inches of a two inch diameter column yields a total of 56.5 in3 or a rate 3
of 5.66 in /sec. The water flow rate through piping leading to the SI pump in the scale model is 3
20.8 gpm or 80.1 in /sec. Dividing the air flow rate by the water flow rate yields a void fraction of 7.1 %
C-2
, _=. =
s
TM 1876A-Initial Air Pocket Volume To support our testing, an initial calculation was'done to determine the appropriate quantity of air for the start of the tests.: We received verbal instruction that the initial air volume in the plant is 1.27 ft' in an 8 inch pipe. This represents 2,195 in' at full scale and 34.3 in' at one-quarter scale. For simplicity, it is also a length of pipe that is 5.6 times its diameter, in the horizontal pipe P2, this was just slightly less than half full of air. The section of 2 inch diameter pipe is 23.5 inches long. Its ratio of length to diameter is 11.75. Italf of this is 5.8 We filled pipe P2 of the one-quarter scale facility _with air equivalent to just under one half of the pipe, matching the ratio 5.6 above, I
i
)
C-3
h[@[Q TM-1876A APPENDIX D SCALING STUDY D-1
N EO TM-1876A On inspection we expected overhead pipe P10 in the one-quarter-scale facility to tend to trap air for a time in the fonn of a stratified How. Moreover, downeomer Pl1 tends to develop a bubbly flow, but has difficulty drawing the air from the run P10. The multiphase flow behavior of a run feeding a downcomer is not well documented in the literature, although we have seen it before (including some 1996 tests for NU for other purpose).
We modeled this pipe segment at two scales, as described in Appendix A. This gave us an opportunity to confirm and illustrate the appropriateness of Froude scaling. And it also allowed us to examine the effect of Weber number on the rate and completeness of bubble formation.
Figure D1 shows the degree of air entrapment as a function of Froude number with steady flow. The data are identical in the tested range, within experimental uncertainty of these quick single-sample tests. We would expect if anything that the rate and degree of air removal might increase slightly at the larger scale, a result of increased flow velocity (and Weber number) at fixed Froude number. This effect, if any, is indistinguishable in the present data.
The actual situation is an unsteady flow. We tested with the overhead run (P10 at one-quarter scale) initially about half full and the water flow stagnant. Then water flow was stepped up and held steady. Air is removed in a spectrum of bubble sizes. Fine bubbles are quickly transported down the downcomer at velocities near that of the water. However, such fines constitute a small fraction of the total (initial) air volume.
We were able to observe " neutral" bubbles that were dragged downward into the downcomer, then tended to stay relatively in place for long periods. At times such bubbles would move downward briefly, then they would rise again. Over time, statistically, such neutral bubbles would tend to progress downward and toward the bottom of the downcomer pipe. But then they would often agglomerate to a larger bubble and rise against the flow. As a result of these multiple and competing mechanisms, air tended to remain for long periods in the downcomer, and the process of air removal from the overhead run above was slow.
Recall that in Figure Dl, very little more air was removed at half-scale than at one-quarter scale, even after waiting for a long period. Qualitatively, bubbles were slightly larger at the larger scale, perhaps by 10% of diameter. That is, bubble diameter was nearly the same in size when the pipe was nearly doubled in diameter. Similarly, the bubble entrainment process was slower at the larger scale, a result of Froude scaling. However, when time was properly normalized, the transients were similar within our degree to define them.
Such details are of some scientific interest. For engineering purposes or the present NU evaluation, summary findings are:
- 1. Froude scaling is appropriate,
- 2. A long chain of small bubbles is observed at both one quarter and one-half scale,
- 3. Projection to full scale tatroduces minimal uncertainty.
D-2 i
3 ]
I re ru.1876A Froude Number Scaling 0.6 ....i....i.... ....
....i....
e 2.5 inch pipe :
E 4.0 inch pipe _
0.5 -
e . .
e .
5 0.4 - -
u -
e .
.9- e .
@; 0.3 - -
n .
.9 -
e c .
u .
5, 0.2 -
m -
. m .
0.1 - E -
~
. e .
0.0 '''''''''''''''''O'8' '
O.2 0.3 0.4 0.5 O.6 0.7 0.8 Froude Number Figure DI. Air Entrapment Declines As Flow Rate Increases D.3