ML20073N004
| ML20073N004 | |
| Person / Time | |
|---|---|
| Site: | Crystal River  |
| Issue date: | 10/31/1982 |
| From: | BABCOCK & WILCOX CO. |
| To: | |
| Shared Package | |
| ML20073M968 | List: |
| References | |
| RTR-NUREG-0737, RTR-NUREG-737, TASK-2.F.2, TASK-TM 77-1137950, 77-1137950-00, NUDOCS 8304220355 | |
| Download: ML20073N004 (57) | |
Text
ATTACHMENT II i
o 8
i i
FEASIBILITY STUDY OF INVENTORY TRENDING f1ETHODS WITH RC PUMPS OPERATING OCTOBER 1982 Prepared by the Babcock & Wilcox Company for Duke Power Company Florida Power Corporation Toledo Edison Company Consumers Power Company Sacramento Municipal Utilities Division ,
B&W Document ,ID. 77-1137950-00 8304220355 830415 PDR ADOCK 05000302 P PDR
_ . _ _ _ _ . ~ __
l i
TABLE OF CONTENTS 1.0 . INTRODUCTION 1.1 Summary
1.2 Background
1.3 Report Organization 2.0 PROPOSED METHODS FOR MEASURING INVENTORY 2.1 Gentile AP 2.2 Pump Current 3.0 SUPPORTING DATA 4.0 COMPARISON OF THE GENTILE AP AND PUMP CURE 6'iT METHODS 5.0 POTENTIAL USE OF INVENTORY MEASUREMENT TECHNIQUES IN l PUMP TRIP CIRCUITRY
6.0 CONCLUSION
S AND RECOMMENDATIONS APPENDIX
i
- s LIST OF SYMBOLS
! SYMBOL DEFINITION f
A Ama g gravitation constant e
H pump head I current K flow coefficient AP pressure differential PF power factor P m tor power m
P pump power p
Q volumetric capacity T pump torque 9 velocity
, V voltage y specific volme W mass flow rate x quality Y net expansion factor 4 a void coefficient a normalized pump speed n
S nonnalized pump torque ny pump impeller efficiency motor efficiency g
l n overall pump efficiency p
i n p ep speed p density I vq normalized pump flow Subscripts or Superscripts G gas phase l
L liquid phase l
TP two-phase l
, f saturated liquid i
g saturated gas
! r.o reference or calibrated state l
I
1.0 INTRODUCTION
1.1 Sumary
~
f Subsequent to the accident at TMI-2, the NRC required a review of instru-mentation to indicate the approach to and onset of inadequate core cooling.
As a result, the owners of B&W nuclear power plants reviewed level measure-ment systems which would operate when the reactor coolant pumps were off.
During industry meetings with the NRC staff in early 1982, the staff indicated that an indication of system inventory with reactor coolant pumps on was desired. This report provides the results of an investigation which demonstrate that either of two methods, a Gentile flow tube AP system or an RC pump current system using presently installed equipment as a base can provide satisfactory inventory trending information with the reactor coolant pumps operating.
1.2 Background
Following the TMI-2 accident, the capability to monitor the primary system water inventory was identified as a potentially useful accident management tool. Review of TMI-2 data and subsequent small break loss of coolant accident (SB LOCA) analyses1 ,2 has revealed that continuous operation of the RC pumps during the transient resulted in a highly voided primary system. When the RC pumps were tripped in this condition, l
I the liquid that was previously dispersed throughout the system via pumping action, collapsed to the low points of the primary system, such as the bottom of the reactor vessel and steam generators. Consequently, the low water inventory at the time of the pump trip could result in an insufficient level for adequate core cooling.
In the fall of 1979, the NRC made a generic assessment of delayed RC pump trips during a SB LOCA.3 They concluded that due to the uncertainties involved in SB LOCA analysis the prudent course of action would be to trip the RC pumps immediately following an indication that a LOCA had occurred (RC pressure dropping below the HPI setpoint). It was also recognized that the imediate pump trip approach was less than optimum.
i <
l For example, in overcooling events (i.e., steam line break) which cause shrinkage of the primary system and loss of RC. pressure, early RC pump trip (and subsequent loss of pressurizer spray) can aggravate these transients and extend the time required to bring the plant into a
( controlled shutdown conditon. However, since these transients did not lead to unacceptable consequences, early pump trip was adopted as a
- course of action.
E Current procedures in existence on operating plants therefore call for manual trip of reactor coolant pumps when the HPI setpoint is reached.
Even though the operators are now trained to trip the pumps, the NRC has requested the additional capability to trend inventory with the pumps operating. This report evaluates potential systems which could provide the trending information required.
1.3 Report Organization In the next section, two methods for measuring primary system inventory which relate measurable physical quantities to RC voiding are presented.
Since both methods rely on various assumptions concerning the performance of the RC system, Section 3 of this report is presented to substantiate these assumptions. This supporting data was obtained from the TMI-2 accident and various experiments conducted in the last ten years.
After review of the supporting data, one preferred model is identified for each of the measurement methods. In Section 4, these models are compared such that the advantages and limitations of each measurement can be assessed. The potential use of these methods in the pump trip circuitry is discussed in Section 5 and the conclusions and reconrnendations are presented in Section 6.
l
i h~
2.0 PROPOSED !!ETHODS FOR MEASURING INVENTORY The mathematical representations of the relationship between RC voiding to Gentile AP and pump current are presented in this section of the report. Justification for the assumptions used to derive i
these models will be addressed in the next section, when the supporting data is reviewed. For each method, two mathematical models are developed to allow for a comparative description of tne phenomena.
t 2.1 Gentile aP A relationship has been shown between RC flow as measured by the Gentile flow tubes and system voiding for the T111-2 accident.
The data is presented in Figure 1. Visual inspection suggests that a linear relationship between system voiding and mass flow rate may exist. Upon further examination it was noted that the relationship between flow and density can be represented by:
W = OVA = oO where W = measured mass flow rate (lb/sec) o = density (1b/ft 3) 9 = fluid velocity (ft/sec) 2 A = cross section area (ft )
3 Q = volumetric flow rate (ft /sec)
If one assumes that the volumetric capacity of the RC pumps is not degraded in any way (i.e., Q is constant), a simple one-to-one I
relationship then exists between flow and density. Specifically, when comparing to a reference (subscript r) or calibrated condition:
W , o (1)
W r 'r
l .
Equation 1 indicates that mass flow rate varies linearly with respect
- - to density. Further, note that the thermodynamic representation i
for void can be expressed in tems of density as:
?
p - of
,, (2) og - of where p = two-phase density (lb/ft 3) 3 of = density of the saturated liquid (lb/ft )
og = density of the saturated gas (lb/ft3 )
a = void coefficient Then, combining equation 1.and 2 will yield a final relationship between the measured flow rate as specified by the Gentile and the primary system voiding (a).
(WW )Pr
-S f
r (3) og - of At a given saturation pressure, a linear relationship does exist between measured mass flow rate and system voiding.
l l Equation 3 represents one mathematical model for the phenomena t
! seen in Figure 1. RC measured flow versus RC system voiding. However, this simplistic model fails to account for any variation in Gentile performance when two-phase conditions exist. In particular, will the Gentile function in the same manner in both single phase and two-phase flow conditions (i.e., will flow rate be proportional to the square root of the generated AP across the Gentile) . Review of numerous papers concerning the measurement of various two-phase fluids L
i '
I, in venturis and orifices have revealed that this is not the case, i.e., aP metering devices have distinct two-phase characteristics which differ from those found during single phase operation.
~
The theory of two-phase flow characteristics through orifices, as
, developed by J. W. iiurdock, will provide the second model upon which the flow-density correlation can be explored. This theory was chosen
, because 1) its development is based on thermodynamic and hydrodynamic first principles, and 2) Hurdock's work is referenced extensively in two-phase literature and thus forms somewhat of a standard. For the sake of completeness, an outline of his model will be repeated here.
The flow rate through a AP metering device can be expressed by:
W = AKY/ 2ggapp (4) where W = mass flow rate (lb/sec) 2 A = cross section area (ft )
K = flow coefficient which includes the velocity of approach factor Y = net expansion factor 2
g = gravitational constant (32.174 lb-ft/lb -sec f
)
AP = pressure differential across device (lb f
/ft2 )
3 o = fluid density (lb/ft )
Noting that Y = 1.0 for the incompressible liquid phase, a set of equations representing all phase conditions can be written.
~
l Single Wg = AKg/29APL c 8L (5)
}. Pnase Wg = @ g g2gcdg pG (6) f
. Two-Phase W L
=A g(K)TPf9aP L TP8 L (7) c Wg=Ag (K Y IgGTPy2g'TP c 0G (8)
Equation 5 and 6 are representative foms of equation 4 for the single phase flow of liquid and sbeam, respectively'. Equations 7 and 8
$ govern during two-phase flow conditions when both components are present in the mixture. The new variables are defined as follows:
Wg = liquid phase flow rate (lb/sec)
Wg = gas phase flow rate (lb/sec) 2
( = area which the liquid phase occupies (ft )
2 Ag = area which the gas phase occupies (ft )
Kg , Kg , (Kg)TP' IKG)TP = flow coefficient for liquid, gas, liquid in two-phase, and gas in L two-phase conditions, respectively.
Yg ,(Yg)TP = net expansion coefficient for the gas in single and two-phase conditions, respectively.
APg, APg, aPTP = pressure differential for liquid, gas, and two-ohase conditions, respectively.
(lb/ft) f 2
og' og = density)of (lb/ft liquid and gas phases , respectively
(
By combining equations 5-8 and representing the ratio of component
, flows by quality (x), Murdock derived a single expression for flow as l
l a function of the two-phase AP measurement.
l (Kgg Y )TP AG2gAPTP e G
( w- (9) gg x+ (1-x) (K Y )TP [#G
. (()TP L I
l Murdock conducted many experiments with various orifices and fluids The g
and concluded that Kg /(Kg)TP = 1.26 and Kgg Y /(Kgg Y )TP = 1.0.
physical significance of these coefficients is important. Since the gas flow coefficient ratio is unity, the presence of the liquid
, phase has no effect on the ability of the orifice to pass the gas phase through the available area, Ag . However, the liquid flow
- coefficient ratio is 1.26, signifying that the presence of the gas i
phase will effectively reduce the liquid flow by s 20% of the value expected (provided aP and o remain fixed).
TP L To apply the theory presented above to the problem of correlating RCS flow to voiding, the following assumptions were made:
- 1. The theory is applicable to any aP flow meter.
The literature revealed that this theory can be applied to orifices or venturis. Gentile flow tubes, orifices, and venturis am all AP metering devices and, therefore, the generic theory 1
presented by Murdock should apply equally for all of these devices.
- 2. The gas and liquid flow coefficient ratios are both unity.
The coefficients found by Murdock where unity and 1.26 for the gas and liquid pnase, respectively; however, the water -
steam data covered the range of 80% < x <100%. The range of interest when measuring RCS voiding to detennine pump trip
~
I L
is 0%<x<5% (0% <a< 70%). Therefore, a literature search was conducted to find realistic flow coefficient ratios in the low
. quality regime. An extensive study conducted by the Atomic f
- Energy of Canade Limited (AECL)5 was found to provide
, such data. Tnis data was obtained for the low quality regime
$ (5%< x <50%) at pressures of interest (1200 psig). The results are as follows: gy Device Diameter Ratio (8) (Kgg Y )TP (K)TPL Orifice 0.45 1.0372 1.0789 Orifice 0.70 1.0818 0.9999 Venturi 0.58 1.0427 1.0779 Review of equation 9 reveals that (K )TP L must approach Kg as quality tends to zero. Considering this boundary condition and the AECL data presented above, a valid estimate for the gas and liquid flow coefficient ratios is unity. This approach will avoid the discontinuity at the saturation line.
l
\ .
- 3. The flow coefficient of the Gentile will be the same for both phases.
The flow coefficients of the gas and liquid phase should be approximately equal for the Sentile since impulse nozzles are used to measure the resulting AP. These nozzles measure fluid kinetic energy and are not sensitive to the phase of fluid being measured.
t l-- - - _ -
6
,6 At sufficiently high Reynolds numbers, tests have shown that the Gentile displays a constant flow coefficient for liquid flow.
However, this value diminishes slightly at low Reynolds numbers.
Consequently, one would expect that the gas flow coefficient will degrade somewhat in the low quality regime at corresponding low Reynolds numbers. Murdocks data reveals that this effect is not large and, therefore, it will be assumed that equal flow coefficients exist.
k
- 4. The net expansion factor in unity.
i
. The range of steam velocities expected are completely subsonic and hence, no compressibility effects must be included. Consr wntly, the next expansion factor can be set to unity.
. If the four assumptions listed above are combined with the
- assumption of constant flow capacity, (as used in the develop-ment of the first model), the development of a second correlation l i is as follows
i
= (as presented earlier in equation 1) r r I
i Since equation 9 provides an expression for the two-phase j flow (W) it can be inserted into the relation above yielding.
A 2g aP e TP 8G (Kgg Y )TP 8 r (STP/#r) (
(1-x)(K gg Y )TP G 8
x + k )TP L
l Applying the simplifying assumptions concerning the flow coefficients and the net expansion factor will pemit reduction i
of the left-hand side of equation 10. . At this time, it should be emphasized that the liquid and gas components are actually saturated water existing in two distinct phases. Therefore, I
the standard subscripts f and g for the liquid and vapor phase properties can be used to replace the L and G subscripts used previously. -
Equation 10 can now be rewritten:
G 2g aP c TPP g
= W r I'TP'r)
I (II) x+(1-x)/c/of g In the above equation, a single flow coefficient, K, is used since it was assumed that all flow coefficients are identical.
The two-phase density on the right-hand side of equation 11 can be expressed as:
TP " IIVTP " IVf II"*)
- V gX)-I (12) also, at the time of Gentile calibration:
W (13) r =KA]29aP/v c r Substitution of equation 12 and 13 into equation 11 yields an expression for quality as a function of two-phase Gentile AP.
- I Y
, r (14) x + (1-x) Vf (I'*)
- Vg*
i I
i
- Or after expanding and collecting terms
s v v
V faPTP Ivfvd V
g g Q APr r
- Assuning no interfacial slip:
a= I (16) 1+v f (b)
Tg Equations 15 anc 16 form the relationship governing the measured AP of the Gentile and RCS voiding. The required inputs are:
- 1. RCS pressure to determine v and v .
f g
- 3. The AP measurement from the Gentile when it is calibrated using the secondary side heat balance (AP g input).
- 4. The specific volume detennined at the time of Gentile calibration (vr.)*
1 2.2 PUMP CURRENT j Two mathematical models can be developed which relate pump current lto RCS voiding. Each model is fonned using an expression for the energy
[ transference fmm the pump to the coolant.
i The input three-phase powr, P,, required to ' drive a constant speed (constant i frequency) squirrel cage induction motor can be expressed as follows:
~
, P, = [IV (PF) (17)
. where [ = accounts for the 3-phase input of power I = line current (RMS Amps)
V = line voltage (RMS volts)
PF = power factor (accounts for energy lost in setting up the magnetic field)
Similarly, the power, P required to drive a pump can be expressed by considering p
the development of head:
p . o0H p (18) n p
where p = fluid density (lb/ft )3 Q = volumetric flow (ft 373,e)
= head generated by the pump (ft) l H
1 n
p = overall pump efficiency (accounts for the mechanical friction at the seals and bearings, hydraulic friction at the impeller vanes and the diffuser vanes, and various other hydraulic losses due to addy fonnation.
l l
l l
. _. . _ _ - . ~ . _ . _ _ . _ _ - _ . _ _ _ _ _ , . _ _ . . _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _
e .
I ,
If one accounts for the windage and mechanical friction losses of the motor q by considering the motor efficiency, n,, the transfer of power to the pump can be stated.
i U
m P, = P, t
or, inserting the previous definitions, n,[IV(PF)=e0H q (19)
- p The change in current relative to varing fluid density can be addressed by denoting a reference point (0 superscript) condition.
' ' 0O
= ' " IV (PF) Q H (20) p np o n,o I V (PF)0 QH Assuming that the motor efficiency, voltage, power factor, capacity, and head remain constant, a simple relationship results.
o " (h) ( o) (21)
Inserting an expression relating void to density.
o = of - a (of - og) will yield the desired relationship between voiding and current.
l
~
i
- l. -
ef - * ( ir ) ("% ) (22)
P f ~ *g i 1 1
A second model can be developed by considering the transference of torque within the pump. In this situation,
- Pp = Ta (23)
"I whem T = hydraulic torque transferred from the impeller to the fluid (ft-lbf) n = pisnp speed (rpm)
'I= pump efficiency accounting for the mechanical friction at the seals and bearings, hydraulic friction and eddy lesses within the impellar.
! Hydraulic friction and eddy losses within the diffuser vanes are not included since only power l
transfer at the impeller is being considered.
l a
.q' P
t .
Coupling equation 23 with the homologous or normalized relation for torque,
" 8 ( #/#R)
~
T i
yields, Pp = STR U/"I'R Where S = norrhalized torque T = rated torque (ft-lbf)
R 3
pR= density corresponding to TR (lb/ft )
Assuming that the motor efficiency, voltage, power factor, and operating point of the pump (6,n) remain constant, a nonnalized relationship results.
c = ) ( o) (24 As before, I
f ~ *o ( T ) 'IS0
)
(25) a=
SfPg This completes the development of the mathematical models. The four relationships involving gentille AP and void fraction, and pump current and void fraction are shown in Table 1.
e
e l.
TABLE 1: VOIDING MODELS o
r
~O AP" f Model 1: a= (26)
- g Of Model 2: ,
V a K TP V
f v7
- [f/ V g
1- V f/ y + V r aP p ( yf- y ) (27)
- g 'P r v"
! -1
- f a= 1 + 79 (1-x) x (28) l 1
i l
q, .
[
Pump Current I
i af - S (I/I ) (n p/n po) p%j 3, ,
(29) l of Og I
,I I
of - 0 (I/I') (nyo ) (30) ibdel 2: ,,
OfOg l
h i
l l
l l__
i In sumary, these models were based on the following assumotions:
Gentil e AP, Model 1 f
l 1. The volumetric capacity of the pumps will not degrade.
I
- 2. The performance of the Gentile flow tube is the same in both single and two-phase flow.
' Gentile AP, Model 2 I 1. The volumetric capacity of the pumps will not degrade.
- 2. The flow coefficient of the Gentile will be the same for both phases.
Pu:ap Current, Model 1
- 1. The motor efficiency, voltage, and power factor remain constant.
- 2. The pump capacity and head do not degrade.
Pump Current, Model 2
- 1. The motor efficiency, voltage, and power factor remain constant.
- 2. Pump speed remains constant.
- 3. The hydraulic torque does not degrade.
! The validity of these assumptions will be discussed further in Section 3.0 when the supporting data is presented.
F
" ' ' ' --^
__ . . . ___ - . . . . . , - _ ~ _ _ _ __.
l.
~
FIGURE 1
% RC FLOW VS % VOIDS DURING THI-2 ACCIDENT .
I 1
i I
i t
ON
\
O N.
N 80, N -
N O
N.
'O 60-3 \
B \
d % ~ ~ '
g 40 7
=
20- ' .
5 20 40 60 80 100 VOIDS, % FR0:1 ECCS ANALYSIS
l..
I 3.0 SUPPORTING DATA To utilize the models developed in the last section, supporting data must i
l be found to substantiate the assumptions used to develop the models.
Consequently, a literature survey was conducted. This section of the report sumarizes the findings of this study.
I f
Review of the assumptions tabulated at the end of Section 2 reveals
, that supporting data is needed in three major areas. The first area
. concerns the questionable performance of the gentille under two-phase conditions. Specifically, does the theory presented by Murdock hold true or will a simpler approach suffice? If Murdock's theory does hold, will the flow coefficients of the liquid and gas phase be equal, or will slip between the phases produce erroneous results, etc.? Secondly, the characteristics of the pump drive motor must be examined. The variation in efficiency, voltage, speed, and power factor must be explored. In the last area, the perfomance of the pump under two phase conditions must be examined. This is a crucial topic since all of the models presented thus far rely on non-degraded pump per-fomance (head, capacity, and torque).
In the review of Gentile literature, sparse information was found concerning single phase flow perfomance,and no information was found concerning two phase flow through Gentiles. In fact, the data ob-
,- tained during ?.he TMI-2 accident may represent the only two-phase perfomance infomation available for Gentiles. Therefore, the two models were applied directly to the TMI conditions and the results were compared with an estimate of system voiding made by ECCS analysis (Table 2 and Figure 2 ). In the ECCS compu'tation, the CRAFT 2 computer l- . ,. ,- .- -,
l .
code was used with best-estimate inputs regarding makeup and letdown flows, leakage flows, and heat transmission throughout the RCS.
I i
Each heading of Table 2 should be explained:
l RCS Pressure - Value taken from ECCS analysis. (Since voiding is pressure dependent, the CRAFT 2 value must be used tomaintainconsistency.)
Measured flow rate - flow rate measured at TMI-2 via the Gentile
, AP Gentile - calculated using typical conversion formula:
LOOP flow (MPPH) = 13.68 [aP(psid) where MPPH = million pounds per hour aj - voiding as estimated by Gentile - model 1
- voiding as estimated by Gentile - model 2 a2
~
By examining Figure 2, the following can be concluded:
- 1. The fluctuating Gentile signal produces a false indication of voiding during the first 10 minutes of the transient.
This is brought about by a combination of reasons:
i) accuracy of the Gentile ii) speed cycling of the RC pumps due to asymetric f
void fortnation within the loops f
l . iii) abrupt addition of auxiliary feedwater at 8 minutes l into the transient producing shrinkage of the primary system.
l
~
- 2. Model 1 for the Gentil'e produces diverging results when compared to the ECCS analysis.
t'
- 3. Model 2 for the Gentile agrees favorably with the ECCS prediction when the transient settles out at 25-30 minutes.
1
i .
\.
This model predicts voiding to a lesser extent than the CRAFT 2 estimate, which may be the result of interfacial slip between the phases or a d slightly degraded flow coefficient for the gas phase as explained earlier. In either case, the flow is over stated and hence the void-l ing estimate is low.
To summarize, the second Gentile model will relate Gentile AP to system voiding when the RCS is in a quiescent two-phase mode with pumps running. The first model does not account for any two-phase effects and, therefore, produces erroneous results. Consequently, this model will not be considered further in this study.
Figure 3 displays the operating characteristics of a typical pump drive motor. Considering the density variation associated with voiding to be on the ortler of 0 to 40'., the operating range of interest lies between 6000 and 10000 HP. In this range, when a constant vol-tage is supplied (6600 volts in this case), motor efficiency and power factor vary by less than M. As a result, current varies linearly with power in this range as shown in figure 3.
i l
l The variations in pump speed are small as load changes. Typically, the induction motor c;; gmrate 1000 HP for every rpm off synchronous.
Therefore, the ptrp ;;y . ar; be assumed w be essentially constant l
when compared to tne 1200 rpm synchronous motor speed.
The assumptions of constant motor efficiency and power factor imply constant supply voltage. It is recognized t?.at off normal voltages may occur due to bus transfers, pump starts or grid disturbances.
These perturbations are expected to be of short duration and would
L not have a sign 1ficant impact on the use of pump current for inventory I
trending indication. Application of this measure' ment for alarm
! or pump trip circuits would require provisions for delaying the alarm or i
trip signal to avoid spurious actuations.
Operating characteristics of the pump drive motor in the range of I interest can be sumarized as follows:
- 1. The motor efficiency remains constant.
- 2. The power factor remains constant.
- 3. The pump speed remains constant.
Experimental data has been generated concerning the performance of the pumps under two-phase conditions. These experimental studies are sumarized below.
Combustion Engineering, Inc./EPRI In an effort to refine the analytical model of the reactor coolant j pumps under hypothetical large break LOCA conditions, Combustion l
Engineering constructed a test system which utilized a 1/5 scale pump. Steady state tests were conducted near the operating point i
l (vn/"n = 1.0) and the results are presented in Figures 4 and 5.
(Note vn is defined as the ratio of the actual flow rate to the r rated flow rate and an is the ratio of the actual speed to rated l
speed.) For low void fraction (0-0.1), the pump head (Figure 4) remained at, or slightly above the water or non-degraded value.
At 0.15 void, the head degrades almost linearly until the worst case (20% of the water value) is reached at .75 void.
i .
Head then recovers as the single phase steam region is reached.
A similar behavior is seen concerning torque, (see Figure 5),
I but the degradation is less severe. The degradation in head j and torque at low pressures is even more pronounced due to the i
larger density differences between the steam and the liquid.
Losses associated with these two-phase mechanisms are three l
times greater at 500 psia than at 1000 psia.
! Babcock & Wilcox Company /EPRI
, With the same program objectives, B&W constructed a test apparatus i
to analyze the perfonnance of a 1/3 scale pump using air / water mixtures to simulate voiding. The results of these experiments are presented in Figures 6 and 7. Restricting our attention to the re-gion around the operating point (v n/*n = 1.0), the same degradation behavior is seen relative to the Combustion Engineering data. However, the degradation effects are exaggerated due to two-phase losses re-sulting from the large density difference between air and water.
, CREARE Inc./EPRI CREARE Inc. worked in parallel with the two studies mentioned above.
In their test rig, a 1/20 scale pump was installed to address the effects of scaling. Figures 8 and 9 show the results of the tests conducted near the operating point for low pressure water / air mix-I i tures. These results compare quite favorably with the B&W data presented earlier.
LOFT Research As part of the Loss-of-Fluid Test (LOFT) Program, RC pump motor power and current measurements, and their utiTity as indicators of RCS
f
- k.
inventory is being explored. Current research shows a large head degradation at approximately 20% void (See Figure 10). This I
phenomena is consistent with the other data presented thus far for
. scaled pumps. It is interesting to note, that power and current i
measurements made during this transient (See Figures 11 and 12,
- respectively) do not exhibit the same discontinuity.
The following can be sumarized concerning two-phase degradation of pumps:
- 1. In all cases, torque did not degrade as appreciably as develop-ed head. This implies that larger losses occur in the diffuser section of the pump than in the impeller section. The LOFT data presented in Figures 10 through 12 shows that this is the case, with the impeller efficiently imparting hydraulic torque to a two-phase fluid at a homogeneous density, while the head abruptly degrades due to two-phase losses in the diffuser sec-tion.
- 2. All degradation effects are minimized at pressures above 1000 psia. This is important since the application of the void measurement will usually be made at elevated pressures (> 1000 psia).
- 3. Review of the RCS flow for the TMI-2 event reveals that little if any pump capacity degradation occurred. Therefore, it must be concluded that scaling effects have not been adequately addressed by the pump testing sumarized in this study. The probable cause for this discrepancy deals with the inability of these tests to scale the bubbles such that the relationship between the size of the bubbles with respect to the vane spacing on the impeller and diffuser is preserved. For example, if ping-pong balls were
I .
suspended in a fluid system, their presence would choke a small pump, whereas a large pump would pass the balls virtually unde-tected.
4
' 4. In developing the technical approach for the pump current meas-urement, model 2 (based on torque transfer) is a stronger approach i
since only impeller dynamics are involved. Model 1 (based on f
head and capacity) would involve resolving the energy losses in
' the pump casing, thus making this approach less desirable. There-
- fore, model 1 will not be considered further.
The review of the data presented in this section has resulted in the elimination of two of the models developed in Section 2.0. The re-maining two models will be discussed further and compared in the fourth section of this report.
l - TABLE 2 ESTIMATES OF TMI-2 SYSTE!1 VOIDING f
t
! Tine RCS Pressure Measured W aP Gentile a a I
(min) (psia) (1b/hrx106 ) (psid) (Vo d) (Vbid) 2186.55 67.4 24.274 0.000 0.000
! 0 l'
1 1731.93 66.2 23.418 0.000 0. 0 01 2 1638.45 65.8 23.136 0.008 0.022 3 1566.36 65.8 23.136 0.022 0.030 4 1503.86 65.6 22.995 0.034 0.042 5 1455.83 65.6 22.995 0.043 0.047 6 1498.91 64.9 22.507 0. 04 9 0.057 7 1560.04 64.0 21.887 0. 051 0.075 8 1616.37 63.5 21.546 0.051 0.074 9 1587.78 65.4 22.855 0.024 0.036 10 1548.81 66.4 23.559 0.012 0.020 15 1374.95 66.4 23.559 0.043 0.034 20 1206.94 66.0 23.276 0.080 0.063 25 1105.19 60.3 19.430 0.176 0.179 30 1104.37 59.5 18.917 0.187 0.1 94 35 1103.30 58.0 17.976 0.209 0.222 40- 1103.57 56.0 16.757 0.239 0.259 45 1104.22 54.0 15.582 0.268 0.297 50 1104.89 53.0 15.010 0.283 0.315 55 1104.46 50.0 13.359 0.326 0. 371 60 1075.98 58.0 12.311 0.358 0.409 65 1074.44 46.0 11.307 0.387 0.446 70 1067.31 44.5 10.582 0.412 0.473 i
J i P AR I 50I4 FIGURE 2:
} v'j ' 2 \/ 0 I D
...s..
l.
1 I -.;=m cu.1,
- n =a' p : n-i _s.
t t
,0 W cer c :: c e L s, L s 6. v ie
^
y,.,-.. .
y .> : C .. L
? J y.
.. s-
,e
, .a l
/
um.
i .Jat,. -
f-Z . 4 f i
/
~
,/
~
Z
/
,/
i :e
.a, a>. err . ,/
,./
- u. I .. c
~
/
- f. ,
./ t-a
. X
- - . 4,,.,,,
,/z. M
/ M
/A C f' M
~ /
T.
~
l e /
/ M c.
sJ
- - -n
_;; 7 -
. /. a
,s
[ M M
/ 5
/Z M a .e, e ,/ M
%a
/E
/
F
/
- g. ,, . e -,
M
" l -
- , c.
/
/ Mr 4-/
th e-p -
.9
' e: u.s av. ...
~6a Ja.
?. ,. s..
.., . a. a- .,. r..--
- ) ..
!MC M ! h.1 e te.L *4!. tj
- 4.mm mi n.
e i, ,
, . . , Is - . i ..,ri
\.J . L.) 1 i m Ji1- 1.\1 _, ;iN w .
FIGURE 3: MOTOR CHARACTERIST!CS
< 800 I
I 700 g .. Efficiency
- 92 o
/ -
Power 600 Factor u 90 -
B U
O 88 < -
L -
,500 3 ^
e 86
,, curren g a
?* O 84 "
- 400 E w- E
- , u o 82 " 3 u
C
[ 80 " ~
300 E.
LJ **
'u 7S E
76 -
200 l
l t
l
- *~
100 i
l r l
l 0 2000 4000 6000 8000 10050 120d0 l
( Load (HP) l
,w_,_ _
~
0
<- :, ;'s/ - - - - 1
~ ~ - e l
/. ,
/, / '
/ ,
y 9
!. p /
I i
0 I
0 0 1
1 3
1 O
1 0
l O
-4,fI -
8 0
F I
R d
S e
e p
O T T O d T n 0 R 0 M. 9 _
. E a 9 0 0 I 7 P w
0
-~ x 0 5 o I
- ~
- R ~ 0 l N N N O 0 F v a a F E
- /
v g
T N N"
E 'N
~ N E
G d
t e
I
/
a N
y G N E
L S Tv yJ. I 6M 0
D 0
R a
r
- lNl l
i a v G JI e
E l A DS N ,
N N
PXI OT
.O ' 5 a p i t
o a
ARA I
I I
0 , R P
05 61 0
0 0 5 5 0
5 N P V O AD I
O N N 0
l d
a e
- - * * + - i l D C l A A A A E S 'S o A s I
I S
P P P I
S S I I S
P A
B o ,N I 4 R 0 F D
u o
g 0 0 0 0 Y I o 0 0 5 8 - O l o
2 1
0 8 4 V m 1
o O oaX N I 3 M
A E
l l
n I
0 R o
-. O
\' T S n P o
- N .
- U i
. ' s6 f6 2
t c
a I _ lo I 0 F r
d i
. o V
f o
Y.
1 v
1 I 0 t c
e f
- f U E
~ - _ - - - - - - 0 2 0 9 8 7 6 5 4 3 2 0
1 O
.0 1 0 4 0 0 0 0 0 0 -
1 1 0 e
r u
5" CMQ s5 { _ZN -
5 i F
g SE= gW go o
I I I I I I I I I 6 1.5 -
O 1200 PSI A l IYO V/"N = 1.0 O 1000 PSI A 150 v N- 0.W TO 1.10 I3 -
-- -- 6 850 PSI A 350 a 0.M 10 1 03 N
1.2 - --
X 480 PSI A !50 vga 0.90 10 1.07 -
1.1 -
O 1.tkx21- -
SC 8 - O '
{$ 0.9 - .
M5 '\'0 ./
o n. 0.8 - N ' / #
EC y \ .i C y 0.7 -
y Yn O O e py'/ _
o g O / O ,
O O.6 -
'd *f./ g[ -
g 0 9 0.5 -
yM ------ .V -- --
- -- -+e -('_ / , -
I<
gz 0.4 -
o
~
~
o
'S APPROXIMATI ADJUSTMNT IOR DEVIAll0NS IN y/aN -
0.2 -
0.1 -
l i I I I I I I I 0
0.2 0.3 0.4 0.5 0.6 0.7 0. 8 0.9 1.0 O 0.1 t'PSTRIAMVOID IRACTION, aF , HONDGENI0llS PER ORIF.
Figure 5: Effect of Vold Fraction on Homologous Torque Ratio Near Rated Flow and Speed
FIGURE 6: Homologous Head at Various Pump Inlet Void Frac-tions for First-Quadrant Pump Operation s
16 i
1a
\
I 's A .
l l-m.'w% wose Fracuens i
'%_'g 12b %'% N N ' = = - _ - .(03) l
. l 3
% f
- s _ _ _ _. _ _ _ o 10 ) ;
- "# = = - _ ' ' == 4 t 10 15) l
'- (0 3) l
",- (3 10) (10 20) I 0e* e",- l (15 20)
! 06 -
~,z j 5
= C4L :
g ,
== ~ ~ (20 30)
(20 30)
[z o==~~
C3 -
{
~
i30-40, (40 50)
< 30.s0 ) 50 70)
= - - _ . --==s 5 m -60s I
C C.5 09 LO
~
L1 C2 03 04 6C 70)
$ a50-60>
o 860 M) \
$ .o y .
- < =. 0 )
.c a .
3 06 .
(40 50)
(30 a0) ce -
(3 10) 10 .
t20 30) f I* to 3) i (10 201 l
1.2 .
1
' 1A Pump-Speed Flow Rate, a /v Or v /a g
1 i ( .
i f
I l FIGURE 7: Homologous Torque at Various Punp Inlet Void Fractions for First-Quadrant Pump Operation 1.4 f
i 1.2 .
voto Fractaons
( 3 to 10E )
1g ,
(ens Alono) ,"*, 10 to 1513
'- s '
to not ,
i O to 3t _ .
_ _ _ _ _ _ .: .: = = ~ ~ --,, y3 g, gg, ,
C.s ,
,,,.,.a- (0 to 32)
, *="*'""~
O6 g5 g pggy _ ., _
(20 to Elt) l a
NE ""
- 40 to Son Oa . (60 to 70s) ,
,,so # . iso to m i I
\ ,-
g #
""" a .
, q go gag 3 N23 C2 .
N e " ,-
80 (30 to SOE)
. (50 to 70s) *
$ 0 ~
O2 0.3 ' .1 0.6 0.7 0.3 og g'a p (S0 to 60s) si .c g (60 to 70s) f 3
o (70 to M )
.c a ,
(3 to 105) 06 . (40 to 50s)
I f
- n. ( t. . ,
(0 to 31)
-1 0 .
't '
(30 to acts 12 j (10 to 205) i l
l' 1.4 Pump-Speed Flow Rate, a /v Or v /c
J n e
I .
1 .
I I Z O
O E
1 - 8'a 1 .
(-) O o o>
n
~
- $ e o
O s' z
h O o N
3 to gr.
H
. b 2
O g c O g
U O g O .
Q" t c 94
- O
- O '
re) ~
. tn emus O O N
j Y
O Oo o
g
- _O O sa o
c
' ' ' ' ' . l <
t D O b O N O o o o o o
. O c o. D o. *- 8-8 -
8
- o '
o n. o.
n' W N N -
o o _ N N &
o, .. . i i
, I W 4 CN < J 1 I 4< N g C
e
.-.-4 , , , , , - , . . , , - - - - - ,-
,n.__,- , - . - - . - . ,y ..,,_,,____,,,,,,,n__,- . . , . , , , , . . , , . _ _ , - _ , , . , -
I.
Z 0
I-O i ! or -
_o k . .
O i O a ;
r
' o> -
2
.m* 3
- O o .
O m z
. e 6- o s
5 o
O -
e m
K O- 8 -
E 2
o o o
S g- g O g m
O-o y o .
M .
O 0 .
m -
m 2
o 2 m -
e f - o h
0 W
. C W
D
, , , , ,m , , , , , I I e
g v
8 8 8 8 8 8 8 8 8 8- 8 8 8 8 m ~ ~ - -
o 0 0 :
a o, m ..
m soeoN<;a.zE<m i m
. E e
E
,---,---,n., s --- - -- - - - -
I -
' M d i i 1 4 l 1 h Saturation (2-phase flow at pump inlet) -
7 g '
to
{
%. I6 2-Phase bubbly flow across pump d
Q
.b n -
- e - 1 L 8
- c- T b / ransition, bubbly-to-churn
- 3e -
~
fir turbulent -
l 3* V -
Churn turbulent flow
_ Bubbly flow -
l l
3 as M r_
is
-Uncertainty 10.28 psi i
l l I \ l I
- e a 188 2s3 3ea 4Ee 500 6as 733
~
Time (s)
Figure 10: Differential Pressure Across Primary Coolant Pumps e m..
i l [
t i
~~
"P'-- , - - - . . _ _ , ,_ , _ _ _ _ _ _ _ _
! - 2 l 1 I i i i i I .
3 IL. Data Uncertainty + 5 lbm/ft _
- f. 45
- i . -
%y . -
t u
-} g _ M ,
. - Data
-V .
f - Y M. -
Calculation 1 M c~~ -
I I I I I i i 15 .
e a les 15a asa z!a see asa e Time (s)
Figure 13. Measured Density Upstream of Pump Compared With Prediction Using Pump Motor Power Assuming Constant Volumetric Flow ~
and Pump Head
~
l I 53 i l i i i i i i 2
a 1._ Datauncertaigty _ .
t 5 lbm/ft O a - .
-} .
h..
, 3 1%
g Data 3e - - _
25 - Calculation e sa tea is zee 2:a 3Ee 3se me Time (s)
FIGU,RE 10 Measured Density Upstream of Pump Compared With Prediction Using Pumo Motor Current Assuming Constant Volumetric Flow and Pump Head
- ~.~.
G ~.'" " ~: . . . . _ . . . . . . -- --- - - - - _ . _ . _ _ l *~.* **.~
.. . ~_ . . , . .
.- ,-, ---------w. , - - -,-- -- - , - - - - -
c--e.<--. . , - - , . - - - - - ,<-.--.-----.,-----,,,---------,--me - --
l .
l 4.0 Comparison of the Gentile AP and Pump Current Methods l
After reviewing the data presented in the previous'section, two inven-I tory trending models remain as viable choices. The first model involves j
correlating the Gentile aP with the system voiding. Murdock's theory of two-phase flow through AP meters forms the basis of this approach. The i
e second model uses a correlation between pump current and system voiding.
I The transfer of torque from the impeller to the fluid forms the basis of this model. The advantages and limitations of the models can be compared
~
I in three areas: technical feasibility, licensing,'and safety.
Technically, the pump current measurement is preferred over the Gentile AP measurement based on the number of assumptions required to support each model. To justify the relationship between current and voiding, the l efficiency of the torque transfer at the impeller during two-phase conditions must be quantified. In contrast, the Gentile aP measurement requires knowledge of the transfer of torque at the impeller, the energy losses in the diffuser section, the degradation of developed head and capacity, and the enhancement of frictional losses in the RCS during
, two-phase operation. Consequently, detailed justification of the Gentile AP method may be more difficult. Although as shown in Figure 2 Model 2 shows good agreement with the best estimate prediction of actual voids in the system.
I t
l l
I
t ,
With respect to licensing, both models may be acceptable. The NRC is currently supporting the LOFT pump power / current work. They have
( also given Westinghouse the preliminary approval for a pumps running void 1
measurement based on the AP across the reactor vessel. Therefore,
'f these types of measurement correlations are neither new, nor radical,
, and should be perceived quite favorably by the NRC.
In the future, the proposed void measurement systems may be recuired to meet lE equipment qualification requirements in order to be licensable.
The ability of either system to meet the guidelines for lE equipment is beyond the scope of this study since hardware has not been addressed.
The behavior of the void models to fail high in degraded conditions would be conservative if these measurements were used in RC pump trip circuitry (see Appendix for details of the analysis). Therefore, either model could be an effective indicator of primary system inventory
( and thus would allow clarification of the trip /no trip decision concern-l ing the RC pumps. The potential use of these models in the pump trip circuitry will be discussed in the next section.
l -
l 5.0 POTENTI AL USE OF INVENTORY MEASUREfENT TECHNIQUES IN PUtiP
( TRIP CIRCUITRY
- As explained in Section 1.0, the level of RCS voiding must be 4
- known to determine an optimum pump trip time in the event that the primary system pressure drops below the HPI setpoint. To use the proposed models and their associated voiding signals in L 7
! pump trip circuitry, the following criterion were used:
i 1., The trip should be delayed sufficiently to allow the HPI to make up the inventory lost for the most probable small breaks.
- 2. The trip should be based on a void fraction sufficiently large to be outside of the normal noise band associated with a = 0 line.
- 3. The trip void fraction should be sufficiently small. such that the trip will occur before pump degradation becomes signi ficant.
Pump current / power tests at LOFT have shown that a setpoint of .15 (15%) void will meet the above criterion. In comparison, since the Gentile flow signal can fluctuate due to other transitory effects, a setpoint somewhat greater than .15 may be desirable to avoid this larger band of noise for the Gentile measurement. This higher
. setpoint can be justified since SBLOCA analysis indicates that a
- system can proceed to approximately 0.4 void (40%) before the pumps must be tripped. In any event, either measurement can provide a conservative indication of void which will yield acceptable pump 4
I i trip times for small breaks in shich the HPI cannot keep up.
When breaks in which the HPI can makeup the inventory being lost I
- or when overcooling events occur, an unnecessary pump trip will be avoided and plant shutdoen will be made more manageable.
A conceptual design for a void measurement system using the Gentile AP model would require three inputs:
- 1. AP signal from the Gentile
- 2. RCS pressure from the wide range taps located on the hot legs.
3, Indication of number of pumps running.
The aP signal will be used as the time-varing AP TP input and the RCS pressure measurement will be used by a calculating module to determine the saturation properties v fand v .g The only unknowns that remain are the calibration constants APrand v .r At 100% power with 4 RC pumps running, AP r can be recorded when the Gentile is calibrated using the secondary side heat balance. The reference specific volume can be determined by a calculating module which uses RCS narrow range pressure l
l and loop temperature measurements as inputs. When switching to other
\ -
l modes of RC pump operation, AP r andrv will have to be updated by an j appropriate logic circuit. Five modes of pump operation should be l considered 4
-1. 4 RC pumps on
- 2. 3 RC pumps on
- 3. 2 RC pumps on in one loop while the other loop is idle l 4, 2 RC pumps on - one in each loop i
- 5. I RC pump on i
i
f-Toobtain accurate values of AP r and rv f r each mode, aP r
will have to be recorded and vr will have to be calculated and recorded when the plant is in these modes. In this fashion the date base for the logic f
circuit is developed.
Similarly, a conceptual design for a void measurement system using pump current would require two inputs:
- 1. Pump current measurement
- 2. RCS pressure from the wide range taps located on the hot legs The current measurement will be used as the time-varying I input and the RCS pressure measurement will be used to estimate the suction pressure at the RC pumps, such that a calculating module can determine the saturation properties of andp . The calibration constants I and p will be recorded using a method similar to the one oroposed for the Gentile AP system mentioned above. A logic circuit may not be required to account for changes in pump efficiency during various modes of RC pump operation. Furthur development of the concept is required to confim this.
The void estimates from either of these systems could be coupled with the ESFAS system to provide automatic pump trip.
O e
I
6.0 CONCLUSION
S AND RECOMMENDATIONS Two methods for determining RCS voiding have been developed and have been shown to be technically feasible. The first method utilizes the Gentile AP signal to infer system voiding. The second method utilizes the RC pump current measurement to determine the inventory of the RCS. Both measurements can produce conservative estimates of voiding to allow for more effective accident management.
Although both the pump current and Gentile AP methods are technically feasible, it is recommended that the pump current approach be pursued for further development. This recommendation is based on the following strengths of the pump current approach relative to the Gentile AP:
a) Fewer assumptions are required to support the model b) An experimental data base from scaled pump tests exists c) The pump current scheme may not require implementation of logic to account for the number of pumps running d) NRC is familiar with and has indicated acceptance of the principle i
7- .
REFERENCES
- 1. " Evaluation of Transient Behavior and Small Reactor Coolant System Breaks in the 177 Fuel Assembly Plant " Babcock & Wilcox Co.,
11ay 7,1979.
- 2. " Analysis Summary in Support of an Early RC Pump Trip," and
" Supplemental Small Break Analysis," submitted as Attachment B to B&W plant Owners Letters in response to ilRC Bulletin 79-05C.
- 3. " Generic Assessment of Delayed Reactor Coolant Pump Trip During Small Break Loss of Coolant Accidents in Pressurized Water Reactors",
NUREG-0623, November,1979.
- 4. J. W. Murdock, "Two-Phase Flow Heasurements with Orifices " Trans, ASf1E E (4), 419-432,1962.
- 5. E. Bizon, "Two-Phase Flow fleasurement with Sharp-Edge Orifices and Venturis", AECL-2273, June,1965.
- 6. J. F. Ripken, fl. Hayakawa, " Calibration Study of a 36 Inch HAH 4EL-DAHL Flow Tube (Oconee No. 3), University of Minnestoa St. Anthony Falls Hydraulic Laboratory, Memorandum No. ti-125, August,1970.
- 7. " Reactor Coolant Pump Motor Power or Current Criteria for Reactor Coolant System Inventory Management in Comercial PWR's During Accidents," LOFT Research flemorandum, March,1982.
a.
APPENDIX i
, To give the reader a qualitative " feel" for these models, a simple computer code, VOIDCOM, was developed. V01DCOM models the RCS as a single homogeneous thermo-dynamic node. Within this node, conservation of mass is determined by considering the addition of mass due to the HPI system and the loss of mass out of a small break. Conservation of energy is maintained by summing the energy contributions to the RCS via the core, steam generators, HPI system, and the break. By considering the RCS as a fixed volume system, the pressure of this large node can then be determined using the conservation of mass and energy principles. A somewhat pseudo primary loop containing a Gentile flow tube and a RC pump is then superimposed within the RCS thermodynamic node. The mathematical modeling of the Gentile flow tube and the RC pump is of sufficient detail to allow the data presented in Section 3 to be utilized. The output of these component models is a simulated signal of AP for the Gentile flow tube and current for the RC pump. The two viable voiding models presented in Section 2 are then applied to yield an
- estimate of system voiding (via Gentile AP and pump current) which can be compared
, to the thermodynamic voiding of the system which is known. ,
i l '
i
(*
O
A TMI-2 type scenario was used to create a voiding transient upon which the models and data could be applied. The chain of events are as follows: . (See Figures 13-18)
- 1. At time zero, the reactor is at 100% power.
f i 2. Loss of all feedwater begins just after time zero.
- 3. As the steam generators provide less of a heat sink to the primary system, the RCS pressure rises until the PORV lifts. (t = 15 seconds)
- 4. The PORV sticks open and water continues to flow out of the break.
- 5. As the RCS pressure drops below the HPI setpoint, the HPI flow is throttled off. (This is done in this example to create significant voiding in a small period of time.)
- 6. Auxiliary feedwater is valved in and reaches full capacity (t = 4 minutes)
, 7. At.20 void, HPI flow is reinstated at full capacity (t = 40 minutes)
, 8. System inventory then recovers and the RCS returns to a subcooled condition. (t = 56.5 minutes) l The transient mentioned was analyzed twice, once assuming non-degraded
! pump performance, and then a second time assuming that the pump would experience some degradation. In the second analysis, a pump head and torque degradation multiplier was applied which approximates the data t
presented in Section 3 for the Combustion Engineerino data at pressures around 1000 psia. The expressions which were used for the head degradation multiplier, Mg , are as follows:
MH = 1.0 0.0 $a5 0.14 MH = 1.4242 - 3.03 a 0.14 < a < 0.25
i.
Similarly, for the torque degradation multiplier, M ,T the expressions i are,
, MT = 1.0 0.0 5 a 5 0.10 M = 1.2 - 2.0 a T
0.10 < a < 0.25 t *
?
l Note the expressions for the multipliers were limited to the range 0 a < 0.25 due to the dynamics of the problem analyzed.
Review of figures 10 through 10 gives the reader a quantitative i
feel for the two models presented,and the impact that the pump degradation can have on the resulting estimates of system voiding.
! With these figures. comparisons can be made which allow the uncer-tainties associated with each measurement technique to be estimated.
1 In figure 13, the transient behavior of the RCS system pressure is shown for both the degraded and non-degraded cases. Due to the
- simplistic nature of the VOIDCOM thermodynamic model, the system pressure is identical for both cases. When a loss of all feedwater occurs, the pressure in the RCS increases from the steady state value until the PORV lifts and sticks open. The pressure then drops
- until it settles out to a saturation pressure detennined by the heat
- sink provided by the steam generators. At approximately 56.5 minutes, Ii the HPI flow, being greater than the leak flow, has brought the PCS back into a solid condition. The pressure then rises until a pressure
- level is reached where the leak flow becomes equal to the HPI capacity (p
- 2050 psia).
1 .
} -
l The loop flow rate versus time is shown in figure 14. In the non-degraSed i case, the reduction in flow is due to changes in system density only (equation 1, Section 2). By comparison, when_the system void exceeds
.14, the flow rate is further reduced due to the now degraded pump head- '
(Figure 15). The degradation in flow rate is not large relative to the '
loss in developed pump head because it is assumed that'the flow 'cate varies as the square root of the AP across the pump.
s In figure 16, the transient behavior of the hydraulic torque is shown. - '
When non-degraded behavior is assumed, the torque varjes as a function '
of density only. Under degraded conditions (- a > .10), Ehe transfer of torque becomes less efficient as void increases.
The void estimates predicted using the Gentile AP and pump current models assuming non-degraded pump performance are compared to the actual thermodynamic void in figure 17. It can be seen that both models produce good estimates of system voiding. Figure 13 depicts the same estimates assuming pump degradation occurs. For both the Gentile aP and pump current method, the degraded pump performances cause the predicted voiding to be overstated.
F i
e
__J
\- .
f n
L h * .d--
I f
e i 25.53_j i
l
\- .
n3.u.
e a-. s t f
cu ..
I O !
- en3.. e,s .y C '
! N c: !
C- i
'~
l c L,, . .,-. w .l M i m .
l f4. -
w i
-~
c- ,'
15.5s ,_
4 .
, 1 i
e /
e7 ;
t >- !
! L7 j
,, -- {
v . s.~ _. .
- }
l l
8 1 !
I
!' t ti0il-DEGRADED AllD ii DEGRADED CASE i.,.5.~_,I.
.. ,, % 2
- W i i 1
- 5 . . -- , , , , i , i
.--- 1.33 23.3= 33 32 '; ;;; S;'U 2 a'-' ' ' ~
~
1 TIME (MIN) r mm m Y
neunt u.. SYS!t.M rn{bS8u u' n NE \i F0 Y, C n&
- + - - - - -e.-,, - . . - , - - y- , , , , , . , . , -
y -e-..
Q e-. .--
4 l
fl0ft-DEGRADED CASE 10 . Jew I
~
- ' .\y /
-s i
N- ,
i
+
D.
.j
. e . DEGRADED CASE l.
o . i x ,_ _ __ i
=_-
l_ %,
=
...a a i e
l d
q
. .. a.
l 1 e !
- t o .
i,
.. ..- i t
i
!* l
... t ._ .._ _. _a.___
l t
TIME (MIN) i nount 14: LOOP FLOW RFE VS IME o
t .
g
.s.v..
f i
i t
i
,,.s__
4 I
NON-DEGRADED CASE !
a . u_ __., _. .
y :
I h.
k 5
_., N.%
o
- B _ ""
t DEGRADED CASE '
x .
I,
' v,
'v I
c_
7.._)*____.,,.
c_
q I e
a i
= ,.
,. 5.____ i
!E- 1 2
, c.. I
, 1
. . . _ . 1
- 3. _ _ _ .
l e
i :
. s., ___ ,
- t. * .
. i
-i
. w ._ 23. = n: e .:= d_.___--
--up igTps i "16.
. t..1J Ilwt.
FIGURE 15: i= UMP DELTR F VS ii
e 1
, e
- A s, u,s_.
I I, '
i !
l !
! I e
35.5 C J t
e 0
t' NON-DEGRADED CASE 2 l
a i
.- / .
- .e
- i u
awmJ. e .at s
c
/// i
- M m i 1 .
/ DEGRADED CASE .
o ~~
- 5=Js L i ;
y ;
C i -
i I ,
w !
i m
_. .a c .-_
J~_ I ,
L_. ! r O l O-
__ i
.O i 15 .u,.,., l C-r-
'D C_
,.--t
.V .wsef l
e- --
J Ja s-J = .J s.
i i a i i , - . -,-
J*dwo Jeddd o --
L = J dd
- ddd W Jed bd-ded J'N*d 8
T. T..L M.C L' .M.. T T, P'1 tf5 rye I
n FIGURE 16: r Mn r i On m 0Lg-t ! lii
j .
. r. :
f 1
. 350 I
f .'
.2 as i !
I I s
255m s
A
'. p
.. + , e 9 /
E 23:_.! GEllTILLE AP /
ESTI!1 ATE g/ .
c j /
$ i
/ [/r
> i
> ./
.laa. _.l ACTUAL VOIDIllG /
/ , .
./ .
. j i PUf1P CURREllT i j ESTI!1 ATE
- i j /r i
.1 c:a p' f i.
' !I \
\
/' x l
, .0M_ \
I .
l
.or '\ -
j 7 i i i ?
(
l c .:= ts.or 20.3 % 33.0 % 4; .:= ss .;; 5; 03 I TIME (MIN)
RERE 17: VOID FRRCT T ON VS TIME ( NON-DEC-R.90ED )
i t .. _ _ _ _ _ -
r I.,
.or s
/-
.353 ] .
/ 1
/
PUt1P CURREilT I
ESTIl%TE i i
I I
.33: .
i I i
\ i GENTILLE AP i'
"'UU-l ESTIl%TE / '
b
/
/
w :
- c) n
- i
\; -
i g ; ,
2 200J
~
f l / ,
o :
/ .-
.- 1, j -
C i
l l ! \
/ / \
,/ ..
.15c_ -
ACTU/d. VOIDING :
[ \;
.1 CO .
j
/
/
. ss_
/ h r
\..
.\ -
.32,d i i i i ,
53.:r 15.0 % 23.Or 33.0 ; 80.0 % ss .:I 3:
I TIME (MIN)
FIGUllE 13: VOID FRACTION VS TIME (DEGRADED) f