ML20205K578
| ML20205K578 | |
| Person / Time | |
|---|---|
| Site: | Davis Besse  |
| Issue date: | 12/31/1998 |
| From: | FAUSKE & ASSOCIATES, INC. |
| To: | |
| Shared Package | |
| ML20205K567 | List: |
| References | |
| FAI-98-126, NUDOCS 9904130277 | |
| Download: ML20205K578 (85) | |
Text
{{#Wiki_filter:. FAI/98-126 WATERHAMMER PHENOMENA IN CONTAINMENT AIR COOLER SERVICE WATER SYSTEMS Client: Toledo Edison Project: Davis Besse Prepared by Fauske & Associates,Inc. d 16WO70 West 83 St. Burr Ridge, Illinois 60521 December,1998 9904130277 990407 'p PDR ADOCK 05000346 p PDR -
ii l TABLE OF CONTENTS faEt j
1.0 INTRODUCTION
.. . . . . . .. .. . .... . . . . . .. .1-1 2.0 PHENOMENA.. . . ... . . . . . . . . . . . .. . . . . .2-1 2.1 The Influence of Dissolved Air in Water... . ... . . . . . . . . . . . . .. . 2- 1 2.1.1 Evidence of Noncondensible Gases Exiting from Solution . . . . .. 2-3 2.1.2 Importance of Small Steam or Gas Voids on the Mixture Sonic Velocity..... .. . . . . . . . . . . .. . ... .2-11 2.1.3 Noncondensible Gases Exiting Solution Due to Subatmospheric Pressures. .. . . . . . . . . . . . .. . . . . . . . . . .2-16 2.2 Waterhammer Loads During the Voiding Phase.. . . ... .2-18 2.2.1 Analytical Approach.. . . . . . . . . . . . . . . . . . . . . .2-20 . 2.2.2 Comparison Data...
. . . . . ... . ... . . ..2-26 l l
2.3 Waterhammer Loads During the Refill Phase.. .... . . .... . . ... . . . . . . . . . . . 2-3 4 j 3.0 TREMOLO BENCHMARKING.. . . . . . .. ... . .. . . . . ... . .. . 3-1 3.1 Purpose.. .. . .. . . . . . . . . . . .....3-1 3.2 Separate Effects Tests. . . . . . . 3 -2 3.2.1 Dynamic Benchmarking of the Containment Air Cooler Heat Exchanger Model . . . . . . . . . .. .. . 3-2 , 3.2.2 Heat Exchanger Model Benchmark Against Davis-Besse Containment Air Cooler Pulldown Data . . . . . . . . . .3-5 3.2.3 Dynamic Benchmarking of the Two-Phase Sonic Velocity Model. . 3-5 l 3.2.4 Benchmarking of the Pipe Wall Heat Transfer Model. .. .3-5 3.3 Integral Code Tests. . . . . . . . .. ..3-11 3.3.1 All Liquid Waterhammer Benchmark Against Method of Characteristic Solution . . . . . . . . . . . ..... . . ... . . .3-11 3.3.2 MOV Test: Condensation-Induced Waterhammer Due to i Sudden Valve Closure. . . . . . . . . . . . . . . . ...... . ...3-14 l 3.3.3 Delft Hydraulics Laboratory Test 180: Pressure Surges in Condenser Cooling Water Systems.. . .. . . . . . ...3-18
i
~
I iii l ! TABLE OF CONTENTS (Contd.) i . f.agg 4.0
SUMMARY
,,4 1 )
5.0 REFERENCES
.. .5-1 d l l s l j l l i l l l l l l
i iv LIST OF FIGURES l 2-1 The formation of bubbles of vapor over cavities in a heated surface
- (taken from Rohsenow,1973) . .
..2-5 2-2 Experimental apparatus for two-phase flow due to flashing (taken from Henry, Fauske and McComas,1970a) .. . . . . . .2-7 2-3 Pressure and void fraction profiles for Run 3 of TS-R7 (taken from Henry, Fauske and McComas,1970a) . . . . . . . . . . .2-9 2-4 Experimental data of Lkingebiel (1964) and Cruver (1963) indicating dissolved gases are coming out of solution. .. . .. .2-10 2-5 Comparison of the homogeneous adiabatic model and Eq. 2-5 with the steam-water data of Karplus . . . .. . . .2-12 2-6 Comparison of available models and the steam-water data . .. . .2-13 2-7 Comparison of Eq. 2-8 and the steam-water data of Dejong and Firey. .2-14 2-8 Test apparatus for observing noncondensible gases exiting solution at subatmospheric pressures . . . .. . . . .. 2 17 l 2-9 Measured void fractions at subatmospheric pressures for room temperature water saturated with air at one atmospheres. . .. . 2 19 2-10 Influence of water attempting to form a separated configuration in a horizontal i piping segment. . . . . .2-22 2-11 Entrapping of a steam bubble to create a waterhammer condition. ..2-23 2-12 Schemati . oasic experiment on condensation induced waterhammer reponed L, (31ock, et al.,1977) . . ... .2-25 l \
l 2-13 Expanded plats of selected waterhammer pressure reported by (Block, et al., 1977).2-27 1 2-14 Experimental configuration for investigating possible waterhammer conditions in the service water system. .2-28 2-15 Typical pressure history during voiding and refill. .2-31 l 2-16 Comparison of waterhammer incidents and the thermocouple response during voiding of the horizontal loop seal. .2-33 I l
v LIST OF FIGURES (Contd.) 1 2-17 Representation of the bubble rise for a steam void during refill.. . .. .2-36 3.2-1 Results of the TREMOLO Revision 1 Dynamic Benchmark of the fan cooler heat removal rate. . . . . . . .. ... . .. . . .3-3 3.2 2 Results of the TREMOLO Revision 1 Dynamic Benchmark of the fan cooler-outlet temperature calculation. . . . . . . . . . . . . .3-4 3.2-3 Davis-Besse CAC-1 Calculated Heat Transfer Rates. .. .. . .3-6 3.2-4 Results cf the TREMOLO Revision 1 Dynamic Benchmark of the two-phase sonic velocity model. . ... .. . .3-7 3.3-1 TREMOLO Revision 1 Calculation of All-Liquid Waterhammer . .. .. .3-13 3.32 TREMOLO Revision 1 integral code benchmark against two-phase MOV closure experiment using lower bound valve for miaimum void fraction.. . .3-15 ' 3.3-3 TREMOLO Revision I results without data smoothing .. .. . .3-17 3.3-4 Comparison ofTREMOLO Revision I results with lower and upper bound on minimum void fraction. . . . .3-19 3.35 Delft Hydraulics Laboratory Test Configuration for Pressure Surge Test 180 with a depiction of the separate voided regions and water columns following column separation . . .3-21 3.3-6 Test 180 Fluid Velocity Near Pipe inlet (Instrument Position VA).. .3-24 I 3 17 Test 180 Fluid Velocity Near Low Level Reservoir (Instrument Position VB). .3-25 1 3.3-8 Test 180 Pressure Near the Closed Valve (Instrument Position P1).. .3-26 3.39 Test 180 Condenser Outlet Pressure (Instrument Position P7).. .3-27 3.310 Test 180 TREMOLO-Calculated Axial Void Profile Prior to Column Rejoining ) in the Condenser (Condenser section is located from 22 to 27 m from the pipe inlet) .. .3-28 3.3-11 Test 180 Water Column Velocities Calculated by TREMOLO (Upstream column refers to the water column between the closed valve and the condenser; Downstream column refers to the water column between the condenser and the low-level l reservoir . .3-29 l
vi LIST OF TABLES
.Pagt 2-1 Solubilities ofNitrogen in Water. . . .. .2-3 i
2 Example Calculation ofNoncondensible Gas (Assumed to be N2) Dissolved in Water . . . . . . . . . .. ..2-4 3.2-1 Comparison efNumerical and Analytical Solutions at Time: 0.500 2 Seconds for Heat Flux = 5.0 W/m ,5 Nodes, Time Step = 1.E-5. .. . .3-9 3.2-2 Comparison ofNumerical and Analytical Solutions at Time: 0.500 2 Seconds for Heat Flux = 5.0 W/m ,10 Nodes, Time Step = 1.E-5. . .3-9 3.2-3 Comparison of Numerical and Analytical Solutions at Time: 0.500 2 j Seconds for Heat Flux = 0.5 W/m ,5 Nodes, Time Step = 1.E-5 .3-10 j 3.2-4 Comparison of Numerical and Analytical Solutions at Time. 0.500 , 2 Seconds for Heat Flux = 0.5 W/m ,10 Nodes, Time Step = 1.E-5. . .. .3-10 3.2-5 Comparison of Numerical and Analytical Solutions at Time: 0.500 2 Seconds for Heat Flux = 5.0 W/m ,5 Nodes, Time Step = 1.E-4. .3-11 I l l l
I I \- l l 1-1
1.0 INTRODUCTION
I l This repon addresses waterhammer conditions in service water systems that provide l cooling for the containment air coolers. The report is directed at the influence cf net steam generation and subsequent condensing of the stearii bubble within containment air coolers as a result of imposed Design Basis Accident (DBA) conditions. The relevant events for these
]
assessments are:
- There is an interruption of the service water pumping capacity such that the service water flow wi'l coast down to a negligible flow rate within a few seconds.
- There is steam in the containment gas space consistent with the DBA conditions and this steam condenses on the fan cooler tubes causing net steam generation in the service water system.
- The service water flow rate is re-established after an interval of sufficient length that' there is substantial steam formed in the containment fan cooler and the associated upstream and downstream piping.
The contents of this repon is selected to provide a technical bai, for responding to the l NRC requests for additional information (RAI) regarding GL 96-06. In panicular, the information assembled in this report addresses the following RAI items: ; i
- RAI Item 2: "
explain why this methodology is applicable and gives conservative results for the Davis-Besse plant."
- RAI Item 1 '
describe the methods used to benchmark the codes . " i e RAI Item 5: " Provide a detailed description of the " worst case" scer.arios for vaterhammer and two-phase flow."
- RAI Item 8: " Determine the uncenainty in the waterhammer and two-phase flow analyses, explain how the uncertainty was determined. "
l l l- . . . . _ . .
I l l 2-1 1 2.0 PHENOMENA
. l The sequence of events for the service water cooling flow through the containment air l coolers begin with a steady state normal operation condition followed by a transient to a second steady state condition corresponding to the DBA condition. Thus, the cooling water flow starts in a single phase cooled condition and experiences a significant heat load due to the LOCA j
condition such that upon the re-estabhshment of electrical power and the service water pump capacity, a new steady state c de an with heated water exiting the fan cooler is subsequently l established. During such an evcat the piping in the open cooling water system for the containment air coolers would experience a transient drain down after the service water pump l stops due to the postulated loss of offsite power. The occurrence of a LOCA in containment would result in heat mpus to the cooling water and given the low service water flow rate lead to steam generation and voiding of the air cooler and attached piping. The void progression into the attached and initially cold supply and retum piping would displace water from that piping. Steam condensation will occur on the exposed cold piping and influence the extent of void i propagation. Upon the restoration of electrical power and the service water pump capacity, the voided region of the air cooler piping as well as the containment air cooler would be refilled. The dissolved gases released from the coolant during the low pressure transient in the cooling piping would result in a residual void following the refilling of the air cooler and piping and the return to a new steady state flow condition. The following sections discuss the key controlling phenomena associated with this sequence of events. 2.1 The Influence of Dissolved Air in Water The Davis Besse service water system is drawn from the lake, passes through various components including the containment fan coolers and is returned to the lake. As a result, the incoming water is saturated with air at one atmosphere with the concentration at saturated conditions being given by Henry's Law, typically stated as P, = K X, (2-1) l
2-2 In this equation, Pgis the partial pressure of the noncondensable gas over the water, K is the Henry's Law constant and X, is the dissolved gas mole fraction. This is revealing from one particular aspect: dissolved gases always exist in cooling systems where the water is exposed to air and no specific actions or processes have been taken to remove the air. For cooling water sy:tems in which the water is taken from a lake, a river, a cooling pond, etc., the incoming water I would be saturated with air at atmospheric pressure. Consequently, any process which causes net steam generation, such as heat addition Or flashing, would also cause air to exit from solution. The existence of the air lowers the waterhammer loads resulting from rapid transients in two-phase mixtures. It leads to the concept of a residual void in the system. When condensation occurs, the air goes back in solution much more slowly than the steam condenses. In the case of the rapid condensation occurring during waterhammer events, all of the air essentially remains in the gaseous state. Hence, this residual air is available to " cushion" this dynamic process. For open systems, one can assess the influence of gas coming out of solution by determining how much gas would have to exit from solution to have a substantial influence on the loads being evaluated. Table 2-1 shows the values listed in the Handbook of Chemistry and Physics (Hodgman, Weast and Selby,1958) for nitrogen. To simplify the following example calculations, one can assume that the dissolved gas is 100% nitrogen instead of 80% nitrogen and j 20% oxygen as is the case for air. Note that the panial pressure of gas in this table is given in terms ci millimeters of mercury (Hg). For a gas which is saturated at one atmosphere, this partial pressure would be 760 mm/Hg. It is noted that there is little difTerence between 0*C and 20*C for the Henry's Law constant Thus, for this sam %e calculation the value for 20*C (68*F) l will be used. l l i l
2-3 l l l Table 2-1 Solubilities of Nitrogen in Water l K x 10 K = P/X P = partial pressure mm. of Hg X = mole fraction Temp. (*C) t=0* 10* 20' 30* 38' 40* 50' 60* 70* l Nitrogen 4.09 4.87 5.75 6.68 7.51 7.60 8.20 S.70 9.20 1 Table 2-2 is an example of the amount of nitrogen that would be dissolved in water at a temperature of 68*F (20*C). Assuming that all the gas were to exit from solution, and that the local pressure was 2 psi (13,793 Pa) the volume that the gas would occupy represents a void fraction of 0.083 (8.3%). 2.1.1 Evic ence of Noncondensible Gases Exiting From Solution Before evaluating the influence of small gas voids on the mixture compressibility, let us consider experimental evidence of noncondensible gases leaving solution. Consider small wall cavities as illustrated in Figure 2-1 (taken from Rohsenow,1973), which arise due to scratches, pits and surface imperfections at grain boundaries. For this discussion it is sufficient to consider the re-entrant cavity shape shown in Figure 2-1 item e. This configuration simplifies the discussion since it removes the wetting angle as a parameter. With an internal pressure Pi in the flowing system, water is forced into the wall cavities until the cavity radius is sufficient to suppon the pressure difference with the water surface tension (c). Inside the envity, the pressure is the sum of the water vapor pressure P (T.) and the partial pressure of nonconder , ole gases. If the temperature of the flowing system is approximately that of the source, the latter value would be essentially one atmosphere (P.,). Therefore, the largest cavity diameter (de) that could suppon the imposed pressure difference is given by d, = 4o{(P1 - (P.(T.)) + P.,)) (2-2) 1
f 2-4 l i Table 2-2 Example Calculation of Noncondensible Gas (Assumed to be N2) Dissolved in Water X = mole fraction ofdissolved gas 1 i P = partial pressure of the dissolved gas (in mm Hg) l K = Henry's Law constant 1 X = P/K 1 1 For a temperature of 68*F (20*C) K = 5.75 x 10' with prolonged exposure to air, P = 760 mm Hg, i.e. I atm. Therefore 760 moles of N, X = 5.75 x 10' = 1.32 x 10-3 mole of water
' 2 8'
= 2 x 10
Mass fraction = X s 18s 1 m' of water = 1000 kg Dissolved N2 mass /m' = 2 x 102 pg For a pressure of 2 psia (13,793 Pa), what would be the gas volume if all the N2 were to exit from solution?
, , m RT ,0.02(8314)(293)
P M. N: 13.793(28) V, = 0.12 m' ! Therefore, the void fraction would be approximately 0.12 if all the gas would exit from I solution. l l 1 l -
2-5 ( 'I' > s, N N
- > 13
>u N
l O l h> 2
\
-i NN o Figure 2-1 The formation of bubbles of vapor over cavities in a heated surface (taken from Rohsenow,1973) l
f 2-6 1 This interface (and all similar to it) provide the site (s) from net steam formation if the imposed pressure decreases dramatically (Dashing), the water temperature increases due to heat addition (boiling) or both. The activation of such cavities is given by the limit of stability condition, i.e. 4cr . P. (T.) + P - P i = d, (2-3) which shows that the noncondensable gases also have an important role in the inception of net steam gener:aion. (Equation 2-2 is the limit of stability since the cavity is stable at this condition but a funher increase in the water temperature (T.) or a decrease in the imposed pressure would cause the steam-air bubble to grow without bound since the surface tension forces would only i l decrease as the bubble grows.) i Rohsenow summarizes es Mence that prolonged boiling (hours) tends to reduce the noncondensible gas influence (degas the cavities) and increase the effective wall superheat. For the open system configuration ofinterest in this evaluation, there would not be such prolonged. boiling conditions. As a result, the available nucleation sites would be in equilibrium with the partial pressure of dissolved gases. The above discussion focused on wall cavities but similar considerations hold for entrained panicles (or motes)in the cooling water. These also have surface irregularities that can serve as nucleation sites given the conditions of flashing or boiling in tb coolant flow. Therefore, the sites for steam formation can be both along the chann, walls and in the water as well. For those situations with heat addition in the fan cooler, the fan cooler tube walls are the
, preferred sites as a result of the temperature gradient from the tube wall to the water coolant.
Given that the incoming water is saturated with dissolved air, it is worthwhile to examine other experimental evidence indicating that noncondensible gases leaving solution would influence the net steam generation Many two-phase critical flow studies have been performed in systems similar to Figure 2-2 Here high temperature water flows through a constant cross-sectional area duct with the coolant pressure decreasing due to wall friction. Considering only steam and water, one would spect the pressure to decrease linearly until it is equal to, or less I
o4 R E W E
% S r N E N G T O b O S T I
Y Y ( R E C R R T E T E E E % I N S W H r N I V L OUI L V L GxI I H L D E D E D E E L @ H B T _ I L N E W O WS y L OS B D E V . R E fE R C Z I U
= L D A "
S R " N E L E A R N I LXN I I T M F L E Xp rd P R E T I A D E E E H ExI L N L w k B 5 8~ mTn % mj 5,a; ] m& F i k 8 S E , 8e?I- F$g xje![grS$E$# .
2-8 than, the saturatien pressure corresponding to the water temperature. At this condition, flashing would begin and the pressure would decrease at an increasing rate as the steam-water mixture was formed. However, as shown in Figure 2-3a, the pressure profile d'eviates from the linear pressure decrease at a pressure significantly greater than the saturation pressure corresponding to tne water temperature. Moreover, gamma ray densitometer measurements of the two-phase density (Figure 2-3b) show that the void fraction has increased to 0.4 at the location where the ' pressure reaches the saturation value (P. (T.)). This measured behavior is due to dissolved gas in the water exiting from solution as the pressure approaches the saturation value. In this experimental system, the water was demineralized and saturated with air at one atmosphere. Hence, the air still exhibited a significant panial pressure after the water was heated to temperature of 300*F (149'C) to 400 F (204
- C). Therefore, tne dissolved air influenced the incipient flashing condition and the magnitude of the steam generation It is also interesting to investigate other such experiments that investigated low quality, high velocity flashing flows. Pressure profiles from two other studies are shown (Figure 2-4) and both of these clearly indicate a departure from the linear pressure profile, typical of all water flow, before the pressure reaches Poi (T.). Consequently, these investigations also experienced the influence of nwondenuble gases exiting from solution. Thus, there are clear examples of the importance of noacondensable gas in the inception and magnitude of the steam formed in the processes of flashing and boiling.
For the system of interest, the water is taken from the lake and circulated through the service water system including the containment fan coolers and the reactor cavity coolers. Depending on the elevation of the respective coolers, these may, or may not, undergo column separation. For those which do, flashing would occur as pan of column separation with the air helping to initiate the steam generation and being liberated into the gas space as well. In the next section it will be shown that this quantity of noncondensable gas can have a significant effect on the subsequent waterhammer loads since it can reduce the mixture sonic velocity by an order of magnitude with very small gas quantities.
2-9 74 l l l g l l l 70 - - ) l 66 - -
.2 8 62 2.
58 - pg I 54 - - 50 l I I I I I I ' 32 28 24 20 16 12 8 4 O LENGTH, in. , I I I
'58 - &
pse - 1.0 56 - o P - 0.8 5 o.
,. 54 - 0.6 52 -
c g - 0.4 l 50 - - 0.2 I I I 48 O l 4.0 3.0 2.0 1.0 0 LENGTH, in. 1 Figure 2-3 Pressure and void fraction profiles for Run 3 of TS-R7 (taken from ' Henry, Fauske and McComas,1970a).
f 2-10 4e i i ; i i i l
, i 44 - -
42 - - 40 - Psat - RUN 24 38 - (KLINGEBIEL',1964) _ 36 - - 34 - 32 h -
.2
~
30 - - A O 28 - - l 26 - -
~
RUN 11 P sat (CRUVER,1963) 22 - _ l 20 18 - - 16 - - l l l l l l l l Y g 18 16 14 12 10 8 6 4 2 O LENGTH,in. Figure 2-4 Experimental data of Lkingebiel (1964) and Cruver (1963) indicating dissolved gases are coming out of solution. I i l l l
I 2-11 For those postulated accident conditions with a LOCA into the containment, the high steam panial pressure caeses energy transfer to the fan cooler coils and possibly boiling in the coils. Here again the dissolved air would be important in determining the inception of steam generation and would also 1e liberated into the steam volume. If subsequent conditions cause the, steam to be condensed, the noncondensable gases would again influence the mixture compressibility by decreasing the sonic velocity. 2.1.2 Importance of Small Steam or Gas Voids on the Mixture Sonic Velocity Numerous experiments (Henry, Grolmes and Fauske,1971, Karplus,1958, Semenov and Kosterin,1964 and DeJong and Firey,1968) have demonstrated the influence of small quantities of steam in the mixture compressibility. In particular, these are illustrated in Figure 2-5 for the experimental data of Karplus (1961) for a pressure of 10 psia, the data of Henry, Grolmes and. Fauske, (1971) for a pressure of approximately 38 psia (Figure 2-6) and the data of DeJong and Firey, (1968) illustrated in Figure 2-7. In these figures, there are different models represented for expressing the influence of the void fraction (c) on the mixture compressibility and these are 1 Isothermal
, 1 ni P asi = < - -
, (2-4)
Ps ni P Ps a' + a (1 - a) P,
+
(1 - a) + a (1 - a) Ps. . P.P 3I t s , Adiabatic i l
)'
! I i yP l al, (2-5) P yP P, , y) A_0, (3 , y ): + y (i , y ) _s 2 a + y (3 . ,
.. sJ . P . P. "I t
l
I 2-12 350 , ;
' E 300 - - -
u N I
$ 250 - - . l C 0 s
oJ 200 o l W
% oO ~
1 > \ 3 \ O o 150 - 0 - i I , i 8 Eq. (3-8) ,. l 100
, ~____ " O 50 **
1 1 O O.10 0.30 0.50 VOID FRACTION, G l Figure 2-5 Comparison of the homogeneous adiabatic model and Eq. 2-5 with the steam-water data of Karplus. f
l 2-13 O o I Q d j e - O 0 m g 000 g _ O O ; OO G 80 l
% h ~
O O l 0 e ! O O O C . l O OE o 0
~
. bSH
- ) - o g
O - L h3 o i
- .g O - g
- o
- a -
R 2M - 0 0 - O
,oo l I I I I O S to IS 20 2S 30 9010 FlWCTION(Q) X M Figure 2-7 Comparison of Eq. 2-8 and the steam-water data of Dejong and Firey.
2-15 i In the preceding equation, a is sonic velocity, p, is liquid density, p, is gas density, ni is gas moles, p is gas pressure, y is isentropic exponent, and all parameters are specified in a t consistent set ofunits. 1 At moderate the low pressures, the liquid compressibility may be neglected for all but the very small void fractions (a = 0.01). This simplifies Eqs. (2-4) and (2-5) to al, l , "2
'T *
- y (2-6) a* 2 a + a (1 - a) Pr P,s and aj. , , 1 (2-7)-
** a + g(3,y) &
2 P. s These are the homogeneous models for the substantial reduction in the sonic velocity with the I occurrence of small voids. The comparison of these models with the experiments show that these represent a strong , decrease in the sonic velocities (the mixture compressibility) with the presence of very small void fractions. In particular, when the propagation velocity is compared with typical values for the propagation of sound in water (approximately 4500 R/sec) it is seen that small void fractions that are much less then 10%, and even for those less then 1%, can reduce the mixture sonic velocity by an order of magnitude. There is the additional expression included in some of the figures that considers the deviation between the measurements and the homogeneous isothermal expression. Considering the vinual mass of discrete vapor bubbles in a water matrix, one can arrive at an expression that the two-phase sonic velocity is represented by the function h = 1.032 + 1.676a (2-8) asi From this further consideration, the results for larger void fractions are better represented than those expressions typically given by the adiabatic and isothermal representations. However, for tnose void fractions less than 10%, the representation of the homogeneous models is sufYicient to indicate the large influence of small steam-gas voids on the mixture compressibility.
1 i 2-16 This substantial decrease in the two-phase mixture compressibility is important when assessing the waterhammer loads throughout a coolant system like the fan cooler and reactor l cavity cooler elements of the service water system. In particular, the exiting of noncondensable gases from solution creates a situation in which the condensation of steam as the system pressure increases in not suflicient to completely eliminate the existence of a gaseous phase. Of particular note is that the air cannot be driven back into solution as efliciently as it exits solution during the l initial vaporization process. Therefore, steam condensation still results in a significant second phase, which is the composite of the remaining noncondensable gases and the steam existing at a partial pressure equa. to the water temperature at the gas bubble-water interface. As a result, the l
" cushioning" of this fa, more compressible region lengthens the time over which the momentum l
is transferred from the water flow rate resulting from the restart of the service water pumps and ) as collision with the existing water slug (s). With this " son landing", the water slug (s) is (are) accelerated over a much longer time than that which would be considered by a water slug, impinging on a second water slug with no compressibility associated with the two slugs other than that of the water alone. Such a " soft landing" enables the acceleration to be slower and the loads to be distributed over a longer time such that the influences on turns in the piping l configuration are minimized. Thus, the respective loads on piping supports and hangers are also minimized. 2.1.3 Noncondensible Gases Exitine Solution Due to Subatmospheric Pressures For the Davis Besse open service water system, the coolant is at 1 atm and would be exposed to subatmospheric pressures with the interruption of service water pumping power. Given the specific pressures, this could be sufTicient to force noncondensible gases to exit from solution To examine the extent of the local void fractions that could be generated as a function of the system pressure, FAI performed simple experiments to assess the possible influences that would be created given column separation conditions. Figure 2-8 illustrates the apparatus used to investigate the void fractions created by subatmospheric pressures imposed on room temperature water. (Note that the approximately saturation pressure for room temperature water is 0.33 psia. In these experiments, the pressures were substantially greater than this value such that the observed void fractions are almost entirely made up of air exiting from solution.).
( l 2-17 I l l l I 1
,3 l
I I Scale ; Manometer. - _ 1
- j
- - l l
4 3 r - - - To Level _ Vacuum - Increase - Pump - I -
* * * * ~ ' " " ' ' ~
es .,,, 0' - O - Water o -
. O -
O *
.RH971035.CDR 1 10-97 Figure 2-8 Test apparatus for observing noncondensible gases exiting solution at l subatmospheric pressures.
I l 2-18 I The results from these experiments are shown in Figure 2-9. As illustrated, depressurizing room temperature water begins to exhibit a void fraction when the system l pressure is decreased to approximately 6 psia with the void fraction reaching about 0.01 (1%) { with a pressure of 3 psia. As discussed earlier.in this section, such void fractions have a I substantial influence on the mixture pressure wave propagation velocity (sonic velocity). Therefore, gas exiting solution would have considerable effect on the waterhammer loads l (magnitude and duration) that would evolve. for piping configuration experiencing column 1 i separation as well as those systems where noncondensible gases would be forced from solution { as a result of heat addition. Obviously, for the design basis conditions considered in the Davis Besse analyses, both of these conditions would be important and should be part of the I evaluations. 2.2 Waterhammer Loads During the Voiding Phase . The potential for waterhammer to occur in a horizontal pipe when cold water is being voided (pushed out) by steam is discussed in this section. Specifically, when considering the two-phase conditions in which waterhammer may occur in the service water system for safety grade fan coolers, it is necessary to evaluate both the voiding and refill phases of the transient. The voiding phase is that interval immediately following the onset of the Design Basis Accident (DBA) (large break LOCA to the containment atmosphere) combined with a Loss-Of-Offsite Power (LOOP). Under these conditions, the voiding phase occurs as steam is produced in the containment fan cooler due to heat transfer from the containment atmosphere, reduced flow due to the temporary loss of the service water pumps and potentially due to column separation for those fan coolers located at a high elevation relative to the service water discharge header. Substantial voiding can occur in the service water piping assuming that the water remains in the cooler tubes to support continued steam generation. Depending on the specific system configuration, steam voiding may progress a substantial distance through the service water ; system and into horizontal piping As the continued steam flow pushes water through this cold piping, waterhammer events have been observed to occur but only to a limited magnitude, i.e. less than 100 psig. 1
L 2-19 4&OL L WC3'C30L4fdW I I I I t N 4
- 0 -
o 4 0
- cu en EEE O
o <l O
.5 m
V c. Y
=i m 1 ng 1 0 40 $
o 40 N O <1 O O o 4 e i I oa I I I o to v co N v-o o o o o o o 6 6 6 6 6 6 6 n 'uopoeJa p!oA 1 l l l Figure 2-9 Measured void fractions at subatmospheric pressures for room temperature water saturated with air at one atmosphere. j
2-20 Numerous experiments have been reported in the literature related to waterhammer in various configurations including those reported by (Block,1980), (Chou and Griffith,1990) and (Rothe et al.,1977). Many of these experiments show substantial pressures developed as a result of waterhammer phenomena (several hundred psi). Most of these situations are I established with water filling a steam void. Those configurations of interest in the service water system (steam " pushing" water) represent a substantially different geometry than has
]
typically been examined. Also, this represents a unique configuration since the steam is I l l displacing water from a given region and the pipe wall under these conditions is cold. As is discussed below, this has a substantial influence on the local two-phase flow characteristics which are responsible for the waterhammer events. l
}
2.2.1 Analytical Approach i I As voiding occurs through the cold piping, the rate of voiding (U,) can be related to the steaming rate in terms of an energy balance on the pipe wall heat sink: 1 x Di l P, U , hr, =P x De Se c3 (T. - Tso) U, 3 (2-9) 4 l where p,, U, and hr, are the steam density, velocity and latent heat of vaporization respectively, De and Se are the pipe diameter and wall thickness with p3, c3 and T3 . being the density, specific heat and initial temperature of the steel pipe wall; T is the saturation temperature at the local pressure. Rearranging this equation in terms of the dimensionless i parameters gives
' Del'P,'~
U, h r,
-= -
U, (46p) spss _ c3 (Tw - Tso), (2-10) l Experiments in a 2-inch Schedule 40 carbon steel pipe result in a measured voiding rate of approximately 1 ft/sec when the pressure is approximately 1 atm. Furthermore, the voiding process appears to occur in a one-dimensional manner even though the water voiding velocity l
2-21 corresponds to a Froude number much less than unity, i.e. an air-water mixture would be expected to stratify. It is instructive to investigate the steam velocity that corresponds to this measured voiding rate. For this size of pipe, the internal diameter is 2.067 in (0.053 m) with a wall thickness of 0.154 in (3.9 mm). Assuming that the pipe wall is initially at ( ' and the system pressure is at 1 atm, results in a calculated steam velocity of 57 ft/sec (17 m/sec). This is nearly identical to the water flooding velocity in a vertical configuration as given by Kutateladze as ga(pr d Ur = 3 d , (2-11) Ps where g is the acceleration of gravity, o is the steam-water surface tension and pr is th] density of saturated water. This is not a coincidence since the water cannot stratify in the presence of a steam flow sufficient to " flood" the water surface. Hence, a situation develops (Figure 2-10) where the continued steam flow " floods" and perhaps entrains the water that is attempting to drain (stratify) and a thermal boundary is developed that moves along the horizontal piping in a one dimensional manner determined by the steam flow. In this configuration, waterhammer events are observed to occur on a regular basis during the voiding phase but with limited magnitude, i.e. pressure increases 0.2 to 0.4 MPa. With a configuration like Figure 2-10, the upper surface can separate from the pipe boundary and expose a cold wall thereby promoting condensation. The pressure difference of importance is that which results in a steam velocity sufficient to " flood" the water surface, cause a wave to form and capture a steam bubble as illustrated in Figure 2-11. This pressure difference is expressed by
2-22
\
Steam , r Water
\
~ /
T Z 1(a) Condensation f
#' ,, ,,- /' i r \
Steam 7 Water
' /
,m333 coa 115*
1(b) Figure 2-10 Influence of water attempting to form a separated configuration in a horizontal piping segment. - ~
2-23 s: s N N N N N N N N l N ; N I N N N N N N g N N N E -N p$+ : N , N 5 N N N l N N b .N N l N N l N N N b .: N N . d: N N m; ' N N ~l.; N g :. . g N N N N N N N N N N N N N N N N8 N E N+ m N
=
N N $ JL Nb N O N W N N N N N N N N g g . D Figure 2-11 Entrapping of a steam bubble ia create a waterhammer condition.
2-24 1 A P, = - p, U! (2-12) Once the wave is formed and reaches the top of tt)e pipe such that a steam bubble is captured, this pressure difference is available to accelerate a water slug. Given this geometry a waterhammer could occur due to the collapse of the entrapped gas space that is being pushed into 4 a region with a cold steel wall surrounded by cold water. For such conditions, the waterhammer can be expressed by the Joukowski equation for one water slug colliding with another, 1 A Pwu = - Pr a. U. (2-13) where a. and U. are respectively the velocity of sound in water and the water slug velocity. Considering the available pressure difference, the water slug velocity is given by A P, P* U. = = Ur (2-14) Pr 2 Pr , 1 Substituting this into the previous equation results in { 1
- a. P, Pc Ur (2-15)
APwu = I 2 Further substituting for the flooding velocity gives the expression 3 w APwu = 1.06 a- Prgo (2-16) This equation enables one to realistically estimate the low energy waterhammer events that are developed for a condition with steam pushing water through horizontal cold piping. l It is to be noted in this evaluation that the waterhammer events discussed in this paper are occurring in a pipe of constant diameter This is an important feature since the controlling mechanism for developing condensation is the rate at which the pipe wall uncovers. This is substantially different than the " water cannon" experiments reported by (Block et al.,1977) as illustrated in Figure 2-12. In these experiments, steam was added to the top of the test apparatus and slowly displaced water out the bottom of the test tube. This could be accomplished since a stable thermal boundary layer separated the steam and water as the water was displaced
2-25 FLUSH WOUNTED TRANSOUCERg % STE AM OVER PRE SSURE) S T AN00FF - MOUNTED TR AN SOUCER ( DE PRESSURIZ ATION ) " I O 9"4 M s
\
\O. x \ \N x x
, >QS<,
1 ', -
, '/
/ I l
'/ /,// // /
l , I I I l l l l l METAL PIPE OR
- i 28' l ACRYLIC TUGE l
I I l 1 ai i s 1.5"10 WATER RESERvolR l l l _ - _ _ _ _ _ -_ _ _ _ -p _ _ - ._ I
- _ - _ = , ,
,. ~ s ._
M e g -
~ '
s - ~ ~ I _ _ _-e __ - I
) ,
A %
-_ - - :_ w m m i
WATER CANNON MODEL : 1 Figure 2-12 Schematic of a basic experiment on condensation induced waterhammer reported by (Block et al ,1977). j l l l
I 2-26 downward. However, when the water cleared the bottom of the tube, the protective thermal boundary layer was pushed away and cold water came into direct contact with steam. This resulted in a depressurization of the steam space and a rapid entry of the' water into the cube due to condensation. Water impacting on the top of the tube generated the large pressure pulses illustrated in Figure 2-13. In this case, the geometry at the bottom of the tube provided for a configuration in whirh the thermal layer could be easily destroyed and cold water could be
- l brought into direct contact with the steam. In the piping arrangements typical of service water systems, this is not the case. Particularly, when substantial voiding of the piping configuration I
occurs there is considerable energy in the pipe wall that acts to develop important thermal layers in the water during both voiding and refill. The combination of this, and the fact that the geometry essentially remains with a constant cross-sectional area, means that there is virtually no way of rapidly removing a thermal boundary layer as was the case for the apparatus shown in Figure 2-12. Hence, the nature of the waterhammer events are strongly influenced by the. flow configuration. Substituting typical values for saturated water in Equation (2-16) results in a calculated waterhammer pressure of about 30 psi which is in general agreement with the data. 2.2.2 Compcrison with Data FAI has performed numerous waterhammer experiments on different cor. figurations related to the geometries of components, in particular fan coolers, for open service water systems. These include one-inch and two-inch diameter experimental configurations with elevated configurations which experience column separation. Moreover, the test programs have net steam generation with no reverse flow permitted in the supply riser as well as with reverse in the supply riser. A typical experimental configuration, is shown in Figure 2-14. 1 l l l
1 2-27
)
y V E
- AL 2
- E mEsouho
$ i - t,vgepotssumt $ Ping : 5PIEE . 2400 -
j :PCAE OFF SCALE) 9 # ~ l[ 4 ' I z" - 3 2000 pg* 8200 0si 33
<=
L p -{-- a ;;i60C y l t
^ .6 Es s lee -
f . W 000 - 1 l 8 8 CEPRESSURifAfl0M b 400 -
.,2 ' ' ' ' ' l ' I ' ' ' '
0 O .co 200 300 400 S00 0 1 2 3 i2 , g ! 9 8 ~ OvtRPRESSURE (PEAK 0FF SCALE) %' SPIR SPirt E "" REtoveso
. 2400 "
g4
/ u 2000 -
l ji 0 m ~ _ _ _, .L__. 01'*00
,, . i 0 0 0 ,..
Q' 2' ? j I g ,1200 - 5 ' e00 - 8 ~8 "DEPRE $$URIZATION O 400-
.i2 i i i 1 i i i I e 9 : I i ! ,
0 10 0 200 300 400 SCO O I 2 3 12
- RESOUNO
~ OvtRMitSSURE SMR SPlaE I* N ~
E 8
~ (75AN OFFSCALE q$
~
g J
._ i g 2000 -
j (l30 a*
~
__ . . _ _ _ _ OI E~al200
8 " N * *'
j' I-* .00 . I .e -OtPRESSURlZAfl088 8 400 .
.it i I , I i l i i i e i i it I i 0 i00 200 300 400 300 Q s 2 4 flutteseen Tiut ta seen SIWtiLTAMEOUS PRESSURE TRACES IN WATER CANNON MODEL l
Figure 2-13 Expanded plots of selected waterhammer pressure reponed by (Block et al.,1977).
I l 2-28 l [ .\ l ! f ,
- i
$.f3
- f. GNN E w23 8 f_$
,4 u
3 ll-*C
'I j[
?l t :
E[! i e-e 1
, u 5 ?. l E
n 2 j., i i l , e f ! 2 2 - r i l -
- e. .
- I 0I l &w '? $-
I ~ sl. l
. Ml w: ge 3 -
r j N-a l
& p-I r . .' . 3:
l i
.. 4................,< .
U .3 i cli l Q l 7 . l i ._ m t l tF * ! l'~ g z' >a[. " j C__gn
~
5
} ,
1 t i T k,'e - R. s
,l ii Figure 2-14 Experimental configuretion for investigating possible waterhammer conditions in the service water system.
l l l L
l 2-29 A typical procedure for these experiments was to set the water Dow rate through the test apparatus (from the water supply vessel to the evacuated receiver vessel) to represent a desired steady-state condition. To accomplish this, the gate valve immediately upstream of the evacuated receiver vessel and that valve immediately downstream of the water supply vessel , were used. This is representative of the plant condition since the gate valve at the exit of the 1 water supply vessel enables the experiment to control the water addition rate during refill in a !. manner that approximates the service water pumps restatting. Once steady-state conditions were established, the solenoid operated ball valve on the water supply line was closed to simulate the loss of service water pumping capabilities. Simultaneously, the solenoid operated ball valve on the line from the steam generator was opened to admit steam to the downstream piping configuration. This net steam addition represents the result of heat transfer from the containment atmosphere to the fan coolers for the-DBA condition. In the experimental test matrix, the rate of steam generation was varied as was the steam addition interval. All of the experimental configurations investigated included tests in which sufEcient steam was added to completely void the downstream piping to the gate valve on i the receiver vessel. 1 Thermocouple measurements were recorded along the centerline of the discharge piping, which enabled one to observe the void progression into the cold piping configuration as
]
well as the water refill transient. This provided measurements on the single-phase /two-phase
- transiet,ts within the pipe and greatly aided the interpretation of the two-phase flow state during the voiding and refill portions of the hydrodynamic transient.
Once a range of net steam generation rates and duration of steaming were investigated, an additional consideration to be addressed was whether an extended length in the horizontal segment of the loop seal for the downstream piping would substantially alter the waterhammer pressures. As shown, the bottom of the loop seal has a length of 23 ft., which corresponds to a length-to-diameter ratio of approximately 138. This is more than sufficient to initiate condensation induced waterhammers from a stratified (steam over cold water) situation if this could occur for the parameters investigated. With the other 2-inch configuration, this extended downstream length also has thermocouples in the center of the flow stream to me.litor the l l 1 i
I 2-30 l progression of the steam void during the voiding phase and the water front during the refill transient. Waterhammer events were observed during the voiding phase. These events had pressure increases of tens of psi, i.e. substantially less than those shown in Figure 2-13. The first concern for such experiments is whether the appropriate pressure increases l could be monitored. In all of the experiments performed this was tested at the end of the l l experiment by rapidly closing the downstream ~ manual ball valve and monitoring whether the measured pressures are sufficient to stagnate the water flow by a single increase (waterhammer) given by A P = p, a, U. (2-17) In this equation p, and a are the water density and scnic velocity with U,, being the steady-l state water velocity before the valve closure. If the velocity in this equation is replaced with' the refill velocity, the calculated pressure is twice the column rejoining pressure assuming l water impacts on water. l As will be shown, the resulting experiments always demonstrated the capability to measure such pressure increases at the end of the experiment. This was a convenient way to assure that the experimental apparatus could monitor the waterhammer events of interest. Furthermore, the measured pressure increases are bounded by the column rejoining pressures. The experiments performed as if there was a check valve on the fan cooler supply piping were those in which there was no drain-down of the supply riser. Figure 2-15 illustrates one of the measured pressure histories foi an experiment in the 2-inch configuration. As illustrated, this particular transient was initiated by opening the manual ball valve l l immediately upstream of the receiver vessel, thereby causing column separation, which was followed by opening of the solenoid operated ball valve on 'he water supply line to establish
I 2-31 i
- 1
-1 ;
g -
!]=:
I 1,5 C i
=
%g___.;- _
a
- ] W __
p .2 - - __ = - j _._ t
- 1 f
- g _ _
1 \ y ;_
^
I
~
U _ i o;
- 5. ry _
S h ___. oY g l E o __ _ .n *- 1 f L ,_ g
- r .
i _g B _ g _g g _s , p _
- 7 .
- _o _-
- . . i . . .a . 1 , 0 CS Cl-
-7 CLI CII oc.
915d >d-Figure 2-15 Typical pressure history during voiding and refill.
2-32 normal flow through the test apparatus. During this normal flow period, the pressure in the loop seal of the discharge piping as mesured at instrument location P4 is approximately 35 psig. After normal flow was experienced for approximately 5 seconds, the solenoid valve for the water supply flow was closed and, simultaneously, the solenoid valve controlling steam addition was opened. This caused a short term pressurization transient as a result of the steam C addition. The subsequent behavior resulted in a depressurization to approximately 15 psig as the steam " void" pushed the water column through the discharge piping into the receiver vessel, which was at a pressure of approximately -28 in. Hg. During the first 5 seconds of the steam addition transient, some small pressurization events were observed, which is typically the interval over which the horizontal run at the highest elevation was voided. After approximately 5 seconds, the steam void was pushing the water column through the downcomer and there were essentially no waterhammer events recorded, which is expected since this is a stable configuration, j After 22 seconds of steam addition, the thermocouples in the flow stream indicated that i I the steam void had penetrated through the downcomer to the bottom of the loop seal. At this stage, some waterhammer events of approximately 10 psi are observed. The voiding rate for this test was a " velocity" of about 1.5 ft/sec. At the end of this interval the steam ,oid j ingresses into the vertical riser and somewhat stronger waterhammer events are recorded with the largest being approximately 60 psig. This was the largest event observed in any of the experiments performed in this configuration during the voiding phase of the test sequence. After this time the entire test apparatus was voided as indicated by the increasing pressure in the system as the piping structural heat sink increases in temperature due to sustained steam flow. During the interval of complete voiding, the thermocouples in the voided region also demor. strate an increasing temperature corresponding to the increasing system pressure. I l l This experiment shows that those conditions in which the supply riser remains full l l would experience waterhammer events during the steam voiding phase, but that these events would be in the range of tens of psi. The strongest event tends to be when the system I encounters a loop seal and the steam void progresses to the vertical riser part of that loop seal. However, even in this case the observed events are tens of psi. Figure 2-16 illustrates typical l l
2-33 ae TER DAMMER TEST e6 (11-!4-96)
=. ~ '
'85- l .
C . d J - . e *. 2
.3
-C J h
%hh h h f l! h o -
= l
- l .
l o* _ J . f I I I I I I
~
50 55 60 65 70 75 80 85 90 j "T!NE* *CSEC)* I k y, ,i ,,.i.... ., i.... , ,
*TCS* :
E :. : TCi2 ---- -
- "TCl3* -
o e -
- 7 l
l
~
I, 3O L: - A e w e.
=
7 9 a- l Y ' ' ' ' ~ j O IT o 1 8a :: I ,
- i C a 7 g
- Ir. , ; :
!: i.:
/ .
O L s' .
* - i . .
' ,.,,,,o 3
e, f f I I I
, j g 50 55 60 65 70 75 80 85 90
' TIME" *CSEC)* ^
Figure 2-16 Comparison of waterhammer incidents and the thermocouple response durin; voiding of the horizontal loop seal.
I 2-34 measured pressures and temperatures during voiding of the horizontal pipe segment. As shown, the waterhammer event are numerous but small. Their magnitudes are consistent with the proposed model. The voiding of cold water from initially cold pipe may occur for the fan cooler system during the postulated LOCA plus LOOP accident due to steam generation in the fan cooler. This leads to the formation of fluid conditions which are not conducive to large magnitude condensation induced waterhammer events and loads. A significant size two-phase and thermal boundary layer region is formed between the voided region and the cold water initially within the cooling water piping. Additionally, the presence of the cold piping provides a significant
- heat sink for condensing the voiding process such that it generally behaves as a one-
! dimensional process with relatively small magnitude waterhammer events. Furthermore, the ; condensation of significant amounts of steam leads to the accumulation of enhanced amounts of' air originally dissolved in the cooling water. The presence of air influences dynamic effects during both the voidir.g and refilling transients. , l 2.3 Waterhammer Loads During the Refill Phase During the water refill stage. the flow rate delivered to the service water system is determined by the pump flow as the service water system is loaded on to the emergency AC busses. Of particular importance in assessing the potential for waterhammer events is whether l the flow rate is sufficient for the water to flow through the service water piping in a " plug flow" l manner. This is also an important consideratiori wi.en deciding the flow rates that should be 1 I used for experimental systems which are smaller than the plant configuration. The most important element of this refill rate is the Froude number given by U" Fr = (2-18) d8-D
2-35 where U. is the refill velocity and D is the pipe inner diameter specified with a consistent set of units. If this Froude number approache; or exceeds unity, the horizontal pipe will nin filled with water (Wallis et al.,1977) and significant condensation induced waterhammer would not occur 1 during this refilling transient (B.iorge and Griffith,1984; Izenson et al.,1988). If this is the case, i the dynamic loads on the piping system and the piping suppons would be those related to the l refilling velocity and the pressure associated with this is given by the expression
. A P = p a U-nu (2-19)
In this expression, water propenies have typically been used for the density (p) speed of sound l and wrter (a). Separate effects experiments show that this generally overstates the pressure increase by at least a factor of two and perhaps an order of magnitude. For typical plant service water flow rates, the Froude number for the refilling of the fan cooler discharge piping is-comparable to or greater than unity. Therefore, this is another key scaling element to be represented in an experimental investigation. Another aspect of the refilling rate is the behavior of the venical piping with the water being added at the top of the piping configuration. In this regard, the drainage bchavior of the piping segment can be related to the refilling rate to determine if the vertical piping can run full ! in the downward direction or whether it is determined by the arainage of film and/or rivulets as well as water falling through the central region of the pipe. l Consider the accumulated water at the top of the vertical segment has no significant downward velocity. Thus, water would accumulate and tend to be pushed downward at the refill velocity. This would tend to form a steam bubble (Figure 2-17) which would attempt to rise against the incoming water flow rate (Wallis,1969) characterizes the rise velocity of such inenially dominated steam bubbles as g: U. = 0.345 < (m
- h (2-20)
P.
2-36 Water Refill . y r/ /
/ 1 r 1 r 1 r 1 r / !
f g Pipe Wall
~
j j f/ N Bubble Rise l
/ Velocity, Um
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ Steam Void /
l
'/
(Steam Bubble) %
/
l
^/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
i , i
/ / l
/ / l
/ / l
/ /
/ g /
/) \/
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
=
D = RM96$033.CDR 9 5 96 Figure 2-17 Representation of the bubble rise for a steam veid during refill.
i 2-37 1 t where the constant 0.345 has been deduced experimentally but is quite close to that derived j analytically by (Davies and Taylor,1950) and others. Since the water density is far greater than that of steam, this essentially reduces to U. = 0.345 M (2-21) l For the 8-inch discharge piping, this rise velocity is about 1.6 ft/sec (0.5 m/sec), which is much' ' less than the refill velocity. Thus, from hydrodynamic considerations alone, the steam void would be " pushed" ahead of the water refill region. Those thermocouples inserted into the central region of the venical and horizontal pipes indicate that the regions are " quenched" sequentially at a rate equal to the refill velocity. This thermocouple information is discussed later in this section. For a 10-inch pipe, the Froude number reference velocity is 5.1 ft/sec. Thus, refill rates which result in velocities comparable to, or greater than, this value would be considered to flow through the reactor avstem in a " plug flow" manner. With this geometry, only minimal interfacial surface area would be exposed for steam condensation between the voided region and the cold water supply. A similar velocity for 2-inch piping, such as that used in the experimental configurations, is 2.3 ft/sec. The experimental programs for the vr,rious configurations investigated flow rates of this magnitude, as well as rates somewhat less than and greater than
- this magnitude.
Relating back to the Figures 2-15 and 2-16 shown in Sectic.n 2.2.2 for the pressure behaviors observed during the entire transient, this shows that the refill phases during the experiments exhibited some small waterhammer events, but, in general, progressed quite smoothly to the water filled condition After 60 seconds of steam addition, the refill transient was initiated as shown in Figure 2-15. With the water addition, the system pressure in the loop seal decreases from approximately 15 to 0 psig as a result of the condensation process. During this time, the pressure transducer at the top of the apparatus (P ) observes some waterhammer events that are approximately 20 psi in magnitude. However, these are not observed at the measurement station represented in Figure 2-13 due to the large compliance of j l the steam void separating the two. In fact, with the measured refill rate in this experiment, the l t
l 2-38 l l thermocouples show a rate of 3 ft/sec, the " .wde number is greater than unity and one would expect the refill process to be proceeding in essentially a " plug flow" manner. It is important to note that the injection of cold water into the experimental apparatus resulted in a depressurization of the system (as expected) and this would tend to increase the refill rate i l compared to the steady-state condition. Thus, as the water or steam-water mixture arrives at the downstream valve, the system should expect some pressurization sufficient to reduce the water flow to that typical of the nominal value. This pressurization would then propagate back through the apparatus causing a decrease in the water velocity to the nominal flow rate. This particular event was observed in several sets of experiments, particularly those in which there was no gate l valve to control the flow rate from the water supply vessel! For those situations with a substantial increase in the refill velocity due to the depressurized state of the experimental ) l apparatus, a significant pressure increase was observed when the refilling mixture encountered the downstream valve. However, these pressurizations were usually in the range of tens of psi, with some pressure increases as large as 260 psi when a refill velocity of 22 ft/sec was used for one particular test in a 1-inch test apparatus. In summary, the results from the numerous experiments performed for different configurations demonstrate that refill velocities which have Froude numbers comparable to unity or greater do not result in strong condensation induced waterhammer events. This is in agreement with the characterization proposed by (Bjorge and Griflith,1984). i l
f~ l f 3-1 l 3.0 BENCHMARK 5 l l 3.1 Purpose l A series of code benchmarks has been performed to determine the applicability of the TREMOLO code to the analysis of complex pipe flow tran:ients stemming from design basis accidents such as those discussed in Generic Letter 96-06. TREMOLO has been benchmarked against relevant experimental data, transient thermal-hydrauliu problems with known analytical solutions, and results from other, established pipe flow codes. The outcome of these benchmarks will ultimately demonstrate that the TREMOLO methodology is applicable to the analysis of two-phase flow and condensation-induced waterhammer events that are postulated to occur in the service water piping systems of nuclear power plants. The following results of the benchmarking activities are divided into two categories: separate efreets tests and integral code tests. The separate effects tests exercise individual physics 'dules to demonstrate that key phenomena are adequately modeled by the code. Some of the < 4 rate effects tests follow an approach referred to as " dynamic benchmarking." The dyn. .c benchmarking concept involves embedding benchmark instructions within the source code such that the "off the shelf" code can be exercised every time the archived code is updated (Henry, et al.,1997a,1997b). Thus, the dynamic benchmarks enable the testing of individual code mode:s and subroutines in the ex- ~ form in which the models and subroutines are i implemented in the archived code. Fu..nermore, embedding benchmark instructions and relevant experimental data within the code provides a permanent record of the benchmark information. This is in contrast to the classic approach of extracting desired subroutines and writing standalone driver modules to perform the testing. In the separate efTects test results that j follow, both the dynamic benchmarking and classic benchmarking approach will be used. The integral code tests exercise the entire code to demonstrate the overall code { applicability to applied problems in transient two-phase pipe flow. The following integral code tests focus on transient flow problems similar to those postulated to occur in the service water
1 3-2 systems of nuclear power plants resulting from a loss of offsite power or a LOCA coincident with a loss of otTsite power. 3.2 Separate Effects Tests 3.2.1 Dynamic Benchmarking of the Containment Air Cooler Heat Exchanger Model l l The TREMOLO code relies on a mechanistic heat exchanger model that a.*mes I conditions of steam condensation on the outside surface of cooling tubes. The cooling water flowing through the tubes may be eith~r a single-phase liquid or a two-phase mixture. l
.1 To determine the heat exchanger model capability to predict heat transfer rates under a range of steady state conditions, a dynamic benchmark of the TREMOLO model against prototypical fan, cooler coil test data (Westinghouse,1969) has been performed. The Westinghouse tests were conducted for a range of containment temperatures, cooling water temperatures, and coohng water flow rates. The dynamic benchmark results, presented in Figures 3.2-1 and 3.2-2, compare the measured heat removal rate and outlet temperature in the fan cooler to that calculated by the TREMOLO code. Each data point represents a single test. The solid line on the figures is provided to show where perfect agreement between the experimental data and calculated quantity would occur. Error bars are provided, when available, for each data point as defined in (Westinghouse,1969). The figures show excellent agreement between the experimental data and the TREMOLO calculations.
I i
r- -- , l l 3-3 l c _\.
> i i
_ a
\ _
-\ -. y
\
N\ _
\ ._ _.
s- __ iO x _
\ ,
_. 7 N s to - c K i V - ev,
\, -
C C ~ v Lc >
- n -
h A
~ -
I l g - 4 - 4 Z - t 1 LJ m
~
C I T 2 J _ o ._ _ Z - _ m C - - 2 c E _ __ tp: un
~,
m
<N g
C rn
\
,, .,,,!,,,,l,,,,l,, ,
q gy e n c_. . ;
- b E'E E EC n;x .-
s V7H 031W~100 % 0 v,, -
- n. -- pi,,n:-mc i
, - a i
Figure 3.2-1 Results of the TREMOLO Revision 1 Dynamic Benchmark of the fan cooler heat removal rate i (
i 1 I t 1 3-4 C ' C
-(,, .
l
' ,i
_. r _\x - s I s - C
- \\, -
c'.
- I r
.sys x
i N _
\ . _
_ g m _. m y 1
- \ _
O s;
-._ s -- <
x; _ m - 1 ! \ C s ev V l \ - C -- 1 i _~ wm l C - x
' - r g_ _
O - o . Z ~ M _ m . 3 _ ___1 - ,- u . o __. y = c., r .- o _ - l _"_. _ y
^
m c r m , _ m ,f g o 'r
- ~ 1 a
i
~
o
\ _ m 1
o
,l.,,,,i,,,,#',,, ,
O '
,n, n;7 pn m
. ,- , , - , - n-- n ,, , , , , , , ,
n ; ,,
--- v-- -=c._- : vet v7c acc n v 7.c,, v c out S
*L-_u
_<,_._ _ ^__ _ . _ .
^r m q;>o.
- u. anon u _q
-v Figure 3.2-2 Results of the TREMOLO Revision i dynamic benchmark of the fan cooler outlet temperature calculation.
3-5 l l 3.2.2 Heat Exchanger b Jet Benchmark Against Davis-Besse Containment Air Cooler l UFSAR Data i TREMOLO Revision 1 provides a mechanistic heat exchanger calculation which is used l to model the containment air cooler (CAC) heat transfer rate under low flow and two-phase flow conditions. The calculated heat transfer rate is a function of heat exchanger geometry, cooling water flow rate, gas temperature, and tube fouling factor. To maximize heat transfer, a lower bound on CAC fouling factor of.00045 is used in the TREMOLO model for Davis-Besse. This fouling factor, combined with the other CAC heat exchanger design parameters input to TREMOLO, yields heat removal rates as a function of containment gas temperature, as shown in { 1 Figure 3.2-3. The TREMOLO results are compared to Davis-Besse data provided in (Toledo Edison,1990). As shown, TREMOLO calculates more heat transfer for the same conditions than the reference data, which is conservative for the current analysis. 3.2.3 Dynamic Benchmarking of the Two-Phase Sonic Velocity Model TREMOLO Revision 1 uses a model for calculating the sonic veloci*.y in single- and two-phase water assuming no heat and mass transfer between the phases. Figure 3.2-4 presents dynamic benchmark results of the TREMOLO sonic velocity model against experimental data from (Henry. et al.,1967) As shown, the TREMOLO model provides a best estimate representation of the experimental data for void fractions belcw 50%. 3.2.4 Benchmarking of the Pipe Wall Heat Transfer Model To test the TREMOLO Revision I transient wall heat transfer model, a standalone executable program was written utilizing the conduction subroutine from TREMOLO. Numerical calculations from the standalone program were then compared to the exact solution for transient conduction into a semi intinite slab with a constant surface heat flux (Holman. 1981). Specific conditions of the sample calculations are, i l t !
1 3-6 l l 1 1 I CAC-1: 85 F,1600 gpm. 00045 Fouling 275 i , , , , ,
)
,A S.
250 -
/
/ -
l B #
-J f
/
2 / 1 t 225 -
/ -
A - 5 /
,/
3 200 -
/
/ -
1
,/ :
2o /
/
U 175 - f p -
/
/
/
i gg I I I f I l 20 30 40 30 60 70 80 90 Heat Transfer to Cooling Water (MBtu/hr)
]
+ TREMOLO Air Cooler Model 4
+ Davis Besse Pulldown Data I FIGURE 3.2-3 Davis-Besse CAC-1 Calculated Heat Transfer Rates I
3-7 - i i l 1
~
l l o.
- I l 4
, _.- T I
i I l l I _ , ,i I - I _ V f 7
,y m = -
c - 7_ i
\
c o 1 Z l w I ; - l m m -
. g. ,
.- v.
g , 2 i t .- _ 00 _ __ c . , os ! = c C C $ i 3 _.
' o i,,, i.
g
^
c-v_L nnt vvv v 2. 7 nr7 vvm OC_'L 001 0 v m-- mm-e r ', Nin;;idn,,a cuds c in . v
^_ . cA ui. ;
I Figure 3.2-4 Results of the TREMOLO Revision 1 Dynamic Benchmark of the two-phase sonic velocity model. i
r 1 3-8 l i e Initial uniform wall temperature distribution = 320 K Wall thickness = 7.138 mm (corresponds to a 0.28 in. wall thickness which is typical of 6 l inch, schedule 40 pipe)
- Wall thermal conductivity = 54. MW/m/C e Wall Specific heat = 465. J/kg/C e Wall density = 7833. Kg/m' e Time step: either 1.E-5 sec or 1.E-4 sec e Number of wall nodes: 5 or 10 2
e Surface heat flux: 0.5 MW/m or 5.0 MW/m2 Tables 3.2-1 through 3.2-5 compare the numerical and exact wall temperature profile = 2 under low (0.5 MW/m') and high (5.0 MW/m ) heat flux conditions and with varying, nodalization schemes. All results are presented at a transient time of 0.5 seconds. As expected, improved accuracy is achieved with more nodes. Less accuracy is obtained with the numerically calculated temperature in the nodes farthest from the boundary with the imposed heat flux, because the numerical solution assumes a finite wall thickness while the analytical solution is for a semi-infinite slab geometry. 1 A sensitivity to the number of nodes used to represent the wall thickness is exhibited, as expected. As the number of nodes increases, the accuracy of the calculated temperature profile increases. Also, when fewer nodes are used, a smaller temperature gradient is calculated to l remove the same heat flux. For instance, from Tables 3.2-1 and 3.2-2, with an imposed heat flux 2 of 5.0 MW/m , and 5 nodes, the temperature drop through the wall is 257 K, while for the 10 node calculation, the temperature drop is 265 K. In the TREMOLO transient analysis, this result implies that if fewer nodes are used, a greater heat loss through the pipe wall will be calculated. This is conservative for the current analysis, since an increase in the pipe wall heat loss leads to increased steam condensation and potentially increased occurrences of condensation-induced waterhammer events. Finally, results presented in Table 3.2-5 indicate that for the time step l sizes considered, there is a negligible sensitivity in the results to the time step.
( l 3-9 l l Table 3.2-1 Comparison Of Numerical And Analytical Solutions At Time: 0.500 Seconds for Heat Flux = 5.0 MW/m*,5 Nodes, Time Step = 1.E-5 NODAL TEMPEPSTURES NODE NUMERICAL ANALYTICAL NODE DEPTH SOLUTION SOLUTION PERCENT ID (M) (K) (K) ERROR 1 0.0000 586.5137 594.4609 -1.3369 2 0.0018 452.8747 459.2443 -1.3870 3 0.0036 375.3574 378.1469 -0.7377 4 0.0054 338.1734 336.7050 0.4361 5 0.0067 329.8552 321.8808 2.4775 Table 3.2-2 Comparison Of Numerical And Analytical Solutions At Time: 0.500 Seconds for Heat Flux = 5.0 MW/m',10 Nodes, Time Step = 1.E-5 NODAL TEMPERATURES NODE NUMERICAL ANALYTICAL NODE DEPTH SOLUTION SOLUTION PERCENT ID (M) (K) (K) ERROR 1 0.0000 592.9850 594.4609 -0.2483 2 0.0008 525.6304 527.0381 -0.2671 3 0.0016 470.1858 471.3912 -0.2557 4 0.0024 425.9277 426.7992 -0.2042 5 0.0032 391.7629 392.1569 -0.1005 6' O.0040 366.3758 366.1002 0.0753 7 0.0048 348.3969 347.1465 0.3602 8 0.0056 336.5676 333.8266 0.8211 9 0.0063 329.8819 324.7910 1.5674 10 0.0069 328.2393 320.1194 2.5365
l 3-10 Table 3.2-3 Comparison Of Numerical And AnalyticalSolutions At Time: 0.500 Seconds for Heat Flux = 0.5 MW/m*,5 Nodes, Time Step = 1.E-5 NODAL TEMPERATURES NODE NUMERICAL ANALYTICAL NODE DEPTH SOLUTION SOLUTION PEPCENT ID (M) (K) (K) ERROR 1 0.0000 337.6514 338.4461 -0.2348 2 0.0018 324.2875 324.9244 -0.1960 3 0.0036 316.5357 316.8147 -0.0881 4 0.0054 312.8173 312.6705 0.0470 5 0.0067 311.9855 311.1881 0.2563 Table 3.2-4 Comparison Of Numerical And Analytical Solutions At Time: 0.500 Seconds for Heat Flux = 0.5 MW/m*,10 Nodes, Time Step = 1.E-5 NODAL TEMPERATURES NODE NUMERICAL ANALYT20.% NODE DEPTH SOLUTION SOLUTION PERCENT ID (M) (K) (K) ERROR 1 0.0000 338.2985 338.4461 -0.0436 2 0.0008 331.5630 331.7038 -0.0424 3 0.0016 326.0186 326.1391 -0.0370 4 0.0024 321.5928 321.6799 -0.0271 5 0.0032 318.1763 318.2157 -0.0124 6 0.0040 315.6376 315.6100 0.0087 7 0.0048 313.8397 313.7146 0.0399 8 0.0056 312.6568 312.3827 0.0877 9 0.0063 311.9882 311.4791 0.1634 10 0.0069 311.8239 311.0119 0.2611 l I f 1
3-11 Table 3.2-5 Comparison Of Numerical And Analytical Solutions At Time: 0.500 Seconds for Heat Flux = 5.0 MW/m*,5 Nodes, Time Step = 1.E-4 NODAL TEMPERATURES NODE NUMERICAL ANALYTICAL NODE DEPTH SOLUTION SOLUTION PERCENT ID (M) (K) (K) ERROR 1 0.0000 586.5500 594.4894 -1.3355 2 0.0018 452.9051 459.2698 -1.3858 3 0.0036 375.3754 378.1654 -0.7378 4 0.0054 338.1812 336.7158 0.4352 5 0.0067 329.8603 321.8871 2.4770 3.3 Integral Code Tests 3.3.1 All Liquid Waterhammer Benchmark Against Method of Characteristic Solution A sample run was performed with TREMOLO Revision 1.0 to simulate an all liquid waterhammer event in a municipal water system. Results of the TREhf0LO analysis are then compared against a method of characteristics (h10C) solution described in (Wylie and Streeter, 1978). The purpose of this benchmark is to demonstrate the capability of TREhiOLO Revision f 1.0 to calculate rapid waterhammer pressure rises, therefore use of an all liquid system provides a limit!ng analysis in terms of the pressure wave transmission speed and pressure rise time. Since this benchmark will be free of any steam-liquid phase transition phenomena, results of the benchmark will present a clear picture of the magnitude of numerical errors present in the explicit integration scheme employed in TREhiOLO Revision 1.0. Datails of the all liquid waterhammer benchmark are:
- total pipe length: 3060 ft
- pipe elevation change from inlet to outlet: 0 ft
- constant pipe inside diameter: 0.25 ft i.
3-12
- constant upstream reservoir pressure: 148 psia e iniFel steady state flow through the pipe: 1.09 fps e total pressure drop during steady state conditions: 2.5 psi Steady state conditions are assumed initially, then a valve located at the end of the pipe is closed instantaneously, giving rise to the waterhammer event as the water column velocity is reduced by the closed valve face. The high pressure wave originating at the closed valve will travel to the upstream reservoir which is held at a constant pressure and a pressure wave will be reflected back toward the closed valve. Because of the large pipe length, instantaneous valve closure, and all liquid conditions, the peak pressure rise observed at the valve face will approach the theoretical maximum pressure ulculated from the Joukowsky equation, AP = p ua, (3.3-1) where p = fluid density u = fluid velocity
- a. = sonic velocity in the ficid Given a sonic velocity of 4722 fps based on water density and bulk modulus at 85 F and an initial water velocity of 1.09 fps at a dens!:y of 62.4 lbm/ff, the Joukowsky equation yields a pressure rise of 69.3 psi.
Results of the TREMOLO Revision 1.0 sample calculations are presented in Figure 3.3-L Also shown in the figure are calculations based on a method of characteristics (MOC) solution assuming a constant liquid sonic velocity of 4722 fps. As shown, while the ideal wave shape approaches that of a square wave, the TREMOLO pressure calculation overshoots the peak pressures shown in the MOC solution then dampens out as the numerical error diminishes. The {-. .. .
w'. ". 5 t
- 4 i
8 ' i _ y ii i i 1 I i ' t I iI I i t i 3 l I ! 4 _ i ' l i y ' _ i i 5 A I IS i - i 3 P _ s _ p8 l > f 4 2 1 i > K R 2 72 i > A4 = _ M: I 3 HY4 _ C T4 I 1 6 _ t NC/ _. - EO2 i i _ B L2 i,# I e, R E 3 i l I , I ' i l i : i f E V1 I
) _
MC 4 1 _ 1 MN6 I 2 l- , _ 5 C _ kO* _ l E _ HS2 _ 2 _ S i hE 2 i 7 _ T F 4 i i ( . A5* 9 _ W8 0 i ! [ DP 1 i > E M I U EM+ S 2 . O l I T 5 C _. I L T LID5 I i i T _ LU4 S i 4 AL1 I R _ F E 0 ,= T 1 sE p C I _ , Nf R A O9 U R 5 A l I S 1 . S 0. S E . H _ 1 I V1 E RVC i
,I ! i i i i i i ' -
E : R t , : I ' ; i i I UE F R YP SR S O 6
-i ,
OIT X EO - L CA OOM RLD POO i 6 - TMH i L MEY EVK EE T LR E 1 I R TD SW I NTM I i , U O E L V FK i , U L O A 4 , J V l - i - I 5 I + l , l , ( , -
- i L f l TI- _ - _ - _ - - _ _ ~ -
0 t o~ ao" o" _ oWm c oo~
) -
m 4_ u O . mmDWWWEA 1 2D5 w'Wa WSTo1 n,E5m.O % >?C.iE 4g"gB38
o 1 I i 3-14
)
I resulting TREMOLO pressure then closely follows the MOC calculation. Furthermore, the initial pressure rise at the valve location in both the MOC and TREMOLO Revision 1.0 calculation is 69.3 psi (recall the initial pressure at the valve is 148 - 2.5 = 145.5 psia), which is precisely the theoretical value calculated from Equation 3.3-1. l Overall, the TREMOLO Revision 1.0 calculations closely follow the MOC calculation and produce a peak all-liquid waterhammer pressure rise equal to that calculated from the l Joukowsky equation. Finally, results of this benchmark indicate that because of the numerical error present in the TREMOLO Revision 1.0 explicit integration scheme, TREMOLO calculated peak pressures bound the theoretical pressure rise. j 3.3.2 MOV Test: Condensation-Induced Waterhammer Due to Sudden Valve Closure j I l l l This benchmark focuses on condensation-induced waterhammer under low void conditions. The two-phase conditions are obtained experimentally by first establishing steady ; I l state flow of slightly subcooled water through a pipe, then suddenly closing a valve at the end of 1 l i ! the pipe. The pressure response at the valve gate face is controlled first by an all-liquid l t I waterhammer event as the closed valve suddenly stops the flowing fluid. Once the resulting l high-pressure wave propagates to the upstream reservoir and returns (with a magnitude now l l based on the upstream reservoir pressure) to the closed valve, a rarefaction wave, originating at the valve disk, then travels upstream. Due to the elevated fluid temperature, the rarefaction wave is limited to the fluid saturation temperature, and, as the low-pressure wave travels upstream, steam voids are formed behind the wave. Once again a pressure wave (with a magnitude controlled by the reservoir pressure) reflects otT the upstream reservoir and begins to propagate toward the closed valve. This time, however, the pressure wave is traveling through a two-phase, bubbly fluid characteristic of flowing fluids with low steam void fractions. As the returning pressure wave moves down the pipe, steam voids begin to collapse, until, finally, the pressure wave reaches the closed valve whereupon a condensation-induced waterhammer event is observed. This condensation-induced waterhammer event is most closely characterized by the trapped void collapse classification of waterhammer events described in NUREG/CR-5220, Vol. 1.
\ w.* 5 e
~
- -- - 2
- ____ _ J- -
i
-i e
ei i i i 4 , ,1 K j -, j 2 R .... A i
= ,-,
i M 4 H i ti
, li;igi )i l u3 C
l iI, e'.:'.;.,'.'.' 6 I '
-lI i N - ,:'
s E 6 ;f ,I ll e i I f
- i. ) _
B I l l / '
., . l 5 S _
/
1 D N 4
/ .1 i N _
O I
- 0 1
0 ',*
, O C
i
- i T E G=
- . E .
S S A i C A RD
- j i tl C
R I T S V O EI ::::,, i:iiiiii,ii - A 6 AV E I . L R P A M M l E G U p 1 I _ T N E T C IN VI
- T S 6 E A O N -
I - i E 2.AR MN - N3H - N 6 R C ED T N0, n
- i M 6 COF MO I
O1 - U F/ P O - i L 6 0SD PO 5 L O AAH .N m,i O TLT GE 1 ! l ii i g.y 1.,.:i o; 5 C l AEE NR sI]' : e * , , . h . :,Il 0 l DRN IT - H , T . I 0 _ o i 0
- N i
e- - S A _C i e - _
- - T _
i A _ i D _ i _ _ e _^
- h_ t 0
D ON O~ o~ t e~ N- O- m e
- 3 nTI0" ym]OnJg1
< LtL t
u>J_c> g ' .- - . .
*gEMYw ,@?O o:=ha"_ggWEgRggz!a aeg 593g 6<
o 5Ea E5 ;r;:aa3ggcg<E lgP n5$a 03 's a cyn
3-16 l l 1 This benchmark is relevant to the Generic Letter 96-06 analyses because its outcome is dependent on the ability of TREMOLO to properly model several key phenomena expected to exist in the Generic Letter 96-06-type transients. For instance, the benchmark provides a test of the models for pressure wave propagation through a combined single end two-phase fluid l mixture, the ability of the phase interface transport model to generate and condense steam in non-equilibrium fashion, and the applicability of the one-dimensioaal, five-equation fluid model incorporated into TREhf0LO to model this type of waterhammer event. Results of this column separation benchmark, presented in Figure 3.3-2, compare the experimental data, RELAP5/ MOD 3.2, and meinod of characteristics transient calculations repoited by (G. Cerne, et. al.,1996) against TREMOLO Revision I calculations. Key features of the experiment include a large reservoir containing water at 325 F (436 K), connected to a 120 ft (36 m) pipe with a nominal pressure of 145 psia (1 MPa), and an initial fluid velocity through the_ pipe of 1.3 fbsec (0.4 m/sec). A valve at the end of the pipeline is closed instantaneously and the ensuing pressure transient near the valve inlet is shown in Figure 3.3-2. The TREMOLO code calculations closely fol'ow the timing, shape, and magnitude of the experimentally observed pressure transient, indicating that the important phenomena ci pressure wave propagation through single and two-phase mixtures, flashing, and st:am condensation are appropriately modeled. Since the TREMOLO code provides output to plot files with a frequency of several plot points per millisecond, the influence of numerical errors inherent in the explicit solution scheme
. err. ployed by TREMOLO is magnified, leading to numerically " noisy" plotted output. To reduce the numerical noise, the results presented in Figure 3.3-2 are time-averaged over 5 plot points.
i That is, the reported value for any given plot point is averaged with the values of the 2 plot points immediately preceding and immediately following the plet point in question. For comparison, Figure 3 3-3 presents the TREMOLO results for the curr ni ber.chmark without the benefit of time averaging.
r 1 3-17
. tn
,,,, g
_'liliiiijaiii;iii,g,,,, l_ l o_ f -
%" l -
M - _ y= - - __ , m t T ~ C - - o ,..- L 1
-- ,1,.- -
t- s _ ~- g -
, .. - w Z .e g
~
laJ CD
- , - - - -e. , - , n Z
~~
$#oz s- _
o
~ ~
g
'l T _ 8 _
I p_ ~ . w c g
- f. - --__ _ m o v g -
- g .
'~
C _; E! > W l 2-Q_ c 5 r h s a t- '~ s O E ): i-- (D - w g 2 I: - wNc r , I ,, zmz (_ 7 - ct' , o w(s O m: oc u. : h, - ro cao L_ N (; e<a., o go es -r o sur zu ~
-l,,q. -
* ~'"**"~~
l.h O ~ f f lf i,...., T..... - o - o
- r
- _
m I(
,i e e I, c c.
o 1: _ 0 : z r. , ~ i., ,
'Iilil!1 l i li,,, ,,, ,l,,,,
ie o AC BI 9[ H El 01 8 9 ' C8V8) 38nSS38d 3AIVA Figure 3.3-3 TREMOLO Revision I results without data smoothing. l i l l l
i 3-18 l l Figure 3.3-4 provides TREMOLO results to investigate the sensitivity in the calculations to the assumed residual void fraction. In TREMOLO, to account for dissolved gas coming out of I i solution plus residual vapor bubbles that remain after the larger scale void' collapse has occurred, ' a user-specified minimum void fraction is retained once the void in a fluid node exceeds the specified minimum void fraction. The void fraction is used to calculate the two-phase sonic i velocity and hence the node pressure. Results presented in Figure 3.?-4 indicate that the minimum void fraction in the range of .001 to .005 produce a good comparison between the TREMOLO calculations and the experimental data. l In general, assuming a lower residual void fraction will lead to larger sonic velocities and hence larger waterhammer loads - a conservative result for the current analysis. For the current j benchmark, the test conditions lead to the production of very low void fractions along most of the pipe length as the low pressure wave traveled along the length of the pipe up to the transient, time of 0.12 seconds. At many pipe locations, TREMOLO ca'culated void fractions less than
.005, thus the use of a minimum void of.005 limited the nodes in which a residual void was considered. This allowed an overdi faster wave transmission time along the length of the pipe 1
and lead to a sharper pressure rise during the condensation-induced waterhammer event at the i valve inlet, as shown in Figure 3.3-4. Thus, while assuming a smaller residual said is considered a conservative approach, for specific cases where only small void fractions are present in the system, using an upper bound value of.005 may be more appropriate. i l 3.3.3 Delft Hydraulics Laboratory Test 180: Pressure Surges in Condenser Cooling Water Systems The experimental investigations of H. H. Safwat (Safwat,1972) at the Delft Hydraulics Laboratory provide insight into the efTects of cold water column separation and rejoining in piping systems that are prototypic of those used in condenser cooling water systems of thermal power plants. The investigations of H. H. Safwat focus on the transient flow events that could occur as the result of a power failure and subsequent loss of the cooling water system pump. Safwat reports that if the condenser is located in an elevated portion of the cooling water system, I l
c 3-19 Ln iliiiiliiiilii.,,,,,,,,,,,1,,, .7 m. l _ t ts _ I - l 2 _ c~~j - cn. . cc ' l 5
^
f -LW - i __ __________________ 3 I o '_ ! z
*s ~.
W - g - CD ; . n
* - tn tn ,
z %,~, . a j i gg 't 'e _ z O Q W _ ~0 ~0 0 ""
~
4 _ o H ' ~~~~~_ . La - c -
-- U1 V
Cr W99 00 ( _ j >> - Lt) ! Q _ C
@rr
._ O0 ~
x w
~
+
w
>EE +
Lo - w e55 rr z H z - e y z..
~OO r _
o gaa - D , o v_a
- u. oo
~~~
~
.. O O ._
o' c ort , zuw ' ! r = tn o - s 255 o _ O ,. - O r i w I : e
. i l c I l H
~
C l o iii,li,iiiiihliiiliiii iiiiliiii o 02 AI 91 kl 21 01 9 9
- (8V8] 38nSS38d 3A7VA Figure 3.3-4 Comparison of TREMOLO Revision I results with lower and upper bound on minimum void fraction.
l l l l
r 3-20 relative to the pump discharge centerline, then a pump trip would initially result in column separation in the elevated ponion of the piping. This occurs as the momentum of the water I column results in continued forward flow of the water through the condenser return piping while at the same time gravity effects cause a flow reversal in the riser of the condenser supply piping. Furthermot ; since Safwat simulated pump trip by closing a ball valve near the piping system I inlet, a column separation was also observed in the piping near the closed valve. Thus, the growth and collapse of two separate steam void regions and the movement of two separate water I
~
columns characterize the experiments of Safwat. This geometry is qualitatively shown in Figure 3.3-5. ! l The pressure surge experiments of Safwat lead to three general types of waterhammer events. First, near the closed ball valve, there is the trapgd void collapse, similar to that t described in NUREG/CR-5220, Vol.1. This occuri. as water co amn 1 in Figure 3.3-5 reverses _ l direction, collapses steam void I, and contacts the closed ball valve. Second, there is a trapped void collapse as the two separate water columns rejoin in the elevated ponion of the piping system. In some instances one water column may be essentially stagnant while the other water column moving with some velocity compresses the voided ragion (void 2 in Figure 3.3-5) and impacts the stagnant column. In other instances, both water columns may be moving toward each other with independent velocities prior to the column rejoining. The third type of waterhammer event is the classic, all liquid waterhammer which could result when water columns 1 and 2 rejoin while column i is in com ete contact with the closed valve. The pressure wave originating at the point of the column - ,ining event would be transmitted along the length of water column 1 and impact the closed valve. This benchmark is relevant to the Generic Letter 96-06 analyses because it considers cold water column separation and rejoining events that could occur in the service water systems of nuclear power plants following a loss of ofTsite power without LOCA. Furthermore, the test apparatus used in the pressure surge tests contains an elevated condenser section that is typical of the condenser cooling water systems found in nuclear power plaats. This benchmark will exercise the TREMOLO capabilities for modeling of flow reversal, draindown from an elevated piping section, the growth and collapse of multiple voided
YS r lai vo ev r Le wse Lo R l m a er n h r t t em) salu2 m n 7 wWo( o C 8 1 _ a Dp E5o n 7 Q I l
'3 ,--
2 '
*~. -
- g'
- J'I d 7 i c l o
i P - V - m 2 sei on
~
i l bc t
'.- LI' ue ';-
TS )-
}
2r : ,' 1 e . '- s i f on F l m e ar n lad etem) cn r alu1 iop t yC pWCo ( s m Tn i U 7 8 1 d QYl 1 i 8 o 9-5 V 3 1 gj 2 E. 1
. ' R D
C - lei r 3 6 vo ;i O ev - d 8 Lr e 9 E hs T ge iR H s . - TE3 u9v. ea.2C:"ich . - ' 8* gRnoq q2 n8mnEEj p' .o a3y we6a y. NO $$ ' f R'5 Eo5mG, y,8. BEhE8 3$ow =E 5=5" o"EBR p:g35 d n$ao xm$.g9
f I 3-22 i regions, the behavior of two independent water columns, and the condensation-induced water hammer events that occur due to column rejoining and trapped void collapse. ; This benchmark focuses specifically on pressure surge Test 180. Details of this test are as follows. The test apparatus geometry is depicted in Figure 3.3-5. As shown, the piping consists of a high-level waar reservoir connected to a 18.7 m horizontal run of 0.09 m inner diameter piping. The ball valve used to simulate pump trip is located near the inlet of the horizontal piping mn. Next, a 5.6 m vertical riser leads to the simulated condenser section. The condenser consists of cylindrical inlet and outlet headers (0.2 m diameter) interconnected by 12 condenser tubes of 0.02 m inner diameter and 2 m length, running in parallel to each other. The outlet header is then connected through a 90-degree elbow to the return downcomer, which run. vertically downward to a 18.7 m long horizontal return pipe. The horizontal runs of supply and retum pipe are at the same elevation. The horizontal return pipe discharges into a low-level, water reservoir. The initial water temperature is 293 K, the steady state flow yields a velocity of 1.11 m/s the high-level reservoir pressure is 1.185 bar and the total pressure drop through the system at steady state conditions is 0.169 bar. To initiate the transient, the ball valve begins closing at 0.458 seconds and is fully closed at 1.158 seconds. l Note that the test apparatus is a closed piping system. Upon completion of the test, water is pumped from the low-level reservoir back to the high-level reservoir and any non-condensible gas accumulating in the condenser highpoints was reportedly vented prior to initiation of the next test run. Thus, the non-condensible gas content in the subcooled water is likely less than that
. expected in the open service water system found in many nuclear power plants. A key feature of the TREMOLO model is the treatment of residual gas bubbles in the fluid following void collapse. Following column rejoining, this small residual void will have an effect on pressure wave propagation speeds and subsequent waterhammer events. The residual void volume used in TREMOLO is an assumed value based on experimental evidence. Therefore, the appropriate value of the residual void for the closed cooling water circuit used for pressure surge test 180 will not be typical of that used for the Davis-Besse open cooling water system.
3-23 Results for pressure surge test 180 are presented in Figures 3.3-6 through 3.3-11. The figures compare TREMOLO calculations and test data for the upstream water column velocity, VA, the downstream wat:r column velocity, VB, upstream pressure, P1, and condenser outlet pressure, P7. Tne instrument locations for the VA, VB, P1, and P7 measurements are indicated in Figure 3.3-5. Figures 3.3-10 and 3.3-11 provide the TREMOLO-calculated axial void and Aow profiles to demonstrate the behavior of the two separate water columns and void regions. While the apparatus for Test 180 used a ball valve, which is characterized by a flow coefTicient that follows an s-shaped curve as a function of valve position, the TREMOLO Revision 1 valve model is designed to model a gate valve. The flow coefficient for a gate valve varies approximately as the square of the valve position. Thus, to approximate the Test 180 ball valve performance, the upstream boundary condition is modeled staning at 0.66 seconds by, , simultaneously ramping down the upstream boundary pressure and closing a motor-operated valve (located in the first fluid node) such that the P1 pressure profile and VA, through time of 1.158 seconds, is closely matched by TREMOLO. Beyond 1.158 seconds, the motor-operated valve is fully closed anc' the upstream boundary is isolated from the remainder of the test apparatus. Thus, for times greater than 1.158 seconds, the TREMOLO calculations of P1 and VA are based solely on the TREMOLO-calculated propagation of pressure waves through the single and two-phase ponions of the fluid. Although data for test 180 is provided out to 8 seconds, the benchmark exercise is terminated at 4.0 seconds. Beyond 4 seconds, the fluid is essentially stagnant and no significant condensation-induced waterhammer events are observed. In Figure 3.3-8, between 1.5 and 2.25 seconds, three significant pressure rises are observed at location P1, and two pressure rises are observed at location P7 as the result of three distinct waterhammer events. The first waterhammer event is caused by trapped void collapse near the closed ball valve as the water column supplying the condenser reverses direction due to gravity effects and impacts the closed valve. This is evidenced by the pressure rise at P1 and the sudden flow reversal exhibited at VA at 1.5 seconds.
3-24 i 1 l 1 l l j , 1 l J CONDENSER PRESSURE SURGE TEST 180 1.5 l - - - - VA DATA 1
~ !
3.o ' VA TREMOLO ) 0.5 a tG \
- I . l\lf\A)/bNif*
iV l i f.i
- l i
' j I
.l.0 .
l I I l
.t .5 0.0 0.5 1.0 1.5 20 2.5 3.0 3.5 4.0 l TIM E (SEC) l l
l l l l FIGURE 3.3-6 Test 180 Fluid Velocity Near Pipe Inlet (Instrument Position VA) 1 1
3-25 CONDENSER PRESSURE SURGE TEST 180 1.5
- - - - VB DATA i .o "~~~% , VB TRDAOLO:
I I
^
~
@ 0.5 -
E ' b e '. .Y ,v. 1 . y 0.0 , lI - -n ; ,:
.' l
'I ,
0.3 1.0 l\1:l - 00 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
; TIME (SEC)
FIGURE 3.3-7 Test 180 Fluid Velocity Near Low Level Reservoir (Instrument Position VB) r
r i 3-26 l i i I CONDENSER PRESSURE SURGE TEST 180 l 60 l
- - - - P1 DATA 5.0 I P1 TREMOLO l f.'
l 4.0 l !*
? I !i '., l 63.0 .
d,Q . l
,' .' l
- n. .'
d ,f ,A { I
.' ' A-2.0 - ., , .
1 , r, 4 1.0 "~~'3 '. N i I i
. f, '. [,' '. ,M 00 ] '
\' i' 0.0 0.5 10 15 20 2.5 3.0 3.5 4.0
; TIM E (SEC) l i
l i 1 1 1 I 1 i FIGURE 3.3-8 Test 180 Pressure Near the Closed Valve (Instrument Position P1) ! I l
3-27 CONDENSER PRESSURE SURGE TEST I80 3.5 , , ,
~~~~
3.0 ,-
-i i P7 TREMOLO 2.5
, l _
2.0 l l a :...dp . !.
- 1.5 ,
4 E I !
,O* l Thh 10 '
0.5
* ~ ~ ~ ' '
. l' r -
- s l .
't
..i. ..
i'.
**1 ' +', ,
- - - ^ ^ ^ ^ ' " "" '**
00 ~ l l l t j 45 0.0 0.5 10 15 20 2.5 3.0 3.5 40 l TIM E (SEC) FIGURE 3.3-9 Test 180 Condenser Outlet Pressure (Instmment Position P7) l 1 L.
1 l 3-28 l i i CONDENSER PRESSURE SURGE TEST 180 ) 0.10 0.09 - PROFILE AT 1.4 SECONDS 0 08 - 0.07 - o 3 0.06 h0.05 - l
.1!0 at a
>0.03 0 02 0.01
^J
- 0W l 0 5 10 15 20 25 30 15 40 45 50 55 Distance fmm Pipe Inlet (m) i FIGURE 3.3-10 Test 180 TREMOLO-Calculated Axial Void Profile Prior to Column Rejoining in the Condenser (Condenser section is located from 22 to 27 m from the pipe inlet) 1 l
p. l 3-29 l l 1 ( l I i CONDENSER PRESSURE SURGE TEST 180 i ! I I I
-TREMOLO UPSTREAM COLUMN N
)1.0 TREMOLO DOWNSTREAM COLUMN -
- - \
y' y gV j- [( s%"v [' es.l .0 N (}l. i
\/ ;l l
~
-1.5 O0 0.5 1.0 15 2.0 2.5 30 35 40 1 TIME (SEC)
FIGURE 3.3-11 Test 180 Water Column Velocities Calculated by TREMOLO (Upstream ; coh.mn refers to the water column between the closed valve and th. concenser; Downstream column refers to the water column between the condenser and the low-level reservoir) l 1 i l l .
[ 3-30 During the next two waterhammer events which both originate in the condenser between 1.75 and 2.25 sec, the supply side is essentially water solid from the valve up to the condenser (see Figure 3.3-10). The waterhammer events are initiated by column rejoining in the condenser region and propagation of the pressure waves through the upstream water column back to the closed valve. Figure 3.3-11 compares the TREMOLO-calculated velocities of the upstream and l downstream water columns. At the instant of the first ca'...ulated void co!! apse in the condenser I at 1.75 seconds, the water columns are approaching each other with calculated velocities at local l maximum values of +0.35 m/s in the upstream column and -0.77 m/s in the downstream column. The experimental data exhibits the same behavior starting slightly later in time at 1.8 seconds. Data from Test 180 indicate that the water columns are approaching each other at a lower relative velocity than that calculated by TREMOLO, hence it is no surprise that the resulting witemammer pressure rise calculated by TREMOLO at location P7 exceeds that observed in the test (see Figure 3.3-9). As the pressure waves originating in the condense region reach the closed ball valve, a pressure increase at P1 ocas. In Figure 3.3-8, the TREMOLO calculations indicate the second P1 pressure rise at 1.8 seconds while the Test 180 data indicates the p essure rise occurs at about 1.95 seconds. The second waterhammer event indicated by P1 is laiger in both the calculated and observed data than the first waterhammer pressure rise obsersed at Pl. This is expected since the upstream water column velocity, VA, at the time of the second waterhammer event is larger in magnitude than the velocity during the first waterhammer esent (see VA in Figure 3.3-6). During the second column rejoining event in the condenser, .alculated by TREMOLO at 2.0 seconds and observed experimentally at 2.1 seconds, the water columns are approaching each other at calculated and observed relative velocities of 0.25 and 0.20 m/s, respectively (upstream column velocity calculated! observed: +0.25/+0.20 m/s, downstream cohimn velocity calculated / observed: 010. m/s). This water column relative velocity is smaller than the relative velocity during the first column rejoining event in both the TREMOLO calculation and Test 180 data. However, the Test 180 data inc. stes a larger waterhammer pressure rise at P7 as j compared to the pressure rise resulting froi., the first column rejoining event, whereas the
)
i I
3-31 TREMOLO calculations show a smaller pressure rise at P7 during the second column rejoining. This is possibly due to the innuence of the residual void on the waterhammer eant. Additionally, the Test 180 data indicates a faster wave transmission time bctween P7 and P1 than . that calculated in TREMOLO, as inferred from the difference in timing of the P7 and P1 pressure rises. The faster wave transmission time and larger pressure rise in the Test 180 data both l indicate a smaller innuence of the voids during the second column rejoining event than that l credited in TREMOLO. l The closed test loop con 0guration and gas venting between test runs likely resulted in a j reduced dissolved gas concentration in the Guid used during the pressure surge tests and therefore less residual void during the test mns. However, such conditions are not typical of those found in the open service water systern design at Davis-Besse and conclusions regarding the residual void fraction required to model the pressure surge test should not be applied directly to the Davis-Besse plant analysis. Although the three signi6 cant condensation-induced waterhammer events discussed previously are present in the Test 180 data and the TREMOLO calculations, the experimentally observed timing of the events r.ad the resulting pressure rises are not precisely predicted by TREMOLO Revision 1. This is primarily the result of two factors. First, it is dif6 cult to match the upstream boundary condition because the in6uence of the high-level reservoir was controlled experimentally by closing a ball valve, while the TREMOLO Revision 1 valve model is designed to match gate valve closure. Second, the residual void model in TREMOLO Revision 1 is more appropriate for use in open cooling water systems, such as that used in the Davis-Besse cooling water system. Despite the differences between the actual test con 6guration and the TREMOLO Revision 1 modeling capabilities, the benchmark does demonstrate the ability of TREMOLO to model the essential aspects of cold water column separation postulated to occur during loss of power events at nuclear power plants Namely, this benchmark exercise indicates that the TREMOLO Revision 1 models can adequately model column separation and rejoining, Guid
1 3-32 ! i I I flow reversals, the behavior of multiple voided regions, the movement of multiple water columns, and the transmission of pressure waves through a combined singe and two phase fluid. l l i l < l 1
p i 4-2 in the waterhammer pressures compared to those that would be representative of an all water state. l Waterhammcr experiments related to the condensation induced waterhammer 3. events during the voiding phase associated with DBA conditions in the fan cooler circuit have been investigated. Results show that waterhammer events certainly occur, but the magnitude of the events are in the range of tens of psi with the peak pressure observed in any such experiments being 60 psi. The relatively low pressure of such events is due to a) increased compressibility of the mixture due to the small void content, b) the extensive thermal boundary layer that is promoted in the associated water phase during such voiding conditions and c) the extensive energy that is resident within the service pipe wall which also contributes to a thick thermal boundary layer in the water adjacent to the steam, bubble.
- 4. Experiments which have been focused on the waterhammer conditions during refill have clearly shown that those conditions with a refill rate producing a Froude number near or greater than unity produce a " plug flow" movement of water through the senice water piping. Consequently, there is minimal interfacial surface to promote condensation induced waterhammer. Thus, while small waterhammer events are measured, the dominant process is that the water " plug" pushes steam out of the service water piping and the resulting transient is benign.
When the water reaches a downstream restriction, a pressurization transient representative of a slowing water system would be expected. But again, this has a I strength that is representative of the system compressibility and is significantly influenced by the presence of noncondensible gases in the water. Here again, the refill associated with the service water pump restart conditions would be expected l to experience a local pressuriza: ion that is in the range of tens of psi above the steady-state pressure distribution as the system develops to the steady-state condition. I l . I
3 4-1 1 I 4.0
SUMMARY
As presented in this repon, a summary of experiences relevant to waterhammer issues considers the role of noncondensible gases exiting from solution, particularly as this may influence the compressibility of the water. Funh'ermore, considerable experiments have been performed relating to the strength of waterhammer transie sts during both the voiding and refill phase. With the background of the experimental data available in the literature and those experiments recently performed at Fauske &' Associates, the following conclusions can be drawn.
- 1. Open service water systems would always have a significant content of dissolved noncondensible gases. Specifically, the service water would be nearly saturated with air at I atm. For those conditions in which column separation andv steam generation as a result of energy transfer would occur, a significas.1 frcction of this noncondensible gas would be driven from solution and could be expected to be resident as individual gas bubbles in the water during any situation in which rapid condensation would be ofinterest, i.e. waterhammer transients.
- 2. Extensive data in the literature clearly indicates the substantial influcsce of small gas quantities voids (as well as the associated steam mass in wch bubbles) on the system compressibility. Specifically, gas void fractions of 0.005 cou!d decrease the mixture sonic velocity, from that of all water, by a factcr of two to ten depending upon the system pressure. Experimems performed wig room temperature water saturated with air at I atm show that such void fractions could be created by depressurizing water to about 3 psia. Hence, meaningful vcid fractions would be created for those situations generated by column separation.
Should the system also undergo energy transfer from the containment (DBA l conditions) additional gas could be driven from solution. For those regions in which the void fraction could be driven a state representing a cont?nuous gaseous media, i.e local void fractions created at approximately 50%, one would expect virtually all of the noncondensible gases to exit solution. Thus, one should expect to observe substantial changes in the mixture compressibility, therefore decreases L.
42 i I in the waterhammer pressures compared to those that would be representative of an all water state.
- 3. Waterhammer experiments related to the condensation induced waterhammer events during the voiding phase associated with DBA conditions in the fan cooler circuit have been investigated Results show that waterhammer events certainly occur, but the magnitude of the events are in the range of tens of psi with the peak-pressure observed in any such experiments being 60 psi. The relatively low pressure of such events is due to a) increased compressibility of the mixture due to the small void content, b) the extensive thermal boundary layer that is promoted in ti e associated water phase during such voiding conditions and c) the extensive energy that is resident within the service pipe wall which also contributes to a thick thermal boundary layer in the water adjacent to the steam bubble.
- 4. Experiments which have been focused on the waterhammer conditions during refill have clearly shown that those conditions with a refill rate producing a Froude number near or greater than unity produce a " plug flow" movement of water through the service water piping. Consequently, there is minimal interfacial surface to promote condensation induced waterhammer. Thus, while small waterhammer events are measured, the dominant process is that the water " plug" pushes steam out of the service water piping and the resulting transient is benign.
When the water reaches a downstream restriction, a pressurization transient I representative of a slowing water system would be expected. But again, this has a strength that is representative of the system ceapressibility and is significantly influenced by the presence of noncondensible gases in the water. Here again, the refill associated with the service water pump restart conditions would be expected to experience a local pressurization that is in the range of tens of psi above the steady-state pressure distribution as the system develops to the steady-state condition.
n 3 4-3
- 5. The fan cooler returns to the CBA heat removal capability essentiaily as fast as the fan cooler coils can be filled with water.
The TREMOLO Revision 1 benchmarking effort summarized in Section 3 has demonstrated the code's ability to analyze column separation and condensation-induced waterhammer events that could occur in the cooling water piping systems of nuclear power plants. In particular, the benchmarking activities demonstrated the code capability to adequately . model key phenomena encountered in transient waterhammer and two-phase flow analyses, such as,
- 1. Heat transfer across pipe walls and heat exchanger coils
- 2. Sonic velocity in single and two-phase mixtures
- 3. Column separation and rejoining .
- 4. Pressure wave transmission in combined single and two-phase fluids
- 5. Movement of multiple water columns
- 6. Void collapse in multiple voided regions )
- 7. Fluid flow reversal Furthermore, the benchmarking validated the TREMOLO approach of using a one-dimensional, five equation fluid model with assumptions of residual void to analyze the types of transient thermal hydraulic events postulated to occur in the service water piping systems of nuclear power plants.
The information presented in this repon provides the technical bases that relate RAI 2,3, 5 and 8. The TREMOLO benchmarks demonstrate the applicability ofits methodology and the conservatisms it provides. The discussion of the controlling physical phenomena provide
- insights that suppon the selection of the worst case scenarios for the waterhammer and two-phase flow analyses plus the parameters and their ranges for use in associated uncertainty and sensitivity analyses.
p 7 5-1
5.0 REFERENCES
l Bjorge, R. W., and Grif5th, P.,1984, " Initiation of Water Hammer in Horizontal and Nearly Horizontal Pipes Containing Steam and Subcooled Water", Transactions of the ASME, Journal of Heat Transfer, Volume 106, pp. 835-840. l Block, J. A., et al.,1977, "An Evaluation of PWR Steam Generation Waterhammer," NUREG-0291. Block, J. A.,1980, " Condensation-Driven Fluid Motions," International Journal of Multiphase Flow, Volume 6, pp. I13-129. Cerne, G, Tiselj, I, Petelin, S (Jozef Stefan Ints., Slovenia),1996, "Modeling of Water Hammer with Column Separation" Transactions of the American Nuclear Society, pp 387-388. Chou, Y. and Griffith, P.,1990, " Admiring Cold Water Into Steam Filled Pipes Without ; Waterhammer Due to Steam Bubble Collapse," Nuclear Engineering and Design,121, pp. 367-378. 1 Cruver, J. E.,1963, "Metastable Critical Flow of Steam-Water Mixtures," Ph.D. Thesis, ! Department of Chemical Engineering, University of Washington. ! l Davies, R. M. and Taylor, G. I.,1950, " Proceedings of the Royal Society", Volume 200, pp. 375-390. DeJong, V. J. and Firey, J. C.,1968, "Effect of Slip and Phase Change on Sound Velocity and Steam-Water Mixtures and the Relation to Critical Flow," I&EC Process Design and 1 Development, Volume 7, (July 1968) Henry, R. E., Fauske, H. K. and McComas, S. T.,1970b, "Two-Phase Critical Flow at Low Qualities Part II: Analysis," Nuclear Science and Engineering, Volume 41, pp. 92-98. Henry, R. E., Grolmes, M. A. and Fauske. H. K.,1971, " Pressure-Pulse Propagation in Two-Phase One- and Two-Component Mixtures," Argonne National Laboratory Repon, ANL-7792. Henry,R. E., Grolmes, M. A., and Fauske, H K.,1971, " Pressure-Pulse Propagation in two-Phase One- and Two-Component Mixtures," Argonne National Laboratory Report ANL-7792. Henry, R. E., Paik, C. Y., Hauser G M ,1997a," Dynamic Benchmarking of Simulation Codes," Transactions of the 1997 ANS Summer Meeting. l \
T 5-2 Henry, R. E., Paik, C. Y., Schlenger-Faber, B., Henry, C, E., McCanney, M. A., Fauske & Associates, Inc, and J. Chao, Electric Power Research Institute,1997b, " Dynamic l Benchmarking Program for the MAAP4 Code," NUREG/CP-0162 Proc. of the U. S. ' th Nuclear Regulatory Commission 25 Water Reactor Safety Information Meeting,1, Bethesda, MD (October 20- 22,1997). Hodgman, C. D., Weast, R. C. and Selby, S. M.,1958, Handbook of Chemistry and Physics: A Ready-Reference Book of Chemical and Physical Data,40th Edition, Chemical Rubber Publishing Company, Cleveland, Ohio. Holman, J. P,1981, Heat Transfer, Fifth Ed., p.116, equation 4-19, McGraw-Hill Book Company, New York. i Izenson, M. G., Rothe, P. H. and Wallis, G. B.,1988, " Diagnosis of Condensation-Induced Water Hammer", NUREG/CR-5220, Creare TM-1189, Volumes 1 and 2. Karplus, H. B.,1958, "The Velocity of Sound in a Liquid Containing Gas Bubbles," Illinois Institute of Technology Repon IIT Repon C00-248. Karplus, H. B.,1961, " Propagation of Pressure Waves in a Mixture of Water and Steam,"' Armour Research Foundation Repon ARF 4132-12. i Klingebiel, W. J.,1964, " Critical Flow Slip Ratios of Steam-Water Mixtures," Ph.D. Thesis, Department of Chemical Engineering, University of Washington. Rohsenow, W. M.,1973, " Boiling," Section 13 of Handbook of Heat Transfer, Edited by Rohsenow and Hannett, McGraw-Hill, New York.
\
Rothe, P. H., et al.,1977, "An Evaluation of PWR Steam Generator Waterhammer," CREARE Repon TN-251, NUREG-0291. l Safwat, H. H., Delft University of Technology, The Netherlands,1972, " Experimental Study of l t Pressure Surges in condenser Cooling Water Systems, " Proceedings of the l International Conference on Pressure Surges, Canterbury, England, pp. D2 D2-32 (September 6-8,1972). Semenov, N. I. and Kosterin, S.1,1964. "Results of Studying the Speed of Sound in Moving l Gas-Liquid Systems," Teploenergetika, Volume 11, pp. 46-51. Toledo Edison,1990, Davis-Besse Pulldown Data Based on Toledo Edison Letter NEO-90-0059, dated 5/30/90. USNRC,1996, Generic Letter 96-06. Assurance of Equipment Operability and Containment Integrity During Design Basis Accident Conditions (September 30,1996).
p 1 I S-3 l Wallis, G. B.,1969, One-Dimensional Two-Phase Flow, McGraw-Hill Book Company, New York. Wallis, G. B., Crowley, C. J. and Hagi,.Y.,1977, " Conditions for a Pipe to Run Full When Discharging Liquid Into a Space Filled With Gas", Transactions of the ASME, Journal of Fluids Engineering, Volume 99, pp. 405-413. Westinghouse,1969, WCAP-7336-L, " Topical Report, Reactor Containment Fan Cooler Cooling Test Coil, " Westinghouse Electric Corporation, Nuclear Energy Systems, Pittsburgh, Pennsylvania (July,1969). Wylie, E. B. and Streeter, V. L.,1978, Fluid Transients, McGraw Hill, New York, p. 39. t l I i o l l I}}